STATE OF CilLIFOEKia DEPAKTMEHT OF NATOHAL HESOUECES CLAYS AND CLAY TECHNOLOGY BULLETIM 16i * DivBiON OF mmm FERRY BunjDmG, SMJ maKasco THE LIBRARY OF THE UNIVERSITY OF CALIFORNIA DAVIS STATE OF CALIFORNIA GOODWIN J. KNIGHT. Governor DEPARTMENT OF NATURAL RESOURCES DeWITT NELSON, Director DIVISION OF MINES FERRY BUILDING, SAN FRANCISCO 11 OLAF P. JENKINS, Chief SAN FPLOlNCISCO BULLETIN 169 JULY 1955 CLAYS AND CLAY TECHNOLOGY PROCEEDINGS OF THE FIRST NATIONAL CONFERENCE ON CLAYS AND CLAY TECHNOLOGY Assembled and edited by JOSEPH A. PASK (Professor of Ceramic Engineering, Division of Mineral Technology, College of Engineering, University of Califomia, Berkeley) and MORT D. TURNER (Assistant Mining Geologist, California State Division of Mines) Price $4,50 LIB-RARY UNlVERSn Y OF CALIFORNIA DAVIS CONTRIBUTING AUTHORS f V. T. Allen. ])i'partm('ut of Geology. St. Louis Univer- sity, St. Louis. Missouri * Isaac Karshad. Leeturcr. Department of Soils, and Assistant Soil Chemist, Agrrieultural Experiment Sta- tion, Tniversity of California. Berkeley, Califoi-nia * Thomas F. Bates, Associate Professor of IMineralofry, School of Mineral Industries. The Pennsylvania State University, State Collepe, Pennsylvania W. E. l>eriinian. Piiillips Petroleum Ciniipany. r.artles- \ille, Oklahoma W. F. Bradley, Illinois Geological Survey. Urbaua, I llinois * (ieorne W. Brindley. Reader in X-Pay Physics, riiysics Lahoratories, Tlie University. Leeds. Enoland ■' T. F. linehrer. Professor of Agricultural Chemistry and Soils; Head. Department of Ag-ricultural Chemistry anil Soils. Colleg-e of Ag-ricidtiu-e, University of Arizona, Tucson, Arizona W. T. Cardwell Jr., California Research Corporation, La I labra, California .1. ::\I. Catterton, Tide AVater A.ssoeiated (Ml Company, A'entui'a, California 11. F. Coffer, Continental Oil Company, Ponca City, ( tklalionia I. I. Cornet. Division of Mechanical Engineering, College of Engineering, University of California, Berke- ley, California * Lannes E. Davis, Associate Professor of Soils and Soil Chemistry. Department of Soils, University of Cali- fornia. Davis, California C. G. Dodd, Research Associate. Production Research Division. Development and Research Dejiartment, Conti- nental Oil Company, Ponea City, Oklahoma J. W. Earley, Gulf Research and Development Com- pany. Pittsburgh, Pennsylvania ]\I. D. Foster, U. S. Geological Survev, Washington, D. C. J. E. Gieseking, Department of Agronomy, University of Illinois, Urbana, Illinois \V. C. Goins.Ir.. Gulf Oil Corporation. Houston. Texas Irving Goldberg. Institute of Transportation and Traffic Engineering. University of California, Berkeley, California * Ralph E. Grim, Honorary Chairman, First National Conference on Clays and Clay Technology ; Research Professor. Dejiartment of Geology, I^niversity of Illinois, Urbana, Illinois J. L. Gring, Sinclair Research Laboratory, Inc., Har- vey, Illinois J. L. Hall, Stanford Research Institute, Stanford, California E. A. Hauser, Highway Research Board, National Re- search Council, Slassachusetts Institute of Technologj% Cambridge, Massachusetts * Edward C. Henry, Professor of Ceramics, and Chief, Division of Ceramics. School of Mineral Industries, Peinisylvania State University, State College, Pennsyl- vania * Asterisk marks the names of authors who contributed papers to this volume. Other persons listed contributed to the discussion of tlie papers. t Affiliation as of date of conference. * F. X. llveem. Materials and Research Engineer, Cali- fornia Division of Highways, Sacramento, ('alifornia * A. L. Johnson, Director of Research, T.^niversal- Rundle Corporation, Xew Castle, Pennsylvania * Xorris Johnston. President, Petroleum Technologists, Inc., ^Montebello. California J. W. Jordan, Baroid Sales Division. National Lead Company. Houston, Texas "NV. 1). Kellei-, De|)artineiit of Geology, University of .Missouri, Columbia, Missouri * W. P. Kelley, Professor of Soil Chemistry Emeritus, Department of Soils, University of California, Berkelej'', California * Paul F. Kerr, Department of Geology, Columbia University. New York. New York * ]\Iyrle E. King. I'etrographer, Petrographic liabora- tory. Engineering Laboratories, U. S. Bureau of Recla- mation. Denver, Colorado E. B. Kinter, U. S. Bureau of Public Roads, Wash- ington, D. C. \V. J. Knapp, Department of Engineering, University of California, Los Angeles, California D. P. Kryniue, CVnisulting Engineer, Berkeley, Cali- fornia * Delmar II. Larsen, Consulting Chemist and Clicmical Engineer, Los Angeles, California * D. R. Lewis, Senior Chemist, Exploration and Pro- duction Research Laboratory, Shell Oil Company, Hous- ton 25, Texas * D. M. C. MacEwan, Pedology Department, Rotliam- sted Experimental Station, Harpenden, Herts, England E. S. JIardock, Well Surveys, Inc., Tulsa, Oklahoma Duncan ilcConnell, Division of Geological Survey, State of Ohio, Ohio State University, Columbus, Ohio G. A. Mickelson, Filtrol Corporation, Los Angeles, California * R. C. Mielenz. Head, Petrographic Laboratory, En- gineering Laboratories, U. S. Bureau of Reclamation, Denver. Colorado * T. H. Milliken. Assistant Director of Research, Re- search and Development Laboratories, Houdry Process Corporation, Marcus Hook. Pennsylvania * G. A. Mills, Director of Research. Research and De- velopment Ijaboratories. Houdry Process Corporation, ^larcus Hook, Pennsylvania II. II. ]\Iurray, Indiana Geological Survey, University of Indiana, Bloomington, Indiana * P. G. Nahin. Research Division, Brea Research Center, Union Oil Company of California, Brea, Cali- fornia * A. G. Oblad, Associate Manager of Research and Development, Research and Development Laboratories, Houdry Process Corporation, Marcus Hook, Pennsj'l- vania H. van Olphen, Shell Oil Company, Houston, Texas * Bernard B. Osthaus, Chief Silicate Analyst, Gulf Research and Development Company, Pittsburgh, Penn- sylvania Roy Overstreet, Department of Soils, University of California, Berkeley, California (3) A. Pabst, Department of Geological Sciences, Univer- sity of California, Berkeley, California * J. B. Page, Professor of Agronomy and Soil Physics, Department of Agronomy, Agricultural and Mechanical College of Texas, College Station, Texas Padraic Partridge, Filtrol Corporation, Los Angeles, California Joseph A. Pask, Division of Mineral Technology, Uni- versity of California, Berkeley, California C. S. Koss, U. S. Geological Survey, Washington, D. C. * Richards A. Rowland, Senior Geologist, Exploration and Production Technical Division, Shell Development Company, Houston, Texas • Asterisk marks the names of authors who contributed papers to this volume. Other persons listed contributed to the discussion of the papers. H. R. Shell, U. S. Bureau of Mines, Norris, Tennessee M. Soldate, California Research Corporation, La Habra, California R. L. Stone, Department of Ceramic Engineering, University of Texas, Austin, Texas M. W. Tamele, Shell Development Company, Emery- ville, California M. P. Tixier, Sehlumberger "Well Surveying Corp., Houston, Texas B. A. Vallerga, Department of Civil Engineering and Irrigation, University of California, Berkeley, California W. J. Weiss, The Texas Company, Long Beach, Cali- fornia * M. R. J. Wyllie, Chief of the Petrophysics Section, Gulf Research and Development Company, Pittsburgh, Pennsylvania (4) LETTER OF TRANSMITTAL The lioxoiiAiii.K Goonwix J. Kn-kutt Governor of the State of California Sir: I liiivc llic lidiidi- to transmit lierewitli Bulletin 169, Chii/s and Cloy Tcchiiolofjif, prepared under the direction of Olaf P. Jenkins, Chief, Division of Jlines. Dei)artment of Natural Re- sources. This volume is the product of a symposium, the First. National Conference on Clays and Clay Technology, held at tlie University of California in Berkeley Jitl.v 21-25. 1052. It con- tains contributions by 25 eminent authorities, which comprise a summary of tlie basic informa- tion on clays. Heretofore, no such summarj- has been available; indeed, much of the information presented here could have been obtained only by diligent library research into rare periodicals, many of them in foreign langitages, b.v study of unpublished manuscripts, and by personal consultation with experts. Bulletin 169 will serve as an invaluable i-eference book for years to come. Its data on the fundamental properties of clays, brought together by research men in the fields of physics, chemistry, geology, engineering, and industry, should serve both science and industry not only in California, but throughout the world. Professor Joseph A. Pask of the Universitj- of California, and ilort D. Turner, Assistant Mining Geologist at the California Division of Mines, both membex's of the organizing committee of the conference, served as technical editors for the book. Respectfully submitted. DeWitt Nelson, Director Department of Natural Resources Jannarv .T. 1955 (5) PREFACE Tlie Fii-st National CoiiforeiU'e on Clays and Clay Teclnidloiiy was held at the I'niversity of I'alifornia at Berkeley July 21-2o, 1952. The primary purpose was to brinji together clay workers from all fields in order that the fundamental knowledge of elays that has been gained in each of these fields, along with its application to clay teehnology, might be summarized. Clay is a basie raw material in maiix' industries, among them agricultui'e, construction, ceramics, food process- ing, ami the petroleum industry. The papers presented at this conference emphasize the continual broadening of clay researcii within these industries as well as within physics, chemistry, geology, and engineering, ilining and ntili/ation of cla.y have always been basic to California economy because of the large and varied clay resources that occur here, and the importance of clay has increased tremendously during the last few years. New centers of clay research have started and grown into prominence in California; this is one of the reasons why the tirst national clay conference was held here. The success of the conference and the economic and scientific value of these ])apers are a recognition of the importance of the sub.iect to California and to the natitui. The conference was organized into two broad divi- sions — one concerning the geology, mineralogy, proper- ties, and identification of clays; and the other concern- ing applications of clay teehnology in soil science, soil mechanics, the petroleum industry, and ceramics. The papci-s i)resented represent the information that has been gathered by workers in each of the various sciences or technologies that deal with clay. These x)apers have been submitteil for i)ublication in order that the discoveries and theories of each field of researcii and application may be made available to the many industries and sci- ences that use or study clay. While each subject was necessarily restricted to the limited space available, some of the more pressing questions are brought out and am- liiitied in the discussions that follow each paper, and bibli(igra]ihies are apjiended for those who wish to go further into any jiarticiilar subject. Universit.v of California groups that sponsored the conference officiall.v were : Division of Mineral Tedinolog.v of tlic DciNirloicnt of Eiigi- npprinK Division i)f Civil Knf;inoeriii^' of tlic Dciiailiiiciit of Eiisi- nceriny: Institute of Triiusportation and Traffic Engineering Department of (Jenlogieal Sciences Department of Soils of the College of Agriculture I'niversit.v E.xtension Coo])erating agencies and societies were: ("alifornia State Division of Mines United States Bureau of Keclamatiou United States Bureau of Mines United States Geological Surve.v American Association of Petroleum Geologists American Institute of Mining and Metallurgical Engineers AuH'ricaiL Society for Testing JIaterials — Subcommittee R-6 of Ci)mmillee D-18 Mineralogical Societ.v of America Soil Science Societ.v of America National Cla.v Minerals Committee The Conniiittce on Clays and Clay Technology, Uni- versity of California, consisted of: Joseph A. I'ask, Division of Mineral Technology — Ceramic Engineering, Universit.v of California, Chairman. Lsaac Barshad, Department of Soils, Universit.v of California. Uarmer E. Davis, Institute of Transportation and Traffic En- gineering. University of California George L. G.ites, U. S. Bureau of Mines, Petroleum Experi- ment Station, San Francisco Irving (Joldlierg, Institute of Transportation and Traffic En- gineering, Universit.y of California AValler P. Kelley, Department of Soils, University of Cali- fornia Alexander Klein, Division of Civil Engineering. University of Califoruia Adolpli I'alist, Dep.-irtment of Geological Sciences, Univer- sity of California .1. SchliH'ker, U. S. Geological Survey, Engineering Geolog.y ISrancli. San Francisco ^Villlur H. Somertou, Division of Jlineral Technology — Petro- leum Engineering, University of California Mort D. Turni-r. State Division of Mines, San Francisco Professor Ralph E. Grim of the University of Illinois served as Honorary Chairman of the conference. The attendance of Drs. (ieorge W. Brindley and Douglas M. C. MacEwan from England was made possible through the assistance of the following companies : American Colloid Company Attapulgus Clay Company Baroid Sales Division, National Lead Company California Kesearch Corporation, Standard Uil Comjiany of California Continental Oil Company Filtrol Corporation Gladding, McP.ean & Company Gulf llesearch iV i >eA'eioi)nient (.'ompany IlMiiitluIu < )il (..'(O'lioraf inn .1. M. lluher ('orporation IliMulile Oil & Refining Company Lane-Wells (Company JIacco Corporation Magnet Cove Barium Corporation Monsanto Chemical Company Ohio Oil Comjiany Pacific Cl.iy Pruducts Company Ricldield Oil Corporation Schlumlierger Well Surveying Ccu'poration Sierra Talc & Cla.v Company Signal Oil & Gas Comjiany Socony-Vacuum Company, General Petroleum Corporation, Magniilia Petroleum Company Sohio Petroleum Company Stanolind Oil .and Gas Compan.y The Texas Ciim|iany Union Oil Company of California Particular thanks are extended to the California State Division of Mines for undertaking the ta.sk of publish- ing the proceedings. (T) ABBREVIATIONS /» Micron (s) A Average effective charge A Angstrom unit(s) Al Aluminum A.P.I. American Petroleum Institute B Boron Ba Barium bbl, bbls Barrel, barrels B.I.P. Bi-ionic potential C Carbon C Centigrade Ca Calcium CAT-A Catalytic Activity Test A cc Cubic centimeter (s) CI Chlorine cm, cms Centimeter, centimeters cp Centipoise(s) Cr Chromium Cs Cesium DTA Differential thermal analysis e Electron emf Electromotive force F Fluorine F Fahrenheit Fe Iron ft Foot, feet ft^ Cubic foot, cubic feet g Gram(s) H Hydrogen K Potassium kg Kilogram (s) kX unit Equals 0.90798 Angstrom unit 1 Liter (s) lb, lbs Pound, pounds M Mol, molal, molar m Meter (s) sq m Square meter(s) md Millidarcy (s) me. Milliequivalent(s) Mg Magnesium mg Milligram (s) ml Milliliter (s) mm Millimeter (s) Mn Manganese M.S.T. Meyer-Sievers-Teorell equation or theory mv Millivolt(s) N Nitrogen N Normal (chemical) Na Sodium O Oxygen P Phosphorus Pd Palladium ppm Part(s) per million psi Pound (s) per square inch Pt Platinum R The "basic" elements (usually aluminum, magnesium, and iron) Rb Rubidium rpm Revolution (s) per minute S Sulfur S.E. Sand equivalent Si Silicon S.P. Self-potential sq Square Th Thorium Tl Thallium wt. Weight Zn Zinc (8) CONTENTS Page INTRODUCTION H Objectives of the First National Conference on Clays and Clay Technology, and Definitions of Terms Used in the Industry, by Ralph E. Grim 13 PART I— GEOLOGY AND MINERALOGY OF CLAYS 17 Formation and Occurrence of Clay Minerals, by Paul P. Kerr 19 Structural Mineralogy of Clays, by George W. Brindley 33 PART II— PROPERTIES OP CLAYS 45 Electrochemical Properties of Clays, by Lannes E. Davis 47 Idu Exi/luinjre Reactions of Clays, by D. R. Lewis 54 Adsorptive and Swelling Properties of Clay- Water System, by Isaac Barshad 70 Interlamellar Sorption by Clay Minerals, by Douglas M. C. MacEwan 78 PART III— METHODS OF IDENTIFYING CLAYS AND THE INTERPRETATION OF RESULTS 87 Particle Size Distribution in Clays, by A. L. Johnson — ' 89 Interpretation of Chemical Analyses of Clays, by W. P. Kelley 92 Interpretation of Chemical Analyses of Montmorillouites, by Bernard B. Osthaus 95 Petrographic Study of Clay Materials, by Ralph E. Grim 101 Dye Adsorption as a Method of Identifying Clays, by Charles G. Dodd 105 Infrared Analysis of Clays and Related Minerals, by Paul G. Nahin 112 Identification of Clay Minerals by X-Ray Diffraction Analysis, by George W. Brindley 119 Electron Microscopy as a Method of Identifying Clays, by Thomas F. Bates 130 Differential Thermal Analysis of Clays and Carbonates, by Richards A. Rowland 151 PART IV— CLAY TECHNOLOGY IN SOIL SCIENCE 165 Role of Physical Proi)erties of Clays in Soil Science, b.y J. B. Page 167 Role of Chemical Properties of Clays in Soil Science, by T. F. Buehrer 177 PART V— CLAY TECHNOLOGY IN SOIL MECHANICS 189 Importance of Clay in Applied Soil Mechanics, by Francis N. Hveem 191 Physical-Chemical Properties and Engineering Performance of Clays, by Richard C. Mielenz and Myrle E. King 196 PART VI— CLAY TECHNOLOGY IN CERAMICS 255 Clay Technology in Ceramics, by Edward C. Henry 257 PART VII— CLAY TECHNOLOGY IN THE PETROLEUM INDUSTRY 267 Use of Clay in Drilling Fluids, bv Delmar H. Larsen 269 Role of Clay in Well-Log Interpretation, by M. R. J. Wyllie 282 Role of Clay in Oil Reservoirs, by Norris Johnston 306 Use of Clays as Petroleum Cracking Catalysts, by T. H. Milliken, A. G. Oblad, and G. A. Mills^_ 314 (9) INTRODUCTION Page Objectives of the First Xatioual Conference on Clays and Clay Technology, and Definitions of Terms Used in the Indnstry. by IJalph E. Grim 13 OBJECTIVES OF THE FIRST NATIONAL CONFERENCE ON CLAYS AND CLAY TECHNOLOGY AND DEFINITIONS OF TERMS USED IN THE INDUSTRY By Ralph E. Grim • Clay mineralogy is a very broad subject that includes work in many different fields : in chemistry, physics, mineralogy, and geology ; in the applied fields of ceram- ics, engineering, and agriculture. ^Yorkers in da.v min- eralogy liave come into it from many different disciplines, and they have many sjiccial interests. However, workers in clay mineralogy and technology, regardless of their special interests, have much in common, because basically their problems all revolve around the structure, proper- ties, origin, and occurrence of clay minerals. An exten- sion of the knowledge of any of these attributes of the cla.v minerals will be of value to all workers in the field. Students of clay mineralogy and clay technology in the United States have i-ealized for some time that much was to be gained by a meeting of all workers in this field, regardless of their special interest or the particular dis- cipline through which they approached their problems. The meeting together of ceramists, engineers, agricul- tural chemists, mineralogists, and physicists, to discuss aspects of clay mineralogy on whieli each has worked and thought, cannot but be .stimulating and lead to new ideas and concepts. In accordance with the thought that progress in clay mineralogy and clay technology woidd be enhanced by tlie cooperative effort of all workers in the field regard- less of their s])ecial interest or discipline, a National Clay IMinerals Committee was established in St. Louis at the time of the 1951 national meeting of the American Institute of Mining and IMetallurgical Engineers to ex- l)lore the desirability and feasibility of starting some kind of an association to include all workers in clay mineralogy. The committee, of wliich I have the pleasure of serving as chairman, unanimously approved a plan to affiliate with the National Research Council, at an open meeting on July 22, 1952. Ohjrcfivrs of the Conference. For the arranging of the present conference, clay mineralogists and technolo- gists everywhere are indebted to the University of Cali- fornia group, and to Professor Joseph A. Pask in par- ticular. It has been a tremendous amount of work, and a magnificent jol) of organizing and planning has been done. The objective of this conference is to bring together workers in clay mineralogy and clay technology so that they may express their ideas and discuss their problems, and in this way learn a little more aboi;t the occurrence, structure, and properties of the clay minerals, and the way in which the properties influence the large rock- properties of clay materials, which in turn determine the manner of use or the way in which they can best be eon- trolled in engineering. The conference provides also an opportunity to discuss matters on which there are differ- ences of opinion, so that some agreement may be reached on controversial points. • Research Professor, Department of Geology. University of Illinois, I'rbana, Illinois. Honorary Chairman, First National Conference on Clays and Clay Technology. Definitions of Terms. In the field of clay mineralogy and clay technology, there are a few terms which have more than one meaning, or are used differently by work- ers approaching the subject from different fields. Thiis, the term clay itself is used to designate a rock with cer- tain physical attributes and a certain general composi- tion. It is also used as particle-size term to designate the fraction of a rock composed of particles finer than a cer- tain size grade. Unfortunately the upper limit of the size grade designated as clay is not the same in all fields. Geologists, for example, place the limit at 4 microns, whereas many others place it at 2 microns. In the case of the size-grade designation, there seems to be a growing tendency to use "clay grade" or "clay size grade" and to reserve clay as a rock term. This is a gratifying ad- vance. Clay, as a rock term, is difficult to define precisely. It is a natural material, very fine grained, ancl essentially composed of silicate minerals — the clay minerals; very fine-grained natural materials of other compositions, such as bauxites and diatomaceous earths, probably should be excluded. Further, cla.ys are earthy and with few ex- ceptions (notably the flint clays) are plastic when mixed with water. A brief examination will permit almost any- one familial- with clays to decide whether a given ma- terial sliould be so classified. There are, however, some borderline materials containing appreciable amounts of minerals other than the clay minerals on which all work- ers would not agree as to whether they should be termed clays. This is bound to be true, since nature did not make all materials to fit into man-made compartments. Soil is another term used differently by different groups interested in clay mineralogy. Soil to tlie geologist is the weathered regolith at the earth's .surface that sup- ports vegetation. To the agronomist, soil is the loose rego- lith at the earth's surface, but it need not be weathered nor support any vegetation — it may be gravel, for ex- ample — which the geologist would never call soil. To the civil engineer, soil is any loose material at the earth's crust regardless of particle-size distribution, compo- sition, or organic material. The engineer (Terzaghi and Peck, 1948) divides the material of the earth's crust into two categories : hard consolidated material or rock, and loose material called soil. The geologist is frequently surprised to hear the engineer speak of a deeply buried gravel zone as a soil. These definitions of .soil are deeply entrenched, and it is my opinion that redefinitions would not be accepted or used. "Workers in the field must know the variation in definition as used by diffei'cnt groups of workers. Shale is _^ another term which has some variation in definition. To the geologist the term indicates a laminated material with the lamination parallel to the bedding and one which has not been metamorphosed to the extent of developing new minerals. To others, shale is merely a hard, indurated clay. (13) 14 Clays and Clay Technology [Bull. 169 I have found the expression "claj- material" very useful in writings destined for several groups which may have somewhat different definitions of clay. soil, etc. It implies auy argillaceous material composed largely of the clay minerals, whether or not it is weathered, at the earth's surface, plastic, or laminated. With regard to definitions of the clay minerals, it seems to me that workers are in remarkably good agree- ment. Halloysite minerals are an exception and the situation is well known. In my opinion the best solution to the halloysite nomenclature problem was that sug- gested by the group meeting in Amsterdam in 1950 at the time of the International Soils Congress (1951). This group recommended that halloysite be used as a general group name and that when the state of hydra- tion of a sample was known a suitable descriptive prefix or suffix be Tised, such as halloysite (2H2O'). It is too soon to tell whether this suggestion will be followed by clay investigators generally. There is still some variation in the usage of a term for the mica clay minerals, excluding the chlorite and ver- miculite types, although the general term illite for all such clay minerals is very widely used. The problem of the nomenclature of the illite minerals is exceedingly difficult becaiise of their range in composition and strue- tui'al attributes, and because there appear to be grada- tionals between forms Avhich at their extremes have very different properties. An example is the probable grada- tion from expanding lattice to non-expanding lattice minerals. A further difficulty is that the illites freqi;ently occur in mixed-layer structures. It is, for example, some- times difficiilt to tell Avhether a sample composed largely of illite shows a slight lattice expansion or contains a few mixed layers of montmorillonite. In my opinion a satisfactory nomenclature and classification for these minerals cannot now be derived. Additional information on the range of composition and structure, and perhaps better criteria for identifying the various possible forms of these minerals are needed before a satisfactory classi- fication can be worked out. In the meantime, it is essential for all of us to describe o\ir samples as precisely as possible from the standpoint of structure and composi- tion. "We particularly need additional data on samples of pure illite. Unfortunately, samples of illite uncon- taminated by other material are hard to find. Clay and soils have been studied in some fashion ever since people became curious about natural things. In- vestigations into the nature of clay materials — the build- ing blocks which compose them — date back abovit to the beginning of scientific research. The clay materials proved particvilarly difficult to study because they are extremely fine grained and because many of them are very complex mixtures. About 30 years ago new research tools, particularly X-ray diffraction, became available, which permitted for the first time precise determination of the identity and character of the components of clay materials. Since the early nineteen twenties, additional analytical tools, such as differential thermal analysis and electron micro.seopy, have been developed and adapted to clay analysis. With the development of the new or impi'oved re- search tools, there came a great expansion of interest in clay research. Many new workers began to devote them- selves wholly to clay studies. In many fields in which elay materials are important, serious basic studies of composition and properties previously neglected, were undertaken. There has been in the last 30 years, but particularly in the last 15 years, a great flowering of clay research. I think everyone will agree that a vast amount of sound fundamental knowledge about clay materials has been gained in this interval. The gain has come from the work of people in many fields — chemists, mineralo- gists, physicists, engineers, and agronomists. Indeed, clay mineralogy is fortunate in attracting workers from many fields. Its ramification into so many fields is per- haps its greatest attraction for many research workers. The flowerino- of clay studies has led to the general acceptance in all disciplines of the clay-mineral concept of the nature ofclaj's. Engineers, soil scientists, cera- mists, and others are beginning to think of their prob- lems in terms of clay mineralogy, although the concept is by no means universally known. Only a few months ago a paper appeared in a leading geological .iournal in the United States in which the author blandly assiimed that all argillaceoiis sediments were composed of kaolinite and proceeded to develop elaborate theories on that basis. At this point I would lik(> to inject a word of caution. Some of us were perhaps led to carry the notion of clay mineralogy to the extent of considering all clay materials to be composed solely of crystalline substances. Recent studies have shown that there are some clay materials — probably not very many, but some — which do have some material which seems to be amorphous. Such material may be very hard to identify and the question of whether a component has a very low degree of organiza- tion or is completely amorphous sometimes cannot be decided. The great strides that have been made in clay min- eralogy do not mean that all problems have been solved. Indeed, the large areas of this field whei-e knowledge is particularly needed, and for which much information is likely to come in the immediate future, can oidy now be seen clearly. Structural studies are, of course, the underlying fundamental work that must precede almost any other investigations. Further information on the structure of montnutrillonite and illite is needed : data on the exact population of the octahedral and tetrahedral positions; the site of the charge; the limits of composition for the expanding lattice ; an explanation of the great difference in the catalytic properties of various montmorillonites; the nature of possible reactions between absoi-bed organic molecules and the montmorillonite structure; and a solution to the apparent discrepancj- between X-ray and electron microscopic data for halloysite. Information on the system clay-minerals-water is essential to an understanding of such properties as plasticity, bonding strength, and sensitivity, which fre- quently determine the utility of a clay material. There is general agreement. I believe, that the initially ad- sorbed water is not liquid water; but further informa- tion on the exact nature of this water — the influence of adsorbed cations and anions — its variation with relative himiidity and distance from the clay mineral surface — is needed. Introduction 15 Textiiral studies of clay iiuitorials in tlieir natural state are the least far advanced. What holds relatively loose elay materials together — what is the nature of the bondinjr force — what is the influence of water, and of various ions — are not known. Engineers are particularly interested in this problem. Tt is not enougli to tell them that a soil is com|)osetl of illite and quiirtz — they need to know how tlie elay mineral particles fit together and what holds them together. I'nfortunately, no one has yet devised adequate teehnitiues for studying the tex- ture of elay materials in their natural state. Another frontier is the change tliat takes place wlien clay minei'ais are heated. What actually lia])pens when the hydroxyls are driven off — what determines the tem- perature at which they are lost — what is the nature of the structural shift when a new phase develops? Such studies have great ]U-actical value in the field of ce- ramics. They also have great general scientific value and .some research going forward in this field may well alter our thinking about the interpretation of phase diagrams. Fortunately the development of high-temper- ature X-ray techniques permitting tlie determination of the ])luise at an elevated temperature while it is in the furnace has greatly enhanced researches in this field. Other frontiers could be mentioned : for example, the essential conditions for the genesis of the various clay minerals; but the foregoing will suffice to show that actually we do not yet know very much about clay materials. SELECTED REFERENCES Interiiiitional Soils Consross, 1!l.")l, The iKmiencIature of clay niin- cral.s: Am. Mini'ralojjist, vdI. 36, pp. 370-371. Terzajilii, Karl, and Peek, II. B., 1948, Soil mechanics in engineer- ing practice : 566 pp., New Yorlc, John Wiley and Sons. PART I GEOLOGY AND MINERALOGY OF CLAYS Page Formaliou and (Jccui'reiice of Clay Minerals, by Paul F. Kerr 19 Structural Mineralogy of Clays, by George W. Brindley 33 FORMATION AND OCCURRENCE OF CLAY MINERALS By Paul F. Kerr * INTRODUCTION Clay minerals may be significant indicators of earth processes. They form through a range which involves at one extreme the action of compressed water vapor at a temperatnre of several hundred degrees centigrade, and at the other extreme, the action of atmospheric agencies at an ordinary temperature. Xot recognized as original ]ir()diicts of magmatic crystallization, clay minerals are j)rini(i facie evidence of a process s\d)sei|Ui'iil to the orig- inal crystallization. Yet their relation to the magmatic (iroeess may be close: the la.st of the fluids and vapors of the magma may react with wall rock to form clay- mineral masses. This is probably tlu' cxlri'iiic of high- tem)iei-atnre. hypogene clay-mineral formation. Between tills and the low temperatiire limit the transition is grad- ual, the minerals formed falling into gradational groups. In time clay aggregates of hydrothermal origin become distinguishable with difficulty from clay aggregates formeil untler atmospheric conditions — that is, those of supergene origin. HYPOGENE PROCESSES Clay minerals formed by hypogene processes result from the action of gases, vapors, or solutions that origi- nate below and force their way upward through rocks of the earth's crust. IMost of the elements of the clay min- erals are contributed by the invaded rocks; few, other than water, are derived from deeper sources. The chief materials removed from the erustal rocks are alumina, silica, alkali or alkaline earth elements, and iron; the.se are transformed into clay minerals at a temperature ranging from slightly below 100°C to about 450°C in an environment that may be acid, neutral, or alkaline, ilepcndiiig upon the pll of the invaded rocks and the acidity of the vapors from the magma. In the conduits of fumaroles, geysers, and volcanic vents the w-all rock may be altered to clay; also, com- pressed vapoi's or mixed liquids and vapors (it is not easy to be certain which) may alter rock-forming min- erals to clay in cavities in pegnuitite dikes or in igneous masses. The most extensive hypogene clay dejiosits, how- ever, have resulted from the action of thermal waters; although some of them are limited to the borders of metal-carrying veins, others are disseminated over a wide area. Temperature Data on temperature prevailing during clay-mineral formation are of three types: (1) measurements in drill holes penetrating hydrothermally active areas; (2) meas- urements of surface emanations; and (3) temperature recorded diiring experiments in clay-mineral synthesis. These data give little information on the temperature of formation of clay minerals in a natural environment, but they are all that are available. • Professor of Mineralogy, Columbia University, New York City, New York. The writer wishes to acknowledge the assistance of Mrs. Roselyn J. Steinhart in assembling the bibliographic data. Temperature in Drill Holes. Fenner (1934) has given measurements of the temperature in holes drilled in liydi-othermally active geyser-basins of Yellowstone Park. A temperature of 180°C was noted in the bottom of a hole in Norris basin drilled to a depth of 265 feet; an- other hole, drilled to a depth of 40G feet in the Upper Geys(>i' Basin, measured 205°C at the bottom. In Italy during the coarse of power develoi)inent at Larderello and adjacent regions in Tuscany (Ippolito, 1947) holes were drilled that yielded emanations and steam adef(uate for a major industrial installation. Keller and Valduga (194(i) reported a temperature of 205°C and a pressni-e of 63.5 jionnds per square inch in a repre- sentative Larderello well drilled to a depth of 87(5 feet. Brannock et al. (1948) reported that at Steamboat Springs, Nevada, the temperature in the bottom of one well, 156 feet deep, was 138°C. MacDonald (1944) has reported that gas collected from drill holes adjacent to a •solfatara at the Kilauea volcano, Hawaii, had the tem- perature of 95.5°C. The temperatures recorded in these hydi-othermal areas are well above the increase attribu- table to the thermal gradient, and are generall.y at- tributed to magmatic activity. Temperature of Svrfacc Emanations. A number of measurements of the temperature of hot springs or fuma- roles have been recorded. One of the most elaboi-ate programs of measurement was carried on by Zies and associates (1924) in the Valley of Ten Thousand Smokes, Alaska, following the Katmai eruption in 1912. There the temperature of gases escaping from fumaroles ranged from less than 1()()°C to 650°C. At Paricutin. IMexico, Zies (1946) observed temperatures up to 640°C in fuma- roles of the older Zapicho flow. The estimated tempera- ture of the lava at its source was given by Zies as 1200° C, in the same range as observed for other basaltic lavas. Bartli (1950) has reported temperature measurements for the springs, geysers, and fumaroles of Iceland. There superheated steam is a great rarity. Although it occurs in hot lava flows and around active craters, it has been reported only twice from hot-spring areas. Tem])erature at these springs was measured at 118°C and 120°C, but temperature in all other steam vents was very close to the boiling point of water; slight superheating probably oc- curred, but was not conspicuous. Byers and Brannock (1949) observed variations in temperature of the emanafifuis from Cone A in Okmok Caldera on northeastern Unmak Island. Temperature dropped from 320°C on July 19, 1946, to 90°C on Sep- tember 5, 1946. Temperature of Synthesis. Ivaolinite has been pro- duced synthetically at different temperatures and under a variety of conditions. It was formed by Noll (1934) at 320° C in tlie presence of 0.5N HCl. Noll expressed the opinion that kaolinite would form below 200°C in the proper acid concentration, if given sufficient time. Early experiments by Collins (1887) were said to have pro- duced kaolinite bv alteration of orthoclase in dilute IIF in 96 hours at "l6°C. Sehwarz and Walcker (1925) (19) 20 Clays and Clay Technology [Bull. 169 stated that kaoliuite must form between pll 4.5 and 5.2, the best range being 4.8 to 5.0. In later experiments, Sehwarz (1933) prepared kaolinite through the altera- tion of feldspar by treating with 0.5N to l.ON IICl or IL.SOj at 300°C for 250 hours. According to Noll (1936), kaolinite forms below 400° C in a neutral solution free of alkali metals or in an acidie solution containing ions of alkali metals. The mineral was produced under acidic, neutral, or slightly alkaline conditions at 300°C and at a pressure of 87 atmospheres. Folk (1947, p. 393), who has studied the alteration of feldspars and their products in the laboratory to gain information on the origin of kaolinite, writes ". . . . kaolin forms in acid solutions up to about 350°C if Al is rather high and K is low; muscovite forms from as low a temperature as 200°C, through 525°C in slightly basic and rather acid solutions if K and Al are high ; and pyrophyllite forms from about 300°C up to 550°C if Al and K are both low. It appears thus that the Al : Si ratio and the K concentrations are very important in deter- mining which mineral will form, since, if the tempera- ture and acidity are proper, that mineral will be stable whose formida most closely resembles the composition of the solution around it." Gruner (1944) reported that in about 30 experiments, largely with feldspars, using HCl solutions at 300°C to 400''C and in one case at 525°C, muscovite formed above 350°C and kaolinite below; and that pyrophyllite commonly formed throughout the range of experiments. The experiments by Ewell and lusley (1935) in which kaolinite was formed from AUOs-SiOo gels in a bomb at 310° C are in accord with other recent syntheses. Experiments in synthesis, therefore, indicate that kao- linite may form inider pressure at about 300°C. Under acid conditions and given an alnindauce of time — both of which nature provides — kaolinite may form at normal surface temperature and atmospheric pressure. The de- hydration curve for kaolinite indicates that a tempera- ture of foi-mation greatly in excess of 400° C is unlikely. Thus, kaolinite probably does not form at the extreme upper limit of temperature prevailing in fumaroles. Montmorillonite was synthesized by Noll (1935) at 300°C and 87 atmospheres; he used ratios of alkali : alu- mina : silica of 0.02 : 1 : 4 to 1 : 1 : 4. The alkalis used were NaOII, KOH, and Ca(0H)2. under conditions slightly more alkaline than required for kaolinite. Noll concluded that montmorillonite forms when solutions are alkaline and alkali-earth metals are present. If an excess of Mg(()II)2 is present, the magnesium enters into the composition of the montmorillonite up to 15.3 percent MgO ; this is thought by Noll to correspond to fuller's earth. Beidellite has been formed by Norton (1939) at 275- 325°C by the action of C02-eharged water on albite ; in 1937 he developed sericite by the same method, using orthoclase, anorthite. and albite. Finely ground ortho- clase was altered with IIoO and CO. in 33 days at 280°C and 1800 psi pressure, and in 150 days at 320°C and 2950 psi pressTire. Noll (1932) found that sericite forms in greatest amount at 250°C, the lowest temperature of formation being 225°C. Ewell and Insley (1935) synthesized diekite from hj'drous alumina-silica gels at 350° and 365°C. Differ- ential thermal curves indicate that diekite is stable at higher temperatures than kaolinite. Thugutt (1894) reported that hydrated nepheline gave a product similar to nacrite. Nontronite was produced in a bomb at 350°C by Ewell and Insley (1935) from coprecipitated gels using FeoOs. 2Si02. Halloysite, according to Noll (1936). is unstable above 50°C ; however, comjilete dehydration is obtained only by heating to about 400" C. Partial dehydration occurs readily at temperatui'es as low as about 75°C, or in a dry atmosphere, or under pres.sure. The intermediate states of dehydration appear to be quite stable because many specimens occur in nature in this form. Adularia is not a clay mineral, but it may be found under similar conditions and ma.v be an anhydrous associate of clay minerals. More.v and Ingerson (1937) indicate that the temperature needed for the synthesis of adularia is in the range 245-300°C. Gruner (1936) heated montmorillonite in gold-lined pressure bombs in aqueous solutions of KHCO3 (10 percent). Seven days at 300°C prodiiced very good orthoclase that gave an X-ray powder photograph identical with that of adularia. At 272°C the feldspar pattern became distinct after 10 days. At 245 C the stronger lines of orthoclase appeared after 42 days of heating. No changes were noticeable after 4 weeks at 200° C. Alunite is another mineral frequently associated with cla.v. Leonard (1927) formed 60 to 90 percent alunite m 7 days in sealed p.vrex tubes with O.IM solutions of HnS04, K2SO4, and Na2S04 with ammonium sulphate at a temperature of 200° C. Alunite (50 to 99 percent) was formed at 100°C in 0.05M solutions with 50g of alumi- num sulphate by heating for 100 days. With similar solutions, 40 to 60 percent alunite resulted in 60 daj's at 22 °C. Zeolites are reported by Noll (1936) to form under more alkaline conditions than either montmorillonite or sericite. In the same type of experiment which produced montmorillonite, analcime was formed with an excess of XaOII, Ca(0H)2, and Mg(0H)2. Gaseous Emanations The vajiors from fumaroles and fissure emanations or hot springs are composed essentially of steam that con- tains some acidic materials, Zies (1929) has given the percentages of acid constituents for the vapors of the Valley of Ten Thousand Smokes as follows: HCl— 0.117 percent ; IIoS— 0.029 percent ; HF— 0.032 percent. B.vers and Braunock (1949) report the constituents of the fumarole gases on northeastern Umnak Island as water vaj)or. carbon dioxide, and sulphur dioxide, and the water from thermal springs to contain as high as 159 ppm of boron and a few ppm of arsenic and antimony. Residual gases in freshly erupted volcanic rock were studied at the time of the eruption of Mt. Pelee by Shepherd and ^Merwin (1927). Samples were heated in vacuo and the gases identified were thought to be char- acteristic of the eruption. The chief volatiles found were : 11/1—80 percent; CO2— 9 percent; and CI. S, and F in smaller amounts. Part I ] GkOLOGY and illXERALOGY 21 Thermal Waters Character of HydroIlK nnul St/liitifni.<;. The character of the solutions responsible for the alteration of a rock to clay depends upon the original source material and the effects of wall-rock contamination. Both are impor- tant. Though fluids from tlic niagmatie source may be acidic, containinjr the vajxirs (if TICl, IIF. and IL-S, reactions witli irroiind-water and wall rucks often produce a neu- tral or alkaline environment for the formation of clay minerals. C)n the other hand, there is good evidence that clay minerals may be formed under acid conditions. Sijiprewald (1932), in studying the alteration of dia- base in igneous intrusions at Ijoveland ^Mountain, Park County, Colorado, concluded that the agents causing the alteration were contained in tlie magma itself and fol- lowed the same paths as the intruded porphyry. Altera- tion was an end of the intrusion cycle of each porphyry. Mineralizers within the sills were augmented by mineral- izers from more deep-seated magma. Acceleration of alteration as a i-esult of acid condition.? has been noted in volcanic regions. Payne and Man (1946) call attention to two types of chemical decom- position in basalt around Kilauea crater. xVt steam vents where no sulphur dioxide is present, silica and soluble bases are leached away, leaving the hydrated oxides of aluminum and iron. At the solfatara, where sulphur di- oxide is present, alteration is accelerated, and alkali and alkaline-earth elements are removed, leaving a yellow siliceous residue. ^MacDoiiald (1944) has reported the develojmient of opal and a lesser amount of kaoliuite as the result of the action of thermal solutions weak in carbonic, sulphurous, and sulphuric acids. Many cavities were lined with sul- phur crystals, and the other minerals deposited include alum, gypsum, mirabilite, kieserite, epsomite. and aph- thitalite. Allen (1935), in studying the geyser basins and ig- neous emanations of Yellowstone Park, foimd it neces- sary to account for bicarbonates of the alkalis in the thermal water. He concluded that there must have been an attack by COo ui)on the wall rocks at deiith. F'enner (1936) found two different ])rocesses causing alteration to clay at Yellowstone Park. Near the surface, by the attack of acid sul]ihate on the felds])ars, kaolini- zation results. At depth, where an attack of CO2 was probably involved, the alteration produced beidellite. An interesting feature described by Fenner is the close as- sociation of pyrite with the beidellite. Casts of feld.spars i-eplaced by clay contain bright litlle cubes of pyrite, which were probably introduced by alkaline solutions re- sponsible for the clay. xVcids of sulphur on the outlet side of an alteration zone, as pointed out by Lovering (1949, p. 53), -would yield a quartz-alunite rock. At a greater distance from the source of the solutions, or nearer the surface, alunite woidd be expected to give way to diaspore. This se- quence was described long ago by Whitman Cross (1896) in the alteration of rhyolite on ilonnt Robinson, Colo- rado. In discussing the East Tintic district. Lovering (1949, p. 491 suggested that alunite forms in a less acid en- vironment than that in which the kaolin minerals form. He postulated precijntation from cooling, nearly sjicnt acid solutions, with the i)ll increased to a little less than 7 by contact with bases in the rocks traversed. Sulphate-bearing solutions, acting on kaolinitic clay, are believed by Koss and Kerr (1934) to form halloysite. Ilalloysite is often formed within or near mas.ses of car- bonate rocks. The mineral is also characteristic of vein de]i()sits formed under acid conditions (see Tjovering, 1949). Ilalloysite deposits at Liege, Tintic, and Park City may be the result of precipitation brought about by acid solutions acting on a carbonate rock. At low levels, in siliceous or argillaceous-siliceous wall rocks, kaoliuite results from the action of acid solutions; but in the car- bonate rocks higher up, halloysite forms. In solfataric action described by MacDonald (1944), in which kaoliuite was formed, the vapors responsible for the alteration contained sulphur dioxide, sulphur vapor, and a trace of hydrpchloric acid. Xutting (1945) reported that soil minerals, particu- larly those of the montmorillonite group, are known to dis.solve or disperse readily in water solutions containing n.ni to 0.04 percent acid. " Hydrothermal Deposits Clay Minerals in Active Thermal Regions. Knowl- edge of clay minerals in active thermal localities is frag- mentary. The solfataric action reported by ]\IacDonald (1944) has resulted in the formation of kaoliuite at the expense of olivine basalt lava and ash. As early as 1851, Bunseu (1851) observed that the transformation of a basalt to clay was no simple soaking process, but that certain constituents might be added, others siibtracted. The alteration usually .starts in the glassy or microcrys- talline groinidmass. leaving the phenocrysts intact. The rock eventually becomes soft. Earth (1950, p. 49) found the rocks in Iceland to be "profoundly altered by hot- spring action; most intense and rapid is the alteration caused by acid springs. The end product consists of opal with small amounts of clay minerals. The mechanical action of boiling and gas exhalation breaks up the rock into a finely divided claylike substance. In areas of in- tense acid-sjn-ing activity such slipjicry warm clay de- posits cover large areas. . . . The basaltic lava is attacked by acid steam and vapors emanating from cracks and fissures in the lava. Around such cracks incrustations usually occur, and the pores and vesicles of the rock are filled with secondary minerals. Simultaneously the pri- mary minerals are altered by metasomatic action, in- ducing changes in the bulk chemistry of the rock." The dome around Geysir, Iceland, he described (Earth 1950, p. 97) as having "been built up by precipitation of silica from the hot-spring water as it flowed out of the basin. It is stratified (consisting chiefly of alternating layers of clay and silica sinter.) . . . The clay (layers) are identical (in composition) with the decomposition product formed by the action of acid hot-spring water on basaltic rocks at other places in Iceland, and it ap- pears that only acid waters can produce this kind of clay. On the other hand, only alkaline waters deposit silica sinter. . . . The hot spring activity in this place has changed many times from alkaline to acid and vice versa." Rock alterations in the Kilauea region of Hawaii have been described bj- MacDonald (1944) and by Payne and Clays and Clay Technology [Bull. 169 Man (1946). The resultant rock is composed largely of opal, with smaller amounts of kaolinite or related clay minerals, and relict magnetite and ilmenite. It has ap- proximately the same volume as the unaltered rock, and original structures and textures are surprisingly well preserved. Anderson (1935) lias described and given many im- portant details of the alteration of the lavas surrounding the hot springs in Lassen Volcanic National Park. Opal, accompanied by minor amounts of kaolin and alunite, is the chief product in the altered exteriors of all lava fragments collected where active decomposition is taking place. The muds from the hot-spring basins are similar in composition except that they usually contain more kaolin ; the mud pots and the sediments from Boiling Lake consist largely of kaolin. Day and Allen (1925) suggested that two types of lava decomposition are in progress: "... The one producing kaolin and some silica without aluminum sidphate. the other producing silica with aluminum sulphate. In the fact that kaolin is de- composed by strong sulphuric acid into silica and alu- niiimm sul)ihate, the key to the ditference is doubtless to be found. If the acid forms in a place where sufficient water is percolating, its concentration is kept down to such a value that the decomposition of feldspars, vol- canic glass, and possibly other minerals is incomplete. The intermediate and comparatively stable compound kaolin results, and this generally occurs in the springs, together with very dilute acid. ' ' On the other hand, if sulphuric acid forms in nearly dry ground, it will accumulate by progressive oxidation of the sulphur gases and the concentration may reach comparatively high values — probabl.v in tlie form of sirupy films." Audei'son (1935) believes there is considerable evi- dence in support of this conclusion, namely the high alumina content of the sediments from Boiling Lake wliere one might expect the greatest dilution of sul- phuric acid. Moreover, opal is being formed from lavas that are being attacked by steam where the acid may be more concentrated. Claii 3rinerals as Cavifij Fillings in Pef/ntafifcs. Pre- siimably cavities in pegmatite dikes, signifying gaseous conditions, may provide channels along which thermal fluids permeate and alter the feldspars. They represent a late, probably hydrothermal, stage in pegmatitie activ- ity. The alteration could be largely either the result of compressed vapors or thermal solutions. Silva and Neiva (1948) attribute niontmorillonite formed in granite peg- matites to hydrothermal solutions. Clay Minerals in BletaUiferons Deposits. In recent ■ years the importance of clay minerals as indicators of mode of origin of metal-bearing deposits has come to be generally recognized. Many of the metal-bearing deposits associated with late Mesozoic or Tertiary volcanism, par- ticularly in the Cordilleran region of the continent, have associated with them prominent zones containing clay minerals. Deposits of tungsten, molybdenum, zinc, lead, copper, and uranium have been observed in a number of instances to occur in sites where the wall rocks have undergone partial or complete replacement by clay min- erals. In particular, montmorillonite, kaolinite, halloy- site, hydromica (illite), and dickite are coming to be recognized as significant by-products of the ore-forming process. The initial stages of hydrothermal alteration may not involve clay minerals, as in the stages described by Lovering et al. (1949) at Tintic, Utah, where the ferro- niagnesian minerals of igneous rocks are converted to chlorite, and limestone is dolomitized. However, Sales and Meyer (1949) consider the montmorillonite at Butte one of the earliest indicators of hydrothermal alteration. Kerr (1951) reports the formation of chlorite at Silver Bell prior to the clay-mineral stage, and also at the cop- per deposit at Santa Rita (Kerr et al., 1950). The initial stage represents the gi'eatest penetration of the solutions and vapors from the magma into the fresh wall rock. It is here also that the alkali and alka- line-earth ingredients of the wall rock react most strongly with acidic fluids from the magma to produce alkaline or approximately neutral conditions. Sales and ileyer (1949) have given a summary of numerous analyses of the rocks of this stage, indicating particularly the re- lease of alkaline earth elements from Cjuartz monzonite bordering the Butte veins. The outer reactive zone in hydrothermal alteration encloses a zone of somewhat lower pll in which hydromica (illite) rather than mont- morillonite becomes the dominant clay mineral. On the inner boundary of the hydromica zone, kaolinite be- comes the dominant clay mineral. Located concentri- cally Avithin the kaolinite zone is an inner zone in which sericite predominates. The chemical and physical conditions giving rise to the formation of the concentric zones of alteration are the same eoiulitions involved in the processes of metallic mineralization. The evidence of the mineral constituents observed indicates an increase in temjierature toward the center; pressure conditions which are more or less uni- form and characteristic of open fractures from a few hundred to a few thousand feet below the surface ; and variations in pH from a slightly alkaline periphery, to neutral, to slightly acid, to slightly alkaline, from the edge to the center. Montmorillonite is foinid in the border zone along the margins of hydrothermal veins, cutting intrusive igneous masses in the Cordillera of the United States. Sales and Meyer (1949) report that at Butte it is the first clay mineral to develop in the hydrothei-mal alteration of the quartz monzonite. Field studies in the ilarysvale district. Utah, have demonstrated that montmorillonite and hydi-omiea (il- lite) may be formed by the hydrothermal alteration of glass.v volcanic dikes. In the vicinity of Lovelock, Ne- vada, solid masses of volcanic glass have been found al- tered to montmorillonite. apparently by hydi'othermal solutions. Volcanic ash is believed to be even more sus- ceptible to alteration in areas of hydrothermal activity. Among the clay minerals, dickite seems likely to form under conditions of highest temperature. The breaks in the static dehydration curve are slightly higher, and field relationships suggest an extension to deeper-seated conditions of occurrence. Exceptions exist, and although is has been suggested that dickite may be formed under supergene conditions, most, if not all of it is hydrother- mal ; it is ordinaril.y the high-temperature kaolin mineral. art Geology and ^Iinlkalogy 23 The orifrinal dickito of Aiifrlesey, diekito from tliree localities near Coluiiibia, ^Missouri, diekite in the lead ores of southeastern ^Missouri ( Tarr and Keller. 1936), the diekite at San Jnanito. [Mexico, dickite formed in the alteration of the Gilman porphyry at Gilman, Colorado, and diekite reported (Fraakcl, 1949) from the fjold- bearinpr rocks of Witwatersrand are all hydrothermal in orifrin. Dickite in the veins of Cerro do Pasco coats coarse crystals of enargite; at Ouray. Colorado, dickite coats sulphide minerals. Sales and ileyer (1!I49) found dickite in the liiulier temperature or central zone in the deejx'st levels of the Butte, ilontana. mines. In the Dag- gafontein mine on the Witwatersrand (Frankel, 1949) the diekite was associated with a greenish-gray clay, a <-hloritic mineral, pyrrhotitc, gersdorflfite. and a lustrous hydrocarbon, identified by Davidson and Bowie (1951) as thucholite. probably also of hydrothermal origin. Where dickite and kaolinite are found together in hydrothermal veins, evidence points to dickite as the earlier of the two. It is a less abundant elaj' mineral, and only in a few places such as San Juanito and Ouray have masses of any considerable size been found. lioness and Williams (103.")) have reported dickite from the Pine Knot Colliery in Schuylkill County, Penn- sylvania. The material is well crystallized and occurs, for the most part, as a white, glistening powder, or in snuill lumps composed of transparent tabular monoclinic crystals, many of exceptional beauty, ranging in size from 0.1 to 0.5 mm along the ma.ior axis. Of the asso- ciated species, clear transparent quartz is most common, altlumgh a small amount of pyrite and magnetic mate- rial is ju-esent. The (piartz crystals are everywhere com- pletely coated with dickite that imjiregnates the crystal sui'faces in a manner suggestive of replacement. The association of dickite with sulphides, such as in the southeastern ^Mi.ssouri lead ores, may indicate neutral or slightly alkaline conditions of precipitation. At Butte. sericite is an associated mineral ; at San -Tuanilo. pyrite may have been precii)itated with dickite. In tlie Tintic district. Utah. Lovering (1949 I found an abundance of supergene kaolin in the oxidized zone of ore shoots. Argillie alteration is found below the lava near channels followed by later mineralizing solutions. In origin, nacrite is probably more closely associated with dickite than with kaolinite. At St. George. Utah, and at San Jnanito, ^lexico, nacrite and dickite are closely associated. Both occurrences are believed to be hydrothermal. as is the well-known occurrence at Brand, Saxony. The sequence of clay minerals appears most signifi- cant in hydrothermal veins. Individual nunerals may occur under a variety of conditions, but the progressive alteration from fresh rock on the wall thi-ough a se- quence such as montmorillonite. illite. kaolinite, sericite. and dickite is probalily characteristic of hydrothermal veins. SUPERGENE PROCESSES The processes by which clays are formed on or near the surface have received much less experimental atten- tion than have the processes of hj-pogene clay formation. It seems clear, however, that atmospheric conditions are resjionsible for the formation of clay minerals under a variety of conditions. Clay-mineral development under normal surface con- ditions may require periods that are long even in terms of geologic time unless some accelerating conditions jirevail. I'nder even slightly favorable conditions, how- ever, leaching and deposition, together with weathering and soil formation, do take place. The chemistry of supergene processes is complex. It is generally thought, however, that even weak concentrations of acid or alkali solutions together with small amounts of alkali or alka- line-earth elements, alumina, silica, and other chemical constituents, given suf'licient time, may produce clay dejxjsits, some of which are very large. Special condi- tions, such as a more concentrated chemical action or moderate increase in temperature, nuiy greatly decrease the time factor. In saline or brackish bodies of water, clay nunerals are thought to be often subject to diagenesis. In regions of heavy rainfall and high temperature, organic acids are apt to accumulate in unusual concentration, accel- erating the formation of clay. Even organic concentra- tions in bogs where the tempei-ature is not extreme may form clay minerals. The formation of clay minerals originally, by either hypogene or supergene action, followed by removal, transportation, and redeposition at a new site, is simple in general concept but involves many features that should receive further scrutiny — particularly the ex- tent to which the alteration or direct precipitation of clay minerals may take ]dai-e in the cycle. Normal Surface Action Normal surface action in clay formation involves leaching and deposition, and weathering emd soil for- mation. Leaching and deposition may take place as a result of special chemical conditions existing in frac- tures in rock masses considerably below the zone of weathering and soil formation. The distinguishing of clay minerals formed by leaching and deposition from those formed by low-temjierature hydrothermal processes may not always be possible, but the attempt is worth- while. In clay acciimulations formed by mechanical transport, the presence or absence of chemical precipi- tates and the origin of the introduced clay minerals prior to transport are the most interesting features. Leaching and Deposition. In the formation of ben- tonite, for example, supergene leaching and deposition are probably involved. As defined by Eoss and Shannon (1926) bentonite is formed by the chemical alteration of glassy igneous material usually tuff or volcanic ash. One large grou]) of bentonite deposits found in the lower ilississippi A'alley and Gulf Coast region (Ilagner, 1939) has probably been formed under supergene con- ditions. Even in many volcanic areas of the western United States, the actual formation of clay may have resulted from supergene action. In recent years, as a result of the search for bentonite to be used as an ad- sorjitive clay, thousands of other bentonite occurrences have been discovered. Nutting (1945) has pointed out that the complete replacement of shells by montmorillonite at Pontotoc, Mississippi, indicates that alumino-silieate sols may be deposited under conditions that allow removal of the original bases present in the shell-forming material. 24 Clays and Clay Technology [Bull. 169 Allen (1944) emphasizes the importance of sedimen- tary processes in elay-mineral formation in certain Pa- cific Coast areas where volcanic materials predominate. Sedimentary processes are more important than vol- canic processes at lone, California ; Castle Rock. "Wash- ington; Whiteware, Montana; Hobart Bntte and Molalla, Oregon. Clays derived directly from volcanic materials are composed dominantly of montmorillonite ; but these clays, with the exception of the beidellite-nontronite varieties, have 'relatively low percentages of available alumina. In contrast, clays derived through leaching of various aluminous rocks in thoroughly drained areas are composed dominantly of kaolinite ; furthermore, the dep- ositional structure of these sedimentary clays favors the formation of gibbsite by weathering or the formation of diekite or kaolinite by hydrothermal action. Large accumulations of fuller's earth that contain attapulgite as the effective clay mineral have been worked for many years in the Georgia-Florida region (Kerr, 1937). The sediments which contain the clays have been derived from the erosion of the crystalline rocks of the highlands. On the other hand, no such accu- mulation of attapulgite as is found in the sedimentary layers has been reported in the source area. It seems likely that the accumulation of attapTilgite may repre- sent more than simple transport and deposition ; per- haps waters unusually high in magnesium, or long-con- tinued action of waters low in magnesium, contributed to the formation of this unusually high-magnesia clay. The unique clay mineral hectorite described by Foshag and Woodford (19.36) is found in the IMojave Desert, California, in folded sediments of Tertiary age. Relic textures are abundant in the clay matrix ; they include minute collapsed cavities lined with layers of the clay mineral, and curved structures which may be altered pumice shards. The rock is probably a greatly altered dacite tuff. The clay mineral of which the rock is now chiefly composed, is chemically and optically almost identical with the magnesian clay mineral, but differs from it by swelling greatly in water. In origin, there- fore, this bentonitic magnesian clay is not essentially different from most normal bentonites. Chemical anal- yses of this clay show that it is essentially a magnesium silicate with only an insignificant amount of alumina. Noteworthy are the high soda content (3 percent) and the litliia content (1.12 percent). According to Tarr and Keller (1937), kaolinite found in several localities in Missouri gives evidence of having been deposited from solution. Kaolinite occurs in the Orongo Circle mine as thin veins in nearly vertical joints of the Pennsylvanian shales, as well as in cavities and as a replacement mineral. The kaolinite at the National Pigment barite mine was deposited in joints in dolomite tliat had been enlarged by solution. At the Reavis mine, kaolinite occurs dominantly in solution cavities and along joints in the Jefferson City dolomite. This mine and the National Pigment mine are in sink holes. Kaolinite in the Keokuk, Iowa, area occurs in quartz geodes and in small solution cavities in siltstone. Allen (1937) has described kaolinite derived from limestone by supergene jjrocesses. The Cheltenham clay of Missouri had its source in an earthy limestone subject to the action of carbonic-acid waters. Erosion into a choked sink hole or closed basin has resulted in the ac- cumulation of a sedimentary clay deposit. ]\IacKenzie (1949) has described a clay mineral of possible supergene origin, which differs somewhat from illite but belongs to the same group. The material occurs at Ballater, Aberdeenshire, in a crush-band in the al- tered rock along a vein in a granite outcrop exposed in road-widening. The alteration is apparently a result of the action of water, but since the band was still highly altered at the base of the exposure (about 25 feet from the surface), it is not certain from the field relationships whether the water was of hydrothermal origin or whether it percolated from the surface. As pointed out by Ross and Kerr (1934) halloysite forms many times by supergene processes. Callaghan (1948) has describecl halloysite 4H2O from Bedford. Indiana, which represents a supergene reorganization of deposited material. Other halloysite deposits were ap- parently at one time a residual soil. Halloysite from Wagon Wheel Gap, Colorado (Cross, 1896), formed as an alteration product of rhyolite, and may have been deposited from solution or suspension by downward- moving waters. Lovering (1949) has described kaolinite, halloysite, and allophane in the East Tintic district. Eureka, Utah, which formed when weathering of sulphide ores pro- duced acid solutions. Away from the sulphide bodies, montmorillonite and beidellite' are the common products of weathering. Bates et al. (1950) have made a number of interesting observations on the morphology of halloysite. Halloysite 4H2O (endellite) is believed to consist of well-developed tubes which split upon dehydration to halloysite 2HoO. Study with an electron microscope has revealed that halloysite 2H2O cr.ystals consist of hollow tubes which have collapsed, or have split and partially or completely unrolled. It is suggested that this change in morphology explains the irreversibility of the dehydration process. There is no apparent morphological transition from these crystals to the pseudo-hexagonal plates character- istic of kaolinite, diekite, and nacrite. Stringham and Taylor (1950) ascribe the formation of nontronite occurring in a contact metamorphic zone to the action of slightly acid solutions. This clay is prob- ably a weathering product of diopside, tremolite, and pyrophyllite. Nontronite (Main. 1950) occui-s within fis- sures and fills cracks and spaces between polygonal joints in basalt in the vicinity of Garfield and Manito, Wash- ington. The mineral occurs chiefly as a filling, although in some cases the basalt is altered to nontronite. The nontronite contains a small amount of sericite, pyrite, and quartz. Presumably neutral or alkaline solutions at normal temperature and supergene in origin, acted on the basalt and deposited nontronite in openings. Weathering and Soil Formation. Clay minerals are produced in various areas on the earth's surface where the normal temperature is low; at such a temperature, however, special physical or chemical conditions are essential, and small differences may be significant. The clay minerals form slowly by rock decomposition. Varve clays deposited during Pleistocene glaeiation along the Hudson Valley are essentially rock flour with a minor clav-mineral content; working near Newburgh, New Part 11 Geology and Mixeralogy 25 York, Alk'ii (1947) I'uuiid a relativelj- small clay-miiu'ral content in the r to Sand and P>ates ('in.")2'i halloysite is formed oidy from the weatlierinjr of feldspars and. imder conditions of intense leachinor, is formed from both plajri- oclase and potash feldspars. Primary mica, however, always alters to vermicular kaolinitc. Potash feldspars, where leacliinntly without passing through an intermediate stage. Rocks can be decomposed by action of normal sx^rfaee waters when sufficient time exists for the completion of the process. Water in soil may occur in cavities and cai)illaries as adhering water, hygroscopically combined water, chemically eonibined water, and even water vapor. The total water content of a soil changes con- stantly thronjrh additions and subtractions. fMohr 1944, 1>. 41") has pointed out in his discussion of tropical soils that the pH may vary from a figure lower than 3 to higher than 9. Even rjuartz (Mohr, 1944, p. 75), which is resistant at less th;iu pTT 7. dissolves in time at a pTI greater than 7.5. Tu tropical weathei'ing, acid water may be responsible for the formation of extensive kaolinitc deposits. Mohr concluded that kaolinitc would form in tropical soils as a result of weathering of feldspar, under the influence of ]iure water or water coutainiuqr carbonic, sulphuric, or humic acid. Tender the influence of alkali or OH" ions, montmorillonite is more likely to form. In the .same cli- mate, if the rocks undergoing weathering are highly calcic, montmorillonite develops: alkali-rich rocks yield kaolinitc. Wherever weathering takes place the possibil- ity of an accelerating chemical action always exists. Correns and Engelliardt (1938) have investigated the mechanism of weathering of potash feldspar and con- clude that potash, alumina, and silica are all at first removed in ionic form with the relative proportions of the three constituents dependent to some extent on the aciditv of the extracting medium. In time, a goodly UTim- ber of rock-forming silicates not particularly soluble may, where present in soil and sub.iected to solutions of suit- able pll. decompose to form clay. An alkaline environment which gives a sufficient sup- plv of magnesium or ferric or ferrous iron in addition to silica and alumina, is apt to lead to the development of montmorillonite. In wet and warm climates, where weatherinp- and oxidizing conditions are comparatively rapid, kaolinitc and lateritic soil are more likely to form than mnntmorillonite. Frederickson (1951) proposes, largely on theoretical grounds, that the mechanism of weathering is essentially a base-exchange process between two similar crystal structures with ditTerent degrees of order. Tlie mecha- nism is visualized as a process whereby the hydrogen ions of crystalline water are base-exchanged for the sodium ions of albite (which is used as a model). The small size of the hydrogen ion and the fact that its introduction into a crystal system is an exothennic reaction makes the l)rocess possible. The niechanieal effect of this base- exchange reaction is a net expansion in the reacting layer of the crystal which causes the rock to exfoliate and be- come more chemically reactive. The minerals will break down into colloids or small clumps of insoluble silica de- pending on the Al : Si ratio in the crystal lattice. Wiklander (1950") has studied the fixation of potas- sium by clays saturated with ditlcrcnt cations. A mi- caceous clav (Clarence soil. Illinois) was treated for 21 days with HCl, NaCl, KCl, NH4CI, and CaClo in order to obtain maximum replacement of K* by the other ions. The subsequent fixation of K by the K*-depleted clays was then studied by the use of radioactive K* ; fixation was found to be low for H+-, K"-, and NH^^-saturated clays but high for Ca**-, and more especially, Na*- saturated chivs. Similar experiments on ground biotite .showed low fixation values for H*- and Xlli^-saturated biotite, but high values for Mg+'-- and Ba**-saturated biotite. Marshall (1948) has reported on ionization of calcium from soil colloids. Dilute suspensions of beidellite, mont- morillonite, and kaolinitc have been titrated with NaOH and with Ca(0H)2 and the activities of H% Na*, and Ca+* in the solution (pll, pNa, and pCa) determined at frequent intervals during the titration. Kaolinite, which probably binds the bases on the surface of the particles only, gives highly ionized "salts," whereas much of the sodium or calcimn taken up by beidellite or montmorillonite is held in an nn-ionized state. This effect is more marked with calcium than with sodium. It was Nutting's (1945) conclusion that montmorillo- nite, given sufficient time, will go into solution in slightly acid water, but that only the bases will be removed in strong acids, and that only free silica will be removed in pure water. Kelley (1939) has pointed out that the nio.st conspic- uous effect produced on clays and argillaceous sediments by base exchange is reflected in their permeability. Gen- erally speaking, calcium-saturated clays tend to be granu- lar and comparatively pervious, whereas sodium clays are highly dispersed and relatively impervious. Kelley indicates that the pII of a clay may be in- fluenced significantly by base exchange. Calcium clays, if free from calcium carbonate, are approximately neu- tral, whereas sodium clays may be highly alkaline, a re- sult of the hydrolytic property of sodium clay by which sodium ions are replaced by hydrogen ions. Ordinary water always contains a low concentration of hydrogen ions and rain water is commonly acidic because of dis- solved carbon dioxide, which substantially increases its hydrogen ion concentration. The hydrogen ions, having strong replacing power, tend to replace sodiiim, with the consequent formation of dilute solutions of sodium bi- carbonate, sodium cai-bouate, or sodium hydroxide, re- sulting in increased alkalinity of the sodium clays. If sodium clav is mixed with calcium carbonate, it may be- 26 Clays axd Clay Technology fBiill. 169 come higlily alkaline upon leaeliing, owina: to the fact that calcium ions, furnished by calcium carbonate, will replace significant amounts of sodium from the clay with the consequent formation of corrps]iondino- amounts of sodium carbonate. According- to Kelley, magnesium, although jjresent in sea water in lower concentration than sodium, takes a more active part in the base exchange of sediments with which it comes in contact. Consequently, sediments in contact with sea water will inevitably contain practically as much replaceable magnesium as replaceable sodium. Tlie decomposition of biotite in soil has been discussed by Walker (1949) with special reference to soils in northeastei-n Scotland. The weathering of biotite causes an increase in the optic axial angle, a decrease in refrac- tive indices, a decrease in specific gravity, the disappear- ance of pleochroism, and a marked color change. This is accompanied by a loss of iron, preceded by oxidation of ferrous to ferric, loss of magnesium, potassium, and sodium, and a gain in water. The products of the weath- ering are hydrobiotite. chlorite, a vermiculite, and ulti- mately kaolinite. In biotite at Glen Buehat, weathered biotite flakes alter to a vermiculite, the alteration proceeding through the crystal layer by layer. The properties of the vermic- ulite differ somewhat from those of the hydrothermal vermiculite previously described. In a study of Australian soils derived from granitic and basaltic parent material. Hosking (1940) has sep- arated colloidal fractious from the subsoils and has identified the clay minerals present by X-ray diffraction. Kaoliiiitic clay was found in several granitic .soils of diiferent conditions of origin. In basaltic soils, however, the type of clay mineral present is a reflection of the soil moisture conditions : kaolin-group clays alone charac- terize the red loams, montmorillonite-group clays pre- vail, together with a little kaolinite, in the red-brown earth, and the montmorillouite group alone is typical of the black earths. Montmorillouite may form in areas of restricted drainage over basaltic rocks, where climatic conditions would lead to the formation of kaolinite if the area were readily drained. Graham (1941) looks upon an acid clay as an agent in chemical weatliering. He believes that hydrogen clay may become an active agent in mineral and rock weather- ing processes. His conclusion, however, is based upon limited data on the transport of calcium from anorthite, a not too common feldspar, to colloidal clay. Grim et al. (1939) conclude, on the basis of data re- lating to the climatic conditions under which various clay minerals in Illinois sediments are formed, that the joint presence of illite and kaolinite in a formation in- dicates a source area, some parts of which had a warm climate and other parts a cool climate. Studies of the clay minerals contained in the fine frac- tion of the insoluble residues from .35 Illinois limestone and dolomite samples representing the major strati- graphic iniits of the Illinois geologic column reveal the presence of illite in all samples, kaolinite in twenty sam- ples, and beidellite ( ? ) in three samples. Much of the illite may be authigenic and probably was derived from beidellite; the kaolinite is probably detrital. Kaolinite characterizes all the Pennsylvanian and Silurian samples studied. Some Mississippian and some Devonian samples contain kaolinite. others do not. The Ordovician samples contained only illite. Transportation and Aci-iii)iixist which suggest kaolinitic alteration, butli before and after final deposition. Diagenesis in Saline Waters It is believed by some that cla.v minerals may form under conditions of marine diagenesis. Tf this theory be correct, small flakes of mica, kaolinite, or other min- erals deposited on the ocean floor undergo progressive change to illite and montmorillouite. Cla.v deiiosited as kaoliuite in brackish water or near shoi'e ma.v disinte- grat(> if dejiosited at depth. It is even suggested that, in a general wa.v, the cla.v-mineral assemblage and tex- tures of crvstallization ma.v indicate the depth at which the sediments were deposited. According to Dietz (1941), montmorillouite may be the original cla.v mineral deposited in some places on the ocean floor, and subsequentl.v alter to illite. rxrim et al. (19491 in their discussin of recent sedi- ments of the Pacific Ocean near California and in the Gulf of California attribute broad-range depth char- acteristics to kaolinite, montniorillonite, and illite. Kaolinite is thought to be gradually eliminated b.v a diageuetic jn-ocess with increase in deptli. Tlie samples studied cover a depth range from near shore to 13,386 feet. While the process suggested is of considerable interest, too little data ai'e available for jjrecise appli- eation. Glass (]9.")1) has apjilied diagenesis in the interpre- tation of the eonditi(ms of origin of the cla.v minerals in the Cretaceous and Tertiary sediments of New Jersey. The minerals involved are kaolinite. moutmorillonite, li.\(lromuseovite, illite. a chloritic mineral, and glau- conite. Transitions ma.v be traced, using differential thermal curves from light brown to dark brown mont- nujrillonite, fi'oni light green dispersed glauconite, through light green glauconite grains, to dark green glauconite grains. Well-ei'vstallized kaolinite in the New .Terse.v sedi- ments is attributed to a fresh-water lake in Triassic shale hollows. The kaolinite in coastal sediments shows a decrease in ciwstallinity because the waters are not entirel.v fresh, and bracki.sh water deposits show a de- crease in crvstallinity of kaoliuite and hvdromnscovite, and an increase in the amount of chlorite formed largely at the expense of h.vdromuseovite. In marine sediments, illite is formed. When marine conditions are non- reducing, an increase in salinit.v is considered to bring about a decrease in the amount of kaolinite and chlorite formed, and an increase in the amount of illite formed. Alteration Aided by Organic Acids Bocj Deposits. Clay deposits have been reported a number of times from bog or swamp accumulations, either of present-da.y or more ancient origin. The clays formed in this way may owe their pnrity to the leaching action of waters containing carbonic and various humie or organic acids tliat are derived from the decay of or- ganic matter and organic processes, as a result of the leaching action of living plants. A residual kaolinite (lej)osit formed by the decomposi- tion in situ of granodiorite, probabl.y as a result of the corrosive action of organic acids acting in old peat bogs, has been described by Kerr (1930). The deposit was enconntei-ed in a subwav tunnel excavated in Brooklvu, New York. Lnterifc. Such a close relationship exists between the formation of cla.v nunerals and laterite or bauxite that discussion of the origin of cla.v minerals woidd be incom- plete without at least a brief consideration of these frequentl.v associated materials. The essential mineralogical constituents of laterite are such aluminian minerals as cliachite, gibbsite, boehmite, or diaspore, and .such ferrian minerals as limonite, goethite, or lepidocrocite. These hydrous ahuninian or ferrian oxides are ordinarily impure, and the extent of their development is an index of the amount of lateriti- zation involved. The tropical weathering of magnesian and ferrian rocks in certain areas suggests that perhajjs secondar.y serpentine or other hydrous magnesian min- erals may form in a similar manner. Man.v divergent solutions have been proposed for the problem of the nature and mode of origin of laterite (Robinson, 1949). ^lore than a hundred years ago, H. Bucliauan (1807) proposed the name laterite for a type of red soil frequently found in southern India. The term lias .since been applied to red soils both in tropical regions and elsewhere. Martin and Doyne (1927), on the basis of chemical anal.vsis. consider laterite a material in which the molec- ular ratio of silica to alumina is less than 1.33:1, while in lateritic soils the ratio is 1.35 to 2.0 : 1. This definition is inadequate as it neglects iron as a cla.v minei'al con- stituent and groups together soils of different profile characteristics and history. Pendleton (193fi) restricted the teriii laterite to specific soil la.vers rich in sesquioxides of iron and aluminum. Man.v such la.vers are characterized b.v the ])resence of pisolitic material and colloform crusts overl.ving mottled or vesicular zones. Campbell (1917) considers laterite of western Australia and Africa to have been developed b.v ground-water. Pendleton (1936) eonsidei"s laterites to result from ]iseudo-i]luvial deposits in the zone of a fluctuating water table under peneplanie conditions. Later uplift nuiy cause these ground-water conditions to disappear, inducing the development of a new profile. Profiles whTch may be explained in this wa.y occur in mauv parts of the Avorld. Harra.ssowitz (1926) lists four horizons for a typical lateritic profile: (1) the parent rock at the base; (2) the horizon repi-esenting the material of initial weathering, which appears to be kaolinitic; (3) the lateritic horizon; and (4) a surface horizon with ferruginous incrustations or pisolites. The physical characteristics are fairly well 28 Clays and Clay Technology IBuU. 1G9 recognized. The lateritie horizon is colorful, consisting of red material mottled witli yellow or violet, and is gen- erally argillaceous. AYlien derived from quartzose rocks, it ma.v be granular in texture, vesicular or cellular, with pores frequently filled witli white or grayish material. When moist, the laterite can be dug, but it hardens on drying and can be used as a building material. The sur- face crust may develop to such an extent that it prevents the growth of vegetation. Pendleton (1086) considers laterites in Siam to be a result of ground-water action. lie distinguislies laterite occurring in deep horizons, where parent rocks are saturated with water and iron compounds go into solu- tion, from laterite formed near the surface where the pores are tilled with air and the ferrous compounds be- come oxidized to ferric. The typical lateritie profile exhibits concretionary or crust-like material uiulerlain by mottled clay, but there are profiles in which, although concretionary material occurs, the mottled liorizons are absent. Hardy and Follett-Smith (1931) describe variations in the lateritie profile. The source material apparently influences the development of laterite. Hardy and Follett-Smith state that acidic rocks such as granite do not yield laterites. This may be because of the substitution of a kaolinitic weatliering product formed from the potash and potash- soda fehlspars occurring in such rocks. Various opinions have been given concerning the mechanism of laterite formation. Harrison (1934) con- siders that two main processes are involved: (1) primary lateritie formation; and (2) resilicification of the lateritie product by deposition of silica from ascending solutions. The latter process occurs in localities affectecl by a high water-table. Campbell (1917) considei's Australian and African laterites to liave been formed by changes which occur in the zones of permanent and intermittent saturation in the vicinity of the water-table. Oxygen required for the oxidation of ferrous to ferric iron implies a near surface formation. Hydrated sesquioxides are a result of hydrolysis of silicate minerals. Alkaline ground-water solutions con- vert the iron silicates to ferrous hydrogen carbonate and the alumina silicates to alkaline aluminates. During periods of drought, dissolved material from the water table migrates upward; the ferrous hydrogen carbonate is oxidized to amorphous hydrated ferric oxide, and the aluminates to amorphous hydrargillate AljO.-j ■ SH^O. Campbell (1917) considers further changes to involve the formation of something he called "turgite" 2Fe203 • 3HoO, and possibly hematite, Fe-jOs, from amorphous ferric oxide hydrate, and crystalline gibbsite AI2O3 • 3UjO from hydrargillite. The greater mobility of ferric oxide in colloidal solution causes lateritie horizons to become progressively more aluminous as laterization continues. Woolnough (1918), in his studies of western Australia, concurs with Campbell. He limits laterization to those situations in which free drainage does not exist, as in a peneplain near sea level. Laterites at higher elevations are products of processes which took place prior to uplift, and of a change in hydrological conditions. Whether alluvial deposition of sesquioxidic material may or may not have occurred in the presence of a high gi-ound-water table, it seems likely that the principal climatic circumstances are high temperature and the alternation of aerobic and anaerobic conditions in the soil profile. The mobility of sesquioxidic constituents in the soil profile has been attributed by Campbell (1917) to the alkalinity of the ground-water, by Harrassowitz (1926) to the protective effect of silicic acid soils. Each of these explanations may be applicable in the wide range of lateritie profiles which exists. Kaolinite is an intermediate product in the formation of Arkansas bauxite. Gordon, Tracey, and Ellis (1949) have recognized residual and colluvial bauxite deposits as the chief source for connnercial production. Kaolinitic cla,y separates the residual bauxite from fresh nepheline syenite, the initial source of the alumina. The colluvial deposits grade into a surrounding kaolinite envelope. Harder (1949) agrees with other students of the sub- ject that bauxite may originate directly by the alteration of primary aluminum-bearing rocks. He also points oiit that botli field and microscopic evidence show that kaolinite may be an intermediate product in the forma- tion of bauxite. He does not think that kaolinite found in association with bauxite represents resilication of the bauxite. Bauxite deposits (Harder, 1949) may consist mainly of the trihydrate, gibbsite, or the monohydrate, boehmite, mixed locally with diaspore. Pressure and possibly heat first change the gibbsite to boehmite, and eventually change the boehmite to diaspore. According to Harder (1949), the weathering of any one of many rock types or their weathered derivatives may produce bauxite or laterite, althoug'h more than normal alumina content and the pi'esence of soluble con- stituents accelerate the process. The rock type, compo- sition of ground-water, topographic relief, temperature, and rainfall are all significant factors in bauxite forma- tion or lateritization. A warm liumid climate with al- ternating wet and dry seasons hastens the process. A land surface of low relief, permitting rainfall to pene- trate the ground, is considered essential. The chemical composition of the ground-water and the solvents present are imjiortant. In many eases, the original constituents of the rock determine whether bauxite, iron ore, or man- ganese ore results. Rankama (1950) emphasized the role of carbonated waters in the formation of aluminum hydroxide from basic igneous rocks. He also attributes the formation of some calcareous bauxite to the action of carbon dioxide- bearing waters in a warm climate. Tlie argillaceous con- stituents of limestone are converted into bauxite and the calcium carbonate is removed as bicarbonate. Goldman (1949), in a detailed petrographic study of specimens from Arkansas, has described the alteration of feldspars to finely crystalline gibbsite, thence to seem- ingly amorphous pisolitic bauxite, followed by kaoliniza- tion of the aluminous hydrates by silica derived from the unaltered core. It is not clear how often this process may take place. Allen (1948) describes three processes in the formation of bauxite: (1) desilication; (2) migration; and (3) resilication. Desilication of the original rock and the migration of the clay and aluminous material have long been recognized. Resilication, or the union of silica with Part I] Geolohv axi> ]\Ii.\i;ualogy 29 f^ibljsite to form clay, however, has not been generally reeofrnized. Allen states that the occurrence of cellular kaolin in Georgia sugjrests formation by addition of silica lo frililisitc. rather tliau by migration of kaoiinite. Tliis conclusion he strongly supports in a more recent publi- cation (Allen, 1952)." Bates (1942, 1945) has described a kaolinitic clay, the Edwin clay, from near lone, California. Field work, mi- croscopic study, and evidence secured from firing: tests sliow that the Edwin clay is residual, derived directly from latorite. Transitions observed in the field, similar textures, and other evidence suggest that the laterite is intimately connected with greenstone (Jurassic meta- andesite). common in the foothills of the Sierra Nevada. Goldich (1948) concludes that in regions favorable for the formation of laterite and bauxite, the position of the water table may determine whether "transitional" clay will form, or bauxite will lie produced directly. Be- low the water table, clay minerals are most likely to be the end product of weathering ; but above the water table, the altei-atiou ]u-odnct will be bauxite or laterite. Below the water table in tropical regions, endellite. hal- loysite, and kaoiinite develop. Why halloysite 2HoO and halloysite 4II2O both develop below the water table is not clear. Lapparent (1936) reports that the bauxitic clays of Ayrshire contain boehmite, diaspore, and kaoiinite formed from basaltic rock. He believes that the altera- tion began as an alumina-silica gel, resulting from the hydrolysis of silicates in a tropical climate. The evolution of the gel first took place in the presence of humic waters that favor the crystallization of boehm- ite; subsequently, it evolved at a lower depth, where higher temperature favored the crystallization of dia- spore. Kaoiinite was formed later than the hydroxides of aluminum. CONCLUSION In view of the complex chemical conditions involved and the wide variety of possible origins, it is remarkable that the clay minerals fall into a comparatively small number of groups. Although great improvement in the nomenclature of clay minerals has been made in recent years, there is still considerable progi'ess to be made. It is hoped that a more general agreement on the nomenclature of .several important groups may be possible before long. Such clarification would facilitate discussion. Ma.jor contributions remain to be made, particularly in accunndating experinu^ntal data on clay-forming processes. Further, a closer correlation of field and laboratory studies and studies of supergene and hypo- gene clay deposits on the part of workers in this field is to be encouraged. Much laboratory work is to be checked in the light of field conditions. DISCUSSION R. A. Rowland: Professor Kerr mentioned some of Dr. Grim's work in the Gulf of California which indicatefl the disappearance of kaoiinite and the presence of degraded illite. It has been noted that flakes of mica larger than most clay particles are common in the soils in the United States. S. B. Hendricks refers to these as "stripped micas" because most of the potash has been removed. Relow the upper soil zone where the stripped micas are found, there is often a gunibotil composed largel.v of a clay mineral which has been called either beidellite or muntmorilloiiito. The material from the gumbo horizon is similar to that in the shales of the (Jiilf Coast region. It is a strii)ped illite which now behaves like montmoril- lonite. These shales aUso contain larger Hakes of potash micas which couhl be called illite, and a variety of chlorites. It is thought that the soil material has been thoroughly weathered and broken down so th.'it only montmorillonite remains to be tran.sported to the Gulf of Mexico. Once in the (5ulf the montmorillonite may change to either illite or chlorite. What is this iiroce.ss of diageuesis in clays? Is there any evidence to indicate that isoniorphous substitu- tion takes place in the process of diageuesis or whether the substi- tution takes place long before the material has heeome a mineral? In other words, do the.se clays undergo diagenesis b.y high-energy- level processes or do they undergo diagenesis by low-energy-level solution jirocesses? R. E. Grim: The American Petroleum Institute is investigating recent sedi- ments in the Gulf of Mexico and a number of samples have been collected off the south part of the Texas coast in deep w:iter. These .samples are being .studied in my laboratory. The water-soluble salts have been washed out of the samples and the amounts of caU'ium, sodium, potassium, and magnesium determined. The deter- minations did not yield absolute vahu's but ratios of sodium to each of the elements. It was found that material deposited on the Gulf floor is largely montmorillonite. with a little illite and kao- iinite. A plot of the ratio of sodium to potassium in the soluble salts against, distance from shore shows a relative decrease in l)otassium close to shore. There is a correlative increase in the proportion of illite-like material close to shore. The .sodium-mag- nesium ratios remain constant until some distance from shore, where the magnesium seems to disappear. P. F. Kerr: I referred to the large accumulation of clay minerals in the Gulf Coast region which has been derived from the erosion of a very heterogeneous assemblage of material. Dr. Grim has pointed out that man.v of these materials come from a variet.v of scuirces in a liasin at least 1.000 miles across, and yet ultimately are reduced I'hiefly to montmorillonite. There is a parallel to this in connection with hydrothermal alteration. At Marysvale, Utah, similar clay- mineral assemblages are produced in surprisingly different rock types. Isaac Barshad: In order to answer Dr. Rowland's question as to the possible diagenesis of the clay minerals upon being deposited in the Gulf by the rivers, it is first necessary to consider the changes which the mica minerals undergo during the course of weathering into the hydrous mica minerals — or if you like — the "stripped micas." Two of these changes are significant in the present discus.sion : (1) the replacement, through exchange, of the interlayer K* by H*, Ca**, Mg**, and possibly Na* ; and (2) the resultant hydration and interlayer expansion of the crystal lattice. Both of these changes bring about a reduction in the density of the weathered mica. It is also important to remember that the changes indicated are readily reversible. With these facts in mind, the process of diagene- sis may be pictured to involve (1) a replacement of the readily exchangeable cation with K*. particidarly in those mica clay min- erals which have undergone the smallest degree of change; (2) a dehydration and contraction of the crystal lattice; and (.3) an increase in densit.v of the crystal lattice. The increase in density is important because it causes this fraction of clay to settle out first; that may be the explanation of the fact that the clay material nearest the shore line contains the most mica minerals. The fraction of the clay which dues not undergo the changes described will settle farther from the shore line and change into a chlorite-Iike mineral. As to whether isomorphous substitution idays a role in dia- genesis — I would say that it probably does not. It is unnecessary to postulate isomorphous changes to explain the results. W. F. Bradley: One difference between kaolin minerals and montmorillonite min- erals is that those of the montmorillonite group are stable in extremely minute flakes, while those of the kaolin group have a large inherent particle size. Alteration resulting in kaoiinite usually ends at that point. When alteration products pass through a degraded mica or montmorillonite stage they are highly mobile. They lose alkalis and magnesium and the particles are then rela- 30 Clays and Clay Technology [Bull, im tively barron (hiring ti'anspoi'tatkm to the sea. The pH of the ocean is higher than that of river water. Through increased pH, cations and protons are extracted from the .sea water, which is essentially a process of precipitation of brncite. Ttie magnesium content of the ocean water is such that it nia.v actually represent a concentration eipial to the solubility of the brucite attached to chlorite at the pll which obtains. When the clay comes into salt water it immedi- atel.T jiicks up some potash and magnesium by ordinary exchange mechanism. It becomes illite or magnesium-saturated montmoril- lonite. As pH increases with distance from shore brucite layers are .synthesized and the montmorillonite changes to sedimentary chlorite. They need not become trioctahedral and ordinarily the chlorite of sediments does not appear to be trioctahedral. The diagenesis takes place between the layers but the layer structure need not be affected. T. F. Bates: There is an interesting analog.v between this situation and that for the Ordovician metabentonites. or potassium bentonites, which occur in abundance in the Appalachian region. C. E. Weaver has established tliat iiotassium bentonite found in I'eunsylvania is a mixed layer complex (Weaver and Bates. T.I."p2). By using Brown and MacEwan's ( Ifl.jl ) curves. 80 percent of the layers were found to be saturated with potas.sium and were non-expanding, and 20 percent of the layers were saturated with calcium and a little sodium and were or could be expande. Bates, T. F.. Hildeliraud. F. A., and Swineford. A., 19.50, Mor- phology and structure of endellite and halloysite : Am. Jliueralogist, V. 35. pp. 403-484. Brannock. W. W., Fix. P. F., Gianella, V. P., and White, D. E., 1948, Preliminary geochemical results at Steamboat Springs, Ne- vada : Am. Geoplivs. Union. Trans., v. 29. no. 2. pp. 211-226. Brown, (5.. and MacEwan, D.M.C., 1951, X-ray diffractmn by structure with random interstratification, in Brindley, G.W., ed., X-ray identification and crystal structures of clay minerals: Chaii. 11. pp. 2llti-2s4. Miueralog. Soc. London. Brush. G. .!.. and I>ana. E. S.. 1880. On the mineral h>cality at Brauchville. Connecticut, spodumene and its alteration products: Am. Jour. Sci., v. 20. 3d ser., pp. 283-284. Buchanan. H.. 1807. .Tcuirney from JIadras through Mysore. Canada, and Malabar. Part II Geology and ^NriXERAi.ocv 31 liiiiisiMi, U. isril. t'l)pr (lie Piuzessc clci- vulUanischeii Gesteius- lpilcliiii!;i'ii Islnnds: I'liKKt'iHli'i'fTs Ann. I'liys. u. Clu'in., v. Ki, pp. 11I7J71.'. I'.yi IS, F. .M.. Jr.. niitl P.raniK.ck. W. W.. 1!I4!I. V..lcaiiic activity on rnmak and (ircat Sitkin Islands, llltti lillS : .\in. (Jcopliys. I'nicin, Trans., v. '.M). pp. 71!l-7y4. Cilla^'lian. F... 1!M."S. Endellite deposits in Gardner Mine Ridfje, Lawrence Conntv, Indiana: Indiana Dcpt, Conservation, l)iv. Ccolii^jy, I'.nll. 1. Camiiliell, J. M., 1017, Laterite, its origin, struetnre, and miner- als : Minini,' Maj;., v. 17, pp. 67-77, 120-12S, 171-17!). 22(1-22!). Collins, .1. H., 1S87, On the nature and \>. .'iS.S-3!)4. Foshag, W. F., and Woodford. A. ().. 1!).'?(!. Rentonitic mag- nesian clav-niineral from Califurnia : .\ni. .Mineralogist, v. 21, pp. 23S-244. Frankel. .T. .L, 1040, Dickite from the Witwatersrand gold mines: Mineralog. Mag., v. 2S. pp. .582-.5S0. Frederickson, A. F., 1051. Mechanism of weathering: Geol. Soc. America Rnll., v. (i2, pp. 221-2.32. (lalpin. S. L.. 1!)12, Studies of flint clays and their associates: Am. Ceramic Soc. Trans., v. 14. pp. .'!01-;!4(>. Gieseking. .T. E., and Mortland. .M. M.. 1!)51, Influences of the silicate ion on potassium fixation: Soil Sci., v. 71. pp. 381-385. (Jlass, Ilerliert D.. 1051. Clay mineralogy of the coastal plain fcirmalions of New .Tersey : Ph.D. thesis, Columbi.a I'niversity. Goldich. Samuel S., 1048, Origin and development of aluminous lalerile and bauxite (abstract) : Geol. Soc. America Bull., v. 50. p. i:!2i;. (ioldmau, Marcus I.. 1949, Petrology of liauxile surrounding a boulder-like core of kaolinized uepheline syenite in .\rkansas (ab- stract) : Geol. Soc. America Bull., v. (iO. pp. 18!)0-1S01. Gordon, M., .Jr., Tracey, .1. I., ,Tr., and Ellis, M. W., 1040, Field relations of Arkansas bauxite deposits (abstract) : Econ. Geology, V. 44, pp. 040(541. Graham, E. R.. 1041. An acid cla.v — an agent in chemical weathering: .lour. (Jeuhigy. v. 40, pp. .302-401. (irim. R. E.. 1033. Petrography of the fuller's earth deposits, ()lmstead, Illinois, with a brief study of siune non-Illinois earths: Econ. Geolog.v, v. 28, pp. 344-3G3. Grim, R. E.. and Allen, V. T.. 10.38, IVtroli.gy ,.f the Penn- sylvanian underclavs of Illinois: Geol. Soc. .\merica Bull., v. 4!). jip. 148.5-1514. Grim. R. E.. and Bradley. A\'. F.. 19:'.9. A unique clay from the (Joose Lake, Illinois, area: Am. Ceramic Soc, Jour., v. 22, pp. 157-1 (U. Grim, R, E., Dietz, R. S., and Bradley, W. F., 1040, Clay mineral composition of some sediments from the Pacific Ocean off the California coast and the Gulf of California: Geol, Soc. America Bull.. V. (111. pp. 178(5-1808. Grim. R. E.. Lamar, J. E., and Bradley, W. F., 1030, The clay minerals in Illinois limestones and dolomites: Jour. Geology, v. 45, pp. 820-834. Gruner, J. W., 193f), Hydrothermal alteration of montmorillonite to feldspar (abstract) : Am. Mineralogist, v. 21, p. 201. Gruner, J, W,, 1044, The h.vdiothermal alteration of feldspars in acid solutions between 300^ and 400''C. : Econ. Geolog.v, v. 34, pp. 578-589. Hagner, A. F., 103!), Adsorptive clays of the Texas Gulf Coast: Am. Mineralogist, v. 24, pp. (57-105. Harder, E. C, 1940, Stratigraphy and origin of bauxite deposits: (Jeol. Soc. America Bull., v. 00, pp. 887-908. Hardy, F., and Follelt-Smith, R. R.. 1031, Jour. Agric. Sci., V. 21, i)p. 73!)-70I. Ilarrassowilz. IL, 102(;, Laterite: Forlschr. Geol. Palaeont., V. 4. pp. 2.5.3-500. Harrison, Sir .1. R., 1934, The katamorphism of igneous rocks under humid tropical ciniditions: Imp. Bur. Soil Sci., Harpenden, England. Henry. .\. V.. and Vaughn, W. H., 1937, (Jeologie and techno- logic asiiecis of the sedinientarv kaolins of Georgia: Am. Inst. Min. Eug., Tech. I'ub. 774, pp. i-11. lioness. A., and Williams, F. .!., 1935. Dickite from Pennsyl- vania : .\m. Mineralogist, v. 20. pp. 4(52-4(5(5. Hoskiug, J. S., 1940, The soil clay mineralogy of some Aus- tralian .soils developed on granitic and basaltic parent material: Jour, .\ustrali.an Council Sci. and Industrial Res., v. 13, pp. 200-210. Ipjiolilo. S., 1047. Energia geotermica per usi industriali : (^riteri di recerca e orientamenti in Italia (Industrial utilization of geo- tbermal energy: Its criteria in research work and its aims in Italy) : Riv. Geominer., v. 8, no. 1, pp. .31-38. Kelliu-. W. I)., and A'alduga. A., 1040, The natural steam at Larderello, Italy: Jour. Geology, v. .54, pp. 327-334. Kelley, W. P., 1030, Recent marine sediments. Am. Assoc. Petro- leum Geologists, pp. 454-4C5. Kerr, P, F., 10.30. Kaidinite from a Brooklyn subway tunni'l: Am. Jlineralogisl. v. 15, pp. 144-158. Kerr. P. F., 1!»37. Attapulgas clay : Am. Miueralogi.st, v. 22, pp. .534.550. Kerr, P. F., 1!I42, Origin of quartz deposit at Fazenda, Pact'i, P.ra/.il : Am. Mineralogist, v. 27, pp. 487-490. Kerr, P. F., 1951. Alteration features at Silver Bell. .\ii/.iiua : Geol. Soc. America Bull., v. 62, pp, 451-480. Kerr. P. F., Kulp, J. L., Patter.son, C. M., and Wright. R. J., 1!).50. Hydrothermal alteration at Santa Rita, Xew Mexico: (!!eol. Soc. America Bull., v. 61, pp. 275-347. Lapparent. .7. de, 19.36, Roehmite and diaspore in the liauxitio clays of Ayrshire: Snmm.-irv of Progress of the Geol. Survey, pt. II, pp. 1-7. Leonard. R. J., 1027. The hydrothermal alteration of certain silicate minerals: Econ. (Jeolog.v. v. 22, pp. 18-43. Lindner, J. L., and Gruner, J. W., 1039, Action of alkali sulplii8. Martin, F. .1., and Doyne, II. ('.. 1!)27. Laterite and l.-iteritic soils in Sierra Leone: Jour. Agr. Sci., v. 17, pp. .530-547 ... v. 20, pp. 135-143. Mohr. E. C. 1!)44, The soils of equatorial regions with special reference to the Xetberlands East Indies. .1. W. Edwards, Ann Arbor, Mich., trans, by R. L. Pendleton, 604 pages. Morey. (J. W., and Ingerson, E., 1037, The pneumatolytic and hydrothermal alteration and synthesis of silicates: Carnegie Inst., AN'.ish., Geophys. Lab., I'aper 937 . . . Econ. Geology, v. .32, pp. (5O7-071. JIurata, K. 'J., 1040, The significance of internal structure in gelatinizing silicate minerals: TI. S. Geol. Survey Bull. 050, pp. 25-33. X'oll, W.. 1!)32, Hydrothermal Synthe.se des Mu.scovits. Ein Beitrag zur Frage Serizitbildung in Tonschiefern : X^ach. Ge.ssell. Wiss. G.-.ttingen, Math.-Physik. Kl., v. 1932, pp. 122-134. X'oll, W., 1!)34. Hydrothermal Synthese des Kaolins : Min. Petr. Mitt., v, 45, pp. 175-1!)(). X'oll, W., 1035, Jlincralbildung im System AUOa-SiOi-HsO : Xeues Jahrh. Min. Geol., v. 70, pp. 65-115. 32 Clays and Clay Technology [Bull. 169 Xoll, W., 1036, Ueber die Bilduns^^liedingungeu von Kaolin. Montmorillonit, Serieit. Pyrophyllit, nnd Aualcim : llin. Petr. Mitt., V. 48, pp. 210-247. Norton, F. H.. 1037, Accelerated weathering of feldspars : Am. Mineralogist, v. 22. pp. 1-14. Norton, F. H., 1039, H.vdrothcrmal formation of clay minerals in the laboratory : Am. Mineralogist, pt. I, v. 24, pp. 1-18 . . . pt. II, V. 20. pp. 1-17. Xutting, P. G., 1945, The solution of soil minerals in dilute acids: Science, v. 101, no. 2633, pp. 619-621. I'ayne, .1. H., and Man, K. T., 1040, A .study of the chemical alteration of l)asalt in the Kilauea region of Hawaii : .Tour. Geology, V. .54, pp. 34.5-385. Pendelton, R. L., 1936, Am. Soil Survey Assoc, v. 17, pp. 102-138. Porter, J. T., 1907, Clays and clay products: U. S. Geol. Snrvev Bull. 315, pp. 268-290. Rankama, K., and Sahama, Th. G., 1050, Geochemistry : Univ. Chicago I'ress. pp. 500-507. Robinson, G. W., 1949, Soils, their origin, constitution, and clas- sification : .John Wiley and Sons, Inc., pp. 409-415. Ross, C. S., and Kerr, P. F., 19.34, Hallovsite and allophane : U. S. Geol. Survey. Prof. Paper, 185-G, pp. 134-148. Ross, C. S., and Shannon, E. V., 1926, Minerals of bentonite and related clays and their physical properties : Am. Ceramic Soc. Jour., V. 9. pp. 77-96. Sales, R. H.. and Meyer, C, 1949, Results from preliminary studies of vein formation at Butte, Montana : Econ. Geology, v. 44, pp. 405-484. Sand, L. B., and Bates, T. F.. 1952. Mineralogy and petrology of the residual kaolins of the southern Appalachian region ; Penn. State College Contract No. N00nr20914. NR 081-098. Tech. Rept. 7. Schwarz, R., 1933, Kiinstliche Umwandlung von Feldspat in Kaolin : Naturwiss, v. 21, p. 252. Schwarz, R., and Walcker, R., 1925, Uber die Genesis der natiirlichen Aluminium-hydrosilicate : Zeit. anorg. Chem., v. 145. pp. 304-310. Shepherd. E. S.. and Merwin. H. E., 1927, Gases of the Jit. Pelee lavas of 1902 : Jour. Geology, v. 35, pp. 97-116. Silva, J. J. R. da, and Neiva, J. M. 0., 1948, Montmorillonite dans les pegmatites granitiques et le problenie de la montmorillo- nitization : Estudos, Notas, e Travalhos do Servico de Fomento Mineiro. Tome IV, fare. 1, pp. 1-7. Singewald. Q. D.. 19.32. Alteration as an end phase of igneous intrusion in sill on Ix)veland Mountain, Park County, Colorado : Jour. Geology, v. 40. pp. 16-29. Stringham, B., and Taylor, A., 19.50. Nontronite at Bingham, Utah : Am. Mineralogist, v. 35, pp. 1060-1066. Swineford, A., and Frye, J. C, 1951, Petrography of the Peoria loess in Kansas : Jour. Geology, v. .59, pp. 306-322. Tarr, W. A., and Keller, W.D., 10.36, Dickite in Missouri : Am. Mineralogist, v. 21, pp. 109-114. Tarr, W. A., and Keller. W. D., 1937, Some occurrences of kans. This is particularly important in relation to problems of clay- mineral identification, for it is clearly necessary for the methods employed to be sensitive both to structure and to composition. • Reader in X-ray Physics. Physics Laboratories, The Universitv, Leeds, England : now Research Professor of Mineral Sciences. Pennsylvania State University, Pennsylvania. T To this list can now be added Grim. R, E., Clay Mineraloav. Xew York. 19.T3. 2—91001 Structurul Groups and Sub-Groups. The minerals with which we are concerned lie in the broad chemical category of silicates, which, following W. L. Bragg (19:37) anil others, may be classified into structural grou])s and su])-groups. Tiie cpiestion of the naming of these structural groups has been tliscussed by Fleischer (1947), Strunz (1941), and others, and need not be con- sidered further here. Clay minerals come mainly in the layer-silicate group and these can be subdivided accord- ing to the type of layer structure. The naming of these layer types will doubtless also be a matter for discus- sion. In addition to the single-layer types, a structural sub-group is also retpiired for the mixed-layer struc- tures, and eventually more than one mixed-la.N'cr sub- group may be necessary. In addition to the layer-silicate clays, there are those which appear to be more closely allied to chain-silicate structures (pyroxenes and amphiboles"). namely paly- gorskite or attapiilgite and jiossibly seiiiolite. These are less fully understood than the layer silicates and exist- ing knowledge of these minerals up to 1950 has been summarized b.y Caillere and Ilenin (1951). Any scheme of cla.ssifieation is in danger of imposing a rigidity which the sub.iect will eventually outgrow and it is therefore desirable to retain flexibility in the scheme. This is especially necessary in connection with the classification of hydrated minerals and of mixed- layer minerals. Hydrated minerals can often be associated most con- veniently with the corresponding dehydrated forms, but on other occasions they may be treated as mixed-layer minerals. At present it seems undesirable to allocate them rigidly in either way. In the mixed-layer group, it may eventually be useful to distinguish between randomly mixed and regularly mixed structures, and particular examples of the latter may be classified at some future date as distinct struc- tural sub-groups. In fact, this is already true of the chlorites which consist of a regular alternation of mica- type and hydroxide-type layers, and. in view of their general importance it is most useful to regard them as constituting a particular structural sub-group. Other examples of regularly mixed layer structures, which are less well understood or which are of rare occurrence, can, for the present, be left in the mixed-layer group. x\s illus- trations, clay minerals showing long s]iaeing regularities may be mentioned, such as a montmorillonite with a 32A s]iaciiig (Alexaniau and Wey. 1951), a mica-type min- eral with a 22 A spacing (Caillere. Mathieu-Sicaud, and Henin, 1950), rectorite with a 25a spacing (Bradley, 1950) and various weathered clays di.scussed bv Jack- son et al. (1952). Chemical Species. "Within each structural group or sub-group the 'chemical species are divisible according to composition. In some cases this gives clearly defined species (e.g. the kaolin minerals, the serpentine minerals) but in others where continuous or largely continuous composition ranges exist, (e.g. the montmorillonite-bei- dellite series, the chlorites) boundary lines must be drawn in a largelv arbitrarv manner. (33) 34 CHEMICAL CATEGORIES Clays and Clay Technology Tdhle 1. A .irhriiie of rlnsxifivntioii of clai/s and relatnl ailicatr niiiicidh- SILICATES I Bull. IG!) STRUCTURAL GROUPS SUB-GROUPS CHEMICAL SPECIES Layer silicates X Chain silicates STRUCTURAL VARIETIES Kaolin ^ype "T Kaolin minerals Serpentine minerals Chamosite Amesite Greenalite Cronstedtite etc. Halloyslte Koolinite Dickite Nocrite Chrysotile(s) Antigorite and hydrated forms Mica Chlorite type type ^r 1" Talc Penninlte Pyrophyllite Clinochlore Muscovite Prochlorite Phlogopite Daphnite Biotite etc. Glauconite lllite(s) Montmorillonoids Vermiculite etc. Polymorphic Polymorphic varieties varieties Mixed-layer *ype "T Anouxite Bravaisite etc. 1 Chloritoid Palygorskife (attapulgite) Sepiolite and hydrated forms There may be some disagreement about grouping to- gether the kaolin minerals proper, the serpentine min- erals, and others placed in the kaolin-type group and also about this choice of a name for the group. The struc- tural scheme is essentially the same for all these minerals and the only point of difference is that some are dioctahe- dral and others trioctahedral. * This distinction, however, occurs in all the sub-groups and appears to be of rela- tively small structural iniiM)rtance. To achieve consist- ency, the same policy must be adopted with all the groups and the simjilest solution is to place di- and tri-octahe- dral members always together; this also eliminates any difficulties with minerals having an intermediate num- ber of octahedral ions. As regards the naming of the group, the term "kaolin-type" suggests the type of structure without actually taking the name of one of the members. The use of a munerical ratio, such as 1 : 1 and 2 : 1 for designating the kaolin- and mica-type layers, has some advantages but it is not easily extended to all the layer types which are involved ; for examjile, is a chlorite layer to be designated 2 : 2 or 1 + 2 : 1 ? It will be observed that in the mica-type group there have been included talc and jiyrophyllite, the micas proper (muscovite, biotite, . . . ), the clay micas (illites. etc.,) and the swelling minerals (montmorillonoids. ver- miculite). The justification for iilacing all these minerals together in a single structural group instead of separat- ing them into "lOA" and "14a" groups or into non- swelling and swelling groups (as is usually done), is that in view of the very close inter-relations of these minerals, the simplest procedure is to regard them as * Tliese term.^ .'5is:nify the number of cations in octahedral coordina- tion per half unit cell. They correspond to the ternrs heptaphyl- lite and (>ctaph\ilite. different chemical species within the mica-type sub- group, having various hydration properties. Strncturol Varieties. The structural Varieties of eacli chemical species differ in the manner in which the struc- tural layers are arranged with respect to each other. To recognize them in clays by the X-raj' powder method, it is necessary to record carefully the details of their dif- fraction diagrams. With well-cry.stallized and relatively pure materials, many of the varieties may be recognized by their powder diagrams (see Brindley, 1951a) but with poorly crystallized and or impure materials considerable difficulties can arise. To the kaolin varieties and serpen- tine varieties, which can be recognized optically, separate names have been given. Polymorphic varieties of the micas, studied by Hendricks and Jefferson (1939), and of the chlorites, studied by Brindley, Oughton, and Robinson (1950), have not been given separate names and require rather detailed X-ray examination to be recognized. They are most usefully described in terms of the unit cell symmetry and the number of structural layers per unit cell, a system of nomenclature which has already been usefully applied to the polymorphs of silicon carbide (Ramsdell 1947; Ramsdell and Kohn, 1951, 1952) and of wurtzite (Frondel and Palache, 1950). Grim and Bradley (1951) have shown that some of the mica polymorphs can be recognized fi-om X-ray powder diagrams, but no such evidence is yet available for the chlorite polymorphs which have hitherto been recognized only from single crystal diagrams. LATTICE PARAMETERS AND CHEMICAL COMPOSITIONS OF LAYER SILICATES A consideration of the relation between the lattice parameters and the chemical compositions of layer sili- I'art I] Geology and Mineralogy 35 cates is of interest in eoniiectioii with i a i tiie identifica- tiou of mineral speeies by means of X-rays, and (b) certain questions related to the morjiiiolof^y of elay min- erals. The question of identification is considered in a second contribution to this symposium ; here considera- tion is friven onh- to the morpholopieal aspects. The variation of the lattice parameters witli chemical composition lias been discussed by several authors in X-ray Identification and Cri/stal SInicturcs of Clay Minciah for particular mineral groups, namely the montmorillonoids by ilacEwan (Ch. IV), the micas by Brown (Ch. V), and the chlorites by Brindley and Robinson (Ch. VI). The development of a more general treatment has been attempted by Brindley and MacEwan (1953) iu a paper in the process of publication, which is briefly outlined here.* Two parameters call for consideration, lu the first place, in so far as the individual layers have hexagonal S3Tnmetry, a single parameter suflScies to express the Figure 1. I'rojcction.s of Si-O I;i.vpi- (left), .tikI X(0H|= la.ver (right) on the ub plane of la.ver lattiee silicates. Small hlack circles represent Si, small open circles X, and lai-ge open cii'cles, O or (OH). Each Si is tetrahedrall.v coordinated to O; the fourth O of each tetrahedral group (not shown iu the diagram) projects on to the Si atom at its center. Each X atom is octahcdrally coordinated to 3 (OH) above the plane of pro.iection. shown b.v heav.v circles, and 3 (OH) below the plane, shown b.v light circles. The right-hand diagram represents a trioctahedral la.ver; one-third of the X atoms are missing in dioctahedral layers. dimensions in the plane of the layer; bo, (see fig. 1), is a convenient value since it can be obtained in most cases from the easil.v observed (060) reflection. For all layer lattice silicates the ratio of the orthogonal param- eters, bo^ao, is equal to or very nearly equal to the value \/'i which corresponds to a perfect hexagonal ar- rangement. The variation of b,, with composition will depend mainly on the effective sizes of the cations in tetrahedral and octaliedral positions in the oxygen and hydroxyl networks (see fig. 1). The interlayer cations, which are * This has now appeared in Ceramics — .-l Symposium, published by the British Ceramic Society, 1953. more loosely coordinated, sucli as the K* ions in musco- vitc whicli are in positions of twelve-fold coordination, are unlikely to have more than a small effect on b„. A second parameter which may be considered is the layer thickness. If Co is the third unit cell dimen.sion and if there are n layers per unit cell, then the layer thick- ness is given by Co/n for orthogonal axes, Co (sin P)/n for monoclinic axes, and Co(l — cos-a — cos-(3) ^/ji for triclinie axes with a ^ 90", p ^ 90". and y = 90". The layer thickness dc|)ends on composition in a more complex way than the parameter b„. The thickness of the silicate network itself depends on the sizes of the tctra- hedrally and octahedrally coordinated ions. The separa- tion of the networks from each other depends partly on the size of the interlayer cations, but also on the magni- tude of the cohesive forces between these cations and the negatively charged silicate layers. It is to be antici- pated, therefore, that the layer thickness will be less easil.v interpreted in terms of composition than the pa- rameter b(i. The ba parameter was first discussed by Pauling (1930) who compared the dimensions of the hexagonal Si-O networks in /3-tridjanite and /8-cristobalite with those in gibbsite, A1(0II)3, and brucite, Mg(0H)2. On the basis of this compari.son he considered that a hy- drated alumino-silicate layer of the kind found in the kaolin minerals was dimensionally feasible, and that a corresponding magnesian silicate was unlikely to exist because of the misfit of the component layers, (see data in table 2), In the light of present knowledge, it seems better to start with the Si-O distance, 1.62 A, found in many silicates. This leads to a value &o = 9.16a for a hexagonal layer of regular Si-O tetrahedra, and when this is compared with the data for Al(OH).-) and Mg(OH)o (see table 2, section 3) there is no difficulty in visualizing the formation of both aliuninum and mag- nesium layer silicates. A difficulty in deriving any relations between bo values and chemical composition is to decide .just how to effect a compromise between the dimensions of the com- ponent tetrahedral and octahedral sheets. Section 3 of table 2 suggests that the arithmetic mean gives values close to the observed values. An important ]ioint for con- sideration is the precise value for the Si-O distance or for 6o for the Si-O tetrahedral net work, since Si-O = 1.62 or &o = 9.16 was never intended to be taken as a precise value. Taylor and his co-workers (see, for example. Cole, Siiriun and Kennard, 1949) have considered the effect of random substitution of Al for Si in feldspars and have concluded that the mean (Si,Al)-0 distance in- creases bv about 0.02a for each Si replaced by Al in a group of four. This would give a corresponding increase in &o of 0.12A. Brindley and MacEwan (1953) have therefore taken the &o parameter of a tetrahedral net- work of comiiosition (Si4_sAlxOio) to be b„=zT + 0.12.r where T is likel.y to be about 9.1-9.2A, the actual value being found from experimental data. m Clays and Clay Teiiixulogy I Bull. 169 Values of ho foi- octahedral layers have been obtainetl from the hydroxide struetiires and are given in table 2, section 1. The calculation of /)o for combined tetrahedral-oeta- hedral layers involves taking the appropriate mean values. Empirically the arithmetic mean has been used and the best value of T found by comparison with ex- perimental data. For dioctahedral structures, T is found to be about 9.16A which agrees exactly with Si-0 = 1.62A. For trioctahedral structures, T = 9.00A. Table 2. Lattice parameters in some layer minerals. 1. Octahedral layers. Al(OH)a gibhsite bayerite re(0H)3 Mg(OH)j Fe(OH). Cation radius Al =0.57 Fe"+=0.67 Mg =0.78 re>+=0.83 6o parameter, A. 8.62 8.68/ * 9.001" 9.36 9.72 *i* Mean value. <2) Interpolated value. 2. Si-0 Tetrahedral layers. From hexagonal networks in ^-tridymite and 3-cristobalite (after Pauling, 1930) On basis of Si-0=1 .62. 60 parameter, A. 8.71 9.16 3. Comparison with Kaolinite and Chrysotile. &o, in A &,i, in A 9.16 8.65 S.flOs 8.03 Si-0 layer Mg-OH layer 9.16 Al-OII layer 9.36 Moan vahie Observed values for 9.26 Observed value for 9.24. 9.18 With these values for T, and on the assumption that a mean fe,, parameter for the combined tetrahedral and octahedral networks is to be taken, the following formulae are obtained : Dioctah edral minerals ho = 8.90 + OMx + 0.09g + 0.18/- + 0.27s Trioctahedral minerals ho = 9.18 + 0.06.r — 0.12/> — OMq + OMs where X = no. of Al for 81 substitutions when the for- mula is expressed in the form (,Si4_xAlxOiii) p zm no. of Al atoms q ^ no. of Fc'^* atoms r = no. of Mg atoms s = no. of Fe-* atoms in octahedral positions p-\-q-\-r-\-s^2 for dioctahedral minerals 3 for trioctahedral minerals. A detailed comparison of &,, values calculated from these formulae and obtained experimentally is given by Brindley and MacEwan (1953), and on the whole a close agreement is obtained. Some of the discrepancies ma.y arise fi'om experimental errors, but others may have real significance; for example, discrepancies found with the micas may arise from the effects of the inter- layer cations. Generally the best agreement is found with minerals having no interlayer cations, such as the kaolin- type minerals and the chlorites, and with minerals like the moiitmorillonoids where the interlayer cations are relatively few in number and are generally hydrated. THE MORPHOLOGY OF CLAY MINERALS Lamellar Forms. Lamellar crystals of hexagonal out- line would be expected from layer structures consisting of hexagonal networks of atoms, but when tlie morphol- ogy of clay minerals is considered in more detail, many ([uestions arise which still require elucidation. In the case of kaolinite, it is probably generally true that well-formed hexagonal flakes give clear X-ray dia- grams indicating a well-ordered succession of layers. The type of kaolin mineral found in many fire clays seems usually to be very poorly crystallized and the X-ray powder diagram indicates considerable disorder in the stacking of the layers, with many displacements of layers by iibo. 'i, n being integral (Brinclley and Rob- inson, 1947). Both in the clay domain and on a mega- scopic scale, disordered sequences of this type can be obtained with material showing a well-developed morphol- ogy. One cannot therefore necessarily associate (piality of morphological development with degree of, crystalline regularity. Kaolinite crystals from different localities show considerable variations in their (thickness size) ratio and in the elongation of the hexagonal forms. No explanation of these variations has yet been offered. The poor development of montmorillonite crystals, few of which show more than an occasional 120° angle between edges, may be associated with the irregularly superposed and easily separated layers of this mineral. On the other hand, vermiculite which is similar in many respects to montmorillonite, exists as mega-crystalline material. Tuhular and Related Forms. The iiou-oricutaf ion of halloysite by simple sedimentation pointed to a different morphology from that of kaolinite and this has been strikingly confirmed and amplified by electron micro- graphs, especially by Bates and his co-workers (1950) who have indicated the existence of tubular crystals as well as split and partially unrolled tubes. The X-ray diagrams are consistent with a randomly disiilaced sequence of layers (Brindley and Robinson, 1!I48). The occurrence of tubular forms has been attributed by Bates et al. to curvature of the layers resulting from the strain imposed by binding the Si-0 and Al-OH layers together. A minimum strain will exLst in a layer which is curved so that the Si-O and Al-OH sheets have, as nearly as ]iossible, tlieir own characteristic dimensions, namely /)ii = 9.16 for the Si-0 sheet and ho = 8.75 for the Al-OII layer. This corresponds to a diameter of about 150A. (This is similar to, but not identical with. Bates' calcu- lation). Following Bates, we may suppose that in hy- drated hallovsite a silicate sheet of this diameter forms Part 1 1 Geolocy and ]\Iixi;ualogy 37 WATER LAYER SILICATE LAYER a. b. FlGURK 2. l)i:is;raiu In illiisli-nlc (») J'.atcs' siiKHOstidii I'm- the stnutur-c .if a tiilii" nf liydraliMl liallnysite, (b) HriiuUi'.v's sugges- tion for the (Structure of a curved fraguiont of imperfectly dehydrated lialloysite; water inclusions lie in the imperfectly fitting, curved silicate sheets. The mean layer thickness is about 7.5A as compared with 7.15 when flat layers are packed together. The diagrams are drawn correctly to scale as regards the water and silicate layer thicknesses and the inner tuhe diameter is equivalent to almut 2fMlA. Silicate layer is represented by white spaces between haehured areas. the iniii'i-iiKist la.x'er. Alternate layers of water and sili- cate may then be added externally with increasing radiii.s and therefore increasing- strain, until the strain imposes a limitation preventing further growth (see fig. 2a). "When placed in the electron microscope, the system will largely dehydrate, ])robably to a metahalloysite with a residual water content given bv the composition Ah.(Si,.0,-,) (OH) 4 :}4 II2O, and a lattice spacing in the range 7.2-7.5A. The suggestion is made that on dehy- dration, the silicate layers (with the possible exception of the* innermost layer) collapse by removal of the inter- layer water, producing broken fragments having radii of about ir)OA, but retaining a certain amount of residual water in the jn-ocess. This is depicted in figure 2b. This view replaces that described by Briudley and Robinson (l!)46b) based on flat layers which was suggested prior to the discovery of the tubidar habit. A similar explanation may be given of the tubular habit of ehrysotile, discovered by Bates, Sand, and Mink (1950), but in this case the Si-0 sheet will be on the concave side of each curved la.yer whereas in halloy- site the reverse is true. No evidence has yet been brought forward for the existence of a hydrated ehrysotile analo- gous to hydrated lialloysite. Brief reference may also be made to a description by Onsager (1952) of a possible explanation of the peculi- arities exhibited by the single crystal photographs of antigorite, which was found by Aru.ia (1945) to have an flo parameter eight times greater than that of ehrysotile. The suggested explanation is that the structural layers are curved alternately in opposite directions so that a sine-curve distortion is impressed on the structure along the f/-axis. This reciuires a reversal of the struclure after every half M'ave. Other Morphological Forms. The existence of lath- like and fibrous forms has been previously attributed to a lattice strain or curvature limiting crystal growth in one direction and permitting growth in a perpendicular direction. While this is feasible in the case of a kaolin- type structure, it is difficult to see how it can appl.v to mica-like structures, such as nontronite and hectorite, which develop as fibrous forms. Beautiful ribbon-like crystals have been observed in electron micrographs of allevardite, a mica-type mineral studied by Caillere, Mathieu-Sicaud, and Renin (1950), but no explanation has yet been given for their occurrence. SOME STRUCTURAL TOPICS Relations Between the Kaolin Minerals. The prin- ci])al kaolin minerals, kaolinite, dickite, uacrite, and lialloysite, are now well understood and rcijaire little discussion. Their structural characteristics are sum- marized in table 3. The best crystalline kaolinite gives a clear powder diagram in which some 60 lines have been indexed with the triclinic parameters given in table 3 and their in- tensities interpreted in terms of a structure with one layer per unit cell (Brindley and Robinson, 1946a). This is ill contradiction to an earlier monodinic structure 38 Clays and Clay Technology Tahle 3. fninerah. [Bull. 169 Mineral Layers per unit cell, symmetry (in A) K (in A) (in A) References 5.14 5.14 5.14 8.94 8.90 8.9.3 8.94 5.14 7.37 14.42 43.0 d(OOl), 7.2-7.0 a91.8°. &104.o<' 696° 50' B90° 20' Brindley and Robinson (1948) Brindlev and Robinson (1946a) Dickite Gruner (1932b) Hendricks (1938) containing two layers per unit cell (Gruner, 1932a). Although the structure is essentially well-ordered, it may (and probably always does) contain some random layer displacements, as Hendricks (1942) has stated. which are more numerous than those occurring in dickite. There is a marked difference in the powder diagrams of well-crvstallized kaolinite and of poorly ciystallized ma- terial such as occurs in many fireclays. The essential crystallographic feature of the poorly crystallized ma- terial is the virtual disappearance of all reflections with k ^ 3m. This indicates the existence of a great number of random layer displacements of the type nbo/S. No evidence of triclinie character remains and such lines as can be observed may be indexed with a single layer, monoclinic cell (Brindley and Robinson, 1947).* Ilalloysite represents a more extreme form of disorder, from which the only reflections observable are the 001 which correspond to the stacking of layers at intervals of about 7.2 — 7.5A in dehydrated halloysite and lOA in hydi-ated halloysite, together with (hJc) bands which arise from the regularity witliin the layers. A treatment of X-ray diffraction from this mineral taking account of its curved or tubular form has not so far been at- tempted, but a good general interpretation of the diffrac- tion data has been obtained on the basis of a randomly displaced laver structure (Brindlev and Robinson, 1948). It is of interest to consider whether a series of min- erals exist extending from an idealh^ well-ordered kao- linite to a fully disordered halloysite, with the fireclay type fully disordered in one direction (?)-axis), as an intermediate stage. This question calls for a more ex- tended survey than it has yet received. The present writer is inclined to the view that all stages probably exist between well-ordered kaolinite and a kaolin mineral highly disordered along the &-axis, but is less certain of the existence of intermediate stages between the Z)-axis disordered type and halloysite. At this point reference must be made to a very careful study by Bramao, Cady. Hendricks, and Swerdlow (1952) of "kaolinite, halloysite, and a related mineral in clays and soils ' ' ; the latter is fine-grained, poorl.v organized material, "which may be the kaolin mineral of fireclay." In the electron microscope, it exhibits fea- tures intermediate between those of kaolinite and halloy- site, and consists of "irregular layered pai'ticles having curved surfaces. ' ' In view of these statements, it is not impossible that a continuous series may exist from kao- linite to halloysite and, indeed, Bramao et al. consider a possible genesis of clay minerals from hydrated halloy- site to the largely dehydrated form, to the disordered varietv of kaolin mineral. The question as to whether a name should be given to the disordered type of kaolin mineral should be viewed in relation to the situation existing in other layer-min- eral groups. The occurrence of ordered and disordered (or, more ordered and less ordered) forms is very com- mon and therefore it seems undesirable to attach a par- ticular name to a form which appears to be characterized mainly by its disorder. The writer advocates ' ' disordered kaolin mineral," or "h-axis disordered kaolin mineral" when a more specific description is required. The writer, in collaboration with R. H. S. Robertson and R. C. Mackenzie, has recentlj^ examined an excellent example of a 6-axis disordered kaolin mineral which under the electron microscope shows very regular hexag- onal crystals about 0.2 micron in size. An account of this work will be published soon f (see also Meldaii and Robertson, 1952). The Kaolin-Type Mifierals. The principal addition to the list of kaolin-type minerals in recent years has been the mineral chamosite (Brindlej- 1951b). This is a fine- grained hydrated ferrous silicate frequently associated t See preceding footnote. 3 Al Kaolinite x 3/2) 3 Mg 3 Fe 2 + • See also Robertson, R. H. S., Brindley. G. W., and Mackenzie. R. C, 1954, Mineralogy of liaolin clays from Pug-u, Tanganyilansiblc vermiculite, and Walker (1949) has studied similar transformations pro- duced by weatherin.ir ]U'ocesses. Thus passage from a swelling to a non-swelling mineral and vice versa seems to be rather easy, and the swelling property to be con- nected primarilv with chemical coiiiposition. Therefore, in accordance with the principles of the classification set out in table 1, these minerals must be regarded as es- sentially different chemical species rather than different structural arrangements. The range of basal spaeings, extending from lOA to 14A or more, is connected pri- marily with hydration which is extremely variable in many of these minerals. The most logical procedure is therefore to place all these minerals in the mica-type structural group, to differentiate chemical species ac- cording to composition, and to re.eard h.vdration as a ]iroperty not affecting the broad lines of the I'lassifica- tion. This perspective does not detract from tlic intrinsic interest of the swelling properties, which will now be considered. The relation of swelling to the type of interlaycr ca- tion has been discussed especially by Barshad (1950) who considers that, "The ionic radii, the valenc.v, and the total charge of the interlayered cations, as well as the nature of the iiitcrlayered substance seem to de- termine the extent of the interlaycr expansion of the mica type of crystal lattice." A useful synthesis of data is obtained by considering the interlaycr cations in order of the potential V = ne/r, where ne is the ionic charge of an interlaycr cation of valency n, and r is the ionic radius. This potential, conveniently measured b.v n/r, determines approximately the tendency of ions to form hydration shells with water molecules and therefore, under humid conditions, to expand the lattice .structure. Table 4 summarizes data by Barshad (1950) and Walker (1951) for the spaeings of, and water layers in, air-dry vermiculite and montmorillonite. It is clear that as n/r decreases, the hydration tendency diminishes and the number of water layers diminishes from two to one to zero. • The tendency of a lattice to swell owing to hydration of the ions is opposed by the eleetro.static force binding the negativelv charged silicate layers to the positively charged interlaycr ions. This force depends on the magni- tude of the layer charge, ne, (which also determines the number of interla.ver ions) ? and also on the seat of the 40 Clays axd Clay Teciixolocy [Bull. ItiO Tahle J/. Lattice .^pnciiigs of. mid iiiiiiilicrs of irnler liii/rrs in iiioiitiiioi illiiiiitr in\il rermiculite iiitli different iiiterlayer cations. Ions H* Mg*+ Ca++ Li + Ba** Na* K* NH4 + Rb' C8 + Radius, r. A 71/r 0.30 3.3 0.65 3.08 0.99 2.02 0.60 1.67 1.35 1.48 0.95 1.05 1.33 0.75 1.48 0.68 1.48 0.68 1.69 0.59 Lattice fBarshad spacings \Walker_ Number of water layers 14.33 2 14.33 14.36 2 15.07 15.0 2 12.56 12.2 1 (i) Data for vermlcuUtes 12.56 12.56 12.3 14.8 1 1.2 10.42 10.6 11.24 10.8 11.24 11.97 Lattice spacings Barshad Number of water layers 14.5 2 14.8 2 15.1 2 13.4 1 (ii) Data for montmoriUonites 12.9 11.9 1 1 12.0 1 1 M 1 12.3 1 12.9 1 Sources of (lata: Biiishad (Ifl.^n) ; Walktr (in,"il). fl]ar>;e. e.a-., whether it arises predominantly in the oc- tahedral or tetrahedral part of the lattice. It is therefore useful to consider the magnitude of the layer charge as determined by eation-exchano-e measure- ments. Some typical Yalues are set out in table 5, from which it is evident that there is no great ditt'erence be- tween illites, vermiculite, and montmorilloiioids as re- gards the charge of the layers. The question now arises as to whether the non-swellino> of such micas as muscovite and biotite is to be attributed to the greater layer charge and therefore greater elec- trostatic attraction or to the kind of interlayer ions. Barshad (1950) has shown that the .sodium mica, para- goiiite, will expand in water when tinely ground (to less than 0.5 micron), but muscovite when similarly treated does not expand. Crystal size therefore appears* to be an additional factor. Table ,5. Layer charge on mica-type minerals. Mineral Layer charge, per unit ceU 4 * 1.0-1.5 (Mackenzie. Walker and Hart. 0.9-1.4 (Barshad, 1948) 0.7-1.0 (Foster. 1951) 0.4-1.1 (Ross and Hendricks, 1943-4) Muscovite, biotite IlUtes 1949) Vermiculites Montniorillonoids Talc, pyrophyllite _ . A comparison of the data for vermiculite and mont- morillonite in table 4 indicates somewhat greater swell- ing by the latter mineral. A full interpretation of this cannot be offered but it may well arise from the smaller crystal size of the montmoriUonites. SOME THERMAL AND CHEMICAL TRANSFORMATIONS OF LAYER SILICATES CONSIDERED FROM A STRUCTURAL ASPECT Structural studies of mineral transformations are con- cerned with the nature of the new phases formed and with the processees by which the changes occur. Chem- ical and optical methods have been employed for many years in the recognition of phases and the X-ray method of identification has provided yet another method. It is particularly useful for studying fine-grained products and for intimate mixtures wliich cannot readily be sep- arated. The second and po.ssibly the more important use of X-rays lies in the possibility it provides of studying the actual transformation process itself. So far such studies have been confined to observing the orientation of a new phase in relation to the old phase. It may then be possible to arrive at a plausible picture of the mech- anism of the transformation. Thermal transformations of layer silicates are of two or possibly three kinds. Dehydration and recrystalliza- tion processes are well known, but a third may be added, namely oxidation processes. Dchi/drafion Processes. Dehydration processes may be subdivided into those concerned with release of water without change of silicate structure and those which involve a structural change. The former- processes in- clude release of adsorbed water from external surfaces, from internal surfaces (ef. montmorilloiioids and ver- miculites) and from channels in a structure (cf. atta- pulgite, zeolites). The latter processes involve the release of so-called "sti-uetural water," generally hydroxyl groups, with concomitant modification or recrystalliza- tion of the structure. A partial release of "structural water" may result in a partial change of structure, but the final release of water is generally followed by recrystallization. As ex- amples of partial release of water, the chlorites may be specially mentioned. Brindley and Ali (IDoO) first showed in detail that the hydroxide or brucite-type layer of the structure can be largel.v dehydrated without modi- fying the mica-type layer and without any major col- lapse of the usual 14A basal spacing of the structure. The X-ray reflections are characteristically modified, especially as regards the basal intensities. The mica-type layer dehydrates at a temperature usually about 100°- 150° C. higher than the first process. Chamosite provides another examjile of a structure which can be partially deliydrated. Two distinct stages of dehydration indicate the presence of hydroxyl ions with different degrees of binding in the lattice, and this is especially clear in the case of chlorites. Similar deductions can be drawn with regard to ad- sorbed water and water of hydration which is shown by the doubled, low temjierature endothermic peaks recorded in differential tlicrnial analyses of some mont- morilloiioids. irt 1 Geoi.ocy and ]Mixerai,(k;y 41 Ricrijstallizatio)! Processes. The final (Icliydratioii of a structure often leads directly to recrystallizatioii, with a structural reorn-auization. Investijratious tend to show tliat wlicu blocks ()]• units of structure of an old jihase i-an be directly built into a new jiliase. the trausfornia- tiou is rapid. Such transformations tend to develoj) by an orderly process. X-ray photop'raphs of single crystals of chlorites, for example, have shown the co-existence of the old and tlie new phases, and in this way informa- tion is obtained of the orientation relations of the jiliases (c.f. Hrindley and Ali. Ifl-jO"). C'hrysotile has bciMi similarly studied by lley and r>auuistcr (104S) and by I'.radlcy and (Jrim (in.'ilV The latter have dis- cussed peiu'rally a number of transformations of this type. Dchiidralinn and Rcrri/stiilliiation of Kaolin Minerals. Space does not permit a detailed discussion of recrystal- lization phenomena p-enerally in la.ver silicates, but reference may be made to the dehydration and recrystal- lization of the kaolin striu'ture. The practical importance of the chancres in kaolinite has stimidated many invcsti- grations (c.f. Kichardson, 19.')1) but in some respects, the residts have been surprisin<>ly discordant. Dehydration at about ooO" C leads to a disordered phase generally said to be amorphotis and reerystallization conunences aroinid 000-9.30° C. with the formation of Y-ahnuina oi- nuillite toi'cther with cristobalite. So unu'h is broadly asr-ertained, but the details seem to lack precision and dilferent workers have emjiliatically maintained different lio'uts of view as rep;ards the details. Ri-indley and Hunter n0.")2') have attempted to obtain further infor- uu'tion by studying: the polymorphic variety, nacrite, wh'ch lias the same layer structure as kaolinite. The onh- differences which can arise, therefore, between ka'>'inite and nacrite are those connected with crystal size. Von Kniiri-iug:, Brindlcy. ami Hunter (1902). work- inu' on luicrite from a newly discovered source, have found that the mineral is well-crystallized and (inter alia) that on dehydration, althoup'h most of the water is lost at about 5oO°C. some water is retained to about 750°C. This is consistent with data for kaolinite. The retention of some water to (piite hiiih tcmi)(>ratiu'es seems well established. The single crystal ])hoto!iraphs of Iicated nacrite crystals (this work is not yet published in detail") show persistence of some degree of strtictural order ahnost to 900°C. The fpiestion, therefore, must be asked whether persistence of order and retention of water can be correlated. A broad diffraction riuf;- indi- eatins' considerable disorder increases in intensitv up to 900°C. At 9.i0°C, tlie diffraction diajTram clarifies and a )iowder pattern of mnllite and some cristobalite is ob- tained.* It was hoped that by workintr with a single crystal of nacrite, a strongly orientated diagram of mull- ite would be obtained. This, however, has not emerged, although there are some small indications of a prefer- ential orientation. It still remains to work out the de- tails of these photographs, but broadly the findings are in agreement with those of Comeforo, Fischer, and Bradley (1948) who empha.sized that mullite rather * Further work by Brindly and Hunter has shown that -/-alumina occur.s as a transient phase, and it anpears to be oriented with respect to the origrinal nacrite. It still remains to elucidate the structural significance of the results. than Y-alumina is th(> essential reerystallization jiroduct. They worked witii kaolinite crystals which were ob- servetl with an electron micrt)Scope and obtained evidence of the development of mullite needles having particular orientations with respect to kaolinite cry.stals. Oxidafion Processes — Chamositc. Oxidation processes have not been studied to any large extent from a struc- tural standpoint, but Hrindley and Youell (19r)2) have completed an investigation of the transformation of the normal ferrous form of chamosite to a ferric form with accompanying dehydration. Detailed X-ray and chemical data indicate two processes of dehytlration, the first being essentially part of an oxidation ])rocess, as follows: Fe"* -^ Fe*** + e^ e-+ (oii)--^o- + n 2H+ (atmospheric) -^ 11,0 The oxidation of Fe^* to Fe*** liberates an electron, c, which attaches itself to (Oil)" forming an 0"^ ion and setting free H which is oxidized to water by atmospheric oxygen. The combined process can be written: 4Fe** + 4(0IT)- + 0-. fatmospheric) = 4Fe*** + 40- + 2II2O Thus the lattice lo.ses only hydrogen, but in a certain sense the process is partly one of dehydration since (OH)-^O--. The oxidation of Fe** to Fe*** is shown quantitatively by chemical analysis, while X-ray powder cliagraras show a shrinkage of the silicate struetui*e which is consistent with the radii of the two ions con- cerned, namely Fe+\ 0.83a ; Fe*+\ 0.67A. That atmos- pheric oxygen plays an important part in the process is proved by the fact that in vacuo or in an inert atmos- phere the reaction does not proceed in this way. Associated with the oxidation-dehydration process, is a normal dehydration of the outermost hydroxyl sheet * of the structure, proceeding as follows : 2(OH)- = H,.0-fO- An important structural aspect of the entire oxidation- dehydration reaction which takes place at about 400°C, is that only the outermost liydrox>is take part in it. Out of the initial (0H)4 of the structural formula, 3 (OH) radicals are dehydrated, but the fourth persists in the structure to a higher temperature, about 450-500°C. Here we appear to have a clear indication of difFei-ence in chemical reactivity of the hydroxyl ions within the silicate layer and those in the external sheet of each silicate layer. It is perhaps wishful tliiukiug to correlate this be- haviour of chamosite with that of kaolinite and to sup- pose that the retention of hydroxyl units by kaolinite is related to the inner positions of one-quarter of the hydroxyl radicals. Although this may well be partly related to the behaviour of the kaolin minerals, so far we lack au\*-clear quantitative explanation because the amount of water retained by kaolinite and nacrite to high temperatures is a small percentage only, of the order of 5 percent, and is not a fraction which might be given a simple structural interpretation. At present, therefore, we can go no further than to note that the release of (OH) ions as water may be related to their I.e., those which lie in the external sheet of eacli silicate layer; the expression does not refer to the external surface of a crystal. 42 Clays and Clay Technology [Bull. 169 structural environment and that in one ease, that of chaniosite, we have clear evidence for such a conclusion. Dissohition of Chlorifcs hi Acids. Another example of chemical reactivity being- related to structural en- vironment has been obtained by Brindley and Youell (1951) from a study of the acid attack on chlorites. It was found that ions such as Mg.Fe** occupying octahe- dral positions in the structure, were removed at the same rate when the amount dissolved was expressed as a frac- tion of the total amount of each ion. Al was found to be removed differently, however, and this was correlated with the fact that Al occupies both tetraliedral and octa- hedral positions in the lattice structure. Octahedral Al was found to be removed at the same rate as other octahedral ions and, in fact, on this basis the proportions of octahedral and tetrahedral Al were separately detei-- mined by a purely chemical method for the first time. The results were found to be in agreement with tlie allocation of Al to the two kinds of lattice positions from structural considerations and provided an experimental proof for the customary procedure. What is more im- portant in the present context is that, here again, there is evidence for chemical reactivity being related to struc- tural environment. CONCLUDING REMARKS In conclusion, it must be confessed that this is a very inadequate account of the structural mineralogy of clays and related minerals. The subject is now sufficiently well developed for a classification to be attempted on mod- erately detailed lines, but it is still growing and is likely to outgrow any classification which is too rigid in frame- work. It is hoped that the scheme suggested will be suffi- ciently elastic to accommodate new developments. DISCUSSION D. M. C. MacEwan: In counoctioii with tbe charge on the structure (if micaceous minerals in relation to their hehavior, some calculations have been done by Brown and Norrish (1952). They have recalculated a number of analyses of micas which were low in potash, and showed that, in many eases, much better agreement can be obtained with the expected complement of octahedral ions and the expected amount of OH" ions if one assumes that the excess water is pres- ent in the form of HsO* ions. In many analyses there is more water than expected on the basis of a normal structure. If the water is assumed to lie present in the form of HsO* ions, the analyses can be recalculated to show that there are two interlarael- lar cations per structural unit, in most cases, with HaO* partly replacing K*. This suggests that more micas than had been sup- posed are fully charged. Nevertheless, there are materials with low charge, such as Barshad has been finding, and his techuicine for determining exchangeable hydrogen directly may be very im- portant. A, Pabst: Much longer spacings than those that have been mentioned here have been reported for chrysotile (Franknchen and Schneider, 1944). The long spacings were interrupted as arising from "parallel fundamental fibril.s — hexagonally packed in cross section". The indicated diameters of the fibrils in various samples ranged from 19.T to 250 A. It may be pointed out that this is in "excellent accord with the later electron microscope work liy Bates and associates (19.50). The dimensions of the fibrils are similar to the dimensions found for halloysite tubes by Bates and co-workers to which Dr. Brindley has referred. D. M. C. MacEwan: I feel sure that you can olitaiu spacings greater than 30 A from montmorillonite in certain circumstances. Norrish, working in our Inlicu-atory. has olitiiincd spacings between .So and 1(X) A. Stephen and I have obtained a material which we think is a type of hydrated chlorite and therefoi-e half-way between mont- morillonite and chlorite. It is a mixture of lioth swelling and non- swelling chloritic material. This material gives 14 A and 2S A spacings when it contains no water, liut hydrated it gives higher spacings. We have also obtained a mixed-layer material which contains both moutmorillonitic material and the chloritic material. This gives spacings of about 30 A and 24 A. The spacings do not necessarily follow in a rational series. G. W. Brindley: I think it is becoming increasingly important to extend measure- ments into the region of long spacings. In some of the early chlorite measurements the longest spacing recorded was around 7 A and the 14 A "reflection"' was missed, and in some of the earliest kaolin measurements the 7 A reflection was similarly over- looked. Several recent investigations tend to show that spacings longer than 14 or 18 A do occur in nature. There is the work of Bradley (1950) on rectorite, which consists of a fairly regular succession of pyrophyllite and vermicnlite layers with a periodicity of about 25 A. There has also been a recent publication (Alexaniau and Wey 19.51) in which sjiacings of the order of 32 A are re- cla- quettes de montmorillonite orienti?e ; Acad. Sei., Paris, Comtes Rendns, v. 232. pp. 1855-1856. Aruja. E., 1945, An X-ray study of the crystal-structure of antigorite : Mineralog. Mag., v. 27, p. 65-74. Barshad, I., 1948, Vermicnlite and its relation to biotite as revealed by base exchange reactions. X-ray analyses, differential thermal curve and water content : Am. Mineralogist, v. 33. pp. 655-(i78. Barshad. 1.. 1950. The effect of the interlayer cations on the expansion of the mica type of crvstal lattice: Am. Mineralogist. V. 35, p. 225-238. Bates, T. F.. Hildebrand. F. .V.. .-ind Swiuctcu-d, A.. 11)50. Mor- phology and structure of endellife and hallovsite : .\ni. Mineralo- gist. V. .35. pp. 463-484. Part 1 1 Geology and Mineralogy 43 Bates, T. F., 8an(l, L. B., and Mii.U, -I. V- ]"J;''0- T_"''"''"' crvstals'of chrvsotile asbestos: Science, v. Ill, pp. ol2-ijl3 "Uradley, W. F.. 1950, The alternatinK layer seciuence of rec- torite: Am. Mineralogist, v. 35, pp. 590-.")'.»."i. Bradley. W. F., and Grim, R. E., l!).".!. llif,'li temperature thermal effects of clay and related materials: Am. MiiieraloRist, V. .St;, pp. 1S2-201. BrasK, W. L., 19.37, Atomic structure of minerals: 292 pp., U.x- foril TJuiv. Press; Cornell Lniv. Press. Hramao, I... Cady, .J. G.. Hendricks. S. I'.., and Swerdlow, M., ]!l."i2. Criteria for the characterization of k.aolinite, halloysite, and a related mineral in clays and soils : Soil Science, v. 73, p. 273-2S7. Brindlcy. G. W., 1951a, Editor, X-ray identification and crystal strnclures of clay minerals: 345 pp., l/)ndon. Mineralo^jical Soc. (Clav Minerals (iroup). Brindley. G. W., 19511i, The crystal structure of some chamo- site minerals: Miueralog. Jlag., v. 29, pp. .502-.525. Brindley, G. W., and Ali. S. Z., l!)."i(). X-ray study of thermal transformations in some magnesian chlorite minerals : Acta Crystal- liiKiaiihica, V. .3. pp. 25-30. P.riudley, (J. W.. and MacEwan, D. M. C, 1953. Structural as- pects of the mineralogy of clays and related silicates, in Green, A. T., and Stewart, G. II., Ceramics, a symposium, pp. 15-59, Stoke-on-Trent. The British Ceramic Society. Brinilley. G. W., Oughton, P.. -M.. and IJohin.son, K.,_ 19.50, polymorpiiism of the chlorites. I. Ordered structures: Acta Crystal- lographica, v. 3, pp. 4<)8-41(). P.rindley, G. W., and Robinson, K.. 194(ia. The slrucl lire of kaolinite : Mineralog. Mag., v. 27, pp. 242-2.53. Brindley, G. W., and Robinson, K., 194(;i). R.-mdoniiiess in the structures of kaolinitic clay minerals: Fnradny Soc. Trans., v. 421!. pp. 198-205. P.rindley. 0. W., and Kobin.son, K., 1947, .\ii X-ray study of some kaolinitic fire clays: British Ceramic Soc. Trans., v. 40, pp. 49-02. Brindley, G. AV.. and Robinson, K.. 1948, X-ray studies of halloy- site and metahalloysite : Mineralog. Mag., v. 28, pp. .393-400. Brindley, G. W., and Youell, R. F., 1951, A chemical determina- tion of "tetrahedral" and "octahedral"' aluminium ions in a sili- cate : Acta Crystallographica, v. 4, pp. 495-490. Brindley, G. W., and Youell, R. F., 19.53, Ferrous chaniosite :in(l ferric ch.amosite : Mineralog. Mag., v. 30. pp. .57-70. Brown, G., 1951, Nomenclature of the mica clay mineral-s, j» Brindley, G. W., Editor, X-ray identification and crystal structures of clay minerals: Chap. 5. part II, pp. 155-172, London Mincral- ogical Society (Clay Minerals Group). Brown, G., and X'orrish, K., 19.52, Hydrous micas: Mineralog. Mag., V. 29, pp. 929-932. Caillere, S.. 1951, Sepiolite, in Brindley, G. W., Editor, X-ray identification and crystal structures of clay minerals: Chap. 8, pp. 224-233, London Mincralogical Society (Clay Minerals Group). Caillere, S.. and Ilenin, S., 1951, Palygorskite — attapulgite, in Brindley, G. ^Y.. Editor, X-ray identification and cry.stal structures of clay minerals: Chap. 9, pp. 234-243, London Mineralogical So- ciety ( C^lay Minerals Group). Caillere, S.. Mathieu-Sicaud, A., and Ilenin. S.. 19.50, Xouvel essai d'identificatiou du mineral de l^i Table pres Allevard, I'alle- vardite : Soc. Francais Min^ralogie Crystallographic. I'.ull., v. 73. pp. 193-201. Caillere, S., and Hcnin, S., 1949. Transformation of minerals of the montmorillonite family into 10.\ micas: Mineralog. JIag., v. 28, pp. Goo-cn. Caillere, S., and Hc'nin, S., 1949, Experimental formation of chlorites from montmorillonite: Mineralog. Mag., v. 28, p. 012-020. Cole, W. F., Sorum, H., and Kennard, O., 1949, The crystal structures of orthoclasc and sanadinized orthoclase : Acta Crystal- lographica, V. 2, pp. 280-287. Comeforo, J. E., Fischer. R. B.. and P.radby, \V. F., 194S, Mulli- tization of kaolinite: Am. Ceramic Soc. .Tour., v. .31. pp. 2.54-2.59. Fleischer, Michael. 1947, Some problems in nomenclature in min- eralogy and inorganic chemistry : Am. Soc. Testing Materials Proc, V. 47, "pp. 1090-1108. Foster, M. L)., 1951, The importance of exchangeable magnesium and cation exchange capacity in the study of montmorillonite clays : Am. Mineralogist, v. 36, pp. 717-730. Frankuchen, I., and Schneider, M., 1944, Low angle X-ray scat- tering from chrysotiles : Am. Cheni. Soc. .Tour., v. 00. p. .500. Frondel, C, and Palache. C, 19.50, Three new pidymorphs of zinc sulfide : Am. Mineralogist, v. 35, pp. 29-42. Grim R E , and Bradley, W. F., 1951, The mica clay minerals, i„ Briniucv G. AV., Editor, X-ray identification and crystal struc- tures of vh.yy minerals: Chap. 5, part I, pp. 138-154, London. Min- eralogical Society (Clay Miner.als Group). (iruner, .7. AV., 1932a, The crystal structiu-e of kaolinite. Zeitschr. Kristallographie, Abt. A, Band S3, pp. 75-88. „ . ^ Gruner, .1. W., 1932b, The crystal structure of dickite : /eitschr. Kristallographie, Abt. A, Band S3, pp. .394-404. Gruner, .1. W., 1930, The structure and composition of greena- lite: Am. Mineralogist, V. 21, pp. 449-4.5.5. , t»ii i Hallimond, A. F., 1951a, Discussion: Clay Minerals Bull., \. 1, ''''kallimond, A. F., 1951b, Problems of the sedimentary inm ores: Yorkshire Geol. Soc. Proc, V. 28. pp. (il-CG. •, ., n Hendricks S B., 1938, The crystal structure of nacnte \hO:>- •^Si0-.-H-.O and the polymorphism of the kaolin minerals: Zeitschr. kristallograiihie. Band 100, pp. 509-518. ■ , ,, Hendricks, S. B., 1942, Lattice structure of clay_minerals aiul some properties of clay: .Tour. Geology, v. .50. pp. 2.0-2'.IO. Hendricks, S. B., and Jefferson. M. E.. 1939 Polymorphism of the micas, with optical measurements: Am. Mineralogist, v. _4, ""m-f "m'.'h.. and Bannister, F. A., 1948, A note on the thennal decomposition of crysotile: Mineralog. Mag v. 2S, pp. 3.5.^-.,... Jackson, M. L., Hseung, Y., Corey, R. B., Evans, E. J., ad Heuvel R C V, 1952, AVeathering sequence of day-size mind. Us in soils' and sediments. II. Chemical weathering of layer sdicates: Soil Sci. Soc. America Proc. v. 10, pii. 3-0. ' Knorring, O. von, Brindley. G. W., and HunteiN K., 1952, Nac- rite from Ilirvivaara, northern Karelia. Finland: Mineralog. Mag., '•^idlen^'Tc, Walker, G. F., and Hart, R., 1949, Blite occurring in decomposed granite at Ballater, Aberdeenshire: M.n- ''t^a^%: ^a'l.'^^ T^. H. S., 1952. Mor,diol^ische EinflUsse auf techni_seh^e Staubeigenschaften : Ber. dent. Keram. ^ Onsa"er L ' 1952,' IMscussion. in Summarized proceedings of a conference' on 'structures of silicate minerals-London, November 1951 • British Jour. Applied Physics, v. .., pp. -M--^- . '"orieL J. Caillere, 's., and Henin, S l^^O- Noiivel essa, de classification des chlorites : Mineralog. Mag., v. 29, pp. 32.I-.«H. p'ulin'' Linus, 1930, The structure of micas and related min- erals : Nat. Acad. Sci. Proc, v. 10, pp. 123-129 Pauling, Linus, 1930. The structure of the chlorites: Nat. Aiad. "IVa^^^lJ-L^'sT-l^fltudies of silicon carbide: Am. Min- "iSeil L: Z !nf Kohn. J. A., 1951. Disagi-eement between crvs^ sv nn^try and X-ra> diffraction data as sln,wn by a new type of •silicn carbid... lOH : Acta Crystallographica. v. 4, pp. ^Ranildell. L. S., and Kohn. J. A., 19.52, Developments in silicon ,.:,rl,ide research: Acta (Crystallographica, v. 5, pp. 21.>— 4. Uii-Inrdson H M., 1951, Phase changes which occur on heating uJ, n cav '» li imlley, G. W., Editor, X-ray identification and ;;.vstal stnu-tures of clay minerals: Chap. .3, pp. 70-8.., London, M'ineralogical Society (Clay Minerals Group). , ^ ^, , Ross C. S.. and Hendricks. S. B., 1931, Minerals of the niont- morillonit.. group, their origin and relation to sods and clays. U. S. Geol. Survey Prof. Paper 205-B, pp. 1:.1-1S0. Strunz, Hugo. 1941. Miner.alogische Tabellen ; eine Klassifizierung ,ler Miueralien auf kristallchemisches Grundlage, ^it einer Ein- fiihrung in die Kristallchemie : 308 pp., Leipzig, Akademische Verlagsgesellschaft. , Taylor J H., 19.51. In discussion on Observations of the chlor- ites 'of iron ores : Clay Minerals BiiU., v. 1, p. 137. AValker, G. F.. 1949, The decomposition of biotite m the soil. Mii.c'ralog. Mag., v. 28, pp. 693-703. , \V-dUer G F 1951, Vermiculites and some related niixed-lajo mi, enils'i,r Brindley, G. W., Editor X-ray identification and crys- tal structures of clay minerals: Chap. 7, pp. 199-223, London, Alineralogical Society (Clay Minerals Group). . , ,. , -AVeaver C. K., and Bates, T. F., 1951, Privately circulated r,.po,t Se'e AVeaver. C. E., 19.53. A classification of the 2:1 clay minerals: Am. Mineralogist, v. 38, pp. 698- (OO. \Vhite 1 L 19.51. Transformation of illite into montmorillonite: S..il Sci' Soc. America Proc. 19.50, v. 15, pp. 129-133. Weyl W. A., 1949, Surface structure and surf.ace properties ot crystals' and glasses: Am. Ceramic Soc. Jour., v. 32, p. .30.. PART II PROPERTIES OF CLAYS Page EU'ctrdclieiiiical I'nipiTtics i)t' (_'l;iys, by Laiines E. Davis 47 Ion Exchaii^'c React ions of Clays, by D. R. Lewis 54 Ails()r])ti\ r and Swclliiii;' I'l-dperties of Clay-^Yate^ System, by Isaac Bai-sliad 71) Intcrlaiuclhir Sorption by Clay ^liiicrals. by Douglas ]\I. C. ^lacEwan 78 ELECTROCHEMICAL PROPERTIES OF CLAYS By Lannes E. Davis • Introduction. The elect rochemical properties of clay resiilt from intei-aetions between eleetrieally charged liarticles, ordinary ions, and molecules of solvent. Floe- eulation and allied phenomena represent the effect of ions and solvent upon the particles. Cataphoresis and related effects are caused by interaction between particles and solvent, althoiifili ions may play an important part in prdduciufr tlie iihenomena. The fabric of the clay and the electrical fields surroundin 53 SELECTED REFERENCES Colciiuui. N. '1'., Williams, D. 10., Nielsen, T. IJ.. and .leniiy, 11., in.")l. On the valiilit.v of interpretations of potent ionietricallj' measured soil jiH : Soil Sci. Soe. America I'roc. li)r>0. v. 15, pp. llltl-114. Dide. llaKolin, I'.Ml. The glass eleclrodi'i nielhnds. application, .ind theor.v : 'X'.2 pp., New York, .John Wile.v and Sons, Inc. lOriksson. E., 10.~)1, The sisnificance of pH, ion activities and iniMiilirane pnientials in colloid systems: Science, v. 1!.".. pp. 41S- 41^0. (iurney, U. W., 1!),TJ, The (pianlnm mechanics i>f elect mcheni- istry : Royal Society London I'roc., v. 18(i, jip. ;!78-3!H. .lenny, H., Xielson, T. R., Coleman, .\. T., and Williams, D. K., 1II.")U, Concerning the measurement of pll, ion activities, and mem- luanc potentials in colloid systems: Science, v. 112, pp. 1(>4-I(i7. Marshall, C. E,, 1948, The electrochemical properties of mineral Miemhranes. VIII. The theory of .selective membrane behavior: .lour. Phys. Colloid Chemistry, v. .")2, pp. 12,'>4-!)."). (Includes bibliog- raphy of relevant papers by Marshall and co-workers.) .Meyer, K. II., and Sievers, .l.-F., lit.'Stia, La permeabilitc des membranes. I. Th^orie de la perm(^abilit6 ionique: Helvetica Chimica Acta, V. j;i. pp. (i49-6C4. .Mever. K. II., and Sii-vers, .1.-1'., ISKilib, I.a permeabilitc des nieml)raues. II. Es.sais avec des membranes .s<''lectives artificielles : Helvetica Chimica Acta, v, 19, pp. (>U."i-G77. Jleyer, K. II., and Sievers, J,-F., 193(jc, La permC'abilit*^ des membranes. Analyse de la structure de membranes v(''gelales et animaies: IIelvetic:i Chimica Ada, v. 19, pp. 9S7-99ri. Scratchard, (leorge. 19.").'}, Ion exchange electrodes: .loui'. Am. Chem. Soc., v. 7~k pp. JS,S.**,-0,v.'{,s. Sta\'erman, A. .J., 19,"t2, Non-e(iuilibriuin thermndynamics of mem- brani' processes: Faraday Soc. Trans., v. 4.S, p|i. 1T('>-I,s."i. Teorell, T., 198"), An attempt to formulate a fiuantilal ive theory of membrane permeability : Soc. Experimental liiol. .Medicim- I'roc. V. :W, pp. L'S2-L'S.-,. Teorell, T., 19.'!7, Discussion, in .Meyer, K. II., The origin of bioelectric phemimena : Faraday Soc. Trans., v. 33, pp. MCh't-^DTui. Wyllie, .M. R. .1.. and I'atnode, W. II., 1950, The laty minerals and, accordingly, a different distribu- tion of the charges on the surface ions. In attapulgite itself a small amount of the silicon is frequently replaced by aluminum ions which give rise to the charge deficiency causing the ion exi-hauge activity of attapulgite (Mar- shall, 1949). Because of its fibrous structure and the pi-esence of channels jjarallel to the long axis of the crystals in which many of the mobile exchange ions are found, the rate of the ion exchange reaction in attapul- gite minerals may be much slower than in platy minerals. This would be expected if the ions along the channel must diffuse into the solution phase to reach an equilib- rium. Illitc Group. The illite group of clay minerals are small particle size, plate-shaped clay minerals distin- guished by their ability to fix potassium irreversibly. The ion exchange activity for the illites is attributed to iso- niorphous substitution occurring largely in the surface tetrahedral silica layers. This gives rise both to more favorable geometric configuration for microscopic coun- ter-balancing of the unbalance in electrical charge and also to the possibility of formation of co-valent linkages. Either coiulition is likely to prodiure an irreversible re- action. MonimoriUoniir Group. The most active clay group in terms of amount of ion exchange reactivity per unit weight of clay is the montmorillonite family. The high degree of their base exchange capacity and the rapidity of their reactions have long been recognized as outstand- ing attributes of this class of clay minerals. Minerals of this group are plate shaped, three-layei- lattice minerals with a very high degree of isomoriihous substitution, distributed both in the octahedral i)ositions in which chiefly magnesium substitutes for aluminum, and in the tetrahedral coordination in which predominantly alumi- num substitutes for silicon (Harry, 19.50; Hendricks, 1945; Ross and Hendricks, 1945). Because of both the large base-exehauge capacity and the widespread occur- rence and economic importance of this group of min- erals, a great deal of the experimental work has been done (Hauser, 1951). As there are these marked differences in the structure both geometrically and in electrical charge density of the principal groups of clay minerals, there will be large variations in the relative contributions of reversible ion exchange reactions, the degree of amphoteric nature of the claj' minerals and physical adsorption to the equi- librium disti'ibution of ions in an acpicous clay-electro- lyte system. EXPERIMENTAL TECHNIQUES Methods of Preparing Ilydroyen-Clan. Although this discussion is more directly concerned with the inter- pretation of the data having to do with ion exchange properties of clays than with the determination of the exchange properties themselves, the usefulness of the data is frequently affected considerably by the exact details of the method of determination of the exchange properties, and, accoi'dingly, some attention must be given to the limitations of various techniques. One group of teehni(|nes which are commonly employed involves the preparation of the hydrogen form of the clay either by dialysis or electrodialysis or by direct action of a solution of a mineral acid. The acid form of the clay is then treated with the base of the desired salt form and the equilibrium distribution determined from the degree of conversion (often measured by the change in pH of the suspension system), or the inflection in the titration curve is used to determine the total exchange eapacity. The difficulties of interpretation of the titration curves of acid clays by either inorganic or organic bases are widely recognized (Marshall. 1949; Jlitra and Rajago- palan, 1948; 1948a; Mukherjee and Mitra. 1946). In the first place, there is no general agreement about the nature of the exchange titration curve. The results of various researchers have varied from the production of definitely diprotic (Slabaugh and Culbertson, 1951) character in the titration curves to curves which have a very broad inflection or none at all and in which the establishment of an end-point corresponding to the com- pletion of a reaction is very difficult even to estimate. Some investigators have titrated to an arbitrary pH which they considered to be an end-point for the reac- tion, assuming that the distribution of proton activity of all the clays in the samples being titrated is the same, and that lep-itimate and reproducible conditions for mea- suring cell potentials in suspension are established in each suspension. The colloidal nature of the system com- plicates both the measui-enient of potentials and the in- terpretation of the potentials in terms of hydrogen ion activities (Mysels, 1951). Moreover, the anomalous be- havior of the hydrogen ion in its reactions with clays has long been known, and recently the behavior of hydro- gen ions in ion exchange reactions of clays has been found to exhibit a pattern that suggests that these ions are held to many clays partly by covalent bonds (Krish- namoorthy and Overstreet, 1950, 1950b). It is likely that studies of the equilibrium distribution ions on clays should not involve the preparation of the hydrogen form as a necessary step (Glaeser, 1946a; Vendl and Hung, 1943). A great deal of useful information concerning the polyeleetrolyte nature of the clays can probably be de- rived ultimately from the studies on the titration be- havior of the hj-drogen form of the clays, but such information is not a necessary and integral part of the study of the exchange behavior of the clays. Method for Preparincj Ammonium-Claii. The most satisfactory experimental technique to employ in a given set of experiments will depend to some extent on the intention of the application of the data. For example, 56 Clays and Clay Techxology [Bull. 169 for the determination of the total exchano-e capacity of the clay minerals, a variety of satisfactory procedures employing' either ammonium acetate or ammonium chlo- ride solutions neutralized with ammonium lij'droxide have been described which differ only in the details of the preparation and manipulation of the sample. (Bray, 1042 ; Glaeser. 1946 ; Graham and Sullivan. 1938 ; Lewis. 1951). The ammonium ions retained by the clay may either be determined directly on the clay or eluted and determined separately. For the determination of the total exchange capacity of a number of clays the nse of an ammonium-form ion exchange resin of suitable character has proved verv satisfactorv (Lewis, 1952; AViklander, 1949, 1951). Experimental Techniques. Experimental technifines may be adapted to micro ((uantities of clay, or methods may be used that permit the colorimetrie determination of the exchange cations rapidly and easily, if less accu- rately. If the equilibrium distribution of ions between a clay jjhase and a solution phase is to be determined, the most direct method involves placing a clay with a known ion population in an electrolyte solntion of known composition. After a .suitable length of time both phases of this system are analyzed to determine the distribu- tion of ions at eqnilibriuiii. This method, so direct in principle, is replete with pitfalls. It may be convenient to analyze chemically only the solution phase before and after the reaction, to determine the distribution of ions accomplished by the exchanger phase, thus reciuiring that the analytical procedure be very accurate in dete rmin- ing a small difference between two large numbers. More- over, because the equilibrium water content of the clay depends strongly on its ion form, the concentration of the external solution changes as the ionic composition of the clay changes, and the degree of exclusion of molecular salts by the Donnan mechanism from the hj-drated clay changes as the ionic form of the clay changes. The effect of the change in equilibrium water "ConfeTvrwith change in the ionic form of the exchanger may be so great that failure to consider it may so dis- tort the results that ion exchange in its ordinary sense does not appear to take place (Lowen, Stoener, and Argersinger, 1951). For the most accurate equilibrium determination, the solntion phase and the exchanger phase should be physically separated in a manner which does not disturb the ionic equilibrium already estab- lished. For accurate work it is desirable to bring the clay to ecpiilibrium with a given composition of elec- trolyte solution, separate the clay phase and repeatedly bring the clay to equilibrium with successive portions of the same solution. In this way the composition of the electrolyte solution is not altered by the contribu- tion of the displaced cations from the exchanger phase, so that the composition of the equilibi'ium solution phase may be determined accurately either from an accurately prepared composition of the equilibriiuii solution or an accurate analysis of the initial solution. The day phase should finally be separated and analyzed directly for the distribution of the ions participating in the exchange reaction or the exchanging pair disi)laced by a third cation and analyzed in the elution product. The direct experimental determination of equilibrium ionic distri- bution can be successful for .studies of ion exchange if careful attention is paid to the details of the experiment, with suitable attention to analytical accuracy and pi'oper mani))nlation of the sample, so that the final data pro- vide an accurate picture of the equilibrium partition of electrolyte ions between the solution phase and the ex- changer phase at ecjuilibrium. Clay Chromatographic Methods. A modification of the column chromatographic technique has been used recently in determining the exchange isotherms for clays. This techni(iue involves the preparation of a column consisting of the clay in an inert matrix (asbestos) that provides suitable flow properties for the column. The exchange isotherm is obtained by measuring the compo- sition of the solution passing through the column as one exchange cation on the clay is displaced by another. This technique in principle possesses the virtues of greatly reducing the amount of analytical work required and of having inherent in the process the separation of the clay and electrolyte phases. If radioactive isotopes are used as tracers for following changes in composition of the eluted solution, the whole process can be put on an essentially automatic basis. The recently reported deter- mination of the cesium-sodimn isotherm at room temper- ature on a montmorillonite from Chambers, Arizona, (API 23) indicated considerable promise for this tech- nique with the clay minerals (Faucher, Southworth, and Thomas, 1952). The colloidal character of the clay min- erals, however, may cause mechanical difficulties in the preparation of suitable chromatographic columns, unless the columns are always operated with solutions having relatively high ionic strengths. Clay-Re.iin Reaction Methods. For the detei-mination of the ionic distribution on clay particles at low solution concentrations, monofunctional sulfonic acid resins may be Tised by bringing an electrolyte solution and resin to eciuilibrium with the clay. After equilibrium is reached, it is possible from only a material balance and an analy- sis of the washed resin phase to determine the e(|ui- librium distribution of ions on the clay equilibrium with the electrolyte solution. It has been demonstrated that the distribution of ions between a clay and a solution is independent of the presence of the exchanging resin. EXPERIMENTAL CONSIDERATIONS There are two major classes of objectives in the exam- ination of the data which are obtained in the study of ion exchange reactions. The first of these requires only that sufficient data be accumulated so that a working ecjuation or graph can describe the data and permit interijolation and extrapolation of the behavior of this system to conditions not precisely covered by the experi- ments. This method permits considerable latitude in the type of parameters and the manner of the mathematical combination to provide a description of the actual be- havior of the particular process. "With such a description the behavior of the distribution of calcium ions and sodium ions, for example, on a specified clay, could be summarized at the temperature and solution strength of the experiments over a relatively wide range of com- positions of the exchanger and solution phases. Such descriptions of behavior serve a useful practical purpose. On the other hand, such descriptions in themselves pro- Part III Properties of Clays vide no dues wliiuh sufifrest eitlier the iiiao;nitvule or direction of ehan-res in selectivity of sodium with respect to eah-iuni as tlio temperature, the total streujrth of the solution, or the iniiieral s|iecies should cliauu-e. The other ohjeclive is that of estahlishinj;- a sound tlieoretieal basis for undei-standinjr tlie ditferent selectivities of the vari- ous ions when reacting' with ditfei-ent exehanjier jihases. The niatheniatieal exjiression of these theoretical views wouhl ])rovide not only a description of the proee.ss, but also a basis for iirediction of chanj-es in the nature of the distribution with chanjies in a wide variety of param- eters which enter either exi)licitly or implicitly into tile e(iuations. (iood expermental data (iljtaiiied from well-diaracterized solutions of electrolytes interacting: with well-defined clay mineral species are necessary for either of these considerations. At present there is a jrreat need for more experimental information on tlie ion ex- chaufre behavior of clays under circumstances which permit the examination of the data with a view to test- in;.;' various hypotheses and theories which have been offered as a basis for the ionic selectivity in ion ex- el-angers. Although the nature of the experimental work which is needed from both practical and theoretical .standpoints in the study of ion exchange of clays was clearly pointed out by IJray in 11142, the present need for tiiese data in clay systems is as great as it was at that time. Both the theoretical and the exjierimental studies designed to establish the contribution of the several conceivable parameters to the actual selectivity of an exchanger for ions in solution has proceeded at a greatly accelerated rate in systems involving synthetic organic resin exchangers. The intensity of activity in the investigation of clay syst<'ms is increasing at ])resent. Those asjiects which Bray |)ointed out as much-needed extensi()ns of the experimental ett'ort involve leaving the convenient range of ion distributions from the stand- point of analytical technicpies in general and extending these studies to very wide ranges of composition of the exchanger phase and over wide ranges of total concen- trations of the solution as well. While both of these directions are now being actively pursued by investiga- tors of resin-electrolyte systems, similar progress has not been maile in clay investigations. Anotlier asjjcct on which Bray felt that considerably more work should be done is that of greatly increasing the number of dif- ferent ions present in a system. From a practical stand- point. ])articularly in connection witli soils, the need for suc'li inxestigations is luidoubtcdly great. From the standpoint of theory, however, our knowledge of the sjiecific interactions between ions in solution and in the exchanger phase is much too inadeciuate to enable us to apply this information theoretically at present. In liis recent review of the theoretical progress being made in the elucidation of the mechanism of ion ex- change reactions. Boyd summarized the curi-ent status of ion exchange equilibrium theory as being somewhat confused and the disagreements in the literature far more numerous than agreements (1951). This sentiment echoed the conclusions expressed by llai-shall in his dis- cussion of the ion exchange reactions of clays when he reported that the only certain conclusion one can draw at present is that better experiments are needed (1949). The various approaches which are presently being made to establish the i)rincii)al mechanisms bv which the solid exchanger phase controls the distribution of exchange- able cations among its available ion exchange sites when in eciuilibrium witii a solution of a given comjjosition may be classified into several broad groups. The ion ex- change equilibrium has been considered (1) as a class of reversible double-decomposition reaction which may be described by the principles of the law of mass action, (2) as an ionic adsorption reaction the behavior of which may be described by a suitable isotherm equation for a mixture of electrolytes, (3) as a (iibbs-Donnan distribution between two jihases, and (4) as reflecting the behavior of solution ions under the influence of a heteropolar ionic solid surface. Most invest igatoi-s have preferred either the ma.ss action or the adsorption de- scription of the exchange process. . In general, there are a number of changes which ac- company the i-edistribution of ions in the ion exchange reaction. These variables must be considei-cd when de- signing experiments to test the various iiyi)otheses of the equilibrium distribution of ions in ion exchange re- actions. They include the following processes which com- pete with tlic exchange reaction or accompany it: .4. Ion-pail- formation l)et\veeu solution ion.s and exchangers. li. Molecular adsorption of partiall.v dissociated electrolytes. C. Complex ion formation in solution. D. Change in distrihiition of ion species with chtniKes in con- centration of electrol.vtes. Ill addition to these processes, the solution concentra- tion and composition may change during the ion ex- change reaction because of the following factors which must be evaluated to permit calculation of the equilib- rium distribution : .1. Variation of ei|nililprinin w.-iter content of excjian^er with change in ion composition. B. Change in solution volume resulting from exchange of electrolvti's haviui; ilifferent p.irtial miliar volumes. MASS-ACTION DESCRIPTION OF ION EXCHANGE REACTION If we consider a reversible reaction of the following form between monovalent cations ^1* and B* in solution and an exchange phase Z, A* + BZ :f± B* + AZ (1) the law of mass-action describes the equilibrium distri- bution in terms of a product (B*) (AZ) K = (2) {A*) (BZ) In this expression the quantities in parentheses represent the activities of the various species. The activity of each species is a quantity »,, such that [li = H° + RT In tti (3) where fi, is the chemical potential of the species i, \i° its chemical potential in some arbitrary standard state, B the universal gas constant and T the absolute temper- ature. If the ion exchange reaction is truly reversible and if the activities o,= (^J can be evaluated, at constant temperature and pressure, the con.stant K can be calcu- lated and the Gibbs free energj^ of the reaction computed from its value. 58 Clays and Clay Technology [Bull. 169 As a first approximation, the concentration of the ions in solution and in the exchanger phase liave been sub- stituted for the activities. In this form, the value of A' is a mass law concentration product which is not ex- pected to remain constant. For practical purposes, a closely related quantity, the selectivity coefficient D is frequently calculated as (AZ/BZ) D = (4) {A*/B*) The type of variation of equilibrium mass law product is illustrated in figure 1. Both the mass law concentra- tion product and the selectivity coefficient are without direct theoretical utility themselves, although they are useful working quantities which ditfer from the thermo- dynamic quantities by suitable functions to convert the concentrations of ions to activities. The evaluation of the activities in both the solution and exchanger phases, however, involves several uncertainties at the present stage of our knowledge of these reactions. A number of approaches have been employed to evalu- ate the activities of the ions which are reacting both in the solution phase and the clay phase. For the solution phase the basic data required are the activity coefficients of the electrolytes in mixed ion solutions over the con- centration and composition ranges employed in the reac- tions. In general, this information is not available, al- though Earned and Owen (1950) have summarized the available data and some rules for computing estimates of activities of electrolytes. The approximation is fre- quently made that the ion activity is that of the single electrolyte at a total ionic strength of the reacting mixed solution. There is the possible objection to all these methods that the activity of the dilute mixed electrolyte solution may not be the correct activity to use on the LOG Cb»/C,. Figure 1. Viirintinn in m.ass-lnw product. grounds that the exchange reaction occurs only in the immediate vicinity of the highlj^ ionic crystalline clay exchanger, where its activity would be expected to be significantly difl'erent from that in the dilute solution both because of the change of dielectric constant of the 10 06 - 0.4 — r- LiCI.^^ ^^ - BiClj 1 1 1 1 1 075 100 I 25 150 FlGUKE 2. Activity coefficients for O.ol m HCl in electrolyte solutions. solvent and the potential energy of the ion in this en- vironment (Davis and Rideal. 1948; Greyhume, 1951; Grimley, 1950; "Weyl, 1950). Since the over-all process is the transfer of ions from the dilute solution to the exchanger, however, and since at equilibrium the chem- ical potential of ions of any species is the same through- out the system, the solution thermodynamic activities should be suitable when they are known. The equilibrium constant for the reaction (1) can be written for the mono-monovalent exchange as m,(B) Y. (-B) {AZ) 'm,(A) Y. {A) (BZ) (5) where m^ (B) is the main ionic molality of the cation B* with the solution anion, and y= i^) is the mean ionic activity coefficient for this electrolyte. These quantities are defined in terms of the molalities of the cation m^. and anion »)_ as mj' = m^^' -{- mJ-'~ (6) In this expression v^ is the valence of the cation, v. the valence of the anion and V = ■!'+ -j- V- (7) Analogously, the mean ionic aetiAity coefficients are y," = y/ + 7-"- (8) The mean ionic activities of ions are influenced by the presence of dissimilar ions. The values of the mean ionic activity coefficients for electrolytes have been determined by emf measurements in suitable cells. The effect of Part II] Pkoperties of Clays 59 ATTAPULGfTE AC COMPANT PROCESSED 45 NORMALITY OF EQUlLtBRtUM NH^ OAc Figure 3. Kxchange isotherms for clay samples. added electrolj'te is shown in figure 2, in -which the mean ionic activity coefficients and hydrochloi'ic acid in the presence of LiCl and BaCl2 are plotted as a function of the total solution concentration. As this figure illustrates, the magnitude of the activitj' correction is significant at relatively low concentration levels and is material^ affected by the nature of the accompanying ions. Another more direct manner of evaluating the ac- tivity of the various ions in solution is to determine them by emf measurements involving reversible elec- trodes (Kressman and Kitchener, 1949a) or membrane electrodes (Barber and Marshall, 1951; Clarke, Marin- sky, Juda, Rosenberg, and Alexander, 1952 ; Marshall, 1949; ilarshall and Barber, 1949) in the exchange solu- tions. The application of the electrode or membrane measurements to systems containing suspended colloids, however, raises other problems (Mysels, 1951). It attempting to evaluate the activity of the exchanger, earlier approaches cousidered the exchanger as an ideal binary solid solution in whicli the activities of the ions in the exchanger were proportional to their molal frac- tions (Boyd, Schubert, and Adams, 1947). This might be satisfactory if the exchange ions are monovalent and similar in size. Ivielland (1935) has pointed out that the assumption of an ideal solid solution is probably not justified and has suggested a method for estimating ac- tivity coefficients for the solid exchanger phase based on the Margules equation, from which the activity' eoefS- cients are calciilated as and log Y. CX^z log ynz = CX„x (9) where Xaz is the molal fraction of the ion A in the ex- changer, and C is a constant value for each particular system and may be positive or negative most commonly varying- from zero to one. On the basis of these considerations, estimates may be made of the apparent activitj- in the exchanger phase, after suitable exchange data are available to permit evaluation of the parameters C in equations. Recently, however, it has been pointed out that the constancy of K o d- - ^ o K °,0 O o ye MONTMOR'LLONITE / LORENfl NEAR FQREST. MISSISSIPPI. o / HUSBAND MINE. 20 ^ 6 - / < o / /o ■^ 6 -/ o / £ / z < 5 K " .2 UJ <. m 1 1 1 1 ] 1 Figure 4. NORMALITY OF EQUILIBRIUM NH, OAc Exchange isotherms for elay samples. in equation (2) is a thermodynamic necessity (Arger- singer, et al., 1950; Bonner, et al.. 1952; Ekedahl, et al., 1950; Hcigfeldt, et al, 1950, 1950a) arising from the definition of activities and independent of the mechanism of the reaction. Accordingly, the expression for the equi- librium constant is combined with the Gibbs-Diihem equation to determine from the experimental data, when the activities of the ions in the mixed electrolyte solu- tions are known, both the true equilibrium constant and the activity coefficients for the ions in the solid phase. The Gibbs-Duhem equation at constant temperature and pressure for a binary sj^stem is ni(7 ^i] + 'n-id \\.2 = (10) In this expression nx is the mole fraction and /xi the chemical potential of the ion species i. Combining this with the definition of activity, equation (3) gives n^z d In a iz* ftnz d In anz = (11) which may be shown to be equal to n^zdlnf Az*nBzdlnf Bz = (12) Combining this expression with the mass-action equation (2), and integrating, with the boundary conditions Iaz -^ 1 as n.iz -^ 1 and fnz -^ 1 as rinz -^ 1 gives and In f IX = — iinz In K^ + I ,- in K^ dn^, J -' AZ In fisx = n_iz ?« A'' CN,z J In K^ dnj (13a) (13b) The substitution of these results into the logarithmic form of the equilibrium equation yields i In K^ d «4z J In K (14) The apparent equilibrium constant A'' is the quantity which is obtained from the experiment coupled with suitable values for the solution activity coefficients. By graphical integration of a plot of In K^ versus n,iz the In K and the true equilibrium constant for the reaction may be calculated. This procedure has not previously 60 Ci.AYs AXD Clay Techxology [Bull. 169 been applied to clay exchange systems, but lias been applied with internal consistency to the sodium liydi'o- p-en exchanue on Dowex-oO (Lowen et al., lOol), although the values of the e()uilibrium constants were somewhat different from those reported by other workers. (Bauman and Eichhorn, 1947 ; Duncan and Lister, 1949 ; Marin- sky, no date). A separate explanation has been suggested to account for the variation of the apparent equilibrium constant in addition to the assumption of non-ideality of the solid solution model of the exchanger (Marinskj-, no date). This view proposes that the reaction does not progress from the initial to final product in the straightforward manner postulated in equation 1. Rather the reaction is assumed to proceed through a number of mixed inter- mediate compounds each of which is in equilibrium with its products. By a suitable choice of steps in the reaction and selection of the proper values for the assumed equi- libria, the deviation from constancy of the experimental equilibrium constants can be fitted. The existence of the hypothetical intermediate compounds, however, would be very difKcult to establish, and any curve can be fitted by the combination of enough variables. LANGMUIR'S ADSORPTION ISOTHERM DESCRIPTION OF ION EXCHANGE REACTIONS The first attempts to describe the ion exchange be- havior on a rational basis were in terms of the Freund- lich isotherm. This was a very reasonable approach to the problem of describing the exchange behavior and many workers employed a form of the equation. kc^/P (15) in wliich x amount of the exchanging cation is taken up by m grams of exchange when immersed in an equilib- rium concentration c of electrolyte. Both I- and /) are constants. The calculation of the exchange behavior iu a form resembling adsorption isotherms has been com- mon practice (see figs. 3 and 4). The earliest mathematical descriptions of the process did not approach a finite limit corresponding to the total capacity of the exchanger, so that a revised form was offered by Wieguer and Jenny (1951) : i/p (16) where the additional quantity a is the amount of electro- lyte added. Another essentially empirical equation which re- sembles the Preundlieli isotlierm is that of Vageler X ~ = k m c c — a sa y — a -\- c (17) in which y is the amount taken up per gram of ex- changer, s the saturation capacity of the exchanger, a the number of equivalents of salt per unit weight of exchanger initially added, and c is the " half -value, " that is, the value of a at which half of s is exchanged. The hyperbolic form of the Langmuir adsorption iso- therm provided an anal.vtical form which represented the observed degree of exchange rather closely. This analogy has been most successfully exploited by Boyd, Schubert, and Adamson (1947), who expressed the Lang- muir ad.sorption isotherm for ion exchange reactions between two monovalent cations A* and B' as kh^Ci* = (18) ' 1 + IhC'c + l.Cu* in which (.r/>H).4, is the amount of A* taken up per unit weight of exchanger, C^^ and Cn. the equilibrium concen- trations of the two ion species and k. hx and ^2 are constants. If the assumption is made that the number of un- occupied exchange sites is small at any moment, the quantity unity in the denominator of eijuation (18) may be neglected in comparison with 6iC.4^ + h-jCu^- This approximation permits the equation to be written in the linear form Ca*/C„* ho 1 f.4* (.T/m)j* i,k k Cb* (19) This form permits a convenient means for plotting the exchange data, and has an added virtue since bi and bo are related to the energies of adsorption of the ions, Ea^ and £■„,. A plot of the quantity on the left (r.u/f bJ/ {x/m)A+ versus CA+fCnt bas been found to give a linear plot in .some systems, permitting the calculation of k and &2 /bi. Another description of the exchange reactions which is closel.y related to the adsorption isotherm equations was initially proposed by Jenny (1936) who derived on the basis of probability the distribution of ions near the interface of the exchanger phase and in the bulk solution on the assumption that different species of ions occupied ditferent oscillation volumes. For ions having the same oscillation volumes this expression reduces to the same form as Vageler 's equation. This equation of Jenny was the first direct approach to the descrijition of ionic dis- tributions on a statistical basis. The statistical method, however, has been extended by Davis (1950, 1950a. and Rible, 1950) who applied the method of Guggenheim (1944) to the ion distribution between the solution and exchanger. The statistical approach was originally de- veloped to describe the mixed adsorption of two species of neutral molecules from solution under conditions of zero energy of mixing. To apply this approach to ion exchange, several assumptions had to be made in addition to those iu the original derivation for molecular adsorp- tion. The most important of these are the assumptions that ions of the same sign do not display specific inter- actions, that all interactions between ions may be neg- lected except for nearest neighbors, and that the internal degrees of freedom of each ion are independent of its neighbors. The mathematical statement of the ionic distribution which is derived on the basis of the statistics and as- sumptions outlined above is of the form (A)^i. [By-' {%qi[i]) '•■--'■''' k = (20) (B)'--- [A]'-" Part 1 1 Properties of Ci.av; 61 Here the (luaiitities in brackets [] refer to number of moles in the cxelianfrer phase while the quantities in ]iarentheses () refer to the activities in the solution pliase. r, is the valence of species i. and r/, is a coefiicieut such that {Z 2) (r,_<7..)=r, — 1, (21) where Z is an intefrer related to the geometry of the ex- ehan t and g moles of solvent from i—*o, then ^F° =BTlMK, = — ■,^[v,:■ {f-g)Vo\ (28) where Vi is the molar volume of species i and i'„ is the molar volume of the solvent. Since the quantit.y [vjt — Ci — (/ — f/)i'o] is the change in volume of the ex- changing system this can be measured directly. It is significant to note that all of the parameters in this ex- pression can be evaluated independently of the exchange jji-ocess itself. In this sense it is unique since all of the other methods of describing the reaction involve un- known values which can only be defined in tei-ms of the exchange process. 62 Clays and Clay Technology [Bull. 169 o® ° ®^. o „o @ cm ® PROCESSES IN AQUEOUS ® ^ O ^ CLAY - ELECTROLYTE INTERACTION 0(-?- 0^0- • o • ]• • o o® ION EXCHANGE WORK OF SWELLING WATER TRANSFER OONNAN EXCLUSION OF ANIONS COMPLEX ION FORMATION ^ 0° • o ^^ - — • — - o • o m^o 0| - ^0 ♦ INCOMPLETE DISSOCIATION ION PAIR FORMATION • o Jl o • O0 o • - _ _ - MOLECULAR ADSORPTION " ° o®o °0 CHANGE IN SOLUTION VOLUME c@ • CATION A CATION B ANIONS FiGiRf; (). Schematic summary nf processes in aqueous clay-electro- lyte interaction. It has been reported by Glueckauf (1949) that the linear relation equation (25) is not valid for resin ex- e-liangers, but no studies of this aspect have been made using swelling- clays. The neglect of activity coefficients of the ions in the exchanger phase is also an oversimpli- fication which will need to be refined as our understand- ing of ion exchange processes grows. For the highly colloidal clay systems, moreover, it is difficult to erect an arbitrary surface in the neighbor- hood of tlie particle such that all ions beyond this sur- face are said to be "outside" the exchanger and those within are said to be "inside." Marshall (1949) has in- dicated some of these complexities and concludes from them that the application of the Donnan principle to clay systems may involve experimentally inaccessible quantities. SUMMARY AND CONCLUSIONS The ion-exchange process is only one of a group of reactions that may occur between a clay and an electro- lyte solution. Figure 6 presents a schematic summary of the processes that take place. For practical purposes it is necessary to know what will be the final result of all of these several processes in any given situation. The reso- lution of the contributions of these several processes requires that the study of each aspect be initially carried out separated from the competing reactions as much as possible. Accordingly, the investigation of the ion ex- change reactions of clay minerals should be made under circumstances in which competing processes make either a negligible contribution or one which can be independ- ently evaluated. The nature of the clay mineral should be known in as complete detail as the techniques for establishing struc- ture permit. As any specific clay material is a member of a structural series, it is important to establish the origin and nature of the exchange sites as fully as pos- sible. For example, the distribution of the isomorphous substitution between the tetrahedral and octahedral co- ordinations is important because the difference in the energies of these two classes of exchange sites would be expected to influence the ionic selectivity, associated with each type. As far as is possible, the distribution of metal ions either as exchange cations or as lattice units should be established. This is especially important for such metals as magnesium and aluminum which play a cen- tral role in the exchange properties of the lattice (Fos- ter, 1951). As Kelley (1948) has pointed out, the ion-exchange properties of members of the major classes of clay min- erals will vary considerably so that it is not to be ex- pected that all montmorillonites or illites or kaolinites can be fitted bj' a type exchange isotherm but it is rather to be hoped that the developing knowledge of mineral structure will ultimately permit us to use the number and distribution of exchange sites in a lattice to predict with useful accuracy the exchange reaction properties of a clay mineral. The electrolyte solution introduces another group of variables. Much more must be known about the specific ion interactions which give rise to the changes in ionic activities with solution composition in mixed electrolytes. This is one aspect of the .solution thermodynamics which must be faced squarely, since all ion exchange reactions with the exception of isotopic exchanges take place in mixed ion systems. More needs to be learned about the change in ionic species distribution with concentration and composition for incompletely dissociating salts and for ions which may dissociate stepwise. For the same direct reason — that equilibria cannot be described ther- modynamically unless the distribution of reacting species is known at equilibrium — the use of ions which may form covalent bonds with the exchanger, ions which react chemically with the clay or the exchange ions displaced from the clay, or ions which form comjilexes with the solution anions all introduce uncertainties which have contributed to the compounded confusion which currently exists in the literature of ion exchange. The transfer of solvent in the exchange reactions of the swelling clays, and the change of volume of the solu- tion iihase which may result from differences in partial molal volumes of the displacing and displaced ions demand that experiments be designed so that these con- tributions may be evaluated directly. Neglect of these contributions has led some investigators to the extreme view of concluding that ion exchange was not an ion- equivalent process. Future experimental work can now be planned in such a manner that the clay exchanger phase and the solution composition can be determined completely enough to give permanent significance to the data. It is to be expected that when the actual behavior of well-charac- terized ion-exchanging clay systems is known quantita- tively, the mechanism of ionic selectivity can be established and the exchange behavior of clays will be capable of prediction to a useful degree on the basis of general thermodynamic considerations. DISCUSSION W. J. Weiss: Wliat forces hold the non-exchangeable cations on clays and cause differentiation between exchangeable and non-exchangeable forms, and how do these non-exchangeable cations affect the physical properties of the clays? Part Propertiks of Ci.ays 63 E. C. Henry: I'lTliaps someone else will answer this i|iiestion directly, but it seems worthwhile at this point to call attention to the work of Fajans and of Weyl and associates on the polarizaliility of ions on surfaces. Weyl (10.")()a) has interpreted results obtained with crushed [lottery flint as follows: The silicon ions at the surface of this material :ire not surrounded by four oxygen ions, as is the case in the interior of quartz. Consequently, the charges of the surface silicon ions are n(tt eoniiiletely neutralized. The result is a force field surrounding the silicon ions. When quartz is crushed man.v silicon ions become exposed, and upon adding water the vinsatisfied charges on these silicons cause dissociation of H^O into H* and OH" ions. The OH" ions are adsorbed, leaving H* ions in the solution and thus a pH below that of jiure water. When the particles of Hint are allowed to settle, the pH rises. A hypothesis by Weyl explains this on the assumption that the adsorbed OH" ions are given up to the liquid becau.se of adequate "screening effect" caused when the p;irticles themselves came close together. This might account for the preferential adsorption of OH ions by colloids as has been previously suggested. I. BarsHad: Faj:in.s' experiments suggest results that have been obtained by grinding kaolinite which show an increase in base-exchange capac- it.v. This may be because the broken bonds that result from grind- ing adsorb cations that are exchangeable; or some of the finely- ground m;iterial may reeombine to form a i)ermutite-like substance which has base exchange properties similar to oi-dinary permutite. Other minerals seem to give the same results. Similarly, the base- exchange capacity of halloysite was increased from 20 me. to .300 me. by heating to boiling in the presence of a solution of XaOH. After this treatment the material was no longer halloysite l)nt amorphous and had the properties of permutite. Duncan McConnell: I also believe that upon grinding, jind also after electrodialysis of unground samples, the material may no longer be a pure mineral but a substance disorganized structurally. Thus, true "base ex- change" should be of the zcolytic type while "base exchange" of kaolinite seems to be adsorption. I. Goldberg: I'r. Lewis referred to several methods of jireparation of hydrogen clays in his paper, among which were electrodialysis and leaching with acid. Another method is that of passing a suspension of clay through a column of cation-exchange resin, whereby the exchangeable ions of the clay are exchanged for hydrogen of the H-resin. The work done in our basic research laboratory has indicated that this method is considerably faster and perhaps more efficient in effect- ing the exchange. In addition, it has been found th:it the miuer- alogiciil composition of the clays treated in this manner is much less changed than by employing the other methods. Has Dr. Lewis had any experience in this method of treatment of clays? D. R. Lewis: We have prepared hydrogen clays by electro-dialysis, ordinary dialysis, acid treatment, and by passing the clays through ccdumns of resin. Electro-dialysis is extremely drastic in its action, because structural elements are removed and cau.sed to pass through the .separating membranes by prolonged electro-dialysis. The method of p;issing the clay through H-resin causes the least damage to the clay and is extremely efficient. Some data have alre;idy been given on the comparative results obtained with a day previou.sly saturated with different cations, using the sulphonated styrcne tyjie of resin. With the Xa-cla.v complete conversion to H-clay was obtained at the highest rate of flow through the resin, indicating extremely favorable equilibrium distribution and a very mobile re;icting ion. Duncan McConnell: On the basis of the hypothesis (McConnell, 19.")0) that tetra- hedral (On)i is present, it is suggested that the cation exchange of montmorillonite could be explained on the b.-isis of highly mobile protons in the tetrahedral (OH) 4 groups. This would avoid the necessity of considering octahedral charge by placing the charge differences in the tetrahedral layer. J. W. Jordon : Why slmiilil llie seat of cation exchange be in the tetrahedral l.-iycr rallirr than in the oi-lahedral layer? Duncan McConnell: If the bonding forces of the exchangeable cations are of a low oriler. it would be well to have them as clo.se to the other ions involved as possible. That is the only reason. The only possible alternative to this explanation would be the shifting of the layers .iiic :il.ove llic olher. G. W, Brindley: The tetrahedral (OH)i hypothesis of McConnell arose in the first place as a nu'ans of accounting for the H:;0 fouml upon de- hydration at high temperature. The same is well eslabllslied for kaolinite; that after cxpulsiou of water about .'"iriO-COO" C there is another sm;ill inflexion in the curve jironnd 800°. Recently (Knorriug et al. 1952) we have found that nacrite showed a very similar distinct loss in weight between 80()-S,">0°. This raises the question as to whether the same hypothesis is applicable to the kaolinite-type minerals. Duncan McConnell: The answer will depend on W'hether or not the- w.itei- content of the sample exceeds the theoretical amount. G, W. Brindley: Chemical jinalyses gave 14.48 percent, 14.21 percent, and 14.40 percent of H^O* in three specimens of nacrite jis compared with 13.00 percent calculated from the ideal formiil:i. J, W. Jordon: H;is anyone delermineil the energy of the ion-exi'hange read ion (reaction entropies)? D. R. Lewis: I consider this to be an area of major confusion. The difficulty seems to be in the determination of the thermodynamically correct exchange eijuilibrium constant which is commonly the i>asis for calculating the entroiiy. In exchange experiments the apparent equilibrium constant seems to be greatly influenced by the ratio of ions in the solution. AVith cation-exchange materials :ind the use of a solution containing two kinds of cations, the selectivity seems to be enlmnced toward the cation that is present in the solution in least amount, a characteristic of all cation exchangers. The opposite has been reported for anion exchange where the prefer- ence is for the ion present in greatest relative amount. Tempera- ture also influences the results. If one calculates the entropy from published data on Xa-Li or Xa-K exchange, the valui^s of TaS vary from approximately 1500 calories per mole to as little as 100 calories per mole. Accurate equilibrium constants are needed. When entropy is calculated we assume that the activities of ions in true solution are known, but the activities of adsorbed ions are less certain. The latter must be taken into account and .several ap- proaches have been proposed none of which seems to be entirely satisfactory. Roy Overstreet: Soil chemists find that by substituting molal fractions, or some- thing quite similar to molal fractions obtained by statistical ap- proa<-h, a very good constant is obtained over a wide range of con- centration down to as low as one percent suspeusi2, Exchange- able cation aualvsis of saline and alkali soils: Soil Sci., v. 73, pp. 251-261. Bower, ('. A., and Truog, E., 1940, Base-exchange capacity de- termination of soils and other materials using colorimetric manga- nese method : Ind. and Eng. Chemistry Anal. Ed., v. 12, p. 411. Boyd, G. E., lOol, Annual revics of physical chemistry, cd. by O. K. Rollefson and R. E. Powell, Annual Review, Inc., pp. 309-342. Boyd, G. E., Adamson, A. W., and Myers, L. S., Jr., 19.50, Ion charge and exchange rates with ion-exchange polymers : Am. Chem. Soc. Jour., V. 72, p. 4S07. Boyd, G. E., Myers, L. S., Jr., and Adamson, A. W., 1947, The exchange adsorption of ions from aqueous solutions by organic zeolites. III. Performance of deep adsorbent beds under non- equilibrium conditions : Am. Chem. Soc. Jour., v. 69, p. 2849. Boyd, G. E., Schubert, J., and Adamson, A. W., 1947, The exchange adsorption of ions from aqueous solutions by organic zeolites. I. Ion-exchange equilibria : Am. Chem. Soc. .lour., v. C9, p. 2818. Bradley, AV. F., 1945, Molecular associations between montuioril- lonite and some polyfunctional organic liquids : Am. Chem. Soc. Jour., V. 67, p. 975. Bradley, W. F., and Grim, R. E., 1948, Colloid properties of layer silicates: .Tour. Phys. Chemistry, v. 52, pp. 1404-1413. Brammall. A., and Leech, J. G. C, 1938, Base exchange and its problems : Royal College Sci., Sci. Jour., v. 8, pp. 43-51. I'.ranimall, A., Leech, J. G. C, and Dunstall, W. J., , Montmorillonite in fullers earth, Nuffield Survey : Geo!. Mag., v. 77, pp. 102-112. lirammall. A., and Leech, .1. G. C, , Some aspects of clay chemistry : Royal College Sci., Sci. Jour., v. 7, pp. 69-70. Brammall, A., 1937, The layer-lattice in relation to mineral chemistry: Sci. Progress, v. 31, no. 124. Bray, R. 11., 1929, Base exchange capacities studies of certain soil types in Illinois : Am. Soil Survey Assoc, Bull. 2, pp. 161-168. Bray, R. H., 1937, Chemical and physical changes in soil colloids with advancing development in Illinois soils : Soil Sci., v. 43, pp. 1-14. Bray, R. H., 1942, Base-exchange procedures : Univ. Illinois, College Agriculture, Agr. Expt. Sta., Urbana, 111. Bray, R. H., 1942a, Ionic competition in base-exchange reac- tions : Am. Chem. Soc. .Tour., v. 64, p. 9.54. Rramao. L., Cody, J. G., Hendricks, S. B., and Swerdlow, M.. 19.52. Criteria for the characterization of kaolinite, halloysite and a related mineral in clays and soils: Soil Sci., v. 73, pp. 273-2S7. Brindly, G. W., 1951, X-ray identification and structure of clay minerals : Mineralog. Soc. London. Brown, D. A., and Albrecht, W. A., 19.50, Complementary ion effects in soils as measured by cation exchange between electro- dialyzed hydrogen clay and soils : Soil Sci. America Proc, v. 15, pp. 133-1.38 (1951). Buswell, A. M., and Dudenbostel, B. F., 1941, Spectroscopic studies of base exchange materials : Am. Chem. Soc. .Tour., v. 63, p. 25.54. de Butts, E. H., 1952, Distribution of ions in solutions of weak electrolytes : Jour. Phys. Chemistry, v. .56, pp. 60-63. Byers, H. G., Alexander, L. T., and Holmes, R. S., 1935, The composition and constitution of the colloids of certain great soil groups : U. S. Dept. Agriculture. Tech. Bull. 484. Caldwell, O. G., and JIarshall, C. E., 1942, A study of some chemical and physical properties of the clay minerals nontronite, attapulgite, and saponite : Univ. Missouri, College Agriculture, Agr. Expt. Sta., Research Bull. 354. Cassel, Marthe, 1950, Donnan equilibrium and ionic distribution between solutions of electrolytes and sulfonated resins : Acad. Sci. Paris, Comptes rendus, v. 230, pp. 2286-2288. 3—91001 Chatterjee, B., 1949, The role of aluminum in the interaction of clays and clay minerals with neutral salts and bases : Indian Chem. Soc. Jour., Ind. Eng. Ed., v. 12, pp. 81-99. Chatterjee, B., and Marshall, C. E., 1950, The ionization of mag- nesium, calcium, and barium clays : Jour. Phys. and Colloid Chem- istry, V. 54, pp. G71-681. Clarke, J. T., Marinsky, J. A., Juda, W., Rosenberg, N. W., and Alexander, S., 1952, Electrochemical properties of a permionic anion membrane : Jour. Phys. Chemistry, v. 56, pp. 100-10.5. Cleaver, C. S., and Cassidy, H. G., 19.50, Cliromatographic adsorption of amino acids on organic exchange — resins II : Am. Chem. Soc. .Tour., v. 72, pp. 1147-11.52. Connick, R. E., and Slayer, S. W., 1951, Ion exchange measure- ments of activity coefficients and association constants of cerous salts in mixed electrolytes : Am. Chem. Soc. Jour., v. 73, pp. 1176-1179. Conway, B. E., Bockris, J. O'M., and Ammar, J. A.. 1951, The dielectric constant of the solution in the diffuse and Ilelmholtz double layers at a charged interface in aqueous solution : Faraday Soc. Trans., v. 47, pp. 756-766. Cornet, I., 1943, .Tour. Chem. Physics, v. 11, p. 217. Cornet, I., 1950, Expansion of the montmorillonite lattice on hydration : Jour. Chem. Physics, v. IS, pp. 023 626. Co.'igrove, J. D., and Strickland, ,T. D. H., 19.50, Cation exchange resins. The applicability of the mass-action law to the exchange of univalent ions: Chem. Soc. Jour., v. , pp. 1845- 18.53. Cunningham, G. E., Gabler, H. E.. and Peachin, W. S., 1936, Adsorption of ions and the physical character of precipitates. II. Ferric oxide and bentonite precipitates : Jour. Phvs. Chemistry, V. 40, pp. 1183-1 1S9. Davis, C. W., 1927, The swelling of bentonite and its control: Indus, and Eng. Chemistry, v. 19, p. 13.50. Davis. J. T., and Rideal, E. K., 1948, The distribution of ions at interfaces. The concept of surface phase : Jour. Colloid Sci., v. 3, p. 31.3. Davis, L. E., 1945, Theories of base-exchange equilibriums : Soil Sci., V. 59, p. 379. Davis, L. E., 1945a, Simple kinetic theory of ionic exchange for ions of unequal charge : Jour. Phys. Chemistry, v. 49, p. 473. Davis, L. E., 1945b, Kinetic theory of ionic exchange for ions of unequal charge : .Tour. Phys. Chemistry, v. 49, p. 473. Davis, L. E., 19.50, Ionic exchange and statistical thermodynam- ics. I. Equilibria in simple exchange systems : Jour. Colloid Sci., v. 5, pp. 71-79. Davis, L. E., 1950a, Ionic exchange and statistical thermody- namics. II, Equilibria in irregular svstems : Jour. Colloid Sci., v. 5, pp. 107-113. Davis, L. E., and Rible, J. M., 1950, Monolayers containing poly- valent ions : Jour. Colloid Sci., v. 5, pp. 81-83. Davydov, A. T., and Lisovina, G. M., 1940, The exchange adsorp- tion of cations by bentonite: KoUoid Zhur., v. 11, pp. .308-310. Deb, B. C, and Chowdbury, A. O., 1949, Failure of the inter- national soda method in estimating clay in a subsoil, as n'vealed by the base-exchange capacity : Soil Sci., v. 68, pp. 251-257. Deule, H., Soms, J., and Anyas-Weisz, L., 1950, The behavior of soluble polyelectrolytes toward ion exchangers : Helv. Chim. Acta v. 33, pp. 2171-2178. Deule, H., Anyas-Weisz, L., and Soms, J., 1951, Contact ex- change in ion excliange : Experientia. v. 7, pp. 294-295. Deule, H., and Hostettler, F., 19.50, History of ion exchange: Experientia, v. 6, pp. 445-4.56. Donnan, F. G., 19.34, Die Genaue thermodynamik der Membran- gleichgewicht II : Zeitschr. Phys. Chem., Abt. 168, pp. .369-380. Donnan. F. G., and Guggenhein. E. A., 19.32, Die Genaue thermo- dynamik der Membrangleichgewicht : Zeitschr. Phvs. Chem., Abt. 162, pp. 346-360. Donnan. F. G., Teller, E., and Toply, B., 1937, Fluctuations of composition in a system of molecules in chemical equilibrium : Philos. Mag., v. 24, pp. 981-1001. Duncan, J.'F., and Lister, B. A. .T., 1948. Ion exchange : Quart. Rev., London, v. 2, pp. .307-348. Duncan, .T. F., and Lister, B. A. J., 1949, Ion exchange .studies II : Faraday Soc, Discussion No. 7. Dyal, R. S., and Hendricks, S. B.. 1942, Formation of mixed layer minerals by potassium fixation in montmorillonite : Soil Sci.. Proc, V. 16, pp. 45-48. Edelman, C. H., 1935, Relations between the cry.stal structure of minerals and their base-exchange capacity : 3d Int. Cong. Soil Sci., Trans., v. 3, pp. 97-99, London. 66 Clays and Clay Technology' [Bull. 169 Ekedahl, E., Hogfeldt, E., and Sillen, L. G., 1950, Activities of the components in ion exchangers : Acta Chem. Scandinavica, v. 4, pp. 556-5.58. Elgabaly, M. M., 1950. Mechanism of zinc fixation by colloidal clays and related minerals: Soil Sci., v. 69, pp. 167-173. Elgabaly, M. M., and Jenny, H., 194.S, Cation and anion inter- change with zinc montmorillonite clays : .Tour. Phys. Chemistry, v. 47, p. 399. Endell, K., Fendius, and Hoffman, V., 1034, Base-exchange characteristics of clay and problems of moldability in ceramics : Ber. deutsch. Keram. Gesellsch., v. 15, pp. 595-625. Endell, K., and Yageler, P., 1932, The cation and water hull of ceramic clays in the unfired state : Ber. deutsch. Keram. Gesellsch., V. 13, pp. 377-411. Ensminger, L. E., and Gieselting, .T. E., 1941, The sibsorption of proteins by montmorillonitic clays and its effect on base-exchange capacity : Soil Sci., v. 51, p. 125. Erbring, H., and Lehmann. II., 1944, Exchange reactions with voluminous organic colloid ions on sodium bentonites : Kolloid Zhur., V. 107. p. 201. Euclien, A., 194S, lonenbydrate in Wiissriger Losung : Zeitschr. Elektrochemie, v. 51, pp. 0-24, Eyring, H., Walter, J., and Kimball, G. E., 1944, Quantum chemistry, John Wiley & Sons, New York. Faucher, J. A. Jr., Southworth, R. W., and Thomas, H. C, 1952, Adsorption studies on clay minerals. I. Chromatography on clays: Jour. Chem. Physics, v. 20, pp. 157-100. Fieldes, M.. King, P. J. T., Richardson, J. P., and Swindale, L. D., 1951, Estimation of exchangeable cations in soils with the BccUman flame spectrophotometer : Soil Sci., v. 72, pp. 219-232. Fireman, M., and AVadleight. C. H., 1951, A statistical study of the relation between pH and the exehangealde-.sodium-percentage of western soils: Soil Sci., v. 71, pp. 273-2S5. Ford, T. F., Loomis, A. G., Fidiam, J. F., 1940. The colloidal behavior of clays as related to their crystal structure : Jour. Phys. Chemistry, v. 44, pp. 1-12. Foster, Margaret D., 1951, The importance of the exchangeable magnesium and cation-exchange capacity in the study of mont- morillonitic clays: Am. Mineralogist, v. 36, pp. 717-730. Fronaeus, S., 1951, The use of cation exchangers for the quan- titative investigation of complex systems: Acta Chem. Scandina- vica, v. 5, pp. 8.59-871. Ganguly, A. K., 1949, Reactivity of exchange spots of silicate minerals: Sci. and Culture, v. 14, pp. 341-342. Ganguly, A. K.. 1951, Hydration of exchangeable cations in sili- cate minerals : Soil Sci.. v. 71, pp. 239-244. Ganguly, A. K., and Gupta, S. L., 1948, Hydration of exchange- able cations of clay minerals and synthetic resins: Sci. and Culture (India), v. 14, pp. 81-82. Gapon, E. N., and Chernikova, T. N., 1948, Chromatographic exchange adsorption of cations by soil minerals : Doklady .... Lenina No. 7, pp. 26-28. Garrison, A. D., 1938, Surface chemistry of clays and shales: Am. Min. Met. Eng. Tech. Pub. 1027. Gedroiz, K. K., 1929, The soil complex as an agent of adsorption : Kolloidchem. Beihefte. Gieseking, J. E., 1939, The mechanism of cation exchange in the montmorillonite-beidellite-nontronite type of clay minerals: Soil Sci., V. 47, p. 1. Glaeser, Rachel, 1946, Determination of the base-exchange ca- pacity in montmorillonite : Acad. Sci. Paris, Comptes rendus. v. 222, p. 1179. Glaeser, Rachel, 1946a, Effect of acid treatment on the base exchange capacity of montmorillonite : Acad. Sci. Paris, Comptes rendus. v. 222, p. 1241. Gliickauf, E., 1946, Contributions to the theory of chromatog- raphy : Royal Soc. Proc, v. A 186, pp. 35-57. Glueckauf, E., 1949. Activity coefficient in concentrated solutions containing several electrolytes : Nature, v. 163, pp. 414-415. Garcia, F. G., 1949, Effect of pll on the cation-exchange capacity of montmorillonite and kaolinite: Anales real Soc. espan. fis. y quim. 45B, pp. 1067-1074. Graham, R. P., and Sullivan, J. D., 19.38, Critical study of methods of determining exchangeable bases in clays : Am. Ceramic Soc. Jonr., v. 21, p. 176. Grahame, D. C, 1951, The role of the cation in the electrical double layer : Electrocbem. Soc. Jour., v. 98, pp. 343-3.50. Gregor, H. P., 1948, A general thermodynamic theory of ion ex- change processes : Am. Chem. Soc. Jour., v. 70, p. 1293. Gregor, H. P., 1951, Gibbs-Donnan equilibria in ion exchange resin systems : Am. Chem. Soc. .Jonr., v. 73, pp. 642-0.50. Gregor, H. P., 1951a, Ion-pair formation in ion-exchange systems: Am. Chem. Soc. Jour., v. 73, pp. 3.537-3.538. Gregor, H. P., and Bregman, .J. I., 1948, Characterization of ion exchange resins. I. Acidity and number of con.stituent cation ex- change groups : Am. Chem. Soc. Jour., v. 70, p. 2370. Gregor, H. P.. and Bregman, J. I., 1951, Ion-exchange resins. IV. Selectivity coefficients of various cation exchangers toward uni- valent cations : Jour. Colloid Sci., v. 6 pp. 323-347. Gregor, H. P., Bregman, J. I., Gutoff, F. Bradley, R. D., Bald- win, D. E., and Overberger, C. G., 1951, Studies on ion-exchange resins. Capacity of sulfonic acid cation-exchange resins : .Tcnir. Colloid Sci., V. 6, pp. 20-32. Gregor, H. P., Collins, F. C, and Pope, M., 1951, Ion-exchange resins. III. Diffusion of neutral molecules in a sulfonic acid cation- exchange resin : Jour. Colloid Sci., v. 6, pp. 304-322. Gregor, H. P., Held, K. M., and Bellin, J., 1951, Determination of the external volume of ion-exchange resin particles : Anal. Chem- istry, V. 23, pp. 620-622. Gregor, H. P., Gutoff, F., and Bregman, J. I., 1951, Ion exchange resins. II. Volumes of various cation-exchange resin particles : .Tour. Colloid Sci., V. 6, pp. 245-270. Grigorov, O. N., and Tarshis, 1949, The reactions of ion exchange involving poorly soluble precipitates : Kolloid Zhur., v. 11, pp. 390- 394. Grim, R. E., Allaway, W. H.. and Cuthbert, F. L., 1947, Re- actions of different clay minerals with some organic cations : Am. Ceramic Soc. Jour., v. .30, p. 137. Grimley, T. B., 1950, The contact between a sidid and an electrolyte: Royal Soc. Proc. A 201, pp. 40-61. Guggenheim, E. A., 1944, Statistical thermodynamics of mixtures with zero energies of mixing : Royal Soc. Proc. A 1.83, pp. 203-212. Gupta, S. L., 1949, Maximum exchange capacities of ion ex- change resins from tlieir titration curves: Science and Culture, V. 14, p. 337. Gupta, S. L., 1949a, Neutralization curves of inn exchangers: Science and Culture, v. 14, p. .5.39. Gupta, S. L., 19.50, Titration curves of amiierlites IR-100 and IR-4B : Science and Culture, v. 15, pp. 325-320. Gupta, S. L., Base, M., and Mukherjee, S. K., 19.50, I,on exchange in synthetic resins : Jour. Phys. Colloid Chemistry, v. 54, pp. 1098-1109. Hamada, Hirosi Kagaku no Ryorki, 1948, The equilibrium of exchange reaction by ion exchange resins I : Jour. Japan Chem. v. 2, pp. 83-84. Hanna, W. J., and Reed, J. F., 1948, A comparison of the NHiOAc and buffered BaCls methods for determining cation- exchange properties of limed soils : Soil Sci., v. 66, pp. 447-458. - Harned and Owen, 1950. Thermodynamics of electrolyte solu- tions, 2d ed.. New York, Reinhold Publ. Co. Harry, W. T., 1950, Aluminum replacing silicon in some silicate lattices: Mineralog. Mag., v. 29, pp. 142-149. Hauser, Ernst A., 1951, The surface structure and composition of colloidal siliceous matter : Jonr. Phys. Colloid Chemistry, v. 55, pp. 68-79. Heller, W., and Thompson, A. C, 1951, The volume recpiirenients of polymer molecules. I. The apparent specific volume of polystyrene in solution : Jour. Colloid Sci., v. 6, pp. 57-74. Hendricks, S. B., 1941, Base exchange of the cbiy mineral montmorillonite tor organic cations and its dependence upon ad- sorption due to Vander Waals forces: Jour. Phys. Chemistry, v. 45. p. 65. Hendricks, S. B., 1945, Base exchange of crystalline silicates: Ind. and Eng. Chemistry, v. 37, pp. 625-630. Hendricks, S. B., Nel.son, R. A., and Alexander, L. T., 1940, Hydration mechanism of the clay mineral montmorillonite satu- rated with various cations: Am. Chem. Soc. Jour., v. 62, p. 1457. Ileymaun, E., and O'Donnell, I. J., 1949, Physicochemical in vestigation of a cation exchange resin (Aml)erlife IR-100) I. Resin equilibria : Jour. Colloid Sci., v. 4, p. .395. Heymann, E., and O'Donnell, I. J., 1949a, Physicochemical in- vestigation of a cation exchange resin (Amberlite IR-100). II. Resin conduct.-mce : Jour. Colloid Sci., v. 4. p. 405. lli<'ster, N. K., and Vermeulen. T., 1949, Kinetics of ion ex- c-liange : Prelim, paper prepared for Los Angeles Regional Meeting. \m. Inst. Cliem. Eng., March 8, 1949. Ililfiger, J. P., 1949, Les echaugeurs d'ions : Chimie et industrie, v. 01 , pp. 559-563. art II PUOI'KUTIKS OK Cl.AYS 67 IIiilTinMii. r., and Kiidell, J., VXH), Die Al)li!inf,'ij;!<('i-I drs Kation- eniiustausches und dci- Quelluiig bci Montmorilloiiite von der \'orcrliitziin!; (Aiisziig): Zcitschr. Anj^ewaiidtc Cliciiiic. v. r>2, p. 708. HiilTniaii, l'., and Gieso, K., 1039, tjbor den Katinniiiaiistaiisch an 'rniunineralien : Kolloid-Zeitsehr., v. 87, p. 21. Ilcifl'niaii. I'., and Klcnien. R.. 1050, Loss on heating of the ahilit.v of lithium ions to exchange in l)entonitc: Zeitselir. Anorg. Chem'ie. v. 202, pp. Oo-Otj. Hogfeldt, K., Ekedahl, E.. and Sillen, L. G., 10.->0. Activities of tile liaiinm and hydrogen forms of Dowe.\ 50: Acta {'hern. iSc.indinavica, v. 4. pp. S20-830. II..gf.ddt, E.. Ekedahl, E., and Sillen, L. 0., 19.")0a, Activities of tile (■(iiiiponents in ion exchangers with multivalent ions: Acta ('hem. Scandinavica, v. 4, pp. S2S-820. Holm, L. W., and Westermark, T., lO.'iO, The exchange of isotopic ions in cation exchangers: Acta Chem. Scandinavica, V. 4, pp. OG8-0(i9. Iloon. U. ('., and Dhawau, C. L., 1048, Eloctrodialysis of soils: The influence of exchangeable bases on the recovery of manganese by <'le<'trodialysis : Indian Cheiii. Soc. .Tour., v. 17, p. 10."). Ilosking. .1. S., 1048, The calion-e.\ch;inge capacity of soils and soil colloids. I. Variation with hydrogen-ion concentration : Council Sci. Ind. Kesearcb .lour., v. 21, pp. 21-37. Ilosking. .J. S.. 1048a, The cation-exchange capacity of soils and soil colloids. II. The contribution from the sand, silt, and cl.ay fractions of the .soil : ("onncil Sci. Ind. Research .Tour., v. 21, pp. .SS-fiO. .r.'icob. .v.. IIolTiiian, IT., Ijoofniaiin. II., and Maegdefrau, E., lO.S"), Chemische und Riintgenographische I'ntersuchungen iiber die inineralische Soiptiiinosubstanz im Itodeii, ISeihefte, pp. 11-19. .renny. Hans. 10;'.2. Studies on the mechanism of ionic exchange in colloidal alnmiiium silicates: .lour. I'hvs. Chemistry, v. 36, pp. 2217-22.58. .leniiy, Hans, 1030, Simple kinetic theory of ionic exchange. I. Inns iif equal valency : .lour. I'h.\*s. Clieniistry. v. 40. pp. riOl-517. .Icnny, H., and Overstreet, R., 10:i0. Surface migration of ions and contact exchange: .Tour. Phys. Chemistry, v. 4.'}, p. 11.85. .loffe, .1. S.. and I>evine. A. K., 1947, Fixation of potassium in relation to exchange cajiacity of soils. II. Associative fixation of other cations particularly ammonium : Soil Sci., v. 03, p. 151. .lolinson" A. Ij., 1040, Surface area and its effect of exchange capacity of montniorillonite : Am. Ceramic Soc. ,Ionr., v. .32, pp. 210-214. .lud.i. \\ '.. and Carron, M., 1048, Equilibrium and velocity of the sodium-ii.\ (Irogen exchange on carbonaceous exchangers in contact with chloride solutions: Am. •'hem. Soc. .Tour., v. 70, pp. 3295-:!310. Karpeii, Vasilesco, 1040, L'Activite des ions, la loi d'action de m.-iase et less lois de Raoult : Acad. Sci. Paris. Comptes rendus, v. 228, p. 1282. Karpen, A'asilesco, 1040a, Sur la nntiiin d'aclivite en cliemie Iili.\siijiie: Acad. Sci. Paris, Comptes rendus, v. 228, p. 1220. Kasper. F.. and Bloch, .1. M.. 194.8, Fixation de nudecnles mine- r.ales par la bentonite; .\cad. Sci. Paris. Coinples rendus. v. 2*20. p. -251. Kayser. F.. and P.locli. J. M.. I'.IIS, The fixation of inorganic molecules upon bentonite : Acad. Sci, Paris, Comptes rendus, V. '220, pp. 251-2.53. Kayser, F.. Rloch, .T. M., and Cominery. (J., 1051. The adsorp- tion of molecules of sodium hydroxide, sodium cliloride and sodium citr;ite bv sodiuni niontmorillonite : Sn<'. cliini. France P.ull., 1051. pp. 402-405. Keenaii. A. C. Mooney. R. W.. and Wood, I.. A.. 1051. The relation lietween exchangeable ions and water adsru-ptimi : .Imir. Pb.vs. Colloid Chemistry, v. 55, pp. 1402-1474. Kelley, W. P., 1035, The evidence as to the crystallinity of soil <-olloids: 3d Int. Cong. Soil Sci., Trans., v. 3, "p. ,88. Kelley, W. P., 10.3.5a. The significance of crystallinity in rela- tion to base exchange: 3d Int. Cong. Soil Sci., Trans., v. 3. p. 92. Kelley, W. P., 1948, Cation exchange in soils, Xew York, Rein- liuld Publ. Co. Kelley, W. P., .Tenny, H., and Brown, S. JI., 1935. Hydration of minerals and soil colloids in relation to crystal structure : 3d Int. Cong. Soil Sci., Trans., v. 3, p. 84. Kiellaud. .Tacob, 10.35, Thermodynamics of base-exchange equi- libria of some different kinds of clays: Soc. Chem. Ind.. .Tour. 1935, PIi. 2.32-2.34. Kimball, G. E., Cutler, M., and Samelson, II., 1952, A theory of li'dyelectrolytes : Jour. Phys. Chemistry, v. .56. pp. 57-00. KiMining. < ). von. lirlndley. G. \V.. and Hunter. K., 10.52, Xacrite friun Ilirvivaara, niuthern Karelia, Finl:iiid : Mineralog. Mag., v. 28, pp. 903-972. Kressman, T. R. E., 1952, Ion exchange separations based upon ionic size: .Tour. Phys. Chemistry, v. .50, pp. 118-123. Kressman, T. R. E., and Kitchener, ,T. .\., 1949b, Cation ex- <-hange with a synthetic phenolsulfnnate resiu. III. Ki|uilibria with large organic salts : Chem. Soc. London, Jour. 1949. Kressman, T. R. E., and Kitchener, J. A., 1040a. Cation ex- i-hange with a synthetic plienolsulfonale resin. II. Eipiililiria with multivalent cations: Chem. Soc, London, Jour. 1049, pp. 1201-1208. Kre.ssman, T. R. E., and Kitchener, J. A., 1040b, Cation ex- change with a synthetic phenolsulfonale resin. III. Equilibria with large organic salts: Chem. Soc, London, Jour. 1949. Kressm.an, T. R. E., and Kitchener, J. A., 1949c, Cation exchange with a synthetic phemdsulfonate resin. IV. A note on equilibria in presence' of nonaqueous solvents: Chem. Soc, London. Jour. 1040, pp. 1211-1213. Krishnamoorthy, ('., and Overstreet, R., 1040, Theory of ion exchange relationships: Soil Sci., v. 08, pp. 307-315. Krishnamoorthy, C, and Overstreet, R., 1950, Behavior of hydrogen ion in exchange reactions: Science, v. Ill, pp. 231-232. Krishnamoin-thy, C, and Overstreet, R., 19,50a, An experimental evaluation of ion-exchange relationships : Soil Sci., v. ()9, pp. 41.53. Krishnamoorthy, C, and Overstreet, R., 19.50h, Beh.-ivior of hydrogen in ion-exchange reactions : Soil Sci.. v. 60, p. .87. Krishnamoorthy, C, Davis, L. E., and Overstreet, R., 1948, Ionic exch.ange equations derived from statistical thermodynamics: Science, v. 108, p. 439. Kunin, Robert, 1951, Ion exchange: Anal. Chemistry, v. 23, PI). 45-46. Kunin, R., and Barry, R. E., 1949, Equilibrium and column behavior of exchange resins — carboxylic, weak acid type, cation exchange resin: Ind. and Eng. Chem., v. 41, pp. 12(!9-1272. Kunin, R., and McGarvey, F. X., 1949, Equilibriuiu and coluinn behavior of exchange resins — strong base anion exchange resin : Ind. and Eng. Chem., v. 41, pp. 1265-1268. Kunin. R., and Myers, R. J., 1947, Rates of anion exchange in ion-exchange resins: Jour. Phvs. and Colloid Chem., v. 51, pp. 1111-11.30. Kunin. R., and Myers, R. J.. 19.50, Ion exchange resins. New York, John Wiley & Sons. Laws, W. D., ami Page, J. B., 1946, Changes produced in kaolinite by dry grinding: Soil Sci., v. 62, p. .319. Lewis, D. R., 1951, Reference clay minerals; Am. Petroleum Inst., Project 49, Sect. 3, Prelim. Rept. 7, Columbia Univ. Press. Lewis, D. R., 19.52, Replacement of the exchangeable cations of clay by means of ion exchange resin columns : Paper pre- sented at Am. Chem. Soc. meeting, Buffalo, New York, March 27, 1052. Lisicki, Z., 19.50. Thermocbemical method of determining the capacity of hydrogen-ion excbangers: Przemyst Chem. 6(29, pp. 4.5, 40)' (19.50), CA 4.5, p. lt>435. Loddesdl, Aasulv, 1932, Factors affecting the amount of clectro- dialyzable ions liberated from some .soils : Soil Sci., v. 33, p. 187. Loosjes, Robert, 1942, pH-meting in suspensies dis.scrtation, Utrecht. Low, P. F., 1951, Force fields and chemical equilibrium in heterogeneous systems with special reference to soils : Soil Sci., V. 71, pp. 409-418. Lowen, W. K., Stoenner, R. W., Argcrsinger, W. J., Jr., David- son, A. W., and Hume. D. N., 1951, Studies of cation-exchange equilibrium on a .synthetic resin : Am. Chem. Soc. Jour., v. 73, pp. 2006-2674. McConnell. Duncan, 19.50, The crystal chemistry of montnioril- lonite : Am. Mineralogist, v. 35, pp. 166-172. MacEwan, D. M. C, 1948, Clay mineral complexes with organic liquids: Clay Min. Bull. 2. p. .30, Aug. 1048. Mackenzie,' R. C, 1951, A inicromethod for determination of cation-exchange capacity of clay : .Tour. Colloid Sci., v. 0, pp. 210-222. Magistad, O. C, Fireman. M., and Mabry, B., 1944, Comparison of base exchange equations founded on law of mass action : Soil Sci.. v. 57, pp. 371-379. Maillard, A., and Hilfiger, J. P.. 1949. New method for study of active dissociable groups : Acad. Sci. Paris, Comptes rendus, v. 228, pp. 1806-1807. 68 Clays and Clay Technology [Bull. 169 Mariiisk.v, J. A., no date. A stud.v of the ion-exebange equilibria l)etween hydrogen ion and sodium, calcium barium, and cerium ions: Massachusetts Inst. Technology, Lab. for Nuclear Sci. on Eng., Tech. Rept. 34. Marshall, C. E., 1949, Colloid chemistry of the silicate minerals. New York, Academic Press. Marshall, C. E., and Barber, S. A., 1949, The calcium-potassium relationships of clay minerals as revealed by activit.v measure- ments: Soil Sci. Soc. America Proc, v. 14. pp. 86-88 [1950]. Marshall. C. E., and Gupta, R. S., 1933, Base-exchange equilibria in clays: Soc. Chem. Ind., Jour., v. 52, pp. 433T-443T. Mattson, S., and Wiklander, L., 1950, The "amphoteric" double layer and the double ionic exchange in soils : Faraday Soc. Trans., V. 36, pp. 306-319. . . . Chem. Abs., v. 34, p. 2990 '(1940). Mayer, S. W., 1950, Cation exchange equilibria at constant ionic strength : Am. Chem. Soc. Jour., v. 72, p. 2292. Mehlich, A., 1945, Effect of type of soil colloid on cation- adsorption capacity and on exchangeable hydrogen and calcium as measured by different methods : Soil Sci., v. 60, pp. 2S9-304. Mehlich, A., 1948, Determination of cation- and anion-exchange properties : Soil Sci., v. 66, pp. 429-445. Mclsted, S. W., and Bray, R. H., 1947, Base-exchange equilib- riums in soils and other exchange materials : Soil Sci., v. 63, p. 209. Merling, J., 1946, The hydration of montmorillonitc : Faraday Soc. Trans., v. 4213, pp. 205-219. Merwin, H. D., and Peech, M., 1950, Exchangeability of .soil potassium in the sand silt and clay fractions as influenced by the nature of the complementary exchange cation : Soil Sci. Soc. America Proc, v. 15, pp. 125-128 [1951]. Mitra, R., and Rajagopalan, K. S., 1948, The apparently tri- basic acid character of hydrogen mica : Indian Chem. Soc. jour., V. 25, pp. 591-592. Mitra, R. P., and Rajagopalan, K. S., 1948a, Titration curves of hydrogen micas : Nature, v. 162, p. 104. Moelwyn-Hughes, E. A., no date, Ionic hydration : Cambridge Philos. Soc, Proc. v. 45, pp. 477-481. Mooney, R. W., Keenan, A. G., and Wood, L. A., 19.52, Adsorp- tion of water vapor by moutraorillonite : Am. Chem. Soc Jour V. 74, pp 1367, 1374. Mortland, M. M., and Gieseking, J. E., 1949, Anion sorption and exchange by amine-clay complexes : Soil Sci., v. 65, pp. 391-.397. Mukherjee, J. N., Chatterjee, B., and Banerjee, B. M., 1947, Liberation of hydrogen aluminum and ferric ions from hydrogen clays by neutral salts : Jour. Colloid Sci., v. 2, p. 247. Mukherjee, J. N., and Mitra, R. P., 1946, Some aspects of the electrochemistry of clays: Jour. Colloid. Sci., v. 1, p. 141. Mukherjee, S. K., and Ganguly, A. K., 1950, Base-exchange caiiacity and crystalline structure of silicate minerals related to soils and clays : Indian Jour. Physics, v. 24, pp. 233-2.56. Mukherjee, S. K., and Ghosh, S., 1950, Cation exchange studies with the collodial clay salts of silver : Jour. Phys. Colloid Chemis- try, V. 54, pp. 1110-1121. Mukherjee, S. K., and Gupta, S. L., 1946, Limiting value of the base-exchange capacity of soils and clays : Indian Jour. Agr. Sci V. 16, pp. 442^44. Mysels, K. J., 1951, (Potentiometric) Measurements in (some) colloidal systems : Science, v. 114, pp. 424-425. Olphen, H. van, 1950, Stabilization of montmorillonitc sols by chemical treatment. I. Properties of sodium and calcium mont- morillonite sols: Recueil travaux chimiques, v. 69, pp. 1308-1312 . . . II., ibid, pp. 1313-1322. Ongaro, D., 1950, Determination of exchange capacity of ion exchangers : Chimica e industria, v. 32, pp. 264-268. Osugi, S., and Matsumura, Y., 1938, The equivalent weight of clay acid of soil : Jour. Sci. Soil Manure, Japan, v. 12, pp. 1-12. Overstreet, R., 1945, Ionic reactions in soils and clay suspen- sions : The significance of soil filtrates : Sou Sci., v. 59, p. 265. Page, J. B., and Baver, L. D., 1939, Ionic size in relation to fixation of cations by colloidal clay : Soil Sci. Soc. America Proc V. 4, pp. 150-155. Paneth, P. A., 1948, Exchange of ions between crystal surfaces and solutions : Jour. Chem. Physics, v. 45, pp. 205-211. Paravicini, Elizabeth D. de, 1932, Sorption phenomena in soils and in silica gels: Soil Sci., v. 33, p. 163. Pedinelli, M., and Bovelacci, F., 1950, Ion exchange and ion exchangers: Chimica e industria, Milan, v. 32, pp. 271-276. Peech, M., 1945, Determination of exchangeable cations and exchange capacity of soils — rapic micfomethods utilizing centrifuge and spectrophotometer : Soil Sci., v. 59, p. 25. Perkins, Alfred T., 1948, Kaolin and treated kaolins and their reactions : Soil Sci., v. 65, pp. 185-191. Perkins, Alfred T., 1949, Reactions of muscovite, bentonite, and their treated residues : Soil Sci., v. 67, pp. 41-46. Perkins, A. T., and Draggsdorf, R. D., 1950, Decomposition of bentonite as influenced by absorbed hydrogen and calcium plus grinding: Soil Sci. Soc America Proc, v. 15. pp. 93-97 [1951]. Plane, R. A., and Taube, H., 1952, Observations of the kinetics of the exchange of water between Cr (H^O) *** and solvent: .lour. Phys. Chemistry, v. 56, pp. 33-38. Pratt, P. F., 1952, Release of potassium from nonexchangable forms from size fractions of several Iowa soils : Soil Sci. Soc. America Proc, v. 16, pp. 2.5-29. Puri, A. N.. and Dua. A. N.. 1938, Hydrogen-ion activity of colloidal acids in soils : Soil Sci., v. 46, pp. 113-128. Puri, A. N., and Uppal, H. L.. 19.39, Base exchange in soils : I. A. Critical examination of the methods of finding base-ex- change capacity of soils : Soil Sci., v. 47, p. 245. Rendig, V. V., 1947, Rapid determination of base exchange capacity of soils with flame photometer : Soil Sci. Soc. America Proc, V. 12, pp. 449-451. Richardson, R. W., 1951, Selective diffusion into ion-exchange resins. A new method for the purification of the direct cotton dyes: Chem. Soc. Jour., 1951, pp. 910-914. Rodda, J. L., 1952, Anomalous behavior of montmorillonite clays in Clerici solution : Am. Mineralogist, v. 37, p. 117. Ross, C. S., and Hendricks, S. B., 1945, Minerals of the montmorillonite group : U. S. Geol. Survey Prof. Paper 205B. Sakai, W., and Seiyama, T., 1952, Theory of ionic exchange equilibria : Kyushu Univ., Faculty of Engineering Mem., v. 13, pp. 95-117. Samuelson, Olof, 1941, Investigations of complex salt solutions with the use of ion £xchangers : Svensk. Kem. Tid., v. 63, pp. 183-193. Schachtschabel, P., 1951, Determination of S-values (exchange- able cations) T-values (sorption capacity and degree of saturation of soils : Zeitschr. Diing und Bodenk, und Pflanzeneruilhr, v. 53, pp. 7-20. Schachtschabel, P., no date, Untersuchungen fiber die Sorption der Tonmineralien und orgauischen Bodenkolloide, uml die Besti- nemung des Anteils dieser KoUoide an der sorption in Boden : Kolloid-Beihefte Bd. 51, pp. 199-276. Schofield, R. K., 1946, Ionic forces in thick films of liquid between charged surfaces : Faraday Soc, Trans., v. 42B, pp. 219-225. Schofield, R. K., 1947, A ratio law governing the equilibrium of cations in the soil solution : 11th Internat. Cong. Pure Applied Chemistry, London, v. 3, pp. 1257-1261. Schollenberger, C. J., and Simon, R. H., 1945, Determination of exchange capacity and exchangeable bases in soil-ammonium acetate method : Soil Sci., v. 59, p. 12. Schubert, Jack, 1948, The use of ion exchangers for the de- termination of physical-chemical properties of substances, par- ticularly radiotracers, in solution. I. Theoretical : .Jour. Phys. Colloid Chemistry, v. .52, pp. 340-3.50. Schubert, Jack, 19.50, Analytical applications of ion exchange separations : An.il. Chemistry, v. 22, pp. 1359-1368. Schubert, Jack, 1952, Ion exchange studies of complex ions as a function of temperature, ionic strength and presence of formaldehyde : Jour. Phys. Chemistry, v. 56, pp. 113-118. Schubert, J., and Richter, J. W'., 1948, Cation exchange studies on the barium citrate complex and related equilibria : Am. Chem. Soc. Jour., V. 70, p. 4259. Schubert, J., and Richter, J. W., 194Sa, The use of ion ex- changers for the determination of physical-chemical properties of substances, particularly radiotracers, in solution. II. The dis- sociation -constants of strontium citrate and strontium tartrate : Jour. Ph.vs. Colloid Chemistry, v. .52, p. 3.50. Schuler, R. H., Boyd, A.C. Jr., and Kay, D. J., 1951, Ex- periments with ion-exchange resins : Jour. Chem. Education, v. 28, pp. 192-194. Siefert, A. C, and Henry, E. C, 1947, Effect or exchangeable cations on hydrophilic nature of kaolin and bentonite : Am. Ceramic Soc. Jour.. V. 30, pp. 37-48. Slabaugb, W. H., and Cultertwon, J. L., 1951. The effect of cer- tain reagents upon the properties of bentonite colloids. I. Inorganic salts : Jour. Phys. Colloid Chemistry, v. 55, pp. 7-14-757. Spiegler, K. S., and Coryell, C. D., 1951, Electro-migration in a cation exchange resin : Science, v. 113, pp. 546-547. Tart II] PUOI'ERTIES OF Cl.AYS 69 Spiegler, K. S., and Coiyoll. C. D., 1052, Elect ro-migration in a cation exchange resin. II. Detailed analysis of two-component systems : .Tour. Phys. Chemistry, v. 56, pp. lOC-11.3. Stout, P. R., 1939, Alterations in the crystal structure of clay minerals as a result of phosphate fixation : Soil Sci. Soc. America, Proc, V. 4, pp. 177. Swietostawski, W., 1950, Classification of sorbents and ion ex- changers: Przemyst Chem., v. (29), pp. 41-43. Teichner, S., 1947, Emplacement des ions exchangeahles dans la montmorillonite : Acad. Sci. Paris, Comptes rendus, v. 225, pp. 1337-1339. Tompkins, E. R., 1950, Ion exchange separation : Anal. Chem- istry, V. 22, pp. 1.352-1359. Tompkins, E. R., Harris, D. II., and Khym, J. X., 1949, A stndv of the effect of a nunilicr of column variahlcs : Am. (^hcm. Soc.'jour., V. 71, pp. 2.">04-2510. Topp, X. E., 1951, Ion exchange resins : Sci. Progress, v. .39, pp. 640-657. Topp, N. W., and Pepper, K. R., 1949, Properties of inn ex- change resins in relation to their structure. I. Titration curves : Chem. Soc. Jour., 1949, pp. 3299-38-3. Toth, S. .1., and Prince, A. I.., 1945. Estimation of cation- exchange capacity and exchangeable Ca, K, and Na contents of soils by flame photometer teclini(|Ues : Soil Sci., v. 67. pii. 4.39-445. Toth, S. .T.. Prince, A. L.. W.illace, A., and Mikkelson, P. S., 1948, Rapid quantitative determination of eight mineral elements in plant tissue by a systematic procedure involving use of a flame photometer : Soil Sci., v. 06, pp. 4.59-466, photostat. Vagcler, P., and Woltersdorf, .7. F., 1930, Pflanzenernah : Diin- gung A 15, pp. 329-342. Vajna, Sandor, 19.50. Ion exchanging resins : Magyar Kem. Folyoirat, v. 56, pp. 88-96. Vanselow, A. P., 1932, Equilibria of the base-exchange re- actions of bentonites. permutites, soil colloids, and zeolites : Soil Sci.. y. 33, pp. 95-113. Vanselow, A. P., 1932a, The utilization of the base-exchange reaction for the determination of activity coefficients in mixed electrolytes : Am. Chem. Soc. .Tour., v. .54, p. 1307. Vasilesco-Karpen, N., 1949, The activity of ions, the law of mass action, and Raoults law : Acad. Sci. Paris, Comptes rendus, V. 228. pp. 1292-1294. Vasilesco-Karpen, N., 1949a, The concept of activity in physical chemistry : Acad. Sci. Paris, Comptes rendus, v. 228, pp. 1220- 1222. Vendl. JI., and Hung, R., 194.3, The role of hydroniuni ions in the exchange behavior of siliceous clays: Palatine-Joseph Univ., Tech. Econ. Sci., Dept. Mining Metallurgy. Pub. 15, pp. 344-354. Waggener, W. C, and Stoughton. R. W., 19.52, Chemistry of thorium in aqueous solutions. 11. Chloride complexing as a func- tion of ionic strength: .Tour. Phys. Chemistry, v. 56, pp. 1-5. Walling, C. 1950, The acid strength of surfaces : Am. Chem. Soc. Jour., v. 72, pp. 1164-1168. Walton, H. P., 1941, Ion exchange between solids and solu- tions: Franklin Inst. Jour., v. 232, pp. .305-337. Wear, .1. I., aud White, .T. L., 1951, Potassium fixation in clay minerals as related to crystal structure: Soil Sci., v. 71, pp. 1-14. Weyl, W. A., 1950, Role of ionic deformation in surface chemis- try : New York Acad. Sci., Trans., v. 12, pp. 245-2,56. Weyl, W. A., 19.50a, Crystal chemical considerations of the surface chemistry of silica: Research, v. 3, pp. 2.30-235. Whiteside. E. P.. and Mar.shall, C. E., 1944, Mineralogical and chemical studies of the Putnam silt loam soil: Univ. Missouri, College of Agriculture, Agr. Expt. Sta., Research Bull. 386. Wieguer, G., 1931, Some physico-chemical properties of clays. I. Base exchange or ionic exchange : Soc. Chem. ludustrv, v. .50, p. 65. Wiegner. G.. 1935, Ionenumtau.s<'h und Struktur : 3il Int. Cong. Soil Sci., Trans., v. 3, pp. 5-28. Wiegner, \. G., and .lenny, H., 1951, Uber Basenaustausch an Permutiten : Kolloid-Zeitschr., v. 42, pp. 268-272. Wiklander. Lambert, 1946. Studies on ionic exchange with special reference to the conditions in soils : Kgl. Lantbruks Hiigskol Ann. 14, pp. 1-171. Wiklander, Lambert, 1949, Adsorption equilibria between ion exchangers of different nature. I. Release of cations from soil by adsorption on exchange resins : Kgl. Lanbruks Hogskol. Ann. 16, pp. 670-682. Wiklander, Lambert, 1950, Fixation of potassium by clays satu- rated with different cation : Soil Sci., v. 69, pp. 261-2G8. Wiklander, Lambert, 1951, Annals of the Royal Agricultural College of Sweden, v. 18, pp. 154-162. Wiklander. L., and Gieseking, .1. E., 1948, Exchango-aliility of adsorbed cations as influenced by the degree of saturation and the nature of the complementary ions with special reference to trace concentrations: Soil Sci., v. 66, p. 377. Young, T. F., and Blatz, L. A., 1949, The variation of the properties of electrolytic solutions with degrees of dissociation : Chem. Rev., y. 44, pp. 93-115. Zebroski, E. L., Alter, H. W., and Hewmann. P. K., 1951, Thorium complexes with chloride, fluoride, nitrate, phosphate, and sulfate: Am. Chem. Soc. .lour., v. 73, pp. 5646-56.50. Zerfuss, S., and Utter, B. A., 1946, Note on reactions of alkali with kaolin and nietakaolin : Am. Ceramic Soc. Jour., v. 29, pp. 205-207. Zhukov, I. I., and Brodskaya, N. I., 1949, Connection between exchange capacity and electrokinetic potential : Kolloid Zhur., V. n, pp. 322.327. ABSORPTIVE AND SWELLING PROPERTIES OF CLAY-WATER SYSTEM Br Isaac Barshad ' INTRODUCTION The subject of the adsorption and swelling properties of the clay-water system may be divided into three parts : (1) clay-water vapor system, (2) clay-liqnid water sys- tem in the gel state, and (3) clay-liqnid water system in the fluid state, i.e. pastes and sols. In studying the swelling of clays in relation to hydra- tion, it is necessary to distinguish between the two kinds of swelling encountered ; namely, the intramicellar swell- ing which involves the expansion of the crystal lattice itself, commonly known as tlie interlayer or interlamellar expansion, as found in montmorillonite, vermiculite-like, and in some of the hj'drous mica clay minerals ; and intermicellar swelling which involves an increase in vol- ume due to adsorption of water molecules between in- dividual clay particles. Intramicellar swelling can be identified and measured onl.y by x-ray analysis, whereas the intermicellar swelling can be determined from a measurement of the total increase in volume of the clay body or of the clay-bearing material with apparatus designed for this purpose (Freundlich et al. 1932; Keen and Eaczkowski 1921 ; von Ensline 1933 ; Winterkorn and Baver 1934). The common feature among the clay minerals is their platy surfaces which consist eitlier of oxygen ions organ- ized into an hexagonal network, or of hydroxyl ions organized into a closely packed network. The oxygen surfaces characterize the montmorillonitic and the mica- ceous clay minerals, whereas both oxygen and hydroxyl surfaces characterize the kaolinitic and the chloritic clay minerals. One of the fundamental differences among the clay minerals lies in the amount and kind of exchangeable cations present on their surfaces, and in the seat of the excess negative charge of the crystal lattice which these cations neutralize (Hendricks 1945; Koss and Hendricks 1945). The scarcity of the exchangeable cations relative to the number of the surface oxygen ions which bear the negative charge (the ratio of the oxygen ions to the cations may range from 3 to 1. as in the micas, to 18 to 1, as in some of the montmorillonites) has been advanced as the possible cause for the jiolarization of these sur- faces and consequently for their reactivity with polar molecules (Barshad 1952). One of the important features, insofar as water ad- sorption is concerned, by which the various clay minerals may be differentiated, is tlie extent of the absorbing sur- face. Table 1 is a summary of the extent of the external ■ surfaces of several of the clay minerals as measured by No and ethane gas adsorption, and of the extent of the internal surfaces of montmorillonite as measured by glycol adsorption (Dyal and Hendricks 1950; Keenan et'al. 1951: Mooney et al. 1952, 1952a; Nelson and Hen- dricks 1942). The external surfaces of the mica-like clay minerals and those of montmorillonite are in the same range of values, but that of kaolinite is somewhat less : the ratio of the internal to tlie external surface in the montmorillonites ranges from 9 to 40. ♦ Lecturer. Department of Soils, and Assistant Soil Chemist. Agricul- tural Experiment Station, University of California, Berlieley, California. Table 1. External and internal surface arrax of xome clay minerals.* External surface Internal surface by ethylene glycol adsorption Ratio of Clay mineral By ethylene glycol adsorption By ethane or N2 adsorption internal to external surfaces Montmorillonite Range sq. m/g 20-80 50 75-180 130 22-37 30 37 sq. m/g 30-90 48 30-100 80 18-44 29 44 •sq. m/g 700-800 750 n 9-40 15 Micalike Range Average Kaolinite Average * Data taken from the following: D.val and Hendricl;s. 1950 Moonoy. et al., 1952, 1952a; Nelson and Hendriclts. 1942. Keenan. ct al.. 1951; CLAY-WATER VAPOR SYSTEMS Early studies of the degree of hydration of clay bear- ing materials or clays upon exposure to a given vapor pressure disclosed that it is affected by the degree of hydration of the adsorbent prior to exposure (that is, whether the adsorbent gains or loses water during the exposure) (Thomas 1921, 1921a), the aggregate struc- ture of the adsorbent (Thomas 1928), the jiind and amount of exchangeable cations on the adsorbent (An- derson 1929, Kuron 1932, Thomas 1928a), the kind and amount of salts and oxides present within the adsorbent (Thomas 1928), and the nature and amount of the clay minerals present in the adsorbent (Alexander and Har- ing 1936, Keen 1921, Kuron 1932, Puri 1949). The usefulness of these early studies is somewhat limited because the precise mineralogical composition of the clay minerals studied was not recognized. However, they laid the foundation for later investigations which deal with the specific clay -mineral species of the mont- morillonitic and kaolinitic groups. The present review deals mainly with the later investigations. Many investigators contributed to the elucidation of the relation between intramicellar swelling (that is, the interlaver expansion) and the degree of hydration (Bar.sliad 1949, 1950; Bradley et al. 1937; Hendricks and Jett'erson 1938; Hendricks et al. 1940; Hofmann and Bilke 1936; Mering 1946; Mooney et al. 1952, 1952a). The discussion that follows is based on these investiga- tions, but the data presented are taken from Hendricks and Jefferson (1938). Hendricks et al. (1940). and Mooney et al. (1952, 1952a). The relation between hydration and interlayer expan- sion can be shown most clearlj' b.v expressing the degree of hydration on adsorption isotherms in terms of water molecules per unit cell (i.e. 12 oxygen ions) of the crys- tal lattice, and indicating the course of expansion by lines drawn across the adsorption isotherms: one line representing expansion equivalent to a unimolecular (70) I'iirt III Properties of Clays 71 03 01 05 06 07 RELATIVE PRESSURE P/Po Figure 1. Adsorption isotherms at IJCC of a Mississippi moiit- morilloiiitc saturated with various monovalent cations. 0' 0! 03 04 05 06 0' OB 09 RELATIVE PRESSURE P/Po PlGURE 2. Adsorption isotherms at ;!0°C ii£ n .Mississippi mont- morillonite saturated with various divalent cations. la.yer of water, two lines representing a dimoleeular layer, and three lines representing a trimolecular layer. The possible number of water molecules per uni- moleeular layer when related to tlie unit cell may vary from 2 to 4, depending on the nature of the organization of the water molecules with respect to the hexagonal net- work of the oxygen surfaces (Barshad 19-49). Further- more, the process of interlayer expansion, apart from hydration, may be described as occurring in two distinct steps : the first step consisting in a separation of the oxygen surfaces during the course of which the inter- layer cations remain attached to the surfaces ; and the second step consisting in a detachment of the cations from the ox.ygen surfaces through their interaction with water molecules. Interlayer Expansion in Relation to Hydration. Any one of the adsorption isotherms, as shown in figm-es 1, 2, 3 and 4, may be chosen to illustrate the course of ex- pansion in relation to hydration. This course is also de- picted schematically in figure 5. The course of hydration and interlayer expansion of dehydrated and contracted montraorillonite particles (1 of fig. 5) along the adsorption isotherm may be described as occurring in five distinct steps. The first step consists of hydration of the exterior sur- faces of the particles to the extent of about 1 to 1.5 water molecules per unit cell of the crystal lattice, which is equal to 50 and 75 mg of water per gram of clay, or 180 and 270 sq. meters of a unimolecular layer of water. As the extent of the external surface of moiitmorillonite 05 06 07 08 09 RELATIVE PRESSURE P/Po Figure 3. Adsorption isotherms at SO'C of a Mg and a Li satu- rated Mississippi niontmcu'illonite. Co Olayiite , RELATIVE PRESSURE P/Po Figure 4. Adsorption isotherms at .j(l°C of two Ca** saturated montmorillonites of varying cation exchange capacity : Otaylite with 120 me. and Volclav with Ot) me. per 100 gm oven dry (100°C) clay. Clays and Clay Technology [Bull. 169 EXTEROR SURFACE O o«oooo o o NTERIOR ooooooooooooo o»O0O0Q«G00riO o«ooooo»ooooo oooo oooo o ""70_OO»C o o o«ooooo«o O QOO ooo o Figure 5. Schematic representation i)f tlio liydratiuii ami inter- layer expansion processes of montmorillonite. Blacli o = excliange- able cation. White o = water molecules. 1, Anhydrous and con- tracted stage ; 2, the initially hydrated stage ; 3, the initially expanded stage ; 4 and 5, advanced stages of hydration after the initial expansion ; 6, beginning of the second stage of expansion. particles of the Na* form, for example, is about 33 sq. meters per gram (Mooney et al. 1952), it follows that the thickness of the initial water layer on the outside surfaces of the particles in terms of water molecules is from 6 to 9. The exact thickness of this water layer may vary from one form to another but in all cases the initial hydration consists of a building up of a multimolecular water layer prior to interlayer expansion (2 of fig. 5). The second step consists of an expansion at the inter- layer position which is equivalent to a thickness of a uni- molecular layer of water ; it occurs after the formation of the multimolecular layer of water on the exterior of the particles. The expansion is accompanied by a distribution of the water from the exterior to the interlayer positions of the particles and results in the formation of a discon- tinuous unimolecular water layer at the interlayer posi- tion. The interlayer cations at this stage of hydration still remain attached to the oxygen sheets (3 of fig. 5). The third step consists of a completion of the inter- layer unimolecular water layers, accompanied by de- tachment of some of the interlayer cations from the oxygen sheets (the extent of the latter varies with the form), and by a reformation of the multimolecular water layer on the exterior surfaces of the particle (4 and 5 of fig. 5). The fourth step consists of another expansion at the ■interlayer position which is equivalent to a unimolecular water layer. As in the second step, the expansion is ac- companied by a redistribution of the already existing in- terior unimolecular and the exterior multimolecular water laj'ers with the formation of a discontinuous di- molecular water layer in which the water molecules are grouped around the interlayer cations (6 of fig. 5). The fifth step consists of a completion of the interior dimolecular water laver and of the reformation of a multimolecular water layer on the exterior of the parti- cles. The data available at present indicate that no further expansion occurs through adsorption of water vapor. However, as will be shown later, further expansion does occur when the clay is immersed in liquid water. The appearance during the second and fourth steps of d(00]) spaeings which do not correspond to a thickness of either a uni- or dimolecular water layer has been in- terpreted to represent the average thickness of mixed layer structures consisting of either unexpanded and expanded zones or zones of both uni- and dimolecular water layers (Hendricks and Jefferson 1938 ; Mering 1946). Interlayer Expansion and Hydration as Adjected by the Interlayer Cations. The effect of the interlayer cat- ions becomes apparent when the expansion is considered in relation to the vapor pressure or humidity at which it occurs. In samples saturated with cations of eqiial charge but varying in size, the larger the ionic radius the higher the relative humidity at which expansion occurs ; but the degree of hydration at the beginning of expansion is about equal for all sizes of ions. It is in- teresting to note that expansion beyond a unimolecular water layer does not take place in the montmorillonites saturated with K*, Rb*, or Cs* ions. In samples satu- rated with cations of equal radius but varying charge, the larger the charge the lower the relative humidity at which expansion occurs; but, again, the degree of hydra- tion at the beginning of expansion is about equal for all of the ions (figs. 1, 2, and 3). In samples saturated with the same cation but varying in amount, the larger the number of the cations, the lower the relative humidity at which expansion occurs ; but here too, when expansion has taken place, the degree of hydration is about equal for the different montmorillonites (fig. 4). The importance of the relationship between the vapor pressure, or the relative humidity, and expansion is that it offers a means of evaluating the interlayer attractive forces which hold together the individual lattice lavers (Cornet 1950; Katz 1933). The course of hydration and expansion as outlined above implies that the adsorbed water molecules are in a constant state of motion — particidarly during the exist- ence of ineomjilete uni- or dimolecular water layers. This is also implied by the fact that the adsorbed water pres- ent on internal surfaces must enter the interior of the particle bj^ wa,v of its edges. The significance of this mode of entry in relation to adsorption appears when the water adsorption properties of montmorillonite and kaolinite are compared. WATER ADSORPTION BY KAOLINITE From tables 2 and 3 it ma>' be seen that the degree of hydration of kaolinite at any given vapor pressure is very much less than that of montmorillonite when the degree of hydration is expressed on a unit-weight basis ; but it is considerably higher when expressed on a unit- area basis — particularly in the range where montmoril- lonite is in an expanded state. The unit-area expression, however, is the more valid one, as water adsorption is a surface reaction. This greater reactivit.y of the kaolinite surfaces with water mav be attributed to two causes : Part II] Properties of Clays 73 Table 2. Water adsorption hy a Ca-kaolinife and a Ca-montmoril- lonite at various relative hutnidities.^ Relative humidity (percent) Adsorption area m*/g clay d(OOl) spacing (A) Water adsorption Exchange cations (mg/g clay) (mg/1000m») (me.- /1000m') K* M* 27.6 437 27.6 810 27.6 810 27.6 810 27.6 810 27.6 810 M 12.9 14.5 14.8 15.2 15.4 15.4 K M 5.5 80 6.5 100 7.9 125 9.3 1G5 10.8 185 11.5 200 K M 199 183 235 124 286 154 337 204 392 228 417 247 K M 1.5G 1.11 1 18. 1.56 1.11 1 . 56 1.11 25 1 . 56 1.11 33 1 . 56 1.11 37... 1 . 56 1.11 t Data derived from Keenan. Mooney. and Wood (1951) and from Mooney, Kecnan, and Wood (1952. 1952a). • K = kaolinito; M = montmnrillnnite. (1) all of the adsorbing; surface in kaolinite is on the ex- terior of the particles, and as a result the water mole- cules in the vapor phase impinge directly on it ; whereas in montmorillonite the larger portion of the absorbing surface is an interior one, and as a result the water molecules in the vapor phase must first be adsorbed on the edges of the particle and then migrate to the interior — a process which tends to retard adsorption; and (2) the larger water adsorbability per imit surface area of the kaolinite could also be attributed, in part, to the larger cation charge per unit area (tables 2 and 3), if it be true that the ability of a surface to adsorb water is proportional to the cation exchange capacity per unit area. Table S. Water adsorption by a 7^a-Jcaolinite and a Na-montmoril- lonite at various relative humidities. f Relative humidity (percent) 8. 14 19 24 32 38 60 70 80 90 95 Adsorption area (mVg clay) d(OOl) .spacing K* M* M 18.6 33 10 18.6 33 10 18.6 33 10 18.6 440 12.4 18.6 440 12.5 18.6 440 12.6 18.6 810 14.2 18.6 810 15.0 18.6 810 15.2 18.6 810 15.4 18.6 810 15.4 Water adsorption (mg/g clay) K 3.0 4.2 4.9 5.4 6.4 7.2 8.0 10.0 13.0 20.0 25.0 M 20 45 65 80 100 115 170 215 240 280 365 (mg/lOOOms) K M 161 606 226 1364 263 1970 290 182 344 227 387 261 430 386 534 489 698 545 1075 636 1345 830 Exchange cations (me.- /lOOOm') K M 1.93 1.11 1.93 1.11 1.93 l.U 1.93 l.U 1.93 l.U 1 . 93 l.U 1 . 93 l.U 1 . 93 l.U 1 . 93 l.U 1.93 l.U 1.93 l.U THERMODYNAMICS OF WATER ADSORPTION AND DESORPTION OF MONTMORILLONITE Thcrmodyiiaiuic data relating to water adsorption and desori)tion reflect the course of the changes which the water and the clay undergo during hj^dration or dehy- dration. The thermodynamic values which reflect the changes of the water system in going from the adsorbed state to the free liquid state or vice versa are as follows: AFi = partial free energy change per gram of water AHi = partial net heat of desorption or adsorp- tion per gram of water AjSi = partial entropy change per gram of water. The thermodynamic values which reflect the changes of the clay system in going from the expanded state to the contracted state or vice versa during hydration or dehydration are as follows : AFo = partial free energj' change per gram of clay AHo = partial net heat of desorption or adsorp- tion per gram of clay. t Data derived from Keenan. Monney. and Wood (1951) and from Moonev. Keenan. and Wood (1!>52. lS.52a). • K ^ kaolinite: M = montmorillonite. WATER ADSORBED, MOLECULES H^O PER CELL FiGUKE 6. Thermodynamic quantities relating to the desorption of water from a Ba** saturated Mississippi montmorillonite at various stages of hydration and expansion (see text for definition of the quantities) . A definition of these quantities may be found in texts on thermodynamics. Figures 6, 7, and 8 illustrate the type of thermodynamic data obtainable. The data presented in figure 6, the desorption values for a Ba**-saturated Mississippi montmorillionite, are believed to reflect the course of Iiydration and expansion as already described. The high AHi value at the moisture content just prior to expansion (end of step 1) reflects an external water layer of well organized water mole- cules, a large proportion of which are grouped around 74 Clays and Clay Technology I Bull. I(i9 20-, Co OtOy ^' :6- O ,4. /^ Ca-M.s$ o < 6- 1 4. *1^ / ^^^3 2 2- r 1 1 1 r 1 WATER ADSORBED, MOLECULES H^O PER CELL Figure 7. Partial free energy change per gram of the solid of three Ca** saturated montmorillonites of varying cation exchange capacity at various stages of hydration and expansion. The numbers on the curves represent d(OOl) spacings in A. The exchange capac- ities are as follows: Ca (Otay) =120; Ca (Mississippi) =110; Ca (Volclay) =90 me./lOO g of dry (IWC) clay. I 23456789 WATER ADSORBED, MOLECULES HgO PER CELL Figure 8. Partial free energy change per gram of solid of a Mg**, a Ca**, and a Ba** saturated Mississippi montmorillonite at various stages of hydration and expansion. The numbers on the curves represent d(OOl) spacings in A. the externally held exchangeable cations. The sudden drop in Affi which follows expansion is believed to re- flect the condition resulting from the redistribution of the externall}' adsorbed water (step 2) with the forma- tion of a discontinuous internal unimolecular layer in which the water molecules are not associated directly with exchangeable cations ; that is, the exchangeable cations ai-e still attached to the oxygen surfaces within the cavities of the hexagonal network of oxygen ions (3 of fig. 5). The increase in the t^Ux which follows an increase in water content reflects the interaction of the water molecules with the exchangeable cations. The decrease that follows the maximum Aifi reflects a pro- portional increase in water molecules which are not associated directly with the exchangeable cations. The changes in entropy of the water, ASi, in going from the adsorbed state to the free liquid state (fig. 6), support the depicted changes in the state of organiza- tion of the water molecules presented earlier. For according to the third law of thermodynamics, an in- crease in entropy reflects an increase in randomness of the water molecules in going from the adsorbed state to the free liquid state, whereas a decrease in entropy reflects a decrease in randomness. Thus the maximum increase in entropy during desorption indicates that the water molecules at that state of hydration were at the maximum state of organization, that is, when grouped around the exchangeable cations ; and the maximum decrease in entropy indicates that the adsorbed water molecules were at the lowest state of organization, that is, having a considerable freedom of motion on the adsorbing surface. The changes in tsU-z, the partial net heat of desorption for the solid state (fig. 6), are believed to reflect the heat of reaction associated with contraction. The positive heat values are interpreted as reflecting the contraction or the coming together of the oxygen sheets, whereas the negative heat values reflect the read- sorption of the exchangeable cations by the oxygen Table .). Integral net heat of trater desorption from tuo montmo- rillonites and a rermictilite as affected hi/ total rhnrge on the crystal lattice and icater content. Sample Cation exchange capacity (me.*) Water content (g*) Mean integral net heat (cal/E HiO) Integral net heat (cal/g) Wet clay Dry clay Na-Vermiculite 200 120 95 4.32 4.32 4.32 670 4.50 280 28 • 18 11 29 19 Na-Volclav 12 Na-Vermiculite Na-0(aylite Na-Volclay 200 120 95 8.64 8.64 8.64 555 295 125 44 24 11 48 26 12 Ca-Vermiculite Ca-Otaylite 200 120 95 17.28 17.28 17.28 440 290 145 65 43 21 76 50 Ca-Volclay 23 • Per 100 g. dry (120° C.) material. Table .7. Mean integral net heat of water desorption from a Mis- sissippi montmorillonite at a given state of hydrntinn as affected hy the exchangeahle cation. Exchangeable cation Water content (g.*) Mean integral net heat (cal/g HiO) Mg** -- - 8.64 8.64 8.64 8.64 8.64 8.64 8.64 360 Ca+* 330 Ba** 245 H* 190 Li* 90 Na+ 100 K* - 40 • Per 100 g. dry clay (120° C.J Part 111 Properties of Clays 75 sheets; that is, the re-entry of the cations into the cavi- ties of the oxygen layers. Thus at low moisture contents, when the contraction involves the coming together of the oxygen surfaces without invqh'ing the rcadsorption oi the exchangeable cations, AH2 is positive. Zero Aff2 reflects the occurence of contraction associated witli the rcadsorption of a part of the exchangeable cations, with the result that the positive and negajjve heat reactions cancel one another. The negative AHo values indicate that the heat of rcadsorption of the cations exceeds the heat of contraction. Since AFo values during water adsorption (figs. 7 and 8) may be interpreted to represent the work done in bringing about expansion and the detachment of the exchangeable cations from the oxygen surfaces, it is not surprising to find that the greater the interlayer charge (fig. 7), the greater is the work done during expansion. The effect of the cation 011 AF2 (fig- 8) ap- pears as follows: For the initial expansion to a thickness equivalent to a unimoleeular water layer but without the detachment of the cations from the oxygen surface, the work done is larger the larger the cation, since as pointed out by the author (Barshad 1952), the inter- layer attractive forces are larger the larger the ex- changeable cation. On the other hand, since expansion equivalent to a dimolecular water layer represents not only the work done in separating the oxygen sheets but also the work done in detaching the exchangeable cation from the oxygen sheets, it is not surprising to find that the total work in the overall process is greater the smaller the cation, since the smaller the cation the stronger the force with which it is attached to the oxygen sheets (Barshad 1952) ; and consequently the work in detaching the cation is also greater. The integral net heat of desorption, which may be determined by differential thermal analysis (Barshad 1952a). is a useful thermodynamic value as a measure of the total change in energy during water desorption. The eft'ect on this value of such factors as the natiire of the exchangeable cation, and the total interlayer cliarge on the crystal lattice at identical states of hydra- tion and exjjansion are shown in tables 4 and 5. CLAY-LIQUID WATER SYSTEM IN THE GEL STATE The differences among the clays in their capacity to absorb water from the liquid state are as pronounced as from the vapor state. Several metliods are in use for measuring water ad- sorption and volume changes of a clay or clay-bearing material when brought in contact with liquid water. The apparatus designed by "Winterkorn and Baver (1934), Freundlich and co-workers (1932), and von Ensline (1933) is particularly' useful for measuring both water adsorption and volume change for small amounts of clay materials, and the Keen-Raczkowski box (1921; Russell and Gripta 1934) is useful for measuring volume changes and water adsorption for large amounts of soil. The water content at onset of gelation of clays which can be brought into stable suspension may also be de- termined b}' evaporating the suspension at a temperature of about 40°-50°C until it sets into a rigid gel. The water and solid contents are then determined by drying a known weight of the gel. From the density of the water and of the clay the total volume, as well as the ratio of the volume of the liquid to that of the solid, may be calculated — assuming that the density of the water is the same as in the free state. The data reported in tables (). 7. and 8 were obtained in tiiis manner. Water adsorption and swelling data for two mont- morillonite clays are given in tables 6, 7, and 9. The data in table 6 .show that gels of montniorillonite, at onset of gelation, contain from 5 to 30 times as much water Tahje G. Water content at onset of gelation of two montmoriUo- nites saturated with various cations, in grants TItO per gram of cJag. Saturating cation Source of montmorillonite Wyoming* California** H* 13-15 18-20 18-20 18-20 18-20 12-14 12-14 26-30 26-30 26-30 13-1.5 Li* 9-10 Na* 9-10 K* 9-10 NH4* 9-10 Rb* 3-6 C8+ 3-6 Mg++ . . U-12 Ca+*--- 11-12 Ba++ ..- n-12 • Clay Spur, Wyoming. •• Otay, California. Table 7. Ratio of liquid volume to solid volume at onset of gela- tion of tuo montmorillonites saturated uith various cations. Source of montmorillonite Saturating cation Wyoming California H+. Li+ - --- 35-40 48-53 48-53 48.53 48-53 32-37 32-37 69-80 69-80 69-80 35-40 24-27 Na* . . -— 24-27 K+ _ ._. 24-27 NHi* -- 24-27 Rb* 13-16 Cs* 13-16 Mg+* 29-32 Ca** 29-32 Ba+* 29-32 as solid. The montmorillonite gels saturated with the divalent cations contain from 50 to 100 percent more water than those saturated with the monovalent cations. A striking difference between the two montmorillonites studied appears in their capacity to retain water at on- .set of gelation. The Wyoming montmorillonite retained nearlv twice as much water as the California montmoril- 76 Clays and Clay Technology [Bull. 169 Table S. Water content at onset of gelation of two IXa-saliirated illites and a Na-saturated kaolinite of rarying particle size, in grams UtO per gram clay. Sample Water content Fithian illite <0.3;i -- 1.5 Fithian illite l-0.5;i . _ -. 0.9 1.4 1.2 1.0 S. C. kaolinite O.Sn-O.Svi... .. . . ... .. 0.7 S. C. kaolinite lii.-2n . .. .. ... ....... ... 0.7 S. C. kaolinite >2n ... 0.5 S. C. = South Carolina. II = micron. Taile 9. Interlayer spacing dfOOl) at onset of gelation of two montmorillonites saturated with various cations. Saturating cation Source of montmorillonite Wyoming California H+ — A 20.9 diff.* di£f.* diff.* diff.* 15.9 14.6 19.8 19.4 19.3 A 19.8 Li* -- - diff.* Na+ diff.* K+ 16.9 NHi* 15.7 Rb+ 16.0 C8+ — -1 12.8 Mg+* 19.5 Ca++ 19.3 Ba++ . 19.4 * diff. = diffused and of extremely weak intensities. lonite in all of its forms, except the Ii+-saturated form. Ill that form the water retained by both montmorillo- nites was nearly the same. The swelling associated with water adsorption at on- set of gelation is shown in tables 7 and 9. It may be seen from the data of table 9 that the intramicellar swelling for several of the montmorillonite forms is limited to about 10 A. This represents only about one-fold increase in volume. lii some of the forms the intramicellar swell- ing is less, whereas in others it may be considerably more, as indicated possibly by the diffused d(OOl) spac- ing. From these data and those in table 7 it may be con- cluded that the large macroscopic swelling at point of gelation of some of the forms is mainly a manifestation of intermicellar swelling rather than intramicellar swell- ing, and that the exterior surfaces of the particles rather than the interior surfaces make the macroscopic swelling possible. Further data which indicate that the hydration of the exterior surfaces of the particles is the cause for the macroscopic swelling are (1) those of Baver and Winter- koru (1935) on water intake and hydration of a Wyo- ming montmorillonite (Winterkorn and Baver 1934). and (2) those of Freundlich et al. (1932) and Mattson (1932) and the author (1950, and table 10 herein), which show the effect of free base and salt on water content and intramicellar .swelling of pastes of montmorillonites prepared either from stable suspensions or from tlie air- dried forms. Baver and "Winterkorn found that the hydration and the macroscopic swelling of montmorillonite are several times larger in gels prepared from suspensions than in gels prepared from the air-dried forms. On the other hand, the data on intramicellar swelling of montmoril- lonite in excess water (tables 9 and 10 herein; Barshad 1950, table 3) show that for most montmorillonite forms drying has no effect on the extent of the intramicellar swelling. This would indicate, therefore, that the de- crease in macroscopic swelling and hydration of mont- morillonite upon drying must be due mainly to its effect on the hydration properties of the external surfaces of the particles. Both Freundlich 's (1932) and Mattson 's (1932) data show that the macroscopic swelling of montmorillonite in salt solutions and in solutions containing free base is smaller the greater the concentration of the solutions. On the other hand, the data of table 10 show that the intramicellar swelling in several of the montmorillonite forms is not affected by the presence of soluble salts. It may be concluded, therefore, that the reduction in the macroscopic swelling in the presence of salts is a mani- festation of the effect of the salt on the water adsorp- tion properties of the exterior, rather than on the in- terior surfaces of the particles. A proper evaluation, therefore, of the over-all water adsorption and swelling properties of montmorillonite should account not only for the nature of the internal surfaces but also for the nature of the external surfaces of the montmorillonite particles. Taile 10. Interlayer spacing d(OOl) in water suspensions and in pastes made with normal chloride salt solution of the saturating cation of two montmorillonites saturated with various cations. Water suspensions Salt pastes Saturating cation Wyoming montmo- rillonite California montmo- rillonite Wyoming montmo- rillonite California montmo- rillonite n* A 20.7 diff. diff. diff. diff. 1.3.9 14.6 19.8 19.4 19.3 A 19.8 diff. diff. 16.7 15.7 16.2 12.8 19.5 19.3 19.4 A 20.9 19.1 19.3 19.5 19.3 19.3 A 19.7 Li+ 19.1 Na* . 19.7 K+ --- - 16.7 NHi* -- 16.7 Rb+ Cs+ Mg+* 19.0 Ca++ .. 19.2 Ba++ 19.4 ]':ll-t II Properties of Ceats 77 Data rctatiug: to water adsorption and swelling of other identified clay minerals arc presented in table 8. It is seen that both illite and kaolinite "gels", at onset of gelation, contain several times less water than any of the montmorillonite gels (compare tables 6 and 8). CLAY-LIQUID WATER SYSTEMS IN THE FLUID STATE The type of data useful for differentiating the clays in such systems include measurements of viscosity, thix- otropy, plasticity, stability, and flocculation (Freund- lich et al. ]!)."!2; Marshall lft49). Sucli measurements, however, will not be presented in tlie j)resent discussion. SELECTED REFERENCES Alexaiidpr. L. T., and Haring, M. M., 1936, Vapor pressure-water content relations for certain typical soil colloids : Jour. Phys. Chemistry, v. 40, pp. ]9.'j-20r>. Anderson, M. S., 1929, The influence of substituted cations on the properties of soil colloids: Jour. Agr. Res., v. 38, pp. 56.5-584. Barshad, I., 1948, A''erniiculitc and its relation to biotite as re- vealed by base exchange reactions, X-ray analyses, differential thermal curves, and water content : Am. Mineralogist, v. 33, pp. 655-678. Barshad, I., 1949, The nature of lattice expansion and its rela- tion to hydration in montmorillonite and vermiculite : Am. Jlin- cralogist, v. 34, pp. 675-684. Barshad. I., 1950, The effect of the iulerlayor cations on the ex- pansion of the mica type of crystal lattice : Am. Mineralogist, v. 35, pp. 225-238. Barshad, I., 1952, Factors affecting the inlerlaycr expansion of vermiculite and montmorillonite with organic substances : Soil Sci. Soc. Am. Proc. 1951, v. 16, pp. 176-182. Barshad, I., 1952a, Temperature and heat of reaction calibration of the differential thermal analysis apparatus: Am. Mineralogist. V. 37. pp. 667-694. Baver, L. D., and Horner, G. JI., 1933, Water content of soil colloids as related to their chemical composition : Soil Sci., v. 36. pp. 329-3.53. Baver, I>. D., and Winterkorn, H.. 1935. Sorption of liquids by soil colloids. II. Surface behavior in the hydration of clays: Soil Sci.. V. 40. pp. 403-419. Bradley, \V. F.. Grim. R. E., and Clark, W. !>.. 1937. A study of the beh.'ivior of mnntniorillonito npun wetting: Zeitschr. Krist.. V. !I7, pp. 216-222. Cornet, I.. 1950, Expansiim iif tbi' niontniuiillnnili' lattice on liydnition : Jour. Chemical Physics, v. 18, pp. 623-626. Dy.-il, R. S.. and Hendricks, S. P... 1950, Total surface of clays in Iiolar liipiids as a characteristic index: Soil Sci., v. 69. pp. 421-4.32. Freundlich, H„ Schmidt. O., and Lindau, G., 1932, tJber die Thixotropie von Bentonite-Suspensionem : Kolloid Beihefte, v. 36, pp. 4.3-S1. Hendricks, S. B., and .Tefferson, M. E., 19.38, Crystal structure of vermiculites and mixed-chlorites : Am. Mineralogist, v. 23, pp. 851- S62. Hendricks, S. B., Nelson, R. H., and Alexander, ly. T., 1940, Hydration mechanism of the clay mineral montmorillonite satu- rated with various cations : Am. Chem. Soc. Jour., v. 62, pp. 1457-1464. Hendricks, S. B., 1945, Base exchange of crystalline silicates: Ind. Eng. Chem., v. 37, pp. 625-630. Hofm.-uin. I'., and Bilke, W., 19.36, t)ber die inncrkristalline tiuellung und das Basenaustchverraogen des Montmorillonite: Kol- lard-Zeitschr., v. 77, pp. 238-251. Katz, J. R.. 1933, The laws of swelling : Faraday Soc, Trans., V. 29. pp. 279-300. Keen, B. A., 1921. The system of soil-soil moisture: Faraday Soc. Trans., v. 17, pp. 228-243. Keen, B. A., and Raczkowski, H. J., 1921, The relation between the clav content and certain physical properties of a soil : Jour. Agr. Sci., v. 11. pp. 441-449. Keenan, A. G.. Mooney. R. W., and Wood, L. A., 1951, The re- lation between exchangealde ions and water absorption on kaolinite: Jonr. I'bys. ••ind Colloid Clicniistry, v. 55, pp. 1462-1474. Kuron. II.. 19.32, Absorption von Hampfen und Gason an Boden und Toni'n und ihre Vermendung zur al)erflachonernutllung diesen Stuffe: Kollod. Beihefte. v. .36, pp. 178-2.56. Maegdefrau, E.. and Hofniann. V.. 1937, Die Kristallstruktur des Montmorillonite : Kolloid. -Zeitschr.. v. 77, pp. 238-251. Marshall, C. E.. 1949. The colloid chemistry of the silicate minerals. Academic Press Inc., New York, N. Y. Mattson. S., 1932. The laws of soil colloidal behavior: VIII. Forms and functions of water: Soil Sci., v. 33. jip. ;!()! 323. Mering, .!., 1946. On the hydration of montuKU-illonite : Faraday Soc. Trans., v. 42B. pp. 205-219. Mooney. R. W., Keenan, A. G., and Wood, L. A., 1952. Adsorp- tion of water vapor by montmorillonite. I. Heat of de.sorption and application of B E T theory : Am. Chem. Soc. Jour., v. 74, pp. 1.307-1371. Mooney, R. W., Keenan, A. G., and Wood, L. A., 1952a, Adsorp- tion of water vapor by montmorillonite. II. Effect of exchangeable ions and lattice swelling as measured by x-ray diffraction: Am. Chem. Scic. .Tour., v. 74. jip. 1371-1374. Nagelschmidt. G., 1936, On the lattice shrinkage and structure of montmorillonite : Zeitschr. Krist., v. 93, pp. 481-487. Nelson, R, A., and Hendricks, S. B., 1942, Specific surface of .some clav minerals, soils, and soil colloids : Soil Sci., v. .56, pp. 285-296. " Pari. A. N., 1949, Soils their physics and chemistry, Reinhold Publishers Corp., New Y'ork, N. Y. Ross. C. S., and Hendricks, S. B., 1945, Minerals of the mont- morillonite group, their origin and relation to soils and clays: U, S. Geol. Survey Tech. Paper 205B, pp. 23-77. Russell, E. W., and Gupta, R. S., 19.34, On the measurement of imbibitional water: .Tour. Agr. Sci., v. 24, pp. 315-325. Thomas, M. D., 1921, Aqueous vapor pressure of soils : Soil Sci., V. 11, pp. 409-434. Thomas, M. D., 1921a, Aqueous vapor pressure of soils: II. Studies in dry soil : Soil Sci., v. 17, pp. 1-18. Thomas, M. I).. 1928. Aqueous vapor pressure of soils: III. Soil structure as influenced bv mechanical treatments and soluble salts: SoO Sci., V. 25, pp. 409-418. Thomas, M. D., 1928a, Aqueous vapor pressure of soils: IV. Influence of replaceable bases : Soil Sci., v. 25, pp. 485-493. von Enslinc, O., 1933, tJber einen apparat zur messung der flus- sigkeitsaufnahme von quellbaren und porosen stoffer un zur charak- terisierung der benezbarkeit : Chem. Fabrik., pp. 147-148 (Chem. Abstracts, v. 27, p. 2.344). Winterkorn, H., and Baver, L. D., 1934, Sorption of liquids by soil colloids. I. Illiquid intake and swelling by soil colloidal mate- rial : Soil Sci., v. 38, pp. 291-298. INTERLAMELLAR SORPTION BY CLAY MINERALS By Douglas M. C. MacEwan * The term "iuterlamellar sorption" was proposed by the author in 1948 to designate the penetration of ions and neutral molecules between the layers of a lamellar structure, resulting in a reversible, one-dimensional swelling. The first system of tliis sort to be studied was graphitic acid (Hofmann et al. 1930), which reversibly takes up layers of water molecules. The phenomenon is mainly met with amongst natural and artificial inorganic substances — in the organic field, a commoner analogous phenomenon is the swelling of polymers by penetration of small molecules between large, chain-shaped mole- cules, although lamellar systems, in some respects analo- gous to the inorganic ones, are met with among proteins and soaps. The interest of the phenomenon is partly due to the light it can throw on adsorptive and surface forces. Alterations in spacing between the layers can very read- ily be followed by means of X-ray diffraction, since sharp reflexions are given by the sequence of parallel planes. This means that a range of relatively powerful methods is available for studying the size and orientation of the sorbed molecules, and hence the nature of the forces holding them. Most of the work on interlamellar sorption, especially of organic molecules, has been done using the clay- mineral montmorillonite as substrate, and hence most of this paper will be concerned with this mineral. However, it will be shown that the phenomenon is of fairly fre- quent occurrence, a number of otlier suitable substrates being obtainable. Attention was first directed towards the reversible swelling shown by montmorillonite on sorption of water (Hofmann et al. 1933). It was shown that the spacing of the layers varies (1) with the equilibrium vapor pres- sure of water; (2) with the exchangeable cations, wliich are always present in samples of montmorillonite in quantities of the order of 100 me. '100 g. The early results obtained by Nagelschmidt, and by Hofmann and Bilke, appeared to show a continuous change in spacing with water content, but Bradley, Grim, and Clark (1937) obtained evidence for the existence of definite hydrates with definite niimber of monomolecular sheets of water molecules between the montmorillonite layers. The eai'lier results can be explained on the basis of the existence of mixed structures, in which, for instance, one-sheet inter- vals and two-sheet intervals may occur in a random man- ner. The phenomenon, however, is more complex than Bradley, Grim, and Clark suggested ; not only may mono- molecular sheets of water molecules occur, but also, as Mering has shown, groupings of water molecules aroiind the interlamellar ions. Tlicse questions are being treated in this symposium by Barshad. who has produced much new work in this field. It is an indication of the com- plexity of the phenomenon that such new work can still be produced, 20 years after the original papers of Hof- mann and his collaborators. In 1941, Gieseking and Hendricks showed that the ex- changeable cations of montmorillonite may be replaced by large organic cations-amines and polyamines. These • Pedology Department. Rothamstead Experinient Station. Harpen- den, Herts, England. large cations give rise to characteristic spacings between the montmorillonite layers, which may be explained by supposing the organic cations to go between the layers, and in general to lie as nearly as possible flat on the surface of the layers. The ions are of course held to the layers by electrostatic forces, the layers being nega- tively charged, and the ions positive ; Hendricks sug- gested that van der Waals forces between the uncharged organic groupings and the montmorillonite layers are also of importance, and contribute towards the increased stability of these comjilexes, as compared with those formed with inorganic cations. Hendricks showed that the observed spacings between the layers are attributable to definite orientations of the large cations, and are com- patible with the supposed type of binding between the cations and the layers. In this he broke new ground in the detailed interpretation of inter-layer spacings. As a logical consequence of Hendrick's postulate re- garding binding forces, one would expect neutral mole- cules of fairly large size also to be strongly adsorbed by montmorillonite, and Bradley (19451 and the present author (1948) have shown that this in fact happens. Apparently any molecule of moderate size, which has a dipole moment may be sorbed, and also certain mole- cules, like benzene, which have not got a permanent dipole moment. The sorbed molecules do not appear to be in a random arrangement, but form well marked la.vers, with probably a fairly definite orientation for the molecides composing a layer. The inter-layer spacings observed with different sorbed molecules are thus in principle capable of being fully explained by "the orien- tation and method of binding of the sorbed molecules (MacEwan 1948), though the explanation may not be obvious in any particular case. These organic molecules are sorbed in neutral form, so the complexes tliey form are quite different from those with large organic cations, investigated by Hendricks. There are in general three elements in the montmoril- lonite sorption complex, namely (1) the silicate layers, (2) the interlamellar ions, (3.) the interlamellar neittral molecules — water, in the natural mineral. To some ex- tent, eitlier (2) or (3) may be replaced independently of each other. Bloch (19ii0) has claimed that neutral inorganic mole- cules may also be sorbed between the layers of mont- morillonite In this ease, the inter-layer cations are supposed to be still present, together with molecules such as, for instance, AgCl, giving a complex of the type Mt-NIl4* (AgCl) where Mt" stands for the negatively charged layer of montmorillonite. The NH4* is replaceable by other ions. However, the amount of AgCl appears to be limited to a maximum of one molecule per ion, so the possibility that the AgCl is here forming a component of a complex ion must be considered. This is also suggested b,v the fact that on replacing the NH4* by Na* or Li*, complexes of inferior stability are obtained, from which AgCl grad- uallj' separates out. The ammonium-saturated complex gives a spacing of 12. 6A, and does not swell further by imbibition of water. ( 78) Part III Properties op Clays 79 The niethotl by which these complexes ai'e obtained consists in saturation with silver to give Mt" Ag*, fol- lowed by treatment with e.g. NH4CI solution. An an- alogous reaction, giving rise to a product with properties similar to those of the natural ehlorites. has been studied by Ilenin and CaiUcre (1!»4;)). The treatment consists in saturation of the montmorillonito with ^Ig", followed by the application of eonoentratetl XII4OII solution, giving rise to the probable reaction. Mt= Mg- + 2NH4OH -^ Mt= (NH4*)2Mg(0H)2 The resulting i)roduct. with interlainellar ]\Ig(OH)o, differs, however, from natural ehlorites in having NH4* cations present as well. These reactions raise the question whether free anions can penetrate between the montmorillonite layers. The existing evidence is hardly sufifieient to enable an exact mechanism to be postulated for the reactions, but it is possible to fornndate both reactions in such a way as to avoid the presence of free anions between the layers. In the sorption complexes mentioned so far, the inter- lamellar material is held in two ways, either by electro- valency (ionic exchange), or by residual valencies (di- pole and A-an de Waals forces). Berger (1941), Gieseking (1949). and Deuel et al. (1950, 1952) claim to have pre- pared a third type of complex in which organic radicals are directly linked to the montmorillonite surface by homopolar bonds. The relationship between this type of complex and the ionic type would be similar to that between dimethyl sulphate (€^3)2804 and sodium .sulphate (Na*)2S04=. Berger claimed to produce his methyl montmorillonite by reaction of the acid clay with diazomethane : Mt-H* + CH3 — R -^ Mt-CHs + H — R Deuel (1952) has investigated a number of other methods of introducing organic and other radicals, and claims to have produced methyl and other derivatives of mont- morillonite, montmorillonite chloride, methoxy-mout- morillonite and numerous other derivatives. The reality of some at least of these derivatives has. however, been questioned by Brown, Green-Kelly and Norrish (1952), on the grounds (among others) that the acid mont- morillonite commonly used as a starting material is prob- ably unstable ; that the difficulty of obtaining truly dry montmorillonite has been underestimated ; and that no adequate evidence based on X-rays has been presented to show that the resulting products are not residual-valency complexes of the type of methanol-montmorillonite. The difficulty of adequate characterization of the complexes is aggravated by the fact that the homopolar derivatives, if they exist, are likely to be highly unstable. Methjd- montmorillonite, for instance, in presence of water, could lirobably decompose to give methanol-h.ydrogen-mont- morillonite: Mt-CHs + H2O -^ Mt-H*-CH.,OH. Confusion on this question has probably been increased by Berger 's original attempt to present the organic re- actions as a "proof" for the truth of the Edelman- Favejee structural hypothesis, which postulates the existence of OH groups linked directly to silicon. These reactions are interesting in themselves, but until much more is known about their mechanisr.i. they cannot be presented as a proof for any structural hypothesis. In particular, careful X-ray work is needed on well- delined material. At present, no attempt has been made to interpret the few spacing measurements that are available in terms of the dimensions and orientation of the organic groupings allegedly present, nor have any one-dimensional fourier syntheses been made from X-ray data. If well-defined complexes do indeed exist, then both things should be possible. At present, the existence of these complexes is possible, but not conclusivel.y proved. If, however, they really exist, then they would open up a most exciting new method for the study of organic re- actions, since a big variety of such reactions involving organic radicals could be reproduced on the surface of montmorillonite, and their progress studied directly by X-ray diffraction. Methods of Study of Interlamellar Sorption Com- plexes. The X-ray diffraction method is at present the most valuable technique for characterizing interlamellar sorption complexes. The simplest measurement one can make by its means is that of the (001) spacing, derived by measi;rements of the inner reflexion of the X-ray diffraction diagrams. A considerable amount can be de- duced from this measurement alone if comparison is made between several related complexes. These may result from variation of any of the several factors in- volved in the formation of these complexes — sizes of the molecule, the nature of the interlamellar cations, the amount of sorbed material, etc. 37 36 35 34 33 ^ / 1 / \ / 1 \ 1 I \ \ \ M \ \ ^ \^ \ i s > \ )^^ • i \ • "-1 \ 8 10 12 14 16 Figure 1. Variation of basal spaciuj; with chain length, for complexes of Ca-montmorillonite, with monoh.vdric alcohols, from data of MacEwan (1948) (open circles) and Barshad (lO.VJ) ( black circlies ) . As an example, two plots are shown illustrating the variation of (001) spacing with chain length of absorbed organic molecules, derived from results obtained bv MacEwan (1948, 1946), Barshad (1952) and Jordan (1949). In the first diagram (fig. 1) the adsorbed mole- cules are alcohols and therefore neutral: in the second 80 Clays and Clay Technology [Bull. 169 of bosol plane O'eo covertd by omine 265 34 415 49 56 5 6 8 10 12 Number of C oloms in amine cnnin Figure 2. Variation of basal spacing with chain length for complexes of montmorillonite with straight-chain amines, from Jordan's data (1949). (fig. 2), they are amines, and present in the form of positive ions. These graphs at onee suggest two things: (a) the spacings do not vary eontinuonsly, but alter in jumps, so that sheets of molecules are probably present ; (b) if we subtract the thickness of the montmorillonite layers, the larger spacing is in each case about twice the smaller one, so that single and double thicknesses of molecules are probably present in the two eases. Further, the actual value of the space available for the molecular sheets suggests that in each ease the molecules are lying as flat as possible on the montmorillonite surface. Further, it will be noted that in the first graph the sj^acing initially goes up with increasing number of carbon atoms, and in the second it goes down. This straightaway .suggests a different mechanism for adsorp- tion in the two cases. In the first case, the number of A(A.u) /y . ^- -- •' - "^ '*~ a ._ — _ rf t 1 / *■ 0.2 0.4 0.6 0.8 P/Po Figure 3. Variation of spacing (A = dooi-9.4A) with partial pressure of acetone vapor in contact with Ca-montmnrillonite, for increasing and decreasing vapor pressures, as indicated li.v arrows. The dotted lines give the corresponding sorption isotherms. From Glaeser (1950). layers is determined by the energy of adsorption, and this goes down as the non-polar part of the molecule becomes bigger ; in the second, the number of layers present depends on the space occupied by a quantity of molecules sufficient to balance the charge on the mont- morillonite surface. Still more detailed suppositious can be made, with a rather smaller degree of conviction. Thus for the alco- hols, it is observed that there is a slight increase in spacing from methanol to ethanol, and a similar increase is observed with the glj'cols ( Bradley, 1945 ) . The author has attributed this to the influence of the more intense binding of the a-carbon atoms to the montmorillonite surface, as compared with the other carbon atoms in the chain (a conception which was formulated by Bradley). Barshad (1952) has recently found very high spacings with nonanol and decanol, and these are shown on the graph. They must correspond to a molecular rearrange- ment, the exact nature of which is not at present clear. The (001) spacing may be plotted against other vari- ables. It may for instance be plotted against the partial vapor pressure of the adsorbed substance, and in this way comijared directly with adsorption isotherms. Fig- ures 3 and 4 show two plots of this type, taken from results obtained by ]\liss Glaeser (1950), for acetone adsorption by calcium and sodium montmorillonites. The dotted lines show the corresponding adsorption isotherm measurements. In each case two spacings are observed. With the desorption isotherm of the Ca-montmorillonite the transition from A ^ 7.4 A to A = 3.5 A is sharp, and this corresponds to a sudden drop in the desorption isotherm. In the case of the Na-montmorillonite a region of mixed spacings is formed between A = 8.8 A and 10 A(A u) J / 8 f 6 / / r/ X ^<^^ / A-—- — ^ ^ A ~~^ 2 — t — / / 0.2 04 06 0.8 P/Po Figure 4. Variation of spacing (A = do<,i-9.4A) with partial pressure of acetone vapor in contact with Na-montmorillouite, for increasing and decreasing vapor pressures, as indicated by arrows. The dotted lines give the corresponding sorption isotherms. From Glaeser (1950). Part II] Properties op Clays 83 Charge per unit cell Figure 5. Basal spacing' as a function of fliai;;i' on ihe layer, for Ca**-water-saturatfcl minerals ( lilaci; circles i and Xa*- glycerol-saturated minerals (open circles). From data of Mering, MacEwan, and Uarshad. Actual minerals represented in this graph are pyrophyllite (charge 0) ; montmorilhmite (charge about S unit per unit cell) ; vermiculite (charge 1 to IJ units per unit cell) ; mica (charge 4 units per unit cell). The actual degree of expansion varies with the exchange cations and neutral molecules present, but the general tendency is towards the sort of curve shown with a region of high maximum expansion, and no expansion for either very low or high charge. A = 3.3 A and this corresponds to a gradual drop in tlie desorption isotherm. The value A = 7.4 is sufficiently near to 2 X 3.5 to make it likely that the two spacings observed with the Ca-niontmorillonite correspond to structures with one and two layers of acetone molecules respective^. The value A = 8.8 A observed with the Na-montmorillonite liowever probably corresponds to a different molecular arrangement. The hysteresis which is shown by the adsorption and (U'sorption isotherm is reflected in the (OOl) spacings. Certain other features of the adsorption and desorption isotherms have no direct correspondence in the X-ray spacing plots. It is clear, however, that these provide valuable additional information in conjunction with the isotherm data. It would be instructive to plot a grajjli showing the variation of spacing with charge on the layer. The last- mentioned quantity is of course a constant, not sus- ceptible of experimental variation, but it is possible to find a range of minerals of similar structural type, and with different charges on the layer. In making a com- parison, of course, all the other factors, in particular the interlamellar cations and sorbed molecules, must be kept constant, and it is therefore difficult to assemble a set of comparable data. The graphs in tig. 5 are based on data for five mineral types — pyrophyllite with zero, or near zero, charge on the layer, which shows no ex- pansion without special treatment (Caillere and Henin, 1950) ; montmorillonite with a charge of about s unit per unit cell layer; vermiculite with a charge of about 1 unit to 1^ units; mica with 2 units; and brittle micas with 4. The last-mentioned group also shows no expan- sion. The graphs show a clear tendency for the expansion to rise to a maximum at a charge near to that of mont- morillonite, and fall away for higher charges. The maxi- nnnn may however occur at lower charges than that of montmorillonite, since the corresponding minerals are not available. There is therefore probably an optimum charge for interlamellar expansion, and this maj' be near to the charge of montmorillonite. Barshad, in his con- tribution to this series of papers, gives some consid- eration to the implications of this fact. We have so far been concerned with what can be learned by a consideration of the (001) spacings only. More information can be obtained when the whole se- quence of (001 ) reflexions is taken into account. This may at the very least serve to draw attention to the cases where the apparent (001) reflexion does not rep- resent a true spacing, but results from a mixed struc- ture of two or more spacings. If a true single spacing is present, the higher-order reflexions will occur at integral sub-multiples of this, of the form d, d/2, d/3 etc. If the reflexions do not form a rational series of orders, then a mixture of spacings is present, and special methods are required for the interpretation of the diagram. If a single spacing is indeed present, it is possible to perform a one-dimensional fourier synthesis, using the measured spacings and intensities of the series of basal reflexions. Bradley (1945) has shown that even a rough estimate of the sequence of intensities may enable a dis- tinction to be made between two-layer and three-layer structures. "We have been trying to improve on this re- sult by increasing the number of available intensities, and the accuracy of their measurement. A small oriented flake of thn treated montmorillonite is used, and a series II 4» mm- lO* 20*^ 30' FniUKE C. X-ray diagrani.s of a gl.voerul-montmorillonite flake, with the flake set at different angles to the axis of the X-ray licara (indicated on the right). The flake was turned towards the right: note that on this side the higher orders of the basal reflexions ap- pear as the angle of turning is increased. At the same time, on the other side, general hk reflexions appear and the basal re- flexions diminish in intensity. 82 Clays and Clay Technology [Bull. 169 FiGUKE 7. One-dimeusional fourier synthesis of Tl*-glycerol- montmorillonite. This diagram covers about one whole repeat distance (17.7A). There are centres of symmetry at O and B, and about these two points the diagram repeats itself. The portion AA represents the silicate sheet, and is about the same for all the montmorillonite complexes. The probable significance of the peak in this region is indicated. The peaks marked G are probably due to glycerol molecules, and the small peak above the numeral 9 to the Tl* exchangeable cations. of dift'raetioii diagrams is recorded, with the flake set at dift'erent angles to the beam. Figure 6 shows a set of such diagrams for glycerol- treated montmorillonite, obtained b.y G. Brown (1950). It is possible from such a set to plot a curve showing intensity of reflexion as a function of orientation, and thus to eliminate the effect of preferred orientation, though certain complicating factors, especially absorp- tion in the specimen, have to be allowed for. To certain extent direct metlu.)ds may be used for de- termining the phase of the reflexions, so that the degree of ambiguity is relatively small. The difficulty arises in interpreting the features observed in the fourier dia- grams. Figure 7 shows a fourier synthesis obtained from a glycerol-moutmorillonite, with Tl* as exchange cation (Brown, 1950). The part on the left up to the point marked A is a uuiver.sal feature of the montmorillonite cui-ves, and corresponds to the silicate layer. The rest of the diagram represents the material between the layers — glycerol molecules, and T1+ and other cations. There is reason for believing that the peak marked B corre- sponds to the Tl* ions. Mechanism, of InterlameUar Sorption. The sorption of large complex ions may be ascribed to an ordinary cation-exchange process, the ions being held to the nega- tively charged sheets by electrostatic forces. Hendricks (1941) suggests however that van der Waals forces play an important part in determining the extra stability of such complexes, and the flat orientation of the molecule. Van der Waals forces are probably of importance in the case of adsorption of neutral molecules, but Bradley (1945) has suggested another mechanism, namely hydro- gen bonds between the a-carbon atoms and the surface oxygen atoms of the mineral. There is considerable evi- dence for the reality of such a bond, tliough it does not rule out the existence of — H — bonds and N — H — bonds as well. These forces provide an explanation of the attachment of neutral molecules to the sheets, and therefore of the repulsion between them. The balancing attractive force may be partly electrostatic in character, and partly van der "Waals attraction between neighboring layers, but the electrostatic force ajipears to be the only adequate one at sliort distances. We have recentlv been making calculations on the probable variation of the electro- static attractive potential with distance, and find that, as one would expect, it dies off very rapidly for a surface with high charge density, less rapidly for one with low charge density (fig. 8). At separations of the order of 100 A or so, the Gouy layer repulsion and van der Waals attraction may be the only forces of impor- tance. It follows from this work that a suitable value of the surface charge density is essential for these inter- A.U FlGtTBE S. Potential energy of the configuration I charged .sheet) -ions- (charged sheet) as a function of distance between opposite surfaces of the sheets. Curve 1 refers to a sheet having ten times the surface density of charge of montmorillonite ; curve 2 to montmorillonite ; and curve 3 to a sheet having one-tenth the surface density of charge of montmorillonite. Any calculation such as this depends of course on a number of assumptions, which cannot all be specified here. Part 11] Pkoperties of C'i.a\> 83 lanipUar adsorption complexes to be formed. If the sur- face charge density is too high, the attractive force will not manifest itself except at very close distances. If the surlace charge density is too low, the attractive force will be effective at fairly large distances, but will not be large enough to equilibrate the repulsive force at any distance. In this case, there will however be a position of equi- librium for very close approach in which the van der Waals attractive force is balanced by Born repulsion. This is the only type of equilihriinu for close appi'oach which is recognized by Verwey and Overbeek (1948) who do not postulate either an adsorption-repulsion, or a short-range electrostatic attraction. A complicating factor which must be taken into ac- count is that the interlamellar molecules interact with the exchange cations as well as with the clay-mineral surface. This is clearly shown by the fact that the num- ber of sheets of certain organic liquids which jienetrate l)etween the moutmorillonite layers is dependent on the exchangeable cation, and may differ even for two cations of the same size. It seems likely that this type of short-range equilib- rium is of general occurrence when charged surfaces are brought together in a liquid medium, and therefore that it ought to be taken into account in explaining colloidal phenomena. The forces considered here are additional, and not alternative, to those considered by Verwey and Overbeek in their book Theory of the Sta- bility of Lyophobic Colloids. Interlamellar Sorption by Other Materials. The phe- nomenon of interlamellar sorption is not characteristic of moutmorillonite alone. A range of materials shows an essentialy similar behavior, though for most of them 4.7 7.5 X Figure 9. Hypothetical structure of the a zinc hydroxide- uaphthol yellow complexes with neutrnl organic molecules, ac- cording to Talibudeen and MacEwan (1949). The double rows of circles represent the zinc hydroxide layers, the long fiat molecules are the dyestuff, and the small shaded molecules are the neutral substance. the details of the process have not yet been investigated. Graphitic acid was in fact studied by Ilofmann earlier than moutmorillonite, and it shows a very similar swell- ing with water, though there is no report of its inter- action with organic li(]uids. At Ixothamsted. we liave studied a-zinc hydroxide (MacEwan and Talibudeen 1949), a system which shows a complicated type of ad- sorption. Apparently the hydroxide surface first covers itself with a double layer of negatively charged dye molecules, and neutral organic molecules are then taken up between these (fig. 9). It is not known whether posi- tive ions are present, though this seems probable. We have also studied the lialloysite system. It is still however, uncertain whether the adsorption of organic molecules is in this case analogous to that by montmoril- lonite, i.e., whether exchangeable ions play a part. A large cation-exchange capacity has been reported for lialloysite by Riviere (1950), but this is in coidlict with the findings of other workers. The hydrated calcium phosphate, autunite lias been investigated here by G. Brown. This shows reversible interlamellar sorption of water, but the organic liquids we have tried, while effective in removing water, are not themselves sorbed. Henin and his collaborators (1950) have recently been doing some very interesting research on a number of materials having layer structures, and have found that many of them will show interlamellar sorption of water after suitable treatment. The materials for which the phenomenon has been demonstrated are hydrargillite, bayerite, brucite, barite, antigorite, and noumeite. The range of materials showing interlamellar sorption has thus been greatly extendecl, but this work can only be regarded as preliminary in character, and a good deal of additional research will be needed to clarify what is happening as a result of the treatments used. For the sake of completeness, one ought perhaps to mention the reversible swelling of protein crystals by penetration of layers of water, as observed by Perutz and others (1947), and the lamellar soap systems studied by McBain (McBain and Marsden 1948) although these systems maj- not be analogous to the ones just men- tioned. Proteins also form complexes, of the ionic type, with moutmorillonite, and these have been studied in our laboratories by 0. Talibudeen (1950), who found that a structure is readily formed in which the protein mole- cules appear to be fully stretched out on the surface of the moutmorillonite sheets, giving a relatively low spac- ing of about 18 A. DISCUSSION W. T. Cardwell, Jr.: There are great differences between different montmorillonites. Some swell no more than kaolinite, so it seems meaningless merely to say that a sample is moutmorillonite. I would like to hear more about the differences between montmorillonites. M. D. Foster: Among 12 samples of montmorillonite that I have studied, from different localities in the United States and Mexico, three-fold ilifferences in macroscopic swelling were found. No correlation was found between exchange capacity and swelling but correlation was found between substitution in the octahedral group of the structure and swelling. In this respect ferric Fe, although it does not affect the charge, had as much influence as Mg. Three of the samples, having the same tetrahedral charge and the same Mg substitution, 84 Clays and Clay Technology [Bull. 169 but containing different amounts of ferric Fe. ranged from 21 to 43 mm per gram in swelling capacity. Two other samples had the same composition and the same total Fe content but differed in the state of oxidation of the Fe. The .sample with 2/3 ferrous Fe and 1/3 ferric Fe swelled only two-thirds as much as the sample with 2/3 ferric and 1/3 ferrous Fe. Thus chemical composi- tion must be considered in arriving at an interpretation of the relative swelling of different samples nt montmorillonite. G. W. Brindley: From what Dr. Foster has said it appears that we may have to distinguish between two kinds of swelling in montmorillonite ; swell- ing that results from water penetrating between layers and swelling arising from water between particles. In the former there would be regularly spaced layers; in the latter the particles would have vari- ous orientations with respect to one another. Barshad emphasized that in regard to the swelling of montmorillonite in water vapor, the factors of importance are the charge on the layers and the size and charge of the interlayer cations. It might appear that Dr. Foster's results were inconsistent with Barshad's views, but she was dealing with Na-montmorillonite in the presence, not of water vapor, but of a considerable excess of liquid water. There is there- fore no conflict. It is of interest that M^^ring has shown that Na-montmorillonite placed in water swells to such an extent that the basal spacing lncrea.ses up to 20 A. Beyond that it is diflicult to follow. The struc- ture swells to the point where there are individual sheets of mont- morillonite with no spatial regularity between them (Mering 1946). The situation then would be quite dift'erent from that referred to by Barshad. Would Dr. Foster state the conditions of her experiments in view of the possibility of the different mechanisms of the swelling? M. D. Foster: One gram of air-dried Na-montmorilbniitp was ground to pass a 20-mesh sieve, then dropped a few particles at a time into a cylinder containing water. Five minutes later, when it was assumed that these particles had come in complete contact with water, a few more were added. This was repeated until all the sample bad been intro- duced into the cylinder. The cylinder was then allowed to stand 24 hours and the height of the clay in the cylinder read. D. M. C. MacEwan: I feel very doubtful whether Na-montmorillonite layers are ran- domly dispersed when it is fully swollen. K. Norrish (1954), at Rothamsted, has been making experiments with swollen systems : with the highest degree of swelling observable in a very dilute NaCl solution there was still orientation. In that conditi(ni the spacing between layers was about 120 A. I have considered the question of the degree of disorientation of neighboring layers and the curvature that are compatible with good X-ray patterns. Both appear to be small. Thus, in highly swollen Na-systems the layers may still be oriented parallel to each other. R. A. Rowland: What is the distinction between the two kinds of swelling? D. M. C. MacEwan: Norrish's experiments were made on a Na-system. The material placed in plastic tubes was thoroughly leached with chloride solu- tions of different concentrations. Plotting the d(OOl) spacings against concentration he found that the spacings go up to about 20 A with decreasing concentrations. But at about 20 A the spacings become broader and weaker until at considerably less concentration a spacing of about 35 A appears and this increases with decreasing concentration up to about 100 A. It appears that at certain concen- trations there are mixed structures with both high and low spacings simultaneously present. At about 20 A the layers seem to separate layer from layer to a new equilibrium jiosition at considerably higher dilution. We have considered i>ossilile explanations for this, but as yet have found none that is entirely satisfactory. W. T. Cardwell: In the suspension where a spacing of about 100 A was reported, were the layers still parallel in the suspension, were they merely oscillating at random, or was it a matter of an average distance? D. M. C. MacEwan: The layers of montmorillonite must have been rather well ori- ented parallel to one another, in order to account for the observed diffraction effects. I. Barshad: In regard to Rowland's question as to where intramicellar swell- ing ends and intermicellar swelling begins — it depends on how the experiment is made. Starting with a dry Ca-system and using water vapor, intramicellar .swelling seems to end with a relative humidity of about 50 to 60 percent, but in the presence of liquid water inter- micellar swelling also takes place. In other words, liquid water ap- pears to be essential for intermicellar swelling. R. A. Rowland: In the dry state there are outside surfaces and many more inside surfaces. Upon swelling to about 20 A, surfaces that were inside .surfaces become outside surfaces. There is, then, a discontinuity in the X-ray pattern and a change in chemical behavior, either gradual or slight. I do not believe there is any justification for the distinc- tion made between inside and outside surfaces unless it can be demonstrated that the inside surfaces take new properties upon be- coming exposed. The effect is probably due to distance from the surface, and the designation of inside and outside is confusing. M. Soldate: In the literature, at least by implication, there is the statement that certain illites swell (Hughs and Pfister 1947). Is this a case of intralamellar swelling in which the spacings change? P. G. Nahin: Graphic illustrations are given in the literature showing that the settling volumes of montmorillonite, illite, and kaolinite diminish in that order. From our experience it seems that one reason why Na- illite swells less than Na-montmorillonite is that the mechanical and chemical forces that bold the layers of illite together are too strong to permit water to enter between layers. R. C. Mielenz: It is well known that montmorillonite-type clay, and some illite types, produce large swelling pressures. As measured in our lab- oratory, undisturbed montmorillonite-type soils containing sm.all amounts of illite and kaolinite produce swelling pressures as high as 2tX) psi. .Jennings (1050) reported experience in South Africa with soil which developed a swelling Ca-illite pressure of 3.5 tons per sq. ft. I wonder to what extent osmotic forces are involved here. W. F. Bradley: The epitactic arrangement of water molecules in the hydration la,vers on a montmorillonite surface is sufficiently similar to their arrangement in ice, particularly in the property of density, that the swelling phenomenon is quite analogous to freezing, and pressures developed in the two processes are analogous. E. 0. Henry: The hydration of dehydrated clay particles can be followed b.v measuring the heat of wetting or by measuring the energy involved in dehydration in differential thermal analysis (Siefert and Henry 1947). By both methods our results agree well with those of Hendricks who found that both the hydration of sjiecitic cations and the wetting of surfaces are involved. C. G. Dodd: What is the relation of water adsorption to the three structures of montmorillonite as proposed by Hofmann, Endell. and Wilm ; Edelman and Favejee; and McConnell (1951)? Duncan McConnell: As was pointed out recently by Mackenzie (1951), the structure proposed by myself may be looked on as a compromise between the other two insofar as the dimension of the c-axis and the exposed hydrogens are concerned. I believe that we are not able to dis- tinguish between the Edelman structure and mine on the basis of base-exchange capacity or swelling properties, except that Edelman's structure will not provide for sufficient contraction accompanying complete removal of low-temperature water. D. M. C. MacEwan: I am not sure if McConnell proposes that the OH ions wliieb he postulates as replacing O ions on the surface of the montmoril- lonite layers should be capable of dissociating protons, and thereby contributing to the cation exchange. However, if this process takes place, then after dissociation, the structure is indistinguishable from the Hofmann one. Of course, this comment does not apply to the (OH)i groups which McConnell also postulates. Tart III Properties of Clays D. McConnell: The ilift'ereuee would be that in my structure there would be certain tetrahedral positions without Si"" or xU"* ions, whereas all such positions are assumed to be filled in Ilofmann's structure. E. A. Hauser: Wliat type of niontmorillonite was used in the experiments with the Ag* ion that were referred to in MacEwan's paper? In the equations, Ag-montmorillouite to which NHiCl was added was represented as giving rise to moutmurillouite-AgCl-NH, ; this from the point of view of colloid chemistry, is diflioult to understand, because the Ag* ion is the counter imi and ncpt unionized as an AgCl molecule. D. M. C. MacEwan: 'I'lie worl< in question was done by liloch (1050), and I was of cour.se merely reporting it. The inontmorillonite used was de- scrilied as "Terre de Taourirt" from French Morocco, obtained from the "Societg franoaise des GlyefTiues." For the reactions described, the starting material was Na-montmorillonite, obtained by reaction between the elcctrodialysed product and NaOH. This was treated with an excess of AgXOa (3 percent) to give a Ag- montmorillonite, and the latter was treated with a 10 percent solution of NH.Cl, and washed. This is presumed to result in an XH. -niontmorillonite, and AgCl, the latter being left between the layers ratlier than being precipitated out in the free solution. The same kind of decomposition has been carried out by Bloch with a number of other substances. In some cases where insoluble compounds are left between layers, a separate pha.se gradually forms as a result of aging. REFERENCES I'.arshad, I., 1952, Factors affecting the interlayer expansion of vermiculitc and niontmorillonite with organic substances: Soil Sci. Soc. America Proc, v. 16, pp. 176-182. Herger, G., 19-11, De struktuur van montmorilloniet : Chem. Weekldad, Deel 38, pp. 42-43. Bloch, .1. M., 1950, Sur quelques proprietes physiques et chi- miques tie la niontmorillonite. University de Nancy, thesis. Boyes-Watson, J., Davidson, E., and Perutz, M. F., 1947, An X-rav study of horse methaemoglobin : Royal Soc. London Proc, V. 191. ser.A, pp. 83-132. P.radley, W. F., 1945, Molecular associations between mont- morillonite and some polyfunctional organic liquids: Am. Chem. Soc. .Jour., V. 67, pp. 975-981. Bradley, W. F., Grim, R. E., and Clark, W. L., 1937, A study of the behaviour of montmorillonite upon wetting: Zeitschr. Krislallographie, Band 97, pp. 21G-222. Brown, G.. 1950, A fourier investigation of montmorillonite: Clay Minerals Bull., v. 1, pp. 109-111. Brown, G., Greene-Kelly, R., and Norrish, K., 1952, Organic derivatives of montmorillonite : Clay Minerals Bull., v. 1, pp. 214-220. CaillJre, S., and H^nin, S., 1949, Experimental formation of chlorites from montmorillonite ; Mineralog. Mag., v. 28, pp. 612-620. Caillfre, S., and H^nin, S., 1950, Hydration de certains mineraux phylliteux : Groupe frangais Argiles, Comptes Rendus Reunions d'Etudes, tome 2, pp. 48-55. Deuel, H., 1952, Organic derivatives of clay minerals : Clay Minerals Bull., v. 1, pp. 205-214. Deuel, U., Iluber, G., and Il)erg, R., 1950, Organische Derivate von Tonmineralien : Helvetica Chimica Acta, v. 33, pp. 1229-1232. Gieseking, .1. E., 19.39, The mechanism of cation excliange in the montmorillonite-beidellite-nontrouite type of clay minerals: Soil Sci., V. 47, pp. 1-13. Gieseking, .1. E., 1949, The clay minerals in soils, in .Norman, A. G., Advances in agronomy, v. 1, p. 187, New York, -Vcademic Press, Inc. Glaeser, R., 1950, Complexes montmorillonite-corps organiques en phase vapeur: Int. Cong. Soil Sci. Trans., v. 3, pp. 65-71. Hen Size Distribution in Clays, by A. L. Jolnisoii 89 interpretation of Chemical Analyses of Clays, by W. P. Kelley 92 Interpretation of Chemical Analyses of Montmorillonites, by Bernard B. Osthaus 95 rcti-noraphic Stndy of Clay Materials, by Ralph E. Orim 101 Dye Ailsorptiun as a .Metiiod of Identifying Clays, by Charles G. Dodd lO.i Infrared Analysis of Clays and Related Minerals, by Paul G. Xahin 112 Identifieation of Clay Minerals by X-ray Diifraetion Analysis, by (4eoroe W. Hrindley 119 Electron ^Microscopy as a Method of Identifying Clays, by Thomas F. Bates 130 Ditl'ereiitiMl Tln'i'iiud An;ilysis of Clays and Carbonates, bv Richards A. Rowland 151 PARTICLE SIZE DISTRIBUTION IN CLAYS By a. L. Johnson • The identification of the structure of a clay mineral is best acconiplishod by X-ray analysis. Under certain con- ditions, however, the differential thermal method (Grim and Rowland 1944), may also be used to identify struc- ture. These are basic methods of determining to what clay fi:ronp a particular sample may belonjr. However, the mere fact that a clay mineral has been identified as a member of the montmorillonite frroup, the kaolinite group, or the illite group, for example, does not solve the problem. Too often, the behavior of one member of a specific mineral family is so divergent from the behavior of another member of the same group, that the question is rai.sed whether or not both may rightly be classified as members of the same family. For instance, two clay samples which indicate by their structure that they are metnlici-s of the 'montmorillonite group can differ so widely in their physical characteristics that one ma.y be eminently suitable for use as an oil-well drilling clay, and the other not at all suitable. Because of such irregu- birities persons interested in clay technology have at- tempted to further define the clay system. As the clay-mineral particles are colloidal in size, fac- tors involving the colloidal properties also must be con- sidered in the definition of the system. Therefore, in addition to determining the structure of a clay mineral, it is necessary to examine other physical properties to more closel.y define the system. The fact that clays are colloidal and exhibit colloidal jiroperties indicates that a variation in these projierties could he attributable to vari- ations in the colloidal nature of the substance, and not necessarily to the crystal structure. It is generally assumed that the factors which affect the colloidal behavior of a clay system are at least four- fold. The first factor deals with the structure ; the second, with the surface area; the third has to do with the electro-kinetic or zeta-potential of the system, and the fourth factor deals with the admixed inorganic and or- ganic impurities in the system. Although the effect and influence of the admixed impurities, soluble and insolu- ble, organic or inorganic, are not fully known, there is little doubt that these impurities can play a ma.i'or role ill altering the properties of a clay. It is not the intent here to consider in detail all four factors which can in- fluence the properties of a clay. Certainly, the subject matter of structure has received adequate attention else- wliere in these proceedings. The consideration of ion- exchange propei-ties and the effect of the type and amount of ions on the electro-kinetic potential still re- quires knowledge which is not j^et available. One reason for the lack of information is the difficulty of deter- mining which of the data on the total ions present in a system applj^ to exchangeable ions and which do not. When the chemical techniques for making this separa- tion have been resolved, then we can expect a clarifi- cation of the role that the ions play in affecting the behavior of members of the clay system. • Director of Research, Universal-Rundle Corporation, New Castle, Pennsylvania. The particle distribution of a given clay sample and the effect of this distribution on the total surface area are of great importance in determining the amount of colloidal activity to be experienced. The measurement of surface area is also of great importance, as colloidal activity can be related to a definite surface-to-weight ratio. But before surface area can be measured, the dimensions of the particles must be obtained. For this reason, the methods of measuring the size and distribu- tion of the primary particles have been of great interest, and activity in this field has increased in the last decade. Early in the twentieth century, Zsigmondi developed the ultra microscope and observed not only gold jjarticles in Brownian motion, but also the size of primary inirtii'les. As a result of his observations, the use of optical eciuip- ment for the measurement of sub-sieve particles received much attention. Direct methods of measurement have been iised with great success. Since his work, the direct methods of particle measurement have been superseded, to some extent, by indirect methods which relate the settling of particles through some medium to the diame- ter of the particle. The relationship between velocity of fall and diameter of spherical particles is given in the following equation : where : d = diameter V = velocity of fall k = constant Although particles of clay ai-e far from spheres, when they are allowed to fall in dilute suspensions without interference, the tumbling motion imparted to the particle by its irregular shape and the viscoeity of the medium make it in effect a sphere. For this reason, probabl.y, the data obtained by using the above formula are in good agreement with direct measurements made with optical equipment (Johnson and Lawrence 1942). Indirect methods of measuring particle distribution fall into at least three classifications : sedimentation, elutriation, or centrifugation. The principal advantage of these methods is that more precise measurements can be obtained than by direct methods losing optical equip- ment. An im])ortant improvement of particle measurement was in the field of direct measurements, where use of the electron microscope, with its high resolving power, superseded some of the indirect methods, espeeialty for determining particle shape. Development of special techniques, such as shadow casting (Williams and Wj'ckoff 1944, Woodward and Lyons 19.51), gave great impetus to the use of the electron miei-oscope for ob- taining concrete information regarding the dimensions of clay particles. Up to this time, the calculations of surface area were based on observations involving the oscillations and scintillations of particles observed in- directly. The electron microscope afforded means of di- rect measurement of the three dimensions of a particle, (89) 90 Clays and Clay Technology llUiU. 169 from which the surface area was readily obtained by simple calculations. In the particle size range down to 25 microns, direct methods involving optical eqnipment (Dallavalle 1948) are sometimes more adequate than methods involving sedimentation. Between the 25 microns and | micron range, sedimentation or elutriation are used to greater advantage than the direct methods involving optical equipment. Below i micron, the sedimentation method must be augmented by some means of centrifugation (Norton and Spiel 1938). For extremely fine particles, such as certain montmorillonites, centrifugal methods (Hauser and Lynn 1940; Lyons and Johnson 1947) are almost universally used. Therefore, depending upon the size-range of particles to be observed, one selects the method and techniques based upon the accuracy desired and the time involved in making the measurement. All indirect methods of particle-size measurement re- quire that the sample be completeh^ dispersed in the medium. This involves not only a careful consideration of the medium and method of dispersion, but a dilution at which particle interference does not take place. This extreme dilution is necessary to insure accuracy in the measurement. In practice, the clay suspension used to determine particle distribution is rarely over 2 percent solids. Such procedures allow a measurement of the size of the primary particles. Data from such measurements, coupled with shape factors, permit the calculation of the surface area of the clay fraction. The clay systems which are utilized liy the soil mechanics, the highway engineers, the agronomists, the ceramists, or others interested in clays, are not those of complete dispersion, however. On the contrary, they are compact mixtures, in many instances containing little or no water. Surface-area calculations which are readily made on completely dispersed systems cannot be made on systems which are compact. The complete definition of the particle size of the system should not take into consideration only the distribution and size of the pri- mary particles, but also the state of aggregation, or the degree of compaction of the primary particles. The need today is for methods of determining the state of aggregation of the clay minerals. Direct methods of measurement appear to lend themselves to this prob- lem. However, no detailed techniques are available which provide the necessary data. The fact that larger particles of other minerals are usually associated with the clay s.ystems, makes the problem more complex. Un- doubtedly, the role of the clay particles with respect to the larger non-clay particles is important, and the state of dispersion of the clay particles greatly influences the system. Methods of determining the absolute size and distribu- tion of clay samples are available and can be relied upon. From such measurements, ealciilations of the surface area can be made. However, the restrictions placed on the methods of measurement such as complete dispersion and dilution limit the application of this information. Data on clay systems as they are used in industry would he more useful. Rarely, if ever, are the clay particles in such systems completely dispersed. Therefore, the state of aggregation is of prime interest. DISCUSSION J. W. Earley: In our laboratory we soijaratt'd a moiitmorilloiiite sample into five particle-size fractions ranging iu size from 1600 millimicrons to 50 millimicrons. By redispersing each fraction 10 to 15 times and rerunning it through the Sharpies super-centrifuge, we were able to increase the minus .50 millimicrons fraction to about 50 percent of the total sample. On the basis of these results would anyone care to comment on the particle size of moutmorillonite, illite, or kaoliniteV E. B. Kinter: Working with the electron microscope, we came to the same con- clusion with resi^ect to montraorillonite. If one continues to disperse moutmorillonite the particles get finer and finer. It would seem, therefore, that coarse particles are essentially aggregates and that their size reflects the degree of dispersion attained. This does not apply, however, to kaolinite and illite, which do not readily break down into such ultra-fine particles. W. P. Kelley: Upon dispersion do individual particles separate from aggre- gates or are individual lattice layers split off from crystals? In other words do crystals subdivide to the jioint where they are thinner and thinner, or is it merely a matter of dispersion of aggre- gates? Possibly the [dates become broken crosswise. G. W. Brindley: From the shape of the X-ray diffraction bands, particularly the one near 4. .5 A. Mering and I believe that the order of magnitude of the average flake or crystallite is from 200 to 250 A. I do not think that crystallite dimensions are as large as 500 A. The particle size, of course, may be quite different. I might add here that the shape of a band, generally speaking, depends both on the variation of structure factors with angle and on the size of the flakes. Some bands are more sensitive to structure-factor variation than others. The strong band at about 4.5 A is dependent mainly on flake size, because in this angular range the structure factor varies rather slowly with angle. If one assumes the flake to lie roughly hexagonal in shape, the variation in the intensity of that band can be used to obtain fairly reasonable estimates of the mean flabe dimension. D. M. C. MacEwan: What would be the effect on the shape of the bands if the layers were bent ; is it possible that the spreading of the bands might be due to bending rather than to crystal size? I would like to draw attention to a paper by Mathieu-Sieaud et al. (1951) discussing the electron microscope technique. They suggest that the funda- mental particles of moutmorillonite are about 300 A across but that these particles can link together on the edges to form larger particles, mainly with parallel orientation. These large particles, in some cases, showed a tendency towards hexagonal edges. The process of dispersion may cause not only the separation of layers, but the breaking of layers into fundamental particles about 300 A across. G. W. Brindley: Regarding the first question : we have not taken account of any bending of the planes. I do not know what the effect of bending of the lattice would be on the X-ray diffraction. Recent papers by Wilson (1940) and Blackman (1951, 1951a) have considered the question but no application has so far been made to a clay mineral. With regai-d to the second question, Mering and I considered that the X-ray diffraction work gave a very satisfactory mean crystal size of the same order of magnitude as the minimum size observed with the electron microscope. The electron microscope indicates that the flakes link together in an edgewise manner but upon dispersion either the edgewise linking is broken or the linking is of such a character that the scattering from adjacent flakes is incoherent. This would indicate that the two linked particles are not crystallographically joined. The flakes are, to a large degree, in parallel orientation but still incoherent from the point of view of X-ray scattering. SELECTED REFERENCES Blackman, M., 1951, Diffraction from a bent crystal: Phys. Soc. London Proc, v. B64, pp. 625-630, 1951. Blackman, M., 1051a, Diffraction from a curved linear lattice : Phys. Soc. London Proc, v, B64, pp. 631-637. Part TTT] ^IeTIIODS of IdENTIFYIXG Ci.AYS AXD IXTKUrKKTATION or Kest'lts 91 I)all:ivalle, J. M., 194S, Mic-nmiciitics, p. 0!), Xew York, Pitman I'ublisliing Corp. Glim, R. E., and Rowlanil, R. A., 1044, Differential thermal .-inalysis of days and shales: Am. Ceramic Soc. .lonr.. v. 27, p. 6.5. Hanser, E. A., and Lynn, J. E., 1940, Separation and refraotion- ation of colloidal systems: Ind. and Eng. Chemistry, v. 32, p. liiiQ. Johnson, A. L., and Lawrence, W. G., 1042, Fundamental study of clay, Part IV : Am. Ceramic Soc. Jour., v. 25, p. 345. Lyons, S. C, and Johnson, A. I^., 1047, Continuous centrifuges in the mineral industry: Am. Inst. Min. Mel. Ens. Tech. Pub. 2105. 11 pp. Matliicu Sicaud, A., JK'ring, .T., and Perrin-Bonnet. I., 11151, fttude au microscope electronicpie de la montniorillonite et de I'hectorite satur<^es par differents cations: Soc. franc, mineralogie et cristalloj;rai>liie P.ull., v. 74, pp. 4:?9-4.56. Norton, F. II., and Spiel. S.. 10.38, Measurement of particle sizes in day: Am. Ceramic Soc. .Tour., v. 21, p. .SO. "Williams, R. C, and Wyckoff, R. W. G.. 1044. Thickness of electron microscopic objects: Jour. Applied Physics, y. 15, p. 712. Wilson. A. J. C, 1040, The diffraction of X-rays l)y distorted- crystal aRiircKates. IL Diffraction l)y bent lamellae: Acta Cryst., v."2, pp. 220-222. Woodward, L. A., and Lyons, S. C, 1051, The mechanism of gloss deyelopment in clay coated sheets, Tappi, v. 34, p. 440. INTERPRETATION OF CHEMICAL ANALYSES OF CLAYS By W. p. Kelley • ABSTRACT As a means of distinguisliini; lictweon tliP liroad groups of clay minerals, t-hemical analysis may be a valuable supplement to other methods, but rarely will it remove the necessity for other methods. X-ray analysis in particular. The two- and three-layer clay min- erals may be distinguished by the Si02 ■ AI2O3 ratio or 810= • ses- quioxide ratio, but these ratios do not provide criteria for dividing either class. Among the three-layer clays, which embrace nmnt- morillonite, nontronite, and the diverse hydrous micas, non-ex- changeable K is a better criterion. It seems probable that mont- morillonite, as found in bentonite deposits, does not contain signifi- cant amounts of non-exchangeable K, whereas all known hydrous micas contain significant, although widely variable, amounts of non-exchangeable K. The greatest value of chemical analysis inheres in the fact that by means of calculation the specific nature of the isomorphism may be determined. But the degree of confidence that can justifiably be placed in the calculated isomorphism depends on the purity of the sample analyzed. Si02 as impurity may be shown by an excess of calculated Si**** ions, or by a deficiency of octahedral cations. It also affects the calculated isomorphism of the tetrahedral layers of the lattice by reducing its Al content. On the other hand, AI2O3 as impurity affects the calculation in the opposite way ; that is, it reduces the number of Si**** ions per unit cell, while increasing both tetrahedral and octahedral Al*** ions. Fe as impurity has essentially the same effect as Al. In some samples it is po.ssible to detect SiO? as impurity by means of calculation. Introduction. The two-layer clay minerals, as typi- fied by kaolinite, can be readily distinguished from the three-layer clays such as the montmorillonites and hy- drous micas, by the Si02:Al20.-i ratio as determined by analysis. But this ratio has only limited value : as a means of determining what specific mineral is present in a given sample, the SiOorAUOs ratio can never be relied on as a certain criterion. This is true even when the specimen analyzed is completely free from impurity. The same ratio is characteristic of all four of the dif- ferent members of the kaolin group of clay minerals, kaolinite, dickite, nacrite, and halloysite. In the case of montmorillonites from different sources, there is a rather wide variation in this ratio owing to the fact that the montmorillonites form an isomorphous series. It seems safe to say, therefore, that the SiOalAloOs ratio, or even the SiOo: sesquioxide ratio, or the Si02:(Al20.'i -f- FeoOs + MgO ) ratio is not a dependable criterion of clay-min- eral species. Isomorphism in Montmorillonite. In the montmoril- lonite group of clay minerals, chemical analysis may be used to determine the essential nature of isomorphism and in consequence to show the origin and location of the charge on the lattice. The isomorphous character of the members of this group of minerals can probably be shown in no other way. The fact that the specific nature of the isomorphism can be calculated from the analysis is what gives the greatest value to the analysis. The evi- dence is that the specific pattern of isomorphism occur- ring in montmorillonite from different sources is rather widely variable. The isomorphism may involve (1) sub- stitution of Al for Si in tetrahedral positions in the lat- tice, (2) substitution of Fe for Al in octahedral coordi- nation, (3) substitution of Mg for Al in octahedral positions, and (4) substitution of cations of various kinds • Professor of Soil Chemistry Emeritus, Department of Soils, Uni- versity of California. Berkeley, California. by other kinds of cations in interlayer positions. The va- riations in the substitution in (1), (2), and (3) are basi- call.y Avhat give the greatest interest to a chemical analysis. Accuracy of Analysis. The statement is often made that the chemical anal3'sis must be accurate. Unfortu- nately, a reliable criterion for accuracy is not now avail- able. Recently, Schlecht (19.51) and Schlecht and Stevens (1951) reported analyses of two silicates by 34 specially chosen analysts. The results showed sur- prising discrepancies. With a sample of granite the dif- ferent chemists reported variations in SiOa of 1.69 per- cent ; with AI2O3 differences of 2.78 percent ; with FcaOs a difference of 1.38 percent; with KoO a difference of 3.41 percent; with Na^O a difference of 1.24 percent. All these differences are expressed as percent of the sample and not as percent variation of the respective compounds. The summation of all constituents ranged from 99.58 to 100.39 which is commonly thought to in- dicate reasonable accuracy of analysis. Since the range of totals was distinctly less than that of the major con- stituents, it is certain that where a given constituent, Si02 for example, was high, one or more of the other constituents must have been low. It is probable that dift'erent chemists would obtain an equally wide variation in analyzing a given clay sample. Such variation would almost certainly make rather large differences in calculated isomorphism and consequently in the distribution of charges in the lattice of montmoril- lonite. For this reason alone we are not justified in mak- ing dogmatic assertations about the specific isomorphism as calculated from a single analysis. Impurities. There is still another reason for caution in the interpretation of chemical analysis of montmoril- lonite, namely, the sample as analyzed may not have been free from impurity. Evidence on this point is presented by Mr. Osthaus in the following paper in this bulletin, who shows that reasonably pure montmorillonite eon- tains only minor amounts of non-exchangeable Ca, Na, and probably little, if any, non-exchangeable K. There- fore, if the sample is found to contain significant amounts of non-exchangeable Ca, Na, or Iv, this should be regarded as evidence that the sample is a mixture. Exchangeable Mg. An altogether different kind of point may have significant influence on the calculated isomorphism in octahedral coordination : that is, a con- siderable part of the Mg may be present as interlayer cations, or exchangeable Mg (Kelley 1945). Recently Dr. Foster (1951) and still more recently Osthaus (see the following paper, herein) have fully confirmed this fact. Therefore, exchangeable Mg must be determined and due allowance for the same must be made in calculating the distribution of cations to lattice positions. Method of Calculation. In calculating lattice posi- tions for the constituents found by analysis, the first step is to convert the analytical data into molecular equiv- alents. This is done by dividing percentages of each con- stituent by its molecular weight. Molecular equivalents are then converted into cation equivalents. (92) Part III] JIethods of Identifying Clays and Interpretation of Results Table 1. Calculated isomorphism and charge per lattice unit (Na-sattirated monlmorillonite). 93 .\nalysi.s Mol equivalents Cation equivalents Charge equivalents Charges per 44 charges Cations per unit cell Distribution in lattice layers Charges SiO. AW.— - FcK)j.... MgO,..- NajO.... 58.96% 22.04% 4.06% 2.50% 2.80% .9817 .2162 .0254 .0620 .0452 Si .9817 Al .4324 Fe .0508 Mg .0620 Na .0904 3.9268 1.2972 .1324 .1240 .0904 30.9042 10.2090 1.1994 .9759 .7113 Si 7.726 Al 3.403 Fe .400 Mg .488 Na .711 Tetrahedral Si 7.726 Al .274 — .274 8.000 Octahedral Al 3.129 Fe .400 Mg .488 5.5908 44.0000 — .437 4.017 — .711 Interlayer Na .711 *.711 The seeoud step is to convert cation equivalents into charge equivalents. This is done by multiplying the values for each cation by its valence. The sum of the charge equivaleuts represent total charge ecjuivaleuts per 100 grams of sample. An alternate method is to calculate tlie oxygen equivalents for each cation. The third step in the calculation is to convert cation charge equivalents into actual charges of each cation per a specified number of total charges ; in this ease the unit cell is taken, which contains 20 oxygens and 4 OH" or a total of 44 negative cliargcs. Since these negative charges must be balanced by cations, the sum of the positive charges must also be 44. The fourth step is to convert charges of each cation into ions by dividing each by its valence. If oxygen equivalents are calculated, the remainder of the calculation is merely a matter of converting oxy- gen equivalents into cations per 22 oxygons, which of course, is the oxygen equivalent of 44 charges. The final results will be the same by both methods. Instead of equivalents, as used herein, "ratios" or "proportions" is preferred by certain workers. For the purpose at baud, these terms are interchangeable. Finally, the ions are distributed into lattice positions. Si goes to tetrahedral positions of whicli there are eight per structural unit. If the calculated Si is greater than eight, tins is positive proof that the sample contained Si02 as impurit}'. But a deficiency of Si (less than eight) is by no means proof that the sample was free from un- combined Si02. Nearly all montmorillonites that have been analyzed show upon calculation less than eight Si ions per lattice unit but the amount of uncombined Si02 may not be sufficient to raise the calculated Si above eight. Therefore, the presence of Si02 as impurity may escape detection by calculation. The deficiencj' in Si is made up by placing Al in the tetrahedral coordination. The remaining Al, all Fe, and non-exchangeable Mg are placed in octahedral coordina- tion. All exchangeable cations go to interlayer positions. Table 1 is presented as an example of the calculation. It should be pointed out that four assumiDtions under- lie the aforementioned calculation: (1) The analysis was accurate; (2) The sample was free from impurities with the exception of minor amounts of Ti02, MnO, etc., and that these were present in the sample in the form of oxides; usually the amounts of these oxides is so small as to atfect the calculated results only fractionally whether or not they are included in the calculation; (3) The unit structure contains 20 oxj-gen and 4 OH" ions ; and (4) All HoO found by analysis was derived from OH ions attached solely to octahedral cations or was ad- sorbed water. Hence all H2O found is disregarded in making the calculations. However, by simple modification of the method, it is entirely possible to include all water found. Brown and Norrish (1952) recently showed that with hydrous mica (H3O*) ions may be calculated from the H2O found. Effect of Impurities. It is not certain what is the most common impurity in samples of montmorillonite. If the sample is merely hand selected, even with the aid of the petrographic microscope, significant amounts of impurities of various sorts may be entirely overlooked. The greater part of the impurit.v is likely to be some form of SiOo. For this reason the eifect on the calcula- tion of SiOa as impurity will be considered. Stated in general terms, SiOo as impurity has the fol- lowing effects on the calculated results: (1) It increases the number of calculated Si ions in a unit quantity of the sample; (2) It decreases the calculated total octa- hedral cations; (3) It decreases the total charge per unit cell. Tlie reason for these eifects should be obvious. The Si ion has four charges whereas practically all other constituent ions have three or less charges. If the sample is contaminated with Si02, obviously the total numl)er of Si ions per unit quantity will be increased propor- tionately. Hence, an undue portion of the total cation charge will be borne by Si and accordingly the charge of all other cations will be reduced. If the sample ac- tually contains tetrahedral Al, the effect of Si02 will be to increase the calculated number of Si ions, and there- fore to decrease the number of Al ions required to fill the remaining tetrahedral positions in the lattice, hence, to decrease the charge in these layers. (Tetrahedra have unbalanced charge only when and to the extent to which some ti'ivalent ion like Al is in tetrahedral coordination.) Thus, it follows that insofar as the tetrahedral layers of the lattice are concerned, the calculated isomorphism will be diminished by SiOa as impuritj-. Silica also affects the calculation with respect to the octahedral layer of the lattice. Since Si02 as impurity increases the number of calculated Si**** ions per unit of sample, its presence has the effect of reducing the calculated numbers of all other cations. This effect is further magnified by the fact that Si has a higher valenc.v than the other cations. The net result is the total number of calculated Al***, Fe, and Mg** ions available <)-t Clays axd Clay Techxology [Bull. 169 for octahedral coordination will be reduced by SiOo as impurity. Consequently, its presence in the sample in- creases the calculated charge on this layer of the struc- ture. Al and Fe as impurity have an effect opposite to that of SiOo, that is. they increase the number of trivalent cations required to supplement Si in tetrahedral coordi- nation and, therefore, they increase the calculated tetrahedral charge. At the same time, Al or Fe as im- purit.v increases the total octahedral cations and accord- ingly reduce the octahedral charge. But Si02, AI2O3, and FcoOs as impurity have one effect in common ; namely, each of them reduces the total calculated charge on the lattice. This may be looked i;pon as the result of diluting the sample. Enough perhaps has been said about principles. The question arises at once : Are the aforementioned effects purely- hypothetical, or do they tincl significant expession in actual analyses of montmorillonite? In the following paper IMr. Osthaus shows that they are not purely hypo- thetical and that, in certain cases, they can be used as evidence of impurities in actual samples of mont- morillonite. SELECTED REFERENCES Brown, G., and Norrish, K., 1952, Hjdrous micas: Mineralog. Mag., V. 29, pp. 929-932. Foster, M. D., 19.T1, The importance of exchangeable magnesium and cation exclianfre capacity in the .study of montmorillonite clays : Am. Mineralogist, v. 36, pp. 717-730. Kelley, W. P., 1945, Calculating formulas for fine-grained miner- als on the basis of chemical analysis : Am. Mineralogist, v. 30, pp. 1-26. Schleeht, W. G., 1951, Comparatiye investigation of precision and accuracy in the chemical analysis of silicate rocks : Anal. Chemistry, v. 23, pp. 1.508-1571. Schleeht, W. G., and Stevens, R. E., 1951, Results of chemical analysis of granite and diabase : U. S. Geol. Survey Bull. 980, pp. 7-24. INTERPRETATION OF CHEMICAL ANALYSES OF MONTMORILLONITES l;i J:bi;.N,U;u B. UciTHAL.S ABSTRACT Particles of monliiKiiillDiiito ropn'sciilinj; a considerablo rauge of sizes were separated from onide l)enloMites by means of sedi- mentation and sniii-reentrifujiation. Except for the effect of impuri- ties, it was found that 'l)ase-excliange capacity was independent of particle size. The ratios of the several exclianf;eal)le cations on the mont- morillonites vary widely as they occur in the crude bentonites. Sodium is the dominant cation in the Clay Spur, Wyoming, and Tielle Fonrche, South Dakota, samples; cjilcium in those from rolkville, .Mississippi, Santa Rita, New Mexico, Chandlers, Arizona, and IMymouth, I'tali ; in the I.ittlo Hock, Arkansas, .sample, hydro- Ben is the most abundant exchanj,'eable cation. The hydrogen ion is reflected in the low pH of thi' sample. The Otay, California, and Merritt, liritish Ci>lumliia, samples contain more nearly equal amounts of exciiangeable calcium, iinignesiuni, and sodium. Formulae calculali-, California 127.5 126.0 127.5 128.3 123.0 122.5 4.7 m/i =z millimicrons. 1)0 "rrcatly infliuMicpd by the impurities that occur in iiioutiiiorilioiiites. Tlic data i)rcspiitcd in this pajxT af- ford stroiifr evidence that insistence on hig-h purity in the interpretation of chemical analysis is amply justified. 1 'nless the sample is largely free from silica as impurity lliere is bound to be nneertainty as to the actual iso- morpliism in the sample, and therefore, there will be ri'asiiiial)l(' doubt as to the origin of the eliarge. Acknowlcrlgmcnis. Dr. W. P. Kelley, Professor Emeritus of the University of California, and i\Ir. Ben 15. Cox, Director of the Geology Division of tlie Gulf liesearch & Development Company, gave generously of their time supervising this study and assisting in the preparation of this report. TJie autlior is indebted to Dr. Paul D. Foote, Execu- tive \'ice President of tlie Gulf Research & Develoiiment Company for permission to publish these results. DISCUSSION C. S. Ross: Willi rctVi'i'iice to the aiial.v.ses of Kranilc ;iiul diaha.-ie tlial Ki'Ui'y citeil. Ilipre are two .schools of thous'it on tlipir evaluation. Xo mattPr how bad they may appear by mere inspection, one group feels lliey should lie liiiuped together with the good analyses and run through a mathematical hokus-poUus and he a part of the final result of the mean. The other school believes analyses that appear to be bad should be eliminated. When the poor analyses mentioned by Kelley are eliminated most of the rest group very nicely around the means. I think that it is ju.stifiable to throw out an analysis where common sense indicates that .somebody blundered. W. P, Kelley: It is perfectly true that the vast majority of the ;inalyses cluster :iround a fairly close median point. However, when an analysis is published, how does one know whether it is a good analysis or a bad one? 1 repeat, we are not justified in making dogmatic assertations about the specific isomorphism when calculated from a single analysis. R. A. Rowland: Has Osthaus applied his technique of analysis to mineral nii-\- tures from sediments or soils? B. B. Osthaus: Xo, we felt tli.'it before tackling sedimenls something about simpler s.vstems. had better learn R. A. Rowland: Eighty percent of the e:irlh crust is made up of chiy, ;ind of thiit 80 percent, .' percent is cla,v of hydrotherni:il origin. Kvery one of us, including myself, is guilty of choosing :i cbiy of hydro- thermal origin or near hydrotheriinil origin on which to do his detailed work. We choose clays that are perfect examples, or nciirly perfect examples, as standards. Yet the ver.v reason for our exist- ence as clay researchers is to le;irn something of the other 75 percent; and until we do there is only sciuit justification for our work, except as ;i purely jicailemic matter. C. S. Ross: 'I'liougli there is neeil for studying clays of the soils and sedi- ments, we need to st;irt first of all with as pure minerals as we can get before we h:iiidle mixtures of niiner:ils. In our work on the montmorillonite group (104.5), it might ;ililiear that we used a disproportionate number of bentonite analyses, but thiit was the best way to secure approximately pure cla.v malerial.s. The neglect of all the other materials was from necessity and not from choice. The Ordovician bentonites are the best source for potash mic:is. Here again we would be inclined to u.se a disproportionate number of s;iinplcs to the neglect of the jiotash cl:iy miiier;ils in soils. I. Barshad: Besides using (diemical analyses for c:ilcnlatiiig formulae, an aujilysis, or even the analysis of a single constituent, is of great v:iliie for other purposes. An jinalysis of the free oxides, carbonates, or soluble salts is indispensible for the classification of soils. In the study of clays an anjilysis of the total pota.ssium or calcium content in samples, which were saturated with Xa* or XHi* prior to analysis, may indicate ii great deal about the presence of hydrous mica or such impurities as the feldspars. In this discussion not enough stress was placed on the distinction between theliualitative and the quantitative aspects of the analysis. Thus for a qualitative ;iiialysis of a sample, the use of any one of the methods may sutfice. but for a quantitative analysis of the same sample several of the methods may be required. Therefore, thorough understanding of all of the methods is e.ssential for an accurate analysis. H. R. Shell: The person desiring a chemical analysis should first make abso- lutely sure that he has a sample which, if analyzed correctly, will 100 Clays and Clay Technology [Bull. 169 give results that directly pertjun to the problem at hand. Further, he should collahoratf with the analyst by stating (ll the elements or constituents to be determined, and (2) the absolute accurac.v which must be obtained for each individual element or constituent. A complete analysis is usually not economically feasible. All samples for scientific or research purposes should be analyzed (1) at least in duplicate, and (2) in conjunction with similar samples of known composition, such as those obtainable from the U. S. Bureau of Standards or the U. S. Geological Survey. The analytical chemist, if at all possible, should u.se methods that are specific, and should guide his work with spectrographic analysis. Interfering elements, such as P and F, should l)e checked unless they are previously known to be absent. F is especially important in minerals where F~ may .substitute for OH", for example, in various micas and amphiboles. SELECTED REFERENCES Foster. M. I).. 1!I.">1. The inipiirtan( >■ of exchangeable magnesium and cation exchange capacity in the study of moutmorillonitic clays: Am. Mineralogist, v. 36, pp. 717-730. Kerr, P. F.. JIain, JI. S., and Hamilton, P. K., 1950, Occurrence and microscopic examination of reference clay mineral specimens : Am. Petroleum Inst. Proj. 40, Prelim. Kept, ."i, .58 pp., New York, Columbia University. Kerr, P. F., et al., lOHOa. Analytical data on reference clay minerals : Am. Petroleum Inst. Pro.i. 40, Prelim. Kept. 7, 160 pp., New York, Columbia University. Ross, C. S., and Hendricks, S. B., 1945, Minerals of the mont- morillonite group, their origin and relation to soils and clavs : U. S. Geol. Survey Prof. Paper 205-B, pp. 23-79. Talvitie, N. A., 1051, Determination of tpiartz in presence of silicates using phosphoric acid ; Anal. Chemistry, v. 23, ])p. 623-626. PETROGRAPHIC STUDY OF CLAY MATERIALS By Kalph E. Grim • ABSTRACT Sonio invpsfisatnrs Imvi- iiscd |ictioKr:i|iliic iiirtli"(ls iis tlic sole means of niakinu' clay iiiiiiiM-al analyses, wlici'oas nthers liavc con- sidered i)elri)f;ra|iliic methods to \n\ useless for materials as fine firained as clays. The value of the method lies between these ex- tremes. It has consideralile value, hut it has limitations and must he used with caution. By the use of v.arious auxiliary techni(|ues it is usually ])ossil>le to obtain fairly precise values fcir tl\e inrliees of refr.nction anil bire- friiiKcnce of the clay mineral components (if a clay m.aterial. Fre- iinently an interference fi;:ure can be obtained which yiehls further d.ata on the ojitical diaracteristics. The presence of very tine Holi- day minerals, such as (|ii:irtz and carbonate, and or;;anic material and iron oxide or hydroxide m;iy prevent the making' of satisfactory optical measurements. It is necessary to decide if the optical data obtained are influ- enced hy contamination with uonclay minerals, and if the.v indicate a sinfile clay mineral or mixture of clay minerals. Fortunatel.y the optical values themselves and the character of the particles on which they were determined give good clues regardinj; purity and inonomineral characteristics. Ojilical data in general will not permit the (lelection of relatively small amounts of a clay mineral in a clay mineral mixture. Also such data may be very misleading in studies of mixed-layer clay minerals. In addition to optical data, petrograpliic studies of cla.v nmterials yield other important information. They are of value in determining the texture of the rod; and the relative ainindance and size of the ncincla.v minerals. In general microscopic studi<'s are cis.y to per- form and the prob,able value from them is so great, that they should be made in all clay material researches. Tvfrodiictioii. A i-cador of the lifrratiifc on clay iiiiiu'i-alogy and clay technology would be told by some authors that optical microscopic methods cannot be ap- plied to clays because of their very fine-grained character. On the other hand he would find some investigators using iiptical iiiicrosi'opic techui((ues as the sole means of identi- fying clay minerals and of providing analytical data for extensive chiy-miueral studies. Tn the writer's opinion the best practice lies between these extreme views. Such methods can frequently be used to produce accurate and precise analytical data. They have limitations and must i)e used with caution. Tender some conditions and for some types of ])rohlems, they may produce very mis- leading results, particularly to the inexperienced investi- gator. l'erhai)s it sliould be remarked at this iioint that no siiio'li" t(>cliiiii|nc for cjax'-minrral analysis is always infallible. The ob.ject of the ])rescnt i>aiior is to consider the ap- plication of ])etrograiihic technitiues (not including elec- tron microsco]iy) to clay studies. Consideration will be given ]iarticularly to the data which can be obtained, their interpretation, and the limitations of the technicpies. Mca.tui-f incut of Optical Ycduca. It is frequently pos- sible to determine with fair accuracy the indices of refraction, birefringence, optical sign, and axial angle of the clay mineral component of clay materials. Sometimes this can be done by simply mounting slivers of the dried clay as received in index of refraction liquids. This can be done only if the clay material contains substantially no nonclay minerals, aiul if the clay-mineral ]iarticles show some degree of aggregate orientation. It can be done on some bentonites, some fairly pure kaolins, and some * Re.search Professor, Department of neology, University of Illinois, Urbana, Illinois. shales composed of illites. For such materials, an experi- enced investigator can obtain a good idea of th(> clay- mineral composition in a few minutes by an examination with the petrographic microscope. Many, perhaps most clay materials, require prelimi- nary treatment before satisfactory optical data for the elay-mineral com]iouents can be obtained. Such treat- ment consists of separating the nonclay-mineral com- ponents and in-eparing oriented aggregates (Grim, TJ. K., i934:) of the clay minerals. The separation of the uonclay minerals i)resent in dis- crete jiarticles, like quartz, is a matter of wet sedimenta- tion. In the author's laboratory a separation is fretinently made at one micrcni; the minus one micron fraction is used for clay mineral identification. Sometimes several separations must be made at other size grades, or perlia])s another size grade is more desirable. The optimum point for fractionation to obtain the best concentration of clay minerals for analysis varies with the clay material and the problem at hand. Sometimes a preliminary micro- scopic examination will indicate the optimum size grade into which to separate the clay mineral fraction. Some- times actual test separations must be made. It is frequently necessary to separate the clay fraction by a series of wet wa.shings so that essentially a complete split is made of a given particle size. In clay composed of mixtures of clay minerals some components may be more easily dispersed than others so that the mineral composition of the first washing is not representative of the clay-mineral composition. In the writer's laboratory the clay is first washed free of soluble salts with distilled water; then a suspension usually results without the use of a dispersing agent. If a dispersing agent is required, ammonia is used to avoid the development of a salt when the material is later dried. Oriented aggregates are formed by allowing the dis- persed suspension to stand. The clay-mineral particles tend to settle on a flat surface — which may be the bottom of a beaker or a horizontally suspended glass slide — one flake on top of another with the flat surfaces substan- tially parallel. In the case of the elongate clay minerals, an aggregate orientation of laths or fibers frequently develops. The accumulated clay is removed from the suspension, dried, and then slivers are mounted in index li(|uids for stud\'. The examination of sucii aggregates will yield usually reasonably accurate values for the a and y indices. If the orientation of the flakes is about parallel and if 2V is small, as in the case of many of the clay minerals, the variation of the values obtained for a and y from the true values is negligible. In cases of a single clay mineral, when there is no contaminating nonclay mineral, aggre- gates may be obtained in which there is orientation par- allel to the a- and &-axes and consequently which yield interference figures giving data on the optical sign and axial angle. The development of such aggregates repre- sents something closely akin to crystal growth. In clay materials containing pigmentary iron or or- ganic coni]iounds it is frequently difficult to eliminate such material from the oriented aggregates. The presence (101) 102 Clays and Clay Technology' lliull. 163 of such pigmentary material may make the aggregates worthless for optieal measurements. Sometimes such material can be removed by mild treatment with oxidiz- ing agents, supplemented in the case of iron, with acids. Great caution must be used in such (reatmeut as many of the clay minerals are susceptible to attack by such reagents. For example, some montmorillouites and the palygorskites-sepiolites are very soluble in acid. What Use Can he Made of Optical Dataf As illus- trated in table 1, the various clay-mineral groups have fairly distinctive optical properties, so that clay-mineral identifications should he possible from optical measure- ments. Before identifications can be made from optical values, it is necessary to decide if such values were ob- tained on material uncontaminated by nonclay minerals and composed of a single clay mineral. With regard to the presence of nonclay minerals, the contaminants are usually quartz (or cristobalite in ben- tonites) and/or calcite. These minerals may be present in particles less than one micron in diameter so that they cannot be se))arated from the clay minerals and their detection may be very difficult or, at times, impossible. Usually such material gives a granular appearance to the aggregates as seen when mounted in the index liquids, and it is impossible to obtain optical values with any precision. No interference figures are obtainable ; in tlie case of quartz suspiciously low birefringence may ap- pear, and in the case of calcite the birefringence may be suspiciously high. An investigator with some experience with the petro- graphic microscope in the study of clays can usually tell when the clay-mineral aggregates are contaminated with nonclay minerals, although it is not always possible. Thus the presence of some cristobalite mixed with mont- morillonite in bentonites frequently escapes detection, but even then experience frequently permits detection of this contaminant. The determination of the presence of a single clay mineral versus a clay-mineral mixture in aggregates is a more difficixlt matter. A single mineral rather than a mixture is suggested by the determination of ]>reeise optical values, and the presence of good interference figures indicating some parallel alignment of particles parallel to the a and b axes. In addition there are par- ticular characteristics of the clay minerals themselves which tend to reveal their presence when seen under the microscope. Thus montmorillonite has relatively low in- dices (except nontronite) and high birefringence. The indices of the mineral vary with the lo.ss of adsorbed water, and may change after they are mounted in the index liquid because of adsorption of the liquid. Also the aggregates of montmorillonite have a unique appear- ance, perhaps because they are composed of very small particles. The individual particles of the aggregate can- not be seen and the aggregate often has the appearance of a single crystal somewhat .strained. Again, the apiiear- ance alone of montmorillonite aggregates under the microscope often is enough to indicate its presence. Montmorillonites are most easily confused with the attapulgite-palygorskite minerals as they may have the same appearance under the microscope and about the same indices and birefringence. Fortunately this group of minerals is rather rare and the indices are likely to be a little higher than for most montmorillonites. One would have to be suspicious of montmorillonite-appear- ing aggregates which had indices in the range of the attapulgite-palygorskites. The optical properties of kaolinite and halloysite are fairly distinctive so that aggregates yielding these values usually permit safe identification. It is difficult to sepa- rate kaolinite and halloysite under the microscope. The hydi-ated variety of halloysite may be i-evealed by the lower indices and by the change in the indices as the min- eral changes to the lower-hydration form. A suggestion of the presence of the lower-hydrated form may be ob- tained from slightly lower indices and birefringence than for kaolinite. Halloysite aggregates may seem to be nearly isotropic. Aggregates of kaolinite and halloysite usually have a di.stinct granular appearance. Care must be taken to avoid confusion of aggregates with low indices because of contaminating quartz. Many kaolinitic clays contain relatively large (10 microns) kaolinite worms which can be seen in a microscopic study of the crude clay. These worms will, of course, yield good opti- cal data, and their character will show the presence of kaolinite. Aggregates with indices and birefringence equal to illite, which have a granular appearance, are almost certain to be composed of this mineral. Nontronite has similar optical properties but aggregates of this mineral do not have a granular appearance. Like other mont- morillonite minerals, nontronite aggregates appear as a single particle that has been subjected to strain. Many shales are composed of illite with a natTiral aggregate orientation parallel to the lamination. The clay- mineral can be identified petrographically in such shales without the special preparation of aggregates. Experience is too scant with clay -mineral chlorites and vermiculites to specify how nuu^h can be done with their identification by petrographic methods. In numy of the shales of Pennsylvanian age that the writer has studied, chlorite occurs in large enough and pure enough particles that it can be identified on the basis of its optical prop- erties. In general its optical properties are sufficiently different from those of the other clay minerals so that any relatively pure material should not be missed. If the mineral is intimately mixed with other clay minerals in small amounts, as seems frequently to be the case, it would probably be missed. Apparently it is frequently mixed with illite where its identification is very difficult. Unfortunately chlorite mixed with other clay minerals in small amounts is very difficult to identify by any available analytical technique. There is even less experience with clay-mineral ver- miculite than with chlorite, so that the problem of its identification petrographically can only be surmised. Its optical properties are about equal to those of an iron- rich montmorillonite, and it is unlikely that the two could be differentiated optically. Perhaps the larger particle size of the vermiculite would be suggestive of this mineral. Optical methods of identification encounter great difficulties when mixtures of clay minerals are involved. In some mixtures where the components are present in about equal abundance, fairly certain identifications can be made. As an example, many Pennsylvanian under- I 'an Hi I ^Ii;rii(ii)s OF Idic.vtifvinc, Clays and Ixteki-hetatiox or Rkst'lts Table 1. llptical properties of the clay minerals. 103 a 1 Y — " Sign 2V Dispersion Remarks 24°-50'' Kaolinite - . 1.5.i3-I.S63 1.560-1.. 570 .006. 007 (-) 42° mean value p>V weak Optie plane and Z i to (010) X A 1(001)= =3°± Di.kif 1.560-1.502 1.566-1.571 .006 -.00!) ( + ) 52° 80° p < V weak Z=b;XAC=-t-I5° to20° X:uiilc- 1.557-1.560 1.563-1.566 .006 (— ) ( + ) 40° 90° p>V P 1.. 526-1.. 532 Hallovsite 4H3O Hallo.vsile2Il!0 ... Mean value 1.548-1.556 .002-. 001 Montinorillonitc 1 . -180-1.. i90 1.515-1.630 .025-. 040 ( — ) 0-30° ± X about 1 (01)1) Hertorite 1.485 1.516 .031 small Sapwnitc- 1.480 1.490 1.510-1.525 .030 .035 ( — ) nioflerate Nontronil<- . 1.. 56.5- 1.60 1.600-1.640 .035-. 040 ( — ) moderate Pleoi-liroir; yellow-brown-jrreen Volchonskoite 1..551 1..585 .034 ( — ) smalt Sauconite- - - - 1.. 5.50-1. 575 1.592-1.615 .035-. 042 (— ) small Illite - 1.545-1.63 1.57-1.67 .022-. 055 ( ) small X about X (001): some pleoclinjic Glanrorito 1.545-1.63 1.. 57-1. 66 .022-030 ( — ) 0-20' ± p>V Pleoehroic; yellow-Kreen Ilydrobiotilr 1.59-1.62 1.64-1.67 .045-1.055 (— ) small Pleochroic: green-brown-j-cllow rhlontp 1.57-1.64 1.57.5-1.645 .003-. 007 (T) small Pleoehroic; green-brown ^'e^Ini(■lIlitf• 1.. 525. 1.. 526 1.545-1.585 .020. 030 (— ) small X about L (001); plcncliroi.-; Eteen-hrown Scpiolili- - _ 1.49U 1..520 1.. 505-1.. 530 .009-. 015 (-) 0-60° Z about^^c (elongation) Attaimltritf 1.510-1.520 1 .540-1 .555 .02.5-. 032 ( — ) small Pal.VHorskile Mean value 1.. 50-1. 555 .025-035 ( — ) small to larRC Z//elongation i-lay.s in Illinois are mixtures of illite and kaolinites. AfTfrregates with substantially no nonelay minerals can be obtained which fiive easily measurable and fairly ])recise iiidiees whicli are intermediate between these elay minerals. In oen(>ral, small aiuounts of a second clay mineral may not be determinable ojitically. This is an extremely serious matter as clay-mineral anal.yses for m;iny inirposes must be complete if they are to be of value. 'J'en percent of montmorillonite in a clay material may larjidy determine its physical properties, so that elay-mineral analyses made to predict physical prop- erties or utility mnst be complete. In many cases, clay- mineral analyses had better not be nuidc unless they are complete. Identifications based on optical data encounter the greatest difficulty in mixed-layer minerals. Sucli mixed- layer materials are likel^y to give almost every indica- tion under the microscope of beiiifr a single clay mineral. The indices and birefringence can be measured precisely and often fairly good interference figures can be ob- tained. Actually the interference figure is likely to afford the best suggestion that the material is a mixture. It can be shown that a stacking of sheet units will uive an interference figure with isogyres sharply defined through- out, only if the sheets are uniformly distributed and there is orientation in the a and b axes. If there is varia- tion in the parallelism in the a and b axes and the sheets are not uniformly distributed, the isogyres will first lose their sharpness at the center and then become fuzzy and indistinct throughout as randomness and mixing in- crease. Thus the character of the isogj'res may give a clue to possible mixing of minerals. The literature is filled with identifications of material as a single mineral by optical methods which has been proven to be mixtures by X-ray and tliermal methods. Bravaisite from the type locality is an example (Grim and Rowland, 1942). Many materials wliich have been listed as beidellite are other examples. Otiirr Values of Petrographic Methods in Clay Investi- (jdtitnix. Studies of thin sections of elay materials reveal information on the texture of the materials and the paragenesis of the minerals whicli may be very valuable in some t.ypes of investigations. In general the clay min- erals are very difficult or impossible to identify in thin sections. Many examples of the value of thin-section studies in claj' problems could be given, but a few will suffice. In the study of beutonites, the presence of shard structures is the best evidence of the genesis of the clay by the alteration of volcanic ash. The study of wall-rock alteration associated with ore bodies, the paragenesis of the clay minerals, their possible zonal arranuement. the relation of secondary minerals to jirimary minerals, and the relation of the clay minerals to the ore minerals all require the study of thin sections. In soil investigations the relation of clay minerals to profile development often retinires the study of thin sections. The value of thiu-section studies would be greatly enhanced if a method were available to cut a section without the elimination of adsorbed moisture by drying. In some cases at least, for example, in loosely bonded sands and silts, this ad.sorbed water is an integral part of the texture and it cannot be removed without chang- ing the characteristics of the material. Thin sections of dried material do not in all cases reveal the texture of the natural material. In the writer's laboratory an attempt was made some years ago to freeze the material suddenly and then cut and examine the section while frozen. The method seemed promising, but other activities forced it to be put aside before it could be worked out. ]:)4 Clays and Clay Technology [Bull. im In the writer's laboratory it is cnistomary to examine all samples niicrost'opieally as the first step in any clay- mineral analysis. Sneh an examination provides informa- tion on the kind and abundance of the nonclay minerals. It often gives a good clue as to the probable identity of the clay mineral. Such a preliminary microscopic study aids greatly in planning the further detailed analysis. It indicates, for example, what the optimum jiarticle size is likely to be for the separation of the clay mineral and nonclay mineral fractions. It indicates the size of sample necessary to provide an adequate amount of clay-mineral fraction for analysis. Also, information on the probable kind of clay mineral suggests the sensi- tivity of the differential thermal apparatus that will be necessary. It may indicate the kind and amount of any pigmentary matei'ial and thereby aid in determining what, if any, procedure may be used safely to elimi- nate it. When one is faced with problem of analyzing a large number of clay samples, a microscopic study may permit a classification of the samples into types so that complete analytical data need not he obtained on all samples in order that the clay minerals in all samples can become known. The first reaction of a novice to the jietrographic study of clays is to conclude that the method is of no value. With experience he finds that more and more can be done with the method, and he should conclude that it is a very valuable tool. It is particulai'ly valuable as an auxiliary procedure and as a preliminary method of evaluating and planning the whole analysis. As a means of identifying clay minerals, it nnist be used with cau- tion. When used alone, it will sometimes give trustworthy analyses. Almost always it should be supplemented by other analytical data and this is particularly true when a complete analysis is required. Experience using the clues mentioned herein will serve as a guide to determin- ing when the method can be used safely. SELECTED REFERENCES Grim, R. E., 1934, The peti-oKi-iiphic stuil.v nf tlio cla.v mineral-s — a laboratory note: Jour, t^edimeiitar.v IV-troloKy, v. 4, pp. 45-46. Grim, R. E., and Rowland, R. A., Differential thermal analy- ses of clay minerals and other hydrous materials : Am. Mineralo- gist, V. 1'7. pp. 74(i-7(il. DYE ADSORPTION AS A METHOD OF IDENTIFYING CLAYS l!v (_'ii.\v:i.r.s i;. I Him) ♦ ABSTRACT Dyestiiffs ;in/.s Dependent on Artificialln Induced Pleochroism. The earliest apjilication of dye staining metliods to the identification of clay minerals appar- ently was made by Grand.iean (1909), following the work of Suida (1904) which was applied to numerous other silicate minerals. This early work was concerned with the artificially induced pleochroism of kaolin- group minerals when aniline cl.ves were adsorbed on their surfaces. I^ater Ross and Kerr (1931) described the natural pleochroism of kaolinite and mentioned the fact that kaolinite becomes more markedly pleoclu-oic when stained bv dyes. In an excellent critical discussion of staining techniques, Faust (1940) described the results of his studies on the artificially induced ple- ochroism of kaolinite, dickite, and nacrite when treated with crystal violet, methylene bine, malachite green, safranine '"0", and basic fuchsin. Faust found that these dyes produced strong artificial pleochroism when applied to the larger crystal aggregates such as books and vermicular forms of kaolinite, whereas dickite crys- tals stained less readilv and nacrite crystals stained very weakly. Both dickite and nacrite exhibited much ( 103) 106 Clays and Clay Thiixolocy [Bull. 169 less pleoehroism. Pleoeliroie tests are not charaeteristie of other elaj' minerals. Observations of artificially induced pleoehroism con- stitute a necessary step in modern dye staining proce- dures, but the pleochroic effects are less valuable diagnos- tic criteria than the characteristic colors which result from the adsorption of aniline dyes and aromatic amines and phenols on claj- granules. Staining Tests Dependent on Acid-Base Eeactions. Dyestuffs which act as hydrogen-ion indicators in aqueous solutions may be adsorbed on clay-mineral sur- faces where the resulting color depends on the avail- ability of hydrogen ions from the clay acid. Such adsorbed dyes may be considered indicators of the hydro- gen-ion content of the water film on the clay surface. G. T. Faust's paper (1940) clearly demonstrated the aeid-base mechanism of the color reactions of aniline dyes on clay mineral surfaces. Weil-Malherbe and Weiss (1948) discussed at length the differences between color reactions dependent on acid-base mechanisms and those dependent on oxidation-reduction reactions. The recent excellent work of ilielenz and King (19.51) indicates that the dyes malachite green, color index No. 6-57, and safranine "Y, " color index No. 841. are the most satis- factory aniline dyes for clay -mineral identification. Following the suggestion of Merwin (Merwin and Posnjak 1938), Faust used aniline as a solvent for his dye reagents. This was done because aniline penetrated the clay granules better than other solvents and led to more intense characteristic colors. Faust also used nitro- benzene as a dye solvent because it has a stable index of refraction approximating that of quartz ; and thus it permits rapid recognition and quantitative determina- tion of the quartz content of a clay-mineral .sample. Nitrobenzene has been found to penetrate clay aggre- gates as well as aniline and does not participate in oxi- dation-reduction reactions with clays as does aniline. It is characteristic of the acid-base staining reactions that colors develop in the absence of any water other than that present as an adsorbed film on clay samples in the oven-dry condition. Samples should be dried immedi- ately before making the test so that the effective concen- tration of hydrogen ions in the adsorbed water film may be as high as po.ssible. If this is not done even the hydro- gen montmorillonoids may be stained the basic color of the dye. On the other hand, colors are developed by oxi- dation-reduction reactions only in the presence of bulk water. Another point of difference is the speed with which chai-acteristic colors are developed. In aeid-base reactions the colors develop immediately whereas the full development of color by oxidation-reduction mechanisms may take from 5 minutes to a number of hours. Prior acid treatment of clay samples leads to uni- formity in the characteristic colors produced by aniline dyes on different clay minerals. The colors used for identification thus ai-e those developed by hydrogen clays. Since the acidity of clay-mineral surfaces is largely a function of the amount of exchangeable hydro- gen, it follows that characteristic colors produced by various clay minerals in acid-base color tests are depend- ent on the respective base-exchange capacities. These conditions obtain, and the test is characteristic, if the proper concentration of dye solution is used together with the proper relative amount of clay sample and volume of dye solution. If an excess of dye is used for a given clay sample, the clay granules will be saturated with the basic color of the dye even in the presence of liydrogen montmorillonoids which produce the acid color under proper test conditions. This effect is the result of base exchange, the replacement of hydrogen ions in ex- change positions b,y dye molecules. The necessity for adjusting the amount of dyestuff added to the sample at hand makes it impractical to attempt to identify very small amounts of montmoril- lonoids admixed with clay minerals of markedly lower base-exchange capacity, ilielenz and King (19.")1), how- ever, have been able to identify montmorillonoids in con- centrations of as little as 5 percent. Sensitivity of the d.ye tests for montmorillonoids is also a function of the size of the granules in the sample. The smaller particles of montmorillonoids may be saturated witli the basic color of the dyestuff whereas the larger grains are stained with the desired acidic color. Detailed procedures for the use of malachite green and safranine "V have been described by Mielenz, King, and Schieltz (]9.")()) and Mielenz and King (1951). An outstanding advantage of the aeid-base tests is that they may be used for semiquantitative determina- tions of the various minerals present in a clay sample. Using nitrobenzene as a dye solvent, it is possible to identify many mineral constituents with the petro- graphic microscope. Samples are prepared on microscope slides with cover glasses. The numbers of grains of the various minerals admixed with clays and the numbers of clay granules or aggregates stained each color may be counted to estimate the amounts of the various con- stituents present. The acid-base tests employing malachite green and safranine ''Y" are the best color tests available for iden- tification of minerals belonging to the kaolin group and probabl.y are applicable to a wider variety of samples than the oxidation-reduction color tests. Heetorite sam- ples, however, are decomposed completely by the neces- sary acid treatment. The tests are useful for the identi- fication of montmorillonoids if samples contain large amounts of these minerals, but are unsuitable if only small ]icrcentages are present. The acid-base dye tests should be used in conjunction with one or more of the oxidation-reduction color tests in order to permit more complete and accurate identification of unknown sam- ples. Hendricks and Alexander's Benzidine Test. In 1940 Hendricks and Alexander (1940) described a qualitative procedure for the identification of members of the mont- morillonoid group. The test depended on the ads(n-|ition and oxidation of benzidine on clay mineral sin-faces. Eisenack (1938) and Hauser and Leggett (1940) inde- pendently described clay color changes resulting from oxidation-reduction reactions, but they did not propose to use them as identification methods. Hendricks and Alexander found that any diamines capable of producing colored senutpunone forms on oxidation exhibit tlie same colors after adsorption and oxidation on montmorillonite surfaces. Semicpiinones are "odd electron" compounds, free radicals stable in acpieous solution, which are formed by tlie loss or addition of one electron at one end of a Pan III] ilETIIODS OF IdKNTUVI.VC Ci.AYS AND INTERPRETATION- OF RESULTS 107 system of conjugated double bonds. Seiniqiiiiioiips are stabilized by resouanee energry and usually are colored. They are stable only in restricted ranjies of jiII. Semi- (|uinone cations residt from one-electron oxidation of various aromatic diamines, jihenols. and certain other orffanic compounds. The cliemistry of semiciuinones lias been diseus.sed in detail by ^lichaelis (1985). Ileiulricks and Alexander believed their test was the result of fortuitous o.xidation of a very small amount of diamine to the semi(|uinone form on a montmorillonite surface. Tiu'v found that fei-ric iron salts and oriianic matter were capable of producing the blue color in the absence of inontmorillonoids. They also found that nuiu- ganese dioxide was a strong: enough oxidizing agent to cause color formation. The inorganic interferences could be r(Mnoved by ]n-ior treatment with hydrochloric acid and the orgfinic mattei- by jieroxides. Keducing agents such as ferrous iron salts and stannous chloride were found to iidiibit formation of the blue color. In addition to the ]irpsi'nce of an oxidizing agent, it was necessary that the reagent be ad.sorbed. presumably on montmo- rillonoid 001 lattice planes. Hendricks and Alexander found i)rior adsorption of large molecules such as codeine or brucine on montmorillonite sam]3les prevented forma- tion of the blue coloi-. Tlu' benzidine test has been the target of nnicli criticism as a I'csult of its indefinite na- ture (Endell et al., 1941; Kriiger and Oberlies 1943; Page 1941; Siegl 1945). Experience with the test has shown that jiositive tests for montmorillonoids can be obtained with clay sam|iles known to contain none; and, ronversely. negative tests have been obtained with known montmorillonite samples. Difficulties encountered in application of the benzidine test to unknown samples probably result from the appar- ently low oxidation jiotential of benzidine in neutral aqueous solution and from anomalous adsorjition effects. Many relatively weak oxidizing agents are eajiable of removing one electron from benzidine in neutral a(|Ueous solutions. The i)resenee of such substances in non-mont- nu)rillonoid claj-s would result in false positive tests. Endell. Zorn. and llofmann n941") thought that mem- bers of the niontmorillonoid grouji containing large amounts of nuignesium or aluminum do not produce blue colors with benzidine, or produce only very weak colors. In agreement with their findings, the writer and others (Mielenz and King 1951) have found that pure lieetorite samples produce little or no blue color when treated -with benzidine. On the other hand, certain pure, typical mont- morillonite samples such as API 11 and :30 from Santa Rita, Xew ^Mexico, and API 24 from Otay. California (Kerr et al. 1949), also give weak or negative benzidine tests. Kerr, Kulp, and Hamilton (1949) calculated, ac- cording to the method of Ross and Hendricks (1945), the amount of ali;minum in tetrahedral co-ordination for 10 of the API collection of niontmorillonoid samples. The above-named montmorillonite samples which give vei-y weak or negative benzidine tests, API 11. 24. autl 30. and hectorite, API 34a, were found to contain essen- tially no aluminum substituted for silicon in the tetra- hedral layers. This suggests that oxidation of benzidine to the semiquinone form in nei;tral solution can occiir on 001 lattice planes of niontmorillonoid minerals only at those active spots where silieou has been rejibiced by aluminum. Mielenz and King's Modified Benzidine Test. Recently Mielenz and King (1951) have suggested an improved modification of the benzidine test which they have effected by adjusting the pll of the reagent solution in contact with a clay sample to approximately one. With moiit- morillonoid clays, this results in a lemon-yellow particle color rather tlian the purple-blue obtained in neutral .solution. The yellow colors also are stable upon drying. This permits mounting the dried granules on a micro- scope slide in immersion oil for semi-quantitative deter- minations in the same manner as the acid-base tests. The newer benzidine test is much more reliable than the older one. It gives positive tests with known niontmo- rillonoid samples and negative tests with samples known not to contain iiii>mbers of the niontmorillonoid group. A tentative reaction equation has been proposed by Weiss (1938) to describe the course of oxidation of ben- zidine to its semiquinone. HsN [O] -le- xa Blue cation? It would be of interest to determine if a different cat- ionic, yellow semiquinone is stable in acid solution at a pll of about 1 such that : H=X / XH; -1- H H.N NH= Yellow cation? Identification of the blue- and yellow-colored substances necessarily is tentative. Clarification of the nature of benzidine semiquinones stable in various pll ranges awaits precise potentiometric studies. In addition to identification of niontmorillonoid min- erals, Mielenz and King have found, emi)irically, that their modified benzidine procedure i)ermits differentia- tion of kaolinite from anauxite. Anauxite is stained yellow whereas kaolinite is not affected. This is useful in conjunction with acid-base dye tests. Haiiihhtoii and Dodd's Para-Amino Phenol Test. In 1951 Ilambleton and Dodd (1953) described the use of 2;-amino phenol as a reagent for the identification of clay minerals by an oxidation-reduction mechanism. Test procedures were developed using various concentrations of alcoholic solutions of the 2)-amino phenol reagent, fol- lowed by drying of the alcohol solvent, treatment of the sample with 1 : 1 hydrochloric acid, aud observation of the sample wet and after drying. Reflected light was most satisfactory. White porcelain spot plates, a stereo- scopic microscope, and a reproducible intense light .source wepe the only equipment items needed. Character- istic colors ■\\ere obtained only in the presence of a highly acidic solution. Another test procedure using a saturated aqueous solution of the /j-amino phenol reagent in a manner similar to the newer benzidine test of Mielenz and King was tried, but it was less successful. The new ^j-amino phenol test is most sensitive and accurate for the montmorillonoids and the hydrous mica 108 Clays and Clay Teciixolociy [Bull. 1(59 minerals. It is less satisfa('tory for members of the kaolin group when these oeeur in mixtures with other clays. The test is particularly sensitive to relatively small con- centrations of montmorillonite or members of the mont- morillonoid grouiJ in mixtures of clay minerals. Tenta- tive equations for the oxidation of p-araino phenol to semiquinone forms are written below : :NH= ■NH= [O] -le- :0H Na :OII Blue cation? :NH. + H :0H Blue cation? ■OH Tlie p-amino phenol reayent is not oxidized to eharac- teristie blue colors by ferric iron or extraneous organic matter, although there is some evidence that the presence of manganese dioxide will give anomalous tests when present in clay-mineral samples. The new reagent ap- pears to have a higher oxidation potential in acid solu- tion than does benzidine in neutral solution. The most sensitive montmorillonoid test is obtained by use of a 0.1 percent alcoholic solution of 2>aniino phenol as a test reagent. A drop of this solution is ap- plied to about 3 to 5 cubic millimeters bulk volume of clay sample in a depression of a white porcelain spot plate, and the sample is stirred with a tootlipick. After evapoi-atiou of the alcohol solvent, the dried grains are mixed with a thin glass stirring rod; and one drop of 1:1 hydi'ochloric acid is added. After 10 or 15 minutes the color of tlie wet sample is observed. [Members of the montmorillonite family are identified by the i^resence of a characteristic dark blue edging around clay granules or aggregates. If 0.5 percent and 2 percent reagent solutions are applied to separate porticnis and the same procedure is followed, it is found that hydrous micas are colored characteristic murky greens, olive greens, browns, yellow greens, or tans on the surface of the main mass of the clay sample. This test often is definitive for the almost invariably impure hydrous micas, but it should be con- firmed by acid-base tests. If 4 percent or 2 percent solutions of the p-amino phenol reagent are used in these tests and the samples ai-e allowed to dry following ap]ilication of 1 : 1 hydrochloric acid, it is found that blue or purph^-blue colors are char- acteristic of montmorillonoids and that various shades of pink are characteristic of members of the kaolin group unless the latter are present in relatively small amounts. When samples are stained with saturated aqueous solu- tions of the 2J-amino phenol reagent less satisfactory differentiations are po.ssible, although clay granules may be colored more intensely. One advantage of the p-amino phenol color tests is that they generally may be applied without any previous chemical preparation of the sample. The best procedi^re is to pretreat unknown samples with dilute hydrochloric acid to remove excess ferric salts and anomalous oxidizing agents such as manganese dioxide. When the unknown sample contains hectorite or one of the nontronites, the usual pretreatment with 1 : 1 hydrochloric acid will de- compose hectorite rapidly and nontronites slowly. For this reason it is advisable to test for the presence of montmorillonoids before making any routine pretreat- ment with acid. The time required to examine the sample for the presence of montmorillonite minerals using a 0.1 percent solution of p-amino phenol is so short after the addition of the hydrochloric acid, only about 10 minutes, that neither hectorites nor nontronites decompose before a characteristic montmorillonoid test has been observed. It seems reasonable to assume that color intensities observed in oxidation-reduction staining tests are a func- tion of the relative densities of adsorbed colored semi- quinone cations. With mixed-layer minei-als in which only a fraction of the total area of 001 lattice planes resembles montmorillonoid crystal surfaces, less intense colors would be expected. The stain would not penetrate those 001 surfaces bonded by potassium ions as in the micas. In the case of relatively pure illite samples, there would be no intervening 001 planes which the stain could penetrate between exterior surfaces. Although these exterior surfaces might resemble montmorillonoid planes, they would constitute only a small percentage of the total 001 planes. Thus there would not be enough of the colored semi(iuinone cations adsorbed to jiroduee an over-all color effect similar to that observed in the presence of montmorillonoid minerals. The result Avould be an opaque color which would be a mixture of tlie semi- quinone color and the body color of the sample. By noting these variations in color, it should be possible to identify illite samples by empirically determi'ning the colors produced with known illites after the removal of extraneous coloring material such as ferric salts. This has been done in the development of the p-amino phenol procedures. Although not detectable by X-ray diffrac- tion, some penetration of 001 montmorillonoid planes by adsorbed reagents probably occurs. Xafure of the Oxidizing Agents on Clay Mineral Sur- faces. Hendricks and Alexander (19401 and Weil- Malherbe and Weiss (1948) have assumed that the oxidizing agent responsible for the one-electron oxida- tion of benzidine to its blue semiquinone on mont- morillonite surfaces is lattice-bound ferric iron. Other oxidizing agents probably contribute to formation of the blue benzidine color also. On the other hand, reagent solutions of p-amino phenol are not oxidized by ferric ions in acid solution, and montmorillonoid clay-mineral samples which contain essentially no iron cause forma- tion of the blue semiquinone color when treated with p-amino phenol and hydrochloric acid. One or more oxidizing agents other than ferric iron must be respon- sible for semiquinone formation by /^-amino plienol. and these same agents probably play a role in the oxidation of benzidine. The empirical nature of oxidation-reduction color tests is in part the result of ignorance concerning oxidation mechanisms which can aecomit for the characteristic formation of colored semiquinone cations only on mont- morillonoid minerals. Experimental results suggest there must be a particular kind of oxidant a.ssociated with the silicon tetrahedral lavers of montmorillonoids, but not Pari Till Mktikids of Tdkntifvim; ("lays axd Txti:i!im;i:tat!ox ok Rkst'lt^; 109 witli those of tho kaolin minerals, jiyropliyllite, or talc. Tile most outstantlinu; strnotnral ditferenees between these minerals resiilt from isomorplious snbstitntions in the montmorillonoids of ahiminnni for silieon in the tetrahedral layers and of mag-nesium and other divalent eations for alnminum in the octahedral layer. These snbstitntions are the eanse of the inherent electrical charfre on the three-layer clay minerals which resnlts in their capacity to adsorb cations on 001 lattice planes. When silicon is replaced by alnminnm in the t^'trahedral layer, tliere is an excess of electrons associated with the oxy-nesinm is sub.stitnted for alnminnm in the oetaiiedral la.ver, the resulting: electron excess is dis- tributed more diffusely in oxygen tetrahedra on each side of the substituted site. Tn eacji case active spots probabl.v are formed on 001 lattice planes. Oxide ions associated with these spots are potential donors of electrons. Ex- cliangeable eations are adsorbed over these sites ; non- exchangeable potassium ions are embedded within the open ox.vgen hexagons adjacent to the activated tetra- hedra. Large cations such as the semiquinones of benzi- dine or /;-amino phenol nuist be adsorbed and lield by both ionic and van der Waals forces over one or more active sjints. Recently W. A. Weyl (1049) has postulated that the surfaces of cla.vs and finely divided silica exert a ". . . stronsl.v nxiilizin;; pffccr which is (■(iniparalilf with thnt of ozoiip. This pffoct is ver.v strong for the dust of freshl.v grouiul quartz, silic.-i gel. el.'iy. ,tik1 for some other silicates and persists even in the presence of moisture. The oxidizing effect of SiOj is enhanced 1).v light, elevated temperature and is proportional to the surface. The quantit.v of ox.vgen availahle is small and its detection reipiires sensitive methods. The oxidation po- tential, however, is ver.v high, as can he seen from the fact that silver is oxidized to its peroxide." We.vl attributed the specific toxicity of silica and clay dust in the lungs, that is, silicosis, to this powerful oxidizing effect, and thus explained the effectiveness of reducing agents such as aluminum metal dust in thera- peutic treatment. We.vl claimed that a specific reagent, 4, 4', 4"-hexametli,vl triamiiio triphen.vl methane, indi- cated the presence of "atomic oxygen" on silica gel at room temperature and on finel,v divided silica or cla.v at 1.50° C. Ilauser, Le Beau, and Pevear (lO.")!) have pro- posed that the relea.se of atomic ox,vgen on cla.v surfaces is responsible for oxidation-reduction color reactions. Weyl ^ considers that all oxide surfaces undergo O2 — Oo" ion exrlumge, one step of which is a one-electron transfer to a I'hcmisorbed oxygen molecule. In the absence of any more plausible theorv, the writer proposes a tentative h.vpothesis based on the assumption that activated ox.vgen ions in montmorillonoid 001 lattice planes are capable of transferring an electron to adjacent adsorbed oxygen atoms to form superoxide or perhy- drox.vl ions, O2'. This reaction might be represented b.v the following equation : O2 + 0- > 0-2 + 0- .\dsorbed oxvEeil molecule .\ctivate(l elt't'tnui- (lonor oxide jon boimd in tetrahedral lattice layer .\vailahlc oxidizing agent (supercxide ion) Bound in tetrahedral lattice layer ^Personal communication, August 13, 1051. The concentration of superoxide ion oxidizing agent on 001 montmorilhnioid planes would be greater if the elec- tron excess resulted from substitution of aluminum for silicon in tetrahedral layers. Thus, in neutral solution it appears that the effective oxidation potential would be too low to oxidize benzidine to its blue semiquinone cation unless tliere were such tetrahedral la,ver substitu- tion. If tli(» montmorillonoid sample c(ui1ained no alumi- num in the tetrahedral la.ver, the oxidation potential in neutral solution ajiparcntl.v would be lowei- than that of the ferric-ferrous ion couple. In acid solution, however, the oxidation potential of superoxide ion would be distinetl.v higher; and this greater potential would be capable of oxidizing benzidine to its .vellow semiquinone and /j-amino phenol to its blue semiquinone. It is necessar.v, also, to assume that the blue semiquinone form of benzidine is stable onl,v in neutral and basic solutions, whereas its .yellow foi-m and the blue semiquinone of p-amino phenol are stable only in acid solution. In acidic solution the oxidation potential apparcntl.v is sufficient to cause semiquinone formation even when the positive charge deficiency arises from octa- hedral la.ver substitution. Thus hectorite produces semi- (|uinone colors in acid s.vstems (before decomimsition) but not in neutral solutions. It is possible that other reagents besides benzidine and p-amino phenol ma.v be found suitable for oxidation- reduction color tests. The selection of these two reagents is based on the emi)irical observation that their semi- quinones are sufficientl.v stable in a pll range which ma.y be utilized in test procedures. Apparentl.y the blue semi- quinone of p-aniino ]ihenol is stable onl.y in a much more acid medium than the yellow benzidine semiquinone. Color Test Procrdiires. llielenz and King (1951) have developed a standard i)rocedure for the identification of clay minerals by staining tests. The.y emplo.v character- istic differences in birefringence to differentiate between the two- and three-layer cla.v minerals, followed b.y ob- servation of the colors produced b.v malachite green and safranine "Y" d.ve solutions. In this manner the.y detect divisions in the kaolin grou]i, that is, kaolinite and an- auxite, dickite and nacrite, and the hallo.ysites. Pleochroie effects aid in distinguishing kaolinite and anauxite from dickite and nacrite. The acid benzidine test ma.v be used (with a different sample) to differentiate kaolinite and anauxite. Montmorillonoids are recognized if present in amounts greater than 5 to 10 percent, and glauconite and cela- donite fail to stain sufficientl.v to mask their .vellow or greenish-yellow body colors. The acid benzidine test then ma.v be applied to distinguish small percentages of moutmorilloiioids. Experience is necessar.v to distinguish between illites, vermiculitcs. and jial.vgorskite or attapul- gite. In place of the acid benzidine test at this stage of the analysi^s, p-amiiio ]ihenol can be used to confirm the presence of montmorillonoids, including hectorite, and to aid in determining the hvdrous micas. In addition, p-amino phenol is helpful in identifying hallo.vsites by producing characteristic mottled brown granule edges, and it mav aid in confirming other members of the kaolin group. Finally, a word of caution is in order. Before appl.y- ing staining tests as a new procedure, it is advisable to 110 Clays and Clay Tecjinology' [Bull. 169 observe results obtained with known mineral samples. Experience is indispensable. Discussion. Staining test proeedvires described in this paper generally are too involved to permit ready adapta- tion to field identification of unknown samples. Much of the unwillingness to apply staining tests stems from un- fortunate experience acquired when the tests were used in the absence of information concerning their limita- tions. Development of color tests has reached the point today where reliable laboratory analytical procedui-es can be utilized to save time and expense. Numerous labora- tories, such as the Bureau of Reclamation Petrographie Laboratory in Denver, have developed routine procedures which have proved to be most dependable in the hands of properly trained aualysts. DifficiTlties in the use of stain- ing tests have arisen when general "shotgun" applica- tions of one ovei-ly simplified procedure have been made to complex unknown clay samples. Enough development work has been done at this time to permit any clay laboratory to set up generally reliable procedures for the identification of most of the known clay minerals. Man.y checks and counterchecks have been developed to improve the aceiiracy of the tests. If the color tests are properly used in conjunction with petro- graphie and other more complex analytical procedures, they will be found to be a most helpfid addition to the tools of the clay technologist. In addition to wider application to analysis, there is a possibility that color tests, particularly those based on oxidation-rt'duction reaction mechanisms, will be useful in elucidating difi'erences in structure among the three- layer clay minerals. Less is known concerning the hydrous mica clays than those of the kaolin or montmorillonoid groups. G. Brown (1951) has presented the hydrous micas as a continiious series of minerals between the well- crystallized micas and the expanding montmorillonoids and vermiculites. ^Mixed-layer minerals were included with other hydrous micas. The use of reagents such as benzidine, ^J-amino phenol, and other similar compounds which can be oxidized to colored, stable semiquinone cations may be used to difl'erentiate between hydrous micas of varying crystallinity and to detect mixed-layer minerals containing various amounts of montmorillonoid or vermicidite layers. Such differentiations might be made by studying the variations in color produced with a given reagent. Ack7iowledgmcni. This paper is based on work done Yvhile the writer was employed at the Petroleum Experi- ment Station of the U. S. Bureau of Klines, Bartlesville, Oklahoma. Thanks are due Dr. C. W. Seibel, Director Hegion VI, U. S. Bureau of Mines, and to the manage- ment of the Continental Oil Company, for permission to publish this paper. DISCUSSION J. L. Hall: Could differentiiitiou between k.iolinite and anauxite li.v the dye adsorption technique be discussed in more detail? M. E. King: An aqueous solution of benzidine is added to an acid-treated sample. After the sample is dried, it is acidified with 1 : 1 HCl and again dried. When such a treated sample is mounted in im- mersion oils and is viewed under a petrographie microscope and transmitted light, anau.xite is lemon-yellow whereas kaolinite is colorless. This procedure is descrilied in detail in a recent paper by Mielenz and King ( 1!I51 ) . The metluid has been checlved on aiuuixite from the type locality at Bilin. Czechoslovakia, and from till' Iiini' formation of California. B. B. Osthaus: What method is used to remiive iron compcniuils and organic matter? M. E. King: A 1 : 1 solution of water and hydrochloric acid is used to re- move iron compounds and a solution i.)f hydrogen peroxide to re- move organic matter. However, caution must be used in treat- ment of samples thought to contain minerals susceptible to break- ilown in this concentration of acid. R. E. Grim: How successful are you in using the dye tests for the identifi- cation of clay minerals in samples which are an intimate mixture of various clay minerals? For example, in a mixture of kaolinite, illite, and montmorillonite? What is the sensitivity of the dye test? M. E. King: A combination of the various dye tests generally enables a satis- factory separation of clay minerals in such a mixture. The main difficulty arises in detection of clay minerals not reacting positively in the tests where they are intimately admixed in a very finely divided condition with clay minerals which react positively. No difficulty is experienced with mixtures if the particles are large enough for ea.sy inspection microscopically. An illustration of the application of staining tests is the differentiation of illite from montmorillonite b.v the combination of Dodd's test using para- amino phenol with the benzidine test. If a material develops a deep prussian-blue color with para-amino phenol, and shows no change in color after acidification and the addition of a drop of benzidine, then the tests indicate that the material consists of montmorillonite. If the color disappears entirely or changes to a very light pink, then the tests indicate that the material contains illite or hydrous mica. With the use of UKLlachite-green and snfranine "Y," identifi- cation of montmorillonite is reliable if montmorillonite constitutes more than 5 percent of the sample, but with the modified benzidine test even one particle of montmorillonite in a mixture can be identified. In general, the dye tests are as sensitive as are any of the other methods. Isaac Barshad: Did you ever use dye tests to differentiate the different mica minerals or a K*-saturated montmorillonite and a K*-saturated illite? M.E. King: Yes, we fouml that mu.scovite. biotite. phlogojiite, ;ind lepido- lite, for exanijile, do not react with ;iny of the dyes regardless of the pH. All of our dye tests are carried out in an acid medium ; therefore, even though montmorillonite might be K*-saturated to begin with, it becomes Unsaturated and reacts with dyes. K*- saturated illite. after acidification, causes only a slight change in the color of safranine '"T" or malachite-green. Moreover, when illite treated with benzidine is acidified to pH 1 and subsequently dried, no coloration is evident with transmitted light. Details of these reactions are described by Mielenz and King (1951). R. C. Mielenz: We have come to feel that the adsorption of dye molecules by clay minerals offers a tool whereby a better understanding of the phenomenon of adsorption may be gained, for upon ad.sorption of dye molecules not only can one measure the amount ;idsorbed and the changes in the crystal lattice dimensions, but also the color transformation in the dye indicates some change in the organic molecule itself which may be related in some way to the forces with which it is bonded to the surface. Regarding the likeness of the montmorillonite and the illite sur- faces, it may be true theoretically that these surfaces should be alike : but, since the dye tests indicate that they are not alike, the task before us is to find out what the difference is between them rather than to ignore it. Part III Mktiiods of Ii)i;NrirviN'(i Clays and Intkrpuktatiox oi- Ri silts 111 V. T. Allen: Has aiiyiiiii' liail difficiiltii's with the l>i'nzi(linc ti'sl dii tlii> miii- ernls nf the kiinliii ;;i'(>iii)V We had sovcral kaolins which h.v all iilhcr tests cniitaiiird nii moiitiniiiillimitc, but the sample still gave a hliii' test with lii'iiziiliiie. M. E. King: I obspi'ved thp same thiiij; with a kaolinito from Macon. Georsia. but X-ray analysis revealed that the sample contained abont ;" percent montinorillonite, thus explaining the anomalous result ob- tain^'d with the n mentioneil? W. F. Bradley: liecau.se of the hemimori>hic nature of kaolinite. it is doubtful whether any surfjices should really be called abnormal. Each cry.st:il must of necessity present both ;in oxygen and a hydroxy! snrf:ice in ;iddition to less extensive edges. The Iarg4' iucre;ise in the b;ise-exchange capacity of katdiuite upfui grinding, as in the experiment reported by K(dley. is not necessarily diU' ti> an increase in surface area. .Vlthough an increase in surface area undoubtedly results from grinding, the new surfaces Avhich are exposed wcuild most likely be :idditional oxygen and hydroxyl surfaces pariilbd to the (Kit ]dane ; and since such surfaces are electrically neutr;il. the large increase in the base-exchange capacity is due most likel.v to some cither pbenonieticui \\-hic'b simulates base exchange. Isaac Barshad: Is it possible tliat a iieniiiit ile-like subst.'iuce is formed from the decomposition products id" kaolinite ndejised during griiuling':' That k.iolinite decomposed during grinding was revealed by differential thermal and X-ra.v anal.vses. This decomposition is reve;iled in I iT.V by a reduction in the magnitude of the endothermic break characteristic of kaolinite and in X-ray analysis by a weakening in tlie intensities of diffra<-lion lines of the kaolinite pattern. I see no reason to question the possibility of the formatiiui of a permn- tite-like material of high exchange capacity from the released silica anil alumin:!. W. P. Kelley: Can pernintite be produced from a mixture of silicon dioxide and aluminuTu oxi7, The Jliueralog. Soe., London (Clay Miiuuals Group). Eisenack. Alfred. 1!).'!S. K;it!ilysche Einwirkung von Toiieu uiul anderen silikatischen .Miner:ilien uud verbiudungeii anf aromatische .\miue uud riienole : /enlralblatt fiir Mineralogie. Geologic und I'aliiontologie. Itand lO.'lS. Abt. .V. pp. .•'.O.'.-MOS. Endell. .1.. Zorn. R.. anil Ilofiuaun. 1".. 1041. Uber die Priifnug auf Jloiitmorillonit niit I'.iMizidin : .Vngew. ('hem.. I'.aud .")4. pp. .370-377. Faust, G. T.. 1040. Staining of cI.mv mii]i>r;Lls as a rapid means of identification in ualnnil :nid benetiiialcd prmlucts : I'. S. Bur. Mines Kept. Inv. .•'..-)22. 21 pp. Grandjean, M. F.. 1000. Coloration des argiles par les couletirs d'aniline: Soc. fran(,ai.se mincnilogie I?idl.. tome ;!2. pp. 40S-410. Ilambleton. W. W.. and I )odd. C. G.. 10."i:',. .V (|Uiilitalive color test for rapid identification of the clay miiuu-il groups: Kcon. Geology, v. 4S. pp. 130-14(!. Hauser. E. A., I.e Be:ui, P. S., ;iiul Pevear, P. P., 1051. The surface structure and composition of colloidal siliceotis matter: .Tour. Phys. Cidloid Chemistry, v. .");">. pp. (!.S-7!l. Hauser, E. A., and I,eggett. M. P., 1040. Coloi- reactions be- tw-een clays and amines: .\m. Chem. Soc. .Tour., v. 02. pp. lSll-14. Hendricks, S. P.. and .Xlexauder. T,. T.. 1040, .\ nualit;itive color test for the nu>ntmorillonite tvpe of ciny minerals: .\m. Soc. Agriui. .Tour., v. .■{2, pp. I.^.t^.'iS. Kerr. P. F.. et al.. 1040 ;ind lO.'.O. Am. I'elrolruni Insl. Pes. Proj. 40. Prelim. Kejits. 1 to S. Xew York. Columbia I'uiversity. (Al.so bound vohime of .H reports. 10.")1.) Kerr, P. F.. Kulp, J. L.. and Hamilton. P. K.. 104!). Differential thermal analyses of reference clay mineral sjx'cimens : Am. Pe- tridenm Inst. Pes. Proj. 40. Prelim. Uept. '■*•, p. ,34, Xew York, Colundiia University. Kriiger. D.. and Oberlies. ]•'.. 104:!. Slruktur und Farbrek.-itionen von .Moutinorilloniterden : Xaturwissensch;iften. li.-ind .'U. p. 02. Jlerwin. H. E.. and I'osn.iak, E., lOSS, Clays and other minerals from the deep sea. hot springs, and weathered rocks : .Vin. .Tour. Sci., ."ith ser.. v. 3.">A. pp. 170-184. Michaelis. I,., lOS."). Semiipiinoues, the intermediate steps of reversible organic oxidation-reduction: Chem. Rev., v. 10, pp. 243-2Sfi. Mielenz. R. C. and King. M. E., 10.">1. Identification of clay minerals bv staining tests: Am. Soc. for Testing Materials I'roc, V. .^il, pp. i21.3-12:!3. Mielenz. R. C. King, M. E., and Schieltz, X. C, lO.'O, Analytical data on reference clay minerals: Am. Petroleum Inst. Res. Proj. 40. Prelim. Rept. 7, p. 135. -Xew York. Columbia University. Page. J. B., 1041, Unreliability of the benzidine color reaction as a test for montmorillonite: Soil Sci., v. 51. pp. 1.33-140. Ross. C. S.. and Hendricks. S. B., 104.5, Minerals of the mont- morillonite group, their origin and relatiiui to soils and clays: U. S. Geol. Survey Prof. Paper 205-B. pp. 23-70. Ross, C. S., and Kerr, P. F.. 1031. The kaolin minerals: U. S. Geol. Survey. Prof. I'aper Kio-E. i)p. 151-180. Siegl. \V.. 1045. Uber den X'achweis von Jlontmorillonit mit Benzidiu : Xeues .lahrb. Mineral. Geol. Mouatsh. A. Baiul 4S. pp. 40-43. Suida. AY.. 10(14. Sitzungsber. Iv. Akad. AYi.ss.. ( Matheiuatisch- naturwissen.schaftliche Klassel. Band 113. Abt. 2b. p. 725. Thomas, C. L.. 1040. Chemistry of cracking catalysts: Ind. and Eng. Chemistry, v. 41, pp. 2.564-2573. Weil-Malherbe. H., and Weiss, .!., 1048, Colour reactions and adsiu-ptions of .some aluminosilicates : Chem. Soc. .Tour., v. 1048, pp. 2104-69. Weiss, .T., 1938, Xote on some free radicals from benzidine and its derivatives: Chem. and Ind.. v. 57. pp. 517-.51S. Weyl, W. A., 1949, Active oxygen formed at surface of silica and clay as. possible cause of silicosis : Am. Ceramic Soe. Bull., V. 28, p." 362. INFRARED ANALYSIS OF CLAYS AND RELATED MINERALS Bv Pall G. Xaiun Introduction. Tlic identification of clay minerals eithei- in pure form or admixtnre with other minerals is not ahvays easy. In some few instances it is sufficient to use only one of the many methods now available. Usu- ally, however, the composition of the mineral sample is complex and requires the use of two or more supple- mentary analytical techni(|ues for establishing the nature of the minerals and their relative amounts. The petrographie microsco]ie, chemical analysis, dye adsorp- tion, determination of cation-exchanjie capacity, differ- , ential tliermal analysis, X-ray diffraction, and the electron microscope provide us with a battery of tools which, employed as a team, is adequate for the sokitiou of most problems of mineral anal.vsis. Recently, a prom- isinii' new method of analysis has been developed ( Hunt 1950; Keller and Pickett i!»-t9, ID.iO; Xahin et al. (1951) based on the absorption of infrared radiation by min- erals. The purpose of the following discussion is to survey the available information in this field, to make some brief correlations among the spectra, and to outline a few problems which, if clarified, might increase the value of the method. The infrared spectrum com])rises the portion of the electromagnetic spectrum lying between the long-wave- length edge of the visible and the short end of the radio- microwave spectrum. Roughly, the infrared region is sometimes divided into two sections: the near-infrared from 0.75 to 25 microns and the far-infrared from 25 to 1000 microns. The spectra obtained in the near-infra- red arise from the resonance absorption of monochro- matic radiations of wavelengths corrcsjionding exactly to those vibrational fre(|uencies of specific atoms or gi-oups of atoms involved in a change of dipole moment of the molecule. Absorption of radiation in the far- infrared reflects the presence of complex molecular rotational or massive lattice vibrational energy levels, for which no data are available for mineral systems. Tlie region of the infrared spectrum which has proved, thus far, to be of greatest interest for most types of molecules including minerals lies in the fiuidamental vibi'ational region from 2 to l(i microns. For detailed discussions of the origin of spectra the reader is re- ferred to text books on spectroscopy (Barnes et al. 1941; Harrison et al. 1948 ; Mellon 1950 ) . Most infrared spectra being reported in the literature are now obtained with conniiercially available instru- ments having resolving powers of the order of a few hundredths of a micron at all wavelengths. The essential elements of all these instruments, in one or another arrangement, are shown schematically in figure 1 for a single beam. The radiation from an infrared glower is focused on the sample ; tiie transmitted portion of the beam is collimated and monochromatized by a prism and mirror optics and measured b^' a thermal detector. A light chopper, CH, with a frequency usually of 10 cycles per second is included to minimize the effect of false radiation within the instrument. The spectrum is constructed either manually or automatically from de- termination of the ratios of the transmitted to incident intensities as a function of the wavelength. Q SOURCE PATH OF IR RAY (SCHEMATIC) THERMAL DETECTORVn # — -"& FOCUS MIRRORS PRISM PLANE MIRROR OFF -AXIS PARABOLOID Ficl I!F 1. MINERALS — DIFFERENT IR SPECTRA 3 4 5 6 7 8 9 10 II WAVELENGTH, MICRONS 12 13 14 IS 16 Figure 2. IMPORTANT CASE OF SIMILAR IR SPECTRA 0.3% K,0 —1 I I 1— ■ 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 WAVELENGTH, MICRONS • Research Division, Brea Research Center, Union Oil Company of California, Erea, California. FiGUEE 3. ( 112 ) ;irt ;METii()ns 01- Idkxtifyi.vo Clays and 1nti:ki'retation or Kisilts 113 MINERALS HAVE FEWER IR ABSORPTION BANDS THAN MOST ORGAN ICS 5 6 7 8 9 10 II 12 13 14 15 16 WAVELENGTH, MICRONS ff^ QUARTZ SPECTRUM VS. • REFRACTIVE INDEX • WAVELENGTH • PARTICLE SIZE 5 6 7 8 9 10 II WAVELENGTH, MICRONS 12 13 14 15 EFFECT OF PARTICLE SIZE ON SHAPE OF MONTMORILLONITE IR SPECTRUM -L DIAMETER > 88 MICRONS -2. DIAMETER < 5 MICRONS S 6 7 8 9 10 II WAVELENGTH, MICRONS HWB l'"i(.ri!K li. Gross Fcaiurcs of Mineral Spectra. As is true for other elasses of eonipoimds. the infrared spectrtim is a "moleeiihir fin^ierpriut "' and can be used to differentiate the various minerals includinir tlie clays. An exanijih^ of this specificity is shown in figure 2. Here, there is no difliculty in noting that the spectra of the two-layer minerals kaolinite and halloy.site and the three-layer mineral nontronite are obvionsly different as emphasized by the indicated absorptions in the 10 to 1 1 micron rei-'ion. This example is one of many which establish the basis for infrared spectroscopy as a clay identitication method: different minerals proeluce different infrared absorption spectra. However, not all combinations of clay mineral spectra are as obviously unique as that of fifrure '1. An important array of similar spectra is shown in fiirure 'i. Here, montmorillonite, illite, and muscovite are compared in the 9 to 10 micron region. All three appear generally similar visually except that nniscovite has a |iron(iuiic(>(l absorption at 9.35 microns. Usually, montmorillonite antl illite are difficult to distinguish, although it has been stated (Hunt et al. 1950) this may be done on the basis of the peak position of the strongest band, 9.6 microns in montmorillonite and 9.7 microns in illite. The potassia contents are included in figure 3 to underscore the value of supplementary data for clay identifications. Another characteristic of mineral spectra is that gen- erally they have fewer infrared absorption bands than have most organic compounds. This is shown in figure 4 for the cases of a loam, an inorganic compound and an organic salt. ]Many organic compounds do not absorb strongly (Hunt and Turner 1!)53) beyond 15 microns as do some of the feldspars, barite, ilmenite. and other minerals (Hunt et al. 1950). In addition to the various instrumental performance factors, the shape of the spectral curve is intimately as- .sociated with the refractive index and particle size of the sample. Since the dispersion curve of quartz for the ordinary ray from 1 micron to 14 microns is known (Henry, 1948) it is possible to exhibit this dependence (fig. 5). using the particle size data of Hunt, AYisherd and Bonham (1950). Figure 5 shows that the transmis- sion is greatest at wavelengths corresponding to refrac- tive indices near unity except at about 9.2 microns where the absorption can be ascribed to an assigned (Plyler, 1929) fundamental mode of vibration. The effect of re- ducing the ]iarticle size to below 5 microns can be seen in the increased transmission and degree of resolution attained throughout mo.st of the spectrum. Experience has shown that the best spectra are obtained with samples having a narrow particle size distribution (Lecomte, 1951; Xahin et al., 1951) with maximum particle diam- eter not greater than, and preferablj' less than, the shortest wavelength used. Figure 6 provides verification of the particle size requirement for montmorillonite. The arrows indicate significant absorptions either absent or almost indiscernable in the coarse powder spectrum. Preparation of Samples. As is clear from the preced- ing discussion, satisfaction of the particle size require- ment is ci'ucial for obtaining definitive spectra of minerals. In some of the eai'lier work on infrared spectra of minerals the sample was triturated in refined mineral oil (nujol), spread on a salt plate and inserted in the 114 Clays and Clay Technology [Bull. 169 sample cell of the spectrometer. Even -with suspension in the nujol of j^roperly sized particles, the technique is unsatisfactory both from the standjioiut of stability of the .suspension and from the necessity for subtracting its absorption spectrum to obtain the mineral spectrum. The technic(ues used by Hunt and his associates at Carter Oil Company and those employed by Union Oil Com- pany are representative of present practice and are here described briefly. In the Carter method (Hunt and Turner, 19.")3) a 5- to 10-gram sample is groimd to smaller than 250 mesh and jet-pulverizecl in a "JMicronizer" to a particle size less than 5 microns. Approximately 10 to 20 milligrams of the powder is placed on a sodium chloi'ide window, made into a paste with a few drops of isopropyl alcohol, smoothed with a microscope slide and dried in air. After obtaining' the spectrogram, the film weight and other data descriptive of sample thickness and extent of coverage are determined. The Union method (Nahin et al., 1951) differs some- what from the Carter method. The clay sample, sus- pended in distilled water, is fractionated with a Sharpies supercentrifuge to yield a colloid fraction of particle size in the range of 0.1 to 1.0 micron diameter as checked by electron micrographs. The fine fraction is suspended in O.OOIN acjueons ammonia, deposited drop by drop on a single crystal silver chloride (Harshaw Chemical Co.) cell window to cover completely the area traversed by the beam. The cell window and suspension are dried in a vacuum oven at 50°C and scanned in a Beckman TR-2T infrared spectrophotometer. The sample weight is determined by calibration of the dilute clay suspension and b.v weighing of the dried film. Recently, the two new techniques of lamination and impregnation for the mounting of solid samples have been described (Sands and Turner, 1952). Although they have not yet been applied to minerals, the authors recommend these methods as potentially valuable for the study of inorganic and non-plastic materials. In the lamination technicine the sample pellet is placed between two sheets of mica or silver chloride. The laminate is inserted between two stainless steel aluminum or chromium foil protected platens in a hydraulic press, heated to plastic temperature and sheeted out under pressure to a total thickness of about 0.004 inch. The impregnation technique is iised when lannnation is not feasible. The sample pellet is placed on a polyethylene sheet, pressed between foil protected platens, rolled into a ball, and sheeted out on the press; the process is re- peated until homogeneity is attained. Discussion of Particular Absorption Bands. A suf- ficiently large number of infrared spectra of solids is now available so that by a process of elimination certain recurring frequencies within series of reasonably ho- mologoiTS minerals can be inferred to represent particu- lar atomic groupings. In this sense, the absorptions in the ranges of 2.7 to 3.2 microns. 6.0 to 6.2 microns (and sometimes the weak absorption at about 7.55 microns (Hunt et al., 1950) ) are assigned to hydroxyl and hydrate water respectively. A careful study (Buswell, Krebs, and Rodebush, 1937; Buswell and Dudenbostel, 1941) of the dehydration behavior of a series of montmorillo- nite salts demonstrated that the characteristic absorp- THE HYDROXYL ABSORPTIONS 2.75a 2.75/t. 2.75/^ 2.75/- WYOMING MONTMORILLONITE SALTS HYDRATED IN SATURATED VAPOR AT 23*C ■DRIED AT ROOM HUMIDITY (M/*a Ca^lN DRY AIR) BD i''iGi-RK : CHARACTERISTIC BAND POSITIONS IN MINERAL SPECTRA I 0-H p-H-O-H I r-O-H-0 1 CO3 \lf ll,< ^^ ^ \jj \ \\ I SO4 a PO4 h 2-119 13.3-14.0 QUARTZ 75 Si04 10 5 (-pOOUBLET 12.5 12.8 x9d9 , , I L. 2 3 4 5 6 7 8 9 10 II 12 13 14 WAVELENGTH, MICRONS » PHVLLOSIUCATES I-'K.IKK S. tion at 2.75 microns represents the presence of essential or lattice hydroxyl of structure sometimes called "un- bonded" or free hydroxyl because of its crystallograjihic similarity to the hydroxyl of the strong alkali metal bases (Wells, i945). The absorption at 2.92 microns is ascribed to hydrogen bonded hydroxyl. perhajis mainly hydroxyl of the clay surface. Figure 7 illustrates the relation- ship between these absorptions and the hydration states of some montmorillonite salts. The curves for the sodium, ammonium and hydrogen analogues are also consistent with this assignment. The hydration water assignment is made from the infrared spectrum of water as well afe by direct correlation with the spectra of the hydrated and dehydrated forms of simple inorganic crystals. Keller, Spotts and Biggs (1952) have just published a thorough study of silicate spectra. Their work indicates that the phyllosilicates or sheet-silicates, of which tlie clays are representatives, are characterized by one main absorption band centered between 9.0 and 10.0 microns. This position is assigned to the Si04 group. Quartz has a well defined spectrum with major absorptions at 8.6 Pai-t nil ilETHODS OF IpENTIFVING CI-AYS AXI) I NTFHCRETATIO.V OF Ufst'lts 115 ami !l.'2() microns and a cliap:nostic-ally useful doublet at V2.'^^) and 12.84 microns. These positions, tofretlier with the |)riMci|)aI absoi'ptions for carbonates, sulfates and piiosphati's. are shown in fiiiiire 8. Foi- the carbonates it has been tlemonstrated (limit and Turnei', 1953) that the 11.2 to 11.9 micron region is diagnostic, the absorp- tion here being shifted to longer wavelengths with in- creasing atomic weight of the cation. Conclusive dis- tinction between absorptions due to silica tetrahedra and ahniiinuin octahedra has not been made although Adier (1950 I has suggested a number of frecpiencies which may apply in the case of montmorilionite and a few related minerals. ATTAPUL6ITE KAOLINITE 6 > o I LUTE MUSCOVITE MONTMORILLONITE ALBITE MINERAL IDENTIFICATION TEMPLATE 8.0 9.0 10.0 11.0 12.0 WAVE LENGTH IN MICRONS I'li.nti: '.I. COMPARISON OF QUALITATIVE AND QUANTITATIVE TYPE SPECTRA 345678 9 10 WAVELENGTH, MICRONS 14 16 16 FlclKK 10. Qiialitdfirc and Qucnititative A)ire ioiocSire>o ' , Ma 1 1 1 i i i 1 ; ; ; i i ; i i i i i i != "1 ; ! ; : i i i ; si;;;; * *'* * 1 I ! 1 ! 1 I II lo ; ; ; ; ; •« »o »o >o ^1 ii iM i ii i i 1 i ii J 1 i MB M M i is:? s = 1 1 CC CO o o o ^, o o as 1:^:1 < SKX s s a a: x -sMx a ii iii isiiii ; ^ [ ■ : : ;; §■11 111 iJJJiiJis & S i • ii f i S - 'S ■£ 2 ; Ji = =?, ; 1 , ^ .3 «ab zise of syiitlietie. oriranie ion-ex- cliaiiirc resins mar he an aiiswei- ( Wicklaiiiler lOol ; Lewis If);"):! 1. 2. The complex relationships among the clay struc- tures and their spectra are almost completely unknown, althoujrli Adler and the Hunt and Keller groups have made a good beginning. Can siiecific absorjitions be correlated with tetrahc- drally bonded ions (Si*"* and AT** in montmorillonite and beidellite) and otliers with octahedraliy positioned ions to give us another mea.sure of structural (listinction between the two- and three-layer clays? Till' manner in whicli the sj)ectra reflect isomorphons replacement within the lattice might be advantageously studied by comparison of highly resolved and sur- face-homoiouized samjilcs of I'casonably homologous series such as pyrophyllite. nontronite, sauconite (Ross 1!)46). hectorite. saponite (Ross and Hendricks 1945). and talc in which the essential oetahedral positions are occupied by Al"*. Fe**', Zu**, Mg", Li*. Mg**. Jig", respectively. Of course, the degree of tetraliedral ]iroxy- ing of Al*" for Si"** will have to be taken into account both in selection of suitable samjiles and in interpreta- tion of the spectra. It appears that the infrared spectra may prove iiseful for classifying the nature of the bonds in minerals in terms of ])ereent covalent or ionic character. Pauling (1948). Gruner (19.10), and Keller. Spotts and Biggs (1952) have called attention to the covalent character of minci-al bonds. Keller and co-workers make the point that wholly ionic-bonded minerals do not absorb appreci- ably in the infrared, indicating thereby that the signifi- cant infrared absorption by minerals is evidence for some covalent bonding. What is the effect of the exchange cation upon the positions of the ma.ior infrared absorptions of the clay minerals? Do we find a shift toward longer wavelengths with increasing atomic weight as in the case of carbo- nates (Adler 1950; Hunt and Turner 19.^3) and non- clay silicates (Keller. Spotts. and Biggs 1952) ? Since the absorption spectra are presumed to be f\inc- tlons of the geometrical configurations not only of indi- vidual atomic groupings but also of the massive lattice, what aspects of the infrared spectra of kaolinite, diekite, nacrite, anauxite, hydrated halloysite, and metahalloysite reflect the order of stacking in the c-aixis direction? Does al'ophane show tip as the silica-alumina analogue of zero order ? Closely i-elated to the preceding suggestion would be a study of the effect of sample orientation ttpon the spectra ; e.g., compare the spectra of kaolinite laminae jiarallel to and perpendicular to the incident radiation. Wliat is the relation between the degree of hydrogen bonding and the exact location of the hydroxyl absorp- tion in the interval 2.7-3.2 microns? Does deuteration of hydrated clays jiroduce stifficient separation of the lat- tice hydroxyl and bouded-hydroxyl absorptions to per- mit estimating the relative amount of each form present ? If the presumption is correct that the characteristic absorption at 2.75 microns represents unbonded hy- droxyl of structure then upon calcining the clays at various temperatures up to 800^ C. there should exist a 1:1 correspondence between its appearance in the infra- red s]iectruni and the intensity of the diagnostic lines in the X-ray diffraction )iatterii of any given clay. Is the calcination temperature dependence of the re- tention of fluoriue-proxied hydroxyl in hectorite similar to that of the OH-montmorillonites? The spectral differences among the various natural hy- drated and non-hydrated halloysites as well as the artifi- cially ]irodiiced complexes such as "glycerol halloysite."' etc., should be examined in detail and att(>mpts made to correlate the spectra with their unusual morphologies as revealed by electron micrographs (Bates et al. 1950). Perhaps this will load to a better understanding of the structural changes accompanying the irreversible dehy- dration of halloysite. 3. The infrared absorption spectra of organo-clays should be investigated. The interest here lies in deter- mining the spectral effects of adsorption (as well as of real exchange in the case of organic cations) of organic molecules on clay surface. Thus, studies of the magni- tudes of the bathochromie and hypsochromic shifts (to- ward longer and shorter wave-lengths, resjieetively) in- duced by the different clay surfaces should be made. It is conceivable that the resultant data might enable a larger number of different minerals in a rock mixture to yield to quantitative analysis through mea.surement of the degree of .spectral shift of one or more of the promi- nent bands present in the orgauo part of the clay in a region of the spectrum which is normally featureless in the unclad clay. Along these lines, the work of Yaro- slavskii and Terenin (1949) on the infrared absorption spectra of ad.sorbed molecules and of Vedeneeva (1947) on changes in the ultra-violet spectra of dyes when ad- sorbed on clay minerals is of interest. The latter found that different samples of bentonite gave malachite green maxima differing by as much as 60 millimicrons. An im- portant conclusion from this work is that intensification of the ionic bond between clay and dye causes a batho- chromie effect, and intensification of the dipole bond a hjiisochromic effect. 4. An important contribution would be the develop- ment of better techniques for quantitative mineral analy- sis, particularly of montmorillonite in the presence of illite, kaolinite, feldspar, quartz, and carbonate. 5. The spectral range of study should be enlarged from the presently used 2 to 16 microns to from 0.75 to 25 microns. It is realized that this involves a difficult instrumentation problem, but the rewards for finding ad- ditional fundamental frequencies could well be important for raising the level of clay mineral analysis. 6. Finally, as an important aspect of the interpreta- tion problem, it is urged that many more spectra of widely diverse but carefully described clays and sedi- ments be obtained in order to expand the available li- bi'ary of reference citrves. Conclusion. It is concluded that the infrared absorp- tion spectra of clays and related minerals are, at present, of limited usefulness for the purpose of qualitative and quantitative analysis of complex and unknown mineral mixtures. However, the value of the method will undoubt- edly become greater as more is learned concerning the variations of the spectra with structure and chemical composition. In the petroleum industry, the infrared 118 Clays and (_'lay Technology [Bull. 169 spec-ti-a of minerals liaYe served as an important supple- ment to the data provided by X-ray diffraetion. eleetnm mierosfopy, and the other well-known methods for ex- amining minerals. Acknowledgment. The author is grateful to Dr. John M. Hunt and the Carter Oil Company for permission to use unpublished data ; to Professor W. D. Keller for his promptness in making available the results of his most recent studies; to Professor Ralph E. Grim for gifts of flay mineral standards; to Mr. James S. Brown of Union Oil Company of California for obtaining a number of the clay mineral speetra ; and to Union Oil Company of California for permission to publish this survey. DISCUSSION A. M.Soldate: What is meant by the phrase "tailored surfaces," and "homo- ionic and heteroionic surfaces"? P. G. Nahin: B.v a "tailored surface" I mean a surface of known composition, prepared in the laboratory according to specifications of the experi- mentor ; a homoionic surface is one on which every exchange site is occupied by a single element, as in a Na-montmorillonite ; heteroionic surface is one on whicli the exchange sites are occupientmorillonite. My Ijurpose in mentioning such surfaces is to urge that studies be made with clays of known and desired surface composition rather than with clays just as they come in nature. R. L. Stone: Dr. Xahiu, do you know of any study relating to the effect of particle orientation on infra-red absorption? P. G. Nahin: Xo. but I think that such :i study would lie very desirable; kaolinite may he the simi)Iest system to Iiegin with in stK'h a stud.v. Adolf Pabst: Tile effect of orientation on ab.sorption was examined extensively by physicists in many countries long ago. The matter has been summarized by Schiifer and Matossi (1!W0). P. G. Nahin: I believe that these deal with relatively large oriented crystals and not with powdered samples, the particle size of which are smaller than the smallest wave length used. The question of the effect on absorption of such oriented particles is still unanswered. SELECTED REFERENCES Adler, H. H., 1950, Infrared investigations of clay and related minerals : Am. Petroleum Inst. I'roj. 49, Prelim. Kept. 8, pp. 1-72, New York. Columbia Universitv. P.arnes, R. B., Gore, R. C, Liddel. U., and Williams, V. Z., 1944, Infrared .spectroscopy, industrial applications and bibliog- raphy : New York, Reinhold Pub. Corp., p. 11. Bates, T. F., Hildebrand, F. A., and Swineford, A., 1950, Mor- phology and structure of endellite and hallovsite : Am. Minerahtgist, V. 35, pp. 463-484. Bray, E. E., and Stevens, N. P., 1050, The preparation of clay samples for infrared absorption measurement : Am. Petroleum . Inst., Prelim. Rept. 8, pp. 7:'.-104, New York, Columbia Uni- versity. .Tour. Sci., V. 250, 12, pp. 827-828. Buswell, A. M., Krebs, K., and Rodebusb, W. H., 1937, Infra- red studies. III. Adsorption bends of hydrogels between 2.5 and 3.5/x : Am. Chem. Soc. Jour., v. 59, pp. 21)03-05. Buswell, A. M., and Dudenbostel. B. F.. 1941, Spectroscopic studies of l)ase exchange materials : Am. Cliem. Soc. Jour., v. 63, pp. 2554-58. Gruner, J. W., 1050, An attempt to arr:inge silicates in the order of reaction energies at relatively low temperatures: Am. Mineralogist, v. 35, pp. 138-148. Harri.son. G. R., Lord, R. C, and Loofbourow, J. H., 1948, Practical spectroscopy : p. 261, Prentice-Hall, Inc. Henry, R. L., 1948, The transmission of powder films in the infra-red : Optical Soc. America .Jour., v. 38, pp. 775-789. Hunt, .T. M., 1950, Infrared spectra of clay minerals: Am. Petroleum Inst. I'roj. 49 Prelim. Rept. 8, pp. 105-121, New York, Columbia University. Hunt, J. M., Wisherd, M. P., and Bonham, L. C, 19.50, Infra- red absorjition spectra of minerals and other inorganic compounds : Anal. Chemistry, v. 22, pp. 1478-1497. Hunt, J. M., and Turner, D. S., 19:53. The deteriuination of the mineral constituents of rocks by infra-red spectro.scopy ; Anal. Chemistry, vol. 25. pp. Il(i9-1174. Keller. W. D.. and Pickett, E. E., 1949, Absorption of infrared radiation by powdered silicate minerals: Am. Mineralogist, v. 34, pp. 855-864. Keller, W. D., and Pickett. E. E., 1950, The ab.sorption of infrared radiation bv clav minerals: Am. Jour. Sci.. v. 248, pp. 264-273. Keller. W. D.. Spotts. J. H.. and Bigg.s. I). L.. 19.52. Infra-red spectra of S(jme rock-forming minerals : Am. pp. 4.53-471. Lecomte, ,1., 1951, Jour. Physics Radium, Lewis, I». R., 19.53, Replacement of cations of clay by ion ex- change resins: lud. Eng. Chemistry, v. 45, pp. 1782-1783. Mellon, .M. G., editor, 19.50. Analytical absorption spectroscopy: p. 440, New York, .John Wiley & Sons, Inc. Nahin, P. G., Merrill, W. c", Grenall, A., and Crog, R. S., 1951, Mineralogic studies of California oil-bearing formations. I. Identi- fication of clays: Jour. Petroleum Technology (Am. Inst. Miu. Met. Eng. Petroleum Trans.), v. 192, pp. 151-1.58. Pauling, L. C, 194S, Nature of the chemical bond. 2d'i'd.. p. 24i), Cornell Uni\ersity Press. Plyler. E. K., 1929, Combination freipiencies of the infra-red bands of (pnirtz : Physical Review, v. 33, pp. 48-51. Ross. C. S.. 1940, Sauconite — a clay mineral of the montmoril- louite gnmp : Am. Mineralogist, v. 31, pp. 411-424. Ross, C. S., and Hendricks, S. B., 1945. Minerals of the inont- morillonite group, their origin and relation to .soils and clays : U. S. Geol. Survey Prof. Paper 205-B, pp. 23-79. Sands, J. D., and Turner, G. S., 1952, New development in solid phase infrared spectroscopy : Anal. Chemistry, v. 24, pp. 791-793. Schiifer, C, anil Matossi, F., 1930, Das ultrarote Spcktnim Springer, Berlin, 4(M> pp. Vedeneeva, N. E., 1947, Izmeneniya spectrov krasitelei pri adsorbtsii na mineralakh glin ( Changes in the S]iectra of dyes when adsorbed on clay minerals): Zhurnal Fizicheskoi Khimi : (Jour. I'hys. Chemistry) (U. S. S. R.), v. 21, pp. 881-891. Wells, A. F., 1945, Structural inorganic chemistry : p. 104, Oxford University Press. Wicklander, L., 1951, Saturation of colloids ami soils by means of exchange resins : Ann. Royal Agri. Coll. Sweden, v. IS, pp. 154- 162. Yaroslavskii, N. G., and Terenin, A. N.. 1949. Infrakrasnye spektry pogloshcheniya adsorbirovannykb molekul ( Infra-red ad- sorption spectra of adsorbed molecules): Doklady Akad. Nauk SSSR (Reports of the Academy of Science of the U. S. S. R.), V. 66, pp. SS5-88S. IDENTIFICATION OF CLAY MINERALS BY X RAY DIFFRACTION ANALYSIS Br Georgb W. Brindlev • ABSTRACT Sincp X-rny (lifFractiim pntti'iiis are diioclly iclalcil lo ciystal stnicturi's. X-i-ay idciililication is. ill inincipal, iH-ttcr suited to till- n'cii;;iiitii)ii (if stnictuial Ki-oups and structural varietii-s tliau lit' clifiuioal species. Well-foriueil k.iolin. mica, and cldciritc structures );ivp rise to cliar.Hli'ristic 7. 10 and 14A spaciu^'s which are relativel.v easily identitied. Ilydrated forms, such as liydrated halloysite (rf = lOA) .•iiid moutmorilliinoids under normal conditions ((; = 14A), are recof;ni/.eil either hy low-temperature dehydration jjiving charac- teristically diminished spacin!;s. or hy the formation of organic comple.xi's (liviiis; characteristically increased spacings. This intro- liuces at once the iirinciple that X-ray identification may. and frenerally does, ent.-iil the study of minenil modification In suitalilc chemical and/(U- thermal treatment. The in.iiii requirements in the X-ray technique are: (1) Aliility to record long sjiacings up to liri.\ or even higher values; |2| well focused lines with good resolution; (.">) alisence of background sc:iltering and white radiation anomalies. Of these. (1) and ('i) lie within the contnd of the investigator, hut (2) depends partly n of micas and chli>rites in any greater detail is difficult. A consideration of basal spai'ings and liasal intensities may make more specific identifica- tion possible. INTRODUCTION Tlic ]ir(il)lciii of ideiitif.viiit;' cIm.x' luiiiri'als b,\' X-ray (lirt'ractioii aiial.vsis may n.sefully be coiisidpi-ed in rela- tion to tlie selienie of classification, already outlined in 'i'alile 1 of the previous paper Structural Mineralorjy of ('l over photo- graphic and micro-photometric deteriuinations. As regards sensitivity in the recording of weak reflec- tions, the Geiger counter method is claimed to be as sen- sitive as photographic recording, but this is jn-obably not valid for all forms of the instrument and it calls for care- ful consideration so that weak but valuable reflections are not overlooked through the use of this very conven- ient instrument. The recording spectrometer is especially useful when it is required to examine a small angular range of the dilTractiou diagram for a range of speci- mens. Long Spacing Measiiremenfx. While the basic^ layer types have sjiacings of about 7. 10 and 14 to 15 A, and glycerol-montmorillonite has a spacing of 17.7 A, a num- ber of larger spacings have been recorded in recent years. Bradley (1950) obtained a basal spacing of 25.0 A with the mineral rectorite which contains a regular alternation of pyrophyllite-type and vermiculite-type layers. Alex- anian and Wey (1951) recorded a spacing of about 32A fnmi a moiitinorillonife from Camp Berteaux (Morocco). Caillere, .Mathieu-Sicaud, and Henin (1950) described a mica-like mineral from Allevard with a spacing of about 22A. These observations show the importance of extend- ing spacing measurements to values beyond 20A as a matter of routine procedure. The study of long spacings from hydrated Ka-mont- morillonite "by Xorrish and :MacEwan. which are reported at this Conference, emphasises still further the impor- tance of develojiing ade(iuate techniques for long spacing measurements. X-raij Powder Cameras for Clay Mineral Studies. Cameras of connnercial design for routine X-ray powder analysis are often inadequate for clay-mineral work since they cannot'be used for spacings as large as 17.7A (the minimum requirement for clay-mineral analysis) while air scattering of the incident X-ray beam and sometimes also scattering by faulty collimation restricts still fur- ther the measurement of long spacings. The use of powder layers in preference to rods or tubes of powder is recommended by some workers. MacEwan (1946) describes the use of a thin powder layer set edge- wise to the X-rav beam, and which may be given a small 122 Clays and Clay Technology [Bull. 169 (a) PLAN (b) VERTICAL SECTION (C) VERTICAL SECTION (d) SPECIMEN HOLDER Figure 1. Semi-focusing powder camera for clay inincral anal- ysis (BrincUey and Robinson). This consists of a lirass cylinder. C, soldered to a massive base, B, having additinnal thickness at the centre to take the conical .ioint, J, %vithin which rests the siiecimen holder. The outer wall of the cylinder and the conical .ioint are accurately co-axial. The powder is pressed into a shallow cavity in a small glass plate. O. which is held by a spring against the vertical face of the right-angled piece, R. This vertical surface is adjusted to lie accurately on the axis of the camera, so that the po\vder surface is automatically set on the axis. A fine .slit, Si, allows an X-ra.v beam having a divergence of about i° to fall on the powder surface ; S2 is a trimming slit. The film is held on the outer surface of C by a slightly stretched, black rnljber baud. I). The reflected X-rays reach the film through the aperture, A. and ti> avuid weaken- ing thereby the cylinder C, a stop, H, is inserted having knife-edges which cut into the direct beam above and below the e(iuatorial plane. The film holder is made light-tight by thin aluminium foil, F, or thin nickel foil when CuKa radiation is employed. The angu- lar setting of the powder surface is determined by a pointer, P, moving over a graduated scale. Q. Diagrams (a), (b) and (c) are drawn (in the same scale, the external diameter of tlie cylinder C being 20.0 cm; diagram (d) is on twice this scale. angular oscillation. The present writer has employed for many years flat powder layers set at an appropriate an- gle to the incident X-ra.v beam, a method which gives partial focusing of the reflected beams over a useful range of angles. Figure 1 shows the type of camera iised by Brindley and Robinson (1946) in their study of the structure of kaolinite. "With the powder plate set at an angle a— 3° to the incident X-ray beam and with CuKa radiation, spacings ranging from about 4A to 20a, or beyond, can be sharply recorded; with a— 9° or 12°, a range from about 7a down to about 1.5a is obtained. *" t'i-^ ' Sandstone JMSBk i S.ltslone I I Mudslone J^ t j < Shole 7.10 14 FiGl'RE 2. Semi-t'ucusing, multi-exposure camera for day-min- eral an.alysis (Brindley and (.'ro(.)ke). This camera takes four pow- der siiecimens which are tightly jii'essed into cavities in perspex or glass holders, jireferably from the back surface to minimize orien- tation of flaky minerals in the front surface. The reflections from the four powders are separated by three .aluminium baffles and are recorded on a film held on the outer surface at the camera b,v a black rubber liand. Four X-ray beams are selected by a slit .system at Si and trimmed by adjustable knife edges at S=. 51 is a lithium fluoride monochromator. The camera is of ma.ssive construction .so that the side plates. A and B, can be removed without disturbing the curvature of the film mount. The jMiwder diagrams show the 7. 10, and 14a lines from kaolin, mica, and chlorite components, and illustrate the results obtained when the camera is set to focus in the range from 5-20A. The increasing diffuseness of the mica Hue in passing from the s.andstone to the shale is readil.v seen. Fi,gure 2 shows a camera designed and used by Brindley and Crooke, with which four powders can be examined simultaneously and separately. The princii^le is closel.v similar to that discussed by de Wolff (1948), but is modi- fied so that surface reflection rather than transmission recording is obtained. Since cla.y minerals give few, if any, reflections at 2(9>!)()'\ it is unnecessary to extend the recording range far be.vond this angle. By cutting off the higher angles. as in the cameras shown in figures 1 and 2, it is possible to use large-diameter instruments (20 em diameter is used bv the writer) and at the same time place the speci- men relatively close to the X-ray source. This arrange- ment, combined with the use of surface reflection and partial focusing, enables high resolution diagrams to be obtained without excessively long exposure times. Tlie spectrometer also uses fiat powder layers with the additional advantage that the reflections are always re- corded under conditions of sharp focusing. The amount of powder required by the spectrometer is considerable, especially if the material has to be hand-picked under a microscope and further purified b.v sedimentation or other methods. "With photographic recording, layers of smaller extent suffice ; the writer connnonly eraploj-s a I'art in Methods of Tdextifyixg Clays axd Ixterpretation of liFsn/rs 123 powder cavity of about 5 by 3 by 0.5 mm but wheu little material is available, a thin layer about 5 by 1 mm (or even smaller dimensions) on a glass plate may be used. While Radiatiou Effects and T'se of Monorhromafor.s. Clearer jiowder diag:rams ;in' obtained by using erystal- nionoclirDmatized radiatimi in jireference to filtered radiation. Curved crystal monochromators, usually of quartz, combined with cameras of a focusing type give shar]ily focused patterns against a low background in- tensity. Plane monochromators nuiy be used with powder specimens, of flat or rod type, mounted at the centre of a circular camera. For clay-mineral investigations, syn- tiietic lithium fluoride is probably the most generally suitable monoi'hromator. It is a very strong i-cflector and is comiilctely stable tinder atmos]iheric coiulitions. The writer has found when using filtered radiation that the peak of the white radiation reflected fi'om the very strong (10.1) qviartz reflection appears in a positicm corresponding to CuKa radiation reflected from a spac- ing of about n to 12A. In consequence, clay materials containing appreciable quartz can prod\ice this spui-ious reflection in the region where basal reflections fi-om clay minerals are found. Such anonuUies are cliniinatfd by using crystal-monochromatizcd radiation. It is necessary to remember, however that many crystal monochromators reflect not only the characteristic wave- length /. from the X-ray tidie. but also the harmonies of wavelengths /. 2, X/S, ... in the second, tliird. and higher orders, which are selected by the monochroniator from the general radiation. The following may be cited as an illustration of what may happen. The writer, examining strong basal reflections from a chlorite using a lithium fluoride monochromator set for CuKu radiation, recorded additional weak reflections a]i]iarently corres|)ouding to odd orders of a 28.A spacing. Xo trace of these reflections was obtained with the usual filtered radiation. Tt even- tually became apparent that these additional reflections were due to /.(CuKa) ''2 radiation. Such effects are largely eliminated by using fluorite or diamond as a monochro- mator since the second and higher order reflections from these crystals are of negligible intensity compared with the first oi'ders. DIFFERENTIATION OF CHEMICAL SPECIES J'ae of Spacliiu and Inl( nsifii M((isiir< )iir)if.i. This is the most difficult phase in the identification of minerals by means of X-rays, but under favourable circumstances considerable progress can be made. The problem to be considered is how the X-ray diagram of a particular mineral structure varies with its composition. It has already been stated that it is principally the sizes and the scattering powers of the atoms which are involved, and that lattice dimensions and reflected intensities are the observable parameters. This means that atoms of similar atomic number, and therefore with similar scattering factors, must be differentiated primarily by dimensional considerations, and atoms of dififereut elec- tronic contents but similar sizes must be differentiated by intensity considerations. Thus in ferro-magnesian min- erals, the Fe : Mg ratio will often be deternuned more sensitively by intensity observations than by spacing measurements, though both may be useful. On the other hand, the replacement of Si by .\1 in tetrahedral coortli- nation and of Al by llg in octahedral coordination can- not be ascertained by intensity measurements, but some progress can be nuule by using dimensioiud arguments. These concepts have been partially applied to the mont- morillonoids by AlacEwan (1951), to the micas by Brown (1951). and to the chlorites by von Engelhardt (1942), and Bannister and Whittard (1945). A more systematic api)roach to the i)roblem is re(|uircd. however, and a first step in this direction has been made by Brindley and MacEwan (1953) in which formulae are derived relating layer dimensions, more i)articularly the /)„ — i)arameter. to the sid)stitutions of atoms in tetrahe- dral and octahedral positions. .\ brief survey is given in the preceding i)a|)er by the writer in this symixisium. Structural Mliirndogij of Clans. The formulae are gen- eral and are intended to be applicable to all Uiyer sili- cates. Comparison of observed parameters with calcu- lated values shows a clo.se over-all agreement, but some exceptions occnr. notably among the micas. This leads to the question whether the interlayer cations, which have been omitted from these calculations, may play an ap- jireciable part in detei-mining the layer parameters. The relations between lattice parameters and composi- tion given by MacEwan, Brown, von Engelhardt. and Bannister and Whittard. were each obtained on the basis of a more restricted argiunent and therefore may be more exactly applicable within the domain considered by each of them. Di- and Tri-oclahedral Minerals. An early attempt to identifv chemical species by X-ray diffraction was made by Xagelschmidt (1937) who showed that the (060) spacing of dioctahedral micas was about 1.50A and of trioctahedral micas about 1.53 to 1.55 A. This important distinction has been confirmed by much subsequent work both with the micas and with other mineral groups, and constitutes the first stage in attempting to identify a particular chemical species. Kaoli)i-t!iiH Miurnds. This term covers all minerals with the same kind of layer structure, namely, the kaolin minerals lu-ojier. the serpentine minerals, chamosite, amesite, etc. Among these minerals, the problem of identification is relatively simple because isomorphous substitution of atoms occurs to only a small extent. Thus the kaolin minerals proper have essentially aluminum atoms and the ser]ientine minerals essentially magnesium atoms in oetaheih'al positions. Chamosite appears to be the only widely occurring ferrous iron member ; its composition is slightly more variable. Other kaolin-type minerals such as amesite, cronstedtite, garnierite. and greenalite are relatively rare. The powder diagrams are sufficiently distinct for "there to be little real difficulty in making a broad separation. 3Iicas, Chlorites and Montniorillnnoids. xVmong these minerals extensive isomorphous substitution of atoms oc- curs and the designation of the chemical species depends principally on the nature and extent of the substitutions. In addition to the work already cited on the recognition of these mineral species by lattice spacing and intensity mea.surements. a short account may be given of a critical test carried out bv F. II. Gillery in the writer's labora- 124 Clays and Clay Technology' I Bull. 160 2(2SiOj. SFeO.ZHjO) JIAIjOj.SiOz. 2Fe0.2HjO) 2l2Si02 .SMgO-ZHjO) 3(Al203.SiOj,.2Mg0.2 H^O) Figure 3. X-ray identification of chlorite species. Black circles represent the chemical anal.vse.s of three chlorites, and the open circles their evaluation by purely X-ray methoiLs ; the shaded areas indicate the estimated uncertainties iu the X-ray evaluations. (Brindley and Gdlery I tory to identify unspecified chlorites solely by X-ra.y aualysis. A Detailed Test of Chlorite lelentificedion hy X-rays. The following- methods and assumptions were used : (i) The octahedral positions were assumed to be fully occupied and the distribution of atoms to be the same in the two octahedral layers of the chlorite structure. The intensities of the first five basal reflections were then calculated with the assumption that Al and Si in tetra- hedral positions scatter equally, that 'Shj: and Al in octa- hedral positions scatter equally, and similarly Fe and Cr. The basal intensities can then be used to determine the ratio of the heavily scattering atoms (mainly Fe) to the lightly scattering atoms (mainly Mg and Al) in the octahedral positions. (ii) The proportion of Al to 8i in tetrahedral posi- tions was determined from the curve relating basal spac- ing to this ratio (Bannister and \Yhittard. ]!U5; Brind- ley and Robinson, 1951). A simplified general formula for chlorites can be written : (Mg,Fer.--.x-yAl.Ji) (Si4-.Ay^0,„) (OH)s An important assumption here is that the iron is wholly ferrous, which is largely true for many chlorites. On this basis, x can be found from the basal spacing curve and the ratio of Fe** to (Mg + Al"), viz. (6 - :r -y)/(x -\- y). from the basal intensities. Thus x and y can both be evaluated. If Alx'^' is partially re- placed by Fe***, an additional variable is introduced which cannot be evaluated at present purely from X-ray data. In figure 3 are shown the final results obtained for three chlorites. Using the well-known diagram of "Win- chell (1936) showing the compositional and optical ranges of members of the chlorite group, the X-ray esti- mation of three chlorites is indicated by shaded areas and their chemical compositions by closed circles. The agreement is sufficiently good to show the potentialities of the X-ra.y method. It is evident that the Fe**/Mg ratio has been cletermined more reliably than the Al^'/Si ratio. This is not surprising, because the basal spacing- will probably depend partly on the substitutions in the octahedral layers. It should be observed, however, that X determines not only the Al'^ Si ratio but also the resultant charges in the layers and this may well be the dominant factor in determining the basal spacing of a chlorite. Tlie results shown in figure 3 justify a restrained optimism as regards the evaluation of the chemical species of a layer mineral purely by X-ray methods. The main requirement is the accurate determination of the basal spacing and basal intensities and this implies that these reflections shall not be masked by reflections from other minerals in a mixture. Herein lies the main experimental difficulty. In the ease of a mica, the lOA and 5a reflections are unlikely to be masked, but in the case of a chlorite the 7a and 3..JA reflections correspond very closely with reflections from kaolin-type minerals, a point which has already been discussed above. The kaolin reflections can be suppressed by heating the ma- terial to about 450° to 550°C., but such treatment may modify the chlorite intensities and vitiate their use for determining the Fe/Mg ratio. The writer has recently encountered the problem of determining the nature of a chlorite which is very fine grained, particle size 1 micron or less, w'hich appears to be closely intergrown with a kaolin mineral. .It has not been possible so far to separate these minerals and the identity of the chlorite remains unknown. IDENTIFICATION OF MINERAL VARIETIES Structural varieties of each chemical sijecies differ principally in the stacking sequence of the layers. Such differences influence the finer points of the X-ray diffrac- tion diagrams. Their recognition turns on the number and kind of components in a material. The varieties of some chemical species are more easily recognized than those of other species. Thus the kaolin varieties, kaolinite, dickite, naerite, halloysite, and the characteristically disordered form of kaolin mineral found in fireclays, are relatively easy to recognize, provided the powder diagram is not too heavily overcrowded with other reflections. The writer has been able to recognize kaolinite and dickite together when quartz was the only contaminant. A small percen- tage of halloysite along with kaolinite would not be easily detected. The extent of the disorder in a kao- linite would not be easily estimated in the presence of mica, quartz, feldspars, and possibly other im]iurities. The structural varieties of the micas have been .studied especially by Hendricks and Jeft'erson (1939) using single crystals. The characteristics of these varieties are not easily seen in powder diagrams, but Grim and Brad- ley (1951) have shown that if mica is the dominant mineral in a sample, then there are characteristic lines in the powder diagrams which enable some varieties, at least, to be recognized. This calls for careful and de- tailed study. ;irl III MkTIIODS of iDE.VTll'YIXf; ('LAYS AND InTKRI'HETATKIN OF ReSIT>TS 125 Tlic stnu'tiiral varieties of cliloi'itcs liave been studied so far only by siii<»:le-ervstal methods (Briiidley, Oiijili- ton, and Robinson, lOoO) and it is donbtful if powdrr (lia-axis so that the only reflections (ihlaiued, other than the basal (00/) reflections, are the two-dimensional (/(A-) bands. The positions of maximum intensity near the low-angle limits can be indexed ap- proximately as (hkO) and the entire sequence of re- liectious then appears to correspond to an orthorhombic (If orthohexagoual cell. This, however, is an incorrect (lcscrii)tion of the lattice. The band of intensity extend- ing to higher angles from each (hkO) position corres- ponds closely to an (hkl) sequence with the h and A- indices having fixed integral values and I taking values extending continuously from 1 = 0. The reason why the ])ositiou of maximum intensity corresponds only approxi- mately with an (hkO) designation depends partly on the variation of structure factor with angle and partly on the fact that each angular position in a band recorded in a powder diagram coi-i-esponds to a small range of / values. Theoretical treatments of diffraction by small, two- dimensional crystal lattices have been given by Warren (1041), by Wiison (1949), and by Brindley and Mering (1951). By making certain reasonable assiunptions, War- ren was able to arrive at explicit formulae for the dis- tribution of intensity in the ditfraction bauds which Brindley and Robinson (1948) have shown give good agreement with observed data for halloysite, Brindley and llering analysed the problem in rather more detail but the final conclusions were that Warren's approxi- mations were, in most cases, amply good enough. There is more difficulty in recognizing the one-dimen- sional disorder which often takes the form of layer dis- placements parallel to the &-axis by amounts of nbo/S. Displacements by this simple fraction of a unit cell side occur in the kaolin- and chlorite-type structures be- cause, in the hydroxyl sheets of these layers, the (OH) radicals are situated at intervals of &„ 3 parallel to the /;-axis and in consequence a displacement by nbu/'i introduces no change in lu'ar-ueighbour relations be- tween adjacent layers. Two effects follow from the random &-axis displace- ment of laj"ers. In the first place, reflections with k ^ 0, 3, 6 , , . remain unaffected, but reflections of type (hkl) with /.• y^ 3» disappear. The two-dimensional regu- larity of the la.vers remains, however, so that bands, similar to those produced by halloysite and moutmoril- lonite, occur when k=^tin. These bands are usnally nuu-h weaker than the strong (hkl) reflections with k. = 3m and in consequence are not easily seen. The .stronge.st band produced by all layers of the type found in clay minerals is the (02, 11, 111/ band, the three components of which completely overlap. The maximum of this baud eorresi)()uds closely with a lattice spacing of bn/2 and occurs therefoi'e at about oiut, atoms differ in size and in scattering jiower. Therefore in determining chemical species, attention must be paid to details of lattice spacings and diffracted intensities. A critical test of the identification of three chlorites is described and it is concluded that in favour- able cases the possibility of identifying species by X-ray analysis can be regarded with restrained optimism. The problems of recognizing structural varieties are briefly outlined and the potentialities indicated. Sometimes the recognition, of a particular structural variety determines at the same time the chemical species. Disorder in clay minerals is of two kinds, namely dis- order due to Tuixed-layer seqxiences and disorder due to layer displacements. The characteristic diffraction phe- nomena arising from these disorders are discussed and the possibilities of studying them in detail are considered. Questions of X-ray technique are treated briefly with special reference to the measurement of long spacings 126 Clays and Clay Techxolooy [Bull. 169 and eliiuinatian of spurious effects arising from the use of filtered radiation and of certain crystal mono- chromators. ACKNOWLEDGMENTS Finallv I Avisli to record my deep appreciation of the invitation to attend the First National Conference on Clays in the United States, and my pleasure at the op- portunities it has provided of visiting many people and laboratories in this coinitry. I wish also to express my best wishes for the success of the Clay Minerals Group which has been initiated at this Conference, and the hope that despite the distance separating Europe and America, there will be useful collaboration between the newly founded United States group and the groups now actively functioning in Belgium. Britain, France, and Sweden. DISCUSSION T. F. Buehrer: Il.)\v accuriitel.v f:iii dih- (Ii-lcriiiiuc the proiinrtioii nf the various minerals present in .a colloidal ola.v of a soil, ami liy what fac-tors would suc-h estimates lie oomplieated? G. W. Brindley: The method for (piantitative mineralosical analysis hy means of X-rays is the well laiown one of adding a standardizing snbstauce in known proportion and comparing the ratios of the intensities of diffraction lines from the different minerals with the intensities of lines from the standard snhstance. The choice of standard substance depends to some extent on the problems to be undertaken ; the dif- fraction pattern of the standard should interfere as little as pos- sible with those of the components to be measured. In the case of day-mineral mixtures, it is not easy to find a standard which meets this requirement satisfactorily, but the writer and R. H. Crooke in work not yet fully published have found ferric ammo- nium sulfate to be a useful substance. The accuracy obtainable is usually not very high but the use of self-recordiug counter spectrometers in place of photographic tech- niques increases consider.-ilily lioth the accuracy and the con- venience of the X-ray method. The principle of the method is to compare suitable reflections from a range of minerals such as kaolinite. mica, montmorillouite. chlorite, etc., with a siiitalde reflection ov reflections from the standard substance in binary mixtures of known proportions. In a multi-component mixture, the intensity ratios will be related in the same way to the composition ratios provided the particle sizes of all the components are sulHciently fine. This proviso raises difticult problems both theoretically and experimentally if the particles are so coarse that they absorb individually an apprecia1)le percentage of the incident X-radiation. This, however, is not likely to arise with the clay fraction of a soil. A more serious difficulty arises in connection with the standard intensity ratios. Such ratios can, of course, be established for particular specimens of kaolinite, mica, montmorillouite, etc., in relation to a chosen standard. The cpiestiou is whether the minerals to be estimated reflect X-rays exactly as the kaolinite. etc., used in making up the standardizing mixtures. This depends on the "crystallinity" of the minerals involved and also on their precise chemical composition. A further source of uncertainty arises from the easy orientation of flaky minerals. The most convenient reflections for intensity measurements are often the liasal. OOl, reflections and these are the reflections most susceptible to orientation effects. The second part of Dr. Uneluer's question is therefore more easily answered than the first part. The accuracy obtainable de- pends so much on circumstances that a simple, direct answer is scarcely possible. I could, however, hazard a guess that a crystal- line component present to the extent of .50 percent by weight in a mixture might well be estimated to within ± 5 percent, i.e., an accuracy of 10 percent in the determined \alue. But a component prescntto the extent of only .") percent might lie estimated with an uncertainty of ± 2* percent, i.e. about .">() percent in the determined value. I feel that X-ray methods for quantitative determinations should be applied with considerable circumspection and that it is a wise precaution always to vary as v.idely as possible the experimental 5 Figure 4. 100 30 -r Ten-A and 12.4A spacings : dioctahedral mica layers. From Broun and ilacEwan lO-il. conditions, in order that the many elusive factors which may influ- ence intensity measurements can be given an opportunity to reveal themselves. Measurements carried out with small variation of the experimental conditions may show a high degree of consistency among themselves that is quite misleading as regards the accuracy of the determination. M. E. King: Another factor beside orientation, which introduces uncertainties in estimating a given crystalline compouent in a mixture, is the presence of substances which highly absorb X-ray.s, such as calcite. In one sample X-ray analysis indicated that quartz was present in an amount of about l.'i to 20 percent, whereas microscopical analysis indicated alMiut (!0 to G.5 percent quartz. After the calcite was removed, the results obtained by means of X-rays and micro- scope were in agreement. I'ait III Mktikids ()i- iuKXTirviNG Clays and Ixteki-rktatiox ov 1\i:sii.ts 12 •<^ CD 03 uj (D CI -^ * CO 03 «I ^ ^ QD 03 e laypi-s. The mimber attaclied to each curve frives the pruixn'tioii l)al)ilit.v of finding a layer at a j;iveii distance from any layer chosen as origin: the (arbitrary) zero level is not shown. The letters indicate the significance of the peaks, .1 meaning a lOA spacing, and /{ a 1:;.4A spacing; thus .l.l/{ is a peak which arises from two sp.acings of the first type plus one of the second (in any order t. The meaning of the dotted curve is evplaiiied in file text. From MtirEiraii. Xiiliirc. r. 777. G. W. Brindley: 'I'he ell'ect of strongly alisorliing constituents in a composite powder is (piite complex and has been discussed by .several writers ll'.rindley. 104."! : Wolff. ]!I37: Wibhinsky. la'il). Mention may also be made y clay minerals is that of random interstratification. This consists in the interleaving of difl'erent layers, in a manner which nuiy be either regular or random. The layers may be either of different structural types, e.g., kacdinite and montmorillonite, or may consist of layers of the same structural type, but with different thicknesses of interl.'imellar material, water or lU'ganic molecules — e.g., partially bydr.'ited balloysite ( F.riudley and Goodyear 1048), partially exjianded montmorillonite, chlorite, and swelling chlorite (Stephen and MacEwan 10.511. The basal series of reflections from .such material may contain broad bands, hut quite often it consists of lines which are not very noticeably less sharp than those from other fine-grained material, but which do not form a rational series of orders, i.e., a series with spaciugs in the ratios 1 : i : i : .... Whenever such an irrational series is present, we may suspect random interstratification. We are then confronted with the jiroblem of finding out what such a non-rational series of lines means. Sometimes they may be modified, e.g., by removing water, or by saturating the material with glycerol or glycol: and the modifications in the jiattern of lines may give us a clue to what was originally present. This can- not alwavs he done however, nor is it always sufficient to give us the clue we were seeking. Iresent and their proportions. In the case where the silicate layers present are of similar nature, and are separated liy interhnnellar material of relatively small scattering power, the prolilem simplifies itself into the de- termination of the inter-layer spacings present, and the proportions in which they are present. This can be done, as Mf-ring (1040) has pointed out, by calculating the fourier transform of a modified cuiTe of scattered intensity from the base, as a function of reciprocal spacing (or, nearly, of angle). Unfortunately, this curve is very difficult to obtain in practice, because the scattered intensity is very low in <-ertain regions, and is liable to he interfered with by non-basal refiecti(ms from the same mineral, and by reflections from other ndnerals. We have found that, where the pattern is one of fairly sharp discrete lines, quite a lot of information cm be obtained by calculating the much simpler function : /, -(•Os2 7r}l- /A where ?r = integrated intensity of rth line; Fi = scattering function for a single layer (this is a function of n) ; iir = recip- rocal spacing corresponiling to rth line; 7f = interlayer .separation in A. This is a very simple summati(Ui, having only as many terms ;',s there are lines. The resulting function of />', when plotted, should show peaks corresponding to the interlayer distances present, their heights giving the ])roporti(uis in whicli they occur. The function may be regarded either as a simidified fourier transform, or as a I'atter.sou function with non-rational indices, i.e.. willi infinite unit cell. Often, owing to the paucity of observations, the series is arbi- trarily cut off at a certain term. This gives rise to diffraction, i.e., to spurious peaks, as w'ith an ordinary Patter.son summation, and the remedy is the same, to smooth the series by arbitrarily reduc- ing the higher-order terms, i.e., by multiplying the intensities by a function which diminishes with increasing /i. This broadens the peaks, making exact spacing measurements difficult, so it may be convenient to calctd.-tte both the smoothed and the uusmoothed series. The sec(uid will give the accurate spacing values, and any pe.-iks not common to the two series will be re.iected. There are other refinements and snags, which cannot be described in detail here. Figures 4 ami ."> show what can lie done, even with very restricted information. Figure 4 gives the diffraction to be expected from mixed-layer structures containing mica-type layers with sepa- rations of 10 and li2.4A, in the proportions shown. These i-nrves are calculated by Brown and MacEwan (lOoO), and do not go beyond m = H."), i.e., r/ = 2.8 A i jj. being arbitrarily defined as 100/rf). there being only three peaks in this region, one of them rather diffuse. Figure 5 shows the smoothed summation from these three peaks. For comparison, an unsmootbed summation corre- sponding to the last of the threi' curves is also shown, in ilotted line. It is easy to .see that the initial portion of the sununaticms gives us very adequate infornnition about the spacings present, and their proportions. Taking the / = 0.7 curves as an example, we see from the dotted curve that the spacings present are 1(1 A at 12.4 A. The heights of the.se peaks are in the ratio of :i4:(>t!, which is near enough to the true value of .30:70. There is some difliculty in determining the true zero level from which to measure these heights, but this cannot be discussed in detail here. Com- parison of the dotted and full curves shows that peaks at 4 A, (!.". A. l.'>..') A and 10 A, may be rejected as being the result of diffraction. Thus only these two si)acings are present. The other peaks at larger values of R result from second nearest neighbors, thii-d nearest neighbors, etc., and from them we may, in principle, determine whether the mixture of layers is completely random, ordered, or partially ordered. Thus, if we had a regular alternation of layers, A BAB .... we should get a large AB peak, but no .4.4 and no BB. If we had a complete segregation of the two types, we should get .4.4 and BB. but no .17?. .\ little consid- eration will show that, with a completely random mixture of layers, we would expect the heights of .1.4. .4 7?. and BB to be in the ratio 0:42:40 with :'.(> percent A. This is clearly very close to what is found, so we may say that the mixture is essentially a random one, and this of course is the correct answer. 128 Clays and Clay TEciiNOLoaY I Bull. l(i!) These summatii)iis are very sensitive to the value of fir but are relatively little afteeteil even by quite larj;e ehauRes in Ir\ Fi\ -'. Thus we may expect that they will give the correct ansvi'er even when there is some uncertainty about the nature of the layers, and th<'rofcii(' about the values of Fi. H. F. Coffer: When is orientation of clay particles in a when is it not? sample ilesiralile ami G. W. Brindley: Orientation is desirable when you want to l)ring out the basal reflections, i)articularly those at higher angles, so that they can be differentiated with certainty from other reflectiims. On the other hand if you want to compare the intensities of the reflections in order to check on a structure analysis, then, of course, you wish to measure the intensities withcnit the comidications introduced by orientation. In a recent structure analysis of amesite (Urindley et al. 1!I51) both single-crystal and powder analyses were employed. In order to interpret the pow.">-.">6. Bannister, F. A., and Whittard, W. F., 1!)4.">, A magnesian chamosite from the Wenlock limestone of Wickwar, Gloucester- shire: Mineralog. Mag., v. 27, pp, !I9-]11. Bradley, W. F., liloO, The alternating layer sequence of rec- torite : Am. Mineralogist, v. 3,j, pp. oyO-o!!.!. Brindley, O. W., 194."), The effect of grain or particle size on X-ray reflection from mixed powders and alloys, c. 200-2S4, ilineralog. Soc, I-oiidcui. iCbiy Minerals Group I . Caillerc. S., .Mathieii-Sicaud, A., and Hciiiu, S., 19.5(P. Xouvel ,essai d"identification du mineral de la Table pres Allev:ird, I'alle- vardite : Soc. Francais Mineralogie Crystallograi)hie Bull.. \'. 73, pp. 193-2(»1. Engelhardt, W. von. 1942. Die Struktiiren von Tburiugil, Bavalit und Cbamosit nud ihre Stellnug in der (_'bIoritgrii]»pc I The struc- tures of tburiugite, iKU'alite, and cluimosite and their position in tlie chlorite gnnip) ; Zeitscbr. Kristallograpliie, Band 104, pp. 142-1.59. (iriin, R. E., and Bradb',>", A\'. F., 1951, The mii-a clay minerals, in Brindley, G. W., Editor. X-ray identification and crystal struc- tures of clay minerals: Chap. 5, part I, pp. 138-154, Mineralog. Soc. London. (Clay Minerals (iroup). Hendricks, S. B., and .leffer.son. M. E., 19.39, polym.uphism of the micas, with optical measurements : Am. Mineralogist, 24, pp. 729-771. Ileuilricks. S. B.. and Ross, C. S., 1938, Lattice limitation of monlniorilbmite : Zeitschr. Kri.stallograpbie, Band 100, pp. 251-2114. ll.'udricks, S. B., and Teller, E., 1942, X-ray interference in p:irtiallv ordered layer lattices: .Jour. Chem. Physics, v. l(i. pp. 147-1(57. I 'an III Methods of Identifying Clays and Interfketatiox of Results 129 Jatkson, M. L., Hseung, Y., Cori'j , K. li., Kvans, E. J., niid lleuvcl, K. C. v., litiJL', WiMtlicriiif; sc(iu('iic(' (if clay size minrrals iu soils and sediiiiPiits. II. Cliemical wcatlieriug of layer silicates: Soil Sci. Soc America I'roc., v. lO. pi). .'i-O. Kerr, P. F., 1050, Analytical data on reference clay minerals: Am. I'etroleum Inst. Proj. -It), I'relini. Ke|it. 7, ItJO pp.. Xew York, Columliia University. MacKwan. D. .M. C. 1040, The iilentilication and estimation of montniorillonite Kroup of niineials, with special reference to soil clays: So('. ('hem. Industry .Iiivir., v. <>.">, pp. 'J98-304. MacEwan, I). M. C. lO.'il, The monlniorillonite minerals (mont- morillonoidsi , in lirindley, (i. W.. Edilnr. X-ray identification and crystal structures of clay minerals: Chap. 1. pp. 86-137, Mineralog. Soc, London. (Clay Minerals Croup i. MacEuan, I). M. C, ami Einch, G. I., V.^^,0. Electron dift'raction by niontmorillonite : Paper read to Clay Minerals Croup, April 29, 1950 I uupuhlished). Mering, J., 19-19, L'intcrf^reuce des rayons X dans les syst&mes it stratification desordonnee : Acta Crvstallo^raphica. v. 2. ]ip. .'hI- 377. JK'ring, J., 1050, Les refie.xions des rayons X par les minereux argilenx inters! ratifies : Fourth Internat. Cong. Soil Sci., Amster- dam Trans., v. 3, pp. 21-20. Nagelschmidt, G., 1937, X-ray investigations of clays. Part III. The differentiation of micas hy X-ray powder photographs: /.eitschr. Kristallographie, Band 07, pp. 514-521. Nagelschmidt, t;., 1941, The identification of clay minerals by means of aggregate X-ray diffraction diagrams: Jour. Sci. Instru- ments, V. IS, pp. 100-101. Nagelschmidt, (',., Gordon. R. L., and Griffin, O. G., 1952, Sur- face of finely ground silica : Nature, v. 169, pp. 538-540. Stephen, I., and MacEwan, I). M. C, 1951, Some chloritic min- erals of unusual type: Clay Minerals Bull., v. 1, pp. 157-162. Walker, G. F., 1949, Distinction of vermiculite, chlorite, and montniorillonite in clays: Nature, v. 164, pp. 577-578. Warren, B. E., 1941, X-ray diffraction in random layer lattices: Physical Rev., v. .50, pp. 693-698. Wilchinsky, Z. AV., 1951, Effect of crystal, grain, and particle size on X-ray power diffracted from powder : Acta Crystallogra- phica, V. 4, pp. 1-9. Wilson, A. J. C, 1049, X-ray diffraction hy random layers, ideal line profiles and determination of structure amplitudes from ob- served line profiles: Acta Crystallographica, v. 2, pp. 245-251. Wilson, A. J. C, 1950, Geiger-counter X-ray spectrometer — in- fluence of size and absorption coefficient of specimen on position and shape of powder diffraction maxima : .lour. Sci. Instruments, V. 27, pp. 321-325. Winchell, A. N., 1936, A tliird studv of chlorite: Am. Mineralo- gist, V. 21, pp. 642-651. Wolff, P. M. de, 1937, A theory of X-ray absorption in mixed powders : Physica, v. 13, pp. (!2-7S. Wolff, P. M. de. 1948, Multii)le (iuinier cameras: .Vcta Crys- tallographica, V. 1, pp. 207-211. ELECTRON MICROSCOPY AS A METHOD OF IDENTIFYING CLAYS By Thomas F. Bates ' ABSTRACT Electron microscope studies sliow tlmt most of tlie clay minerals have morphological characteristics which can be effectively used to aid in their identification. In addition, detailed studies at high niagnification have provided structural information not obtainable by other means. In the kaolinite group each of the common minerals has a clear-cut and diagnostic morphology. However, much more work is needed on flint clays and the kaolins in soils before the same can be said of these varieties. Continued study of halloysite reveals significant details pertinent to the form and structure of the tubular crystals. Illites from different sources show interesting differences in form which reflect variation in the degree of crystallinity. An illitc from the Ordovician Oswego formation shows an unusual development of narrow to broad lath-shaped crystals. Except for lower content of iron and magnesium the material has all the characteristics of other illites. The thinnest laths are of the same order of thickness as the flakes of more common illites. Certain mixed-layer minerals are identical in morphology with some varie- ties of illite. The minerals of the montmorillonite group have been the most difficult of the clay minerals to characterize on the basis of morphology. The lath-shaped crystals of certain beidellites, sauco- nite, nontronite, and hectorite are diagnostic. The metal shadowing technique shows that many of the laths approach one unit cell in thickness. The morphology of many montmorillonites is dependent upon the mode of sample preparation. The replica method offers a means of studying clay particles in their natural state but this technique is still in the early stages of development as it relates to the clay minerals. Introduction. The method of electron microscopy relies for its chief advantage over other techniques upon the application of the well-worn principle that "seeing is believing. ' ' Two clays may give the same X-raj* pat- tern, the same difiPerential thermal curve, and so on, biit if thej' look different, we know that something either of a fundamental nature or related to environment, growth, and possibly subsequent alteration has resulted in the morphological traits which now distinguish them. In the electron micrographs which follow, significant details of mori^hology will be pointed out, and those ■which aid in distinguishing one clay material from an- other will be emphasized. In general, the method of preparation of the material to be photographed was the simplest possible. A small amount of clay was mulled in distilled water, the suspen- sion diluted the proper amount, and a drop allowed to evaporate upon a substrate of polj^vinyl formal (form- var) which was supported by the usual disc of 200-mesh screen. Most of the pictures were taken on a model EMU electron microscope by Mr. Joseph Comer, electron mi- croscopist of the College of Mineral Industries at the Pennsylvania State University. Most of our work on the morphology of fine-grained minerals has been supported by a Penrose Grant from the Geological Society of America and a contract with the Geophysics P>raiich of the Office of Naval Research. The electron micrographs included here were chosen either because they are representative of a particular variety of clay or because they illustrate some significant morphological feature. With two exceptions our own pic- tures were used, chiefly because it was convenient to do • Professor of Mineralogy. The Pennsylvania State Universitv State College, Pennsylvania. so. There is no intent to minimize the very important contributions made in this field by other laboratories. Except where noted otherwise, the scale on all figures represents one micron. Kaolinite Groiip. The first electron micrograph (fig- ure 1) is that of kaolinite from the Woodbury clay de- posit near Huntingdon, Pennsylvania. The difference between thin and thick crystals (A and B respective^) can be distinguished by variation in density. This picture shows the difference in appearance of single crystals as opposed to oriented aggregates. Studies of kaolins found in fireclay, flintclay and soils are more ditficvdt, less satis- fying, but of considerable importance. The relatively small amount of electron-microscope work that has been done on these materials shows that discrete particles and aggregates are in some instances flaky but do not com- monly show the perfection of cr>-stal habit illustrated by the china clay type of kaolinite pictured in figure 1. Figures 2 and 3, which show a mixture of halloysite and kaolinite from the Maria Elizabeth bauxite deposit in British Guiana, illustrate several interesting points. The large tubes are not typical of most halloysites but were chosen to illustrate certain interesting features which are not apparent in smaller crystals. Evidence obtained by several workers indicates that kaolinite plates, if thin enough, will curl at the edges and possibly roll up into tubes. This is consistent with the theory that curvature results from the misfit of the tetrahedral and octahedral layers and the presence of weak interlayer bonds (Bates, Hildebrand, and Swineford 1960 ; Bates 1951). In figure 2 the crystal marked A might either be an endellite tube unrolling or kaolinite rolling iip. The hexagonal character of the plates suggests the latter. The tube at B also has hexagonal terminations and in this respect is not typical of halloysite tubes in general. Figure 3 is a composite of three electron micrographs of material from British Guiana. An interesting differ- ence may be noted in the morphology of the cross sections shown in the top aiul bottom portions of the figure. That at A gives the appearance of overhqiping plates, while at B the exterior of the tube appears smooth. In both cases the polygonal rather than truly circular aspect of the sections is of interest. The difference in thickness along the center tube at Point G represents a change of wall thickness of 100 A. Figure 4 is of a more typical halloysite from Alex- ander, North Carolina. Here a large number of the tubes have unrolled, giving rise to flakes of irregular shape. Anauxite (figure 5) is another mineral of the kaolinite group which needs more study. Like dickite it commonly occurs in crystals somewhat large for electron microscope investigation. Figure 5 shows how much some of the material from the lone formation of California resembles kaolinite. Illite Group. The clay minerals in the illite group are not as photogenic as those described above ; neverthe- less they have distinctive and interesting morphological characteristics. ( 130) Part III] :^lF.TnoDs OF Identifvixg Clays and Interpkktation of Rksvlts 1:31 Figure 1. Ka„liiiite from Woodbm-y day deposits, Huntingdon, Pennsylvania. .). thin, and B, thick single crystals. Graphic scale repre- sents one micron. 132 Clays axd Clay Techxology I Bull. 169 Figure 2. Kaolinite, li;illu>sitp mixture from Maria Elizabeth bauxite deposit, British Guiana. A, flalie with rolled edge; B, tube wilh hexagonal terminations. Graphic scale represents one mioi-on. Pari Iltl ;\It:t!1()I)s (IF 1i)i;n TiKVixf; Clays and TxTKKi'iiKTATiox of Rkst'lts ]:« FioiRi: :;. S:iiiH' sample sh"wii in li^'iirc 2. .1, tulie showiu^ overlappins; effect; B, another end view of a tnlie ; C, lOOA increase in wall tliickness. Graiihic scale represents 0.1 Ciieron. 134 Clays and Clay Teciixolooy [r.iill.l(i9 1''IG1_1!E 4. ll;illnysite from Alex;ni(lci-. Xm-tli (':irciliii:i. Criipliic scale rfpresents one micron. J'art nil Methods ov Tdkntifvixo Ci.ays and Interpretation- of Results 135 FiGUKi; 5. Aiuuixite from lone, California. DrtsiUiiiK of sulisti-ate lias prohalily caiiscl apiiariMit roliiii;,' up of hir^c particles al \<,k,-v right. (Jrapliic .scale represents one micron. 136 Clays and Clay Technology [Bull. 169 Fii.rm: (i. llliit lium Fitliian. Illinois. .1. flake with lif.xasoiuil oiitliiii'; /;, liix.Tsoiial-appcarins aggre;;atP. Jlotnl sliailowod w i Grajiliic scale repre.^ents one mioruu. Pt.-I". Part 111 .Mkthods of Identifying Clays and Interpretation of Results 137 ^ «■■.* V FiGUKE T. lUiti; frum Juniata formation, Jlilc-lnu;;, T.-iin^ M«t,l. (Iiaphic scale represents one micron. 138 Clays and Clay Technology [Bull. 169 Figure 8. Brammallite, from Coal Measures of South Wales. Cr-shadowed. Graphic scale represents one micron. Part III] IMethods op Identikyixg Clays and Interpretation of Results 139 Figure 0. lllur ii-m Oswego formation, State Colleg... I'-nnss Ivania. Pt-Pd shadowed. Graphic scale represents one micron. 140 Clays and Clay Technology [Bull. 169 Figure 10. Fullers earth from Surrey, England, Ba-saturated. Arrows indicate tiny spherical units which form larger two- and three- dimensional aggregates. Cr-shadowed. Graphic scale represents one micron. Part III] Methods of Identifying Clays and Interpretation of Results 141 Fl(,i i!l. 11. J[..iitinui-ill'ii.itt; Iroiu \Vf.-t inH' walcr i)f swcIliuK iniKlit have l)een removoil iluriiiK llic evaporation of tlio niotal on tlic cla.v and that the observed swellin;; represents onl.v a minor part of the actual swellinfT of the particle in its wet state? T. F. Bates: The work with double shadow casting is very preliminar.v, and we are enoounterinj; nian.v dilticulties. We have no idea as to the orientation of the individu:il plates in the aKKregate and therefore no real calilmition of the length of the doulile shadow in terms of expansion. ]5,v working with some of the phity niontmorillonites (hectorite, nontronite, etc.) which i)resnmal)ly are oriented with the C-axis parallel to the beam, the production of two shadows would permit a measurement of the expansion. This, however, demands some very precise work. Padraic Partridge: Have you seen .-my dendritic patterns in your (dectron micro- graphs which could have been formed from evaporation of soluldes in the water, or that came from the clay or its alteration products? I have seen such patterns and wondered about the cause. T. F. Bates: We sometimes run into them but so far have been able to eliminate their occurrence by using freshly distilled water in the preparation of our suspensions. H. F. Coffer: What wouhl .vou estimate to be the thinnest sheet of a moiit- inorillonite that you have photographed? T. F. Bates: We have not as yet seen an.v montmorillonite in well defined sheets that can be measured. Thus far we have found montmoril- lonite to consist of fluffy aggregates without crystalline pattern. «■■■■> RephcQ (R) from Sample (S) Metal shadowinq of replica W ;i m\\i\\ii.. Microqraph of shadowed replica Figure 18. Negative replica technique. 148 Clays and Clay Technology' [Hull. 169 FiGUKE Itl. Keplica of nontruuite fruin Sandy Rid^'e, Xnrtli Caruliiia. Pt-I'd shadowed. GrapLic scale represents one micron. Part 111 ilKTiious OF Identifying L'lays and Interpretation of Results U9 Figure 20. Keplica of nontronite specimen showing interesting rolls. Pt-Pd shadowed. Graphic scale represents one micron. 150 Clats and Clay Technology [Bull. 169 R. L. Stone: In the pictures you showed, with few exceptions, there is a con- siderable amount of aggregation of the particles. Will it be possible in the future to take a picture of a clay in its most dispersed form? T. F. Bates: This has already been achieved for minerals of the kaolinite group. With proper control of the sample, such as regard to pH and exchangeable cation, it should be possible to take photographs of any clay in a highly dispersed condition. In the case of mont- morillonite there has been more success in this respect in France than anywhere else. H. F. Coffer: I believe that this can be done by use of the freeze-dry technique. A 2 or 3 percent highly dispersed suspension is prepared and frozen rapidly at about minus 80° C. The water is then driven off while the suspension is in the solid state. I have not applied this to montmorillonite but believe it would be successful. M. D. Foster: Recently Ed Dworuik made an electron micrograph of Mg- saturated montmorillonite from Santa Rosa, Mexico. It gave a lieautiful crystalline pattern, much better than any other electron micrograph that I have seen of montmorillonite. T. F. Bates: I have recently received a manuscript by Mr. H. P. Cahoun of the University of Utah which describes a Utah saponite with very well-formed, large, thin, and almost equi-dimensional square plates. They are crystalline, and have well-defined angles, but these angles do not seem to be related to the usual crystallographic directions. Padraic Partridge: What is Murray's technique of dry-sampling for the electron microscope? H. H. Murray: A toothpick wrapped with lens paijer is dipped into the powdered .sample and rolled on a glass slide coated with collodion. The collo- dion film is removed from the slide by floating it on distilled water and then mounting it on a ^-inch disk of 200-mesh -screen. The sample is then shadow-cast and electron micrographs taken. SELECTED REFERENCES Bates, T. F., 1947, Investigation uC the micaceous minerals in slate : Am. Mineralogist, v. 32, pp. 625-636. Bates, T. F., 1951, Morphology of layer lattice silicates : Deni- son Univ. Bull., v. 42, pp. 83-91. Bates, T. F., Hildebrand, F. A., and Swineford, A., 19.''i0, Mor- phology and structure of endellite and halloysite : Am. Mineralo- gist, v.' .35, pp. 463-484. Grim, R. E., Bray, R. H., and Bradley, W. F., 1937, The mica in argillaceous sediments: Am. ilineralogist, v. 22, pp. 813-829. Mathieu-Sicaud, A., Mering, J., and Perrin-Bonnct, I., 1951, fitude an microscope glectronique de la montmorillonite et de I'hec- torite saturees par difffrents cations : Soc. frang. mineralogie Bull., V. 74, pp. 430-456. Weaver, C. E., 1953, A lath shaped non-expanded dioctahedral 2:1 clay mineral: Am. Mineralogist, v. 38, pp. 279-289. DIFFERENTIAL THERMAL ANALYSIS OF CLAYS AND CARBONATES By Richards A. Rowland*' ABSTRACT DiffiTi'iilial tlicrmal analysis (DTA) hpsaii soon after tlie de- vt'lci|iiiiciil (if tile tlu'i-miii'Duplc. It lias iironri'ssod tlimiigh tlie systematic ilcvcliipincnt of liPtter equipment and the catalofjiiinR of typical DTA curves for a variety of materials until K<>od teolinique now rei|uires control of the composition and pressure of the furnace atmosphere as well as consideration of the thermo- dynamics and kinetics of the reactions involved. Although dif- ferential thermal analyses have heen made for many materials, the major ajiplications have been concerned with clay and car- bonate minerals. In I>T.V curves for clay minerals the low-temperature endo- thermic loop associated with the loss of water, and the high- temperature exothermic loop accompanying the formation of new compounds, are changed in shape, temperature, and intensity by the kind of exchange cations. The midtemperature-range endo- thermic loop has a temperature dependence on the partial pres- sure of water in the furnace atmosphere. For the anhydrous normal carbonates the dissociation tempera- ture and its deiiendence on the i>artial pressure of C0= are in the decreasing onler ("a, Mg, >In, Fe. .and Zn. The lower temperature loop of dolomite, the reaction for which nuist be preceded by an internal i-earraugement, is independent of the jiressure of Cll^. Iiul may be shifted to a lower temperature by prolonged fine grinding which aeeomiilishes a similar rearrangement. INTRODUCTION Differential tliennal analysis (DTA). although not a very aecurate or definitive method, has found an impor- tant place amonp' teehniques whieh allow the characteri- zation of materials, l^imited only by the sensitivity of the apparatus, the differential thermal curves record all transformations in which heat is taken up or given otf. This includes the dehydration of elaj's, the decarbona- tion of carbonates, the reversible change from a- to p-qiiartz, the burning of materials, and the recombina- tion of elements into more stable forms. "When employed alone, the technique can be used to identify a number of reasonably pure compounds and to follow changes in mixtures for control purposes. When used in conjunc- with X-ray diffraction, microscopy, and chemical analy- sis, otherwise difficult identifications can be made. The technique is not easily standardized, however, and the factors which frequently make it difficult to compare DTA curves prepared in different laboratories are sum- marized by Ahrens (1950), The development of differential thermal analysis has progressed through several stages. As early as 1887 le Chatelier described the use of his thermocouple as a difference thermocouple and published DTA curves of kaolinite. From that time until Oreel (1935) began the systematic differential thermal analyses of clays, about twenty miscellaneous DTA papers appeared. Another stage began with the design of good furnaces, sample holders, and photographic recording equipment by Norton (1939) and Hendricks (1939). Refinements of this design by Grim and Rowland (1942) were followed by further developments by Berkelhamer and Spiel (1944). Throughout this period many papers appeared which repeated the thermal curves of the same clay samples and related oxides, and a portable apparatus * Publication No. 25, Exploration and Production Technical Divi- sion, Shell Oil Co., Houston, Texas. •* Senior Geologist, Exploration and Production Technical Division, Shell Oil Company, Houston 2.5, Texas. was developed by Hendricks (1946)^ for use in study- ing bauxite deposits in the field. The last development in the basic apparatus was the visual recording of the DTA curves of a number of samples being heated in the same furnace. Simultaneous development of DTA tech- niques for the elementary study of carbonate minerals took place in the U. S. A., .lapan, and the U. S. S. R. Reconsideration of the thermodynamics of the sj'S- tem gave rise to a very sensitive sample holder (Gruver, 1948) (Kaufman and Dilling, 1950) made of platinum foil. Ilerold (1948) developed a thin sample holder half platinum and half platiniiin-lO percent rhodium in which the thermocouiile .in net ion. built into the sample holder, was a ring around the middle of the e.ylindrical sample. Development of static atmosphere control within the furnace was introduced by Saunders and Giedroyc (1950) and Rowland and Lewis (1951). Dj'uamie at- mosphere control within the sample was introduced by Stone (1952)- Presently the trend is toward atmosphere control at elevated pressures where DTA reactions liegin to approach equilibrium reactions. From the simple ap- proximate mea.surement of the effective temperature dif- ference obtained by comparing the temperature of the reaction of a sample in its own atmosphere with that of an inert standard, the technique has now progressed to a consideration of the heat exchange under controlled conditions of an inert atmosphere or of a participat- ing gas. KINDS OF TRANSFORMATIONS The endothermic and exothermic deflections of a DTA curve record many kinds of changes of state. The only limitation is that ^ ^ , the rate of change of enthalpy (A£r), be sufficient for the temperature difference to be registered before dissipation in the system. First-order phase changes, which involve discontinuities in volume, entropy, and the first derivatives of the Gibbs fiuiction (AF) are represented by two kinds: the reversible al- lotropic inversion of alpha to beta quartz (Faust 1948) (Grimshaw, et al. 1948) and the irreversible monotropic change of aragonite to calcite (Faust 1950). The change from endellite to lialloysite probably is a monotropic phase change. Definite second-order phase changes, in which there is no discontinuous change in volume and entropy while the second derivatives of the Gibbs func- tion change discontinuously, are rather rare. One which is habitually recorded in DTA, employing a nickel block as a sample holder, is the change from ferromagnetic to paramagnetic nickel (Curie point) at 353° C. Murray and White (1949) have discussed the kinetics of thermal dehydration curves. Alost of the chemical reactions recorded by DTA are first-order reactions in which the 'rate of reaction is directly proportional to the concentration of the reacting substance. The dehydration of claj' minerals such as kaoliuite_ and the dissociation of 1 This apparatus is available commercially from the Eberbach Cor- poration, Ann Arbor, Michigan. 2 Variable pressure DTA apparatus is available from Dr. Robert L. Stone, Austin, Texas. ( 151 ) 152 Clays and Clay Technology [Bull. 169 carbonates are chemical reactions of this type. The very poor curves obtained for rauscovite — because the rate of dehydration for the usual heating rates is very slow — also represent a first-order reaction. Second-order reac- tions in whieli the rate depends on the concentration of two molecules, and third-order reactions where the con- centration of three molecules controls the rate, are not common in the interpretable DTA reactions. Combina- tions of first- and second-order reactions, and perhaps some third-order reactions, probably take place after the final breakdown of the clay mineral lattice when new high-temperature products are formed. The kinetics and thermodynamics of the DTA method are actually too complex to permit the application, in any sense other than approximate similarity, of these physical-chemical terms for better-known reactions. This rather incomplete discussion of jjhase changes and order of chemical reactions is included because it has become increasingly popular to refer to DTA curve deflections as representing a specific kind of chemical reaction or phase change. LINE FOR KflOLINlTE aASEO ON 5P HEAT OflTA TAKO a CORNWALL KAOLINS 1000 / 'K VAN "T MOFF LINES FOR SEVERAL MINERALS [AFTEH STONE. J A CER S 15. I9i2 1 FiGUKE 1 THERMAL THEORY Spiel (1945) and Kerr and Kulp (1948), by opposing the thermal efi'ects — the heat of the thermal reaction and the differential heat flow between the block and the sample — arrived at an expression to show that the area enclosed b.y the loop and the base line is an approximate measure of the total heat effect and, under certain condi- tions, is proportional to the amount of thermally active material in the sample. By making a set of calibration curves with prepared mixtures of dolomite and calcite, Rowland and Beck (1952) were able to show that this relationship can be used to determine dolomite in lime- stone when as little as 0.3 percent is present (fig. 13). "Wittels (1951) varied both the heating rate and the mass of the sample to obtain an expression and calibra- tion so that precise calorimetric measxtrements can be obtained from DTA curves. M. Void (1949) has derived equations for the calcula- tion of heats of transformation from differential heating curves, which are independent of external calibration, by using the rate of restoration of a thermal steady state to CURVE "A" IN AIR 500 400 CURVE "B" IN COa A K h — ^ ..'.. '• 300' .n 100 200 CURVE "C" WITH COVER -^ h ^-T- 100* 200 300 400 CURVE "D" diluted i 1- 500 600' 700* 600' 900' 1000 C 300' '°0 DTA CURVES OF SIDERITE FiGUKE 2 establish a relation between the differential temperature and the heat adsorption producing it. Valid results were obtained for such widely dift'ering processes as tlie melt- ing of stearic acid and the vaporization of water. A highly significant contribution to the understanding of differential thermal analysis was made by Murray and White (1949). They point out that a Clausius-Clapeyron DOLOMITE 0T4 CURVES AT I MM TO 760MMC0j PRESSURE (AFTER MAUL 6 HEYSTEK. AMER MIN 37, I9S2) Figure 3 I'arl 111] Methods of Idexthyixg Clays and Interpretatiox of Rksults 153 RAH ID AIR DTA OF ORGANIC -CLAY IN NITROGEN FlGUKE 4 typft equation can be reduced to a plot of In Ph2o vs 1 r to obtain a straigrht line of slope — Ai/ 2/?. By select- ing a number of partial pressures of H-jO and observing from the DTA curve the value of t°C. at which the loss of liydroxyl water begins. Stone (l!)o2) assembled data for a van't Hotf line from the slope of which the heat of reaction can be calculated (fig. 1). Comparison of these heats of reaction with values obtained from specific heat data shows that, for minerals of the kaolin group, the temperature at the beginning of the deflection of the DTA curve is considerably higher than equilibrium tem- perature up to a partial pressure of In p == 6.50 (665 mm). Above In p = 6.50 better agreement is obtained. For calcite, good agreement is obtained at In p = 3.8 (447 mm). Stone concludes from these experiments that at temperatures close to equilibrium in dry air the kaolinite decomposition reaction must be very slow in- deed. These experiments show that, even though the clay minerals have very similar structural arrangements, tlieir DTA hydroxyl-loss loops can be shifted selectively by control of the partial pressure of water vapor. Hence, clay mineral DTA cui-ves so obtained should resolve the midrange endothermic loops which interfere when the furnace atmospTiere is uncontrolled. ATMOSPHERE CONTROL Atmo.sphere control in differential thermal analysis has taken several different forms. Wlien a sample is heated in air. it biiilds up its own atmosphere, but not in excess of one atmosphere pressure. A typical example is the dissociation of siderite (Rowland and Jonas 1949) (fig. 2), in which the DTA curve is a compromise be- tween the endothermic effect of CO2 liberation and the exothermic effect of iron oxidation, until the CO2 evolu- tion is violent enough to exclude oxygen and the endo- thermic effect predominates. Oxidation resumes when CO2 evolution slows down, and the endothermic loop is interrupted by an exothermic loop. A similar effect is shown by the DTA curve when dolomite is heated in air. The curve in air resembles the curve at about 360 mm of COo (Haul 19511 (fig. 31. When a cover is used on the sample holder, the main oxidation loop of siderite is dis- placed to a higher temperature. Except when the sample well is covered, the pressure of the evolved gas probably never attains one atmosphere pressure and is quickh' re- duced by diffusion to a much lower concentration. These atmospheric effects are not controlled but are a function of the sample dissociation. The atmosphere of a furnace may be maintained at about one atmosphere partial pressure by allowing a gas to fiow through the furnace (Rowland and Lewis, 1951). This technique is sufficient for many applications where approximately one atmosi)he7'e of an inert gas, or a par- ticipating gas, is required. A better technique, using a sintered block for a sample holder, has been described bj'' Saunders and Giedroye (1950). Thii^ method insures that the gas surrounds the individual grain of the sample from the beginning of the analysis. Neither of these methods y)ermits control of the partial pressure of the gas. and the composition is maintained only so long as no air is swept in with the gas. Actual control of the pressure within the furnace has been used as a vacuum technique by Whitehead and Breger (1950). A dynamic system for control of the pressure and composition of the atmosphere surrounding the particles of the sample was described by Stone (1952) who included the sample holder in the gas- handling system. With this arrangement it is possible to maintain a continuous supply of fresh gas moving through the specimen at a predetermined pressure. Atmosphere control can be used to eliminate unwanted exothermic reactions resulting from the burning of or- ganic matter in clays (fig. 4). DTA curves of some car- bonates, particularly calcite and dolomite, are greatlj' improved by an atmosphere of CO2. From DTA curves made in a dynamic steam atmosphere van't Hoff lines can be constructed. While van't Hoff lines constructed from DTA curves only approximate equilibrium at ele- vated pressures, they are a summary of the DTA curves at several pressures and as such may be more charac- teristic of the material than the original DTA curve. DTA CURVES OF CLAYS A.side from a number of papers describing systematic studies of the collections of clays and carbonate minerals to learn what differences could be observed by this tech- nique, there have been a number of studies involving the factors controlling the individual parts of the differ- ential thermal analysis curves. The geometry of a differen- tial thermal curve of a clay is usually made up of three di.stinct parts. The first is a low-temperature endothermic loop which is registered when atmospheric water departs from the material. A second or midrange endothermic loop accompanies the loss of bound water or the dissoci- ation of hydroxyls from the lattice. The third is a high- temperature combination of loops accompanying the final breakdown of the lattice and the formation of one or more new materials. Low-temperahire Loop. The low-temperature loop, which may cover the interval from 50°C. to about 240°C., is dependent on the kind of clay mineral for its pres- ence ; on the type (bivalent-univalent) and amount of exchange cations for its shape; and on the moisture content, or the relative humidity surrounding the clay 154 Clays and Clay Teciixoi.ogy [Bull. 169 ooo ooo 5% 10% 25% 40% 50% 70% 90% DTA CURVES OF MISSISSIPPI MONTMORILLONITE WITH SEVERAL COMMON CATIONS AT DIFFERENT WATER CONTENT [AFTER HENDRICKS. NELSON 8 4LEK4N0ER. J A C S 62,'9401 prior to analysis, for its size. In general, members of the kaolinite group do not show a low-temperature peak. The exception is hydrated halloysite; its peak can be irre- versibly destroyed by storage over a period of time in an atmosphere of low relative humidity at room tempera- ture, or by heating to slightly more than 100° C. The three-layer lattice clay minerals invariably have a low-temperature endothermic loop. Of these, the mont- morillonite loops are the largest and most sensitive to moisture content, humidity, and type and amount of exchange cations. Although the illites also exhibit a low- temperature loop, the true micas, such as muscovite and biotite, do not. Chlorite in clay-mineral particle size has a low-temperature endothermic loop, but chlorite from metamorphic rocks does not. The effect of exchange cations on montmorillonites and illites is frequently rather marked. Hendricks (1940) pointed out the effect WYOMING BENTONITE 300 ■'00 5Q0 200 UNTREATED of a number of different exchange cations on different bentonites (fig. 5). In general, clays with monovalent cations exhibit one endothermic loop at about 150°C. ; most clays with bivalent cations have a second loop or a shoulder on a loop similar to the monovalent loop at a higher temperature (220°C.). Various organic com- pounds, particularly those which blanket the space be- tween the layers of the lattice, also have their particular effect on the hydration loop, but this is frequently ob- scured by the immediate volatilization or burning of the organic material. As yet, no one has succeeded in making use of the area of the low-temperature endothermic loop to deter- mine either the total moisture content or to make a quantitative estimate of the type and amount of exchange cations on the elav. DTA CURVES OF DICKlTE I OURAY, COLORADO) AT DIFFERENT PRESSURES OF WATER VAPOR I AFTER STONE. JACERS 35.19521 Hiyh-femperafure Loops. At the high-temperature end of the differential thermogram most of the recorded loops are the combined heat effect of several reactions, both endothermic and exothermic in nature. Grim (1948) and Stone (1952) have pointed out that, even in kao- linite, a very small endothermic loop occurs and is inter- rupted by the large exothermic loop usually associated with the formation of mullite. The high-temperature zone for members of the montmorillonite and illite groups is largely controlled by the chemical composition of the material. This involves the amount and kind of isomorphic substitution within the lattice and the nature of the exchange cations. ^lost of the three-layer lattice clay minerals undergo an endothermic i-eaction associ- ated with the final breakdown of the clay mineral lattice (Grim, 1948) and with the loss of a small amount of water which supposedly results from the loss of the last hydroxyls. Different persons have different ideas as to just what happens during this endothermic reaction. McConnell (1950) theorizes that tetrahedral hydrosjds give rise to the small water loss, and occur in groups of four, substituted for silicon in the tetrahedral layer. It is also possible that the hydroxyls are supplied from local substitution of magnesium in the octahedral layer. While there appears to be no reason for one part of the octahedral layev to retain its character at temperatures Part II ri ^Methods of Identifying C'l.ws and Intf.kpketation of Results Table 1. Firing products of several clays. 155 High alumina 900OC. 1000" C. 1100° c. 1200* C. 1300" C, a-Al;03 (a) Y-AhOo (a) spinel (a) spinel Cb) mullitc (a) mullite (a) a-AljOi (a) g-quartz (a) anorthitc (?) (c) spinel (b) cristobalite (e) g-quartz (a) enstatite (c) spinel (a) spinel (a) at-quartz (b) spinel (a) g-quartz (b) spinel (a) a-quartz (b) cristobalite (a) mullite (b) spinel (c) &-quartz (a) nristobalite (c) anorthite (7) (c) cristobalite (a) spinel (a) cristobalite (a) 0-quartz (a) enstatite (b) spinel (a) quartz (c) cristobalite (a) spinel (a) cristobalite (a) spinel (a) spinel (a) cristobalite (b) mullite (a) cristobalite (b) mullite (a) cristobalite (b) mullitc (a) cristobalite (a) cristobalite (a) spinel mullite (b) cristobalite (a) cordierite (a) cristobalite (a) spinel (a) mullite (a) cristobalite (a) spinel (a) cordierite (b) Endellile Diaspore Gibbaite_ Bauxite (Kaolinite anil pibbsite) Monttnnriiloiute group Beidcll. Ccilo mullite Cheto Fairview. Utah. cristohalito cristobalite (a) cordierite (a) mullitc (a) Otay. Calif.-- Palmer, Ark _, mullite (b) cristobalite (c) cordierite (a) periclase (c) cristobalite (b) Sierra de Guadalupe cordierite (bj cristobalite (a) Tatatila, Vera Cruz cordierite (a) cristobalite (a) mulUte (b) cordierite (b) Wagon Wheel Cap. Coin Woody nontronite cristobalite mullite cristobalite spinel Parenthetic letters siEnify: (a) important, (h) moderate, and (c) minor. (After Bradley & Grim, 1951.) higher tliaii that attained by other parts of the same layer, it is still possible to draw the parallel between the temperature at which gibbsite loses its hydroxyls versus the temperature at which brucite loses its hj'- droxyls. Other magnesium-bearing minei'als, such as talc and chlorite, seem also to lose their hydroxyls at tem- peratures somewhat higher than encountered in mate- rials consisting primarily of aluminum in the octahedral layer. Bradley and Grim (1951) have described manj^ of the factors controlling the nature of the immediate high- temperature products (table 1). They point out that the DAYS STANDrNG 66 146 200 400 600 600 OTA CURVES OF SODIUM MONTMOBILLONITE AFTER HEATING TO INDICATED TEMP- FOR I HOUR AND STANDING FOR DIFFERENT PERIODS (AFTER GRIM 8 BRADLEY. AMER MIN 33.1946) KlOL'ltE 8 — MONTMOniLLONlTE ^^^ZKGLiSH KAOLIN -Die KITE DTA SHOWING EFFECT OF STEAM INJECTION ON DRIED CLAY MINERALS lAFTen STONE. J A CCR S ii. '9il } Figure 9 156 Clays axd Clay Technology [Bull. 169 exchange cations can give rise to a vai'iety of spinels and cordierite. When the exehange ion between the layer positions is blanketed with an organic compound so that at elevated temperatures the only exchange cation present is hydrogen, the formation of mullite occiirs even with a three-layer lattice clay mineral. In figure 6 the exothermic loop at 930°C. accompanies the formation of a spinel in the untreated sample, mullite and spinel in the NH4 sample, and mullite in the remaining sam- ples. In some cases where there is a return to the base- line between the endothermic and exothermic reactions and where lithium is present in the clay minei'al, the accompanying excess silica appears in the form of beta quartz instead of cristobalite. Midrange Loop. The endothermic loop occurring at midtemperature range and associated with the major loss of hydroxyls from the octahedral layer varies con- siderably from clay to clay. In the kaolinite group this is an intense reaction which probably starts at a much lower temperature but is sufficiently strong to cause deflection at about 450°C. and to peak at about 600°C. Dickite, the most highly organized member of the kaolin- ite group, has a slightly different differential thermal curve through the range of loss of hydroxyls. The low- temperature side of this loop is quite steep, while the high-temperature side is at a somewhat lesser slojDe. The result is a loop skewed toward the low-temperature end. The starting and peak temperatures of the midrange loop of both dickite and kaolinite can be shifted by Ph2o of the furnace atmosphere (fig. 7). Wyoming bentonite and other bentonitic materials in which the order of stacking and the organization of the crystals are very good, have a loop beginning at about 57.5°C. and peaking at about 700° C. When the organization is poor, as is the case with most sediments containing mont- raorillonite, this loop is approximately 100°C. lower. The loop for nontronite, the iron analog of montmorillonite, also occurs at a somewhat lower temperatui-e. Members of the illite group lose their hj-droxyls over the same appi-oximate range as do some of the less well- PERCENT CALCITE 100 300 500 700 900*C 100 300 500 700 900 IIOO'C MAGNESITE AT .-^^ DT« CURVES FOR SOME RHOMBOHEORAL CARBONATES l*FTER KERR a KULP. AMER, MIN, 33, 1948) FUilKE 10 EFFECT OF DILUTION — DTA CURVES OF CALCITE ALUNDUM MIXTURES (flfTER KULP. KENT. KERR. AMER MIN 36, 1951) Fkiikk 11 organized montmorillonites. In sediments which may contain both illite and montmorillonite, it is seldom pos- sible to distinguish between montmorillonite and illite with differential thermal curves. In fact, the shales and clays of the Gulf Coast, at least to the base of the Terti- ary, appear to contain both an illite and a very poorly organized montmorillonite which may be in effect a de- graded illite in which a large portion of the potassium has been lost. Previously this loss of hydroxyls was considered to be an irreversible reaction. However, Grim and Bradley (1948) (fig. 8) demonstrated that clays heated to a temperature just below the end of their differential thermogram dehydration loop will reabsorb a consider- able amount of moisture as hydroxyls when exposed to an average relative humidity over a period of time. From his experiments using steam atmospheres. Stone suggests (fig. 9) that more rehydration may be obtained at ele- vated steam pressures. DIFFERENTIAL THERMAL ANALYSIS OF THE CARBONATE MINERALS The carbonate minerals are especially amenable to dif- ferential thermal analysis. Normal anhydrous cai'bonates undergo dissociation in an atmosphere of CO2 at progres- sively lower temperatures in tlie order Ca, Mg, Mn, Fe, and Zn (fig. 10). The temperature of the dissociation of calcite is very sensitive to the partial pressure of CO2. In the absence of CO2 in the surrounding atmosphere the dissociation starts at about 500°C. When one atmosphere of COo surrounds the sample, the dissociation starts at about 900 °C. The other normal carbonates are much less sensitive to change in pco2. Rowland and Lewis (1951) have shown that the order of decreasing sensitivity to change in pco2 is also Ca, Mg, Mn, Fe, and Zn. DTA curves of the anhydrous normal carbonates, with expla- nations of the reactions represented, have been published by Cuthbert and Rowland (1947), Kerr and Kulp (1948), Gruver (1950), and Beck (1950). In addition to the normal anhydrous carbonates. Beck included DTA curves of samples representative of most of the other carbonate minerals. ]'art nil .MiniioDti 01' Idextu-'yixg Clays and Interpkktation oi' Results 157 DT4 CURVES OF C4LCITE ARAGONIT E MIXTURES (AFTER FauST. flMER MtN 55,1950) FlC.l-RK 12 A review of the interpretations of necessity for: (1) determining liy other nature of the iiroduct fornieil by each whetlier each thermal loop represents compromise heat effect resulting; from vestigatinc the effect of varyiiiR the gas to establish the temperature dependence phase. The data from (3) when plotted uniquely describe the thermal character DTA curves indicates the methods, usually X-ray, the reaction; (2) estalilishin;; a single change or is a several reactions; (3) in- pressure within the sample of the reaction on the gas as van't Hoff lines almost istics of the materials. Galcite. Tlie dissociation of calcium carbonate is used in physical chemistry as a classic example of the effect of the partial pressure of a participating gas on heterogene- ous equilibria. Perhaps it is for this reason that very little attention has been given to the DTA curves of cal- cite. Fau-st (1950) and Kulp, Kent, and Kerr (1951) have shown that the peak temperature and the initial decomposition temperature of pure calcite decrease when the sample is ground to an extremely fine particle size. Kulp et al. (1951) (fig. 11), also show a drop in both temperatures when the sample is highly diluted with alundum. These results were obtained in an ambient fur- nace atmosphere without control of the CO2 and are therefore not definitive. Dilution reduces the opportunity for the buildup of a back ijrcssure of COo and conse- quently lowers the dissociation temperature. This effect is frequently observed in unwashed Ca-clay samples which have been allowed to stand in water open to the atmosphere. The DTA curves exhibit a small endothermic peak at about 750° C, resulting from the calcium car- bonate formed from calcium in the solution and CO2 dissolved from the air. DTA curves of the aragonite -^ calcite transformation have been examined by Faust (1950) (fig. 12), who has pointed out that this monotropic transformation does not take place at a constant temperature, and is subject to further variations resulting from the presence of barium, strontium, lead, and perhaps zinc. The transformation temperatures range from 387 °C to 488 °C at a heating rate of 12°C per minute. Magnesite. DTA curves of magnesite have been pub- lished by Cuthbert and Rowland (1947). Faust (1949), Gruver (1950), Beck (1950 V and Kulp, Kent, and Kerr (1951). Pure coarsely crystalline magnesite heated in air yields a simple endothermic peak at 680 to 700°C. The temperature of the peak varies somewhat in the presence of impurities. The magnesite from Stevens County, Washington, shows an exothermic peak at the end of the endothermic peak. Kulp attributes this peak to the presence of small amounts of iron substituted in the lattice. It may also be the heat effect accompany- ing the organization of magnesium oxide as periclase. Siderite. Cuthbert and Rowland (1947), Kerr and Kulp (1947), Frederickson (1948), and Rowland and ■liinas (1949) liave discussed the DTA curve of siderite. Diluted and heated in air, this carbonate yields a small exothermic loop (fig. 2). In an atmosphere of CO2 the loop is large, endothermic, and at the proper tempera- ture for the Ca, Mg. Fe, Mn, and Zn series. Undiluted and heated in air, the curve first swings in the exother- mic direction until enough CO^ has been liberated to prevent oxidation of the iron. The dissociation of CO2 is then registered by an endothermic loop which is in- terrupted by an exothermic loop representing the oxida- tion of the FeO when the back pressure of CO2 begins to subside. At a higher temperature the partially oxi- dized iron is completely oxidized to hematite. "V 9X DOLOMITE 99 5% C*LCiTE -v^ 3t DOLOMITE 99 7% C*LCITE DTA Calibration Curves of Small Percentages of Bureau of Standards Dolomite and Iceland Spar Calcite FiGiTtE i:i 158 Clays and Clay Technology [Bull. 169 effect of prolonged grtnding on ota of dolomite in co; atmosphere Figure 14 Dolomite. Of all of the carbonate minerals of the Ca-Mg-Fe group (Kulp, Kent, and Kerr, 1951) 'flolo- mite has received the most attention. Berg (1945) at- tempted to use the areas under the loops as a quantita- tive expression of the amount of dolomite in the sample. Rowland and Beck (1952) (tig. 13) succeeded in doing this for samples heated in an atmosphere of COo. Haul and Heystek (1952) (fig. 3) have shown that DTA curves for dolomite have only one loop at 1 mm 2^002, two loops, resembling the curve made in air, at 300 mm pco2, and two distinctly separated loops at one atmosphere of CO2. This is accomplished entirely by shifting of the second or CaCO.s peak. The apparent immobilitj' of the first peak leads them to suggest that this peak is formed only after a certain amount of diffusion of lattice constituents has taken place. The requirement for this activation energy explains the formation of this peak at a higher temperature than the peak for magnesite dissociation. Actually, the first dissociation peak of dolomite is not immobile. Bradley, Burst, and Graf (1952) (fig. 14) have shown that during prolonged grinding (250 hours) there first appears another peak about 100°C. lower, which grows in size until the usual first peak is ex- hausted. At any stage the ratio of the siun of the areas of these two peaks to the area of the calcite is constant. These authors demonstrate by X-ray diffraction studies that, by a process of twin gliding and translation glid- ing, the Ca and Mg of the dolomite lattice which at first occiipied alternate positions around any CO.s group have now been rearranged so that most of the Mg has magnesium for its nearest neighbors and vice versa. Hence, the temperature delay required to activate these atoms to sufficient mobility so that dissociation can occur is no longer required. The first loop of a dolomite DTA curve is the algebi'aic sum of the A// required to dis- sociate both MgCOs and CaCOa (endothermic), to re- form most of the CaCOa (exothermic), and perhaps to form periclase and some calcium oxide (exothermic). Dolomite furnishes an excellent example of the effect of small crystallites (not fine grain size) on DTA curves. In figure 13 the endothermic loop beginning at 925~C is preceded by a small shoulder. This shoulder accompanies the dissociation of the extremely fine crystallites of CaCOs formed from the products of the first loop which dissociate before the more coarse-grained calcite frag- ments. Berg (1943) and Graf (1952) have shown that the presence of soluble salts such as encountered in brines will materially affect the shape and size of the first loop of the dolomite curve. In addition, the presence of a sericite-like mica will completely eliminate the second or calcium carbonate peak in a CO2 atmosphere. MISCELLANEOUS APPLICATIONS OF DTA Soaps. Void and Void (1941) established that, in- stead of melting directly from crystal to liquid, sodium salts of long-chain fatty acids pass through a series of forms, each constituting a definite stable phase existing over a definite range of temperature. They calculated heats of transition from the DTA curves of these soaps and have since (Void, Grandine. and Void, 1948) de- lineated the polj-morphie transformations of calcium stearate and calcium stearate monohydrate by their technique. Greases. B.y the same technique Void, Hattiangdi, and Void (1949) have delineated the phase state and thermal transitions of numerous samples of aluminum, barium, calcium, lithium, sodium, and mixed base com- mercial greases, and of the corresponding oil-free soaps. CONCLUSION Differential thermal analysis is well established as a technique for the characterization and control of ma- terials which undergo characteristic changes on heating. It is less well established as a method for investigating the products obtained when such a material is heated, since equilibrium is an inherent impossibility of the method. However, the latter is not an obstacle when thermodynamic considerations control the design of the apparatus and when good recording eqiiipment is em- ployed. With the addition of dynamic atmosphere con- trol nnich useful information about the products of heat- ing can be assembled in a short time. Because differential thermal analysis is most useful when the apparatus is designed so that several different techniques can be employed, there should be no standardization of materials, heating rates, etc. Instead, a flexibility shoidd be maintained so that due considera- tion can be given to the details of the kind of change being analyzed, and these considerations must be pre- sented as a part of the data. DISCUSSION J. A. Pask: 111 tlio DTA curves of montmorillonite Rowlnnd mentioned that 111!' exotliermic loop at 0.30°C. is accompanied b.v the formation of a spinel in the untreated material, mullite and spinel in the NHi*-saturated samples, and mullite in the nieth.vlamine-saturated samples. Could this be discussed? R. A. Rowland: I believe the explanation lies in the nature of the exchangeable cation. When the exchangeable cations are Ca** and Mg**, Spinel is formed, but when these are completely absent, as in the case of the nieth.vlamine-saturated samples, mullite is formed. The forma- tion of liotli spinel and mullite in the XH, '-saturated sample would indicate that the sample was not completel.v saturateil with \II,* ; some of the original exchangeable cations must have remained on the clay. I'ai-t nil jVIethods of Identifying Clays and Interpretation of Resi'lts 159 J. A. Pask: Is Ihf spinel I'oniied by ii CDinbiiKitioii of the exchau;;t'al)lG oalicm anil llic aluminum of the lattioeV R. A. Rowland: This appears lo be so from the series of curves which I sIidwciI anil from other curves run in similar fashion. G. W. Brindley: 1 feel lliat progress can be made in the use of the various nielhoils of day identitication and estimalion by a cooperative effort whereliy type mineral specimens would be examined by the various methoils by those persons wlio have hail a ureal ileal of experience with a given method. J. A. Pask: 1 think thai any one of the nieilioils for clay identification is as flood and as useful as any other, provided the operator is thoron^hly familiar wilh the method which he uses. Isaac Barshad: Each method yields data which another method does not. That is precisely why the various methods of analysis were developed. Thus, while X-ray analysis is indispensable for crystal structure analysis, DTA is undispensable for recording changes which occur in a mineral during the course of heating. It would l)e practically impossible to identify and estimate amounts of the various clay minerals in a clay .sample derived from a soil unless various methods of analysis are used. T. F. Bates: This discussion has further indicated the need for ailililional fundamental research and for the exch;inge of clay samples be- tween workers on both sides of the Atlantic. SELECTED REFERENCES Compiled bt Frank J. Sans Agafauov, V., and Jourausky, G., 1934, The thermal analysis of the soils of Tunisia : Pedology, Acad. Sci. Paris, Comptes rendus, v. 108, pp. 1:!.jG-.58. Agatonolf, V., 103."i, Mineralogical study of soil : 'M Internat. Cong. Soil Sci. Trans., v. 3, pp. 74-7.S, Ahrens, P. L., 10.50, Differential thermal analysis; a conventional method: 4th Internat. Cong. Soil Sci. Trans., v. 4, pp. 2ti-li7. Alexander, L. T., Hendricks. S. ]',., and Nelson, R. A., 1939, Minerals present in soil colloids; II. Estimation in some repre- sentative soils : Soil Sci., v. 48, pp. 273-270. Alexander, L. T., Hendricks, S. B., and Faust, G. T., 1941, Occurrence of gibbsite in some soil-forming materials : Soil Sci. Soc, v. (!, pp. ~)2-iM. Allaway, W. H., 1948, Differential thermal analyses of clays treated with organic cations as an aid in the study of soil colloids: Soil Sci. Soc. America Proc, v. 13, pp. 183-188. Asada, Yahei, 1040, Alunite; VIII, Mechanism of thermal de- composition of alunite: Inst. Phys. Chem. Research (Tokyo) Bull., v. 10, iip. 07(i-091. Ashley, H. E., 1911, The decomposition of clays, and the utiliza- tion of smelter aiul other smoke in preparing sulfates from clays: Ind. Eng. Chemistry .lour., v. 3, pp. 01-04. I'.ailly. V. II., 10."i2, Thermal dift'ereutiril curves reflect subsurface geology: World Oil, v. l.'!4, jip. 77. Balandin, A. A., and Patrikeev, V. V., 1044, Differential thermo- couple method in contact catalysis: Acta Physiocochim. (USSR), v. 10, pp. 57(i-.501. Balandin, A. A., and Patrikeev, V. V., 1944, Differential thermo- couple in heterogeneous catalysis: Jour. Gen. Chemistry (USSR), V. 14, pp. 57-69. Barshad, I., 1948, Vermiculite ami its relation to biotite as revealed liy base-exchange reactions. X-ray, differential thermal curves, and water content : Am. Mineralogist, v. 33, pp. 6.55-678. Barshad, I., 1950, The effect of the interlayer cations on the expansion of the mica type of crystal lattice : Am. Mineralogist, V. 35, pp. 225-230. Bar-shad, I., 1952, Temperature and heat of reaction calibration of the differential thermal analysis apparatus : Am. Mineralogist, V. 37, pp. 667-605. Beck, C. W., 1046, An improved method of differential thermal analysis and its use in the study of natural carbonates: I'h.D. Thesis, Harvard Univ. Beck, C. W., 1950, An amplilier for differential thermal analysis: Am. Mineralogist, v. 35, pp. 508-524. Beck, C. W., 1950a, Differential thermal analysis curves of carbonate minerals: Am. Mineralogist, v. 35, pp. 0S5-1013. Beck, W. R., 1040, Crystallographic inversions of the aluminum orthophosphate polymorphs and their relation to those of silica: Am. Ceramic Soc. .lour., v. 32, pp. 147151. Helyankiii, I». S., and Deodot'ev, K. JI., 1040, The heating curve of kaolin in a new light: Doklady Akad. Nauk. (USSR), v. 65, pp. 3o7-;!(!(l. Berg, I,. C!., lOl.'i, Influence of salt admixtures upon dissociation of dolomite: Doklady Acad. Sci. (USSR), v. 38, pp. 24-27. Berg, L. G., Nikolaiev, V. I., and Rode, E. Y., 1911, Tlienii.i- graphia: Acad. Sci. (USSR), v. 25. Berg, L. G., 1045, On area measurenuuits in thermograms for quantitative estimations and the determination of heats of re- action: Doklady Acad. Sci. (USSR), v. 49, pp. 64S-651. Berg, U. G., and Rassonskaya, I. S., 1050, Rapid therm.-il analysis: Doklady Akad. Nauk. (USSR), v. 73, pp. 113-115. Berg, L. G., and Ras.sonskaya, I. S., 1051, Thermographic analysis under elevated pressures: Doklady Akad. X'auk. (USSR), V. 81, pp. 855-858. Berkelhamer, U. H., 1944, Differential thermal analysis of quartz: U. S. Bur. Mines Rept. Inv. 3763, 18 pp. Berkelhamer, L. H., 1945, An apparatus for differential thermal analysis: U. S. Bur. Mines, Tech. Paper 6G4, pp. 38-55. . . . 10-14, U. S. Bur. Mines Rept. Inv. 3762, 11 pp. Berkelhamer, L. H., and Speil, S., 1945, Differential thermal analysis: Mine and Quarry Eng., v. 10, pp. 221-225. Berkelhamer, L. H., and Speil, S., 1045, II. DilVerenti.al thernuil analysis: Mine and Quarry Eng., v. 10, pp. 273-270. Bradley, W. F., and Grim, R. E., 1948, Colloid properties of layer silicates: .Tour. Phys. and Colloid Chemistry, v. 52, pp. 1404- 1413. Bradley, W. F., and Grim, R. E., 1051, High temperature thermal effects of clay and related materials: Am. Mineralogist, V. .30, pp. ] 82-201. Bradley, W. F., 1952, The alternating layer sequence of rec- torite : Am. Mineralogist, v. 35 (7, 8) , pp. 590-596. Bradley, W. F., Burst, .7. F., and Graf, D. L., 1953, The crystal chemistry and differential thermal effects of dolomite : Am. Mineralogist, v. 3S. ]ip. 207-217. Bramao, L., Cady, .1. G., Hendricks, S. B.. and Swerdlow, JI.. 1052, Criteria for the characterization of kaolinile, halloysite, aiul a related mineral in clays and .soils : Soil Sci., v. 73, pp. 273-2S7. Burgess, G. K., 1908, On methods of obtaining cooling curves : Electro-chem. Metal Ind., v. 0, pp. 360-371. Burgess, G. K., 1908-00, Methods of obtaining cooling curves : U. S. Bur. Standards, Tech. News Bull. 5, pp. 199-225. Caillere, S., 1033, Study of the thermal dissociation of serpentine minerals: Acad. Sci. Paris, Comptes rendus, v. 196, pp. 62S-(>3l). Caillere, S., 19.34, Observation of the chemical composition of palvgorskites : Acad. Sci. Paris, Comptes rendus, v, 19S, pp. 1795- 1798. Caillere, S., 19.36, Study of the serpentine minerals: Soe. franc. Mingralogie Bull., v. 59, pp. 103-326. Caillere, S., and Henin, S., 1939, Differential thermal analysis of kaoliuite : Acad. Sci. Paris Comptes rendus, v. 209, pp. 684-686. Caillere, S., and Henin, S., 1944, New observations of faratsi- hite : Acad. Sci. Paris Comptes rendus, v. 219, pp. 48.5-489. Caillere, S., and Henin, S., 1044a, The origin of some anomalies presented by the thermal curves of certain montmorillonites : Acad. Sci. Paris Comptes rendus. v. 210, pp. 685-686. Caillere, S., Henin, S., and Ture, L., 194G, Investigations in differential thermal analysis of clay.s — significance and .specificity of the phenomenon of recrystallization : Acad. Sci. Paris Comptes rendus, v. 223, pp. 3S3-384. Caillere, S., and Henin. S., 1947, The application of differential thermal analysis to the study of the clay minerals found in soils : Ann. Agron.,'v. 17, pp. 23-72. Caillere, S., Guennelon, R., and Henin, S., 1049, Thermal behavior of some phyllites at 14 angstrom units: Acad. Sci. Paris Comptes rendus, v. 228, pp. 033-035. Caillere, S., Henin, S., and Esquevin, J., 1950, The hydration of certain phyllitic minerals — metahalloysite : Acad. Sci. Paris Comptes rendus, v. 230, pp. 1190-1192. Caillere, S., and Henin, S., 1951, Observations on the chlorites of iron ores : Clay Min. Bull., v. 5, pp. 134-138. ItiO Clays and Clay Technology [Bull. 169 Caillere. S., and Henin, 8., 1951a, The properties and identi- fication of sapunitp ( liowlingite) : Clay Min. Bull., v. 5, pp 138-145. Callaghan, E., 1048. Endellite deposits in Gardner mine ridge, Lawrence County, Indiana : Indiana Div. Geology Bull. 1, 47 pp. Chiang, Y., and Smothers, W. J., 1952, Differential thermal analysis in the general chemistry laboratory : Jour. Chem. Ed., y. 29, pp. 308-309. Chukhrov, F. V., 1950, Beudantite from the Kazakhstan steppe : Doklady Akad. Nauk. (USSRi, y. 72, pp. 115-117. Chukhroy, F. V., and Anosoy, F. Y., 1950, Medmontite, a copper- bearing montmorillonite mineral: Zapiski Vsesoyuz. llineral. Obshchestva, y. 79, pp. 23-27. Chukhroy, F. V., and Anosoy, F. Y., 1950a, On the nature of chrysocoUa : Mem. Soc. Russe Min., y. 79, pp. 127-136. Cohn, W. H., 1924, The problem of heat economy in the ceramic industry : Am. Ceramic Soc. Jour., y. 7, pp. 475-4SS. CoUini, B., 1950, The mineralogical composition of our (Swedish) Quaternary clays: Geol. Foren. i Stockholm Forh., y. 72, pp. 192-206. Cuthbert, F. L., 1944, Clay minerals in Lake Erie sediments : Am. Mineralogist, v. 29, pp. 37S-3SS. Cuthbert, F. L., 1946, Differential thermal analysis of New .Jersey clays : New Jer.sey Dept. Conseryatiou, Misc. Geol. Paper. 20 pp. Cuthbert, F. L., and Rowland, R. A., 1947, Differential thermal analysis of some carbonate minerals : Am. Mineralogist, y. 32, pp. 111-116. Dean, L. A., 1947, Differential thermal analysis of Hawaiian soils : Soil Sci., y. 63, pp. 95-105. Dennis, T. W., and Hunt, J. M., 1949, Application of certain instrumental methods on production research : World Oil, v. 129, pp. 152154, 158. Dubois, P., 1936, Thermal balance analyzer \yith photo-recorder : Soc. Chem. France Bull., y. 3, pp. 1178-1181. Efremoy, N. E., 1940, The problem of classification of serpen- tine minerals by the method of thermal analysis : Acad. Sci. USSR, Comptes rendus, v. 28, pp. 442-445. Ewell, R. H., Bunting, E. N., and Geller, R. F., 1936, Thermal decomposition of talc: U. S. Bur. Standards, Jour. Research, y. 15, pp. 551-556 (Research Paper 848 I . Faust, G. T., 1944, The differentiation of raagnesite from dolo- mite in concentrates and tailings : l^con. Geology, y, 39, pp. 142-151. Faust, G. T., 1948, Thermal analysis of quartz and its use in calibration in thermal analysis studies : Am. Mineralogist, y. 33, pp. 337-345. Faust, G. T., 1949, Dedolomitizntion and its relation to a pos- sible deriyation of a magnesium-rich hydrothermal solution : Am. Mineralogist, y. 34, pp. 789-823. Faust, G. T., 1949a, Differentiation of aragonite from calcite by differential thermal analysis : Science, new ser., y. 110, pp. 402-403. Faust, G. T., 1050, Thermal analysis studies on carbonates ; I. Aragonite and calcite : Am. Mineralogist, y. 35, pp. 207-224. Faust, G. T., 1951, Thermal analysis and X-ray studies of sau- conite and some zinc minerals of thesame paragenetic association : Am. Mineralogist, y. 36, pp. 795-823. Fedot'eo, K. N., 1940, Modern methods of thermal analysis. A method for registering heating curyes : Trudy Tret' ego Soye- schaniya Eksptl. mineral i Petrog. Inst. Geo. Nauk, pp. 83-94 . . . 1941, Khim. Referat. 3 hur 4, no. 2, p. 57. Fenner, C. N., 1913, Stability relations of silica minerals : Am. Jour. Sci., y. 36, pp. 331-384. Perrandis, V. A., 1949, Differential thermal analysis of some Spanish clays and kaolins: Anales edafol. v fisiol. yegetal (Ma- drid) , v. 8, pp. 33-58. Fink, W. L., Van Horn, K. R., and Pafour, H. A., 1931, Thermal decomposition of alunite : Ind. Eng. Chem., y. 23, 1248-50 . . . 1932, Ceramic Abst., y. 11 No. 4, p. 274. Pranzen, P., and Van Voorthuysen, .1. J. B., 1950, Synthesis of nickel hydrosilicates : 4th Int. Cong. Soil Sci. Trans., y. 3, pp. 34-37. Frederickson, A. J., 1948, Differential thermal curye of siderite ; Am. Mineralogist, y. 33, pp. 372-374. Frueh, A. J., .Jr., 1950, Disorder in the mineral bomite, CusFeSi : Am. Mineralogist, y. 35, pp. 185-192. Gad, G. M., 1950, Thermochemical changes in alunite and alu- nitic clays : Am. Ceramic Soc. Jour., y. 33, pp. 208-210. Gilard, P., Jr., 1950, Several particular aspects of the treatment of kaolin and metakaolin in autoclave : COBEA, no. 4. Ginzburg, A. I., 1950, Kruzhanoyskite, a new phosphate mineral : Doklady Akad. Nauk. (USSR), y. 72, pp. 763-766. Gorbonoy, N. O., and Shurygina, E. A., 1950, Thermal curves of minerals encountered in soils and rocks: Pochvovedenie (pedology) (USSR), pp. 367-373. Graf, D. L., 1952, Preliminary report on the variations in dif- ferential thermal curves of low-iron dolomites : Am. Mineralogist, V. 37, pp. 1-27. Granger, A., 1934, Thermal analysis of clay : Ceramique, v. 37, p. 58. Granquist, W. T., and Amero, R. C, 1948, Low temperature nitrogen adsorption studies on attapulgite (Floridin) : Am. Chem. Soc. Jour., v. 70, p. 3265. Griffiths, J. C, 1946, Clay research and oil development prob- lems: Jour. Inst. Pet., v. 32, no. 265, pp. 18-31. Grim, R. E., and Bradley, W. F., 1940. Investigation of effect o£ heat on clay minerals, illite and montmorillonite : Am. Ceramic Soc. Jour., V. 23, pp. 242-248; Illinois Geol. Survey. Rept. Inv. 66, 13 pp. Grim, R. E., 1942, Modern concepts of clay minerals : Jour. Geology, v. 50, pp. 225-275. Grim, R. E., and Rowland, R. A., 1942, Differential thermal anal.vsis of clay minerals and other hydrous materials : Am. Min- eralogist, v. 27, pp. 746-761, 801-18 . . . Illinois Geol. Survey, Rept. Inv. 85. Grim, R. E., and Rowland, R. A., 1944, Differential thermal analysis of clays and shales, control and prospecting method : Am. Ceramic Soc. Jour., v. 27, pp. 65-76. Grim. R. E., Machin, J. S., and Bradley, W. P., 1945, Amena- bility of various types of clay minerals to alumina extraction by the lime sinter and lime soda sinter processes : Illinois Geol. Survey Bull., v. 69, pp. 9-77. Grim, R. E., 1947, Differential thermal curves of prepared mix- tures of clay minerals : Am. Mineralogist, v. 32, pp. 493-501 . . . 1948, Illinois Survey, Rept. Inv. 134. Grim, R. E., and Bradley, W. F., 1948, Rehydration and dehy- dration of the clay minerals : Am. Mineralogist, v. 33, pp. 50-59. Grim, R. E., Dietz, R. S., and Bradley, W. P., 1949, Clay min- eral composition of some sediments from the Pacific Ocean off the California Coast and the Gulf of California : Geol. Soc. America Bull., y. 60, pp. 1785-1808. Grimshaw, R. W., and Roberts, A. L., 1944, Study of the clay quartz system — estimation of quartz by thermal methods : Gas Re- search Board, Communication 19, pp. 31-38. Grimshaw, R. W., Heaton, E.. and Roberts, A. L., 1945, Con- stitution of refractory clays ; II. Thermal analysis methods : British Ceramic Soc. Trans., v. 44 (6), pp. 76-92. Ceram. Abstracts, 1046, April, p. 66. Grimshaw, R. W., and Roberts, A. L., 1946, Study of the clay quartz system ; II. Thermal analysis methods — experiments with tridymite and cristobalite : Gas Research Board, Communication 25. pp. 58-62. Grimshaw, R. W., and Roberts, A. L., 1948, Study of the clay quartz system ; IV. Extension of the thermal analysis method to quartzite rocks: Gas Research Board, Communication 41, pp. 21-6. Grimshaw, R. W., Westerman, A., and Roberts, A. L., 1948, A symi>osium on silica inversions ; I. Thermal effects accompanying the inversion of silica : British Ceramic Soc. Trans., v. 47, pp. 269-276. Gruver, R. M., 1948, Precision method of thermal analysis : Am. Ceramic Soc. Jour., v. 31, pp. 323-328. Gruver, R. M., Henry, E. C, and Heystek, H., 1949, Suppres- sion of thermal reactions in kaolinite. Am. Mineralogist, v. 34, pp. 869-873. Gruver, R. M., 1950, Differential thermal analysis studies of ceramic materials; I. Characteristic heat effects of some carbon- ates : Am. Ceramic Soc. Jour., v. 33, pp. 96-101. Gruver, R. M., 1950a, Differential thermal analysis studies of ceramic materials; II. Transition of aragonite to calcite: Am. Ceramic Soc. Jour., v. .33, pp. 171-174. Gruver, R. M., and Henry, E. C, 1950, Differential thermal analysis, a useful tool in ceramic research : Pennsylvania State College, Mineral luds., v. 20. pp. 3-4. Gruver, R. M., 1951, Differential thermal analysis studies of ceramic materials : III. Characteristic heat effects of some sulfates : Am. Ceramic Soc. Jour., v. 34, pp. 353-357. Haffray, J., and Viloteau, J., 1948. The thermal and dilatometric analysis of chromic tixide : Acail. Sci. Paris. Comptes rendus. v. 220 pp. 1701-1702. Part ITT I MioTiioDs (IK Identifyixg Clays axd Ixteri'retation of Results 161 II:ill, .1. 1,.. 1941. Socoiidiiry expansimi of liiKli Mlumina refrac- tories: Am. ("eramic Soc. Jour., v. 24. pp. .■U!t-:!.")C. Ilarnian, ('. (!.. and Fraulini. Felix, 1940, Properties of kaolinite as a function of its particle size: Am. Ceramic Soc. .lour., v. 23 (!M. pp. 2.">2-3r>;i. narni:in. ('. (I., anil I'arnialec. ('. W.. 1JI42, Testing and classi- ti<-;ition of hallcla.vs ; Thermal lustor.v : Am. Ceramic Soc. Rull. 21 ill), pp. 2s:!-.S(i. Hattiangdi. C. S.. Void. M. .1.. and Void. H. !>.. T.H'.I. differential thermal anal.vsis of metal soaps: Ind. Kns. Chcui., v. 41, pp. 2320- 2.TJ4. Han!. Kohert A. W.. and Heystek. Hendrick. l!)r)2. Differential Iliernial analysis of the dolomite decomposilion : Am. Mineralogist. V. 37. pii. 1(1(5-180. Iledvall. .1. A.. Lindner, R., and Ilarller, N., 19.">0. A thermal- analysis study of the crystalloitraphie transition of .silver sulfate: .\ita t'hem. Scand., v. 4. pji. loil'.l llOS. Hendricks. S. B.. and Alexander. L. T.. 1039. Minerals present in soil colloids; I. Descrijition and meihods for ideiitifii-ation ; Soil Sci.. V. 48. pp. 257-271. Hendricks. S. B.. and Alexander, L. T., 1940, Semiiiuantitative estimation of montmorillonite in clays : Soil Soi. Soc. America Proc. V. 5, pp. 9.5-99. Hendricks. S. B.. Xelson. U. A., and Alexander, L. T.. 1940. Hydration mechanism of the cla.v mineral montmorillonite satu- rated with various cations: Am. Chem. Soc. Jour., v. 62, pp. 1457-1404. Hendricks. S. B.. 1941. Base exchanse of the clay mineral mont- morillonite for orfjanic cations and its ilependence upon ;ihsorption line to van iler Waal's forces : .Tour. I'livs. Chemistrv. v. 45. pj). 0.5-81. Hendricks, S. B., Goldich. S. S.. and Nelson. R. A.. 1940. A portahle differential thermal analysis unit for bauxite exploration: Econ. Geolo^-y. v. 41, pp. 04-76. Herold. P., and Planje. T. .T.. 1948, Modified differential thermal analysis apparatus: .Tonr. Am. Ceramic Soc. v. 31. pp. 20-22. Heystek. R.. 1952. Differential thermal analysis of f;an!;ne min- erals in chrome oven : .Vm. Ceramic Soc Bull., v. .31. pp. 133-138. Hill. W. L., Arminger, W. H.. and Gooch. S. D.. 19,50, Some properties of pseudonavellite from Florida : Min. Eng., v. 187, pp. (i99-702. Hollings. H.. and Colih. .T. W'.. 1915. Thermal phenomena in carhonization ; .1. Gas Lighting, 120. 917-24: Gas World, v. 60. pp. .S72-S7.S. Hollings. H., and Cobb, J. W., 191.5. Thermal study of the carbonization process: Chem. Soc. ,Tonr.. v. 1(17. pp. 1100-1115. Honeyborne. D. P... 1951. The clay minerals in the Keuper marl: Clay Min. Bull., v. 5. pp. 1.50-1.57. Houldsworth. H. S., and Cobb, .T. W.. 1922-23, Behavior of fire- clays, bauxites, etc. on heating: British Ceramic Soc. Trans., v. 22. pp. 111-137. 344.348. Howie. T. W.. and Lakin, .T. R.. 1947. A note on the application of the differential thermal method to some basic refractory ma- terials: British Ceramic Soc Trans., v. 40. pp. 14-22. ITnmmel, F. A.. 1949. Properties of some substances isostrnc- tnral with silica : Am. Ceram. Soc. .Tour., v. .32, pp. .320-326. Hnrlbut, C. S.. Jr.. 1946. Artinite from Luning, Nevada: Am. Jlinerologist. v. 31. pp. 36.5.369. Inslcy, H.. and Ewell. R. H.. 1935, Thermal behavior of the kaolin minerals : U. S. Bur. Standards. .lour. Research, v. 14, pp. 615-627 (Research Paper 792 I . Ivanova, V. P.. 1946, On the mineralogy of hydrated ferrialu- mosilicates: Acad. Sci. (USSR I, pp. 93-103. Ivanova. V. P.. and Tatarskii. V. B., 19.50. Thermograms of mixtures of dolomite and kaolin: Dokladv Akad. Nauk. (USSR), v. 73. p).. 341-343. Jaffray. .7.. 1947, The investigation of higher order tr.ansforma- tions by thermal anal.vsis: Jour. Recherches centr. Nat'l Recherche sci.. pp. 1.53-163. Jeffries. C. D., 1944, Quantitative approach to the stnd.v of ther- mal characteristics of clavs : Soil Sci. Soc. America Proc, v. 9, pp. 80-91. .Tohnson. A. L., 1949, Surface area and its effect on exchange capacity of montmorillonite : ,\m. Ceramic Soc. Jour. v. 32, pp. 210-214. .Tourdain, A., 1937, Studies of the constituents of refractory clays by means of thermal analysis : Ceramique, v. 40. pp. 13.5-141. Kagan. T. B., aud Bashkirov. A. N., 1948, Use of the differential thermocouple for kinetic measurements : Izvest. Akad. Nauk. (USSR), Otdel, Tekh. Nauk., pp. .349-3.58. 6 — 91001 Kauffman. .\. .7.. Jr.. 1948. Differential thermal analysis: World Oil. V. 12S, no. 3. pp. US. 120. 122. 124. 12(). Kauffman, A. J., Jr.. 1949, Differential thermal analysis as applied to the lime-soda sinter process : U. S. Bur. Mines Rcpt. Inv. 4585. Kauffman, A. J., Jr., and Dilling. E. D., 19.50, Differential ther- mal curves of certain hydrous and anhydrous minerals, with a description of the aiijiaratus used : Econ. Geology, v. 45, pp. 222-244 Kazakov, A. V., and Andrianov, K. S.. 1930. Methods of thermal analysis: Acad. V. I. Vernadskomn k Pyatides-syatiletiyn Nauch. Devatelnosti, 2, 833-58 . . . 19:!8, Chem. Zentralbl. I, pp. 2305- 2306. Keller. W. D.. .-11111 Westcott. J. F., 1948, Differential thermal analysis of some Missouri fireclays: Am. ("eramic Soc. ,7our., v. 31, pp. 100-105. Ivelley, W. P.. and Page. J. B., 1942, Criteria for the identifi- cation of the constituents of soil colloids: Soil Sci. Sf>c. America Proc, v. 7, pp. 17.5-181. Kerr. P. F.. and Knlp, J. L., 1947, Differential thermal analysis of siderite: Am. Mineralogist, v. 32, pp. 678-680. Kerr. P. V.. and Knlp. J. L.. 1948. Multiple differential thermal analysis: .\m. Mineralogist, v. 3.3. pp. 387-419. Ki-rr. P. F.. Knlp. J. L.. and Hamilton, P. K., 1949, Differential thermal analysis of reference clay mineral specimens: API Proj. 49. Prelim. Rept. .3, Columbia I'niv. Kerr. P. F., and Holland, H. D., 1951, Differential thermal analysis of davidite: Am. Mineralogist, v. .36. pp. .56.3-.573. Kind. V. A., Okorokov, S. D.. and Khodikel. E. P., 19.39, The use of the method of thermal analysis: Tsement 6, no. 7, jip. 32-37. Kiyinira, R.. and Sata T.. 1950, The quantitative analysis of the CaCOs — CafOHl^ — Mg(Oir): system by differential thermal analysis: Ceramic .\ssoc. .lapan .7our., v. .58. Knizek, J. O., and Fetter. H., 1946, Properties of natural alunitic clays : Am. Ceramic Soc. .lour., v. 29. pp. 308-313. knizek, J. O., and Fetter, H., 1947, Alunite and clays : British Ceramic Soc. Trans., v. 46. jip. 22-46. Kournakov, N. S., and Tcbernick. V. V., 1928, Physical-chemical investigation of several hydrated iron oxides and cla.v silicates: Soc. Russ. Mineral. Mem. I^YII. pp. 02-73. Kracek. F. C, 1929, The polymorphism of sodium sulfate; I. Thermal analysis : Jour. Phys. Chemistry, v. 33, pp. 1281-1303. Kracek, F. C, Bowen. N. L.. and Morey, G. W., 1929, The system ; potassium metasilicate-silica : .Tour. Phys. Chemistry, v. 33, pp. 1857-1879. Kracek. F. C. 1940. Phase relations in the system sulfur-silver and the transitions in silver sulfide : Am. Geophysical Union Trans., v. 27. pp. 304-374. Knlp, J. L., and Kerr, P. F., 1947, Multiple thermal analysis: Science (new ser.). v. 105, pp. 413-414. Kulp. J. I^., Wright, H. D.. and Holmes. R. J., 1949, Thermal study of rhodoi-hrosite : Am. Mineralogist, v. 34, pp. 19.5-219. Kulp. J. L.. and Kerr. P. F., 1949, Improved differential thermal apparatus : Am. Mineralogist, v. 34, pp. 839-845. Kulp, J. L., and Adler, H. H., 1950, Thermal study of jarosite : Am. .Tour. Sci., v. 248. pp. 475-487. Kulp. J. L., and Trites, A. F., 19.51. Differential thermal analy- sis of natural hydrous ferric oxides : Am. Mineralogist, v. .36. pp. 23-45. Kulp. J. L.. Kent. P.. and Kerr. P. F.. 1951. Thermal study of the Ca-Mg-Fe carbonate minerals: Am. Mineralogist, v. .36, pp. 643-671. Kumanin. K. G.. and Kalenen. N. S.. 1936. The effect of some experimental factors on the geometrical elements of heating curves: Jour. Phys. Chem. (USSR), v. 7. pp. 405-417. Kurnakov. N. S., and Uraxov, 6. G., 1924, Heating curves of Tichvinski bauxites: Ann. Inst. Phys. Chem. Anal. Leningrad, V. 2. pp. 495-496. Kurnakov. N. S.. and Uraxov. G. G.. 1924. Thermal an.alysis of clajs and bauxites : Ann. Inst. Phvs. Chem. Anal. Leningrad, v. 2, pp." 496-498. , Kurylenke, C, 19.50, Therm.al analysis of several tourmalines: Soc. Franc, mineral. Bull., v. 73, pp. 49-54. Laird, J. S.. and Geller, R. F.. 1919. Rehydration of calcined clay: Am. Ceramic Soc Jour., v. 2 (10), pp. 828-832. Lazarenko, E. K., 1950, Mineralogical characteristics of donbas- site: Doklady Akad. Nauk. (USSR), v. 72, pp. 771-774. Le Chatelier. H.. 1887, The constitution of clay : Zeitschr. physik. Chem., V. 1, p. .39(5. Le Chatelier, H., 1887, The action of heat on clays : Soc. Franc. Mineral. Bull., v. 10, pp. 204-211. 162 Clays and Clay Teciixolocy [Bull. 169 Lcout'eva, A. A.. 1944, An atti'uipt to apply thermal analysis to the study of adsorption processes: Jour. I'hys. Chem. (USSR), V. 18, pp. 469-472. Linseis, M., 1950, Relation between mineralogical structure and ceramic properties of kaolins and clay : Sprechsaal, v. S3, pp. 253- 256, 389-391, 409-410, 433-436, 456-458. Luzhnikov, L. P., and Berg, L. G., 1948, Use of differential thermography in the study of aging in aluminum alloys : Zavods- kaya Lab., y.'l4. pp. S24-82S. McConnell, Duncan, 19.50, The cryst.-il chemistry of montmoril- lonite : Am. Mineralogist, y. .'^'i, pp. 166-173. McConnell, Duncan, 1951, The crystal chemistry of montmorillo- nite; II Calculation of the structural fornuila : (lay Min. Bull., V. 6, pp. 179-189. McConnell, D., and Early, .1. W., 19.52. Apparatus for differential thermal analysis: Am. Ceramic Soc. Jour., y. 34, (6), pp. 1S3-1S7. Maegee, A. E., 1926, The heat required to tire ceramic bodies; Am. Ceramic Soc. Jour., v. 9, pp. 206-247. Mackenzie. R. C, 1949, Nature of free iron oxides in soil clays : Nature, v. 164. p. 244. Mackenzie, R. C, 19.50, Differential thermal analysis of clay minerals : 4th Int. Cong. Soil Sci. Trans., v. 2, pp. 55-59. Mackenzie, R. C, 1951, Differential thermal analysis and its application to industrial powders : Toninntrolled atmosphere ; British Ceramic Soc. Trans., y. 49. pp. 365-374. Schafer, G. M., and Russel, M. B., 1942, The thermal method as a (luantitative measure of clay mineral content : Soil Sci., v. 53, pp. 353-364. Schwob. y.. 1949, The simple and complex rhondiobedr.-il car- bonates of calcium, magnesium, and iron ; Their thermal dissocia- tion : Rev. materiaux construction trav. publ.. Edit. C, no. 411, pp. 409-420. Schwob, Y.. 1950. I. Simple and complex rhombohedral carbon- ates of calcium, magnesium, and iron ; Their thermal dissociation : Rev. m.iteiiavix construction tray, publ.. Edit. C, no. 413, pp. 33-43, 85-00. Schwob, Y., 19.50a, II. Simple and complex rhombohedral carbon- ates of calcium, magnesium, and iron; Their thermal dissociation; Rev. materiaux construction trav. publ.. Edit. C, no. 415, pp. 128-1.33. Sedletskii. I. D.. lf>49, X-ray characteristics of the mineral "mouothermite"; Doklady Akad. Napk. (USSR), v. 67, pp. 3.53- 35.5. Sedlet.skii, I. D., 1949. Mineralogy of white clays of the Rostov region; Doklady Akad. Nauk. (USSR), v. 69. pp. 60-72. Sedletskii, I. D.. and Samodurov, P. S.. 1949, Magnesium mono- tliermite; Zapiski Vsesoyuz. Mineral Obshechestva., v. 7S, pp. 274- 27(i. Shorter, A. J., 1048, Contribution to the theory of differential thermal analysis ; I. Measurement of the heat required in firing clays ; British Ceramic Soc. Trans., v. 47, p. 22. Siefert, A. C, and Henry, E. C, 1947, Effect of exchangeable cations on hydropliilie nature of kaolin and bentonite ; Am. Ceramic Soc. .Tour.. V. .'11 . pp. 37-38. I'arl ^Methods ok Ii)i:xTnviN(; Clays axd Interpretation- OF Results 163 Smcitliirs, W. J., auil Dziemianowicz, T., 1951, Refraotoi-.v jiiid tliiTinal iiliscivaliiins ou an Arkansas clay deposit: Am. Ceramic Sue. Bull., V. .'iO, pp. 7-4-75. Smothers, W. .1., Chiang, Yao, and Wil.son, A., 1051, Bihliogra- pli.v of differential thermal analysis: Univ. Arkansas Inst, of Sci- ence and Tech.. IJesearch ser., no. 21, 44 pi>. Smothers, W. J., and Chian;,', Yao, 1!)52. Differential tlieiinal curves of selected lignites: Kcon. Oeology. v. 47. pp. o84-:;!l7. Smyth, II. T., 1!)51, Temperature distrihution p. 9-10. Soveri, U., 1950, Differential thermal analysis of some Qua- ternary clays of Feuuoscandia : Ann. Acad. Scicnl. Fennicae, A. HI, V. 23, 10.3 pp. Speil, S. 1944, Application of thermal analysis to clays and aluminous minerals: U. S. Kur. Mines, Kept. Inv. .3764, 3t! pp. Speil, S.. Berkelhamer, D. II., Pask, .1. A., and Davies, V... 1945, Diffi-reiitial thermal analysis; Its application to clays and other aluminous minerals: U. S. Rur. Mines, Tech. Paper Gti4, SI pp. Splichal, .r., SUramovsky, S., and Goll, .1., 1930, Thermal decom- position of carbonate minerals: Veda prirodiui, v. 17, pp. 2(16-213. Stone, It. L., 1951. Differential thermal analysis uuder controlled thermodynamic conditions: Ohio State Univ. Fngineering Experi- ment Sta. Bull., no. 149. Stone, I{. L., 1952, Apparatus for differential thermal analysis under controlled partial iiressure of H:(). (■(>• or other gases: Am. ('eramic Soc. .lour., v. 35 (.3), pp. 76-S2. Stone, I{. L., 1952. Differential thermal analysis of kaolin group minerals under controlletl i)artial pressure of H-O : Am. Ceramic Soc. .Tour., v. .35 (4), pp. !K)-99. Tereshchenko, A. V., and Dudarskii, I. E., 1935, Regulating the plastic propi'rties of clay : Ogneuporui, v. 3, pp. 127-134. 208-21G, 29S-304. Theron, .T. .T.. Ileydrenrydi, ,T. C, and Anderson, F., 1949. A lliermocuuiile microvoltmeter for use ill the differential thermal analysis of clays: .lour. Sci. Inst., v. 26, p. 2.3.3. Thilo, E., and Schunemann. II., 1937. Chemical studies of sili- cates; IV. I'.ehavior of pyrophyllites AL( Si.C),,, ) ( Oil 1= ou heat- ing and the existence of a "water-free" pyrophyllite, Al-f SiiOio)0 : Zeitschr. anorg. allgem. Chem., v. 230, pp. 321-3-35. Trombe, P., 1938, Estimation of quartz by differential thermal analysis: Acad. Sci. Paris. Comptes rendus, v. 207, pp. 1111-1113. Vakhrushev, V. A.. 1949, On the ferrihalloysite from the Ana- tolsky silicate-nickel ore deposits in the middle Urals: Soc. Russe -Min. Mem,, v. pp. ;74. Vaseniu, F. I., 1937, Thermal analysis of silicates: Vsesoyuz. Xauch.-Issledovatel'. Inst. Tsement., Byull., 1937, No. 1, pp. 79-83. Void, M. J., 1949. Differential thermal analysis: Anal. Chem., v. 21, pp. 683-(iS8. Void, M. .1., Ilattiangdi. G. S., and Void, R. D., 1949, Phase state and thermal transitions of greases: Ind. Eng. Chem., v. 41, pp. 2539-2546. Void, M. .7., and \"oi(l. R. I)., 19.50. The phase beliavior of lithium stearate in ci'Line and in decalin : .Jour. Colloid Sci., v. 5, pp. 1-19. Void, 1{. I)., 1941, .\nhydrous sodium soaps; Heats of transition and cla.ssification of the phases: Am. Chem. Soc. Jour., v. 63, pp. 2915-2924. Void, R. D.. and Void. M. J.. i;i45. Thermal transitions of the alkali palmitates : .lour. Phys. Chemistry, v. 49, p. 32. Void, R. D., 1945, The polymorphism and transitions of anhy- drous sodium stearate: .Jour. Phys. Chemistry, v. 49, pp. 315-328. Void, R. D., 1947, I'liase boundaries in <'oneeutrated systems of sodium oleate and water: .Jour. I'livs. Cidloid Chem.. v. .51. pp. 797-815. Void, R. p., Grandine, .1. D., ami Void, M. J.. 191S, Polymorphic transformations of calcinm ste.-irate mcniohvdrate : ,Jour. Colloid Sci., V. 3, p. 3399. Wallach, R., 1913, Thermal analysis of clays: Acad. Sp. 276-277. Wilcox, R. L., and B.dlard, .1. R., 19.36, A self-reconling appa- ratus for thermal analysis: Metal and .\lloys, v. 7. pp. 221-224. Wittels, M., 1951. Thi- dift'ereiitial thermal analyzer as a micro- calorimeter : Am. Mineralogist, v. .3(5, jip. 615-622. Wittels, M., 1951, Some asjjeets of mineral calorimcliy : Am. Mineralogist, v. 36, pp. 760-767. Wittels. M.. 1951, Structural transformations in ampliiboles at elevated temperatures : Am. Mineralogist, v. 36, pp. 851-859. Wittels, M., 19.52, The structural disintegration of some amplii- boles : Am. Mineralogist, v. 37, pp. 28-37. Wohlen, R., 1913, Thermal analyses of clays, bauxites and some allied materials: Sprechsaal, v. 46, pp. 719-721, 733-735, 749-751, 767-769, 781-783. Yamauchi, T., and Suzuki. S.. 1942. Thermal analysis of .Taiia- nese raw clays: Japan Ceramic Assoc. Jour., v. 51, pp. 211-221. Zakharov, M. V., 1940, Establishing optimum conditions for heat- ing and cooling during differential thermal analysis : Zavodskaya Lab., V. 8, pp. 968-973, 19.39. . . . Kliini. Referat. Zhur., no. 3, p. 50, 1940. Zhuravlev, V. F., and Zhitomir.skaya, V. I., 1950, Binding prop- erties of crv.stal hvdrates of the sulfate type: .Tour. Applied Chem- istry (USSR), V. 23, pp. 113-117. Zhuravlev, V. F., and Zhitomirskaya, V. I., 1950, Binding properties of crystal hydrates of the sulfate type: Jour. Applied Chemistry (USSR), v. 23, pp. 230-232. PART IV CLAY TECHNOLOGY IN SOIL SCIENCE Page Role of Physical Properties of Clays in Soil Science, by J. B. Page 167 Role of Chemical Properties of Clays in Soil Science, bv T. F. Buehrer 177 ROLE OF PHYSICAL PROPERTIES OF CLAYS IN SOIL SCIENCE Bv J. B. r.vGE ■ ABSTRACT The iihysicnl propci-tirs nf r\:i\ -.wv of rxlrt-me imiiortnucc iu soil scii'iicc. I'lniil Kriiwlli. :umI liciici" criip production, within any cnvininnifnt:il coiidition is l:irf,'«'ly oontrolleil by soil structure which results from ri'aclions involving clay. The active clay material in soil, particularly in combination with small amounts of organic mat- ter, exerts a tremendous effect on soil properties. This effect may be on structure (the arrangement of soil particles), or on con- sistence (the response of the soil to mechanical manipulation). Where structure is favorable soil grains are clumped together into effectively larger aggregates so that soils have a more open arrange- ment anil water and air can move freely and roots function normally. Where structure is unfavorable, soils tend to be heavy and impervious, and both the physical and chemical properties of the .soil become unfavorable for plant growth. Soils which are low in clay, such as sands and silts, exhibit a rather narrow range over which i)hysical properties can change and may be unfavorable for plants, being droughty and lacking in fertility. Structure of soil may change through action of natural forces, management prac- tices, or cropping systems, and it is of great importance that we undiM-sland how structure affects plants, and how it is formed and stabilized by reactious involving cla.v. Combination of clay with relatively small amounts of certain organic compounds greatl.v changes the physical properties of the system and the lutture of the combination and the mechanics of the soil structure-forming process are but little understood. Such changes greatly affect soil consistence and soil-water relationships as well as soil structure. The problem is made difficult since the results of any particular ccunbination or change in clay character- istics must be interpreted not in terms of the clay system alone but in light of the resultant effect on the complex and dynamic system which constitutes a soil. Solution of such jiroblems, however, may be of great importance to our future abilit.v to produce food and liber abundantl.v and efficiently from our limited soil resource. INTRODUCTION Clay is involvod in almost eveiy roaotioii in soils wliich affeets plant o;r<)wtli. Both cheiuical and phy.sical prop- erties of soils are controlled to a very large degree by properties of clay, and an understanding of clay prop- erties is essential if we are to arrive at a full understand- ing of soil plant relationships. Soil scientists liave in the past placed most emphasis on clicmical iiroperties of soils; increasing interest is now being shown in physical properties and reactions as well. Poor physical conditions such as tight impervious laj-ers, crusts, or over-all high density and unfavorable porosity cause soils to be quite unsatisfactory for plant growth. Thi.s of course has long been known, but in re- cent years an increasing amount of interest has been shown in studying such conditions and in attempting to arrive at an understanding of how and why phy.sical conditions affect plant growth. Such characteristics as soil temperature, soil aeration, soil consistanee. and soil- Avater relationships and their effects on plant growth are now being studied intensively. In some cases plant growth has been shown to be severely limited because of physical conditions and addition of fertilizers has had little or no effect in improving yields on such soils. These conditions result almost entirely from reactions involv- ing clay. At the same time most of the favorable condi- tions occurring in soils are also associated with clay so that it is important to understand how clay enters into and controls soil properties if high and efficient * Professor of Agronomy and Soil Physics, Agricultural and aiechan- ical College of Texas, College Station. Texas. levels of pro(liictii)ii of food and fi])er are to be main- tained. Tn soil science we arc primarily concerned with deter- mining and controlling tliose factors which affect plant growth. The physical properties of soil control the sup- ply of water and air (and to a certain extent, nutrients) to' the plant roots and also modify the environment in which roots grow and function. Since roots grow in spaces between soil jiarticles or granules, and water and air are supplied to the root by movement through these pore spaces, soil porosity is an extremely important char- acteristic. "When porosity is favorable water and air can move freely and plant roots find a favorable environ- ment. Where pores are small, strong capillary forces tend to keep them filled with water so that air cannot diffuse freely. Obviously, where pores are discontinuous or sealed off at the surface free movement of water and air would be restricted. The physical characteristics of the soil are largely con- trolled by reactions in which clay plays a leading part. Clay is the active part of the soil both chemically and physically. Tn a soil where content of clay is low and sands or silts predominate most of the pores will be large and continuous so that water and air may move freely. In tliis ease porosity may be favorable, but chemical properties would be unfavorable, as nutrient suiiplyiiig capacity would be low. Where clays are more abundant and the chemical properties are more favorable, physical characteristics of the soil may be either good or poor depending upon the arrangements of the soil particles. The final environment in M'hich the plant grows is the resultant of the balance of all forces acting on and witliin the soil. In .soil physics, studies of clay must always l)e made with the ob.jective of arriving at an understanding of how the clay properties will affect the whole .soil com- plex. Many chemical properties and reactions of clays can best be studied on separated clay samples, since sand and silt are comparatively inert and act only as diluents to the more active clay. Studies of the physical proper- ties of pure clays, however, although of extreme signifi- cance in other phases of clay technology, are of only limited value to soil physics. There are two reasons for this: (1) in soil, clay is alwaj-s mixed with other mate- rials such as sands and .silts and these materials strongly affect physical characteristics; (2) physical and chemical characteristics of clay are greatly changed by absorp- tion or condjiiuition with certain types of organic com- pounds which exist in soils. Because of this we are interested in clay as it exists iu soils and controls soil properties, rather than clay as a separate material. SOIL STRUCTURE The physical characteristics of a soil which have the greatest influence on the growth of plants are those asso- ciated witli soil structure. Soil structure can change as a result of either time or management. The range of soil characteristics associated with soil structure, in fact structure itself, is controlled largely by clay. Definition. Soil .structure has been defined in many ways (Baver 48). The simplest definition is that soil (167) 168 Clays and Clay Technology' [Bull. 169 1' likii^.. ^^,"«v ' .f.--. '•*•«=»- -'JT',^^^ ••^-*- •f» u .•• , .,A ---H^^Tr^ Figure 1. A«lj;ii-eiU corn iiluts mi I'iiuUling clay, July. lU-i'.J, shovviiiK offfct of suil .structure nn orop jji'uwth. Top, corn on plot yrown in crop rotation in which soil structure improves. Bottom, corn grown continuously ; struc- ture has deteriorated ; water is standing as a result of inadequate porosity for water and air movement ; poor soil structure has definitely resulted in seriously reduced crop production. structure is the arrangement of soil particles. Such a general definition may be adequate for strictly classifi- cation purposes, but in terms of plant growth the sta- bility of the structural arrangement and the over-all vol- ume relationship of tlie strtu'ture need to be specified. Other writers have used a different definition in which structure is defined as the extent to which soil is aggre- gated (Anonjonous 1938; Nikiforoff 1941; Zakhai-ov 1927). An aggregate is a cluster of soil particles held together more or less loosely but with sufficient strength so that it behaves in the soil as a unit. The forces hold- ing particles together are stronger witliin the aggregates than the forces between the aggregates. An aggregate of agricultural significance is usually taken to be between a quarter of a millimeter and five millimeters in diameter aiul to liave at least a moderate degree of stability even wlu^n saturated. "When a soil is aggregated the clay par- ticles no longer act as individuals but cause the sand and silt grains to be held together in larger units. This causes an increase in the proportion of larger pores in the soil, thus improving movement of water and air. In some soils with proper management, stable aggre- gates of ideal size have developed and it has been found that as the soil becomes more thoroughly aggregated it becomes more productive (where other factors are favor- able). Where aggregates are quite stable and exist as Part I VI Cl-AY TECnXOLOGY IN SoiL SciKXCE Kit) (lisLTiite units of characteristic shape and size, a deter- mination of the proportion of the total soil combined into these afrprregates can be used, to a large extent, as an index of the characteristics of the pore space. In such soil a fair degree of correlation has been found between the degree of aggregation and crop yields or productivity (Alderfer and Merkle 1!I44; Bayer l!t48; Olmstead 1047; Page and Willai-d 1!I47). In many soils, liowever, the ag- gregates may differ greatly in size, shape, or stability and quite different volume-i)ore space relationships may re- sult, depending upon how tlie aggregates themselves are arranged in the soil mass. In such cases deternnnation of tlie amount of aggregation may be of limited value and give little indication of the volume relationshi]is in the whole soil, since tliese may be subject to sudden eliange. In tliese soils, determination of aggregation ' may be of limited value in describing the environment in which plants grow, since the over-all structure of the soil is necessarily destroyed in the separation and measure- ment of the aggregates (Robinson and Page 1951). Soil Porosiiii. The important characteristic of soil structure is porosity. It is important to linow: (1) the propoi'tion of soil volume existing as pores, (2) the size of these pores, and (3) their configuration and whether they are connected into cliannels or exist as pocl^ets in the soil. Porosity in the soil may be of two kinds — tex- tural and structural. Textural porosity is that porosity associated with tlie primary mineral grains in the soil and can be inferred from a mechanical analysis. "Where sands or silts pre- dominate the geometry can be calculated fairly well as- suming cubes or spheres and closest packing (Dallavalle 1948). In these eases soils usually have an excessive pro- portion of large pores, and too little w-ater-holding ca- pacity, so that they tend to be droughty and low in fer- tility (because of absence of clay). In contrast to this. when appreciable amounts of clay are present and the clays are not aggregated, the pores between the larger grains may be filled with dispersed clay so that the soil pore space is preponderantly in the very fine capillary range. Such soils are iiractically impermeable to water or air and are quite unsatisfactory for growth of plants. Under more favorable conditions the clay particles may combine with one another and the silt and sand grains, causing them to be clumped together into aggre- gates so that the porosity is quite different than in a soil where each particle acts individually. The favorable porosity (because of the increased proportion of larger pores) associated with aggregation is called structural porosity. This will of course change as structure changes. \Yhere structural porosity is favorable tlie soil will usu- ally have sufficient clay to hold water and nutrients, yet at the same time will have a higli proportion of lai-ger pores. Under this condition excess water can drain read- ily from the soil, air will diffuse freel.y into the .soil and roots can develop normally. Soils having this more favor- able arrangement of their particles are said to have good structure. Characteristics of such soils, the transforma- tions of structure with time or with different cultural treatments, and the influence of various structures on crop production all constitute a major area of study in soil physics. There are large gaps in our knowledge of how soil structure is formed, but it is known that the reactions are almost all associated with the properties of clay ; in fact, clay is essential for formation and main- tenance of favorable structure in soils. Importance of Soil Structure to Plants. That soil structure is of extreme importance to plant growth has been amply attested by much research and many pub- lished reports (Baver 1949; Page and Bodman 1951; Kussell 1950).- Figure 1 shows an example of how com- pletely crop production may be limited by soil structure. Figure 2 shows restricted root growth resulting from very poor structure in the lower layers of the soil. In this case only a small fraction of the total soil volume is available to the roots for absorption of nutrients or water. Our knowledge of these important physical fac- tors however, is still largely in the beginning stage, partly because of the complexitj' of soil-plant relation- ships and partly because there has not been adequate realization of the importance of the physical factors in plant growth and crop production. Good soil structure has a profound effect upon plant growth in at least three ways : First, because of favorable porosity associated with good soil structure, roots find a favorable environment and can penetrate and spread through the soil readily. Where structure is unfavorable roots may be restricted to growth in only a small part of the soil. Thus but little soil is contacted and water and nutrients available to the plant are limited even though the bulk of the soil con- tains water and nutrients in what would appear to be normal amounts. Second, for normal growth and function roots need a coutinued supply of oxygen. This is supplied primarily by fri'e diffusion through open soil pores and channels into the root zone. If soil pores remain saturated or are not continuous to the surface, oxj'gen will not be .sup- plied to the roots at the required rate. Roots deprived of ' The publications in this field are too numerous to list. The refer- ences given are to recent survey articles which adequately re- view the literature and present current thinking on the subject. 1 Degree of aggregation is determined by sieving after moderately severe mechanical treatments. Sieving is usually done under water to measure stability when wet, but dry sieving is some- times used. Figure 2. Tree i^iMiv ,.\|iM>r(i in i-m.-i.1 rut iii-:ii- Cnll.-r Si.i : Texas. Roots have cvteuded borizoutall.v, rather than ijeneti'are tiRht, less porous subsoil. 170 Clays and Clay Teciixolouy [Bull. 169 oxygen stop fuiu'tioninu- and are unable to assimilate water or nutrients at required rates regardless of how abundantly these essentials may be supplied (Page and Bodman 1951). Wherever tight layers or crusts occur in soils free movement of water and air into the soil may be hampered so that normal supplies are not made avail- able to the plants (Winters and Simonson 1951). Under conditions of poor aeration plants frequently show symp- toms of deficiencies of nutrients or of water. The usual practice in such cases has been to add more fertilizer or more water, either of which would increase the diificulty. The correct solution in this case would be to improve soil structure. Third, soils having good structure usually allow free drainage of excess water. This not only favors diffusion of necessary oxygen and permits free root growth but modifies heat capacity and conductivity so that soils remain more nearly at the optimum temperature. "Water has a much higher heat capacity than other constituents in soils; poorly drained soils remain wet and cold in the spring. Formation of Sfruciure. Since most of the factors which control plant growth are intimately associated with physical characteristics of the soil, soil structure is of prime importance. In many cases unfavorable soil structure limits or restricts plant growth and response to application of fertilizers. Since soil structure has such a profound effect on crop production, it is important that a sound basis for understanding the mechanics of structure formation be developed. Both genesis and stabilization of .structure are important, and in both of these clay plays the primary role. Various theoi-ies have been proposed to explain the processes by which soil structure develops. The earlier attempts at an explanation pictured the soil particles as being held together by cementation with the amorphous oxides and hydrous oxides of iron and aluminum which were then considei-ed to be the materials of clays. These were thought to surround the larger soil grains and hold them together by cementation. With the introduction of the concept of the crystallinity of clays and advances in knowledge of their colloidal properties, the way has been opened for a fuller understanding of the role of clay in soil structures. The mechanisms which have been proposed to explain formation of aggregates in the soil are : (1) Direct effect of living microbes and fungi to bind soil particles to- gether (Martin 1946; ilartin and Waksman 1940; MeCalla 1947; Olmstead 1947; Peele and Beale 1942), (2) Encapsulating or cementing action of gelatinous organic material, gums, resins, and waxes which occur quite commonly in the soil (Martin 1945; McCalla 1945; Sideri 1936) and (3) Action of clay particles them- selves to cohere and enclose or even bridge between larger grains (Peter.son 1945; McHenry and Eussell 1944; Russell 1934). There is good evidence that each of these types of binding may contribute to formation of aggregates in different soils and it is probable that they all operate either singly or in combination to different degrees in different soils. Where numbers and activity of micro-organi.sms are high the first mechanism is undoubtedly of importance, but this is usually short lived (McIIenrv and Russell 1944) and either of the other two ]U'oeesses would have to operate to give the stable, long-lived structural units commonly found in many soils. The second type of process in which gelatinous or resinous compounds enclose and bind soil particles is an attractive and rather obvious explanation. The signifi- cance of this type of process has not been established, but available evidence does not strongly support it. Such cc-mpounds are quite resistant to microbiological attack and do accumulate in the soil, but careful exam- ination does not reveal the cajisules or matrix of organic matter one would expect. Even with the electron micro- scope no evidence was found (Kroth and Page 1947) of coatings or capsules on microaggregates. The third mechanism, in which clay plays the primary role, now appears to be b,v far the most important. Un- der certain conditions cohesive forces between clay par- ticles may be very great, leading even to solidification. This last condition would obviously be unfavorable for agriculture but the same types of forces between clay particles appear to be involved in producing desirable structure in agricultiTral soils as are active in solidifica- tion of puddled soils. There are three ways in which clay particles in soil are thought to be held together: (1) by linkages of water dipoles or bridging with divalent absorbed cations (Russell 1934; Henin 1937; Peterson 1945, 1946; Sideri 1936) ; (2) by bridging or tying together with certain tvpes of polar long chain organic molecules (Kroth and Page 1947; Martin 1945; McCalla 1947; Myers 1937); (3) by cross bridging and sharing of intercrystalline ionic forces and interactions of exchangeable cations be- tween oriented clav plates (Hauser and leB'eau 1938; Henin 1937; Sideri" 1936). It is quite likely that the first of these is of importance under moist conditions and jirobably accounts for some of the resistance to dispersion observed in some soils. It is diiScult to see, however, how such a mechanism could opei-ate over the dry part of the moisture range or account for the coherence of dry soil granules. Such a mechanism may be active in causing or at least affecting orientation of ad,jacent clay particles as they are drii-d out. The second mechanism in which polar, probably long- chain, organic compounds hold cla.vs together may prove to be of great significance and certainly needs to be thoroughly investigated. There is evidence to show that many such compounds can be strongly adsorbed by clays. It appears logical that they could serve as cementing or binding agents to hold soil particles together either by liydrogen bonding or direct bridging. It is known that different compouiuls vaiy ti-emendously in the degree to which thej* are held by clays and likewise that the clays differ in the force with which polar compounds are adsorbed (Bradley 1945; Gieseking 1939; MacEwan 1946). Many such compounds are held very tightly and it has been reported that certain clay-organic complexes are resistant to redispersion or crushing after drying (Gieseking 1949). Recently synthetic long-chain poly- mers have been introduced for use in stabilizing soil structure. These have jiroduced verj- striking results on certain types of clay soils and much interest is being shown in tliese materials not only for practical use, but Part IV Clay Tkchxology ix Soil SriExcE 171 for rcsearL'li cm soil stmicture as well. Thi' nici-hanisiu by which these materials stabilize soil ajigregates has not been determined. Snoh a study should prove valuable in lieljiiiifr to reveal the nature of the combination between clay and similar active orpanic compounds and their effects on phxsical ju-opcrties of the clay. It would be of considerable ini|) most important, i-ole in soil sti-ui'ture is played by the pol\ iiro]iitlcs and polysaccharides. Tlie.se types of eom- l)oumls are not only abundant in soils exhibiting good structure but they are also strongly adsorbed by clays. Much interest is being shown in this problem and elec- troplioresis and infra-red adsorption techniques are be- ing applied iu the identification and characterization of the organic compounds active in iiroducing soil struc- ture. It seems apparent that adsor|)tion of polar organic molecules on clays is an essential for structure formation. This is based on the fact that soils low in clay show little improvement in structure upon addition of organic ma- terials wiiicli would otherwise be effective when clay is liresent. Recent work with synthetic soil additives mentioned above has shown that these highly polymerized straight- cliain compounds are extremely tightly held by clays. They do not appear to be replaced by ordinary exchange and are quite resistant to microbial attack. The extent to Figure 4. Aggregates screened (air ilr.vj from soil in plots shown in figure 1 hefore (hnttoni) and after (top) adding water, show extreme difference in stability. The amount and kind of clay are the same in the two samples; presumalily the only difference is in the adsorbed organic matter. which they are adsorbed varies roughly witii the adsorp- tive capacity of the clay and the adsorption may be either anion or cation adsorption depending upon how the mole- cule dissociates. It is significant that these materials will not create good structure but act instead to stabilize whatever structure is present when the material is ap- plied. If the soil can be prepared into favorable sized aggregates or fragments the materials do an effective job of stabilizing the aggregates so that they do not tend to run back together upon further wetting. If structure is jioor at time of application the material acts to stabilize that condition and may be of little or no benefit. Enough work has been done to show that different types of clay respond quite differently to these materials, but further work is needed to elucidate the mechanism by which the stabilizing effects are produced. It has been shown that many organic compounds when adsorbed on clays are still subject to attack by micro- organisms. As would be expected this results in loss of their stabilizing effect on soil structure and striking changes can occur in a short time if decomposition of fresh supplies of organic matter does not continue to supply new polar compounds for adsorption. Most of the changes in soil structure which are observed in the field can be explained on the basis of changes in the nature or amount of adsorbed organic compounds and the ac- tivity of micro-organisms. One special t3'pe of soil-structure change needs special mention. Saline and alkali soils are normally low in organic matter and the structural changes which occur can best be understood from the standpoint of floccula- tion. Saline soils normally have an excess of calcium or magnesium or sometimes sodium ions which act to keep the colloids effectively flocculated. Such soils usually are thoroughly desiccated from time to time so that floceules are quite stable although the cohesive forces are not ex- hibited strongly enough to cause solidification. Crumbs resulting from drying of floes are not highly oriented and the number of points of contact and total area of contact are not large. Such soils usually remain perme- able and present no physical problem, although chem- ically they may be quite unfavorable for plant growth. If, however, the excess salts are leached out and sodium predominates on the exchange positions the colloids are readily dispersed and the former rcindom orientation is lost. Depending somewhat upon the type of clay, the soil swells, becomes plastic, and usually is quite im- permeable to water. If desiccation occurs the soil solidi- fies into a hard intractable mass but readily redisperses and seals off again upon wetting. Correction for this con- dition can be had by replacing the sodium with calcium so as to again produce floeeulation. Earlier literature stressed the importance of floeeulation in soil structure on all soils, but it has been found that colloids in most all non-alkali soils tend to be flocculated. Both Ca** and H* ions produce floeeulation and further, adsorjitiou of most polar organic molecules causes complete floeeula- tion. It is now considered that most soil clays are al- ready flocculated and that changes occurring in soil structure are not primarily changes in degree of floeeu- lation, but rather in degree of expression of cohesive forces between already flocculated clay particles. It should be re-emphasized that clays are essential in structure formation and that the primary role of organic 174 Clays and Clay Technology I Bull. l&J matter is in modifying the ph.vsical properties of the clay. Sinee the mechanism involves an adsorption process, only very small amounts of the active compounds may be involved at any one time, but the effect on clay and hence on soil properties is tremendous. The amount and composition of the organic materials in the soil at any one time are dependent upon activity of micro-organisms so that physical properties of the clay organic matter system may change rather rapidly. During decomposi- tion the micro-organisms themselves exert a direct effect, usually favorable, on structure but the effects produced through adsorption of the compounds produced are thought to be much the most significant. The specific organic compounds which combine with and modify the characteristics of the clay are not yet known but their importance is tremendous and studies of the nature of these compounds and the clay-organic matter combina- tion should prove to be very fruitful. Swelling and PlasticUy. Major emphasis has been placed on the role of clays in soil structure. This ap- proach was chosen because soil structiire is of first im- portance as far as plant growth is concerned and it is through soil structure that the various materials in soil affect agriculture. Clay plays the major role in soil structure even when present in comparatively small amounts. "When clay is a major component of soils its characteristics may predominate and the soil will then exhibit many properties of the clay. Structure is still of prime importance to plant growth but such soils may also exhibit marked tendencies to swell or be plastic. These properties are exhibited in practically all soils containing appreciable amounts of clay to a degree de- ])ending upon the amounts and properties of clay present and upon the adsorbed cations or organic compounds. Swelling is an extremely important soil characteristic, but one that has received comparatively little study. Since swelling of claj^s is discussed in another paper in this symposium the mechanics of swelling will not be discussed in detail here, but it is important to indicate a few ways in which swelling is of significance in soil science. Soils which swell usually swell rather slowly, and probably only reach maximum swelling of the whole soil profile during the winter or early spring months when the soil is saturated for long periods of time. Dur- ing the growing season excess water drains away and such soils are usually characterized by cracks, which in some cases are very deep and very large. Most such soils are heavy clays, and it is probable that agriculture could not be practiced on these soils if they did uot crack, since the cracks serve to ventilate the soil and facilitate water and air movement into what would otherwise be a tight, solid mass, quite unsuitable for plant growth. It appears that most soils do not swell to any great ex- tent during the growing season of plants, although there may be some localized swelling. An example of the latter is swelling and sealing of the soil surface which some- times occurs during excessively long periods of flood irrigation, or where intense rains coupled with poor surface drainage cause water to stand on the soil for several days. In general it does not appear that swelling has a very" significant direct effect on plant growth since conditions are not favorable for swelling to occur during most of the growing period. Shrinkage and formation of cracks is, however, of considerable direct importance. Swelling and shrinking of soils influence formation or destruction of favorable structure in the soil. Swelling in some soils helps to cause destruction of aggregates since the tendenc.y is for adjacent aggregates to be forced together strongly so that they rejoin and lose their identity. On the other hand shrinkage favors for- mation of aggregates from large masses of soil initially in poor structure. Swelling of soils also is of great im- portance in engineering as well as agriculture, for it may cause destruction or damage of building founda- tions, bridges, roads, etc. It has usually been inferred that swelling soils are composed primarily of clays having an expanding lattice and that the swelling can be accounted for on the basis of the increased volume resulting from expansion of the lattice. It has in fact been suggested that swelling characteristics in soils indicate that expanding lattice type of clay predominates. It appears that this is an over-simplification. Most explanations of swelling show that swelling is a result of adsorption and orientation of water molecules on the surfaces of the individual particles. The property of an expanding lattice is usu- ally associated with elays which have a very high aiflnity for water and exhibit high surface activity, but the volume increase occurring during swelling and the quantity of water involved are usually much greater than could be accounted for by water associated with expansion of the lattice alone. It is in fact likely that mider most conditions in agricultural soils where the lattice is free to expand (not held together by "fixed" potassium or organic molecules) expansion would al- ready be essentially complete. There is no knoMu direct experimental verification of this assumption but then- is abundant evidence that montmorillonites will expand to 20A or more at relative humidities above 90 percent. Soils at a moisture content even .somewhat below the point at which plants will wilt and die are in equilib- rium Avith an atmosphere in the soil pores of more than 98.5 percent relative humidity, so that one would expect the lattices to be rather completely expanded. It is surprising that so little work has been done cm swelling in soils. Probably the most important direct effect of swelling is the eft'eet on water infiltration or entry into soil under conditions of saturation as prevail (luring irrigation. The almost universal finding in meas- urements of infiltration rates is that with many soils high initial infiltration rates soon change to low rates or the soil becomes completely impervious. Serious re- duction of rate of infiltration usually occurs (where it occurs at all) within a time which will be of ]U'actical significance during heavy irrigations or high rainfall. Work has been reported (McCalla 1947) showing that large organic molecules and many polar organic com- pounds are strongly adsorbed by highly hydrophyllic clays, and that such adsorption interferes with further adsorption of water. This, of course, markedly reduces swelling. Eecent work with synthetic soil additives in- dicates a similar action. These findings suggest the possi- bility of eft'ectively reducing the tendency of certain soils to swell by treatment with additives or active organic matter, and help explain some of the results observed on soil structure when such materials are adsorbed on the clay. Much work is needed on the me- I'art IV Clay Tkciinology ix Soil Science 175 fhanics of the swelliiifr react ion. not alcMie on pure clays and in systems haviiip: conmiercial imiiortanee. but in soils as well, and the exact role of swelliiijr in peiicsis or breakdown of soil struetnre needs to be determined. Mueh that has been said eoneerninp: swelling could also he said coneerninjr plasticity. Plasticity of soils is entirely a property of clay; soils containin condition for growth of plants. As would be expected, organic matter is extremely important in modifying plasticity of clays, apparently having two roles in soils: (1) because of its bulk and absorptive capacity, organic matter absorbs water and insulates or coats soil granules so that the clays do not become plas- tic until quite a high moisture content is reached (thus ]iermitting plowing at higher moisture contents) ; (2) by vii'tue of direct adsorption, polar organic molecules modify cohesive forces or bind clay particles together so that they are not so free to move and thus are less ])lastic. Again, these reactions have not been studied in detail and need to be thoroughly investigated. In most ca.ses the effectiveness of any favorable structural condition in the soil will depend upon the resistance of the struc- tural units to breakdown by swelling or as a result of tilling. Too often tilling is done at excessively high moisture levels and with too great intensity so that there is a strong tendency toward breakdown of structure and orientation and consolidation of clay particles. In the last analysis, the clianges which occur naturally or dur- ing tillage and which are so extremely important to crop production, ai-e changes in the arrangement of the col- loidal clay particles. The arrangement is greatly affected by interactions of surface and cohesive forces of the clay ]iarticles and hence can only be understood on the basis of knowledge of the colloidal properties and reactions of the elay. CONCLUSION It has been the purjiose of this discussion to show how clay affects physical properties of soil and how such l)roperties are important to the growth of plants. It is evident that clay is of vital significance in controlling the physical properties of soils and that intensified studies of colloidal properties and adsorption reactions of clays in soils are needed. It might appear at first glance that studies of surface properties of clays and nature of the adsorption re- actions which occur on soil clays woitM be chiefly of academic interest; but it is hoped that this discussion has shown that such studies have an immediate and im- portant i)ractical value as well. Much that is done at the prt^tseiit time in the way of tillage, or management of soil tilth is an art and as such is dictated largely by whim, fad, or custom. It seems quite apparent that if these operations, which are so vital to our economy and well being, are to be put on a soi;nd scientific basis, studies of the colloidal properties of soil clays and the effects of adsorption of organic compounds on these properties nuist be carried forward to the point where a clear picture can be obtained. Little mention has been made in this discussion of how specific types of clay minerals contribute to the physical properties of soils. This was done jiartly be- cause it was not felt that enough good information was available for a sufficiently wide range of soils to permit generalizations, but mainly because of the strong con- viction that identification of clay minerals is of much less importance for these purposes than characterization. In many cases traces of polar organic molecules when adsorbed have profoundly changed expression of physi- cal characteristics of clay in soils. There is no intent to imply that identification is of little importance; rather what is meant is that identification, without character- ization and study of the combinations in which clay is found in the soil, maj' be quite inadequate to explain soil properties. Soil-plant relationships are extremely complex, but at the same time very challenging. The benefits to be gained by increased knowledge in this important field of study justify the best efforts of colloid cliemists, clay mineralogists, plant jihysiologists, and soil scientists. The best techniques developed in these fields should be applied to solution of soil problems: it is hoped that this can be done so that soil management can be put on a truly scientific basis. DISCUSSION T. F. Buehrer: K(';;:ii(lins the use of pol.vmefliracrylates and other corrective.s for impermeable soil conditioning, we have some work going on, testing various Kriliums put out by Monsanto Cheniieal Com- pany: CRD 18G, the Ca-saturated Krilium ; CRD 189, the Na- saturated form ; and several others. I'radlcy. MacEwau, and others have shown that such ions can be adsorbed between the layers of montmorillonite. Gie-seking has done a great deal of work on the adsorption of substituted organic ammonium groups on montmoril- lonite and other clays. I understand that the polyanions are ad- sorbed on kaolinite and the polycations into montmorillonitic and beidellitic clays. I wonder if there might be some discussion on the operation of these polyanions and pol.vcations. J. E. Gieseking : We think these amino compounds are adsorbed as exchangeable c.'itions or, in many cas<'s, as very difficultly exchangeable cations. There are so many of the Krilium-like compounds, that we have not had a chance to find out very much about them. In general, these materials are hydrolyzed polyacrylonitriles, which means that they must have many carboxyl groups on the molecules — if we might be permitted to call them molecules. The amino compounds 176 Clays and Clay Technology [Bull. 169 are large catimis mid the Krilium-liUe imilei-iiles are large anions. The.v are pol.vmers. and we have indications that ahout 100 units of iior.vlonitrile iiolynierize to give the Krilium-like materials. I think that the Krilinm-like molecules are adsorl)ed li.v h.vdrogen lionding in a manner similar to the .situation descrilu'd liy Bradley. J. B. Page: The people from Monsanto Chemical Conipany have worked with adsorption of the polyacrylonitriles and find no variations in the hasal spacing of montmorillonites. It should he noted that an is.sue of Soil Science [vol. 7:'.. Xo. (!. -Tune l!>."iL'l is devoted to these materials. T. F. Buehrer: Gieseking and Ensminger showed that when such things as gelatin and other large molecules are adsorhed hy montmorillo- nite. the exchange capacity is greatly reduced (Gieseking l',)40). J. E. Gieseking: In polymerizing, the molecules must he very highly hranched, and I suspect that they would not be ahle to get into the variable spaeings in the montmorillonite. We have done a little work which indicates that they do not get in at all. However, this work on base exchange has not been applied to Krilium. Krilium-like mate- rial should have an exchange capacity of its own. It should not greatly affect the base-exchange capacity of the clay except to increase it slightly. E. B. Kinter: In practice, Krilium is not ordinarily added in large enough quantities to contribute much to affect the base-exchange capacity of clay soils or to affect all of the clay particles, internally or externally. F. N. Hveem: What i.s the relative efficiency of Krilium-like materials in pro- ducing aggregates or flocculation in comparison with other mate- rials? J. B. Page: The chief action of these materials is in stabilizing rather than jiroducing aggregates. Most experiments have shown a tremendous increase in stability with these materials as compared with lime or large quantities of organic matter. Irving Goldberg: I'reliminary studies in our laboratory have shown that the addi- tion of 0.1 percent of a Krilium-type compound to a highly expan- sive clay soil is approximately equivalent, in reducing swelling pressure and increasing stability, to the addition of 2 percent calcium hydroxide (Cat OH) 2), This improvement, however, was not particularly impressive when compared with additions of higher percentages of calcium hydroxide. No higher percentages of the Krilium-type compound were investigated. SELECTED REFERENCES Anonymous. ItlM.S, Soils and men : U. S. Dept. Agr. Yearbook of Agriculture lil.38, 12:i2 pp. Alderfer, R. 1!., and Merkle, F. G.. 1944, The comparative effects of surface application vs. incorporation of various mulching mate- rials on structure, permeability, runoff, and other soil properties : Soil Sci. Soe. America Proc. W4.3, v. S, pp. 79-86. Baver, L. D., 1948, Soil physics: 2d ed., pp. 126-135, 181-192, John Wiley and Son. Haver, L. D., 1940, Practical values from physical analyses of soils : Soil Sci., v. 6S, pp. 1-14. Bradley, W. F., 194,"i, ]\Iolecular associations between montmoril- lonite and some polyfuuctioual organic liquids : Am. Chem. Soc. Jour., v, 67, pp. 975-981. Dallavalle, J. M., 1948, Micromeritics : Chap. 6. pp. 123-148, Pitman Publishing Corp. Gieseking. .T. E., 1939, The mechanism for cation exchange in the montmorillonite-beidellite-nontronite type of clay minerals : Soil Sci., v. 47, pp. 1-13. Gieseking, J. E., 1949, The clay minerals in soils, in Norman, A. G.. Advances in agronomy : v. 1,, pp. 159-204, New Y'ork, Academic Press, Inc. Grim, R. E., Allway, W. H., and Cuthbert, F. L., 1949a, Reac- tion of different clay minerals with some organic cations : Am. Ceramic Soc. Jour., v. .30. pp. 137-142. Grim, R. E.. Allway, W. H., and Cuthbert, F. L., 1949b, Reac- tion of clays with organic cations in producing refractory insula- tion : Am. Ceramic Soc. Jour., v. 30, pp. 142-145. Hauser, E. A., and le Beau., D. S., 1938, Studies on gelation and film formations of colloidal clays. I : .Jour. I'hys. Chemistry, V. 42. pp. 9()l-9()'.l. Hauser, E. A., and le Beau, D. S., 1939, Studies in gelation and film formation. II. Studies in clay films: Jour. Phys. Chem- istry, V. 43, pp. 1()37-104S. Hendricks, S. B.. 1941. Base exchange of the clay miner;il montmorillonite for organic cations and its dependence upon ad- sorption due to van der Waals forces: Jour. Phys. Chemistry, V. 45, pp. 65-M. Henin, S., 1937, Asymetrie et orientation des miscelles argilenses: Aced. Sci. Paris Comptes rendus, v. 204. pp. 1498-1499. Kroth, E. M.. and Page, J. B.. 1947. Aggregate formation in soils with special reference to cementing substances : Soil Sci. Soc. America Proc. 1946, v. 11. pp. 27-34. MacEwan, D. M. C, 1946, The ideutiHcati(ui and estimation of montmorillonite group of minerals, with special reference to soil clays : Soc. Chem. Industry Jour., v. 65, pp. 298-304. Martin. J. P., 1045. Microorganisms and .soil aggregation. I. Origiu and nature of some of the aggregating substances : Soil Sci., v. 59, pp. 163-174. Martin, J. P., 3946, Microorganisms and soil aggregation. II. Influence of bacterial polysaccharides on soil structure : Soil Sci., V. 61, pp. 157-166. Martin, J. P., and Waksmau, S. A., 1940, Influence of micro- organisms on soil aggregation and erosion : Soil Sci., v. 50, pp. 29-47. McCalla, T. M., 1945. Influence of microorganisms and some organic substances on soil structure : Soil Sci., v. 59, pp. 287-297. McCalla, T. M., 1947, Influence of some microbial groups on stabilizing soil structure against falling water groups : Soil Sci. Soc. America Proc. 1946, v. 11, pp. 260-263. McHenry, J. R., and Russell, M. E., 1944, Elementary mechanics of aggregation of puddled materials : Soil Soc. America Proc. 1943, V. 8, pp. 71-78. Myers, H. E., 1937, Physicochemical reacticuis between organic and inorganic soil colloids as related to aggregate formation : Soil Sci., V. 44, pp. 331-3:59. Myers, H. E., and JlcCalla, T. M., 1941. Changes in" soil aggre- gation in relation to bacterial numbers, hydrogen-ion concentration, and length of time soil was kept moist ; Soil Sci., v. 51, pp. 189-200. Nikiforoff, C. C, 1941, Morphological classification of soil struc- ture : Soil Sci., v. 52, pp. 192-212. Olmstead, L. B., 1947, The effect of long-time cropping systems and tillage practices upon soil aggregation at Hays, Kansas : Soil Sci. Soc. America Proc. 1946, v. 11. pp. 89-92. Page, J. B., and Bodman. G. B., 1951. Mineral nutrition of plants: pp. 133-166, Univ. Wisconsin Press. Page, J. B.. and Willard, C. J.. 1947, Cropping systems and soil properties : Soil Sci. Soc. America Proc. 194(5, v. 11, pp. 80-88. Peele, T. C. and Beale, O. W., 1942, Effect on runoff and ero- sion of improved aggregation resulting from the stimulation of microbial activity : Soil Sci. Soc. America Proc. 1941, v. 6, pp. 176-182. Peterson. J. B., 3045, The effects of montmorillonitic and kaolinitic clays on the formation of platy structures : Soil Sci. Soc. America Proc. 1944, v. 9, pp. 37-48. . . . 1948: Soil Sci. Soc. America Proc. 1947, v. 12, pp. 29-34. Peterson, J. B., 1946, The role of clay minerals in the formation of soil structure : Soil Sci.. v. 61, pp. 247-256. Robinson, D. O., and Page, J. B., 1951, Soil aggregate stability : Soil Sci. Soc. America Proc. 1950, v. 15, pp. 25-29. Russell, E. W., 19.34, The interaction of clay with water and organic liquids as measured by specific volume changes and its relation to the phenomena of crumb formation of soils: Royal Soc. London Philos. Trans., v. 233, ser. A, pp. ;!61-389. Russell, E. W., 19.50, Soil conditions and plant growth: Long- mans, Green, and Co. Sideri, D. I.. 1936, On the formation of structure in soil. I. II, IV, and V: Soil Sci., v. 42, pp. 381-393, 461-481. . . . 1938: v. 46, pp. 129-137, 267-271. Winters, E., and Simonson, R. W., 1951, The subsoil, in Nor- man, A. G., Advances in agronomy : v. 3, pp. 2-92, New York. Academic Press, Inc. Zakharov, S. A., 1927, Achievements of Russian science in morphology of soils: Russian Pedological Investigations, Acad. Sci., U. S. S. R., V. 2, pp. 1-47 (in English). ROLE OF CHEMICAL PROPERTIES OF CLAYS IN SOIL SCIENCE l:v T. F. BUEHRER • The applu-ntioii ol' petrograpliii- aiul X-ray teeliiiiques to the study of iniiu'i'als in eoUoidal chiy of soils dates baclv over a I'einarkahly bi'ief span of years to \9'.i0, when the epoeh-makinp- papers of Hendricks and Fry (1930) and Kelley, ]3ore, and Brown (1930) appeared independently and almost simultaneously, announcing that the inorganic colloidal material of soils was not an aniorjihoiis. iiKletei'iiiinate mixture of oxides, as had hitherto been supposed, but a definitely crystalliiu> ma- terial. This iinportan! finding was, in effect, a confirma- tion of the results announced in 1!)27 by C. S. Ross (1927) to the effect that the clay minerals, notwith- standing their colloidal dimensions, exhibit definite crystallographic constants determinable with the petro- graphic microsco])e. Hendricks and Fry (1930), apply- ing both petrograjthic and X-ray technirpies, found that the colloidal clay from soils of man.v parts of the United States contained moutmorillouite, Ordovician bentonite, and lialloysite. Kelley, Dore and Brown (1930) in their extensive chemical and X-ray studies demonstrated with phenomenal clarity and rigor the close relationship between the crystalline character of the constituents of soil clays and their cation exchange properties. In nu- merous subsc(|nent jiapers by these and other investi- gators the relationship between the mineral content of the colloidal clay and the chemistry of various soil processes has been firmly established. Importance of Clay Chemistry in Soil Science and Technology. The clay fraction of the soil and in par- ticular the kind and amount of the respective clay minerals present, determines in large measure the chem- ical and physical properties of the soil. The principal physical effects manifest themselves in soil structure and aggregation and in the movement and retention of soil moistin-e. By reason of its predominant role in cation and anion exchange, the clay fraction is the ]iriniary factor which controls soil acidity and alkalinity, and as such must be takcti into account in any program of reclama- tion of alkaline and saline soils. To the extent that it influences the ease or difficulty of release of plant nu- trient ions, the clay fraction is important in plant nutri- tion. In the field of soil morphology, Jenny and Smith (1935) have elucidated the part which the clay fraction plaj's in the formation of clav'pans. There are other respects, too, in which the clay fraction influences the character of the profile. Thus the chemistry of soil is essentially the chemistry of clay. Chemical Properties of Clay. Chemically the colloidal clay is a highly polar, reactive system. Although its polar nature stems in considerable measure from its large surface per unit of weight, the seat of the polarity per se lies in the net charge on the crystal framework and on the openness or compactness of the lattice struc- ture. The charge on the framework may be of relatively large dimensions, as Marshall (1937) has found, it being of the order of 4 x 10^^ electrons per sq. cm in the * Professor of Agricultural Chemistry and Soils : Head, Department of Agricultural Chemistry and Soils. College of Agriculture, Uni- versity of Arizona, Tucson, Arizona. beidellites, and 7 x ]()'■' in the moutmorillonitcs. The charge ma.v be inherent, due to lattice substitutions, or it may be accidental due to broken bonds. Thus the 1 : 1 and 2 : 1 clay minerals have varying degrees of polarity. in kaolinite the inherent charge is negligible due to the virtual absence of lattice substitution, whei-eas its acci- dental charge may be considerable, lu the 2:1 minerals there may be both a high inherent and a high accidental charge. The clays behave as acids and the salts of acids and exhibit amphotcrism in greater or lesser degree. Bradfield (1923) has shoM'u that the clay acids behave as though they were monobasic. The clays sorb water and organic molecules through hydrogen bonding. Jenny and Reitemeicr (192fi) and Baver (1929) have shown that the clays exhibit large zeta jiotentials according to the size and charge of the adsorbed cation. 'I'liese poten- tials are the primary factor in the dispersion and floccu- lation of ela.v. Underlying all of the foregoing properties is the propertj- of cation exchange which is the dominant chemical characteristic of clay and intimately linked with nearly every important problem in soil science and technology. Distribution of the Clay Minerals in Soils. The mag- nitude of the contribution of the clay minerals of soils to cation-exchange capacity, buffer capacity, cation fixa- tion, and various physical properties depends upon the nature and amount of the various mineral species present. Attempts have been made by several investi- gators to estimate the quantitative proportions of the clay minerals in the superfine colloid from the results of chemical. X-ray, and thermal anal.vses (Alexander, Hen- dricks, and Nelson 1939 ; Buehrer, Robinson, and Deming 1949; Shaw and Humbert 1942). Alexander, Hendricks, and Nelson (1939) included in their anal.vses the determination of the iirincipal lattice constituents : silica, alumina, combined ferric oxide, magnesium oxide, and potassium oxide. From the intensities of the 7.1 A line for the second order reflection from kaolinite, the 10.0 A line for illite, and the 15 A line for the (001) reflection for air-dry montmorillonite, the.v estimated the amounts of these minerals present. Except for min- erals in which iron substitutes extensively for aluminum, isomorphous replacements do not affect the diffraction pattern appreciably. Since hydrous mica is the only clay mineral having an appreciable percentage of non- exchangeable (crystal-lattice) potassium, these authors have proposed calculation of the amount of illite using an average of 6 percent KoO as a basis. Kelley (1948), however, has pointed out that in the hydrous micas the pei'centage of crystal -lattice potassium varies appre- ciably from sample to sample, hence the calculated result can be little better than a rough approximation. AlexandcB et al. (1939) regard the percentage of free ferric oxide as determined chemically to be the best measure of the amount of iron oxide minerals present. They appear to attach greater significance to the result derived from X-ray analysis for hydrous mica, and indi- cate in their tabulated results that the values calculated from percentage of K-0 may be too low. Montmorillonite (177) 178 Clays and Clay Techxolooy I Bull. 169 is difficult to estimate ijuautitatively by X-ray ditfraetioii because of variation of the c-spacing with water content and because its second most intense reflection is at 3 A, which coincides with that for the mixed-layer minerals. MacEwan (1946), who forms a complex of montmoril- louite witli j;lycerol, thereby sharpening the X-ray lines and making it possible to identify as little as one percent of the mineral with accuracy, pi-ovides a reliable method for soil scientists working in this field. Although kaoli- nite would seem to be easily determined quantitatively since its 7.2 A spacing is not influenced by either ad- sorbed water or exchangeable cations, the determination is luicertain if the mineral is present to the extent of only 5 percent or less. Jloreover, its determination by thermal decomposition is complicated by the fact that the crystal lattice water of both kaolinite and hydrous mica is given off at temperatures which lie verv close to one another (610-620° C). Kelley and Dore (19:38) have devised an ingenious method of (juantitative mineral analysis that consists of heating the oriented sample to 500° C prior to X-ray analysis. Upon heating to constant weight at this tem- perature the variable 15 A line for montmorillonite dis- appears. Kaolinite and halloysite being stable at 500° C, do not decompose, and their lines can be measured with accuracy. Upon heating to temperatures above 500° C, the kaolinite and halloysite lines disappear but the (110) line at 4.49 A for montmorillonite does not disappear until the temperature is considerably above 600° C. The lines for the more resistant minerals such as quartz can be determined from the diffraction patterns obtained by heating the clay above 600° C. Marshall's method (1937) involves calculation of the ultimate analysis of the colloidal clay to gram-atoms of the individual elements on the basis of the number of oxygen atoms in the theoretical anhydrous formula of the mineral assumed to be present in predominant amount. Thereafter the elements are assigned to tetra- and octahedral positions consistent with the known positions in the mineral chosen. The deductions as to the dominant mineral species present in bentonite, Putnam clay, and Kothamsted claj', arrived at by this method, were in agreement with the results found by petrographic methods. Kelley (1945), however, has pointed out that the assignment of atoms to the various positions may be arbitrary, and the existence of inter- leaving and the presence of small amounts of non- exchangeable calcium, sodium, and potassium may to- tally vitiate the calculation. Shaw and Humbert (1942) and Buehrer, Kobinson, and Deming (1949) have made estimates from chemical and thermal data alone, based on determinations of non- exchangeable potassium, crystal-lattice water content of the colloid, and its base-exchange capacity. These calcu- lations involve numerical values for these three proper- ties for what are presumed to be standard mineral samples. In some instances the estimates so found have agreed very closely with those derived from X-ray de- terminations. The deviations may be a result of vari- ation in potassium content of the hydrous mica due to lattice substitutions, and to possible presence of mixed- layer minerals. None of the above-mentioned authors regard the numerical values obtained as being more than rough estimates, in the absence of more precise technique. Some remarkably suggestive deductions have grown out of such estimates, notwithstanding the rela- tively empirical nature of the method. The weaknesses of the method of Shaw and Humbert (1942) have been pointed out by Kelley (1948). who states that the cation-exchange capacity of mineral mix- tures is difficult to interpret and is not necessarily a simple summation of the exchange capacities of the com- ponent minerals. Calculation of mineral content by use of cation-exchange capacity can have significance only if the colloid happens to consist predominantly of only one group or type of clay mineral. The exchange capacit.v of any so-called standard mineral that might be chosen for these calculations is a variable depending greatly upon particle size and lattice substitutions. Page (1943) in a thermal study of the montmorillo- nites found considerable variability among 12 bentonites which on the basis of X-ray patterns were considered to be largely montmorilhniite. He fovnid that they gave widely differing thermal curves, in some cases exhibit- ing an enilothermic inflection at 850° C, a temperature considerably higher than the 800° C found generally for the removal of the last OH" water from mont- morillonite. In one instance the sample was a mixture containing considerable illite. He concludes that mont- morillonite is thus not a definite entity but one whose properties depend in large measure upon lattice sub- stitutions. He suggests that the thermal curve, never- theless, may be of value in arriving at the nature and extent of such substitutions. Coleman and Jackson (1946) in studies on coastal- plain soils estimated kaolinite, montmorillonite, and quartz from X-ray patterns, hydrous mica "from the percentage of non-exchangeable potassium, and hema- tite from fusion analvsis. Kelle.y, Dore, and Brown (1930) and Kelley, Dore, Woodford, and Brown (1939) in their extensive studies of mineral comjjosition of California soils, obviate all of these assumptions by not reporting numerical percentages, but merely indi- cating whether a given clay-mineral group is present or not, and if present, whether in considerable or small amount. The table shows that the red and yellow podzolic soil colloids are highly kaolinitie and contain considerable amounts of hydrous oxides, chiefly iron. The gray- brown podzolics contain predominant amounts of illite, and smaller amounts of kaolinite and hydrous oxides. The prairie and chernozem soil groups contain about 70 percent montmorillonite, and the remainder consists of kaolinite and illite in about eqiial proportions. In the alkaline desert soils the dominant constituents are mont- morillonite and illite, with appreciable amounts of kao- linite, as well as quartz, hydrous oxides, and carbonates in the solonchaks. The solonetzic soil colloids similarly contain all of the mineral groups, with montmorillonite and illite predominating. Coleman and Jackson (1946) report a laterite from Puerto Rico as consisting entirely of minerals of the kaolin group. Parent material does not appear to influence the mineral composition of colloidal clay appreciably. Under humid conditions, whether the parent material be gra- nitic or basaltic, the dominant mineral appears to be kaolinite. Under ai-id conditions, montmorillonite and illite are the dominant minerals, regardless of parent Pari IV Clay TKciiNor.ocY in Sou, Sciexce Table 1. Estimated clay-mineral composition of soil colloids. 179 Soil series Great group Parent rock Horizon SiOj/ RiO> Kao- linite percent Mont- morillo- nite percent Hy- drous mica percent Quartz percent Hy- drous oxides percent References Cecil (N.C.) Decatur (Ala.),- Granitic.- Limestoni- Clarial till Glacial till Granitic C Bi Bi B, C 23-36 18-27 18-27 A B B: 0-12' 0-12" 20-37" 4-14" 12-24" 0-10" 10-24" 1.88 1.95 2.69 2.77 2.60 1.18 1.44 2.85 2.73 1.93 2.73 2.80 2.05 2.48 2.04 2.63 1.95 80 80 10 30 20 00 65 10 20 75 100 Present None High None 20 4 6 None ""'20 80 None 10 80 40 15 13 7 None 2 27* 22* 7* Alexander, Red podzolif CJ. B. podzolio Chernozem Red desert Yellow podzolif Red-brown podzolic Red-brown earth ,. . Rendzina Red podzolir Laterite Solonchak Miami (Ind.) Barnes (S.D.) Mohave (Calif.) Nelson (1939) Denmark (W.A.) Granitic. Hosking(1940) Gunnedah (N.S.W.) Basalt 65 55 20 None Present None Present 34 33 57 44 26 3 None Present High 6 5 5 None High Present Houston (Miss.) Catalina (P.R.).,_ (1946) Imperial (Calif.) Redding (Calif.).- (1941) Red desert Red desert 65 46 44 52 Mixed - Tubac (Ariz.) (1949) * Summation of rerric, aluminum, and titanium oxides. material. Montmorillonite is dominant in the younger soils. altliou<>h tliere are without doubt exceptions to tills rule. Coleman and Jaekson (1946") found that the clay eolloid from eoastal-plain soils was fairly abundant in quartz illite, kaolinite, and hematite in the coarse clay fraction, but low in montmorillonite. In the fine clays, montmorillonite was the dominant mineral (50-80 percent) ; there were appreciable amounts of illite, kao- linite, and hematite, but there was less than 5 percent quartz. Soil j>n and the Xiilurc of the Clay Acids. The pH value of a soil as determined by means of the glass electrode is assumed to be a measure of, or a function of. the hydrogen- or hydroxyl-ion activity in the sense that Sorensen originally defined it. Because of the hy- drolytic equilibria in soils containing various exchange- able cations that directly affect the hydrogen-ion ac- tivity of the soil-water system, because of their effects on the soil, and because of the ease of its determinaticni with vacuum-tube pH meters, pH determination has become the popular approach to the solution of many field problems. The pH value is influenced by a variety of inherent as well as external experimental factors. Such inherent factors as the nature of the clay, the pro- portions of various exchangeable cations present, the presence of organic matter, carbonates, gypsum, and soluble salts, combine to form a very complex system. Variation in any one of these factors may result in appreciable variation in pH value. The conditions of measurement which involve the soil: water ratio, time of standing, and characteri.stics of the electrode use eon- tribute to make this determination one of the most characteristically empirical quantities measured in soil science and technology. Numerous papers have been pub- lished in recent years dealing with these variables too extensively to justify detailed reference here. The ad- vantage of the glass electrode in making possible direct measurement of pH on soil pastes and under direct field CLMiditious, and the fact that it comes to equilibrium with the soil system almost instantaneously, have put such measurements on a more reproducible basis than before. There has. however, appeared considerable polemic thinking and writing in regard to what the glass elec- trode actually measures. Coleman, Williams, Nielsen, and Jenny (1951) have pointed out that the process involved in the cell reaction involves an unknown, vari- able, and probably indeterminable .iunction potential, as a result of which the actual cell process does not in- volve a simple diffusion through the glass membrane, but a junction potential near the surface of the calomel electrode which contains saturated KCl. The pH value is a potential measurement, hence seriously affected by the unknown junction potential when the latter becomes large. Colman et al. (1951) conclude that "Em])irical correlations obtained in the past between soil pH and other variables are still valid," but a considerable error will be introduced when Ej becomes large. Empiricism is, of course, unavoidable in such a complex system, hence the discrepancies in measurement reported in the literature. The most serious problem in this connection is that soils investigators have not yet agreed upon a standard and uniform condition of measurement, par- ticularly with reference to moisture content. The function of the colloidal clay in these systems is to influence the H* or OH" ion activity by dissociation, hydrolysis, cation exchange, or buffer effects. Bradfield (1925) defines the point of maximum buffering as that at which the soil is half-saturated with cations. The underlying cause of these effects is, of course, the strength of the clay acid. Bradfield (1923) shewed that the clays 'give buffered pH titration curves similar to those of weak acids, and that in a 1 percent colloidal suspension, the normality of an electrodial.yzed clay acid lies between 0.002 and 0.004-N. For concentrations of clay between 1 percent and 12 percent, he found the value of Ka to be of the order of 3 x 10"', assuming it to be monobasic. The clay acid therefore exhibits a strength of the order of magnitude of carbonic acid. 180 Clays axd Clay Technology [Bull. 169 Jenny (1932) termed the apparent dissociation con- stant of the clay acid which Bradfield had calculated from the pH value given by the clay at the point of half -neutralization, the ' ' avidity index, ' ' on the grounds that the constant so found was not a true constant. Bradfield (1923) and Peech and Bradfield (1948) have shown that the dissociation constant so calculated for Ca- and Na-salts of a clay acid are correct if the activity of these salts is included in the equation as follows: pM = pK -\- log - ttM-clay (iH-clay log Ui, where Oj, is the activity coefficient of the cation of the added base adsorbed on the clay. Marshall (1949) points out that the clay acids are not pure H-sj'stems but mixtures in which at low pH values aluminum plays an important part. These acids are only partially dissociated and their degree of dis- sociation varies with the cation present and the degree of saturation with cations. Under alkaline conditions the clay minerals decompose to a greater or lesser extent, yielding aluminates and silicates. He holds that the clay acids are not similar to the simple weak acids, nor to the simple non-mineral colloidal electrolytes. In other words, they cannot be treated as compounds with one or more sharply defined dissociation constants. Their con- ductivities are generally less than the sum of the con- ductivities of the charged clay particles and the cations which constitute the "outer atmosphere." The shape of the pH titration curve for clay acids varies with the nature of the clay, the concentration of the clay suspension, and the base used for titration. In most instances a single inflection point is observed, cor- responding to the equivalence point, as in similar titra- tions of the simple monobasic acids. If the suspended clay at various concentrations is so titrated, the curves obtained intersect in a common point. This has been found by Marshall (1949) to be true for beidellite, illite. and kaolinite, but not for montmorillonite. In the last- mentioned case, the higher the concentration of the sus- pension, the greater the base required per unit qviantity of clay for neutralization, and the higher the pH value at the point of inflection. Both conductance and poten- tiometric studies have been made on H-saturated clays. They indicate a pH value of 2.85 for H-montmorillonite, 3.58 for H-beidellite, 3.69 for H-illite, and 4.60 for H-kaolinite under comparable conditions. Mehlich (1942-1943) in extensive studies on the rela- tion between base saturation of soils and their pH vahies investigated this relationship also for pure clay minerals and their mixtures. His results based on pH titration cui'ves of the H-saturated minerals are in agreement with the findings of Bradfield, Jenny, and Marshall previously cited, concerning the dissociation of these clay acids. Kaolinite over the entire range of saturation percent- ages yields higher pll values ; bentonite and bentonite- halloysite mixtures yield y)H values in the lower range and exhibit greater buffering power. Illite and montmo- rillonite yield nearly identical titration curves but they lie within an appreciably lower pH range. A very close correlation was obtained by Mehlich between the ti- tration curves for pure minerals and mineral mixtures and similar curves for soil clay colloids whose mineral composition had been determined. Cationic Activity of Clays and Soils. The dissociation of clay salts is of great importance in plant nutrition since it determines the rate at which cations on the col- loidal clay can be released for absorption by the plant. The term "available" plant nutrient is usually defined in terms of that portion which is water-soluble and that which is exchangeable. The degree of dissociation is a function of the character of the ionizing surface, hence related directly to the nature of the minerals which make i;p the colloidal system. The bonding energies with which cations are held in equilibrium with the ionizing surface are expressed quantitatively in terms of cation activity as measured by Marshall (1951) by means of clay-mem- brane electrodes. They are no doubt also involved in the oscillation volumes of exchangeable cations taking part in contact exchange in terms of the theory of Jenny (1936). By use of clay membranes prepared from differ- ent clay minerals and a cell arrangement similar to that of the glass electrode, in which the clay membrane per- forms essentially the same function as the glass mem- brane, Marshall and his eo-workers have determined what they consider to be the activity coefficient of various clay minerals for various uni- and di-valent cations. Their work is too extensive for detailed discussion at this point. Reference to the most significant findings will, however, be made. Marshall and Bergman (1942) have studied the dissociation of montmorillonite, beidellite, illite, and kaolinite saturated to A^arious degrees with potassium and ammonium ions. For beidellite in 2-4 per- cent suspension, thev report an activitv coef@cient for Ca** ion to be 0.01-0104, for K* ion, 0.18," for NH4-, 0.08, Na* ion, 0.09, and H* ion 0.04, from which the relative ease of dissociation may be deduced. ^Marshall (1948), com- paring the dissociation of Na-, Iv-, and NH4-salts of kaolinite and montmorillonite, found that the kaolinite salts, in contradistinction to those of montmorillonite, are quite highly dissociated, to an extent of 20-30 percent. The order of the minerals as to activity coefficient of the cation is : kaolinite> montmorillonite> beidellite> illite with respect to the univalent cations other than hydro- gen. For the same clays saturated Avith hydrogen ion, the order is : montmorillonite > beidellite> illite> kaolinite. The feeble dissociation of the kaolinite acid and its strong dissociation as a salt is interpreted in terms of the lattice structure. Kaolinite has a compact structure with no unbalanced inherent charge on the lattice. Dissociation of OH" groups is feeble and hence the hydrogen-ion activity is low. In the form of its salt, however, the cations are on the outer edges of the crystal and readily accessible. In montmorillonite, bei- dellite, and illite, dissociation is enhanced by lattice substitutions. As a i-esult, their acids are much stronger than that of kaolinite, despite the fact that secondary reactions liberating aluminum occur. The salts, however, are feebly dissociated because of the strong attractive forces holding the cations to the lattice, and of penetra- tion of certain cations into the silica layers. Part IV] Clay Technology ix Soil Science ,181 The nature of the cation-exchange reaction, principles and quantitative equations relating to the equilibrium involved, and the numerous respects in which this im- portant reaction enters into soil science and technology liave been set forth in comprehensive form by Kelley (]!148') and for reasons of space limitation will not be treated here. Fixation and Release of Potassium in Soils. One of the most clialleiiging pro])lems in soil fertility that en- gaged the attention of soils investigators as far back as 1848 is the fixation and availability of potassium. Its importance is indicated by the large number of papers, some 200 or more, listed in the scholarly review of the subject by Reitemeier (1951) in Advances in Agronomy. A full hundred or more papers on both field and labora- tory studies on this problem were not included in his review. Pota.ssium occurs in soils in the form of primary minerals (to the extent of 90-98 percent as the feld- spars and micas), in the claj-s and other secondary minerals, and in water-soluble form. In the clays it exists in both exehangeablo and non-exchangeable form, the latter being generally greatly in the predominance. A sample distribution of potassium in its various forms in calcareous and uon-caleareous soils, as found by McGeorge (1933), is shown in table 2. Table 2. Diitrihution of potassium in calcarrous desert soils.* (p.p.m. K on dry soil basis) Soil Total K (fu-sion) Water- soluble COi- soluble Ex- change- able Non-ex- change- able Neubauer value Litchfield Kyrene I.aveen 23.100 30,100 26,400 25,500 21,800 24,400 19,200 83 134 75 60 69 48 60 200 440 70 130 310 110 130 670 1.730 200 910 1,310 320 910 22,350 28,240 26,060 24.530 21,600 24,030 18,230 188 308 154 300 Tempe 240 Mesa - 340 • McGeorge, W, T., 1933. The data in table 2 illustrate the fact that by far the greatest proportion of the ]iotassium is nou-ex- changeable, as measured by replacement with ammo- nium acetate, this form ranging between 90 and 99 percent of the total potassium present. The COs-soluble fraction represents the amount of potassium replaced at a pH value of about 6 in a 1 : 5 water suspension of the soil and, as would be expected, is considerably greater than the amount soluble in water alone. The Neubauer values in most instances exceeded even the amount of potassium brought into solution by saturated carbonic- acid solution, but were in some cases considerably less than the total exchangeable potassium. On this basis JIcGeorge concluded that the potassium supph' of desert soils is of such a magnitude that this element will not need to be applied in fertilizers for some years to come. Verj- few mixed fertilizers sold in the State of Arizona carry potassium, and in few instances have crop re- sponses to its application been reported. On the other hand, the fixation of potassium in an unavailable form is a problem that has assumed great importance in recent years. Volk (1933) was apparently the first to demonstrate what happens when potassium becomes fixed in non-exchangeable form in soils. He showed that when the 0.3 micron fraction of Ilagers- town silt loam w'as treated with KCT solution and evaporated to dryness, the potassium became fixed, whereas little potassium was fixed while the soil was kept moist. Tliirty-two wetting-and-drying cycles at 70" C resulted in fixation of To percent of the i)otassium added. X-ray examination yielded a diffraction pattern which resembled that of muscovite so closely that he announced fixation to be due to the synthesis of musco- vite. Subserjuent research has indicated that the mineral identified was not native muscovite but hydrous mica, or illite. A surprisingly large proi)ortioii of arable soils fix potassium in a form unavailable as measured by ordinary eation-i-eplacement methods. It is fixed primaril.y by the 2 : 1 group of clay minerals with expanding or partially expanding lattice and not by the kaolin group. By rea- son of its stable 12-coordination as in the micas, as well as in a 14-coordination. potassium has now been defi- nitely shown to be fixed in the variable spacing of the expanding-lattiee minerals. A series of fundamental researches subsequent to Volk's announcement have served to elucidate the mechanism of the fixation of potassium. Truog and Jones (1938) found that the fixation of potassium re- duced the cation exchange capacity as was to be expected if the binding together of layer packages reduced the number of exchange spots on the lattice of the mineral. A brief summary of their data, obtained after removal of oi-ganic matter from the soil sample, saturating with K* and subjecting it to 20 wetting-and-di-ying cycles at 80°C, is shown in table 3. It is evident that the nuni- Table 3. Reduction in cation-excJiange capacity of soils and clay minerals accompanying potassium fixation.* Soil type or mineral Horizon of treatment K fixed me./lOO g Reduction in C.E.C. me./lOOg Miami silt loam . A B .A, B A B 1.5 4.7 2.7 3.4 4.0 5.0 1 9 4.8 2.7 Richfield clay _ 3.2 3.8 5.1 Bentonite powd 20 cycles @ 80^*0 14.8 49.4 27.0 31.5 14 5 Bentonite <0. In 20 cycles ©SCO 51.7 Bentonite, powd. 75 hrs. @ 110°C 28.0 Nontronite <0.1iJi.lOhrs. @ 145°C 34.0 ♦ Truog. E., and Jones, R. J., 193S. ber of milliequivalents of K* fixed is equal to the number of milliequivalents by which the cation-exchange ca- pacity was reduced ; a fact particularly striking in the case of bentonite under a variety of conditions. Martin, Overstreet, and Hoagland (1946), working with Vina and Ramona colloidal clay, found that the reduction in exchange capacity was considerably less than the amount of K* fixed in the coarse clay fraction, but equivalent to it in the clay size fraction. When the whole soil was considered, the amount of K* fixed was found to be 182 Clays axd Clay Technology [Bull. 169 equal to the siun of the exchangeable bases liberated duriiip: fixation. Page and Baver (1940), in a study of fixation of K+ by bentonite and Miami clay on drying at 105°C, found that NH4*, Rb*, Cs*, and Ba*"^ can undergo similar fixa- tion. The results of their studies are summarized in Tahle Jf. FUiition of cations bii Miami rnlloid as a function of ionic si::e.* Ionic diameter A Percent r epiaceable Difference Percent of exchange- Cation Moist Dry able ion fixed by drying Li+ 1.36 1.96 2.42 2.66 2.96 2.34 1.42 1.96 2.30 2.62 89.7 89.4 71.7 77.0 54.4 43.8 65.0 82.0 59.5 72.4 74.6 72.4 34.6 35.1 26.7 24.1 46.1 68.7 43.2 38.3 13.1 17.0 37.1 41.9 18.7 19.7 18.9 13.3 16.3 34.1 16.8 Na+ 19.0 NH.+ K+ 51.7 54.5 Rb+ 51.0 Cs+ - 45.2 Mg++ ^ 29.1 Ca++ 16.2 Sr+* -.. 27.4 Ba++.- - 47.2 • Page, J. B., and Bavcr. L. D., 1939. table 4. They found that ions which were of such a diameter that they could fit into the hexagonal 0-eavity whose "diameter" was 2.80 A. could become fixed. Thus the cations whose ionic diameters were of the order of 2.80 A can fit snugly into such positions and be firmly held, whereas much smaller cations such as Li+, Na*. Mg**, and Ca** are loosely held and therefore remain in large measure replaceable. Wear and White conclude (1951) from studies on Wyoming bentonite that the charge resulting from substitutions in the tetrahedral layer of montmorillonite is responsible for the fixation. Beidellitic and illitic clays have more tetrahedral sub- stitution than does montmorillonite, hence show a greater fixation capacity. Wear and White hold that during fixation some of the interlayers remain contracted upon re-wetting, while the remainder expand. Stanford (1948'i in a study of calcareous soils foimd that potassium may become fixed immediately upon its addition to the soil, even without drying. Working with electrodialyzed illite and acid-washed bentonite, and a constant level of added potassium, he varied the pll value of a series of samples with NaOH, thereafter de- termining the amount of exchangeable potassium on the moist and dry soil. He found that illite fixes potassium in increasing amounts as the pH increases from 3 to 10, whether the soil was moist or dry. Bentonite fixed no potassium in the moist state, but on drying, the amount of potassium fixed was 5 me. per 100 g at pH 3, the amount decreasing to a limiting value of 2 me. per 100 g at neutrality, and thereafter remaining constant. Thus illite fixes potassium in calcareous soils even under moist conditions, and its capacity to fix potassiiim is in- creased by di-ying. Acid illite fixes relatively little potassium because hydrogen, iron, and aluminum ions, present in the inter-layer, appear to inhibit fixation. Removal of these ions by addition of a base, a phosphate or fluoride, increased the capacity of illite to fix potas- sium. Fluoride increases fixation by montmorillonite, through removal of aluminum which blocks the exchange positions at the interplanar surfaces. In acid soils, hydro- gen, iron, and aluminum must be replaced before potas- sium can be fixed. In calcareous soils, Ca**, Mg**, and Na* are replaced readily by potassium, hence the latter is fixed readily even under moist conditions. Stanford thus holds that potassium fixation by illitic clay is essen- tially an exchange reaction. The question as to whether the so-called fixed potas- sium can again become exchangeable and available to plants has likewise engaged the attention of investiga- tors. McGeorge ^ cites the case of Superstition sand near Yuma, Arizona, which contains potassium which cannot be released in the ordinary cation-exchange pro- cedures but which is extensivley available to plants as indicated by Neubauer determinations. This soil has an exchange capacity of 4 me. per 100 g. Only 0.7 me. po- tassium per 100 g is found to be exchangeable. For 12 years this soil has been cropped to alfalfa, yielding 6 tons of alfalfa per acre per year, and the crop removing 225 lbs of potassium per acre per year. To date, 2,700 lbs of potassium have been thus removed, yet the soil originally contained only 2,320 lbs of exchangeable po- tassium per acre to a 2-foot depth. This soil has never shown a response to potash fertilizer in terms of in- creased yield of alfalfa. McGeorge concludes that the non-exchangeable potassium must gradually be becoming excliangeable. Fine, Truog. and Bailey (1941) have found that al- ternate freezing and thawing will release fixed potas- sium. Hoagland and Martin (1933), Drake and Scarseth (1940), Reitemeier, Brown, and Holmes (1951), and others have found by Neubauer determinations that ])lants can extract more potassium from soils contain- ing fixed potassium than can be extracted bj' replace- ment with ammonium acetate, dilute nitric acid, and other solutions. Heating to 200°C, alternate wetting and drying and eleetrodialysis release more potassium than can be replaced by cation exchange. Williams and Jenny (1952), in experiments with plants grown on Ramona loam, a non-calcic brown soil with high potassium-fixing capacity, find large amounts of non-exchangeable potassium to be absorbed, indicating that such a conversion takes place. In leaching experi- ments at various pH values of the leaching solution, it was found that the potassium replaced between pH 3 and 7 is that which is present in normally replaceable form in the soil. Below pH 3. however, more potassium appears in the leachate. which they conclude must be part of that potassium which is non-exchangeable. Hence the pH value of the leaching solution appears to be one of the controlling factors in this release. In leaching experiments with the chlorides of H*, Na*, Li*, Mg**, Ca""", and NH4*, Williams and Jenny found that all of the cations except ammonium are able to release considerable amounts of non-exchangeable potassium. Ammonium Ion Fixaiion. The first report on the pos- sibility of ammonium-ion fixation by soils was in 1917 by JIcBeth, who found that not all of the ammonium ion with which a soil had been treated could be removed either by distillation with alkali or extraction with 10 percent HCl. Bower (1951) observed that the subsoil of an alkali soil showed an exchangeable-sodium content ' Private communication. I'ait IV I Clay Technology ix Soil Sciexce 183 40 percent great(M- than the exchange capacity of the soil as determined liy replacement with neutral aiiimo- nium acetate. This anomaly pointed to the possibility of error in the determination of exchange capacity nsing ammonium ion as the replacing cation, and to fixation of the ammonium ion by the clay. lie found that a faii'ly high percentage fixation occtirs iindcr moist-.soil condi- tions. Analogous to observations of other investigators working on the fixation of potassium. P>owcr fomid that the amounts of K' and XII4* ions fixed by the same soils were practically identical in magnitude, suggesting that this could occur oidy if the mechanism of the fixa- tion were substantially the same. He observed that a i-ednction in cation-(>xchange capacity a('coiii])anies the fixation of annnoniuiii ion and concluded from this that fixation is itself an ion-exchange reaction or involves cation exchange somewhere in the ])roccss. As to the mineral responsible for the fixation. Bower holds that the amount of illite present in the colloidal clay would not account for its magnitude. It appears likely that another clay mineral is resjionsible for it. Stanford and Pierre (]il47) find that the calcareous soils of Iowa fix e()nsirable amounts of ammoniiun ion under moist conditious. They showed that if the soil is saturated in part with K* ion prior to addition of Nri4* ions, the amount of NH4* ion fixed was reduced; like- wise the amount of potassium fixed by the soil is reduced by prior fixation of ammonium. The mechanism of the fixation of NII4* ions has been studied by Allison. Doetch. and Roller (19ol) who find that the fixation which occi;rs under moist conditions involves intei'action with illite whereas the additional fixation brought about by drying at 100"C (six times as much being fixed on drying as under moist condi- tions) must be attributed to montmorillonite and illite. Barshad (1948) has shown that vermiculite. heated repeatedly to 70° C, in presence of normal ammonium salt solutions, fixes large amounts of Nn4* ion which is not replaceable by K", Kb*, or Cs*, but readily re- placed by Li*. Xa% Mg**, Ca**, and Ba**. It is apparent that the ions which are smaller than NII4* are able to replace it from the position where it is fixed in the lat- tice, whereas the large cations cannot. Barshad there- fore concludes that fixation involves the orientation of the XIL* ion in the liexagonal openings between the layer packages of montmorillonite analogous to the man- ner in which fixed potassium is held. Barshad (1951) further found that in replacing the adsorbed NII4* ions from a soil, distillation with a base is more effective than leaching with a salt solution. The difference between the amounts of NH3 obtained by distillation with NaOH and K(_)II respectively, represents the fixing capacity of a soil for NH4* ion. The results of these studies are rather far-reaching, since they cast doubt upon validity of conclusions based upon many hundreds of cation exchange capacity deter- minations that have been made in the past, in which neutral normal ammonium acetate replacement, followed by distillation of the adsorbed ammonium, has been accepted as a standard method. Since such a distillation results in the liberation of both exchangeable and non- exchangeable ammonium, the exchange-capacity values so obtained must in many instances have been nnich too high. Bower. Reitemeicr. and Fireman (1952) have sought to obviate the diffictdty by using sodium acetate as the rei)laciug reagent, and determining the number of me. per 100 g of Xa* ion adsorbed by the colloid. However, Larson and AUaAA'ay (1950) find that Na* ion can likewise become fixed in non-replaceable form under cei-tain conditions. Phosphaie Fixation. The subject of phosphate fixation in soils has been comprehensively reviewed by Dean (1949). In view of the extensive litei-ature on the sub- ject, the present discussion will be limited to the mech- anism of i)hospliate fixation oidv. ^Vlurphv (1939), Stout (1940). Coleman (1945). Dean 'and Rubins (1947), Low and Black (1948), McAuliffe et al. (194S), and others have shown that phosphate can enter into anion exchange-reactions with clay. The mechanism of the fixa- tion has, however, been a subject of some controversy in recent years. Dean and Rubins (1947) have shown tliat anion ex- change in soils increases with clay content and specific surface, and involves the lattice cations that are located at the broken edges of the mineral crystals. Clays have a definite anion-exchange capacity, as shown by the fact that they can alternately exchange arsenate and phos- phate groups. Phosphate enters into such reactions and is readily exchanged by F . OH", tartrate, borate, and silicate ions. Exchange from kaolinite is rapid with dilute solutions of phosphate, and there is evidence to show that the exchange occurs between the phosphate ions and the free -Al-OH groups at the crystal edges. In the case of the kaolinitic clays, and anion- and cation- exchange capacities are approximately equal to one another. This is as might be expected, since the -Si-OH groups at the edges are responsible for cation exchange, and the -Al-OII groups for anion exchange. In mont- morillonite and illite clavs. anion adsorption is only a small fraction of the cation-exchange capacity, since the latter is the result of ionization which occixrs at the planar surfaces. In the kaolin clays there is no charge resulting from lattice substitutions, hence the adsorption of both cations and anions is attributed to the accidental charges due to bi'oken bonds. Stout (1940), on the basis of chemical and X-ray evi- dence with phosphated kaolinites, found that the OH" gronps made accessible by grinding must have reacted with the phosphate ions, as evidenced by the loss of watei- at 150°C. Kaolinite lost 13.5 percent water, hal- loysite 21.2 percent, and bentonite only 0.2 percent, indicating that the interaction must have occurred be- tween the phosphate and OH" groups of the — Al — OH groups of the gibbsite layer of kaolinite. There being no corresponding OH^ groups available in montmorillonite, there was no interaction. Further, the equilibrium shifted to a more alkaline pH value as would be ex- pected if OH" groups were liberated in the interchange. The X-ray pattern of the fraction greater than 1 mici-on gave the characteristic pattern of kaolinite. However, when the <1 micron fraction was examined, the kao- linite pattern was destroyed, and the phosphated kaolinite was amorphous to X-rays. When the phosphated kao- linite was dephosphated, the X-ray pattern was the same as that of the original material. Stout therefore concluded that the amorphous character of the pattern for the phosphated kaolinite indicated that there had 184 Clays and Clay Technology [Bull. 169 been a disruption of the kaolinite layer packages result- ing from the siibstitution of larger phosphate ions for OH ions within each cleavage plane. Restoration of the pattern on dephosphating indicated that the original alumino-silieate units of the kaolinite lattice had not been destroyed. Thus the X-ray evidence obtained by Stout pointed to the fact that the fixation process was reversible. This fact was further confirmed by a calculation, from the altitude of the PO4 tetrahedron and the diameter of OH" ion, of the plane of repetition in kaolinite, and the value found was 9.4 A. Black (1942) and later Coleman (1945) in studies of phosphate fixation have reported evidence to show that the sorption of phosphate might be due to the presence of aluminum hydroxide on the surfaces of the kaolinite crystals. Both of these investigators removed the hy- droxides and found that not only did the purified kao- linite adsorb negligibly small amounts of phosphate, but that the adsorption was not greatly dependent upon the pll value. Other clay minerals, montmorillonite, halloysite, and illite gave the same results, although on long standing, kaolinite gave a much higher adsorption at pH 3 than at pH 7. These investigators therefore con- cluded that Stout's results were less a property of the kaolinite, per se, than of the material that had resulted from the fine grinding. Stout (1940) had found that the solution at equilib- rium was less acid after phosphation. Coleman (194-5) observed, however, that after removal of the ferric and aluminum hydroxides, there was little or no increase in pH value, and in most instances the pH value actually decreased. Hence he concluded tliat there had been little or no anion exchange between iihosphate ion in solution and the OH groups on the clay minerals. Coleman's curves of phosphate fixed b^- montmorillonitic and kao- linitic clay, respectively, as related to the final pH value of the phosphate-clay mixtiire, show that these clays are able to fix only a small portion of the phosphate which was fixed before the free iron and aluminum hydroxides were removed. The phosphated clays were then ex- tracted with Tri;og's reagent and the amounts of alumi- num and iron hydroxides so removed were determined. The results showed that the phosphate was removed simultaneously with the hydroxides, and further, that the amounts of each were equivalent to one another, indicating that the fixation was the result of a simple interaction between the phosphate and the hydroxides of iron and aluminum. Moreover, the clay minerals themselves did not break down either during or after removal of the hydroxides, indicating that the hydrox- ides were not an integral part of the mineral lattice. Black (1942) assuming that during contact with the phosphate the clay might decompose, liberating soluble aluminum that might fix phosphate, treated the clay with acetic acid at the same pH as that of the phosphate solution used. The aluminum rendered soluble was far too small to aceoinit for the phosphate fixed. Hence the fixation must be attributed to the hydroxides natively present in the soil. Low and Black (1948) obtained evidence to show that the phosphate does, in fact, react with kaolinite, but the product formed is not a phosphated kaolinite; instead. it is alumininn phosphate. If this were true, it would necessitate the liberation of silicic acid during the re- action. They assume a solubility product as follows: [Al(OH)*-y]^- [SioOs— y] = Ksp in which the concentrations are molar concentrations, and Y the activity coefficient. Any ion capable of reduc- ing the activity of either ion represented in the eqiiation should therefore cause the clay to dissociate. Low and Black accordingly digested kaolinite at 60° C for 20 hrs with a solution containing 8-hydroxy-quinoline buffered to a pH of 4.7, and the determined amount of silica which appeared in the filtrate. They found silica to be released by this treatment, the amount increasing con- tinuously to a maximum of 105 mg per 100 g of clay. This amount was over twice that obtained with am- monium chloride extraction, which does not precipitate the ahiminum at 10 times the concentration and twice the time of digestion. The evidence is conclusive that kaolinite dissociates and yields aluminum Avliich is then free to combine with any phosphate added, forming aluminum phosphate. ]\IeAulift"e, Hall, Dean, and Hendricks (1948) have studied this problem by use of the isotopic exchange with P*"04""". "When a clay or soil is suspended in a phosphate solution, there should be an equilibrium es- tablished between the PO4 — ions in solution and those associated with the solid phase. If P^-04 is intro- duced, keeping the total PO4 — concentration constant, it should be possible to determine the equilibrium exist- ing between the PO4 — ions on the surface and those in the solution. Since this is a simple isotopic interchange, the equilibrium constant should be unitv. Knowing the total P^-Oi"- added and the P^^Oj— and P31O4— in solution, it should be possible to calculate the amount of P^'04""" in solution. They found that two distinct changes are involved in the interaction between the phosphate in solution: the first is rapid and reaches equilibrium in 32 hours. It corresponds with the ex- change of phosphate in solution with the phosphate on the surface. In the second stage, which is much slower, the ratio of P^- in the surface to the P^- in solution increases continuou.sly with time. The phosphate which comes to equilibrium rapidly at the surface in the first step, correlates closely with the amount of available phosphate as determined by Truog's method and Neu- bauer determinations. McAuliffe et al. (1948) reasoning that this difference in rate of phosphate adsorption must be related to the extent of the hydroxylic surface as well as to the lability of the OH-groups, measured the specific hydroxylic sur- face by way of an isotopic reaction involving deuterium oxide. Total surface was determined by ethane adsorp- tion after the method of Brunauer, Emmett, and Teller. Their results show that in kaolinite and halloysite the percentage of hydroxylic surface relative to the total surface is about 2/3 of that which may be expected from the crystal structure of these minerals. The high values for the other (2:1) minerals probably result from diffu- sion of deuterium into the lattice where the deuterium atoms exchange with labile hydrogens. Hence these meas- urements indicate something as to the lability of the OH-groups within the crystal lattice. A comparison was I'art IV I ("i.AV Technology in Son, Science 185 also made between the readily exehaii'reable P the NaOH-soluble phospliorus. The data show that the per- eentagre of the total siirfaee occupied by P^' atoms is very small, which makes it appear that the fixation of lihosphate by kaoliiiite does not occur by way of simple anion exchanjj-e with the soil colloid. Only a small frac- tion of the phosphorus that was presumed to be present as exchantreable anions was in direct equilibrium with the phosphorus in the soil solution. Since such an equi- librium apparently does not exist, it is not justifiable to conclude that phosphate is fixed by anion exchange on the kaolinite lattice. DISCUSSION W. p. Kelley: In .soil clicinistry we :irp i>nil)alily (li'aliiiK. in a f,'rpat man.v and pi'rha|)s all soil.'--, with mixtures of cla.v minerals together with a great variety of other substanoes. Very commonly these clay min- erals, particularly the layer types, are either interleaved or else so intimately mixed that sejiaration is pra<*tically impossible; with the result that we rarely fin:)rdinarily be considered a very stable structure. I think the material had been given the usuiil calcination (1100° F) which gives a thorough dehydration. By using water that con- tained oxygen 18 instead of oxygen 16, in the form of steam going past the cracking catalyst, he found a very rapid exchange between the oxygen in the silica-alumina lattice and that which was in the steam going past. It is of interest that even in such a stable structure the oxygen is quite mobile and free to exchange between steam and a solid structure. 186 Clays and Clay Technology [Bull. 169 E. C. Henry: In legartl to the degradation of illite as a source of potash in soil, C. D. Jeffries made a very thorough study of the soils of Puerto Rico where he had an example of almost every type of soil, and where records had been kept for many generations (Jeffries, Bonnet and Abruna 1052). He .showed an excellent correlation betwien tlie plant .vield. the potash availability, and the amount of feldspar in the .soil. Although illite was largely absent in Puerto Kicau soils, the potash made available by the weathering of feld- spars apparently would have made it unuece.s.sary to call upon the ili'U'r.-Klatiou of illile even if lioth had been present. T. F. Bates: Leonard Sand, a former graduate student at Pennsylvania State College, bas just completed a thesis on the weathering of feldspars and other minerals to kaolinile and halloysite in the Spruce Pine district of Xorth Carolina (Sand 10.")2i. A large proportion of the clay in that region is endellite. yet there is a considerable amount of coarse mica. Sand has established that in this area the Uaolinite is formed directly from the mica, which is either primary in the pegmatites of the district, or formed by alteration of some of the feldspar. I'nder conditions of optimum leaching all of the feldspar went directly to halloysite. It is apparent that structural control imposed by the platy mica was the dominant factor which caused the form.ation of Uaolinite under conditions more suitable fiU' the formation of endellite. When these clay minerals exist in such very fine particles, as I believe they nuiy, it seems possible that the individual particles ma.v be entirely crystalline and yet be so very small and with such tremendous surface area that they may act in relation to many processes as very poorly crystallized materials. As far as the particular particle is concerned, however, the lattice may be well dex-eloppd. SELECTED REFERENCES Ale-ander. I.. T., Hendricks, S. P.., and Nelson. R. A., ]!«il. Minerals in soil colloids. II : Soil Sci., v. 48, pp. 273-79. Allison, F. E., Doetsch, J. H.. and Roller, E. M., 19.il, AmnKuiium fixation and availability in Ilarpster clav loam: Soil Sci., V. 72. pp. l,S7-200. Bar.shad, I., 194S, Vermiculite and its relation to biotite as revealed by base exchange reactions. X-ray analyses, differential thermal curves, and water content : Am. Jlineralogist, v. 33, pp. 6r).-,-78. Barshad, I., lO")!, Cation exchange in soils. I. Ammonium fixa- tion and its relation to potassium fixation and to determination of ammonium exchange capacity : Soil Sci., v. 72, pp. 361-71. Baver, L. D., 1929, The effect of the amount and nature of exchangeable cations on the structure of a colloidal clay : Missouri Agr. Exper. Sta. Research Bull. 129, 48 pp. Black, C. A., 1942, The penetration of phosphate into the kaolinite crystal : Soil Sci. Soc. America Proc. 1941, v. 6, pp. 157-161. Bower, C. A., 1951, Fixation of ammonium in difficultly exchange- able form under moist conditions by some soils of .semiarid regions : Soil Sci., v. 70, pp. 375-383. Bower, C. A., Reitemeier, R. F., and Fireman, M., 1952, Ex- changeable cation analysis of .saline and alkali soils : Soil Sci., V. 73, pp. 251-261. Bradfield, Richard, 1923, The nature of the chemical reactions of colloidal clay : Colloid Symposium Mon., v. 1, pp. 369-.391. Bradfield, Richard, 1923, The nature of the acidity of the col- loidal clay of acid soils : Am. Chem. Soc. Jour., v. 45, pp. 2669-78. Bradfield, Richard, 1924, The effect of the concentration of colloidal clay upon its hydrogen ion concentration : Jour. Pbys. Chemistry, v. 28, pp. 170-175. Bradfield, Richard, 1925, The chemical nature of colloidal clay : Am. Soc. Agronomy ,Tour., v. 17, pp. 2.j3-270. Bradfield, Richard, 1931, Some chemical reactions of colloidal clay: Jour. Phys. Chemistry, v. 35, pp. 360-373. Buehrer, T. F., Robinson, D. O., and Deming, J. M.. 1949. The mineral composition of the colloid fraction of some southwestern soils in relation to field behavior : Soil Sci. Soc. America Proc. 1948, V. 13, pp. 1.57-165. Coleman, N. T., Williams, D. E.. Xielsen, T. R.. and Jenny, H., 1951, On the validity of interpretations of potentiometrically measured soil pH : Soil Sci. Soc. America Proc. 1950, v. 15, pp. 106-114. Coleman, R., 1945, The mechanism of phos])hate fixation by montmorillonitic and kaolinitie clavs : Soil Sci. Soc. America Proc. 1944. V. 9, pp. 72-78. Coleman, R., and Jackson, M. L., 1946. Jlineral composition of the clay fraction of several coastal plain soils of southeastern United States: Soil Sci. Soc. America I'roc. 1945. v. 10, pp. 381-391. Dean, L. A., 1949, Fixation of soil phosphorus, in Xorman, A. G., Advances in agronomy : v. 1, pp. .391-411. New York, Academic Press, Inc. Dean, L. A., and Rubins, E. J., 1947, Anion exchange in soils. I. Exchangeable phosphorus and the anion exchange capacitv : Soil Sci., V. 63, pp. 377-387. Drake, M., and Scarseth, G. D., 1940, Relative abilities of dif- ferent plants to absorb potassium and the effects of different levels of potassium on the absorption of calcium and magnesium: Soil Sci. Soc. America Proc. 1939, v. 4, pp. 201-204. Fine, L. O., Bailey, T. A., and Truog. E., 1941, Availability of fixed potassium as influenced by freezing and thawing : Soil Sci. Soc. America Proc. 1940, v. 5, pp. 183-186. Hendricks, S. B., and Fry, W. H., 19.30, Tb.> results of X-ray and microscopical examination of soil colloiils : Soil Sci., v. 29, pp. 457-476. Hoagland, D. R.. and Martin, J. C, 1933, Absorption of po- tassium by plants in relation to replaceable, non-replaceable, and soil solution potassium : Soil Sci., v. 36. pp. 1-.33. Hosking, J., 1940, The soil clay mineralogy of some Australian soils developed on granitic and basaltic parent nuiterial: Council Sci. lud. Research (Australia) Jour., v. 13, pp. 206-216. Jeffries, C. D., Bonnet, ,1. A., and Abruna, F., 1952, Mineral characteristics of some soils of Puerto Rico : Soil Sci. Soc. America Proc. 1951, V. 16, pp. 310-311. Jenny, Hans. 19.32. Studies (ui the mechanism of ionic exchange in colloidal aluminum silicates: .lour. Phys. Chemistry, v. 36, pp. 2217-2258. .lenny, Hans, 1936, Simple kinetic theory of ionic exchange. I. Ions of equal valency : Jour. Phys. Chemistry, v. 40, pp. 501-517. Jenny, H., and Reitemeier, R. F., 1934, Ionic exchange in relation to the stabilit.v of colloidal .systems : Jour. Phys. Chem- istry, v. 39, pp. 593-604. Jenny, H., and Smith, G. D., 1935. Colloid chemiea,l a.spects of clay pan formation in soil profiles : Soil Sci., v. 39, pp. .377 389. Kelley, W. P., 1945, Calculating formulas for fine grained min- erals on a basis of chemical analysis: Am. Mineralogist, v. 30, pp. 1-26. Kelley, W. P., 1948, Cation exchange in soils; Am. Chem. Soc. Mon. Ser. 109, 144 pp., X^ew York, Reinhold I'ublisbing (I'orp. Kelley, W. P., and Dore, W. H., 1938, The clay minerals of California soils: Soil Sci. Soc. America Proc. 1937, v. 2, pp. 115-120. Kelley, W. P., Dore, W. H., and Brown, S. M.. 1930, The nature of the base-e.xchange material of bentonite. soils, and zeo- lites, as revealed by chemical investigations and X-ray analysis : Soil Sci., V. 31. pp. 25-45. Kelley, "\V. P.. Dore. W. H.. and Page, J. P.., 1941, The colloidal constituents of American alkali soils: Soil Sci., v. 51. jip. 101-124. Kelley, W. P., Dore, W. H., Woodford, A. O., and Brown, S. M., 1939, Colloidal constituents of California .soils : Soil Sci., v. 48, pp. 201-2.55. Larson. W. E., and Allaway, W. H., 19.50, Release of sodium from nonreplaceable to replaceable forms in some Iowa soils: Soil Sci., V. 70, pp. 249-2.56. Low, P. F.. and Black. C. A., 1948, Phosphate-induced de- composition of kaolinite : Soil Sci. Soc. America Proc, 1947, V. 12. pp. 180-1S4. MacEwan, D. M. C 1946, The identification and estimation of montmorillonite group of minerals, with special reference to soil clays: Soc. Chem. Industry Jour., v. 65, pp. 29S-.304. Marshall, C. E., 1935, Layer lattices and the base-exchange clays: Zeitschr. Kristallographie, Abt. A, Band 91, pp. 4.33-449. Marshall, C. E., 1937, The colloidal properties of the clays as related to their crystal structure : Jour. Phys. Chemistry, v. 41, pp. 935-942. Marshall, C. E.. 1948, Ionization of calcium from soil colloids and its bearing on soil-plant relationships : Soil Sci., v. 65, pp. 57-68. Marshall, C. E., 1949, Colloid chemistry of the silicate minerals: 19.5 pp., X'ew York, Academic Press, Inc. Marshall, C. E., 1951, Mineral nutrition of plants : Chap. 3, T'niv. Wisconsin Press. I'arl I VI Clay TKriixoi.ociv in Sou. Scikxce 187 Miirslnill. C;. E., iuul I'.i'i^'iiiuii, \V. I)., 11I4L'. 'I'lic flciMio-cliciiu- cal pioiiertios of niiiioral niciiiliiancs. H. .Measiiri'meiU of potjis- siiiin-idii activitips in culloidiil clnvs: Jdur. I'livs. Clieniistrv. v. 4t'i, pp. r,2(il. -Miirshall, C. E., and Bergman, \V. K.. I!i4-Ja, 'I'lio clcctii)- ilu'inifal propprties of mineral niciiilnani's. I\'. Tlip nwasuicnicnl lit" aiiMMoniiiin ion activitips in colloicial clays: .lour. I'hvs. Col- loiil Clu inislry, v. 4(!, pp. '.\2~-X'A. Martin. .1. ('., Overstrcct, U., and IIoaKlaud, I). K., 1'.I4(!, I'o- tassium fi.\ntion in soils in rppU-acpahlp and nonrpplacoahlp forms in relation to oliomical reactions in the soil ; Soil Sci. Soc .\rnerii;[ Troc. I!t4.">. V. 10, pp. !»4-10]. Mc.\nliffe, (". I)., Hall. N. S., Dean. L. A., ami Hendricks. S. I"... 1!I4S. Kxchanjie reactions between phosphates and soils, hydroxylic snrfai'es of soil minerals: Soil Sci. Soc. America I'roc. 11147. \. ^-. pp. 11!I12.-.. .Mcl'.<'tli. I. (I.. 11I17, Fixation of ammonia in soils: .lonr. .\ur. Kesearch, v. !l. pp. 14M.>j. -McCJeorye, W. T., "iU'.V.i. Potassium in calcarcons soils: .\rizoii.i A^-r. Kxper. Sta. Tech. Hnll. .".(). 42 piJ. Mehlich, A., 1042, Hase nii.satnration and plj in relation to soil type: Soil Sci. Soc. .\merica Proc. 1941, v. G, pp. ].'')0-l.">(i. Mehlich, A.. li)4o. The significance of percentage base saturation and pH in rehation to soil differences: Soil Sci. Soc. America Proc. 1!)42, V. 7, pp. 1(!7-174. Mnrata, K. .1., 1!M(), The sis^nificancp of internal strnciiire in Sel.atiniziiiK silicate minerals: T'. S. fleol. Survey Hull. !l."i(l. pp. 2.-,-:{:{. -Muridiy. H. F.. 1!«!), The role of kaolinite in pli.isphalc fixa- tion: Hilgardia. v. 12, pp. 34:i-.'iS2. Oblad, A. (;., 19.")1, Advances in catalysis: v. 2. pji. 1'.i:i-247. \e\v York, Academic Press, Inc. ■Pase. J. H., 1948, Differpntial thermal analysis of mimlmorillo- nite: Soil Sci., v. 50, iip. 27."l-2So. I'aKC .1. 15., and Paver, L. D., 1940. Ionic size in relation to lixalion of cations by I'olloidal clay: Soil Sci. Soc. America Proc. I9:!9. V. 4. pp. l.-)01. -..-). Peech, .\l.. and P.radlielil. U., 1948, Chemical methods for esti- mating' lime needs of soils: Soil Sci., v. (i.^, i)p. ,'5.^-.").j. IJpitemeier, R. F., 1951, Soil potassium, in Xorman, A. G., Advances in ai^ronom.v : v. ;». j>i>. ll.*Mri4. Xew York, .Vcaiieinic Press, Inc. Ueitemeier, H. F., P.rown, I. C. and Holmes. K. S.. 1951, Re- lease of native and fixed nonexchaiiKeable iiolassinm of soils con- lainiuK hydrous mica: U. S. Dept. Afir. Tech. Mull. 1(149. Ross, ('. S., 1927, The mineraloKy of clays: Soil Sci. Comm. First Intprnat. Cons., v. 4, pp. 555-.5()l. Sand. L. 1!.. 19,52, Mineralogy and petroloK.v of the residvial kaolins cd' the scuithern Appalachian rpf;ion : Ph.D. Thesis, Div. Mineralogy, The Pennsylvania State CoUese. Shaw, P.. T., and Humbert. R. P., 1942, Klectrcm micrographs of clay miru'rals: Soil Sci. Soc. America Proc. 1941, v. C, pp. 14(1-149. Stanford, (',.. 194.S, Fixation of pot.assium in soils under moist conditions and im dryiuf; in relation to type of clay mineral: Soil Sci. Soc. America Proc. 1947. v. 12, pp. 1G7-171. Stanford, (}., and Pierre, W. II.. 1947, The relation of potassium fixation to ammonium fixation: Soil Sci. Soc. America Proc. 194(), V. 11, pp. 1.5.5-1(!0. StcMit. p. R., 1949, .Mterations in (he crystal strucliire of clay minerals as a result of phosphate fixation : Soil Sci. Soc. America Proc. 19S9, V. 4, iip. 177-1S2. Troujf, K.. and .Tones, R. .1., 108S, Fate of soluble potash ap- plied to .soils: Ind. and Kni;. Chemistry, v. 'UK pp. 882-88.5. \*olk. X. .1.. PKi.'l. I''ormalion of musco\itP in soils, and refine- ments in siiecific ^I'avitv si'paral ions : .\m. .Tour. Sci.. 5th Ser., V. 2(i, pp. 114-129. Volk, X. .1.. 1984, The fixation of iiotash in dillicnlily avail.ihle form in soils : Soil Sci., v. 87, pp. 2(17-287. AVear, .1. I., and White. ,T. T,., 1951, Potassium fixation iu clay minerals as related to crystal struperties which distinguish the more satisfactory from the less satisfactory soils indicates that in the majority of cases clays are detrimental to stability. It is apparent that wet clay has the effect of a lubricant in diminishiuj; the natural resistance due to friction that would otherwise exist. It is necessary that the civil engineer responsible for construction of .'iny form of earth work should be informed not only concernins the (|uantity of clay minerals that are present but al.so about their n.-ilure and their potential influence on the cngiucerins properties of the soil. Civil eiigineerino; covers many activities; as a result, with tlio advanecraoiit in constrnetion techniques and accumulation of knowledge, a continuous subdivision into specialties becomes necessary. Iliohway engineering is one specialized branch of civil engineering; but it too is now being subdivided into specialized fields. If there is such a thing as an "average engineer" he is likely to feel that having spent a number of years in studying the strengths of materials, struggling with the necessary courses in the basic sciences, et cetera, it is asking too much for anyone to expect him to also bet'ome an expert ill .soil mechanics and conversant with all of the ramifica- tions of that extensive subject. Indeed, not many years ago, most construction engi- neers felt that they were making adequate distinctions when eartii work or excavation materials were classified as either rock or dirt ; they were somewhat inclined to feel that life was getting altogether too complicated when certain specialists with nothing better to do began to classify soils and to draw di.stinctions between various kinds of sand, silts, and loams, to say nothing of the cabalistic signs and symbols involved in the various soil classitication schemes. However, few engineering and technical lu-oblems retain the appearance of simplicity if one delves very far into the factors wliidi iiitliience performance under service conditions. Most civil engineers are faced with the necessity of dealing with soil problems in some phase of their work and the field of soil mechanics covers a variety of appli- cations. Time will not permit even a brief discussion of these numerous fields even if my knowledge were ade- quate ; therefore it should be understood that this dis- cussion is primarih' concerned with the effects of clay in highway pavements, bases, and foundations. In order to establish a relatif)nship between the effects or influence of clay and over-all performance, a chart, figure 1, is offered, which attempts to break down the over-all jjroblem into its primary, secondary, and even more remote contributory factors. AVhile it may appear from this chart that the position of clay is rather remote from the main point of interest, such is not the case; however, in order to establish the importance of the clay content in soils and gramdar base materials, it is first necessary to outline tlie important properties of such material and to describe the means or mechanism which enables soil masses to support loads and resist deformation. The immediate surface of a highway or airfield pave- ment is subjected to a variety of destructive influences. Among these are the thrusts which result from the trac- tive effort of vehicles either in accelerating or decelerat- ing, wliich may be described as a combination of abrasive action and rolling friction. The surface of the pavement must also witlistand the effects of wind, variation in temperature, and moisture, in addition to supporting the weight of the vehicle. The material beneath the pave- ment is also subjected to several influences, among which are the effects of pressures resulting from vertical load- ing and the effects of moisture. It may be argued that low temperatures are also damaging to subgrades, but the effect is an indirect one as there is no reason to think that any damage would result from freezing or thawing FACTORS AFFECTING THE DESIGN OF PAVEMENTS I 2 A i- To de t ermine the type and thickness of 9ose a Surface Moisture Density Equilibrium of Soil Plostic -Def ormotion of Soil Foligue - of Povement Unit Weight Base a Pave. Exponsion Pressure Type of soil <§) oislure content Compoction - Availabili ly -4-M t- Availabili ly _ Absorption 1 H„,oscopicily 0' Woter L p,,meobility T'" ^Ploslic Deformation (•) Thesf (odors otiected 6y cloy conrenf ® Resilience of Foundotion Flexibility •- Bose a Surf Porticle friction ertio Cohesion Surchorge lertio Slob strength Speed Wheel load Repet 1 1 ions ECompoclion Moisture content Soil composition (•; ® • Materials and Research Engineer, California Division of High- wa.vs, Sacramento, California. FiGURIO 1. in the absence of water. Of all the factors or influences which affect the supporting power and load-carrying capacity of soils, moisture is by far the most important variable. In the absence of water, virtually all soil mate- rials, whether crushed stone, sand, loam, or adobe, will support any vehicle load that can be carried on pneu- matic tires; therefore, if the soils beneath highway pavements could be prevented from becoming wet, soil mechanics would have little meaning and would not be of much concern to the highway engineer. Actually, so far as practical applications are concerned, the very title ".soil mechanics" is misleading, as the term makes no reference to the most important and influential cle- ment: namely, water. When engineers mix as little as 5 percent of liquid asphalt with mineral aggregate, the resulting combination is called an asphaltic mixture; if 15 percent of portland cement and water is combined with sand and gravel, the combination is called portland- (101 ) 192 Clays and Clay Technology [Bull. 169 cement concrete; but when faced with the necessity of predicting the behavior and supporting power of soil which may contain as much as 25 percent of water, the engineer still tends to tlelude himself by thinking that he is concerned only with soil mechanics. As stated before, it was not many years ago that engi- neers generally paid very little attention to differences in soil types. Most railroad and highway engineers learned long ago that classified excavation is usually a source of trouble and controversy between the contractor and the engineer ; and agencies in many states — includ- ing the California Division of Highways — do not classify excavation materials. In other words, the contractor is paid on the basis of his bid price whether he is moving loose soil or solid rock. In spite of these simplifications, engineers began to realize some 25 years ago that mate- rials encountered along the highway routes were not always giving equally satisfactory performance and one of the first factors to receive consideration was compac- tion. The California Division of Highways pioneered in setting \ip requirements to control the compaction of soils during construction. It did not take long to discover that in order to secure the maximum efficiency with field-compaction equipment, careful control of the moisture content was necessary. However, with the passage of time and increasing study of the problem, it was evident that thorough compac- tion alone was not enough to guarantee satisfactory sup- port in all types of soils ; it also became apparent that there was need for some laboratory -test procedure that would evaluate the capacity of the soil to sustain loads when in its most vulnerable condition of density and moisture content. Dating from Coulomb's time (1736- 1806, most theoretical discussions on soil mechanics recognize that the entire structural strength of soils rests upon two distinct properties : namely, internal fric- tion and cohesion. The stability of slopes and embank- ments, the pressures against retaining walls, and the ability to sustain vertical loads, all depend upon a com- bination of these properties — friction between the rock or soil particles and any cohesive forces which may exist. The laws governing the frictional resistance between solid particles are probably very complex and it is to be doubted whether the nature of this phenomenon is clearly imderstood, even today; but for practical pur- poses variations in frictional resistance may be well enough defined in the terms used by Amontons, who was one of the first to investigate frictional phenomena. Amontons' law states that the resistance to sliding be- tween adjacent particles in contact varies with the na- ture of the surfaces, directly with the pressure which holds the surfaces together, and is independent of the speed and the apparent area of contact. The common definition of cohesion as used in soil-mechanics texts does not correspond to the dictionary definition ; it is generally stated that the cohesive resistance in soils is that part of the resistance which does not vary with the pressure. This is a rather negative definition, but it agrees with the observed behavior of viscous liquids. The internal friction of liquids is virtually independent of the pressure : the resistance developed by a film of viscous material between two solid bodies varies directly with the area and with the speed of action, biit is largely independent of the pressure. While the civil engineer considers that the term "soil" includes all materials of the earth's mantle, including gravel and sand, as well as the silt and loam, it is true that virtually all such materials contain a greater or less percentage of extremely fine particles having the special properties of plasticity and mobility that are characteristic of the clay minerals. If there is a complete absence of the clay fraction, the material finer than a No. 4 sieve is to all intents and purposes a sand, even though some of the particles are very small. Of course, most clays possess properties that are not solely attribu- table to small pai-ticle size alone. It is not the purpose of this paper to attempt any discussion of the intimate shape and composition of the clay particles. From the standpoint of the highway engineer who is attempting to evaluate the behavior of soils as an engineering mate- rial, it is sufficient to recognize that Avhen combined with sufficient water most clays are very effective lubricants and when a sufficient quantity of clay is intermixed with the coarser granular portion of a soil, lubrication will develop whenever enough water is added. As the ability of soils to resist deformation depends very largely on the internal friction, wet clay has the effect of reducing or canceling out the frictional resistance. It may also be pointed out that the so-called cohesive resistance is induced almcst entirely by the clay fractions, and there- fore that clean sands are noneohesive. Again we must note the important part played by water, as finely ground dry clay particles exhibit no cohesive properties. If water is added to a dry soil, the cohesive resistance will normally increase with the addition of moisture and in most cases the frictional resistance will not be greatly impaired until a certain amount of moisture is added. Beyond this point, the friction will diminish but the cohesive resistance may continue to increase up to some point of higher moisture content, after which both values will diminish as the soil approaches a completely fluid state. As the wet clay fraction reduces the internal resist- ance by lubrication but may increase the resistance by improving the cohesion, it is necessary to determine something of the relative importance of these two prop- erties. From a practical view point, both field and laboratory experience demonstrate that of the two basic properties, resistance resulting from friction is greater in magnitude and therefore the most important when dealing with typical soils or granular materials. Beds of clean sand, even when wet, have long been known to furnish excellent foundation support provided a rea- sonable thickness of surfacing material is placed above them. Crushed .stone or gravel with no measurable cohe- sion is excellent for base course construction and is able to withstand virtually any vehicle load if covered with relatively thin surfaces having the requisite tensile strength. On the other hand, plastic soils or asphaltic mixtures having high cohesive values but little internal friction are rarely adequate to sustain vehicle loads. It is realized, of course, that if the cohesive or tensile strength could be made sufficiently high, internal fric- tion would not be necessary. Metals are typical sub- stances having little or no internal friction; in them, resistance values are almost entirely a result of the cohesive or tensile strength. However, natural soils con- I'art Vl Clay Technology in Soil Mechanics 193 taiiiiiijr appirciaUie amounts of water are not capable of developing such high cohesive values and therefore Tlie internal friction is the most important property. This conclusion is in accord with the observation that an excessive amount of clay is detrimental and soils or gravel containing high percentages of clay invariably become unstable and lack supporting- power when wet. Having recognized the gcnci-ally adverse iiiHuence of clays, it is pertinent to in(iuire whether there are dif- ferences in performance which result from the type of clay, as it is well known that there are numerous types and classes among the clay minerals. While specialists and experts in clay teelinology must perforce recognize many of the variants and peculiarities of clays, the high- way cngiiu'cr has gone a long way if he recognizes three main groups : namely, kaolinite, illite, and montmorillo- nite. Again he is only concerned with their physical prop- erties and it is well known that most clays of the mont- morillonite class possess a very high affinity for water, tyjiicaily showing considerable expansion or swelling; when wet they are very effective lubricants. Kaolinite is at the other end of the scale, having in general a much lower capacity for water, and retaining a greater in- ternal resistance from friction than does bentonite, for example. Illite clays appear to be somewhat intermediate. In addition to the particle shape or structural differ- ences that are characteristic of each of these types of clays, it is also evident that the physical properties may be markedly affected bj^ small additions of water-soluble salts or other organic or inorganic compounds. A knowl- edge of these behavior patterns and the effect of such elements on the fluidity and plasticity of clays is, of course, one of the essential branches of clay technology, and requires specialized knowledge which the average liiglnvay engineer or even a well equipped highway ma- tci'ials department can hardly hope to connnand. The engineer 's interest, however, is primarily involved in the question of the over-all effect of the particular clay when the soil becomes wet and the primary question is the effect on the structural stability which, as stated above, realh' involves the possible reduction of internal friction. \ MUD or TUT/N6 MAClflNC. \— p ■ v^i PI3T0H FOR AfFLYINO LOAD TO yiCIMCH ADJU^iTARU jrAOf FLEXIBLE DIAPHRAGM ■Liquid under jmall initial pressure 'fiATiN or Tfjmo im:HINE Figure 3. Figure 2. In order to measure the ability of soils and granular materials to sustain loads, many tests have been devised ; one employs an instrument called the "stabilometer," which is used in the laboratory of the California Division of Highways. This testing procedure makes it possible to prepare soil specimens by introducing sufficient water to fill the void space, compacting to a state of density comparable to that developed in most highway bases and basement soil layers, and then by measuring the resist- ance to deformation. Basically, the stabilometer is a form of plastometer. The test reflects primarily the internal friction or degree of lubrication; cohesive resistance plays a minor part. A compacted sample being tested in this instrument, is sub.iected to a vertical load which may be varied at will. For highway purposes it is typically 160 psi. The instrument makes it possible to measure the lateral pressure transmitted by the specimen (see fig. 3). Figure 2 is a chart showing characteristic curves illustrating loss in stability or internal resistance of a crushed sandy gravel due to the addition of increments of plastic clay. This test is used as a basis for calculating the supporting value of the soil; by use of suitable formulas it is possible to compute the tliickness of cover courses of bases and pavement which will be necessary to support vehicle loads of a given magnitude and num- ber of repetitions. These test precedures have been re- ported in considerable detail elsewhere and will not be discussed further here. The stabilometer test and the elaborate compaction equipment required to produce specimens that are characteristic of soils in place in the roadbed mean that this test procedure is virtually re- stricted to a fairly large, well-equipped laboratory. It is essential however, that the engineers in charge of construction should have some ready and convenient means for detecting the presence of excessive amounts of adverse clay or fine material. In view of the fact that the lubricating effect of clay or of any other material is -91001 194 Clays and Clay Technology [Bull. 169 critical for most natural soils; therefore, a working solu- tion of 0.05N has been adopted and will be used imtil accumulated experience may warrant a change. After some experience with the calcium chloride solution, it was found that the addition of a small amount of glycerin produced a stabilizing effect and test results were more readily reproducible when made on carefully quai'tered samples. Finally, it was noted that the cal- cium-chloride-glycerin solution was not sterile and cer- tain molds tended to grow. In order to sterilize the solu- tion, formaldehyde was added. When the sand equivalent test was first developed, it was hoped that it would furnish a good indication of the overall resistance value of the soil. A correlation 5% Beotoni'e So Gr * 2 3 Figure 4. dependent upon the volume or the thickness of film between the particles, the most fundamental relationship depends upon the effective volume of clay that exists in each soil. In order to speed up the testing operation by avoiding the need for weighing the sample and drying it in an oven, a test called the "sand equivalent deter- mination" has been developed. This test is applied to a sample of soil passing a No. 4 sieve. The relationship between the quantity of claj- present and the amount of coarser sand particles in the soil is developed on a volume basis, and the test results indicate whether the volume of sand is high or low — hence the name "sand equivalent" (S. E.). Essentially, the sand equivalent test is performed by shaking a sample of the fine aggregate vigorously in a transparent cylinder containing a special solution (fig. 5) and noting the relative volumes of sand and partially sedimented clay after standing for 20 minutes. The entire operation can be carried through in less than 40 minutes. In order to speed up the sedimentation of the fine clays or colloidal particles, a flocculating agent was required and a solution of calcium chloride was selected on account of its relatively low cost, stability, and non- irritating properties. As illustrated in figure 4, a small amount of bentonite is in lubricating effect equal to a much greater weight of kaolinite and the strength of the CaCl2 solution was adjusted to the point where 5 percent of bentonite woiild give an S.E. reading approximately equal to that produced by 21 percent of kaolinite after a sedimentation period of 20 minutes. This relationship appeared to be best established by using a 0.02.5N CaCI. solution. However, the strength of the solution is not Fir.iiRF. 5. I'ait V Clay Tkchxui.ogy ix Soil ilixiiAxio; 195 iloes exist, but it is not sharply defined tlirouphout the seale. The reasons therefor arc not diffiL-ult to under- stand, if it is reeofrnized thiit tlie ability of a mass of soil or eranular material passinii' a No. 4 sieve to resist deforniatinii will de]iend upon the followinii' factors: (1) The amount of lubric-ant mixed with the saud frac- tion; i.e., asphalt, clay water, etc.; (2) The effectiveness or efficiency of the lubrieatinfr fraction wet bentonite is a better lubricant than kaoliuite, for example; (3) Tlie deirree of roufihness or irretrularity of the sand araius or rock particles; (4) The amount of void space in the saud fraction of the soil; and (a") The amount of iutermiufjled coarse rock retained on a No. 4 sieve. "We readily perceive that of these five variables the sand equivalent determination is primarily an indica- tion of no. 1. It attem]its to compensate for no. 2 by means of the type of solution used. It cannot indicate the variation caused by item :3, and as presently per- formed does not make allowance for no. 4. althoup'li it seems possible that means for makinir this correction may be worked out. Allowance for the effects of no. 5 need to be made if the coarse aggregate exceeds 25 or 30 percent of the total. Therefore, in order to evaluate the combined effect of all factors, some test such as the stabilometer test is necessary. However, experience has shown that one of the principal variables is the amount of clay present, and it may readily be determined that when the sand equivalent value is greater than 30, the clay fi-action is not sufficiently large to have much in- fluence on the resistance value of an untreated soil. Very small amounts of clay may be detrimental to the performance of bituminous mixtures, especially when the clay exists as a coating on the surfaces of sand grains. As the sand equivalent determination furnishes a read.v means for detecting the presence of such fine materials, a tentative scale of values has been set up to permit i-apid testing and quick determination in the field. A comparison of sand equivalent test values with other test results indicates that tlie majority of soils showing high expansion under soaking may be identified by means of the sand equivalent. It has been the general practice to consider that an}' soils showing an expansion of greater than 5 percent when tested in tiie California bearing ratio procedure will be unsuitable for placing in the U]iper levels of a road bed. It appears that the same class of soils could be identified and segregated by stipulating that any soils having a sand ecpiivalent less than 10 should not be placed in the upjior layers, as they are also likely to develop excessive expansion when saturated. In conclusion, it may be stated that the suitability of soils for engineering purposes depends largely upon their ability to remain in place and to support wliatever loatls may be ))laced upon them either by a i)ennanent engi- neering structure or by transient vehicle loads. A study of the properties which distinguish the more satisfac- tory from the less satisfactory soils indicates that in the majority of cases clays are detrimental to stability and that wet clay has the effect of a lubricant in diminish- ing the natural resistance due to friction. It is necessary tluit the civil engineer responsible for construction of Any form of earth work should be informed not only concerning the quantitj' of clay minerals that are pres- ent but also should know something of their nature and their potential influence on the engineering properties of the soil. DISCUSSION D. P. Krynine: For those interested in permafrost some Russian l)o()lis trans- lated into English ma.v be recommended. The Steffenson Lil)rar.v translated for the Army Engineers a book by V. F. Zhidiov en- titled Earthuork for Foiiniiations in Permafrost Zones, Moscow, 1946. The U. S. Geological Survey translated the liook l)y N. A. Tsitovitch and M. I. Soumgin entitled Principles of the Mechanics of Frozen Giotmds, Moscow. 1937. Neither translation has l)een printed, however. Russian scientific work on permafrost has been tised in engineering practice, but much is still in the experimental stage. A still unsolved i>rol)lem in soil mechanics is the shearing resistance of clays, both saturated and non-saturated, especially of the latter. Mineralogists and physicists probably eoukl eon- tribute considerably to this study, as jiractically nothing is known of the behavior of the lattice during the shear process, particu- larl.T in brittle clays and in clays flowing plastically. A phenome- non which is well known to the engineers is the difference in shearing resistance of a clay sample as determined by slow and tiuick shearing tests. Further clarification of the causes of this difference is needed. PHYSICAL-CHEMICAL PROPERTIES AND ENGINEERING PERFORMANCE OF CLAYS By Richakd C. Mielenz and Mtrle E. King • ABSTRACT Fabric (texture and structure i and miueralogic compositiou de- termine tlie response of clays and shales to events occurring dur- ing construction and operation o£ engineering works. A new system of classification of the fabric of earth materials is proposed. Char- acteristic mineralogic composition of clays and shales, especially in the western United States, is described. Fabric and composition are correlated with soil mechanics properties and engineering per- formance. Needed research on clays and shales as a basis for de- sign, construction, and maintenance of engineering structures is emphasized. INTRODUCTION Clays are important to the designer and the construc- tion engineer because their structures frequently rest Tipon clayej' formations, excavations commonly must be made into clayey materials, vast quantities of earth materials containing clays are used in embankments and linings, and clays occur commonly as constituents of engineering materials such as aggregate, pozzolan, and grout. Moreover, ela.ys present many unique problems to the engineer primarily because their physical and chemical instability renders masses of earth susceptible to ready and repeated change of form and volume in response to loading or unloading, vibration, and changing moisture content. As though these qualities were not suiSciently annoying, the degree to which clays respond to these actions commonly changes with adsorp- tion of ions or molecules from solution, precipitation interstitially of granular substances, and alteration of internal texture and structure. Through recognition of these characteristics of clays, the engineer can minimize unanticipated difficulties in construction and operation of engineering works. Also, intriguingly enough, the susceptibility of clays to change in response to their physical and chemical environment leads at once to methods by which undesirable properties can be improved. With increasing size and complexity and hence cost •of structures, success of the works must be more assured in spite of the fact that available sites commonly are less desirable from the standpoint of stability. Only the first chapters in the tale are yet told, but advance is being made in this field of soil mechanics under the leadership of engineers and scientists in many countries. Of particular note are the contributions of H. F. Winterkorn, A. Casagrande, E. F. Preece, B. K. Hough, T. W. Lambe, E. A. Hauser, D. T. Davidson, and their as.sociates in the United States; K. Endell, E. H. Ackermann, E. Schmid, H. 6. F. Winkler, and their associates in Cxermany; L. F. Cooling, L. Casa- grande, A. W. Skenipton, K. E. Clare, P. C. T. Jones, and their associates in England ; J. E. Jennings and liis associates in South Africa; and others in Russia, Holland, France, Switzerland, Sweden, and elsewhei-e. Engineers also are concerned with clays in areas other than soil mechanics. Work is in progress in study of claj'S as they affect both natural and manufactured aggregate, as thej' can be used in grouting or drilling fluids, and as they contribute to activity of pozzolanic • Head, Petrographic Laboratory, and Petrographer. respectively, Engineering Laboratories, U. S. Bureau of Reclamation, Denver, Colorado. admixtures for concrete. These investigations are being conducted in governmental, university, and industrial laboratories in the United States, England, France, Ger- manj', Russia, Italy, Mexico, Japan, India, Brazil, and other countries. Development of physical-chemical concepts in engi- neering has drawn heavily from knowledge in related fields, particularly agriculture, mineralogy, geology, ceramics, and chemistry. Workers in Germany, United States, England, France, Russia, Australia, and Japan have been especially fruitful. This paper describes the properties of clays in relation to their mineralogy and fabric and their performance during construction and operation of engineering works. Particular emphasis is given areas of ignorance with a view to stimulating research directed to their explora- tion. SOIL MECHANICS AND THE PHYSICAL-CHEMICAL APPROACH TO ENGINEERING INVESTI- GATION OF CLAYS Developing hand in hand with the applied science of soil mechanics, principally during the past 25 years, has been an increasing awareness of the need for a better understanding of the physical-chemical processes affect- ing the response of earth materials to engineering activities, such as loading or unloading of foundations, changing of the ground-water regime, introduction of new .substances, and excavation and recompaction. Soil mechanics is based upon physical, mechanical, and hydraulic laws. In engineering application, proper- ties of earth materials are obtained from field and laboratory tests which are carefully designed to be quick and comparatively low in cost. These tests are the basis for most design and construction practice for structures resting upon or composed of earth materials. To be classed among the common procedures are the more or less standardized tests for particle-size distribution, con- solidation, shear resistance, penetration resistance, per- meability, and many others. Consistent with the point of view permeating all engineering, these tests are applied to measure the performance of materials under condi- tions simulating those on the job. Most remarkable of the empirical procedures is the so-called airfield classifica- tion (AC) system, developed by Arthur Casagrande (1948; Abdun-Nur, 1950) which is based primarily upon visual inspection and observation of the response of earth materials to simple manual tests without equip- ment. This system with some modification is now used widely in engineering practice. The general success of the metliod demonstrates that its empirical concepts take root in fundamentals of performance of earth materials in engineering. Physical-chemical investigation of clays is based upon analyses of fabric, mineralogic constitution, and the chemical composition, and the response of the materials to chemical or physical-chemical treatment. Although physical-chemical research on soil mechanics problems was begun more than 15 years ago, progress has been slow. Knowledge of physical-chemical properties of soils (196) Part \' Clay Technology in Soil .Mlciiaxics 197 developed in the fields of mineralofry. •reolop:y, aorronom}% ceramics, and chemistry has been adajited to study of enfrinoerinij problems and especially devised investi- gations have been conducted in several laboratories, most notably by H. V. Winterkorn, T. W. Lambe and their associates' These methods have been applied successfully in several large engineering organizations to aid in plan- ning of soil investigations, as supplements to standard soil mechanics tests, to elucidate anomalous behavior, to forewarn of possible difficulties, to aid in selection of de- sign and construction methods, and to develop i-emedial measures to correct failures in opei-ation. They do not replace the physical and mechanical tests. Nevertheless, examples of the successful application of physical-chemical knowledge of soils to the solution of significant engineering problems are few. Lee's control of the permeability in the lining of the lagoon on Treas- ure Island in San Francisco lUiy ( 1(141 h and the in- vcstitration of soil stabilization by Winterkorii (1940; l!)4(i; l'J53; and Eckert, 1940; and Chondbury, 1949), Casagrande (1937; 1939; 1947; 1949), Endell (1935), David.sou (1949; and Glab 1949), Preece (1947), Lambe (1954) and others are good examples of the application of fundamental knowledge of mineralog3% fabric, and physical-chemical characteristics of soil. They portend an enlarging future. The paucity of application of fundamental informa- tion to the solution of engineering problems is explained by difficulties iidierent in the solutions and by the in- ability of most of those capable of effectively analyzing the physical-chemical conditions controlling earth ma- terials to specify practical remedial measures which will overcome the engineering difficulties. In turn, this lack of ability is in part the result of the limited opportunity afford(>d individuals skilled in the physical-chemistry of earth materials to follow continuously the development of engineering projects from the earliest stages of in- vestigation and preliminary design. Indeed, only in the ])ast decade has the geologist gained acceptance in this regard. The opportunity for tlie earth materials scien- tist still is to come. In general and with rare exception, the physical- chemical conditions controlling earth materials at engi- neering sites are elucidated, if at all, only by .special investigations conducted after difficulties are encoun- tered and only after the time is past for successful appli- cation of methods based upon physical chemistry. The imminent destruction of costly structures or the need for reconstruction of damaged facilities usually demand direct means, such as excavation of the oft'ending un- stable materials, driving of piles, or others, including, all too commonly, the restoration of the condition which originally resulted in failure. Scientific investigation of pi-ol)lcms of soil engineer- ing cannot replace empirical mechanical tests, for these supply entirely adequate data for most projects. Rather physical chemistry, petrography, and geology permit critical evaluation of data developed by the empirical tests at individual projects, because inquiry along these lines reveals the precise cause of given performance, the continuity and extent of materials with given properties, and structural relations within soil masses, such as atti- tude of strata, faults, and lithologic discontinuities. Moreover, they form the basis for improvement of exist- ing tests and define the need for additional new standard tests. Most important, however, the scientific study of soils for engineering purposes will (1) forewarn of dangers inherent in specific engineering situations, (2) make possible the development of superior means of stabilization and selection of the most appropriate method for use on individual projects, and (3) dictate means by which the danger of failure of soil masses in service can be minimized, as for example, through appro- priate design of drains, vents, or moisture barriers where control of water is essential. To be of maximum service to the designer and the construction engineer, the soil scientist engaged in engi- neering problems always must recognize that the ulti- mate objective of the effort is the constructing, rebuild- ing, or maintenance of a structure. Plis investigations and recommended procedure should be consistent with that end. The criterion of successful engineering is adequacy at minimum cost. PHYSICAL-CHEMICAL PROPERTIES OF CLAYS Physical-chemical proiierties of earth materials are those qualities and responses which arise in the mutual spacial relationships of the constituent molecules. Physi- cal-chemical properties merge imperceptibly into chemi- cal properties with rising significance of phenomena involving oidy individual molecules, atoms, and ions. On the opposite end of the dimensional scale, physical- chemical properties grade into physical properties as the primary phenomena come to involve to greater and greater degree the mutual relations and mechanical interaction of major units, such as rock particles, whole cr3'stals, masses of earth material, or bodies of fluid. Clearly, all of the observable properties of mate- rials arise by combination of physical, physical-chemical, and chemical properties, each to greater or lesser degree. Physical-chemical properties of interest in soil me- chanics of earth materials depend primarily upon electrostatic and gravitational forces existing at the surface of the solid phases, the area of surface per unit volume of material available to fluids permeating the mass, and the amount and kind of interstitial liquids and gases. Consequently, physical-chemical properties of earth materials arise in mineralogical or compound com- position and fluids content. By control of cohesive and adhesive forces between solid and solid and between solid and fluid, the physical-chemical properties influ- ence or control development of fabric of materials. The nature and magnitude of physical-chemical properties of the mass depend upon both mineralogic or compound composition and the fabric at all levels of dimension at and above the molecidar. FABRIC The fabric of an earth material is the pattern estab- lished by the arrangement and mutual relationships of the constituent particles and amorphous masses and the discontinuities. Fabric constitutes a more or less con- tinuous three-dimensional repetitive pattern in space, and includes such factors as packing, boundary relation- ships, discontinuities, grain size or granularity, the presence or absence of a matrix, shape and roundness of particles, and orientation. As used in petrofabrics, fabric also includes the symmetry of arrangement; 198 Clays and Clay Tecuxology [Bull. 169 TaHe 1. A tpj-turul clussificdlion of cailli iiiiiterinls (major subdirisions only). Texture Definitive character Clastic... - ., , .. Granular, without interlocking of grain boundaries Crystalline _ Biof ragmental '_ Particulate, with particles conchoidal. fibrous, radiate, acicular, spindle-shaped, reticulate, or spiral further work may make possible the application of these principles to earth materials. In this paper, "fabric" will include large or small physical features of an earth material and so encompasses concepts of both "structure"' and "texture," as commonly used by many soil scientists, petrographers, and geologists. The much-debated term "sti'ucture" is reserved here for the gross geologic features, such as bedding, deposi- tional attitude, and the over-all form or shape of a geologic body. A discussion of these structural features is not included. For the purpose of this paper, a grain is defined as a hard jiarticle whieli acts as a unit iinder stress. An aggregation is defined as a coherent assemblage of grains or argillic particles so joined as to act as a unit under stress. In the nomenclature of petrofabrics, grains have been called elements or units, whereas aggregations have been called superindividuals (Fairbairn, 1949). Based Tipon the mutual relationships of the constituent particles and amorphous masses, the fabric of earth mate- rials can be classified as clastic, biofragmental, crystal- line, and amorphous (table 1). Classification of materials according to this system does not depend upon a knowl- edge or an assnmi:)tion of the origin of the material, al- though by observation of the fabric, the origin usually can be deduced. The clastic materials are particulate and the bound- aries of the particles do not interlock. This group includes the terrigenous sediments (inch;ding some limestones), pyroclastic materials, and breccias. The biofragmental materials are a variety of clastic mate- rials, but are sufficiently distinctive in the shape of the particles to justify separation. Biofragmental earth materials are composed of grains which are spindle- shaped, spiral, fibrous, radiate, rod-like, conchoidal, or cylindrical. They usually consist of the hard parts of organisms, such as diatoms, radiolaria, sponges, mol- lusks, and corals. The cry.stalline fabric is composed of particles whose boundaries interlock. Materials belonging to this group typically are those precipitated from solution, and include crystalline igneous rocks, hydrothermal vein deposits, metamorphic rocks, and chemically precipi- tated sediments, such as many limestones and saline deposits. The amorphous fabric is not particulate, being essen- tially composed of noncrystalline masses. This fabric is exemplified by volcanic glass, opaline chert, and certain organic materials. Clearly, gradations exist between the four types of fabric. Clastic and biofragmental materials commonly are admixed with crystalline or amorphous substance*. Also, crystalline materials are combined with amorphous substances in widely A'arying amounts. Clastic, biofrag- mental, amorphous, and crystalline rocks and soils are completely intergradational. Clastic fabric may be interrupted by discontinuities, such as fractures, abrupt changes in grain size, holes produced by plant and animal life, changes in matrix, and natural openings produced by leaching. Classifications of fabric may be based most definitively upon (1) the nature and frequency of discontinuities (table 2) and (2) the relationships of the particles and amorphous masses constituting the material (table 3). The writers wish to oixtline the principles of the system primarily to stimulate investigation and research in this neglected field of soil mechanics. No new nomenclature is introduced. Also to be recognized is the point that the objective of the classification scheme is the systematic examination of fabric. Consequently, no attempt is made to correlate the various fabrics with specific petrographic types. Indeed, it is readily evident that with the common system of petrographic nomenclature now in use, earth materials of widely different fabrics might be given the same lithologic name, such as sand, silt, clay, loess, and so on. With a clear concept of fabric in mind, those in- terested in soil mechanics might ultimately develop a .system of nomenclature which will designate a variety of fabric tjrpes. However, this is for the future. A systematic study of the fabric of earth materials is important to the soil mechanics engineer because the properties and performance of earth materials depend upon the mutual arrangement of constituents and the discontinuities (A. Casagrande. 1932). Conseq^iently. rational analysis of the properties and performance of materials in tests or service can be accomplished only if differences of fabric are recognized. By correlating fabric with results of tests and experience, the engineer can predict more reliably the properties to be anticipated in new ground, the need for specific tests will be recog- nized, and the data obtained fi-om tests can be extrapo- lated more definitively over the area of interest. Fabric Based on Discontinuities When a fabric is interrupted by discontinuities, the resulting forms and patterns alter the over-all fabric of an earth material and change its properties. For example, two materials having similar composition and particle-size distribution in bulk, but differing in dis- tribution of the coarse and fine particles in the fabric, as with stratification, will exhibit widely different physi- cal and phj^sical-ehemical properties. Similarly, small fractures, plant root holes, or pores affect the mass properties of the earth material. The forms produced by fracturation are quite diverse. Fracturation may be caused by unequal stresses, such as result from differential heating, heating and cooling, wetting and drying, freezing and thawing, swelling and .shrinking, chemical dissolution, consolidation or shear failure, and the action of plants and animals. The dis- tinctness of fracturation and stability of the forms are influenced markedly by the type and kind of clay minerals, organic matter, and other cementing materials. In order to elucidate the di^-erse forms which an earth material can exhibit, a fabric classification of clastic earth materials based on discontinuities is proposed Part V] Cl.AY Tuf'IINOLOGY IX SoiL MECHANICS Table 2. A classification of fabric of clastic earth materials based on discontinuities. 199 Group Type Kind Size of features Aggregation Fracture." snbnrdinate Isotropic Massive Nonporous Porous Highly porous Fine Medium Coarse Absent Poorly defined Moderately defined Well defined Anisotopic Planar Foliated Laminated Lenticular Scaly Fine Medium Coarse Absent Poorly defined Moderately defined Well defined Isotropic Blocky Spheroidal Cuboidal Irregular Fine Medium Coarse Absent Poorly defined Moderately defined Well defined Fractures prominent Anisotropic Columnar Prismatic Cylindroidal Fine Medium Coarse Absent Poorly defined Moderately defined Well defined Planar Foliated Laminated Lenticular Scaly Fine Medium Coarse Absent Poorly defined Moderately defined Well defined Tahle 3. A classification of fahric of clastic earth materials based on particle relationships Packing Grains in con- tact Clastic Grains not in contact[or Ar- gillic Granularity Coarse Granular >4.76 mm Medium Granular <4.76 mm >0.074 mm Micaceous Fine Granular <0.074 mm Micaci Coarse Granular >4.76 mm Medium Granular <4.76 mm >0.074 mm Fine Granular < 0.074 mm Medium ArgiUic <4.76 mm > 0.074 mm Fine Argillic <0.074 mm >0.005 mm Ver>' fine ArgiUic <0.005 mm Coatings Grains coated Grains not coated Grains coated Grains not coated Grains absent Grains coated Grains not coated Type of matrix No matrix Clastic matrix Granular Argillic Crystalline Amorphous Biofragmental Intergranular braces Argillic Amorphous Grains separated by matrix ArgiUic Amorphous Biofragmental ArgiUic Amorphous Orientation Kandom Preferred Random Preferred Random Preferred Cementation None Points of contact Intergranular filling None Points of contact Intergranular filling None Interparticle filling Particle form Rounding: Roimded Subrounded Subangular Angular Shape: Equant Tabular Rod-Uke Irregidar Rounding: Subangular Angular Shape: Equant Tabular Rod-like Irregular Rounding : Rounded Subrounded Subangular Angular Shape: E(|uant Tabular Rod-Uke Irregular Rounding: Subangular Angular Shape : Equant Tabular Rod-Uke Irregular Shape: Tabular Rod-like 200 Clats and Clay Techxology [Bull. 169 FiGUKE 1. Claystone, Wellton-Mohawk Canal, Arizona, showing change from Ca-beidellite (massive) to Na-beidellite (cracked). Chiy asgregations show within Na-beidellite fragments. Magnifi- cation ox. (table 2). This classification is somewhat similar to that proposed by the National Committee on Soil Structure (Russell et al., 1929) but differs in that this classifica- tion is not limited to soils. It encompasses fabrics of both loose and indurated soils, sediments, and rocks of all origins. The first broad division of earth materials depends upon the prominent or subordinate character of the fractures. Fractures interrupt the continuity of the fabric ; hence, two earth materials having similar grain relationships, but differing fracturation, will differ in physical properties. For example, a claystone near Yuma, Arizona, is in the process of changing from a Ca-beidellite to a Na-beidellite. Both materials have similar internal grain relationships, but differ in fabric discontinuities as the result of excessive shrinkage of the Na-beidellite (fig. 1). Each of the two broad divisions of the classification is subdivided into isotropic and anisotropic groups. Mate- rials classified as isotropic exhibit three almost equal axes or are uniform in all directions ; whereas, aniso- tropic materials possess distinctly different dimensions in different directions. The isotropic fabric can be mas- sive or bloeky. Massive types are devoid of any per- ceivable form ; whereas, the blockj^ type is characterized by forms resembling a rough cube or other equidimen- sioiial forms. The anisotropic fabric can be columnar or planar, depending upon whether the vertical axis is longer than the horizontal axes (columnar) or is shorter (planar). The massive type of fabric can be nonporous. porous, or highly porous (fig. 2). The pores may be larger or smaller than the grains or aggregations. The bloeky type exhibits three kinds of form ; namely, spheroidal, euboidal, and irregular. The columnar type is character- ized by prismatic or eylindroidal kinds of form. The prismatic form can be further differentiated into those which are hexagonal, square, or trigonal in cross section and into long or short forms. Figure 2. Massive, highly porous fabric (grains in contact). Navajo sandstone, Glen Canyon Dam site, Arizona. Magnifica- tion 10.x. The planar type exhibits discontinuities which repre- sent either differences in particle size between layers, cleavage in one direction, or other two-dimensional fabric changes. This type is differentiated into four kinds of form. A foliated planar fabric consists of very fine dis- contiinious layers, which can be almost parallel or gently undulating and can be easily split into flakes. A lami- nated planar fabric includes more or less eontinuous layers of varying thickness (figs. 3 and 4). A lenticular planar fabric is composed of aggregates which are shaped like a double convex lens and are discontinuous (fig. 5). A scaly planar fabric is composed of aggregates shaped like a concavo-convex lens. Three size grades are recognized for each of the var- ious kinds of form, being designated as fine, medium, and coarse. The fine grade includes forms less than 2 mm in diameter. For platy, bloeky, and massive types, the medium grade is 2 to 10 mm, and coarse grade is greater than 10 mm in size. Being usually larger features, the columnar types range from less than 1 cm for the fine grade, 2 to 10 cm for medium grade, and over 10 cm for the coarse grade. The degree of aggregation of each of the kinds of form is designated as absent, poorly defined, moderately defined, and well defined to denote the distinctness with which the various aggregations manifest them- selves. Aggregation is absent in the Navajo sandstone (fig. 2). Aggregations may or may not be water stable. Aggregations commonly are water stable when coated or impregnated by lignaceous or proteinaceous sub- stances or iron oxides. The void spaces between aggre- gations might be empty or filled by amorphous or crystalline material. The Porterville claystone in the vicinity of Lindsay. Exeter, and Porterville, California, shows well-developed aggregated fabric, the aggregations being composed mainly of beidellite and coated with iron oxides (fig. 6). The aggregations resist slaking in water unless worked mechanieallv. Part \^ Clay Technology in Soil ilicciiANics 201 ^i^ )PSi6WJ O.^iU'**?? ■' - ?^[K*S*^' I''i(a UK .'{. riauar, laminated fabric witli fractures subordinate. Nespeleiii furniation, near Coulee City, Washington. Magnifica- tion 6x. Fabric Based on Grain Relationships Packing. Packing is the spacing and mutual arrange- ment of particles constituting a fabric. The response of jjarticulate earth materials to stress depends priiiuirily upon the jiacking of the particles, and secondarily upon the boundary relationships, particle shape, properties of the particles, and amount and properties of the matrix. Consequently, a classification of fabric based upon particle relationships sliould indicate at once the mutual spacial relations of tlie j)rimary components of the fabric. Accordingly, in table 3, clastic materials are subdivided into those in wliieh (1) grains are in contact and (2) grains are not in contact, or the material is argillie. In materials in which the grains are in contact, there are theoretically six different ways spherical particles can be arranged. (Graton and Frazer, 1935) Inasmuch as soil particles are not spheres, only two of the six pos- sible types will be considered ; namely, cubic and rhom- boh(>dral. These two tyi)es of packing represent the least efficient and most cfiicient modes of arraiigeinent. When grains are cubically jiacked, a moderate amount of con- solidation with reorientation of the particles will result if no matrix is present. If a matrix is present, consolida- tion with reorientation may or may not take place de- pending upon the nature of the matrix. When grains are efficiently packed, consolidation will be small even with- out a matrix. Earth materials in which grains are not in contact are the materials which are of the most concern to engineers. The stability of these materials is for the most part con- trolled by the type and amount of matrix, and the stability of water films. When particles are not in con- tact they are separated by intergranular braces, a matrix in excess of the volume necessary to fill the intergranular spaces, or water films. The volume of the matrix mate- rial in some instances may constitute only 25 to 30 percent of the whole, yet the grains may not be in contact j^'i^^Sf, Figure 4. Laminated fabric with fractures prominent. lUitic sliale, Yangtze Gorge, China. See fig. 10 for microscopical details. Magnification 6x. Figure 5. Aggregations composed of silt-sizid j;i:iiiis mIikIi ;iri' cemented together b.v a montmorillcmite-typc mineral. I.oi'ss, Asliton Dam site. Nebraslnds rather well to a genetic classification of soils. However, for soil mechanics ]ir()iierties, packing is found to be the most fundamental as|)cct of fabric, the nature of grain coatings being of considerably lesser significance. Type of Matrix. The absence or presence of a matrix exerts a great influence on the physical properties of earth materials. A matrix .just filling the voids between grains at maximum hydration tends to add stability to the materials under stress. This point is elucidated in subsequent sections of this paper. Eartli materials compos(>d of gravel and sand may contain no matrix or any one of several matrices (table 3). P^iiu' granular materials (silt size), however, can contain (mly argillie or amori)hous matrices. Materials having a granular matrix are quite different from mate- rials containing an argillie matrix. Similarly, a material with a ci-ystalliue matrix is quite different from material with a clastic or amcn-phous matrix. ^laterials contain- ing a granular matrix may display pro]iertics similar to those of otherwise similar materials containing a bio- fragmental matrix. Also, argillie matrices may produce properties similar to those resulting from some types of amorphous matrices. However, nuiterials containing a niontnu)rillonite-type clay mineral in the matrix will beha\-e (piite differently from materials containing the othei- Jiinds of matrix. Variations in i)i'operties will occur in materials containing montmorillonitc accord- ing to the kind of exchangeable cation present. Any type of matrix may occur intermixed with one or more ethers. However, one type is usimlly |)resent alone or strongly predominates over others. In cei'tain materials in which the grains or particles are not in contact, the fabric is supported by inter- granular braces. This type of matrix is of considerable engineering importance. In some loess, granular com- ponents are widely separated by intergranular braces comjiosed of vei\v fine silt intermixed with beidellite or montmorillonite (fig. 8). This fabric is characteristic of one type of loess in which the strength is high when the material is dry, but with wetting the fabric col- lapses and a large consolidation takes ])lace. Intergranu- lar braces may also be composed of argillie or amorphous material. The medium, fine, and very fine argillie materials may contain an argillie or an amorphous matrix. For ex- ample, in the lone formation near Friant, California, sand-size argillie aggregations composed of kaolinite are bound together by a matrix of kaolinite (fig. 9). Study of argillie materials from many areas may reveal that clastic argillie aggregations cemented by argillie or -ii^ .« lllilc line aiiiillic Yangtze Gferred. In petrofabrics, the synnnetry of the fabric is also deter- mined and further work may justify more definitive designation of orientation in clastic fabrics. By com- plete study of the orientation and symmetry of the fabric of earth materials and rock, better static and dy- namic test procedures are being developed and a better nndei'standing of the mass properties of the earth ma- terial inevitablv will result. FlGTRE 11. Mcdiiiin anil fine ai'gillic labrit. Jii'idt'llitf clu.ystunf, Porterville cla.v, near Lindsay, California. Dark areas and aggre- gations of beidellite cemented mainly by argillie matrix (thin, white, irregular streaks between dark aggregations). Magnifica- tion .50x. 204 Clays and Clay Technology I Bull. 169 "When the elements of an earth material are arranged in such a manner that no repetitive pattern is observ- able, the fabric is said to possess a i-andom orientation. "When the elements are arranged in a repetitive pattern, the fabric possesses a preferred orientation. Random orientation is well illustrated by the Navajo sandstone, Crlen Canyon Dam site, Arizona (fig. 2). A detailed study of the fabric of this rock probably would reveal some type of orientation. Preferred orientation, in which all of the elements are aligned with their longest axes parallel to each other, is observed in many shales. As viewed with polarized light, the whole fabric acts more or less as an anisotropic crystal. This type of orienta- tion is illustrated by an illite shale from near the Yangtze River Gorge, China (fig. 10). Materials with a preferred orientation are more likely to fail at lower shear values than are similar materials in which the orientation is random. Some degree of preferi-ed orientation of the constituent particles may in turn orient the pore spaces in an earth material in such a manner that directional permeability and porosity result, as is typical of loess. The determination of direc- tional permeability in sands is of importance in the pri- mary and secondary recoveiy of oil. Cementation. A great variety of mineral substances may occur as cements in clastic earth materials. Cement- ing substances can be classified into two general cate- gories: (1) reversible, and (2) irreversible. A cement in an earth material is said to be reversible if the prop- erties of the material changed by manipulation and de- hj'dration are regained with wetting and remolding, and restoration of the original conditions of temperature and pressure. Otherwise the cement is said to be irreversible. Common cements found in earth materials are: Reversible Irreversible Clay ("arbonates (calcite and dolomite) Soluble salts Iron oxides Water Aluminum oxides Ice Silica Organic matter Any one of these cementing substances may occur as the main cement, but commonly both reversible and ir- reversible cements occur together in almost any com- bination. Thus, water and clay, caleite and clay, or "water, caleite, and silica, may occur together. Loose, clean sands are essentially cohesionless when dry ; con- versely, when water is present, the sand can be molded to any desired shape, and the molded shape will be retained with drying. However, only a minute force is sufficient to destro.y the molded form. When clay min- erals (except halloysite, 4H2O) are added to the water- sand system, the material is easily molded, and when dry, the molded form possesses many times the cohesion of the water-sand mixture. Both these clay minerals and the water act essentially as reversible cements; the dry material can be rewetted, remolded, dried, and the mate- rials will develop cohesion again. When siibstances like gelatinous silica and hydrated iron and aluminum oxides are precipitated in the soil in amounts large enough to constitute the major cement- ing material, the soil may be molded to a desired shape if not dried. Diiring drying the material retains its shape and is resistant to shearing stresses, but when broken the material resists remolding. The cohesion of the mass may or may not be increased significantly by the addition or presence of these irreversible cements, depending iipon the type of irreversible cementing mate- rial, degree of induration, and the amount of water present. Thus, dry heavy clays may be as strong as or stronger than some earth materials cemented with cal- cium carbonate or hydrated oxides. Cementation may occur in various degrees. Grains may be cemented together at the points of contact (fig. 2), or the interstices may be partially or completely filled with cementing matrix. Claj's can be cemented by various cementing materials, but in contrast to granular materials they possess a natural ability to cohere. Cohe- sion between particles of claj' minerals maj' be high, even while adhesion between the clay mineral and granular constituents is relatively weak. The cementing materials may be deposited with the grains or thej- may be introduced later by meteoric or hypogene water. The interested reader is referred to various articles on the origin and mineralogy of cement- ing materials (Waldschmidt, 1941; Johnson, 1920; Frye and Swine ford, 1946). Particle Form. The particle forms in an earth mate- rial are quite diverse and difficult to evaluate. The ma- jority of the components of earth materials can be said to be irregular in shape, but some grains, such as zircon, hornblende, some quartz crystals, sillimanite, and dia- toms, possess regular shapes. In order to describe parti- cle shape, the form of the components of an earth mate- rial ma.v be compared with regular geometrical foi-ms as follows : Equant — Dimensions in all three mutually perpendicular direc- tions are equal or the dimension in one direction is less than I3 times that o£ the other directions Tabular — The length and width of the particle are more than lA times the thickness Rod-like — The length of the particle is more than 1 J times the width or thickness Irregular — Xo geometrical form More specific detailed geometical forms can be used as a basis for description of particular shapes. Thus, equant could be subdivided into spherical, cubical, and octa- hedral shapes ; tabular into disk-like or platy ; and rod- like into jirismatic, cylindrical, or bladed. In order to express the shape of particles numerically, the concept of sphericity of particles is used in sedi- mentary petrography (Pettijohn, 1949) and can be introduced in more detailed investigations of earth ma- terials. Sphericity is a measure of the degree to which a particle approaches a spherical form. Sphericity is expressed as the ratio of the diameters of the grains. If the grain diameters are a, b, and c (length, width, and thickness, respectively) four ditt'erent shape classes; namely, disks, spheroids, blades, and rods can be dis- tinguished. Rounding, in contradistinction to sphericity, is a meas- ure of the angularity of the edges and corners of particles. Roundness is expressed as the ratio of the summation of radii of the individual corners of the particle, divided by the number of corners, to the radius of the maximum inscribed circle. Depending upon the degree of rounding of the corners and edges, particles can be classified as rounded, subrounded, subangular, and angular. Part VI L'lav Tkciixology IX Soil Meciianic-s 205 The shape of the particles materially affects some of the physical properties of earth materials. Fraser (1935) has shown how porosity is affected by chanjre of irrain shape in sands whose gradation is the same. Permeability is also dependent to some dep'ree upon the shape and de^ee of rounding of the particles, as well as other aspects of fabric, such as size, packing, sorting, and continuity of voids. Materials containing diatoms exhibit high compressibility because of their peculiar shape. Diatom tests are exceedinglj- thin, as compared to tlieir length and width, and in a sense act as platy units, much like the micaceous minerals, except that their strength and elasticity are not so great as those of mica. The degree of rounding of the granular components affects the stability of a loose sand. The greater the angularity, the more the components tend to interlock and thus to increase internal friction and the angle of repose of the material. However, because of irregular distribution of stress, angular grains tend to fracture and granulate more readily than do similar materials composed of rounded grains. MINERALOGIC COMPOSITION Granular Components The grains (granular components) of earth materials are the hard particles of rocks, minerals, and organic structures which act as individual units under stress. Such components are best illustrated by quartz, feld- spar, calcite. dolomite, the hard remains of diatoms, radiolaria. mollusks. and such rock types as granite, rhyolite, basalt, volcanic ash, limestone, sandstone, schi.st. and gneiss. These components, when predom- inant, form the fundamental framework or skeleton of earth materials. By means of this framework, pressure is transmitted from grain to grain throughout the mass. When they are tlie oidy components present, the material is a loose gravel, sand, or silt. Granular components also almost always are present in argillic clastic materials. Granular mineral and rock fragments are derived from igneous, sedimentary, and metamorphic rock by chemical and mechanical weathering. They commonly reiiresent the rocks and minerals most resistant to chem- ical and mechanical weathering. Quartz is the dominant granular component of most soils ; feldspars are wide- spread ; rock particles almost always are present, and may dominate, as in areas underlain by fine-grained rocks. TTigh content of feldspars commonly is found in residual soils where feldspar predominates over quartz in the original bedrock. In areas of dry or cold climate, feldspars remain abundant through long distances of transport by streams, as for example in the sands and gravels of the Great Plains states, which were derived from granitic rocks and gneisses of the Rocky Moun- tains. The minerals commonly observed in soils as .sand- and silt-size particles are (in order of decreasing abundance) : t. Quartz 11. Olivine 2. K-feldspars 12. Sillimanite 3. Na-Ca-felclspars 1?.. Zircon 4. Calcite 14. Kutile .^. Dolomite 1">. Magnetite fi. Muscovite 10. Tourmaline 7. Biotite 17. Staiirolite .S. Chlorite IS. Garnet 0. Hornblende 19. Sphene 10. Augite 20. Apatite Calcite and dolomite are present in soils and in rocks as detrital grains more often than is generally recog- nized. In earth materials, calcite and dolomite may be present also as precipitated crystals which act as detri- tal grains, or calcite and dolomite may cement the grains together, forming concretions, caliche, or calcareous hardpan. For purposes of this classification, micaceous minerals, such as muscovite, biotite, and chlorite, are considered as granular components of earth materials when present in sand- or silt-size particles. However, when these minerals ai-c present in clay-size particles, they should be classed as clay-like because their crystal- lographie structure usually is degraded. Hence, mi- caceous minerals may be regarded as intermediate between granular minerals and argillic materials. Argillic Materials Cla]/ and Clay Minerals. The clay minerals are one of the most important constituents of earth materials and many of the properties of earth materials are in- fluenced by the identity, amount, particle size, and chem- ical composition of the clay minerals. Three large classes of clay minerals are recognized, namely : the kaolinite, montmorillonite, and illite (hydrous mica) groups. In any discussion of the mechanical components of earth materials two conflicting conceptions arise con- cerning the word "clay," one referring to particle size and one to mineralogic composition. In general, "clay" is meant to comprise the components of earth materials which are very fine-grained (say, less than 5 microns in size) and plastic when moist. More definitively, how- ever, unless mineralogieally identified, these fine-grained components should be referred to as clay size. During the past two decades, largely as the result of analysis of soils by X-ray diffraction methods, it has been de- termined that clays almost universally contain one or more hydrous aluminum and magnesium silicates which are crystalline. Plasticity is usually associated with the definition of clay, but clay minerals vary greatly in their plastic properties. Kaolinite and halloysite may exhibit mod- ei-ate or only slight plasticity, aiul halloysite commonly is nonplastic. On the other hand, montmorillonite is highly pla-stic. Illite is widely variable in plasticity. Jloreover, the term "clay" cannot be restricted justifi- ably to minerals which possess plastic properties by short-time manipulative tests, since clay minerals can acquire plastic properties after a long period of time of hydration For purposes of this paper, clay will be defined as earth materials which are A'cry fine-grained and plastic when wet, or which contain high proportions (25 per- cent or more b.y volume) of minerals of the montmoril- lonite. illite. or kaolinite groups. Clay minerals may be absent from earth materials, such as certain gravels, sands, and silts (glacial rock flour), or thej' may be almost pure, as in some clay de- po.sits. In most earth materials gravel, sand, silt, and clay minerals are mixed in various proportions. The clay minerals may occur as coatings on grains, cement- ing materials between grains, or as silt- and sand-size aggregations (fig. 11). Mineralogic composition in rela- tion to engineering properties of many .soils is reported by Ijambe and ^lartin (1953). 206 Clays and Clay Technology [Bull. 169 Crystallograpliy of Chnj Minerals. The crystallo- grapliie structure and chemical composition of clay minerals have been extensively investigated (Pauling, 1930; Gruner, 1933; Hendricks. 1936; Hofmann et al., 1933), and are disciissed in detail elsewhere in this volnme. The definitive relationships of crystallography, chemical composition, and atomic .substitution should be reviewed in order that the inter-relation of these prop- erties and engineering performance of various clay min- erals can be recognized. Suffice it to say that clay minerals may be subdivided into three main categories: (1) the micaceous or platy clay minerals; (2) the fibrous clay minerals; and (3) the amorphous clay minerals. The clay mineral species may be arranged as follows : Triori(thcih-nl ** Antigorite Chrysotile Cronstedite Chamosite Ame.site Triocfnhedfdl Saponite Heetorite Vermieulite ( jefferisite) Trioctahedral Hydrobiotite Mg-illidromiea 1. Micaceous Clay Minerals a. Kaoliriite (iroup Diociahedral * Kaolinite Halloysite. 2H=0 Hydrated halloysite, 4H»0 Nuerite Dickite b. Montmorillonite Group Diocfahedral Montmorillonite Beidellite Nontronite c. Hydrous Mica Group Diociahedral Illite Finite Glauoonite Celadonite Branimalite d. Mixed Layer Group Anauxite Bravaisite Vermiculite-clilorite Kaoliiiite-nontronite Montmorillouite-illite 2. Fibrous Clay Minerals a. Sepiolite Group Sepiolite Garnierite Pilolite b. Palygorskite Group 3. Amorphous Clay Minerals a. Allophane Because so little is known concerning fibrous clay minerals, a brief comment will be made on their miner- alogy. Fibrous clay minerals include two subgroups. One subgroup comprises minerals with a sepiolite structure, the second those with a palygorskite structure. These two minerals, palygorskite and sepiolite, have been studied by DeLapparent (1935) and Longchambon (1937). Sepiolite is a hydrous magnesium silicate, whereas palygorskite is a hydrous magnesium alumi- num silicate. The possibility that these two minerals belong to one isomorphous series is quite unlikely in view of their totally different crystallographic structure. The name " attapulgite " was proposed by DeLap- parent (1935). Later work by Bradley (1940) showed * Dioctahedral (Brindley. 19 51) clay minerals are those in which approximately two of tliree positions in the octohedral layer of the half unit cell are occupied by cations. ** Trioctahedral (Brindley, 1951) clay minerals are those in which approximately three of thi-ee positions in the octohedral laver of the half unit cell are occupied by cations. that the structure of "■ attapulgite "" is analogous to that of the amphiboles. Palygorskite with a very sim- ilar, if not identical, chemical composition and X-ray diffraction pattern was named by Fersman (1908) in 1908. Consequently, the name "palygorskite" should take precedence over "attapulgite." As shown by X-raj' diffraction analysis, the main mineral in deposits of so-called "attapulgite" in Florida and Georgia is paly- gorskite. Varying amounts of a montmorillouitc-type mineral also are present. In some occurrences, palygor- skite is described as mountain cork and mountain paper. The mineralogy and paragenetic relationships of paly- gorskite are described by Macksoud (1939). Sepiolite, often referred to as meerschaum, is of lim- ited occurrence but is apparently widespread. Garnierite is a nickeliferous member of the sepiolite group. Since the ionic radius of nickel is close to that of magnesium. nickel can substitute for magnesium in the lattice quite easily. X-ray diffraction patterns of garnierite from New Caledonia, one of the type localities, are almost identical with the X-ray diffraction patterns for sepio- lite. Pilolite or hydrous magnesium aluminum silicate from Euboea, Greece, as shown by A.S.T.il. X-ray Dif- fraction Data Card No. 11-41, also yields an X-ray diffraction pattern very similar to that of sepiolite. Other minerals probably belong to these two groups, forming a sequence analogous to the montmorillonite group. Occurrence of Clay Minerals. The formation of clay minerals is discussed in detail by Ross and Hendricks (1945), Grim (1942), and Hoskings (1940). Clay min- erals form in response to physical-chemical conditions, the determining factors being acidity or alkalinity of solutions, drainage, oxidation or reducing conditions, and the availability of chemical elements. The available elements, as determined by the original rock, dominate the early stages of alteration. For example, a tuffaceous material at Fena River Dam, Guam, contains montmoril- lonite and halloysite; the halloysite forms the matrix, whereas the montmorillonite constitutes the bulk of the embedded tuff fragments. At Pierce damsite near Singa- pore, Malaya, granite is weathered to depths in excess of 60 feet ; the feldspar crystals being converted in place to fine kaolinite and the biotite to vermieulite, the quartz remaining intact. With progress of alteration and differ- ential leaching of original elements of the parent mate- rial, the physical-chemical conditions come to dominate and similar clay mineral suites may be developed from divergent rock types (Hoskings 1940). Kaolinite in earth materials is much more common in the eastern and southern parts of the United States than in western United States. Many of the occurrences of kaolinite in western United States are of hydro- thermal origin. Kaolinite clays of high purity of sedi- mentary origin occur in parts of the Dakota and Lakota formations near Mesa Alta, New Mexico. Kaolinite is widespread as a minor constituent of many clays and shales, as in the Tongue River formation near Barnes, South Dakota, in which it constitutes 25 percent ; shale of the Benton formation. Golden, Colorado (10 percent) ; shale of the Mowry formation, Tiber Dam site, Montana (15 percent). Kaolinite clays and sandstones containing intermixed anauxite are abundant in the lone formation of California. Vermicular kaolinite and flakes of anaux- I'art VI Clay Tkciinology ix Soil ]\Ieereent) ; the Pierre shale formation near James- town, Nm-th Dakota (80 to 85 percent) ; the "Wasatch formation near Ivie, Utah (80 percent) ; the Creede for- mation near Wagon "Wheel Gap, Colorado (60 percent) ; the ilowry shale formation near Laramie, Wyoming (50 percent) ; the Eagle Ford formation near Dallas, Texas (50 ]iercent) ; and elsewhere. Jlontmorillonite-type clay minerals also occur in small proportions in many forma- tions, as in the Valley Springs formation. Valley Springs, California (3 percent) ; the ^Monterey forma- tion. California (5 to 25 percent) ; the Brule, Arikaree, and Frontier formations in Wyoming (25 to 45 percent) ; and the Ringold formation, near White Bluffs, Washington (15 to 35 percent). Nontronite is widespread in altered basalts, where it develops from palagonite, ferromagnesian minerals, and plagioclase. Space does not permit the listing of additional occur- rences of montmorillonite-type clay minerals encoim- tered in the western United States. Other occurrences of montmorillonite are cited bv Ross and Hendricks (1945) and by Kerr and Kulp (1949). Tllite-type clay minerals are fairly conunon in earth materials of the western United States. The estimation of their amoinits in earth materials is difficult as they often occur intermixed or interlayered with montmoril- lonite, vermiculite, chlorite, and other micaceous min- erals. Hlite has been observed in the Pierre formation, Boulder, Colorado (40 to 50 percent) ; the Lower Dakota formation. Carter Lake, Colorado (50 to 60 percent) ; the Morrison formation near Tenderfoot Mesa, Colorado (40 to 50 percent) ; the Carlisle formation near Cedar Bluffs, Nebraska (50 to 60 jiercent) ; the Benton formation near Golden, Colorado (40 percent) ; the Nespelem formation near Grand Coulee Dam, Washing- ton (10 to 40 percent) ; the Flathead formation near Trident, Montana (30 percent) ; and in recent glacial deposits at Eklutna, Alaska (40 to 50 pei'cent). Hlite is also observed in small amounts in the Green River formation near Rifle, Colorado (15 percent) ; the Eagle Ford formation, near Dallas, Texas (20 percent) ; and the Chadron formation, Weld County, Colorado. The occurrence of illite in manv other earth materials is de- scribed by Grim (1942). Certain earth materials contain crystallographically interlayered nunerals, such as montmorillonite and illite (Bradley, 1946); vermiculite and hydrobiotite (Gruner, 1934) ; and vermiculite and chlorite (Brindley, 1951; Bradley, 1946). Other combinations are probable. Interlayered vermiculite and chlorite are an important constituent of the glacial lake deposits of the north- western T'nited States, especially in the Nespelem forma- tion of AVashington (fig. 3). These combinations also occur in physical mixtures. It is often difficult to deter- mine whether a given clay is a physical mixture or is composed of interstratified cla.y minerals. Walker (1949) has devised a method to differentiate various clay min- erals, and the differential thermal mi'tliod is useful in identifying interlayered minerals. Crystalline Hydrated Oxides. Most of the earth mate- rials contain small amounts of crystalline hydrated iron oxides (goethite) or simple oxides (hematite) and possibl.y hydrated aluminum oxides (boehmite and gibbsite). In soils occurring in subtropical and tropical regions, the iron and aluminum oxides cpiite commonly occur together, as in laterite, or aluminum oxides may predominate as in bauxite. These compounds occur mixed in various proportions, with or without clay and detrital minerals, dejiending upon the climate, degree of weathering, type of parent rock, and transportation agent. The hydrated oxides of aluminum and iron represent the end products of rock and mineral decomposition; that is, they are the compounds which remain after the more-soluble constituents have been leached out. In general, these oxides act as cementing agents for other constituents of earth materials. Organic Matter Organic matter fomul in sedimentary rocks, water, and recent soils, usually is very complex chemically and physically, inasmuch as many factors, processes, and types of orgainsms have contributed to its formatioiL Natural carbonaceous organic matter can be classified into five main groups, namely: (1) carbohydrates; (2) proteins; (3) fats, resins, ancl waxes; (4) hydrocarbons, such as coal, asphalt, petroleum; and (5) carbon. Ben- zoic acid, vanilliiL and hydrobenzoic acid were found by Shorey (1914) in an organic hardpan in the Leon soil of Florida. In addition to the above constituents, earth materials contain living organisms and newly formed products of their metabolism. Carbohydrates, such as cellulose, hemicellulose, and lignin are the most abun- dant organic compounds in earth materials. These com- pounds are derived by decay of plant structures. Earth materials vary greatly in content of organic compounds. Desert soils contain the least amount of organic matter. Soils of humid, temperate, or cold re- gions commonly contain abundant organic matter, peat being an extreme example. Laterites contain only small amounts of organic matter. In prairie soils, organic mat- ter is usually confined to the upper 3 feet and ranges from 5 to 10 percent. Organic matter found in marine 208 Clays and Clay Technology [Bull. 169 shales may consist of petroleum, liydroearbon, carbon, and "kerogen." According to Trask (1939) only a few- marine sediments contain more than 10-percent organic matter and only a few contain less than 0.5 percent. Organic matter is most common in the finest fractions of sediments. On the average, cla}-e.y materials contain four times as much organic matter as do sands and twice as much as silty materials. Much of the organic colloids or "humus" in soils are derived mainly from plant lignins. The organic colloids vary greatly in their properties according to the parent material, climatic conditions, and state of decomposi- tion. Lignin-humus, as these organic colloids are often designated, possess outstanding cation-exchange capac- ity, whereas cellulose and hemicellulose possess small exchange capacity. Soluble Compounds Soluble compounds, such as inorganic salts, are pres- ent in many of the soils of the arid and semiarid regions of the world. In recent unconsolidated sediments, inor- ganic salts accumulate in arid and semiarid regions wherever inland or poor subsurface drainage exists or where irrigation is practiced. Soluble salts also accumu- late on low alh;vial flats along seacoasts in humid areas. Soluble salts in soils are of interest because tlieir presence affects the value of land, its potential use, the feasibility of irrigation projects, the durability of con- crete in engineering structures, the permeability of earth or bentonite linings of canals and reservoirs and soil stabilization practice. Saline and alkali .soils occupy considerable portions of certain areas and according to Zakliarov (1927), alkali soils underlie 24 percent of the area of the Fnion of Soviet Socialist Republics. They are widespread in the Dakotas and parts of Colorado, Wyoming, and eastern ]\Iontana, and in the southwestern part of the United States. Soluble constituents of earth materials originate by decomposition of other minerals or by precipitation from hypogene, connate, or saline surface waters. Through mechanical disintegration and chemical processes, such as hydration, carbonation, oxidation, reduction, and dissolution, the elements composing rock- forming minerals are made soluble and are carried to the ocean or to inland basins by surface or ground wa- ter. Some geologic formations yield soluble salts directly by leaching of soluble compounds. Many different salt compounds are present in saline and alkali soils. The most common compounds are the chlorides, carbonates, and sulfates of sodium and the sulfates and chlorides of calcium and magnesium. Cal- cium and magnesium carbonates are not significantly soluble in pure water, but they become soluble when acted upon by weak acids. In general, potassium chloride and carbonate are not common, although in certain oc- currences they may be the dominant salts. Borates and nitrates, although concentrated as crusts on soils in variotis regions, are not common in most areas. In some soils, double and triple salts form by wetting and dry- ing. Glacial soils from northern Montana and southern Alberta, when saturated with water and allowed to air dry, yield a triple salt, tentatively identified as Chile loeweite. The original soils contain gypsum (CaS04- 2H2O) which partially dissolves with wetting. Cation exchange then takes place, Na*, Mg**, and K* being re- placed in part by Ca**. The released Na*, Mg*+, and K* then combine with the sulfate radical to form the triple salt upon drying. Similarly, gypsum occurring natu- rally in sodium bentonite will effect partial cation ex- change upon wetting of the material, and impedes swell- ing. For purposes of classification, soils containing salts are designated in the United States as: (1) saline, (2) saline-alkali, and (3) nonsaline-alkali (Anon., 1947). Saline earth materials contain abundant but widely variable amounts of chlorides and sulfates of calcium and magnesium and, less commonly, nitrates of sodium. Sodium in exchangeable form is subordinate. Calcite and magnesite may also be present. These saline earth materials may be developed from normal earth mate- rials under poor drainage conditions. The excess of soluble salts maintains the claj- minerals in flocculated condition and tlie fabric of the soil is open and perme- able. The saline-alkali soils contain both soluble salts and exchangeable sodium. These saline-alkali soils are formed by both salinization and alkalization. The fabric of these soils is similar to that of saline soils so long as excess salts are present. However, when leached, the clays dis- perse and the fabric 'beeomes similar to that of nrtnsaline- alkali soils. The nonsaline-alkali-earth materials contain sodium as the main exchangeable cation; small amounts of chlo- rides, sulfates, and carbonates of calcium and magnesium are present. These earth materials develop from the saline-alkali earth materials by leaching of soluble salts or by erosion, transportation, and redeposition of mont- iiKU-illonite-rich fractions. Hydrolysis of these .soils forms sodium hydroxide (and subsecjuently sodium carbonate by carbonization) and any organic matter becomes dis- persed through the soil, producing a dark color. If large amounts of organic matter are present, the soils are referred to as "black alkali." These soils are dense and possess a massive fabric. Amorphous Constituents Tile aniorplious compounds in earth materials are mainly hydrated oxides of iron, manganese, aluminium, and silicon ; the hydrous aluminum silicate, allophane ; volcanic glass ; the altered basic glass, palagonite ; and, as mentioned above, certain organic compnunds. Amor- phous phosjihate compounds are rare. The amorphous hydrated oxides are present in earth materials in small amounts. They generally occur inter- mixed with clay at seams, as coatings on grains, and in intergranular voids. Allophane also occurs in small amounts in earth materials, but, because it is difficult to identify, allophane is rarely reported. Volcanic glass is present in many of the earth mate- rials in the westena United States, ranging from only a trace to virtually the entirety of the nuiterials, as in pumicite. As these glasses are very susceptible to chemi- cal alteration, montmorillonite is commonly present. Kaolinite is the less-common ai'gillic alteration product in the western United States. Palagonite is an indefinite hydrous mineraloid containing MgO, FeO, FcoOj, AUOa, and SiOi, which alters readily to montmorillonite, non- tronite, sapouite, or kaolinite minerals, depending upon the environmental conditions. Basaltic glass aiul palago- nite are widespread in certain areas such as the Pacific Tart V ("lay Tkciinology IX Soil ]\Ie(Iianics 209 Xortlnvost ; nuntrunito is ul)i(iuit()us in association with palagoiiite, sapouite occurs sparsely. In the Hawaiian Islands palagonite tvpicallv alters to kaolinite (Dean, 1947). SURFACE CHEMICAL PHENOMENA Adsorption Adsor])tion is tiie ehanfre in coneontration ocourring at the interface or phase boundary between solids and lirpiids, solids and gases, and li(|nids and gases, with or without eheniieal reaction. To be included are such interrelated phenomena as adsorption and exchange of molecules, cations, and anions. Inorganic and nonpolar or polar organic compounds, take part in the three adsorption processes. Adsorption may be the result of coulombic or gravitational (van der Waals') foi-ces. The electrical unbalance (coulombict causing ad.sorption arises in ]5art thi-ough partial or conii)lete substitution in a crystal of ions of differing valency, at broken elec- trostatic bonds, at areas of polarization, and by chenii- sorption upon accessible hydroxyl groups or other ions at the surface. In the clay minerals, adsorption may take place at the edges and side surfaces and between the clay packets, as in expanding lattice minerals like montmorillonite. The clay minerals act as adsorbents for liquids, gases, and suspended solids in liquids. Only limited data are available on the adsorption of gases by clays, but Nel- son and Hendricks (1943) have shown that inert gases are adsorbed only on external surfaces of clay crystals. Abundant data are available, however, on the adsorption of liipiids by clays. Molecular Adsorption. The adsorption of water mole- cules by clay minerals of the montmorillonite group results in an increase in the c-axis spacing withoiit any apj)reciable change in dimensions of the a- or b-axis. This ])henomenon was first studied crystalldgraiihically by ilofmann, Eudell, and Wilm (19:13) and by Ilof- mann and Bilke (1936) who measured the variation of the c-asis spacing with different vapor pressures. Brad- ley, Grim, and Clark (1937) postulated that H-mont- morillonite forms definite hydrates which contain fi to 24 ILO molecules per unit cell. Hendricks and Jeffer- son (1938) suggested that different degrees of oi'derli- ne.ss existed in the water held between the packets of the crystal and that the greatest degree of orderliness is attained in H-montmorillonite. The adsorbed water molecules are thought to assume a hexagonal arrange- ment, analogous to the crystallographic structure of ice, and to develop one, two, three, or four layers with in- creasing vapor pressure. The spacing along the c-axis increases stepwise with formation of each new water laver (Bradley, et al., 1937 ; Hendricks and Jefferson, 1938). The exchangeable cation in a clay mineral of the ex- panding lattice type determines the amovuit of water vapor taken up under equilibrium conditions (table 4"). In 1940, Hendricks, Nelson, and Alexander (1940) demonstrated that the water adsorbed at low relative humidity by Ca-, Mg-, and Li-montmorillonite hydrates the exchangeable cation which lies in the interpacket space on the (001) plane. Ca and Mg cations are hy- drated by 6 molecules of water, whereas Li* is hydrated by 3 molecules of water. Other cations, such as Na*. K*. and IP, apparently are not hydrated. Development of Table -}. Effect of exchaniituhle cation on adsorption of icater by Mississippi bentonite." Exchangeable Equilibrium water content (g/g dry clay) rehitivo hu- midity of atmosphere (percent) cation .5 50 90 Na+ 0.025 0.120 0.28 K* .015 .100 .20 H* .045 .190 .36 Ca++ .070 .22 .36 Mgt* .08.5 .22 .34 '"Prom Hendricks. Nelson, and Alexiinder (1940). the hexagonal water net is eomiJeted only after hydra- tion of available Ca**. Mg**, and Li* in exchange ])osi- tions. At relative humidity in the range of 5 to f)() per- cent, Na- and K-montmorillcnite adsorb less water from the vapor phase than do Ca- and Mg-montmorillonite. However, Na-montmorillonite adsorbs more water from the liquid state than do Ca- or Mg-moutmorillonite. The significance of differing adsorptivity of clay minerals containing the common exchangeable cations is discussed in greater detail in the section dealing with consistency and the compressive strengsth of clays. The uptake of polar organic molecules and the for- mation of complexes by montmorillonite and halloysite have been investigated by Ilofmann and associates (1934), Bradley (194.5), and MacEwan (1944). Brad- ley (1945) has shown that glycols and glycol ethers produce two molecular layers between the montmoril- lonite micelles. Both Bradley and ^MacEwan (1944) have shown that hydrated halloysite forms complexes with polar organic compounds, such as the mono- and dihydi-ic alcohols, glycol, and many others. Vermiculite (Walker, 1950) also forms complexes with glycol, the glycol forming single molecular layers between the packets. Dyal and Hendricks (1950) have devised a gravimetric method whereby the amount of minerals having the montmorillonite type of expanding lattice can be estimated by the amount of ethylene glycol re- tained. Cationic Adsorption. Either organic or inorganic cations may be adsoi-bed by clay minerals. If the ad- sorptive process is reversible, that is, if an equivalent number of previously adsorbed cations are released into solution while the introduced cations are being adsorbed, tlie term cationic exchange is used. All clay minerals Table .7. Cation exchange capacity of clays." Mineral Capacity (me. per 100 g) 60 to 160 20 to 40 2 to 15 6 to 70 25 to 30 65 to 146 •' Aftpr r.rini (1!I42), B.irsliad (lii4S). and Lewis (1950). 210 Clays and Clay Technology [Bull. 169 possess a cation exchange capacity. Init tlie capacity varies sreatly (table 5) due to ditfcrcnces in crystal- lographic structure, differing atomic substitution, par- tiele-size distribution, and the location in the crystal lattice at which atomic substitution occurs, that is, in the octahedral in contrast to the tetrahedral portions of the lattice. As onl.y limited substitution occurs in the crystal lattice of kaolinite and because kaolinite clay minerals are comparatively coarse, cation exchange ca- pacity is low. Spiel (1940) has shown that the cationic exchange capacity of kaolinite increases with fine grind- ing. This increase most likely arises in the greater num- ber of broken bonds and possibly in partial destruction of the clay lattice. The exchange positions for kaolinite are located mainlj' on the external surfaces. Atomic substitution in the lattice is much greater in montmorillonite than in kaolinite minerals. This, to- gether with characteristically very small particle size, produces high exchange capacity in montmorillonite. Fine grinding does not increase the cation exchange ca- pacity. Illite is intermediate in exchange capacity. The over-all substitution in the lattice structure of illite is as in montmorillonite, but differs in that the atomic sub- stitution is mainly in the tetrahedral position. The sur- face charge is compensated by potassium, which has such an ionic radius that it fits closely into the surface on the 001 plane of opposing packets and ties the packets together. Consequently, potassium is more or less fixed in illite, as in the micas. Beidellite appears to be intermediate between illite and montmorillonite. Vermiculite is reported by Barshad (1948) also to pos- sess a high cation exchange capacity (table 5). Investigations by Jenny and associates (1932; 1936) of release of exchangeable cations from beidellite follow- ing treatment of monovalent homoionic modifications with an ammonium salt and of divalent homoionic modifications with a potassium salt, indicate the rela- tive ease of release of exchangeable cations to be Li*>Xa*>NH4*>K">Mg**>Ca"*>H*. Thus, the ca- tions to the right of the series such as Ca^"^ and H% are more strongly held by the clay mineral and have a greater replacing power than do the cations to the left in the series. The relative replacing power of various cations then is seen to be H*>Ca*+>Mg^+>K*>NH/> Na*>Li*. If a clay mineral is suspended in a solution containing ecpiivalent concentrations of sodium and potassium ions, the potassium ions will be adsorbed in the greater amounts. Based upon treatment of an XH^-montmorillouite, Schaehtschabel (1940) concluded that the relative ease of replacement or release of NH.i* from NH4-montmoril- lonite at various concentrations of replacing cations is : me/100 ml of icater 10 Li+- in discussion, properties and the related perf'onnanco of clays and shales are discussed sinml- tanconsly. STATIC PROPERTIES OF EARTH MATERIALS Particle-size Distribution Clays are composed ])re(loniinantly of particles less than 20 microns in diameter, but sand, silt, gravel, and fragments of organic remains in varied proportions al- ways ai-e present. In soil mechanics analyses for engi- neering purposes, various limiting sizes for gravel, sand, silt, and clay are used by recognized engineering organ- izations (Burmister. 1949; Terzaghi and Peek, 1948). This leads to some variation in nomenclature of earth materials. Particles classified hy mechanical analysis involving sieving and ditferential settling include individual min- eral grains, organic matter, rock fragments and pebbles, concretions, coagulated clusters of grains, and fragments of ceraentitious materials released b.v disaggregation of the sample. The fractions including particles larger than 5 mieroiis are composed mainly of: (1) rock fragments; (2) terrigenous and secondary granular minerals, such as quartz, carbonates, feldspars, jiyroxenes, and amphi- boles; (3) micas and micaceous minerals; (4) organic remains, including plant fibers and hunnis; (5) concre- tions; and (6) aggregations of grains and argillic ma- terials. The fractions containing particles less than 5 microns in size are composed of: (1) clay minerals; (2) coagu- lated clusters of grains; (3) organic matter; and (4) very finely divided particles of rock aiul minerals, in- cluding fragments of cementitious substances. Clay min- erals occur primarily as ]iarticles less than 2 microns in size and they can occur in particles as small as 0.1 milli- micron, approaching the dimensions of the unit cell (Grim, 1942). However, crystals of clay minerals com- monly occur in sizes larger than 5 microns and at least as large as 100 microns in undisturbed materials. Or- ganic materials constitute particles down to 0.1 micron in size. Silica is common as (piartz and cristobalite in particles at least as small as 0.1 micron (Grim, 1942; Soveri. 19.")0). Feldspars are abundant in sizes coarser than 2.0 microns, although they are found in sizes smaller than 0.2 micron. Ilydrated oxides of iron and aluminum can be detected in particles as small as 1 micron. Calcite is commonly abundant in fractions as small as 2 microns. The particle-size distribution of clay is susceptible to considerable change, depending upon the ph.vsical-chem- ical properties of the cla.v minerals (table 6), the me- chanical processes involved in preparation of the sample for anal.vsis (Anon., 1941), agitation of the suspension during ditferential settling procedures (Wintermyer, 1948), and the kind and amount of dispersing agent used. Particle-size distribution indicated by most proce- dures of size analysis demonstrates the response of the Table 6. Mineralogic contposition of size fractions of days." Minerals in order of quantity Particle size. microns PredominatitiK Common Rare constituents constituents constituents <0.1 Montmorillonite Mica Illite Beidellit* intermediates (traces) 0.1-0.2 Mica Kaolinite Illite intermediates Montmorillonite Quartz (traces) 0.2-2.0 Kaolinite lUite Quartz Mica Montmorillonite intermediates Feldspars Micas Halloysite 2.0-11.0 Micas Quartz Halloysite lUitca Kaolinite (traces) Feldspars Montmorillonite (traces) •iVfter Soveri (l'J50). material to a mechanical and a chemical process, rather than the particle-size distribution of the original ma- terial. Cla.v minerals of moutmorilloniti' type, especially with sodium as the exchangeable cation, are particularly prone to cleave and disintegrate along cleavage surfaces during agitation in water. Consequently, particle-size analysis made by ditferential settling, almost alwa^'S in- dicates smaller sizes than actually are present in the natural clay. Montmorillonite minerals segregate into fractions less than 0.08 micron during fractionation of clays and shales (Pennington and .Jackson, 1947). Cal- cium, magnesium, potassium, and hydrogen montmoril- lonites are not so dispersible, especially after drying, as are sodium montmorillonites so that apparent particle size will be greater if cations other than sodium predom- inate (Winterkorn and Moorman, 1941). Also, ad.sorp- tion of organic molecules may inhibit dispersion of clay particles. Because of their variable nature as the result of both chemical composition and exchangeable ions, illites vary in tendency to disperse. Some illites readily disperse to jiarticles less than 0.1 micron, but illite connnonly re- mains as particles greater than 0.2 micron and is abun- dant only in sizes greater than 2 microns. Mica is com- mon in the range 0.2 to 2.0 microns ; the degraded micas transitional to clay minerals are common even in sizes less than 0.1 micron. Kaolinite tends to concentrate in the size range 0.2 to 2.0 microns. Many kaolins are difficult to dispense to particles as small as 1 micron. Yet individual kaolins release kaolinite into fractions as small as 0.08 micron. In a detailed engineering investigation of soils of the southeastern portion of the United States, Havens, Young, and Baker (1948) have demon.strated that, in soils containing both kaolinite and illite, illite typically becomes the more concentrated in the finest fractions. Dehydrated halloysite and allophane resist disaggrega- tion and dispersion more than do the other clay minerals. The primary cause of differing contribution of the sev- eral claj'- minerals to soil mechanics properties lies in the facility with which thej' are dispersed to particles less than 0.05 micron. The proportion of the fraction less 214 Clays and Clay Technology [Bull. 169 - Ill III 1 ' "30 01 • - Si - r\ c o20 a. . \ " \\ \ >- o z UJ 1 \ \ - o / \ llJ £'0 K\ \ Montmonllonite - ^V-Uuortz -/!!/V^,.i^/ N/V Illite ond Hydrous Mica - ,„„- Feldspors^::^ — i 1 I omite \^^^M,3e M.nerols r^-^-^ J 1 10 5.0 PARTICLE SIZE 2.0 ( Microns) Figure 12. Particle size distribution and mineralogie composition of sodium montmorillonite clay, near Yuma, Arizona. PARTICLE SIZE (Microns) Figure 14. Particle size distribnlion and mineralogie composition of tlic Eagle Ford shale, near Dallas, Texas. - 1 - - - - - - /MontmorillonitA J /lllite and\ \ - y/^ / Hydrous Mica\ \ /^^iJ^Quartz y^ \ \ \\ f -^-"""^ Calcite \ \ \ y^ \\ V ^ ,Misc Minerols,„„— — -....^^ \\ / __- Dolomite'""'\^ W. ^/----''Feldsparsv ^ ^CS^ 10 50 2 1.0 PARTICLE SIZE (Microns) Figure 13. Particle size distribution and mineralogie composition of calcium montmorillonite clay, near Tuma, Arizona. 10 5 2 10 PARTICLE SIZE (Microns) Figure 15. Particle size distribution and mineralogie composition of the Porterville clay, near Lindsay, California. than O.Oo micron decreases from montmorillonite to illite to halloysite to kaolinite (Marshall, 1949). Of all these minerals, the particle-size distribution of montmorillonite clays is controlled most strongly by the exchangeable cation. The relation of mineralogie composition to particle- size distribution is indicated by X-ray diffraction and microscopical analj-sis of separated fractions in accord- ance with procedures used by Grim (1949) and by Winkler (1938; 1949) using a minimum of agitation (figs. 12, 13, 14, 15, 16, and 17). No separation was attempted on size fractions below 1 micron. The cut-oft' point at 0.12 micron is arbitrary. The proportion of ma- terial in any size fraction can be determined as a ratio of the area between vertical lines drawn at the i;pper and lower limit of the fraction to the total area beneath the uppermost curve. Two clays from near Yuma, Arizona, are of interest (figs. 12 and 13) because they illustrate the relative eft'ects of exchangeable sodium and calcium in facilitating dispersion into the finest frac- tions. Granular components, including quartz, calcite, dolomite, and feldspars, are abundant onh- above 1 micron. With the sodium clay, illite becomes relatively more abundant in the less than 1 micron fraction. The Eagle Ford shale (fig. 14) illustrates a complex clay miueral composition, including montmorillonite, illite or hydrous mica, and kaolinite. The bulk of the clay minerals is concentrated in the size range from 20 to 2.0 microns, the kaolinite disappearing and the montmorillonite and illite-ly-drous mica clavs increasing Part V] f"i,AY Technology ix Soil ilEciiAxirs 215 - 1 1 I III 1 Mont morii Ion ite 1 - ^30 5* • * - m L- Illite ond Hydrous Mico - A - j \ \L Vermiculite-Chlorite - c « - 1 A 1 - > // / Quarlz \ \u Ul o lii £io // /colcite^^ \\\ - 1 1// A \\\ - fs) // /FeldsporsN \ v| 1 ^ //Misc Actinolite. Minerols. ' ^s^ ■* 00 50 20 10 50 20 10 0.5 0.2 0.1 PARTICLE SIZE (Microns) Figure 1G. Particle size distribution and mincralogic composi- tion of varvod clay and silt, Nesiiplpm fonnnlion. near Coulee ("ity, AVasliington. Koolinite Biotfte ond Hydrobiotite 10 5-0 2.0 1.0 PARTICLE SIZE (Microns) t'KiURE 17. Particle size distribution and mineralosic composition of kaolin. lone formation. Valley Springs, California. in proportion in the less than 1 mii-roii fraction. Such a gradation produces a dense fabric capable of excessive pressure development if expansive clays are present. The analysis of the Porterville clay (fig. 15) demon- strates the presence of the well-developed, aggregated, argillic fabric which is evident petrographieally (fig. 6 and 11). The montmorillonite-type mineral (beidellite) is concentrated in the coarse fraction (44 to 74 microns) and in the less than 7 micron size fraction. The coarsest fractions represent water-stable argillic aggregations whicli break down progressively witli agitation and dis- appear entirely with usual metliods of mechanical or hydrometer analysis. For example, in repeated hydrom- eter analyses on the sample, effecting mechanical disper- sion merely by .soaiving and inversion of the hydrometer tube, the analysis of the coarse fractions is seen to vary as follows : .[iiiniiiit (percent hi) weight) Paificle siz e (microns) Trial 1 Trial 2 Trial 3 Trial /, 74 50 37 to to to ->o .37 10 - 13.R - 4.4 oo ■> 10.2 ."1.2 is.s 8.4 5.4 1S.0 1.6 4.4 17.4 t'(nisc(iuently, tlie form of the gradation curve i'or tliis material varies remarkably with manipulation of the sample. The fraction less than 1 micrcm is composed almost completely of beidellite. Kaolinite occurs in very small pr()])ortions in all size fractions. The Xes]ieU>m varved clay and silt (fig. Ki) is com- posed of intermixed montmorillonite, illite, and hydrous mica and a micaceous clay-like mineral intermediate between vermiculite and chlorite. Minor aggregation is indicated in the fraction greater than 44 microns; very small amounts of material occur in fractions less than 3.5 microns. Calcium and magnesium are the main ex- changeable cations. Analyses of sam]>les of Xespelem silt are reported by Grim (11)49). The Valley Springs clay, a member of the lone for- mation of Eocene age, widely exploited in the footliills of the Sierra Nevada is a ceramic clay composed pre- dominantly of kaolinite (fig. 17). The kaolinite M-as derived by weathering and erosion of the Sierran grano- diorite and was redeposited together with fine sand and silt in small lakes. The kaolinite is concentrated in the less than 5 micron fraction and most is less tliau 1 micron in size. Biotite and hydrobiotite show a similar distribution. The relative amount of hydrobiotite in- creases with decrease in particle size. Additional analyses of this type for clays containing lialloysite and allophane are published by Grim (1949). In experiments with a beidellite clay (Putiuxm clay) involving replacement of tlie natural cations by hydro- gen, sodium, ]iotassiinH, magnesium, calcium, and aluminum, Winterkorn and Moorman (1941) found variations in apparent size fractions as follows : Size fraction H Na K Mg Ca Al 33.0 48.0 45.0 47.0 40.0 61.0 44.0 49.0 45.0 44.0 4S.0 21.0 25.0 22.0 23.0 Silt (0.05-0.O0.5mm) 46.0 Clay (<0.00.5mm) 48.0 Colloids (<0.001mm) 26.0 The apparently greater content of particles of col- loidal dimensions in the sodium-treated soil is out- standing. The relationship is an indication of both greater dispersion and greater adsorption of water by the particles. Adsor|)tion of water by tlie particles ef- fectively reduces the specific gi'avity and increases the dimensions of each pai'ticle. In independent work on the Putnam clay, Marshall (1949) found that the quantity of material in the size range below 50 millimicrons was distinctly greater for the clay treated with lithium, sodium, and potassium, than for 'the clay treated with magnesium, calcium, barium, or hydrogen. "Wyoming bentcnite showed only slight variation in this regard after similar treatment. This difference in behavior of beidellite and montmoril- lonite seemingly relates to the site of the electrostatic deficiency leading to the adsorption of the external cat- ions. Differences in apparent particle size probably will be found also upon detailed investigation of the effects of replacements of anions. For example, adsoi'ption of 216 Clays and Clay Technology Table 7. Average size composition of clay samples.' [Bull. 169 No. of saniiiles Grade limits (mm) — composition (percent) Mineral > 1/256 1/256- 1/512 1/512- 1/1024 1 1024- 1/2048 < 1/2048 Kaolinite ,. 20 4 4 5 1 2 2 27.0 49.7 84.9 39.7 21.6 31.3 25.8 10.8 8.2 9.0 14.6 14.6 20.0 4.9 13.4 6.3 2.2'' 14.1 15.8 17.1 18.8 12.3 4.5 I.IK 7.0 11.2 5.4 5.2 36.5 Hallovsite- , .- . - 31.3 Dickite -. ... 2.8' 24.6 36.8 Illite 26.2 45.3 "From California Researcii Corporation (1950). ■^ Average of tiiree samples. the formic anion by moutmorillonite markedly decreases dispersion and increases particle size tlirough aggre- gation. Studies of particle-size distribution by the pipette method of samples of the important clay minerals from many localities (table 7) were correlated by the Cali- fornia Research Corporation (1950) with electron mi- ei'oseope observations. The particle size typically was indicated to be smaller by the electron microscope than was indicated by settling velocities, probably because of aggregation of grains in the suspension. Badly needed is a method of particle-size analyses which will reveal the size of the grains or masses which act as structural units of the fabric during volume change or strain. A similar need already has been recog- nized in agriculture (Yoder. 1936). This is a property of concern to the engineer because he wishes to know the manner in which the material will respond to stress, to entrance or drainage of water, or to other actions. Unlike the geologists or pedologist, he is not concerned with the process by which the material has formed ; hence, in an analysis of earth materials as a basis for engineering interpretation of the probable performance of the soils in place or in fill, determination of particle- size distribution by rigorous mechanical and chemical treatment and excesses of water will produce misleading results. I'liit Weight and Void Bafio. Unit weight of eai'th materials is detei'mined by the specific gravity of the solids and fliiids composing the material, and their rela- tive abundance. The ratio of the volume of fluids (typi- cally water and air) to the total vohime is the porosity. A more common factor in soil mechanics is the void ratio, representing the ratio of the volume of the void to the volume of the solid constituents in the soil. Porosity and void ratio, and to a lesser extent unit weight, are expressions of the fabric of the material. Porosity and void ratio are greater (1) with departure from a well-graded condition, either through irregu- larities in gradation or because of concentration of ex- ceptionally uniform sizes, and (2) with the introduction of fractures as the result of shrinkage or jointing. Porosity and void ratio are less if the soil has been impregnated with secondary deposits of mineral matter, either by infiltration of colloidal material or precipita- tion of substances from solution. Except for infiltration or precipitation, porosity and void ratio arise in the grain-to-grain relationships established as the material accumulated in past geologic time or during construction of fill and modified by consolidation or shear. As has been indicated previously, clay minerals are of prime significance in the development of the fabric of soils primarily because of their adhesive, cohesive, and plastic qualitj' and their capacity for volume change with changing moisture content. For example, the highly porous structure of loess would be completely unstable and incapable of preservation if rims of montmorillonite or illite with intermediate moisture content were not present upon grains of silt and fine sand to act as inter- granular binding at time of deposition. Reworked hal- loysite soils typically are very low in unit weight and high in void ratio possibly because of the tubular form of the ultimate particles of halloysite (Bates et al., 1950). For example, a halloysite soil proposed for use as fill in an earth dam being constructed by the city of Nairobi, Kenya, East Africa, possesses a natural unit weight of 50 pounds per cubic foot and optimum dry density of only 73 pounds per cubic foot with corre- sponding void ratio of 1.8 and 1.3, respectively. Similar values were obtained on halloysite soil used in construc- tion of Fena River Dam, Island of Guam. Investigations by Johnson and Davidson (1947) dem- onstrate the influence of exchangeable ion and clay mineral type upon moisture-density cuiwes of kaolinite and moutmorillonite soils. The density of the sodium- kaolinite soil is higher than that of the calcium-kaolinite soil, probably because the greater dispersion of the former permits development of a denser fabric at given compactive eifort. Sodium-montmorillouite soil achieves a somewhat lower density at slightly higher moisture content than does kaolinite soil, presumably because adsorption of water by the moutmorillonite reduces plas- ticity of the soil and swells the clay fractions. As will be subsequently discussed, porosity and voids ratio of earth materials are changed with loading and consolidation of the material. This is true at high mois- ture content particularly. The degree of change effected by loading depends upon the gradation of the soil com- ponents, the swelling or shrinking properties of the clay mineral component, the amount and rate of loading, and Part VI Ci.AY Tfx'hnology IX Soil ;Mkciiaxics •217 tlio (Icfrroe of water resi.staiK'»> of tlu' iiulividual particles or ar^illic ag-rregations eonstitutiiip the soil. If the iiiiit load is maintained constant, the rate of chanjze in po- rosity, voids ratio, and unit weight depends primarily upon the permeability of the soil. Specific Gravity. Specific gravity of eartli materials is important to engineering considerations primarily be- cause it is used in calculation of other important factors, such as void ratio and degree of saturation. In stand- ard tests for speeifie gravity of soils, as for example A.S.T.M. Designation: D S.'ii (1949). the basic assump- tion is made that the solid particles of the sample possess tlie same volume in air as in water and that no chemical or physical-chemical reaction takes place with the li(iuid used in the pyenometer. As is manifest from the discussion of adsorption, clay minerals which are partially or completely dehydrated will adsorb water added to the pyenometer during the determination of specific gravity. With adsorption of the water and its consequent in- crease in specific gravity, additional water must be added to fill the pyenometer. The calculated volume of the solids consequently is lessened and the determined specific gravity is increased. This error can be corrected by use oC a nonpolar liquid instead of water. Compara- tive values for two samples of Colorado shale from Tiber Dam site containing high proportions of sodium-mont- morillonite are as follows : Specific (pa fit!/ Specific gravHij Sample \o. (using tetrahydro naphthalene) (using water) nF-263 2.683 llF-264 2.T02 2.72 The diiferenee found in specific gravity using polar in contrast to nonpolar liquids will vary with the kind and amount of clay mineral, the exchangeable cation, the ratio of liquid to solid used, and the particle size of the clay minerals. The effect should be significant only for montmorillonite group and illite clays. Comparatively small changes in specific gravity pro- duce significant differences in the calculated value of void ratio and degree of saturation of earth materials. Using the data for Sample No. llF-263, above, the void ratio is calculated to be 0.36 if the specific gravity is taken as 2.72 ; whereas, the void ratio is 0.35 if the specific gravity is taken as 2.683. "With a moisture con- tent of 19.8 percent (the observed natural moisture con- tent), the first result woiild indicate the sample to be 97.1-percent saturated, whereas the second indicates complete saturation. Actually, however, because of the great adsorptivity of sodium-montmorillonite, virtually all of the water in Sample No. nF-263 was adsorbed by the clay mineral constituent. Under tiiese conditions, specific gravity of the solid constituents should liave been indicated as 2.103, assuming a void ratio of zero. Conclusions regard- ing void ratio, degree of saturation, and the relative amounts of free and adsorbed water are significant with regard to shear resistance and possible volume change with loading or change in moisture content. Depending upon the moisture content of the clay crystal, the specific gravity of montmorillonite can vary within wide limits. Calculations for hydrogen mont- morillonite based upon adsorptive phenomena described WATER IN INTERLAYER SPACE (Percent of dry waight) 10 20 30 40 50 60 1 1 \ 1 1 A 1 \ - 1 /^ 1 / A - ~ > 1 \ < \ \ 4 y^, - - v '1> \ ■ \ \ ^ 1 ' 4 12 16 20 Z* MOLECULES OF WATER PER UNIT CELL Figure IS. Theoretical variation of specific gravit.y of H-mont- morillonite with adsorption of water along the 001 interlayer space. by Bradley and associates (1937) are summarized in figure 18. They indicate a range from 1.775 to 2.683. Fluids Content. Water and air are the fluids always present in naturally occuring earth materials. Water occurs in varying degrees of fluidity ranging from the free water existing in large voids to the water subjected to significant capillary tension to the adsorbed water held bv forces of varying intensity upon external and internal surfaces with or without intimate association with exchangeable cations. The water in these various occurrences possesses differing properties, such as re- sistance to shear and vapor pressure. Loss of the several Table 8. Free svell ie.its of Wyoming bentonife in i-arious liquids. Liquid Tempera- ture of liquid (°C) Dielectric constant * Free swell (percent) 25 25 21 25 26 27 26 26 27 27 25 58 23 20 2.2 2.3 4.3 4.5 6.1 6.8 20.6 24.2 33.4 35.6 42.5 56.5 78.5 109 8 8 Diethyl ether 35 5 20 20 145 80 165 Nitrobenzene 22 115 40 Water _ 1240 770 ' From Maryott and Smith (1951). 218 Clays and Clay Technology [Bull. 169 kimis of water at diffpring temperatures demonstrates the need for close control of temperature during drying of soils ill preparation for physical and chemical tests. Air is an important constituent of earth materials, particularly ^vhen application of heat or penetration by capillary water deyelops internal pressure in excess of the atmospheric. Air is the source of oxygen, which contributes to many alteration processes, and of CO2, a compound participating in reactions which contribute greatly to changes in properties of earth materials. As will be indicated at appropriate places in the following discussion, the physical-chemical properties of earth materials yary markedly with yariation of dielec- tric constant, yiscosity, and electrical conductivity of interstitial water, and wnth changes of concentration of dissolved solids. Moreover, these and other properties of aqueous solutions at¥ect the hydration and surface- chemical properties of soil particles. In studies of clayey loess, Denisoy (1951") found con- solidation with given load to be a function of the wetting fluid as follows: Additional settlement jrith wettitiff undp7- load of Dielectric ^-5 '^P/'"! <■>» (percent) Liquid constant Undisturbed Distiirhed Water 81.07 11.0 13.6 Acetone 26.60 5.2 8.5 Benzene 2.29 0.02 0.87 Electrolytic aqueous solutions decrease consolidation in the sequence: water>NaCl solution>AlCl3 solution >FeCl:) solution. Also, the angle of stable slope for un- Tahle 9. Atterhern limits of earth materials in relation to the clay mineral." Plastic limit Liquid limit Plastic index lUite 35.7 24.8 23.9 28.8 25.4 26.3 36.3 29.9 25.0 41.6 26.3 97.0 55.0 63.8 48.0 81.4 40.0 116.6 61.2 35.9 29.1 .54.0 45.2 74.0 58.4 35.0 52.0 86.7 43.8 625-700 501.0 .537.0 227.0 117. .5 86.0 177.8 Not plastic Not plastic Not plastic 25 5 LaSalle Countv, Illinois ,,_ 11.1 5 2 25 2 Pretoria, South Africa 19 8 London, England''.. 47 7 Kaolinite 22 1 Twiggs County, Georgia Near Bath. North Carolina S.l 27 Cornwall, England , _. 45 1 17.5 Na-Montmorillonite Belle Fourche, South Dakota Clay Spur, Wyoming 528-603 466 473 . 2 Near Shelby, Montana 179 Ca-Montmorillonite 36.1 46.0 Attapulgite 61.2 Halloysite Eureka, Utah (hydrated) LawTCnce County, Indiana Allophane Lawrence Countj-, Indiana ' After White (1949). with addiUons. " Contains 5 percent of montmorillonite disturbed loess varied as follows for several wetting liquids : Dielectric Angle of stable Liquid constant slope Wiiter 81.07 26" Ethyl alcohol 24.1 29° Benzene 2.29 35° In slaking tests of Nebraska loess in which the binding agent is a montmorillonite clay, Karpoff and Gibbs (see Denisoy, 1946) obsei-ved immediate and complete slaking in w^ater, partial slaking in acetone, and little disturbance of the fabric in benzene. All of the above observations demonstrate the effect of dispersion and swelling of the clay binder in the loess in decreasing friction at points of contact of the grains. Swelling of Wyoming bentonite was indicated pre- viously in relation to dielectric constant and molecular structure of various organic liquids (table 8), the degree of swell tending to increase with increase of dielectric constant for compounds of similar molecular structure. Similarly, closeness of packing during sedimentation of finely divided solids, such as fuller's earth, alumina, iron oxide, silica, chalk, and talc, has been found to increase with increase in dielectric constant of the liquid (Fischer and Gans. 1946). Dielectric constant of the liquid is effective in these i-egards because the magnitude of attraction or repulsion by electrostatic charges be- tween the solid particles is inversely proportional to the dielectric constant of the separating medium. Dynamic Properties of Earth Materials Plastic and Liquid Consistency Limits The plastic limit is the minimum moisture 'content at which the soil is plastic* The liquid limit is considered to be the ininimum moisture content at which a soil is liquid. t In routine soil testing, the plastic and liquid conditions are established according to arbitrary criteria in empirical tests. The plasticity index is the difference between the plastic and liquid limits and thus defines the range of moisture content in which the soil is plastic. Stated differently, the plasticity index is the weight of water necessary to carry 100 grams of originally dry soil from a plastic to a liquid condition. Tlie magnitude of the plasticity limits depends pri- marily upon the content of particles less than 1 micron in diameter and the nature of tlie clay mineral. Havens, Young, and Drake (1949) demonstrated the influence of size range by comparison of the plasticity limits of numerous soils in both the natural condition and fol- lowing separation of the fraction less than 1 micron. The data reveal that the types of clay minerals which exist in particles smaller than 1 micron or are sus- ceptible of ready dispersion to this range, raise the plasticity limits significantly, particularly the liquid limit and jilasticit.v index (table 9). Sodium bentonites, such as are widespread in South Dakota and Wyoming, are characterized by liquid limit of 400 to 700 and plastic limit ranging from 50 to 100. Widelj' divergent limits reported for sodium bentonite • The plastic limit is the lowest moisture content (percent by weight of oven-dry soil) at which the soil can be rolled into threads one-eighth inch in diameter without breaking into pieces (Allen, 1942). t The liquid limit is the moisture content (percent by weight of oven-dry soil) at which the soil will just begin to flow when jarred slightly (Allen, 1942). Part V] C'LAV TECIlNOLOtiY IX SoiL ^lErllANICS 2111 relate at least in part to thixotropie strenprth developed with rest. E. J. Kileawley * found the li(|uid limit of a AVyoiiiJii'j: l)cntoiiite to be 4.")() percent immediately after mixiiifr, l)iit about 700 pereeiit if the sample was allowed to set for a week and tested by deereasinj? the water eon- tent through drying. The very high plasticity index of sodium montmorillonite, ranging from 450 to 600, dem- onstrates the enormous surfaee area to be satisfied before tlie adsorbed water films are sufficiently thick to permit their outermost zones to api)roach the characteristics of a li(|nid. At the liquid limit, water films on particles of Xa-montmorillonite are calculated to be 100 to 200 A thick. P'or calcium bentonitc, the liquid limit ranges from 115 to about 140, and the plasticity index ranges from about 35 to 90 percent. Sodium bentonite should show little influence of particle size upon Atterberg con- stants because of the easy dispm-sion effected in the par- ticles of clay mineral regardless of the natural size gi'adatiou. Calcium bentonite should show a greater variation of plasticity with fineness because of difficult dispersion, particularly after laboratory drying of the sample. Illite is widely variable in i)roperties. deiiciuling upon its ])ai'ticle-size distribution and its ]>ositiou in the montmorillonite-beidellite-hydromic-a-nutscovite series. White (1949) has demonstrated the marked increase in Atterberg limits of illite soils with increasing fineness, a eharaeteri.stic consistent with the difficulty of thorough dispersion of the mineral. Kaolinite contributes to only modest degrees of plasticity which increase but slightly with decrease in particle-size distribution. llalloysite. hydrated halloysite. and allophane are re- ported by "White to be nonplastie. However. Grim (1949) reports very high plastic properties for materials con- taining halloysite, in transition state between the hy- drated (4II:.6) and partially hydrated (2H2O) forms. The plastic properties of grani;lar earth materials or those containing illite. kaolinite. or halloysite as the pre- dominant clay mineral, are changed nuirkedly by addi- tions of even very small proiiortions of montmorillonite (table 10). Of course, the effect is greatest if sodium is the main exchangeable cation of the montmorillonite, due to its influence on dispersion. Winterkorn and Moorman (194T) have pi'oved the infltience of exchangeable ion on the plasticity range of a beideilite soil. AVith introduction of sodium, calcium, aluminum, magnesium, hydi-ogen. aiul potassium in por- tions of the soil, values ranged as follows: Iii(Hiiealcinm>.nhiminum>mjij.'nesium = li.Vflrosen>iKitassiinn IMa.stic limit : pcitassinm>ralcinni>aluminum>nia<;nesiiim = sodi um > li y (1 r»»j;en I'la.sticify index : sodium>caleium>alnminum>liydro>;en> magnesium>potassium By comparison of these values with results of water intake measured on similar or identical materials by Baver and Winterkorn (1935), a close relation between water affinity and plasticity index is indicated, the higher intake correlating with liighcr liijuid limit and * I'ersonal communication. plastic index values. The observation indicates that, with increased affinity of the beideilite for water, a greater l)roportion of water is required to produce adsorbed films whose outer periphery reacts to shear as a li(|uid. Data compiletl from several sources (Winterkorn and Moorman, 1941; Sanuiel.s, 1950) demonstrate the marked reduction effected in liquid limit and plasticity index of montmorillonite (Wyoming bentonite) by re- placement of naturally occurring sodium. The liquid limit and plasticity index decrease in the order: Na->H->K*>Ca**>Mg**>Al***>Fe*** = Th"** (fig. 19). Change in licpiid limit and plasticity index is far less marked for kaolinite, illite, and beideilite. For beideilite, the action of potassium in binding together adjacent packets of the cry.stal is evident in the low value of the liquid limit and plasticity index for the K-beidellite modification. Sodium influences properties of clays far beyond relative abundance in the exchange positions. According to Winterkorn (1953), if as little as 15 percent of the exchange capacity is occupied by sodium, montmorillonite exhibits the properties of homo- ionic Na-montmorillouite. Introduction of any of several eationic aliphatic amines to montmorillonite clay engen- ders hydro]iliobic qualities which radically decrease the liquid limit and plasticity index (fig. 20), and change significantly other physical and physical-chemical i)rop- erties (Dav'idson. 1949; and Glab. 1949). The physical-chemical properties controlling Att(>rberg limits relate to many fundamental soil mechanics prop- erties. According to Winterkorn and Moorman (1941), Koegler has found an almost linear relationship between plastic iiulex and the angle of friction of cohesive soils, ilaterials with a licjuid limit up to 35 percent and values of plastic index in the range to 15 will have high in- ternal friction if well graded. With plastic index less than 5, cohesion is slight or absent. With higher plastic index, cohesion is high. Liquid limits in excess of 35 percent and plastic index greater than 15 correlate with increased content of clay minerals or organic matter, the necessary amount of clay to increase the magnitude of these properties into this range being least with the montmorillonite-type clays of sodium type. In this range dry weight is low (generally less than 100 pounds per cubic foot) and shrinkage, expansion, and capillarity are high. Burmister (1949) classifies plasticity according to plastic index as follows: Pln.iiirHy PUislicity index Xonpla.^tic SliKllt Low Medium Ili^'li Very hiKli Permeability In their usual occurrence, clay minerals in earth materials ai'e exceedingly fine and consequently tend to fill void spaces lying between grains of silt, sand, and gravel. Consequently, clay minerals characteris- tically decrease permeability of earth materials, es- pecially of remolded clays in which structural features, such as shrinkage cracks, joints, stratification, or shear zones are absent. Permeability decreases with the regu- larity of particle-size distribution of the material, es- pecially with the extension of the size range into 1 to o 5 to 10 10 to 20 20 fo 40 40 ;ni d urr ■ater 220 Clays and Clay Technology [Bull. 169 Ht K* Co'* Mg* EXCHiNGEABLE CATION Th«*<-* Figure 19. Effect of exchaugeable cation on Atterberg consist- ency limits of clays (Winterkorn and Moorman, 1941; Samuels, 1950 ; Kilcawley, personal communications). fractions below 1 mieroii. Aceoi'ding to Lambe (1954b), the major factors influencing the permeability of fine- grained soils are: (1) soil comjjosition, (2) character- istics of the fluid, (3) void ratio, (4) fabric, and (5) degree of saturation. For remolded clays, under ordinary circumstances of saturation by relatively pure water, permeability of earth materials is least if montmorillonite-type clays are present and greatest if kaolinite is the clay mineral component. In tests upon beidellite clay, Winterkorn and Moorman (1941) found the permeability approxi- mately ecjual for the hj^drogen, calcium, and magnesium modifications, and distinctly less than that of the potas- sium modification at void ratio less than 1.3; but at void ratio greater than 1.3, the permeability of the magnesium modification rises rapidly and greatly ex- ceeds that of the calcium soil. Although not reported directly, other tests of the sodium modification indicate that it possesses the lowest permeability of the several soil modifications investigated. In comparable tests of modifications of Wyoming bentonite, Samuels (1950) determined the permeability relation to be thorium > aluminum>caleium>sodium at loads less than 1 ton per square foot. At greater loads, the permeability of the calcium and sodium modifications is virtually nil (fig. 21). The permeability of the aluminum modification becomes negligible at loads of about 5 tons per square foot. Permeability of treated kaolin is similarly affected by exchange of ions but the permeability is 4 to 20 times greater than for the bentonite modifications. For the kaolin the permeability decreases in the series aluminum>calcium>hydrogen>sodium, for given load. The permeabilit}' of kaolin can be decreased greatly by decrease in particle size and increase in surface area (Harman and Fraulini, 1940). With increase of ap- parent surface area from about 30 square meters per gram to 240 square meters per gram the permeability of kaolin decreases by a factor of 3.5. As has been implied, permeability of clay formations in place usually departs greatly from the permeability of the material when remolded. Structural features, such as shrinkage fractures developed by drying or syneresis, joints, bedding planes, and shear zones, lead to high permeability in many clays, claystones, and shale formations. Sealing of such fractures as the result of swelling of the clay mineral should not be depended upon in design of hvdraulic works. In a study of coarse gravel containing abundant interstitial clay, Cary (1949) found a high degree of permeability apparently as the result of the intimate fracturing of the clay by shrinkage through syneresis. These fractures occurred both above and below the water table. The foregoing discussion is based upon consideration of essentially pure water as the penetrating medium. Electrolytic solutions will greatly increase permeability of clay soils and formations by inhibition of swelling as the result of hydration. Newly introduced solutions can progressively effect ion exchange, in addition to the -cv-ARMAC T -TARMAC I8D -o-ARMAC I2D ^ROSIN AMINE- D ACETATE -•-AMINE 220 -O-AMMONV XT £0 40 60 80 DEGREE OF SATURATION OF CATION EXCHANGE CAPACITY (PERCENT) WITH CHEMICAL ADMIXTURE FiGURK 20. Eirpct of cationic surface active agents on plasticity index of a montmorillonite soil (Davidson, 1949). Part \' Ci.Av Tki iiNdLoiiY IN Soil ilEciiAxirs 221 fioeculaliiij; or coasrulatiiig effect. Flushing of the elec- trolytic solution by water containing only small amounts of dissolved solids ajrain will restore a reduced permea- bility, altlioujrh semipernianent increases may be de- veloped bv the ion <'Xcliange process (Lee, 1941 ; Hod- man, 1941). In engineering practice, clays are used to control permeability of formations by linings of various types and by grouting. Linings include use of bentonite, loose earth blankets, and conqiacted earth. Loose earth and compactctl earth linings may contain any species of clay mineral, the prime reciuisite for satisfactory installation being the low permeability of the lining in place. The permeability is a function of gradation, in-place density, the thickness of the lining, and the hydraulic conditions of operation. Bentonite is used in various ways in linings: (1) as membrane I to 2 inches thick, depentliag primarily upon the montmorillonite content, swelling properties, and fineness of the material; (2) mixed with soil and placed in a layer about 2 to 3 inches thick; and (3) mixed into soil in place b.v disking, harrowing, or rototilling. Sev- eral successful applications have been made by sowing of commercial bentonite over the surface. Elsewhere, canals and reservoir bottoms have been treated by silting action, the bentonite being tirst s])read over or mixed in the water. iMembraue linings reciuire adequate cover for pro- tection against drying and consequent cracking, curl- ing, and erosion. Cracking as the result of syneresis might occur in membrane linings, even under water or moist cover material, but the conditions under wliich syneresis is significant I'emain to be determined. Admix- ture of bentonite with granular material increases the shrinkage limit and thus jn-events formation of the penetrating cracks wliicli iire so typical of lean and fat clays. Of critical significance in use of bentonite to develop impermeability, is the limiting expansion developed in sodium montmorillonite with adsorption of water. Al- though the osmotic adsorption of water by sodium montmorillonite is great, complete dispersion of the clay crystals is resisted by interlayer forces. This re- sistance to dispersion is easily demonstrated in simple free-swell tests in which jiarticles of sodium bentonite expand to limiting bulk volumes up to 17 times their dry volume. However, with agitation, complete dis- persion takes place readily. Permeability of linings can be changed b.v ion ex- change through continued contact with saline waters. As is evident from the review of laboratory tests pre- viously described, exchange of sodium for calcium will increase permeability of linings owing their impervious- ness to clay minerals. In practice this effect is developed particularly by saturation of bentonite linings from below by waters carrying dissolved gypsum (CaS04- 2H2O). The swelling of sodium bentonite is reduced greatly by contamination of the raw material with gyp- sum. Apart from ion exchange, permeability is influenced by coagulation or dispersion of clays as the result or change of electrolyte content. Lee's (1941) treatment of the lagoon lining on Treasure Island in San Francisco Bay with sea water is an excellent and now classic demonstration of effects of both ion exchange and the in- fluence of electrolytes on soil structure and permeability. Upon flooding of the lining with sea water sufficient to l)crmit seepage tlirouiih the lining of a total of 40 inches of deptli, the ratio of exchangeable sodium to calcium was changed from 1:1 in the original soil to an average of 2.2:1. IJpon drainage of the excess salt water from the lagoon and introduction of fresh water, permeability of the lining was found to decrease as the result of collapse of the fabric of the clay and dispersion of the finest fractions with decrease of electrolyte content of con- tained water. The seepage rate at completion of the operation was found to be 0.10 inch per day in con- trast to 0.90 inch ])er day pi-ior to treatment. Control of electrolyte concentration of cla.y -water mix- ture was put to advantageous use in the construction of Alexander Dam, Island of Kauai, Hawaiian Islands, (Cox, 1929; Anon., 1930; Anon., 193()a^ when coagula- tion resulting in excessive pei-meability and low unit weight was prevented by a carefully controlled addition of soda ash to the sluri-y introduced to the core pool. The resulting dispersion decreased void ratio of the deposited material from 2.52 to 1.98 and the permeabil- ity decreased from 0.6 X 10"^ to 0.04 X 10"^ centimeter per minute. Sodium bentonite is used as grouting material either with water alone, with liquid petroleum pi-oducts, or as an admixture with |)ortlaiid cement. Bentonite-w'ater grouting mixtures usually contain 4 to 15 percent bj' weight of the clay. In bentonite-portland-cement mix- tures, bentonite constitutes 10 to 20 percent by weight of the total solids, the optimum proportion depending upon water-to-solids ratio and the nature of the formation to be grouted. Similarly, the ratio of bentonite to petroleum product depends upon the grouting conditions and the liquid used (Cordon, 1944). Bentonite or bentonite- cement grouts are ineffective in grouting sands contain- ing fractions passing the No. 50 or No. 30 sieve, respec- tivel.v (Burnett, 1936). They find best application in grouting of formations containing comparativel.y large and interconnected voids, channelways, and fractures. Eft'ectiveness of the bentonite water grout depends upon expansion of the clay in place within the openings to be sealed and upon thixotropic setting of the slurry. If hydraulic pressure is applied before setting occurs or if failure begins, permeability rises rapidly because of destruction of the established thixotropic gel struc- ture. Best results are obtained if injections occur as coagulated matter so that swelling is completed after em- placement. Formations grouted with bentonite must be kept wet if the seal is not to be lost by drying shrinkage. Admixture of bentonite with portland cement, aids in- jection of the grout into the formation at given con- sistency, retards hardening of the cement, and improves the seal effected. Thixotrop>^ As originally defined by Peterfi and by Freundlich (Freundiich, 1935; Green and Weltmann, 1946) thixot- ropy is an isothermal, reversible, sol-gel transforma- tion. The phenomenon is well exemplified by marked increase in viscosity of sodium-bentonite sols with rest, and the subsequent decrease in viscosity with agita- tion. This sequence may be repeated indefinitely. Later 222 Clays axd Clay Technology Tahle JO. Coiisiatenci/ uiid shrinkage Umits of synthetic bentonite and kaoUnite soils. I Bull. 169 Composition (percent by weight) Liquid limit (percent) Plastic limit (percent) Plasticity index (percent) Shrinkage limit (percent) Shrinkage ratio Wyoming bentonite * Kaolin •> Sand and Silt = lOD 50 25 10 5 2.5 3.0 10.0 25 40 45 47.5 100 50 25 10 5 47.5 45.0 40.0 25 10 5 2.5 50 75 90 95 50 75 90 95 50.0 50.0 50.0 50 50 50 50 501.0 253.0 128.0 52.5 30.7 52.2 18.0 16.3 17.0 31.0 45.0 65.0 146.0 217.0 235.0 234.0 35.0 21.0 15.0 24.9 25.8 25.3 n.i 15.1 17.0 12.0 13.0 15.0 16.0 14.6 19.0 19.0 466.0 232.0 113.0 27.6 4.9 26.9 6.9 1.2 0.0 19.0 32.0 50.0 130.0 202.4 216.0 215.0 8.1 6.9 24.9 34.8 29.3 25.3 14.9 15.1 17.9 17.2 15.4 15.2 15.4 24.4 10.3 8.8 8.0 1.78 1.91 1.47 1.36 1.48 1.43 1.90 1.88 1.77 1.75 1.81 1.80 1.80 1.84 1.81 1.79 1.88 " Natural bentonite, near Osage. Wyoming, containing about 85 percent of Na-montmorillonite. •• Kaolin, near Batli. Suutli Carolina, conl;iining about 9H percent liaolinite. *" Screened and washed from natural sand. Clear Creek, near Denver, Colorado, and containing equal parts by weight of No 50 to No 100 and No 100 to No 200 fractions. Table 11. Atterberg consistency limits and stiffening limit for several clays and shales. Geologic formation Petrographic type Predominant clay mineral Consistency limits (percent of dry weight) Liquid limit Plastic limit Plasticity index Sti£fening limit Eagle Ford shale, near Dallas, Texas lone formation. Valley Springs, California, - Graneros formation, near O.sage, Wyoming.. Nespelem formation, near Coulee City, Washington AlluWum, near Yuma, Arizona Porterville clay, near Lindsay, California. Shale Claystone Claystone Varved clay and silt Clay Claystone Beidellite KaoUnite Na-Montmorillonite Vermiculite-Chlorite-Illite, and Montmorillonoids Ca-Beidellite Ca-BeidelUte 49.0 49.0 501.0 60.0 80.0 69.3 25.0 24.0 35.0 27.0 28.0 33.0 24.0 25.0 466.0 33.0 52.0 36.3 132.0 136.0 1560.0 120.5 123.0 134.0 work questions the necessity for complete transfor- mation of gel to sol as a prerecjuisite for tliixotropy, the property being identified with a variation in vis- cosity with rate of shear (Green and "Weltmann. 1946). The data strongly suggest that tliixotropy results from development of a structure involving the suspended solids and the licjuid phase. Thi.xotropy is a characteris- tic only of systems in which large volumes of liquid phase are adsorbed upon and held between particles. Tiie upper limit of size of particles participating in the thixotropie structure are about 10,000 A. After cessation of agitation, the particles supposedlj^ are shifted about by Brownian movement, until a condition of balance of electrostatic and gravitational forces is attained (Lar- seii, 1946). Orientation of the solid particles and mole- cules of the liquid proceeds slowly, so that thixotropie strength develops progressively and is destroyed only through finite and measurable periods of time. Thixot- ropy in relation to shear resistance of earth materials is discussed subsequently. Thixotropy develops to a degree in almost all sols, but the rate of formation and break-down and its strength vary greatly with the particle size, particle shape, surface activity, exchangeable cation, concentra- tion, nature of the liquid phase, electrolyte content, and other factors. Among clay minerals, the property is most easilj- demonstrated in bentonite-water slurries. For sodium-montmorillonite, 80 milliliters of water can be added to 1 gram of clay and the mixture forms a gel which will not flow from a test tube 8 mm in diameter 1 minute after cessation of agitation. In similar tests 30 milliliters of water can be added to 1 gram of kaolinite; whereas, only 16 milliliters of water can be added to 1 gram of muscovite of the same grain size (Winkler, 1949; 1943). Thixotropy of bentonite is reduced if alkali ions are replaced by Ca" or Ba" and is eliminated by complete removal of alkali ions and their replacement by H* or pol^'^'alent ions (Freundlich, 1935: Houwink and Burgers, 1939). As is well known, bentonite-water slur- ries are used as deep-well drilling fluids, because thixo- Part V 1 Clay Technology in Soil Mechanics 223 y WYOMtNG BENTONITE \ \ V / \ \ ^ - NaXri Co, ^^^^^ — ^ PRESSURE (Tons per squore foot) 0. u. o z u 20 o u. \. KAO LIN V \ ~^~- *'', ti"> ^ No- ---— ^^-^ PRESSURE ( Tons pep squope foot ) FKiiiii: I'l. EITt'Ct of pxchanspalilc Ciition on permoability of kaolin ami Wyoniin;; bentonitc (Samuels, ]0."i()|. tropic setting of the slurry prevents rapid settling- of the cuttings after temporary stopping of the rotation of the bit. In a tlionght-provokiug article, Ackermann (1948) describes tlic sio-nificance of thixotropy in soil mechanics prdpcrtics of glacial clays of Norway. The clays range i'roiii fat clay to fine sand of low clay content, and con- sist predominantly of particles in the range from 2 to 20 microns. The finest fractions are predominantly com- posed of particles in the range from 0.05 to 0.20 micron. The .sand and silt fractions are mainly quartz and feld- spai-s. ;\Iica and clay minerals are concentrated in the finest fractions. No montmorillonite clays were identi- fied. Ltdioratory tests of clay-water mixtures demon- strate the thixotropic setting of sols of the glacial clays with rest. Because of the great significance of thixotropic strength in establishing stability of these soil masses at high water content, Ackermann (19-18) proposed an ad- ditional consistency limit comparable to those proposed by Atterberg. He designates the (punitity "stiffening limit." The "stitfcning limit" is the boundary between the quick and liquid consistencies. At water contents greater than the liquid limit an apparently stable soil is main- tained by thixotropic strength and hence can be liquefied suddenly by vibration, overloading, or other disturb- ances. The "stiffening limit" is defined as the water content (expressed as percent by weight of diy soil) at which q thoroughly stii'red thixotropic soil still fiows under its own weight in a test tube of 11 mm diameter after ex- actly 1 minute of rest. The liquid limit of thixotropic Norwegian fat clays generally is above 40 percent, whereas the "stiffening limit" ranges from 65 percent to very high values. Lean coarse clay (silt) of the same electrolyte content, and containing the same clay min- erals as the fat clay, becomes syrupy at water contents as low as 20 percent, and the "stiffening limit" always is lower than 100. The lean coarse clay has a gi'eater ability to How because it liquefies at a lower water content. Ability to flow is inversely proportioiml to thixotropy. Ijow liquitl limit (mostly in the range :i0 to fiO percent) indicates weak thixotropy, whereas high "stiffening limit" (mostly over 100 percent) indicates high thixotropy. Norwegian (piick clays possess a liquid limit less than 56. Becau.se of its high degree of thixot- ropy, a fine clay with high optimum moisture content and optinunn concentration of electrolytes may become more highly litpiid. and thus more likely to flow, than the coarse clay or silt. The "stitt'ening limit" of several clays and shales occurring in the Ihiited States is indi- cated in table 11 in relation to mineralogy and Atter- berg consistenc.v limits. Gelling time is closely related to concentration of the sol. Winkler (121) summarized data for sodium mont- morillonite-water slurries as follows: Volume of the liquid Volume of the solid 3.3 34 36 Gelling time (minutes) 1 6 45 Thixotropic properties occur even in cohesive soils composed largely of sand and po.ssibly containing less than 5 percent of clay minerals. DistincJ thixotroi)y has been demonstrated in certain quick sands and loess, the latter containing about 3 percent montmorillonite (Freundlich and Julinsburger, 1935). It is entirely likely that thixotrojjy contributes to stability at low moisture content of loess in western Nebraska. The structure of the loess is preserved by films of mont- morillonite upon the grains of silt and fine sand. "With wetting, this loess is subject to considerable consolida- tion as the result of collapse of the original open fabi-ic. In claj-ey soils, thixotropy is strongly influenced by electrolyte content. AVinkler (1943) demonstrates in- crease in thixotropy of aqueous suspensions of mica and kaolinite by addition of NaCl and KGl; and reduction by addition of BaCU and CaCL. For aqueous sols of sodium montmorillonite, originally air dried, increasing concentration of NaCl, KCl, NaOH, and KOH up to 1.0 N continuously decreased thixotropy. When main- tained moist without drying prior to preparation of the slurry, sodium montmorillonite attained a nuiximum thixotropy' with potassium hj-droxide at 0.65 normal. 10 meters GRAVEL QUICK CLAY BLUE CLAY, VERY SOFT |-_-g_j BLUE CLAY, SOFT (S:mJ BLUE CLAY. STIFF BLUE CLAY, SEMISOLID Figure 22. Devplojiment of a zone of quick clay throuKli leach- ing of electrolytes by drainage into an underlying gravel stratum (Ackermann, lO.jO). 224 Clays and Clay Technology [Bull. 169 For hydrogen montmorillonite a definite maximum in thixotropy is observed in water slurries in which the concentration of KOH is about 0.05 X. Simultaneously, sediment volume and swelling attain a maximum. At this concentration of KOH, the H* of the original acid montmorillonite is neutralized (Fi-eundich, 1935; Hou- wink and Burgers, 1939). As is noted by Marshall (77) two zones of thixotropy can be developed in aqueous slurries of montmorillonite : one at low electrolyte con- tent and one at high electrolyte content, the two zones being separated by a range of concentration in which thixotropy is small. Rosenquist (1946) found thixotropie strength de- velopment in clays with addition of up to 3-pereent sodium chloride solution, but with further additions the quick clays lost their ability to flow. The quick condition is recovered with removal of the excess salt. Rosen- quist 's experiments correlate with observation of zones of quick clay containing water of minimum electrolyte content, and the occurrence of quick clays overlying permeable gravels (fig. 22). Thi-ough decrease in thick- ness of adsorbed water films with decrease in concentra- tion of electrolyte, a plastic clay can become quick without changing total water content. "With other clays and other solutes, quick clays might develop through increase in salt content of interstitial water. Shear Resistance Resistance to shearing stress is developed in soils by particle interlocking, meshing of irregularities on par- ticle surfaces, adhesion, cohesion, and cementation by secondary minerals, such as iron oxides, carbonates, quartz, opal, and chalcedony, which are deposited inter- stitially. Clay minerals are of fundamental importance in establishing shear resistance of earth materials. At low water content, restricted water films subject the mass to compressive stresses of considerable magnitude and thus increase shear resistance. However, increasing degree of saturation by introduction of water decreases the internal compressive stress arising in water films, ^lo-reover, if degree of saturation is increased by rapid consolidation of the earth material with loading, shear resistance of the material is reduced by pore water pressvire. Consequently, shear resistance of earth ma- terials depends not only ixpon the structural integrity of the solid constituents, but also upon the ability of the material to drain contained water and air as read- justment cf the structure takes place in response to load. Clay minerals and the fabric of granular earth ma- terials play critical roles in both phenomena. Winterkorn and Moorman (1941) have found the shear resistance of cohesive soils to be approximatelj"^ a logarithmic func- tion of the moisture content in the lower part of the plastic range. Shear tests of soils containing differing clay minerals and clay minerals with differing exchange- able ions demonstrate the interrelation of structure and permeability in control of deformation of the test speci- men. "With loading of specimens of a beidellite (Putnam) clay, Winterkorn and ]\Ioorman (1941) show greatest deformation in the sodium modification and smallest for the modifications containing calcium, hydrogen, and potassium. In triaxial shear tests performed on the same soils at maximum density, and optimum moisture content, but without achievement of equilibrium, de- 1 WYOMING BENTONITE 1^ Al", ^ -^ --(Za ^ Na-, - ^-^ ^^ 10 20 30 40 50 60 NORMAL PRESSURE (pounds per squore inch) KAOLIN Ca- ^ ^ --'rja ^^ Al- -^ ^ ^^ NORMAL PRESSURE (pounds per square inch) Figure 23. Effect of exchangeable cation on shear resistance of kaolin and Wyoming bentonite (Samuels, 1950). formation of the sodium clay is less than that of the hydrogen clay because drainage of the sodium clay was greatly impeded by its very low permeability. Cor- respondingly, the internal friction of the sodium clay is low because of the high pore pressure and ponsequent anomalously low compressive stress linking elements of the fabric together. Loosely cemented silts and sands possessing an open fabric are especially susceptible to loss of shear resist- ance with disturbance by vibration or rapid changes of load w'liile saturated. Intergranular braces, so typical of many loesses, soften with wetting and lose cohesion. Loose or poorly cemented sands and silts are subjected to '"spontaneous liquefaction" with shifting of the grains to more stable positions while the voids are filled with water (Terzaghi and Peck, 1948; Terzaghi, 1950). In reverse, marked resistance to shearing is developed with rapid application of shearing stress to saturated, loose, noncohesive or slightly cohesive earth materials w-hich possess a closely packed fabric ( Terzaghi and Peck, 1948; Green and Weltmann, 1946). Disturbance of the fabric necessarily results in increase in bulk volume and dilation of the voids. The material then appears dry and hard, as, for example, in the familiar experience of walking upon saturated sand on the ocean shore. In their undisturbed condition, dilatant materials flow readily if not stressed rapidly. "With release of the applied stress, the original compact fabric is restored, presumably by compressive stress arising in the capil- laritj' of interstitial water. These phenomena are devel- oped especially in silts and sands. Dilatancy is significant in drilling because the mass of cuttings settled com- pactly in the hole commonly expands (dilates) against the drill rods or pipe when an attempt is made to withdraw the tools (Larsen, 1946). Jarring momentarily re-establishes the compact fabric and permits progressive Part V] Clay Technology in Soil Mechanics 225 UNDISTURBE D CONDITION THIXOTROPIC FIN4L SHEAR RESISTANCE X> // < REMOUL )ED CONDITION 80 160 24-0 320 TIME (days) Ku.uuE 24. Changes in shear resistance of plastic glacial lake clays with remolding and rest (Ackermann, 1948). TIME (HOURS) FiGURK 25. Thixotropic development of penetration resistance in clays. freeing of the pipe or rods. Dilataney as a factor in soils mechanics is described further bv Terzaghi and Peck (1948). Clay mineral constituents of earth materials increase sensitivity of shear resistance to moisture content. The range of moisture content at maximum shearing resist- ance decreases as the clay niinei-als become increasingly abundant, and as iutergrauular friction is reduced. The role of shear resistance in controlling densitj' of soils remolded with given compactive effort must not be overlooked in study of stress-strain phenomena in earth materials. The moisture content yielding maximum co- hesion is so low the workability of the material is poor and low void ratio and high densitj- in the remolded specimen are attained only witii great difficulty. To obtain both high density and higli shear resistance John- son and Davidson (1947) recommended compaction at optimum moisture content, followed by drying to the moisture range producing maximum shear resistance. In studies of Wyoming bentonite, Samuels (1950) dem- onstrated increasing shear resistance with the exchange of cations in the series, sodium < calcium < aluminum (fig. 23). Shear resistance of sodium, calcium, and aluminum modifications of kaolin is essentially identical and similar to the shear resistance of Wyoming bentonite containing aluminum as the exchangeable cation. By 8—91001 replacement of naturally occurring hydrogen, calcium, and sodium in the Putnam soil with potassium, Winter- korn and JMoorman (1941) increased the angle of fric- tion from 19° to 22°. The effect of exchangeable cation upon the angle of friction of kaolin is negligible. Shear resistance is increased in clayey materials by development of thixotropie strength, but this element of shear resistance is lost with disturbance of the fabric by vibration or rapid strain (Ilvorslev, 1939). With disturbance of natural structure by crushing and knead- ing, plastic clays become softer; and clays owing their plasticity largely or entirely to thixotropy become liquid. With remolding and rest, thixotropie strength increases shear resistance at a decreasing rate, the shear resist- ance ultimately reaching a constant value which is only a fraction of the shear strength of the undisturbed ma- terial (fig. 24). Strength gain is very slow after 4 weeks of rest. The proportion of original strength re- gained varies widely with the nature of the material. The original strength arises in cementation, particle interlocking, interpartiele cohesion, and stable water films developed tlirough geologic time. These features are regained only partially or not at all. Strength regain with time was observed in illitic and micaceous Nor- wegian glacial clays by Ilvorslev (1939) and by Acker- mann (1950) ; and for montmorillonite clays in Germany by Winkler (1949), and in Mexico Citv. Mexico, by Zeevaert (1947, 1949) and by Cummings (1930). Con- struction experience with a thixotropie fill material is described by Hirashima (1948). Thixotropie stiffening of clay develops increased re- sistance to penetration. In laboratory tests, resistance to penetration is increased most rapidlj' for Wyoming bentonite and least for a kaolin ; whereas intermediate values were obtained for a glacial lake clay (Nespelem formation, near Coulee City, Washington) containing montmorillonite, illite and hydrous mica, and a vermicu- lite-chlorife (fig. 25). In nature and in engineering practice, landslides or slope failures are the commonest demonstration of inad- equate shear resistance in earth materials. Although any natural material will fail on slopes under extreme con- ditions, poorly cemented sands and silt, with or without clay iniiicral fractions, and fractured expansive clays FiGCKE 2 Mg f Fe [/ / y 10 ■ C( y y y y / ^ i ^ Y / A f> H y y / / '/, ^ / ^ ^ ELECTRICAL POTENTIAL (Volts) FuaRK 28. Effect of exchangeable cation on electro-osmotic yield of a kaolinite soil (Winterkorn, 1047). Part VI Ci.AV Tk(IINOlo«y in Sou. .Mi;i hanks niitrlit be increased l)y ehaiiy:es in (•(ineeiitral ion ol' elec- trolytes in intei-stitial water. Increased shear resistance in lii'rinvay sul)i;rades lias been obtained economically by varions tyjies of cheniical and iihysical-clieniical treatment. Stabilization of clays re()uires bindinp: of tlie particles by agents resistant to water or waterproofing by agents which permit main- tenance of sntficient cohesion by watei- originally con- tained in the material. Cementing action can be obtained by several inorganic and organic binding auents, each with particnlar ranges of application, conti'olled lai'gely by the snrface-chemical properties of the earth material. "Winterkorn (1948) has classified earth materials and complementary agents as follows : ( ''nnploineiitHry chemical juiinixtiire Siirf.ico-clicmical chnractcr of the soil 1. Complex iron and aliimiiitiiii 1. (>r;;aMic catiou.s (ammonia silicates with SiO;:K-..()n ra- derivatives anil snhstances rio ;;reater than two: the whieli may react with par- clay particles are typically tide surfaces throngh hydro- charged iiecatively fien linkage) J. ("iiniplex iron and aluminum 2. Organic anions, such as solu- silicates with SiO^. : KcO, ra- hie soajis. saturated and un- tio less than two and com- saturated fatty acids, and plex hydrous iron and alu- resin acids. Antio.xidauts are niinum oxides. The negative recommended to assure dura- charges are less and positive hility charges are greater than for soils of (Jroup 1 .■>. Particle surfaces covered hy .'?. Snhstances forming synthetic lignaceous and proteinaceous resins with humus materials organic matter 4. S:iline and alkaline .soils A. I'ii-lcl unexplored ~i. I'r.it anil muck ."i. Suhslances forming synthetic resins with organic matter or wliirli themselves form resins For quantitative data and the limitations ])ertriiiiing to these methods, the reader is referred to the definitive writings of II. V. Winterkorn and his associates (i;)4(3). Recent work on stabilization of soils bv dispersants is reported bv Lainhc (lli.")4a' and I.amhe and Michaels (li)r)4). Stabilization of highway snbgrades liy inti-nduction of divalent or trivalent iiiortianic ions, sndi a -i calcinm liydro.xide oi' salts of calcium, aluminnm, or iron, has been investigated extensively but best results have been obtained so far with calcium hydroxide (Callaway and Buchanan, 1951). Although these methods have yielded satisfactory results in highway snbgrades, whei-e only a few inches of material must be treated, stabilization of clay foundations of bnildini;s. dams, bridges, and simi- lar large works by mechanical admixture or hydraulic injection (^f chemical agents, has not proved feasible be- cause of the quantity of material in place to be treated and the characteristically very low permeability. For successful treatment of a soil mass the agent must per- meate the soil fabric rather than merely |)(>net rating fractures transecting the formation. Physical-chemical treatment of loess in place has been described bv Deni- sov (1951).. Extensive investigations have been conducted on elec- tro-osmotic and electro-chemical stabilization methods in many laboratories of the United States, England, Ger- manv, France, Eussia, and elsewhere (L. Casagrande, 1937; 1939; 1947; 1949; Endell, 1935; Preece, 1947; Erlenbach, 1936; Endell and Hofmann, 193fi; Kolbn- szewski, 1952; Schaad and Ilaefeli, 1946: Vey, 1949; Spanulei-, 1!149; Karpofl', 1953 i. In accordance with ob- servations originally made by Renss in 1808, adsoi-bed cations, together with the adjoining water in inter.stices of clays, will move toward the cathode in an induced electrical field, a process designated as electro-osmosis. If continued sufficiently long, the naturally occurring cat- ions are replaced by hydrogen and significant quantities of watei' are moved from portions of the material, or may be I'eniox'cd entirely, if desired, at the cathode. Shear re- sistance is increased by replacement by hydrogen of ions, such as sodium, which tend to increase zeta potential and water-adsorbing cajjacity of clays, and also by re- moval of the interstitial water. If a temperature gra- dient exists or is induced by the electrical potential, thermo-osinotic transfer of water will occur sinniltane- ously. Winterkorn ( 11)47 I has demonstrated widely differing etfei-tiveness of electro-osmosis with change of exchange- able cation in a kaolinite soil from ncai- llagerstown, \ew Jersey (fig. 28). For given power consinnption, aluminum, ferric iron, and hydrogen modifications re- leased more water than did the calcium, potassium, and sodium soils, magnesium soils being intermediate. How- ever, the calcium, ferric iron, aluminum, and hydrogen soils re(|uire greater voltage at given current intensity (30 milliamjieres) for given water release than did tlie. sodium, iiotassinm, and magnesium soils. At given electrical potential and dielectric constant and vi.scosity of the pore water, the volmne of electro- osmotic flow increases with zeta potential and decreases as the size of the capillary decreases. Conseiiucutly. in sodium clays with high zeta potential antl high dis]ier- sion. conductance is high and electro-osmotic yield is low. Conversely, for ferric iron and aluminnni modi- fications, the zeta potential is low and dispersion is less; but when the low zeta potential is compensated by higher induced potential gradient, electro-osmotic flow is moi'c rapid and is accomplished with less jicwcr con- sumption than foi- the sodium modification. During electrochemical treatment, electro-osmosis is sup|)lcinented by action of chemicals introtals toward the anode. ^loreover. complex crystals, such as montmorillonite. can be partially decompcsed at higher potential gradients. Electro-o.smotic treatment is especially applicable to dewaterino: of silts and clays in which electro-osmotic permeability is equal to or greater than hydraulic per- meability (L- Casagrande. 1949). In clays, electro- osmotic permeability (expressed as centimeters per second at } volt per centimeter) usually is several thou- sand times tlie hydraulic ]iermeability at unit hydraulic trradient. In their present state of development, electro-usinotie methods are not competitive with ordinary methods of foundation or slope control, such as dewatering b.v drains or well points, or use of sheet piling or. for foundations, use of any of several types of piles or foot- ing designs to support the structure. However, where other methods have failed, electrical methods of soil 228 Clays and Clay Technology Table 12. Compressive strength of synthetic soils. I Bull. 169 Mixture (percent by w eight) Moist specimens .\ir dry specimens Wyoming Kaolin •* Sand 1- Density (lbs. /ft.") Water content (percent) Compressive strength (psi) Density (lbs./ft.») Water content (percent) Compressive strength bentonite » Initial Final Initial Final (psi) 100 77.7 17.1 55.5 79.5 87.2 20.0 7.0 195.2 25 75 105.8 14.4 42.0 108.2 112.1 18.1 1.3 .541.0 25 25 50 107.8 13.9 85.9 107.5 116.6 19.6 2.6 553.5 ..._ 25 75 111.2 13.6 9.1 108.9 114.7 17.8 0.1 76.5 - — 100 — - 89.9 14.2 100.3 86.1 85.4 17.8 0.4 65.6 "^ Predominantlj' Na-montmorillonite from ne.ir Osage. Wyoming. *» From near Bath. South Cirolina. ' Equal parts by weight of No. 50 to No. 100 and minus No. 100 fractions of sand from Clear Creek near Denver, Colorado. stabilization have proved successful at reasonable cost. For examples of application, cost, procedures, and theo- retical treatment the interested reader is refei-red par- ticularly to the writings of L. Casagrande (1947, 1949), of Preece (1947), and of others (Steinfeld. 1951; Kar- poff, 1953). Unconfined Compressive Strength Unconfined compressive strength depends upon many aspects of fabric and composition, including size, shape, and gradation of constituents; their mutual relations; size, shape, and distribution of voids ; and the content of water and air. These factors control the response of earth materials to loading, including elastic and plastic adjustments, as well as shear failure. Clay minerals are significant in response of earth ma- terials to loading because, when dry, they develop rela- tively high strength through formation of adhesive and cohesive bond throughout the mass. Introduction of small proportions of clay minerals to a sand or silt greatly increases compressive strength, the maximum strength of the mixture exceeding that of either the sand or silt or the clay alone (table 12). The strength of such mixtures increases more rapidly with small additions of montmorillonite minerals in contrast to kao- linite, halloysite, or illite types, in response to the finer particle size distribution and greater dispersion of the montmorillonite minerals. Johnson and Davidson (1947) determined the maxi- mum unconfined compressive strength of sj'ntlietic soils containing calcium kaolinite, sodium kaolinite, and sodium bentonite, with results as follows : Mnj-imutn compressive strength, psi At optimum After air drying moisture and (at about 70° F ) Mijrture maximum density for 7 days Calcium-kaolinite mixture 14.2 92.4 Sodium-kaolinite mixture 20.2 138.5 Sodlum-bentonite mixture 70.8 952.5 The increased strength of the sodium-kaolinite mixture in contrast to the calcium-kaolinite soil, is regarded as due to greater dispersion and hence more effective dis- tribution of the sodium clay component. The much greater fineness and dispersity of the sodium bentonite continues the same trend, the clay mineral being very efficiently distributed as a binding agent over sur- faces of granular constituents of the mixture. With di\v- ing, the kaolinite mixtures increase six to sevenfold in strength, whereas the bentonite mixture increases more than thirteenfold, a clear demonstration of the effective- ness of the compressive action of restricted moisture films in reinforcing the specimen. This type of structural control of strength of earth materials is exemplified by loess (figure 29). In tests of sodium bentonite and bentonite in which the sodium is replaced by calcium, Samuels (1950) has demonstrated the relationship of compressive strength to moisture content. At a moisture content less than 50 percent of dry weight, sodium bentonite shows the greater strength ; with moisture content between 50 per- cent and 75 percent, the strength is about equal; and at moisture content over 75 percent, the sodium bentonite again exhibits the greater strength. The comparatively high compressive strength of the sodium bentonite at moisture content below 50 percent probably is the result FlGtTRE 29. Photomicrograijh!, iil tliiu sections shciwiiis faliric (if loes.s (upper left), clayey, silty, fine sand (upper rifrht), remolded loess (lower left), and clayey sand (lower right), (load-consolida- tion curves a, e, e, and f in fig. 30). Nicols crossed. Magnification 60x. Part V Clay Technology in Soil IVIechanics 229 5 10 20 50 100 200 500 COO LOAD-PSI Fkii'rk ?A). ('onsolidiilion of undisturbed loess, remolded loess, :iud siiud with londing juul wetting (cf. fis. 29). of cohesiou of the specimen produced bj' tension iu wa- ter films. With moisture content in the range 75 to 143 percent, calcium bentonite approaches the liquid limit ; whereas throughout the range of moisture contents in- vestigated (up to 143 percent), sodium bentonite alwa.vs is well below the liquid limit. Studies by Hendricks, Nelson, and Alexander (1940) on Missi.ssippi bentonite modified by excliange of various cations, demonstrate that calcium moutniorillonite con- taining 36 percent of water is in equilibrium with an atmosphere at 90-percent relative humidity (table 4). Consequently, at and below this water content, the wa- ter is adsorbed and does not possess the properties of free water. This is consistent with Samuels' (1950) observation of modest compressive strength (28 psi) at water content slightlj' above this value (40 percent of 1000 10000 Figure 31. 10 100 TIME (MINUTES) R.-ite of eonsolidatiou of kaolin and Wyoming ben- tonite modification (Samuels, 1950). dry weight). However, sodium montmorillonite contains only 28 percent of water while in equilibrium with air at 90-pereent relative humidity, and the compressive strength of the sodium-bentonite tested by Samuels is seen to decline precipitously at moisture contents just above this value (44 to about 60 percent). If the calcium bentonite were as fine as the sodium modification, com- pressive strength of the calcium modification would be considerably higher than that of the sodium type. With moisture content equivalent to one-fourth that of the respective liquid limits, the strength of natural ben- tonite is about 2 psi ; whereas, tliat of the ealcium modi- fication is about 23 psi. Consolidation Consolidation of the earth materials with loading represents adjustment of the internal structure of the solid framework by rotation and sliding of the consti- tuents and displacement of fluids (air or water or both) from the consolidating mass. For natural earth mate- rials, consolidation as a result of structural adjustment is rarely large and is significant only in materials of very low unit weight, such as loess, volcanic ash, and organic soils. The rate and magnitude of this consolida- tion are increased greatly by wetting and conscfjuent decrease in shear resistance of the soil constituents, especially if expansive clay minerals are an essential binding agent. Curves in figure 30 demonstrate the re- sistance of dry loess to consolidation with loads as high as 100 psi, and the great consolidation of the material with wetting while under load. The related fabrics are illustrated in figure 29. The low clay mineral content and high void ratio permit rapid drainage of pore fluids with consolidation of loess. At higher content of clay minerals, the size of voids and their continuity are decreased so that pore pressures develop unless application of load is very slow. Consoli- dation is dependent mainly upon drainage of air and water from voids. Consequently, permeability is the critical control of consolidation rate. Samuels (1950) found consolidation rate of Wyoming bentonite to in- crease in order Na < Ca < Al < Th, and consolidation rate for kaolin to be Na < Ca (figure 31). This sequence is the order of increasing permeability' in these clays at this loading (fig. 21). As reported by Preece (1947), Cooling has demonstrated the dependence of consolida- tion of bentonite upon exchangeable cations. At given load up to 5,0 tons per square foot, void ratio decreases in the order sodium > calcium > thorium > aluminum. At slightly greater loads, the void ratio of all modifica- tions is in the range of 1.3 to 1.5. The compressibility is greatest for sodium and decreases in the order sodium > calcium > aluminum > thorium, the latter two ex- hibiting similar rates. Similar data for beidellite soil were developed by Winterkorn and Moorman (1941), the consolidation varying as follows : sodium > calcium > magnesium > potassium > hydrogen. With normal fabrics, earth materials containing kao- linite or illite as the predominant clay mineral consoli- date less than do those containing members of the mont- morillonite group, Samuels (1950) demonstrates the lesser consolidation of kaolin in contra.st to bentonite (fig. 32), the change of void ratio for the sodium kaolin with loading from to 8 tons per square foot 230 Clays and Clay Technology [Bull. 169 14.0 12.0 10.0 O 20 I— < o > SODIUM \ \ \ \"' Bentonite \ >\ "v. -..._ _ Koolin--^ 1 i \ CALCIUM ...Bentoni+e ^ 1 ^,_ '^JKaolin 1 ALUMINUM V 1 .-•Kaolin ^ =ssi^^^_i - Ben-tonite in 0.0 THORIUM 30 V ,-Ben+Dnite .x..__^ 1.0 EFFECTIVE PRESSURE ( TONS / SO FT) Fldi'iH': 'A'2. Uelatiun of t'xoli;iiiut"ilile cation ti> cuii.soliilalinu nf li.iolin and \Vyoniin.;i hentonite { S;nniU'Is. 19.")0) being only ^^4 that of the sodium bentonite. The com- pressibility decreases in the order aluminum > hydro- gen > calcium > sodium. This sequence is essentially the rever.se of that exhibited by bentonite, the difference apparently arising in the distinctly different proportion of particles less than 1 micron in the cationic modifica- tions of the two soil types. For the bentonite, the fraction smaller than 1 micron is about 90 jjereent for the sodium modification, 81 percent for the calcium type, and 60 per- cent for the aluminum type. For kaolin, the less than 1 micron fraction i-anges only from 82 percent for the hydrogen and aluminum types to 88 percent for the sodium type. Thus, for the kaolin, the influence of de- creasing zeta potential from sodium to alumininn is not masked by widely differing particle size. The initial void ratio increases from the sodium modification of the kaolin to the aluminum modification because of increasing ran- dom aggregation with increase in attractive forces be- tween the particles. The high initial void ratio of the aluminum modification correlates also with the higher liquid limit (figure 19) and probably with a higher coii- FlGURE .33. Settlement of a l>uildinK as the result of consolida- tion of inontmoi'iUonite clays in the foundation, Me.xico City, Jlexico. Courtesy of A. E. CnnuniuKs (1!)47). tent of pore water. This jxiint has been empliasized by Johnson and Davidson (1!)-17). Earth materials containing montniorillonite-type clays may consolidate or cxjiand with addition of water, depending upon the adsorptivity of the clay mineral and its swelling potential, the original density of the mate- rial, and the load. P^r example, the sodium bentonite from the Mowry formation, Tiber Dam site, increased from 109 jionnds per cubic foot to 112 pounds per cubic foot, with loading to 4.3 psi at natural moisture content. Witli addition of water, the unit weight decreased to 111 pounds per cubic foot as a result of hydration of the montmorillonite constituent. With decrease of load to 3.5 psi, the sample expanded to 9-1 pounds per cubic foot. With increments of load above 45 psi, after addi- tion of water, comparable samples increased as much as 7.5 pounds per cubic foot during increase in load to 220 psi. Most spectacular of settlement problems in engineer- ing are found in Mexico City, Mexico, which is under- lain by layers of montmorillonite clay, tuff, gravel, sand, silty clay, and boulder clay. Zeevaert (1947, 1949) re- ports progressive settlement of the area of the city at an average rate of 5 inches ]icr year. M;iny large buildings have settled as much as 10 feet since constrtiction (fig. 33), as a result of compression of zones of soft to stiff claj', which are as much as 130 feet thick. Typical water content of the clays ranges from 500 to 700 jiercent of dry weight, and void ratio is as high as 14.0. Rate of settlement has increased since ancient times not only because of loading incidental to growth of the city but also because of drainage of groundwater. Removal of groundwater increases loading of deep-lying strata by loss of buoyancy of zones emerging above the water table and increases the ease of drainage of loaded clay layers. Successful protective measures so far dejiend upon design of the foundation, compensation for loading by variable excavation and use of piles to equalize settlement for each structure (Zeevaert, 1947; Cum- mings, 1947). Loe.ss deposits also are subject to great consolidation although their granular fabric is very different from tliat of montmorillonite clays and shales. Their unexpectedly high compressive strength while dry is readily lost with wetting because of breakdown of the argillic intergran- ular braces of montmorillonite. Tender the load of a 50- foot embankment at Trenton Dam, Nebraska, wetted loess consolidated a total of 1.54 feet dtiring the fir.st 2 months following completion of construction. Swelling with Hydration Adsorption of water by clays leads to expansion or swelling. Its magnitude varies widely, depending upon the kind and amount of clay minerals present, their ex- changeable ions, electrolyte content of the aqueous phase, l)rirticle size distribution, void size and distribution, the internal structure, water content, superinqiosed load and other factors. Two mechanisms are involved: (1) a re- laxation of eff'ective compressive stress related to en- largement of capillary films and (2) osmotic imbibition of water by expanding lattice clays. Expansion of clays as the result of enlargement of cajtillary films is essentially tiie reverse of drying shrink- age, althouoh the original \oluinc at given water content Part V] Clay Technology in Soil Mkchank -^ 1 231 -^ COMPOSITION OF MIXTURES { Percent by weight) 1. Wyoming Ben+onite 2. 25% Bentonite,75%Sand 3. 10% Bentonite, 90% Sand 4. Kaolin 5. 25% Kaolin, 75% Sand 6. 25%Bentoni+e, 25% Kaolin, 50% Sand 7. 10% Bentonite, 40% Kaolin, 50% Sand LOAD = I RS.I. SPECIMEN CONFINED LATERALLY 8 16 Tl ME 24 (Days) 32 36 Figure 34. Volume cIi.-iiikc i>f lientonite-kaolin-saiid niixturcs with wi'ttiu^ usually is not recovei-ed because of periuauciit changes in internal texture and structure during tlie shrinkage process. In tests of calcium and sodium kaolin — syn- thetic .soil mixtures at optimum moisture content and maximum densit}% Johnson and Davidson (1947) ob- served water absorption of 1.5 percent and 1.7 percent, respectively, during a 4-day soaking period with result- ing linear expansion of 0.12 percent and O.O.So percent, respectively. L'nder identical conditions Wyoming ben- tonite expanded 0.94 percent while sorbing 2.4 jiercent by weight of water. Tests of synthetic mixtures of Wyoming bentonite- kaolin-sand, reveal the expansive potential of sodium bentonite. Under a load of 1 pound per scpiare inch, with lateral restraint, ^\\voming bentonite expanded 66 per- cent during 33 days of sorption of water and was ap- parently still expanding at the conclusion of the test (fig. 34). A mixture containing 2o percent sodium ben- tonite expanded 0.94 percent while sorbing 2.4 percent under the same conditions. In a test of a mixture of 25 percent bentonite. 25 percent kaolin, and 50 percent fine sanil, the specimen exi)anded 33..") |)crccut. The greater expansion of the latter mixture in contrast to the first mixture arises in a more dense fabric and hence greater effectiveness of given expansion of the bentonite constituent. The data demonsti-ate similar relationships for the mixture containing 10 percent bentonite. With sorption of water by expanding lattice clay min- erals, such as members of the montmorillonite group and some types of illite, expansion as a result of enlarge- ment of water films is supi)lemented by expansion as a consequence of adsorption of water along the 001 inter- layer space and upon edges of broken crystals. The capillary ^compressive forces increase inversely as the radius of the capillary decreases. Consequently, the ex- pectable compressive forces are greatest for finest -grained materials. Therefore, the greatest expansion is expected from niontmorillonite-type clays, not only because of expanding lattice but also because of relief from higlaer compressive stresses. With sifting of materials containing montmorillonite into polar liquids, expansion takes place, the magnitude 232 Clays and Clay Technology [Bull. 169 Table 13. Free aivell data on clay and clay-like tninerah. Clay minerals Location Free swell » in water (percent) 145 125 Willow Creek Dam, Colorado 95 45-85 Na-Montmorillonite _^ 1400-1600 Na-Hectorite 1600-2000 Ca-Beidellite - Friant-Kern Canal, California Weil ton-Mohawk Canal, Arizona Wellton-Mohawk Canal, Arizona 80 Na-Beidellite 110 310 K-imte 115-120 60 15 Grand Coulee Dam, Washington 120-140 5 60 TTftlIny<;itf Langley, North Carolina Santa Rita, New Mexico 20 70 Muscovite . . North Carolina 25 Vermiculite Pyrophyllite Libby, Montana 50 40 Talc Serpentine.- . . Providence. Rhode Island Warren County, New York 30 10 ■ Test basoil upon swell in water of 10 cc of crushed material passing No. 30 screen and retained on No. 50 screen. ** Ca-inontniorillonite prepared in laboratory from Na-montmorillonite (bentonite) . <^ Contain montmorillonite, illitc. jefferisite, and clilorite. ■* Contains small amount of montmorillonite. of which is dependent upon the amount of clay present, the composition of the clay, the presence or absence of cementing- materials, the exchangeable cations, electro- lyte content of the liquid, the molecular size and struc- ture of the licjuid, and the interaction of molecular forces of the liquid and solid. For the clay minerals, the free swell in water is in the series montmorillonite > illite > halloysite > kaolinite (table 13). The free swell of beidellite is distiiictl.y less than that of montmorillite and commonly e(iual to or greater than that of illite. For materials such as pyrophyllite, talc, and museovite, the reported free-swell values are probably too high, for the platy shape of the particle prevents efiScient pack- ing of the sample. With claj^s of montmorillonite type, the swelling de- creases greatly with exchange of sodium for other univalent ions or for divalent or trivalent ions. For Wyoming bentonite, Baver and Winterkorn (1935) observed decreased swelling in the sequence : sodium > lithium > potassium > calcium > magnesium > hydro- gen. For beidellite soil, the sequence was found to be lithium > sodium > calcium > barium > h.ydrogen > potassium, the latter demonstrating binding of the lattice by potassium in the 001 interla.yer space of beidellite. A series of free-swell tests were performed using Na-montmorillonite (Wyoming bentonite) in various liquids (table 8). No definite correlation between the dielectric constant and the degree of swelling was TIME AFTER INTRODUCTION OF WATER (UIN) Figure 35. Uplift pressure developed by hydr.ation oi AVyoming bentonite and an altered rhyolite containing montmorillonite, after initial loading (after Holland, personal communication). observed, althougli the swelling tended to increase with an increase in the dielectric constant. The particles of montmorillonite remained intact in most of the liquids. Very slight dispersion occurred in carbon tetrachloride, diethyl ether, acetic acid, acetone, and glycerol. Dis- integration was almost complete in water and form- amide. Examination of table 8 indicates that the dielectric constant of the liquid will not account alone for the variations in swelling. Of significance also are the dimensions of the liquid molecules; the ability of the liquid molecules to build layers between the packets of the clay mineral crystals; and characteristic molec- ular forces, such as distortion polarization ahd orien- tation polarization, both of which influence the dipole moment of the molecule. In a moist atmosphere, montmorillonite soils adsorb water, and the crystals swell as the result of hydration of certain exchangeable cations and development of layers of water molecules in the 001 interlayer space. Hendricks, Nelson, and Alexander (19-10) found the relative adsorption of a Mississippi montmorillonite clay to be as follows (table 4) : Relative humidity, 5 percent — magnesium > calcium > lithium > strontium > barium > hydrogen > sodium > cesium > potassium Relative humidit.v, 40 percent — calcium > strontium > magne- sium > hydrogen > lithium > barium > sodium > potassium > cesium Relative humidity, 90 percent — calcium > hydrogen > lithium > strontium > magnesium > barium > sodium > potassium > cesium The avidity for water of magnesium, calcium, lithium, strontium, and barium clays at very low humidity is evident and is correlative with hydration of the ex- changeable cations. The work demonstrated that sodium, potassium, cesium, and hydrogen ions are not hydrated. With increase of relative humidity from 5 percent to 90 percent, the adsorption by the modifications contain- ing nonhydrating cations increases eightfold to thirteen- fold, whereas the adsorption by modifications containing hydrating ions increases only fourfold to sevenfold. The increase in adsorbed water correlates with an in- crease in the dimension from a fraction of an Angstrom iinit for the potassiiun modification to 5.7 Angstroms for the sodium type, the latter correspond- ing to an increase of 58 percent in the volume of the Part V] Clay Technology in Soil IVIechanics 233 unit cell. At saturation without loading, the "c" dimen- sion exceeds 30 Aiijrstronis and dispersion of the sodium 1o individual packets follows with ag:itation. For hydro- gen and calcium modifications, the "c" dimension docs not p:reatly exceed 20 Angstroms, even in a saturated atmosphere. With immersion of sodium moutmorillonite in dis- tilled water, complete dispersion does not occur in the absence of ajritation. Rather, a maximum swellinp: occurs when equilihrium is achieved between the attractive forces between the plates of the clay minerals and the osmotic pressui-e developed in the intervenin-s are sub.iected to initial loads of high magnitude, tremendous swelling pressures are developed. Unpublished data obtained bv W. Y. Holland (fig. 35), in the Bureau of Reclamation labora- tories during design studies for Davis Dam, Arizoua- Xevada, demonstrate pressures as high as 540 psi de- veloped in Wyoming beiitonite originally compressed at 5,000 psi and subsequently wetted following a period of unloading sufficient to permit relaxation of the specimen. An altered rhyolite containing high propor- tions of montmorillonite developed pressures in excess of 330 p.si under the same conditions. Initial loading of the clays at 3,000 psi instead of 5.000 psi produced lesser swelling pressures. Consolidometer tests of undisturbed clays and shales containing illite or montmorillonite-t.vpe minerals in the clay fraction indicate potential hydration pressures as high as 15 tons per square foot (table 14). Expansive cla.vs and shales have caused significant distress in engi- neering structures in many parts of western United States and also in other countries, notably South Africa, Palestine, Burma, Australia, and India ("Wooltorton, 1950; Jennings. 1950; Holtz and Gibbs, 1953; Felt, 1950). Distress arises with introduction of water to the foundation materials after con.struction. Water can be inti-oduced by hydraulic flow, capillarit.v, vapor transfer in response to gradient in relative humidity, and thermo- osmosis. The action commonly follows seasonal c.vcles of wetting and desiccation (Wooltorton, 1950). Criteria for recognizing expansive soils from soil mechanics properties are described by Holtz and Gibbs (1953). Spectacular distress is widespread in Texas in the vicinity of Aiistin and Dallas whei-e the Eagle Ford shale and Tavlor marl underlie the surface (Felt, 1950; Gieseeke, 1922; Simpson, 1939: Daw.son, 1953). The Eagle Ford shale contains about 35 percent of clay min- erals including calcium-montmorillonite. illite. and kao- |t-t-— : ^ / / House M,-/ / ^ ^ House ^^y^ ^ A ^^y • Mouse 95/'''^ ^ /^ ""^ .-House 23 ^ .1 ! , 1 , , FiGUKE 36. Uplift of buildings by expansion of illite clay as the result of hydration, Pretoria, South Africa (.T. E. Jennings, 1950, with additions). Unite (fig. 14). A very high swelling pressure arises in the combination of the swelling clay and a compact fabric which develops full eft'ectiveness of the expansion of the moutmorillonite constituent. Drying as the result of exposure and growth of vegetation extends to depths as great as 15 feet. Dawson (1953) reports differential uplift as much as 2.4 inches of structures with result- ing cracking of walls and foundations and misalignment of doors and M'indows and other ilistress. Satisfactory foundations thus far require the founding upon piles penetrating to stable material below the weathered layer. Similar distress related to a calcium-illite soil are re- ported bv Jennings (1950) in Pretoria and adjacent areas of South Africa. Uplifts as much as 6.4 centimeters in about 3 years are reported for light structures, the uplift relating only generally to incidence of rainfall (fig. 36) because of extraneous supply of water to the foundation by drain pipes broken as the result of move- ment of the soil and buildings. Uplift typically is a dom- ing action with maximum deflection occurring in the center of the building. Shattered slickensided clay af- fected b.y drying and atmospheric temperature varia- tions extends to depths as great as 27 feet. In addition to hydraulic flow the increase of moisture content re- sults from' vapor or capillary transfer, possiblj' in re- sponse to reduced temperature in the ground beneath the building in contrast to the temperature of the sur- rounding ground at equivalent depth. Jennings reports a temperature difference of 5° F. The uplift probably is the result of both expansion of clay minerals and relief of capillary stresses. The largest reported uplift resulting from swelling of expansive clays or shales is described by Mielenz and 23-1 Clays axd Clay Technology [Bull. 169 Figure 37. Displacements of concrete lining by swelling of bei- dellite clay in the suBgrade. Friant-Kern Canal, near Lindsay, Cali- fornia. Okesou (1946) in an in\-esti,oation of foitiidation dis- placements on the Malheur River Siphon near Cntario, Oregon. Seepage of irrigation water and run-off into parts of the foundation resulted in progressive but ir- regular uplift whieli loeally exceeded 1 foot within 3 years after eonstruction. The sodium bentonites of the Idaho formation developed uplift pressures of 7,200 pounds per square foot and at one point lifted not only the siphon but 38 feet of alluvial sand and silt as well. Where saturated prior to coustruetion, the foundation yielded to small consolidatioiL Most susceptible to uplift and displacement are light slabs such as highway pavements, canal linings, spill- ways, and basement floors (figs. 37 and 38). Distress along the Friant-Kern Canal involves the Porterville clay. Heaving and cracking involving side slopes only occurred along the Delta-Mendota Canal, California, in sections traversing the exjiansive clays of the Tulare formation. Dift'erential uplift of 3 to 7 inches for con- crete slabs are connnon in numy parts of the western United States, and are reported locally to exceed 12 inches in areas of Texas underlain bv the Tavlor marl (Felt, 1950 and 1953). Figure 38. Cracking and bnlging of concrete lining by .swelling of beidellite clay in the subgrade. Friant-Kern Canal, near Lindsay, California. Remedial measures are difficult and expensive in most situations. Three approaches are possible: (1) pre- venting increase in moisture content beneath the struc- ture by moisture barriers, drains, and vents; (2) pre- wetting to equalize moisture content and reduce later expansion of foundation materials; (3) excavation of the expansive earth materials and their replacement by stable materials in sufficient depth to minimize uplift by the remaining foundation materials and (4) chemical treatment to prevent penetration of water into the foun- dation or to reduce swelling characteristics. Moisture barriers and drains are feasible if the offending waters penetrate from the surface or move in well-defined sub- surface channels. Unfortunately, however, many exam- jdes of distress relate to vapor or capillary transfer vertically or laterally from zones of higher water con- tent. Control of such moisture movement frequently is economically unfeasible so that appropriate measures require either special design or expedient repair. As a result of extensive studies of differential uplift of highwav pavements overlving the Tavlor marl near Austin and Dallas, Texas. Felt (1950, 1953) concluded that severe heaving can be prevented by prewetting by ponding areas underlain by dry expansive soils. It Avas found necessary to continue the ponding at a given site for 1 month or longer. During ponding the moisture content of subgrade materials increased 6 percentage points on the average (from 13 to 19 percent) in the top 7.5 feet. Satisfactory wetting of subgrade materials could not be achieved by introduction of water at 4-inch- diameter drill holes on 5- or 6-foot centers. Minor heav- ing was found to be controlled by early clearing of vege- tation from the right-of-way and proper •control of moisture content and density of subgrade materials dur- ing grading operations. Overexcavation and replacement of expanding clays with stable material is effective only if the ujilift pres- sure is so small that loading sufficient to maintain njilift within design limits can be accomplished with moving of a reasonabl(> amount of earth. This is not true for many structures where uplift pressures of several tons per s(iuare foot occur in the foundation materials. How- ever, uplift pressure is reduced greatly with slight ex- pansion of clays and shale ; and moreover, expansion is reduced greatly by even slight increase in loading (Holtz and Gibbs, 1953). Experience with fotuulation design of structures located upon expansive clays and shales is described bv "Wooltorton (1950), by .Icuuiugs (1950), and by Holtz and Gibbs (1953). Dramatic results have been obtained in laborator.y tests by chemical treatment of expansive soils. T'nfor- tunately. introduction and distribution of significant quantities of treating agents into the formation in place partii-idarly the relatively impermeable expansive clays and shales has so far proved impossible. Davidson ( 1949 ; and Glab. 1949) has reduced swelling pressure of the montmorillonite soil by factors ranging from 4 to 18 by introduction of 1.29 to 9.66 percent of Armac T, a water-soluble cationic amine acetate. In other work, calcium hydroxide admixed with montmorillonite and beidellite clays has proved effective in greatly reducing hydration expansion. Goldberg and Klein (1953) demon- strate reduction of swelling pressure from about 9 to about 5 psi by admixture of 8 percent of calcium hy- droxide with sodium bentonite, and from about 7 to 1.5 l^irt V (■|.A\ 'I'll II \ill.(ii,\ IN Soil. MllllWK |isi for I III- l)riili'liili' i-niitaiiiiiii; I'urliTx iilc i-hi.\- > tin. :ii' ' . Siinil;ii- ctl'i'cts ;iri' .iilaiiii'i! liy llir iiil rmltirt inn of ili\aliMiI or tri\ali'iil iiioi'janii- i-alioii-. In i'\ paiivix .• ria.x in wliirli Miijinni i-. lln- ilmiiinanl rNi-lianuralilr i-alioii. Laws aiiil i'a;.:i- IVIii arliir\i'il fa\oralili' rrvnh-, in I riMl iiM'iiI of kaoliiiiti' anil illilr voils willi soilinni ^ili i-ali'-.. liiil llir swi'llinu ot' nioiil niorilloniti' was iiirrrasi'il. I III roilnrt ion of siahilizinu i-alioiis into iornialions in piai-i' ronlil III' ari-oniplislii'il In a ili"_'i- illirr liy iiy- ilraiilir injfclion or liy rln-l rin-hrniiral 1 rral ini'iit. ijow- .■\rr, sm-i-rssfnl Irralnn'iil ol' si.; nilii-anl inassrs of clay oi- shall' in plaro lias yrl lo lir i li'iiiinisl ratnl. Drying Shrinkage I'ryinu' of oi-i'jinall,\ inoisi i-la,\s jraiK to a iliiiiinnlioii of \o|iinio as a ii'siill of 1 ili'i-roasiirj sizi' of tlir unit ri'll of .■Npaiiilini; lallirr iniinTals if pirsiMiI . ami L' inrrrasiiiii' roinpri'ssi\i' slrrssrs ori'j inat ini: ni rapillarx foi-ci's anil ai|so|-li,-i| watiT lilins. ( 'la\ s of ihr kaolinili- 'jronp. ami to a li'ssci' I'Nlriil Ilir illili' 'jroiip, show liirlr or no volninr i-liainji' of iniluiijna! i-i'\sials ami rssni- lially all of tin' ilryimj shrinka'jr arisrs in roiiipi-i'ssi\i' capillary forces ami cNlcmally ailsorlicil walcr lilnis. I'mlcr iilcal comlil imis. ihc capillar\ forces arc invci-scly proportional to ihc railiiis of the channclw a\ s pcrnical- iiii;- the clay. Sim-c the size of the small. 'sl capillary inlimalcly |ieiiet rat ini:' the cla\ will I n the sai inlei' of size as t lie liiiest part ides ami i|siiall\ will lie sniallei\ the coinprcssi vc capillary forces increase rapiilly with (lecrcasiiic- particle size. In tests of syiitlietii- calcium ami simIiiiih kanliinle soils .lohnson anil l)a\ii|soii llilT' oliservcil hiijher I WYOMING BENTONITE Co(OH )C0MTEMT0% .■\pansi\c sniK are ilit'ticiilt to reproiliice lieeaiise of i r rc.j 11 1.1 r fractnrine nf ihe ma tcrials. W'interkorii ami .Monrman i I'MI > report a raicje of shriiika.je limit of a li.'nlellitc soil from lit. I to 1 1 ..S percent 111 the foMowin.j oriler: pottissiiim .■ aliiminiim > liyilrii.jeii in.i'j iiesi iini c;i leiiiiii ■ sodinni. Shriiika'je ratio \arieil from •_^(l^ to l.sd in the order soiliiim ■ iiiaLinesinm ■ ealciiim • aluminum ■ liy- ilriiijcii ■ polassium In tests of moililical imis of Wyo- miiiL' licntonile. Samuels 'l'i."iii foiiml the shriiik.'i'ji' limit to he as fi.llows; calcinm ■ siMlium ■ al iiiiii mint for spei-imciis colli a 1 II iii.. .Iry \iil|||iics of .'ili.i, .ili. ami ."i:; s re |!ei- hllllilreil .jrapis i|r\ weijllt. Still!) of shriiika.je of syulhciic miNtiircs of soiluini heiitoiiite ami kaolinile willi tine saml re\ea|s suj u ilica lit iiillneiices of fahric upon \oliime chaime with i|r.\ii|.j- (tallies 1(1 anil 1 .'i > , In the heiil mi it e saml mixtures. i|e \elopiiieiil of jraiii lii.jraiii i-onla.-l pr.'\enls e\eessi\i- shriiil- .■..literil \\\\\ r.-iii,^.- ail in.-ri-,-o.' in \..liini.. ..I lli-- lea-,:--. Shrinkai;.- rail., is tlie l.nlk siii-.ilic Kiavily ..f ll..' .iii.-.l -<.il c-. •! in flelirininiim tlie -linnUaK.' Iilnil (.\ll.-n. lul:'' 236 Clays and Clay Techxology Tahle 1^. Uplift pressure observed on various expansive clay and shales. [Bull. 169 Clay mineral Petrographic type Geologic formation and locality Maximum observed viplift pressure" (tons/sq ft) Idaho formation, near Ontario, Oregon 3.6 Shale- - 2.0 Clay- 11.0 Shale 15.0b Clay Porterville clay, near Porterville, California 10.5 Mixed cation montmorillonite__ Clay - Alluvium, near Patterson, California 8.0 Clay -.. Alluvium, near Antioch, California 4.9 Ca-IUite Clay 5.00 J* Load required to prevent vertical expansion of laterally confined specimen. ■^ From Dawson (1953. in press). «■ From Jennings (1950). tions of kaolin do not fill the intergraimlai- spaces. Coat- ings on the grains are poorly developed. Voids content at the shrinkage limit of bentonite- kaolin-sand mixtures increases progressively, although slightly, with decrease of bentonite content from 47.5 to 40 percent. The values obtained in this range are similar to the .shrinkage limit of bentonite alone be- cause of separation of the granular elements of the fabric. With bentonite content of 25 percent (kaolin consti- tuting 25 percent and sand 50 percent by weight of the mixture) shrinkage limit is high (24.4 percent), as the result of stabilization of the fabric by grain coatings composed of montmorillonite, and the presence of inter- mixed montmorillonite and kaolinite in intergranular spaces. Although the shrinkage limit of this mixture almost equals that of the bentonite-sand mixture con- taining 25 percent of bentonite, the shrinkage ratio is much higher. This relationship demonstrates the re- straint of internal shrinkage cracking by mixed ex- pansive and nonexpansive clays in the fine fractions of a sandy soil. Such a mixture is capable of excessive expansion with rehydration (fig. 34) and development of swelling pressure. With kaolin content above 25 percent and bentonite content less than 25 percent, both shrinkage ratio and shrinkage limit are reduced. This relationship indicates some restraint of volume shrinkage by the sand and coarse fractions of the kaolin, but it shows as well the development of internal cracking in the soil mass. Such a mixture is capable of but limited expansion with re- hydration (fig. 34). Shrinkage of earth materials can be altered by treat- ment with organic surface-active agents. In tests of a kaolinite soil, Davidson and Glab (1949) decreased shrinkage by admixture of a fatty acid amine acetate, whose dissociation in water leads to release of a large organic cation according to the equation RNHsAc ?^ (RNHa)* + Ac". With addition of the agent in amounts up to 3 percent (by dry weight of the soil), shrinkage limit increased progressively from 14.9 to 25.1 percent, and shrinkage ratio simultaneously decreased from 1.77 to 1.46. In similar tests of a montmorillonite soil, Davidson (1949) produced a decrease in shrinkage by use of six water-soluble cationic organic compounds.* Eosin Amine-D Acetate and Ammonyx T were particularly effective, the shrinkage limit being increased progres- sively from 6.0 to about 37 percent and 28 percent by addition of 16.11 percent and 13.08 percent by dry weight of the soil of the two compounds, respectively. Shrinkage is critically important in the development of properties of earth materials and in engineering per- fcrmaiice. Excessive shrinkage accompanying drying of cohesive earth materials containing expanding lattice clay minerals leads to the development of crumbs or aggregations, and thus gives rise to fine and medium argillic fabrics. These are exemplified by the beidellite- eontaining Porterville clay (figs. 6 and 11). Drying in expansive clays can occur to depths in excess of 25 feet. Wooltorton (1950) has summarized the role of drying shrinkage in control of fabric and permeability of ex- pansive montmorilloiiitie and illitic soils of Burma, and the associated distress of engineering structures. Cracking as the result of drying shrinkage of earth materials leads to disruption of structural integrity of the mass (fig. 40). Drying is accelerated by the ac- companying increase of exposed surface, and water easilj' penetrating to depths following precipitation re- sults in excessive slaking and expansion in the forma- tion in place. Such cyclical volume change accelerates creep of superficial deposits on slopes. Moreover, loss of cohesion within the mass can aggravate development of landslides, particularly following long periods of drought (Ladd, 1934). Slides of this type have been observed in canal sections lined with expansive clays exhibiting high drying shrinkage. Drying cracks ex- tending to depths of several feet commonly become par- tially filled with sand and silt from the surface. With expansion of the clay, during, rehydration, the jacking action can be destructive to adjacent retaining walls or foundations. Shrinkage of soils adjacent to structures decreases support, as, for example, under highway pavement or canal linings. Certain montmorillonite "adobe" soils in central California recede as much as 2.5 inches from adjacent concrete during dry weather. As a result, dif- ferential settlement of the structures is a common phe- * Armac T, Armac ISD, Armac 12D, Rosin Aniine-D Acetate. Amine 220, and Ammonyx T. All yield monovalent cations on dissolu- tion in water, except Amine 220 whose cations are divalent. I'arl VI Clay Technology ix Soil Mechanics 237 -*• — FiGURK 40. Shrink;!;;*' cra-ivs m iiir t■\|M)^.•^| Miii;irf Hi un_- i'or- terville clay, near Lindsu.v, Califi)iiiia. The cracks are easily visible to depths of more than two and a half feet. nonienon. Resultiiig eraeks permit excessive penetration of water from the surface into foundation or subgrade materials. Moreover, the linear shrinkage can cause cracking of asphalt membi-anes designed to prevent seep- age into subgrades. For example, shrinkage of the mout- morilloniti'-containing Denver and Laramie formations, near Denver, Colorado, necessitates repeated resealing of edges of pavements to maintain adequate surface drain- age. Excessive shrinkage of clays destroys many types of protective coatings on buried pipe installations. Syneresis Syiieresis is the spontaneous separation of an initially homogeneous colloidal sj'stem into two phases — a co- herent gel and a liquid (Heller, 1937). Shrinkage of the gel phase is equal to the volume of the liquid ex- pelled (P"'erguson and Applebey, 1930). Syneresis is reversible unless coagulation of the colloidal phase pro- ceeds too far (Heller, 1937). Three varieties of synere- sis were recognized hy Heller (1937): (1) Syneresis by reduction in swelling as the result of dehydration of hydrophilic particles; (2) syneresis by agitation through development of gels from dense geloids settled by grav- ity or centrifuging, or through formation of an emul- sion of geloids; and (3) syneresis by contraction. The phenomenon is best developed in concentrated and strong gels and is pronounced in fine-grained sedi- ments, such as chalks, marls, and colloidal clays, espe- cially those of sodium montmorillonite type (Jting.st, 1934; Braune and Richter, 1949). Shrinkage as the re- sult of syneresis is accompanied by cracking, distortion, and hardening of the clay-water system. The process gives rise to an appearance suggestive of drying shrink- age. In laboratory experiments Jungst (1934) reports Assuring to depths up to 3 cm in sediments 12 cm thick, the cracks being as much as 3 mm wide at the surface. The sediments were continuously covered with water during the test. He also observed numerous special de- tails of the secondary fabric in clays, including mounds, pits, cones, grooves, and spirals, which might easily be mistaken for features of sedimentation or fossils in geo- logic formations. Considerable research has been per- formed in study of syneresis in silica gel, agar-agar, and dye stuffs (Heller. 1937; Ferguson and Applebey. 1930; Plank and Drake. 1947; Rossi and Mareseotti, 1936; Bonnell. 1932: 1933: 1933a: Gapon. 19.30). but little has been accomplished in elucidating this property in clays and shales. The rate and degree of development of sjnieresis de- pends upon the original concentration of the colloidal jihase, tlie temperature, the acidity or alkalinity, and the concentration of electrolytes (Ferguson and Apple- l)ey, 1930; Jungst, 1934; Braune and Richter, 1949; I'lank and Drake, 1947; Rossi and Mareseotti, 1936; Bonnell, 1932; 1933; 1933a; Gapon, 1930). In clays, fabric is important. Jung.st (1934) found that an inter- mediate content of colloidal clay and electrolyte is con- ducive to most rapid and pronounced syneresis. An ad- dition of sand to colloidal claj' increased syneresis, a ina.ximum being attained with 30 percent by weight. No shrinkage or expulsion of water was evident in clay- sand mixtures containing 80 percent or more of sand. Syneresis is a condensation of gels which, in clays, apparently demonstrates the drawing together of hy- lirated colloidal particles of clay minerals under the influence of van der Waals' forces and electrostatic attraction, with partial destruction of adsorbed water layers (Talmud and Suchovolskaya, 1931). In a clay- water mixture of low electrolyte content, fragments of crystals of clay minerals are surrounded by a diffuse cloud of cations held by the net negative charge on the lattice and by anions (especially OH ) fixed at broken prism edges of the lattice. "Water molecules are held in this zone in such volume as to cause osmotic equilib- rium between it and the surrounding free water solu- tion. For weakly held monovalent cations such as Na*, the diffuse cloud of adsorbed ions is thick and the re- sidual negative charge of the system is high. With more tightly bonded monovalent cations, such as IP and K*, especially with illite and beidellite after drying, and with divalent and trivalent ions, the diffuse cloud of ions and water molecules is less thick and the negative charge of the cla.v mineral is more nearly balanced by the charge of the cationic counterions. Similarly, addi- tion of electrolytes to the clay-water system will extract water from the diffuse cloud of cations and a new osmotic equilibrium will be established with a water hull of lesser volume and thickness. Although a net negative charge remains upon the entire hydrated clay particle (including both clay solid and counterions). the peripheral position of the cations causes a repulsive effect in the immediate vicinity of the peri]ihery of the adsorbed water hull toward cations in the free water or toward the cations similarly associated with other particles of clay in the clay-water system. In a suspension of Na-montmorillonite and distilled wa- ter, the repulsive forces at this peripheral position in the adsorbed water hull are sufficiently great to prevent coagulation. However, in clay -water S}"stems of more concentrated type, proximit}' of the particles gives rise to van der Waals' forces of significant magnitude, which tend to draw the particles together in spite of the repulsion of the diffuse cationic hulls of adjacent particles. Moreover, according to Hauser and Le Beau (1946) the free water contains a net excess of positive charge which repre- sents cations in equilibrium with cations within the adsorbed water hull upon the clay particles. They pos- tulate that these "interact and form groups of ions which have a greater charge associated with them. Thus. 238 Clays and Clay Teciixology Tnhle l.'i. Shriiikftfie 'mil ijihiiiskjii iif xi/iitlietic soih irith Ions iiiiil ii"i>i of wiiter ronlciit. [Bull. 169 Mixture (percent) by weight Drying shrinkage Expansion with wetting \Vyoniing Kaolin !■ Sandt Density (Ibs./ft.') Water content (percent) Shrinkage (percent) Density (Ibs./ft.') Wat«r content (percent) Expansion (percent) Total volume change •* bentonite" Initial Final Initial Final Initial Final Initial Final 100 2'> 10 75 90 50 50 75 70.2 107.0 108.3 105.5 106.7 108.0 97.8 38.1 19.7 19. G 20.1 19.5 19.3 19.3 29.6 5.4 2.3 11.5 12.9 11.2 —0.8 72.0 107.0 107.8 104.7 106.7 109.2 96.4 38.9 85.8 107.1 78.4 100.4 113.2 89.2 43.4 20.0 20.5 20.4 20.0 19.4 19.3 99.5 35.0 20.8 40.1 24.3 17.0 31.0 66.0 24.7 0.7 33.5 6.3 —3.5 8.1 95 6 112.9 111.3 117.9 123.1 121.7 97.5 1.9 0.8 2.3 0.6 0.0 0.0 30.1 3.0 2o 10 25 40 25 100 45.0 19.2 7.7 7.3 ' » Piedomiiiiiiilly N;i-iuontiiiorilloiiltt; from near Oy-lt**. Wyoming. ^ From near Bath, South Carolina. « Equal parts by weight of No. 50 to No. IIIO and minus No. 100 fractions of sand from Clear Creek near Denver, Colorado. "> Sum of shrinkage and expansion. these »roups may act as nuclei about wliieh the colloidal particle-s may condense, owing to forces of an electrical nature. On the basis of this idea, if orientation and condensation do take place, this will occur up to the point where the attraction forces between the basic cliarui' on the particles and the ions in the dispersion mediuiu are balanced by the forces of repulsion between the ions in the diffuse layer and the ions in the disper- sion medium. Thus, a rigid gel may form, which may be made fluid by redistributing the ions in the system." The redistribution might be aecompli.shed by shaking or stirring. To the attractive electrostatic forces mentioned by Ilauser and Le Beau (194(5) must be added the van der AVaals" gi-avitati(inal forces. The close approach of the particles will cause a contraction of the cationie diffuse layer of each particle. This contraction of the diffuse layer can be accomplished only by expulsion of water. This excess water is extruded from the clay -water mass as the volume decreases in the course of syiieresis. The electrolyte content of the water expelled during process of syneresis is different from that of the original suspen- sion (Rossi and Mareseotti, 1937). Glasstone (1946) believes the extruded water to be water held by capil- lary forces between heavily hydrated particles constitut- ing the framework of the gel. Much work remains to be done before the mechanism of syneresis and its causes and conditions are explained fully. Syneresis is a widespread jilienomenon whose develop- ment usually passes unrecognized because of its super- ficial resemblance to drying shrinkage. Its significain'c in engineering performance of earth materials remains to be evaluated. Failure of bentonite membrane linings, even while under water or buried beneath saturated earth, might occur as a result of syneresis. The phenome- non develops cracks which are so typical of claystone and shale formations in place, especially those containing members of the montmorillonite and illite groups. Syn- eresis is a common cause of difficulty in testing and con- ti'ol of drilling muds (Larsen, 1946). The relation of syneresis to other properties of clays and shales is dis- cussed at appropriate places in the text. Frost Heaving Frost heaving is raising of the ground surface by de- velopment of bodies of ice within the soil. The magnitude of the uplift depends upon the capillarity and trans- mittancy of the soil, the rate of freezing, surcharge or loading, and the ground-water conditions. Detailed studies of frost heaving are reported by Taber (1929; 1930) and by Beskow (1947; 1948) upon whose work many of the following comments are based. Taber demon- strated conclusively that fro.st heaving is caused by movement of water into the freezing zone in an' amount equal to the uplift. Expansion of water with freezing virtually is insignificant in tlie process. Indeed, lenses can be produced by freezing of soils saturated by liquids whicli contract upon freezing, such as nitrnbenzcne and benzene. Under given conditions of temperature, surcharge, and water availability, the rate and magnitude of heaving depends upon the fabric of the .soil (table 16). From detailed studies, Beskow (1947~l concluded that heaving will not develop in soils containing less than 30 percent by weight of fractions ])assing tlie 0.062 mm sieve or less than 55 percent by weight of fractions passing the 0.125 mm sieve. For most soils, heaving will not develop if fractions passing the 0.062 mm sieve constitute less than 50 percent of the material. Jlore fundamentally, frost heaving depends U])i)n the capillarity of tlie soil, that is its ability to sustain capillaries of considerable vertical coiitiiuiity. For the loose packimr achieved at the liquid limit, a capillary rise less than 1.0 meter identifies non- heaving sediments ; at maximum density of packing, the minimum capillary rise for frost-heaving soils was found by Beskow to be 1.25 meters. Fuder small loads, such as are sustained by mo.st soils at the freezing zone, capillary rise must exceed 1.5 meters with loose packing or 2.0 meters with packing at maximum density. Xormal frost- heaving soils sustain a capillary ri.se in tlie range 2 to 20 meters. Admixture of clay with saiul makes frost heaving possible (Plank and Drake, 1947). AVith a 5-pereent ad- dition of fat clay, frost heaving can occur; with 10- to 20-percent addition, frost heaving is well developed. I'ait V t'l.Av Ti;riix()i,oi;v ix Son. .Mkciianhs Tahle 16. delation of soil properties to frost licariitg 239 Soil group Average particle size (mm) Amount passing sieve (percent) Capillarity *» Nature of frost hea\'ing Soil type 0.062 mm 0.125 mm (meters) Sediment 0.1 30 55 <1 15 22 Froet hea\ing only superficial and with verv high ground-water table 0.1-0.07 30-50 i-iji 30-50 Frost heaxnng through road base with very high ground-water table 0.08-0.05 ! .-. -25 22 ■'.'• lJi-2}^ 15-25 22-36 <0.05 750 2-20 water up to IH meters for sediuients, and up to 1 meter for 725 730 Frost-heaving clays, but not liable to boils Sediment. 20-? Non-frost-heaWng stiff clays Sediment • From Beskon (1035). '' Capillary hright for packing at the Iinuiy this process large volumes of ice can accumulate in one or more lenses if the move- ment of the freezing isotherm is slow compared to the rate of introduction of water from an external source. If the rate of freezing is rapid, oulj' small lenses result or the mass is frozen homogeneou.sly before water trans- fer can be accomplished on a significant .scale. Clearly, even with a slow freezing rate, lens formation caii be restricted or prevented by iiiade(piate transmi.ssion of water to the freezing zone, either through want of water or low iiermeahility of the fabric. It is interesting 1o note that fine stratified ice lenses can develop in stiff moist clays and shales by redistribution of water within the material, even if external sources of water are not avail- able (Beskow, 1947). In this process heaving is practi- cally absent. However, extraction of water from portions of the material simidates drying as a consequence of evaporation. Shrinkage cracks may form, disrupting original fabric and stimulating the onslaught of slaking ■with rewetting. The size and spacing of ice leuses also depend upon fabric of soils (Beskow, 1947). In silty soils ice layers are clean and parallel to the freezing surface. The 240 Clays and Clay Technology [Bull. 169 coarser the silt, the less ice will accumulate, both in thickness, continuity, and spacing- of the layers. In a fine silt, the ice layers typically are a few tenths of a milli- meter thick, a few centimeters in lateral extent, sep- arated a few millimeters, and oriented mutually par- allel, giving a streaked appearance. The coarser the silt, the finer the structure of the ice layers until, with a grain size predominantly in the range 0.06 to 0.10 mm, the ice layers disappear entirely. Coarser soils freeze homogeneously. In clays, the ice layers are thick and widely spaced, uniform groups lying parallel to the freezing surface and being more distinct and thicker, the more colloidal the clay. Thick ice layers commonly develop in clays at discontinuities, especiallj' planes of stratification. In varved clays subjected to slow freezing, many thick and uneven ice layers grow simultaneously in several laminae. Cracks that occur in the dried upper part of clays commonly are filled with ice above the frost line, but aeeiunulation of ice causes the cracks to widen and ex- tend downwardly several centimeters below the freezing zone. In undisturbed claystones and shales, the imperme- ability frequently prevents development of ice lenses in the mass but yet significant heaving can occur by growth of ice layers along joints and other fractures by draw- ing of water from external sources through the fracture system. Variations in fabric and water availability cause dif- ferential growth of ice lenses, hence can produce differ- ential heaving of overlying or adjacent engineering structnres. Clearly, differential displacement is more critical than uniform movement. Fortnnately, the poten- tialities for stress development in materials as the result of frost heaving are relatively small. Taber (1929 ; 1930) was able to inhibit ice lense formation bj' application of surface loads of about 215 pounds per square inch. Sensitivity of Clays to Remolding Excavation and recompaction of earth materials par- tially or completely destroys original fabric of clays and shales. Simi^ltaueously, capillary and adsorbed water films are disrupted. The degree of permanent change effected in the soil properties by manipulation depends upon the work done in disturbing a unit volume of the material, the original fabric, and the nature of the bind- ing agent originally supporting the fabric. If the bind- ing agent is mineral matter of low solubility, such as iron oxides and silica, originally deposited from solu- tion in voids and upon gi-ain boundaries, the change will be essentialh' irreversible. If the original structure is stabilized by moisture films, the remolding might effect only temporarj- changes, which diminish as the moisture distribution progressively approaches that of the original material. Similarly, if the binding agent is composed of highly soluble salts, strength lost by wet- ting and working will be regained wholly or in large part with drying, or, indeed, the strength of remolded material may be greater than that of the original soil (Winterkorn and Tehebotarioff. 1947). Reversibility of binders was discussed previously under "Cementation." The strength of clay stabilized by thixotropic develop- ment in the solid-water sj'stem can be lost almost com- pletely by agitation, vibration, or application of shearing stress. With moisture content in excess of the liquid limit, soils liquefy as a result of disturbance. The thixo- tropic structure redevelops with rest at rates depending upon the physical-chemical properties of the constitu- ents, the fabric of the soil, electrolyte content of the water phase, and the water content. In a study of the sensitivity of clays to remolding. Winterkorn and Tschebotarioff (1947) conclude that decrease in compressive strength with remolding is due to destruction of cementing minerals ; and that increase in compressive strength with remolding is due to (1) break-down of secondary aggregates with consequent in- crease in surface area available for adsorption. (2) re- arrangement of particles into a more stable fabric, and (3) destruction of planes of weakness or fractures. Clays unaffected by remolding appear to owe their cohesive properties to capillary water films and to the action of hydrophilic reversible colloids. Loess is especiallj' sensi- tive to remolding, the disturbed loess t.vpically possess- ing about one-half the compressive strength of the un- disturbed loess at the same porosity (Denisov, 1946). Sensitivity of soils to drying and wetting depends in large part upon the hydrophilic or hydrophobic proper- ties of the clays and their volume change with change in water content. The phenomena associated with slaking are discussed subsequentlj'. Slaking Slaking is the disintegration of materials as the resvilt of change of water content. Under natural conditions, slaking proceeds until single grains or particles and water-stable aggregations are produced. In the natural evolution of geologic deposits of clays and shales, slak- ing results from rewetting after partial removal by evaporation of stable adsorbed water films, which are at equilibrium within the fabric. A complex series of actions is involved in the slaking process. With initial drjdng the shale or clay first is compressed by capillary forces whose magnitude de- pends upon the size of the pores, their abundance, and the surface tension of the inter.stitial water. As com- pressive forces of large magnitude develop, shear failure with resultant cracking will occur if discontinuities of fabric create surfaces of weakness and variable capillary potential within the material. With continued drying below the shrinkage limit, the larger capillaries are progressively emptied by the smaller. In this process differences of compressive stress related to differences of fabric are accentuated and ci-acking increases. These cracks disrupt the fabric and facilitate rapid entry of water when rewetting occurs. With complete removal of capillary water by drying, water is confined to films upon surfaces of the solid particle and at points of contact. Tension in these films and other forces (see below) effect strong cohesion in the shale or clay, even in the absence of mineral cements. At equilibrium with ambient air the thickness of the water films depends upon the relative humidity and the surface-chemical properties of the particles. With rewetting of dried cla.v or shale, the adsorbed water films are increased in thickness, expansive clay minerals swell, and air is compressed in the interior under the force of penetration of water along capillaries (Yoder, 1936). All of these actions contribute to inter- nal stress and slaking. In the absence of expansive claj- minerals, the increasing thickness of adsorbed water films is significant in slaking onlv in loose sands and Tait V Clay Technology in Soil Mechanics 241 silts. AVith rewetting-, weak aggregations of sand and silt grains, stabilized only by water films, collapse under the force of gravity. Similarly, in clays and shales, minor splitting along fractures might occur in response to the decrease in cohesion effected by enlargement of adsorbed films. All significant slaking of clays and shales results from swelling of clay minerals or development of internal air pressure. The relative imi)ortauce of each depends upon the amount and swelling potential of the claj- minerals and upon the fabric. Clays and shales containing ex- pansive clay minerals will swell with or without spalling and disintegration if penetrated by water. Clays and shales containing only nonswelling clay minerals will spall and disintegrate witlunit notable swelling if pene- trated by water. Kaolinite. halloysite, and many hydrous mica clays and shales commonly are so compact as to defy penetration by water. They remain intact indefi- nitely when submerged. Dry shales and clays contain- ing expansive minerals might react very slowly because of their impermeability, but swelling or slaking idti- mately occurs. With immersion in water, air-dry sodium bentonite does not disintegrate ; rather, it slowly swells and floc- culant growths appear upon the surface. With swelling, the size of the piece increases severalfold, and ultimately its form is lost bj- slumping. However, complete disper- sion of the bentonite through the water phase does not occur, even though the bulk volume of the bentonite increases manyfold. Air-dry calcium montmorillonite is prone to rapid expansion with sinndtaneous disintegra- tion into small aggregations. With brittle montmoril- lonite-type clays and shales, the spalling can be ex- plosive. Air-dry potassium montmorillonite begins at once to swell and spall with considerable release of air bubbles as each new crack is formed. The spalled pieces continue to disintegrate until only small aggregations remain. Air-dr\' illite shales and clays begin at once to disintegrate by spalling. Air is released with the forma- tion of each new crack. The slaking proceeds until a mass of chips and flakes of moderate size remain. Porous kaolins spall and ravel at varying rates, air usually e.scaping with each spall. If the air-dry clays are evac\uited prior to immersion ill water, spalling and slaking occur in expansive cla.vs in a manner similar to that exi)erienced without evacua- tion. However, slaking commonly is more rapid, although less complete. After evacuation, kaolin containing no expansive minerals does not slake. These observations demonstrate that compression of air in capillaries is the only significant force causing slaking of nonexpansive clays. In expansive clays air pressure and differential expansion combine to effect slaking. Sands and silts containing expansive or nonex- pansive clay minerals will respond as do the related clays. For example, loess bound b.v grain coatings and interstitial montmorillonite slakes immediatel.v either with or without previous evacuation. In granular fab- rics, expansion of the clay constituent decreases cohesion and intergranular friction. Significant slaking occurs on wetting onlj- if the air- dry condition is approached. In tests of the beidellite- eontaining Carlisle shale occurring at Cedar Bluffs Dam site, near Ogallah, Kansas, a minimum of 16 to 32 min- utes of drying at 50 percent relative humidity and 70° F was necessary to induce softening and .slaking follow- ing immersion. Slaking increased with further drying time, being complete following immersion in water after 24 hours of drying. Cla.vey materials usually do not slake in nonpolar liquids, such as bcnzeiu\ If a variety of liquids is used, rapidity and degree of slaking tends to increase with increasing dielectric constant of the liquid. This is true even after drying of the ela^^ at 200° C, a temperature found to completely remove the adsorbed water. The fact that clays and shales, susceptible to slaking under ordinary conditions, remain cohesive after drying at 200° C indicates that forces other than the tension of water films are involved in the binding of particles to- gether. Increasing degree of slaking witli increasing dielectric constant of the penetrating li(|uid suggests that the forces are electrostatic. Consequently, it is not neeessarj" to assume that materials, which slake in water but resist slaking in liquids not miscible with water, do so only because stabilizing water films remain in the fabric. Slaking of clays and shales in water depends upon both weakening of elecfro.static binding forces be- tween particles and development of internal tensile stresses in the fabric by expansion or air pressure. Slaking of clays, claystones, and shales cause many minor difficulties at engineering sites (Burwell, 1950). Slopes ravel progressively following exposure by exca- vation aiul are prone to erode badly with run-off or if acted upon by waves or currents of reservoirs or canals. Shales and clays permitted to dry befcn-e placing of concrete slake if later saturated, tlius destroying bond of the concrete to the foundation. Terzaghi (1950) cites slaking of newty exposed clays in open cuts as causing landslides by decreasing shearing resistance of the ma- terials. ^Measures to prevent excessive slaking usually involve placing a protective cover, such as concrete, earth materials, or asphalt, innnediately after excavation to grade. Slaking also is reduced by electro-osmotic and electrochemical treatment and b.y introduction of organic cations w'hich develop hvdrophobic qualities in the clav (Davidson, 1949; and Glab, 1949). CLAYS AND SHALES IN CONCRETE MATERIALS Clays in Natural Aggregate and Crushed Stone Clay minerals are almost universal constituents of natural sand and gravel and crushed stone. They occur as disseminated crystals or masses, or as veinlets in sedimentary, metamorphie, and igneous rocks of all types (Rhoades and Mielenz. 1948; Knight, 1949). For example, granites and granite gneisses of the Rocky Mountains of Colorado and Wyoming commonly contain disseminated crystals, veinlets, and seams of clay minerals, including kaolinite, illite, montmorillo- nite, and halloysite, either singly or in various combi- nations. Basalts of the Columbia Plateau of Washington almost universally contain nontronite with or without montmorillonite and saponite as a result of deuterie, hydrothermal, and epigene alteration of original glass, palagonite, ferromagnesian minerals, and plagioclase. This alteration may be uniformly distributed through pebbles, or it may be accentuated at the periphery of pebbles as a result of weathering after the gravel was deposited (fig. 42). In that area, sands typically carry nontronife-containing basalts, grains of almost pure 242 Clays and Clay Technology I Bull. 169 PlGUKE 42. Basalt gravel showins (levrloinneiit of weathered rims in wliich palagonite is leacliprl ami iiiiitially ;iltpreil to nontro- nite and iron oxides. iiontrouite, and, le.ss I'oiiinKiuly, Hakes of montmoril- loiiite witli or without iiitennixed halloy.site. Lime- stones from many areas contain clay minerals, either dis.seminated within or between crystals of calcite or concentrated along solution channels (Loughlin, 1928). In the vicinity of San Diego, California, the Poway conglomerate locally contains* pebbles of dacite and andesite which are deeply altered to montmorillonite- type clay. Alteration evident in natural aggregates may devekip in the original rock formation or after deposition of the sand and gravel. Alteration of natural aggregates in. place is especially dangerous because deepest decom- position occurs at the peri]ihery of the particles where the cement must bond. Depending upon the kind, amomit, and distribution of clay minerals in aggregate, several deleterious actions might arise in the concrete. Adsorption of water is increased and resistance to freezing and thawing and wetting and drj^ing of the concrete is reduced. If expan- sive clays are present, wetting and drying can produce deleterious volume change of concrete (Rhoades and Mielenz, 1948; Knight, 1949). Strength and wear re- sistance of the concrete can be decreased critically. Break-down of the aggregate during handling and mixing is aggravated, and water reqitirement is in- creased and the strength is reduced. Changing gradation during mixing is a characteri.stic of the Poway eon- glomerate where excessive amounts of altered volcanic rocks are present. Clay minerals occur also in uatui-al aggregate as con- stituents of highly clay particles, such as shales, clay- stones, argillaceous concretions, and clay balls. Clay balls are rounded lumps of clay occurring in gravel or coarse sand, having been developed by rolling and molding of plastic clay by stream action ; they generally contain sand and fine gravel particles. Argillaceous ferruginous concretions are particularly abundant in sands and gravels of North and South Dakota and adjacent parts of Wyoming and Nebraska. Chijjs and flakes of shale are common, minor constituents of sands and gravels, particularly those carrying locally derived materials not yet subjected to long continued attrition and impact. Claj' balls are common in Colorado in coarse sand and gravel or small streams draining areas in which the Mancos shale is exposed. In addition to accentuating the deleterious action exi^erienced with disseminated clay and clay seams in aggregate, the highly argillaceous materials commonly occur as lightweight particles which segregate towards the surface during placing and finishing of concrete. Moreover, shales commonly produce flat and slabby pieces which may segregate and produce oriented zones which decrease workability and require tise of additional sand, cement, and water in the mix. Sodium and potas- sium clays of high cation exchange capacity, such as montniorillonoids, can release alkalies into the solutions permeating the mortar by exchange reactions with cal- cium, thus possibly contributing to alkali-aggregate reaction. Clays occur also in natural aggregates as coatings upon the particles. These coatings decrease bond with cement and their removal during handling and mixing increases the fines in the aggregate by continuous break- down. Removal of plastic clay coatings fre(iuently is costlt, both in development of the necessary plant and in its operation and control. Phelps (1952) reports progress in use of deflocculating agents to displace clay coatings from aggregate more effectively than could be accomplished by washing with water alone. Various kinds and concentrations of reagents are used, most wide-spread success being found with sodium silicates. The aggregate is fed into a pool or into a .classifier where the water is added together with the dispersing agent. The aggregate is agitated continuously and the clay is discharged with effluent water over a weir. In one installation, aggregate containing as much as 25 percent by weight of material passing the No. 325 sieve is said to be deslimed with use of a deflocculating agent at a cost of $0.04 per ton of feed, iloreover, capacity of the plant is increased by treatment of the aggregate, one plant reporting an increase of 75 percent in capacity at a cost of .$0.01 per ton. The type of agent, its rate of addition, and other featitres of the process depend upon the composition and amount of clay and its mode of occurrence. Conseqitently, the requirements nmst be established individually for each installation. Clays and Shales as Sources of Lightweight Aggregate Production and use of both natural and manufactured lightweight aggregate for concrete, mortar, and plaster is a rapidly expanding industry. Clays and shales are processed into satisfactory lightweight aggregate of sev- eral types which are marketed conunercially under the trade names such as llaydite, Rocklite, Gravelite, Cel- Seal, and others (Cordon and llickev, 1948; Price and Cordon, 1949; Tuthill, 1945; Kruge et al., 1949; Peter- sen, 1950). Firing is accomplished in rotary kilns or sintering machines. In either process the raw material is heated rapidly to the range between incipient and complete fusion. Bloating or vesieulation is accom- plished by entrapment of released gases by the viscous, partially fused clay or shale. Clays and shales are fired alone after crushing and sizing, or after admixture with iron oxides or carbonaceous material. Research I 'art Vl C'l.Av TiaiiNOLoiiY IX Soil Mkcuaxic; 243 FiGuiiE 43. Coarse and fine Rocldite ( left I :inil ll;i.vilitc li^lil woifiht assi-pg.-ite. and elevek)piiR'iil oL' materials, iiictlunls, pi'udiK'tioii equipmont, and additional uses is activelj' in progress. Ilaydito is niannfactured as both coarse and fine aufirefiatc. and has hem used most extensively of all clay- and slialc-type li<,ditwtMp:lit ag^rcg-ates. eomniercial niannfactnrc iiavinii- be<>Tm in 1!)2(). After being crashed to pass the 1.')- or 1-ineh sieve, the shale or elav is firetl in a rotary kiln at 2,000" F to 2,100' F until "a desired degree of vesieulation is achieved (Tnthill, 1945). The clinker material is then crushed to desired sizes (fig. 43). N'ery good stuctural concrete can be made with properly iiiMiiiiracI ured Ilaydite. the unit weight ranging from 1(1(1 i(i 110 pounds per cubic foot, and the compressive strengtii ranging from 4,000 to 5,000 psi, depending on the cement content of th(> nux (Cordon and llickey, 1!)48; Price and Cordon, 1949). The concrete is resistant to freezing and thawing and wetting and drying. Ex- pansion as the result of alkali-aggregate reaction is negligible. Bulk dry weight of the coarse aggi'cgate ranges from 35 to 49 |)ounds jjer cubic foot. Rocklitc is produced by firing of crushed and screened shale 01- clay in a rotary kihi at about 2.170 F (Tnthill. 194.")). Dnring the firing, originally iri'cgnlar pieces vesiculate and expand to rcmnded pebble-like forms, and a finely porous skin of high strength about 1 nun thick is developed at the peri])hery (fig. 43). Very good structural concrete can be produced with Hoeklite ag- gregate, the unit weight varying from !tO to 100 pounds |)ei- cidiie foot, and the strength ranging' fi'oni 1.000 t- O en 2 X UJ a: < i- (£ O O Q UJ TEMPERATURE (C) CALCINATION TEMPERATURE Figure 44. Effect of calcination of clay and shale pozzolans ou expansion of mortar as the result of alkali-aggregate reaction. The clays and shales classified by Conley and associates as "bloaters" contain less than 25 percent of AI2O3, but three-fourths of the nonbloaters likewise contain less than this amount of AI2O3. Their work demonstrated that determination of pH is 90 percent correct in iden- tifying propensity for vesiculation, the nonbloaters pro- ducing a water slurry with pH less than 5, whereas bloaters produce a water slurry with pli greater than 5. This relation probably relates to the nature of the ex- changeable cation and the. cation exchange capacity of the clay or shale, the alkalies and alkaline earth serving constructively as fluxes in the firing process. In the firing of natural clays and shales, gas undoubt- edly is evolved by several ijroeesses (Jackson, 1924; Hostetter and Koberts, 1921; Orton and Stalev, 1909). Austin, Numes, and Sullivan (1942) identified HoO, CO2, and SO3 as gases evolved from several bloating clays. In addition, it is likely that CO, O2, and possibly H2 are evolved in vesiculation of many clays and shales. Riley (1951) concluded that the most significant re- Part 100 90 80 VI Ci.AY Technology in Soil Mechanics 245 70 60 to z s. X c 50 4 a: o £ -No- BENTONiTE.neor Osage, Wyo 40 z o 30 20 10 ,-lLLlTE SHALE, near Trident , Mont. I4da 28da 8 9 10 2 3 4 5 6 7 AGE (Months) Figure 45. Effect of clay and shale pozzolans after calcination at 1400° P upon expansion of mortar as the result of alkali-aggregate reaction. 12 action involved in generation of gas is partial reduction of ferric oxide according to the equation : GFeoOa ^ 4PeFe204 + Oo. The ferric oxide is furnished by original limonite or hematite present in the raw shale or clay or by decom- position, with or without oxidation, of original ferric or ferrous compounds, most notably including biotite, amphiboles, montmorillonite-type or illitic clays, chlo- rite, and other clay and clay-like minerals. These rela- tionships demonstrate the need for mineralogic analysis in the investigation of clays and shales as sources of lightweight aggregate. Calculation indicates that at least 7.6 percent by weight of Fe20;i must be available from these sources for adequate vesiculation if this source alone furnishes gas (Riley, 1951). Complex silicates, such as micas, am- phiboles, and the clay and clay-like minerals, probably also release O2, Ho, and IL.O as the result of other re- actions. Experiments by Conley and associates (1948) and by Riley demonstrate that gases are evolved by re- actions of dolomite, pyrite, sulfur, sulfates, and car- bonates of tlie alkalies and alkaline earths, and possibly calcite. Less than 1 percent by weight of these minerals is required for adequate vesiculation. Carbon undoubt- edl.v contributes to gas formation during firing of some carbonaceous clays and shales. The fabric of the original clay or shale is significant in the expansion process. Most beneficial is a dense, rela- tively impervious fabric which resists shrinkage during heating and retards relea.se of vapors and gases before fusion effects a seal in the particles. The fabric is es- pecially significant in firing of carbonaceous clays and shales, inasmuch as a porous, open fabric permits ready burning out of the carbon, whereas a dense fabric re- tards oxidation by the kiln atmosphere and retains CO and COo produced by reaction with interstitial water or with water or oxygen released by dehydration or de- 246 Clays and Clay Technology [Bull. 1(19 fompositioii of hydrated eompounds or hydroxylated silicates. Most promising as sources of lightweight aggregate are shale.s and clays containing illite, beidellite-type mem- bers of the montmorilU)nite group, and vermiculite- chlorite. These minerals approximate the composition found by Riley ( 1951 ) to yield the melt of optimum vis- cosity. They almost always contain significant amounts of iron in their lattice and are tyi^ically admixed with varying amounts of limonite and hematite. Unlike mem- bers of the kaolin group, they typically contain moderate proportions of alkalies and alkaline earths which serve as fluxes in the firing process. The potentialities for ex- pansion and vesiculation of clays and shales of these mineralogic types have been demonstrated both in the laboratory and iu connnercial enterin-ise. Clays in Pozzolans Pozzolans are silieeous or siliceous and aluminous ma- terials, natural or artificial, processed or nnproee.ssed. which, though not cement it ious in themselves, contain constituents that will combine with lime in the presence of water at ordinary temperatures to form compounds which have a low solubility and process cementing prop- erties (U. S. Bur. Reclamation, 1949; Lea, 1938). Use of pozzolans as a replacement for part of the portland ce- ment in concrete has come into prominence in recent years and is increasing. "Within recent years, certain pozzolans have been found to participate in chemical and physical-chemical reactions with alkalies (NasO and KoO) released during hydration of portland cement (Hanna, 1947; Mielenz et al., 1950). As a consequence, deleterious reactions involving the alkalies, such as the alkali-aggregate reaction, are inhibited or prevented. Mielenz, Witte, and Glantz (1950) and :\Iielenz, Greene, and Schieltz (1951) have classified pozzolans into five "activity types," depending upon the sub- stances responsible for the pozzolanic action, as follows : Activity Type 1 Volcanic glass Activity Type 2 Opal Activity Type 3 Clay mineral.^ Activity Type 3a Kaolin group Activity Type 3b Montmorillonite si'oiip Activity Type 3c Illite group Activity Type 3d Mixed clays with vermiculite-chlorite Activity Type 3e Palygorsldte Activity Type 4 .Zeolite Activity Type 5 Hydrated aluminum oxides In the natural condition, clay minerals are nonpoz- zolanic or only weakly pozzolanic. However, with calci- nation, particularly in the range 1,200° to 1,800° F, partial dehydration and crystallographic changes result in significant reactivity with lime and alkalies for most types of clay (figs. 44 and 45). Although certain cal- cined clays control or markedly reduce expansion of mortar as the result of alkali-aggregate reaction, water requirement typically is excessive and strength develop- ment is usually low, espeeiallv for pozzolans of Aetivitv Type 3b (figure 46). The kaolin group of minerals disintegrate with partial dehydroxylation to one or more aluminous and siliceous substances which are largely amorphous microscopically and by X-ray diffraction. With slow heating of kaolinite, water is lost progressively between 600° and 930° F, at which temperature collapse of the crystallographic struc- 6 S So 01 a: o Kaolin (Cole. 1400° F) aCo.-Beidelli+e Cloy ( Cole. 1400' F ) Note; Pozzolon is o mixture of calcined, ground cloy and pulverized quartz and reploces 35% by volume of the Portland cement 25 50 75 PROPORTION OF CLAY IN POZZOLAN (PERCENT) Figure 40. Effect of calcined clay pozzolans on water require- ment and compressive strength of portland cement mortar. ture begins (fig. 47) (Nutting, 1943). Disintegration may occur at temperatures as high as 1,225° F if the kaolinite is heated rapidly (Parmelee and Rodricjuez, 1942). This disintegration proceeds rapidlv between 930° and 950° F, with liberation of II and Oil ious as water. With disintegration of the crystalline structure an amorphous aluminous and siliceous substance is pro- duced. As the temperature increases, an exothermic re- action occurs at about 1,800° F, possibly as the result of crystallization of gamma-alumina, and gamma-alumina can be identified by X-ray diffraction analysis. At about 1,825° F, mullite" (3Alob:i-2Si02) and/or gamma-alu- mina form, the relative amount of each dependiug largely upon the degree of order existing in the (n'iginal clay mineral. After calcination at 2,000° F, the gamma-alu- mina has disappeared. With continued heating at 2,750° F, a mixture of mullite, cristobalite, and silica glass is produced, the precise temperatin-e of complete nuilliti- zation and the content of glass depending upon the kind and abundance of impurities (Parmelee and I!odri(]uez, 1942). For jnire kaolinite. the weight loss during calcina- tion exceeds 14 percent ( Nutting, 1943). The course of dehydration and recrystallization of anauxite is similar to that of kaolinite, except that de- hydration begins at about 625° F and proceeds rapidly with disintegration of the crystallographic structure between 750° and 950° F. virtually complete dehydra- tion being accomplished between 1,200° and 1,475° F, Part VI ri.AV TriiiNHi.ocY IX Sim. .Mi:( ii.wic 247 400 600 800 1000 1200 1400 CALCINATION TEMPERATURE (F) 1600 1800 2000 FiGUKK 47. Wci;,'lit change of clay miner;ils iliniiit: cjilcinatiiiii i.Xultiii;;. 1',I4:!|. depending upon the composition of the original anauxite (Nntting. 1943). The pozzolanic propertit-s of kaolin undergo remark- able change, both physically and chemically, with calci- nation (Mielenz et al., 1950). Water requirement de- creases from 72.8 to 46.0 percent for tested materials calcined at 2.000° F. Neither initial nor final set of the lime-kaolin paste occurs at an age up to 14 da.vs unless the kaolin is calcined at or above 800° F. Beginning of crystallization of gamma-alumina and later of muUite during calcination at 1.600°, 1,800°, and 2,000° F in- creases the time necessary for set of the lime-kaolin paste to be accomplished. At an age of 28 da.ys, the mortar strength increases almost iiniformly with calcina- tion of the kaolin ; but the maximum strength is equiva- lent to only 85 percent of the strength of the control mortar. Test data indicate that crystallin*- kaolinite is com- paratively ineffective in controlling alkali-aggregate reaction, but the amorphous siliceous and aluminous substance produced with collapse of the crystalline structure after calcination between 1.000° and 1,600° F is highly effective, the optimum calcination temperature being 1,400° to 1,600° F (fig. 44) (Mielenz et al., 1950). This substance apparently is a comiiouiid rather than a simple mixture of siliceous and aluminous glass, inas- iiiiich as the solubility of the silica remains very small until gamma-alumina forms at 1,800° F, probably with resulting decomposition of the amorphous compound aiul formation of readily soluble silica (Mielenz et al., 1950). However, formation of gamma-alumina causes a decrease in the effectiveness of the kaolin in controlling mortar expansion. At 2.000° F, mvillite forms, and the effectiveness of the material in reducing alkali-aggregate reaction probabl.y would be decreased further. With calcination, montmorillonite first loses free wa- ter and, then, betw^een about 300° and 950° F, water adsorbed on or between lattice planes of the clay crystals is released (fig. 47) (Xutting, 1943). Beginning at about 950° F, H* and OH ions of the lattice structure of the clay are lost, the rate of loss being high as the tempera- ture rises to about 1.200° F. Dehydration continues slowly at higher temperatures and is essentially com- plete at 2,000° F. However, the rate of dehydration with progressive heating depends upon the chemical composi- tion, degree of alteration, and exchangeable cations of the clay. Typical montmorillonites lose from about 4.6 to about 9.8 percent in weight between 212° and 1,475° F. 248 Clats and Clay Technology Table 17. Effect of calcination on 001 interplanar spacing of montmorillonite-iype clays and illite.' [Bull. 169 Calcination Interplanar spacing 001 (kX units) *> temperature CF) Na-MontmoriUonite Beidellite Beidellite Nontronite. near Coulee City, Washington Hectorite, near Hector, California Illite, Illinois • Raw 14.01 IS. 11 14.50 15.11 14.54 10.27 800 15.11 and 12.84 d 14.50 and 9.76 9.88 14.54 10.07 1000 12.53 10.27 9.64 9.88 14.54 10.07 1200 9.88 and 12.53 9.83 9.69 9.88 9.64 and 13.78 10.07 1400 9.76 9.83 9.69 9.88 9.64 10.07 1600 — ' _. 9.69 9.88 _. 10.07 1800 — — — • — " — » 2000 — — — — — — ' From Mielenz. Witte. and Glantz (1950). t One AnRstrom =1.00202 kX units. « Crystal structure destroyed. ^ Line too blurred for unique measuremeat. • Sample supplied by W. F. Bradley, Illinois State Geological Survey, t'rbana, Illinois. With the release of H and OH ions from the lattice of montmorillonite, irreversible shrinkage occurs along the c-axis of the crystal (table 17). For example, for the soclium-montmorillonite d(OOl) decreases progres- sively with calcination. After calcination at 1.600° P, the clay constituent has disintegrated so that the char- acteristic X-ray diffraction pattern of montmorillonite no longer can be obtained from the material. Upon dis- integration of the erystallographic lattice of the mont- morillonite of "Wyoming bentonite, spinel and beta- cristobalite form in progressively increasing amounts as the calcination temperature increases from 1.500° to 2,000° F. The general course of dehydration of beidellite is similar to that of montmorillonite. but loss of H* and OTT- ions and shrinltage of the lattice take place at lower temperatures (between about 750° and 950° F), dehydration being essentially complete at 2,000° P (fig. 47) (Nutting, 1943). Depending upon the chemical com- position, degree of alteration, and exchangeable cations, beidellite may lose more than 12 percent by weight of water during heating from 212° to 2,000° P. Irreversible shrinkage of the crystal lattice of beidellite begins dur- ing calcination at a temperature about 800° F, being completed during calcination at 1,200° P. Disintegration of the lattice apparently begins with calcination at 1.400° to 1,600° F, and after calcination at temperatures of 1.600° to 1,800° F, the atomic arrangement within the crystal of beidellite is destroyed (table 17). During calcination at 1,600° F, beidellite commonly disinte- grates completely and spinel cr.ystals of colloidal dimen- sions form. "With heating to 1,800° F, the crystals of spinel increase in size and cristobalite is formed and both minerals persist after calcination at 2.000° P, but a small amount of mullite and other unidentified crystal- line compounds also are developed. In general, mont- morillonite-type minerals fire first to a spinel if sub- stantial proportions of Al*** occur in tetrahedral coordi- nation, whereas quartz or cri.stobalite occur before spinel if the proportion of Si**** approaches 4.00 (Bradlev and Grim, 1951). Nontronite and saponite rarely occur in amounts suffi- cient to be of interest as a source of pozzolan. Pozzolans of Activity Type 3b are changed greatly by calcination in the range 800° to 2,000° F (Mieleuz et al., 1950). Water requirement is markedly decreased with calcination. For materials of this type water required to produce a lime-pozzolan paste of normal consistency ranges from 32 to 165 percent. The setting time of lime- pozzolan paste decreases markedly with calcination of the pozzolan, usually to a minimum after calcination at 1.000° to 1.400° P. With calcination at higher tempera- tures, time of set increases. Comparative strength of portland-cement-pozzolan mortar increases with calcina- tion of the pozzolan, the maximum strength being ob- tained after calcination at 1,200°, 1,400°. 1,600°, and 1,800° F, depending upon the composition of the poz- zolan. The effect of calcination upon the alkali reactivity of pozzolans of Activity Type 3b depends upon the miner- alogic identity, chemical composition, and exchangeable cations present in the clay minerals. Sodium bentonite has been found to be comparatively ineffective in con- trolling alkali-aggregate reaction (Mielenz et al., 1950), apparently because the presence of exchangeable sodium prevents the clay from reducing the concentration of alkalies in the solution permeating the portland-cement mortar (figures 44 and 45). If naturally occurring sodium is replaced by calcium in Wyoming bentonite, the clay is etfective in controlling alkali-aggregate reac- tion after calcination at 1.400° P (fig. 45). The change in activity of montmorillonite-type clays with calcination relates to the erystallographic changes occurring with the heat treatment. Shrinkage of the montmorillonite and beidellite crystals from 14 or 15 kX units to 9.69 or 9.88 kX units (table 17), during calcination at temperatures ranging from 800° to 1,200° F, relates to increased reactivity wnth alkaline solu- tions and particularly a greater capacity of the clay to adsorb and hold alkali ions. Calcination from 1,200° to 1,600° P leads to destruction of tlie clav and formation Part V Clay Tih unology ix Soil ilEciiAxirs 24;) of eoinparativcly stable compounds with resulting de- crease in alkali reactivity of the pozzolan. "With calciuation, illite clavs lose free water between about !10° and 212° F (fig. 47) (Xuttinjr, 1943). Ad- sorbed water is lost between 212° and 575° F, and de- hydroxylation of the lattice occurs with heating above 575° F. Complete dehydration is accomplished at about 1,300° F. The details of dehydration and lattice change with calcination vary with composition, alteration, and exchangeable cations of the illite. The dimensions of the crystallographic lattice respond less to calcination than do those of the montmorillonite-type clays (table 17). Calcination from 212° to 1.600° F produces little change in X-ray diffraction patterns of two specimens analyzed, other than slight changes in position and in- tensity of line (table 17). After calcination from 1.600° to 1,800° F, the lattice has disintegrated with produc- tion in one specimen of spinel, quartz, and other un- identified compounds. Calcination at 2,200° F of a shale containing illite-type clay with muscovite, together with considerable calcite, dolomite, and quartz, produced a mixture of glass (n greater than 1.548), diopside (CaMg SiO.T)2. quartz, and a small proportion of un- identified substances. Pozzolans of Activity Type 3c are affected signifi- cantly by calcination in'the range from 800° to 1,800° F. "Water requirement is decreased moderately, especially after destruction of the clay by calcination at 1,800° F (Mielenz et al., 1950). For pozzolans of this activity type producing set of lime-pozzolan paste, time of set is re- duced to a minimum by calcination of the pozzolan at 1,400° F. Compressive strength of lime-pozzolan mortar is improved slightly or moderately by calcination, max- imum strength occurring after destruction of the clay component by heating at 1,800° F. Illite-type clay poz- zolans investigated show little change in alkali reactivity with calcination in the range 800° to 1,800° F, and fail all tests (figs. 44 and 47). The mixed clays with vermieulite-ehlorite are clayey silts and clays which contain a complex mixture of clay minerals which constitutes 30 to 50 percent of the whole. They are distinctly inferior as pozzolans. X-ray diffrac- tion analysis of a typical sample of the material indi- cates that the vermieulite-ehlorite constitutes about 66 percent, beidellite-uontrnnite about 22 percent, and illite about 11 percent of the mixture. During calcination, the beidellite-noutronite and illite respond as indicated for these minerals in the previous paragraphs. Vermieulite- ehlorite represents a series transitory between vei'micu- lite and typical chlorite, the position of any given member of the sequence depending upon the relative proportions of hydrated exchangeable magnesium and water as against brucite (^Ig(0II)2) in tiie interlayer space (Brindley, 1951). Irreversible dehydration may begin as low as 800° F if excessive interlayer water is present. At 1,600° F disintegration usually is complete. Pozzolans of Activitj^ Type 3d show only slight change in reactivity with calcination up to 1,800° F. None developed set with lime while raw and many show no final set with lime after calcination at 1,000° F. Com- pressive strength of mortar ranges from inferior to good, the maximum strength being developed after calcination of the pozzolan at 1,800° F. X^o data are available on alkali-reactivitj' of pozzolans of Activity Type 3d. Sparse data are available for palygorskite (Activity Type 3e). Tests have established the effect of calcined (1.400° F) and ground Floridin (Attapulgus clay) on mortar expansion resulting from alkali-aggregate reac- tion. Satisfactory results were obtained (fig. 45). During differential thermal analysis, Floridin loses adsorbed water in two stages at about 300° and 480° F (Kerr et al.. 1949). Dehydroxylation proceeds between about 845° and 1,025° F. A sharp exothermic reaction at about 1,650° F marks the development of spinel. Clays and shales are the source material of several pozzolans produced commercially or in pilot operations at several plants (Mielenz et al., 1951). Most notably, the siliceous shales of the Monterey formation of Cali- fornia, which owe their pozzolanic activity both to opal and beidellite, have been produced in a calcined condi- tion and marketed by the California Cement Company at Colton, California, and by the Santa Cruz Portland Cement Company at Davenport, California. An oil- impregnated shale of Activity Type 3b occurring in the Monterey formation near Casmalia, California, has been used after calcination both in Cachuma Dam, near Santa Bai-bara, by the Bureau of Reclamation, and in Big Creek Dam No. 7, near Fre-sno, California, bj- the South- ern California Edison Company. ]\Iowry shale has been produced experimentally in "Wj'oming, and was used after calcination in highway test sections by the Kansas State Highway Department. The Eagle Ford shale, near Dallas, Texas, also a representative of Activity Type 3b, has been investigated as a source of pozzolan. Altered pumicites owing their activity to both rhyolite glass and clay minerals have been produced. An altered pumicite owing a significant portion of its activity to montmorillonite was used in a calcined condition in con- struction of Falcon Dam, near Laredo, Texas. In in- vestigations for Hungry Horse Dam. Montana, an al- tered pumicite owing its activity primarily to opal, anauxite. and kaolin was found to be satisfactory. CONCLUSIONS Physical-chemical properties control the engineering properties of earth material through mineralogic and compound composition, fabric (texture and structure), and fluids content. Much more remains to be learned about the interrelation of physical-chemical phenomena and engineering performance, but already application of known principles is lagging behind advances being made in the laboratory. Continuing progress is possible only with close coordination of laboratory investigations and field trials. Study of physical-chemical properties will not replace the soil mechanics tests now widely used to establish quantitatively the response of earth materials to specific physical and hydraulic conditions. However, knowledge of physical-chemical properties of earth materials can aid the engineer by : 1. Anticip.iting probable performance of earth materials under varying conditions of service and thus aiding development of minimum testing programs 2. Selection of design or construction methods by which difficul- ties can be avoided or minimized 3. Explaining causes of failure so as to avoid recurrence in new construction 4. Selecting effective procedures for repair and maintenance in the face of difficulties 5. Outlining needed research 250 Clays and Clay Technology [Bull. 169 Areas of civil and materials engineering in which con- tinuing fundamental and applied research on clays and shales is required to answer immediate engineering needs are : 1. Movement of moisture under structures a. Conditions eontrollinfj movement of moisture in eartli mate- rials b. Means of controlling movement of moisture in earth mate- rials c. Methods minimiziiiR effects of moisture ehanse on founda- tion materials 2. Drainage of foundations, sulj^radcs, natural slopes, and em- bankments a. Conditions controlling drainage of earth materials b. Means of effecting drainage, including electrical methods 3. Stabilization of foundations, subgrades, natural slopes, and embankments a. Physical-chemical phenomena in binding and waterproofing actions, including sensitivity to remolding b. Methods of controlling shear strength, bearing capacity, water resistance, and durability of water-bearing clays in place, including use of admixtures and agents, and electro- chemical procedures c. Methods of controlling settlement of structures on clay formations 4. Control of permeability a. I.ow-cost earth canal linings b. Methods of reducing permeability of formations in place c. Methods of increasing iiermeability to improve drainage and reduce development of pore water pressure in clays and shales. 5. Channel stabilization and erosion control a. Conditions controlling erosion of clays and shales in jilace or in embankments b. Conditions controlling establishment of cohesion in newly deposited clays e. Methods of controlling erosion of clays, including preven- tion of erosion of clays and shales and inducement of scour in newly deposited clays (J. Calcination of clays a. Production of lightweight aggregate from clays and shales b. Production of pozzolans by heat treatment of clays, shales, and clayey materials Successful completion of these programs of research demands coordinated investigations in the field and in the laboratory. The properties and the performance of clays must be evaluated in the light of their composition and fabric and the geologic relations of the formations in which they occur. Evaluation of performance must be made with due regard to conditions of service to which the installation might be subjected. Continuing advance in engineering involving either use of clays as construction materials or construction of works at sites underlain by clays can come only with close cooperation of the engineer, the engineering geolo- gist, the engineering petrographer. and the clay tech- nologist. ACKNOWLEDGMENTS The authors wish to express their gratitude to their colleagues of the Bureau of Reclamation for assistance and helpful di.scnssion in the development of this paper. In particular, thev recognize the generous cooperation of W. G. Holtz, H. J. Gibbs, and N. C. Schieltz. E. J. Benton and P. H. Geier assisted in compilation of the data. The authors also are indebted to many individuals and organizations for their generosity in supplying or per- mitting use of data and illustrations. In this regard they recognize the cooperation of J. E. Jennings, Director, National Building Research Institute, South African Council for Scientific and Industrial Research; F. M. Lea, Director, Building Research Station, Department for Scientific and Indu.strial Research, Garston, Eng- land ; The American Society for Testing Materials ; The Iliglnvay Research Board; The Reinhold Publishing Corporation ; A. E. Cummings, Raymond Concrete Pile Company ; H. P. Winterkorn, Princeton University ; and Irving Goldberg and Alexander Klein, Institute of Transportation and Traffic Engineering, University of California, Berkeley, California. The Engineering Laboratories are under the direction of W. H. Price. All design and construction activities are supervised by L. N. McClellan, Assistant Commis- sioner and Chief Engineer. The Bureau of Reclamation, United States Department of the Interior, as a whole is directed by W. A. Dexheimer, Commissioner. DISCUSSION C. G. Dodd: Has there been any practical application of electro-osmosis to the removal of water from clays for practical soil-mechanics opera- tions? R. C. Mielenz: Most of the work has been done in Germany and Russia and very little in this country. I understand that the principle of electro-osmotic dewatering to effect stabilization was used at La Guardia Field during its construction. Preece (1047) has done a great deal of work on electro-osmotic and electrochemical sta- bilization. The most extensive practical applications were made in Germany during the war, for stabilization of embankments, excavations, and U-boat pens. Electro-osmosis theoretically involves only withdrawal of water without addition of materials to the soil. Conversely, electrochemical stabilization involves both electro- osmotic withdrawal of water and addition of new substances to the soil, either by decomposition of the electrodes or by addition of chemicals added at the electrodes. Electro-osmosis is not per- manent unless some supplementary means of preventing return of the water to the dewatered area is installed. The electrochemical process may or may not be permanent depending upon the specific reactions that have taken place. R. E. Grim: The process of electro-osmosis is used by the railroads in Hol- land for dewatering in some of the low areas. There the process goes on continuously. In Sweden paper wicks for dewatering were punched into the ground in the construction of the International Airport between Stockholm and Upsala. The wicks were 5 inches wide and about i inch thick and were inserted by a specially developed apparatus. R. L. Stone: There has been some work done on a]iplicalion of electro- phoresis to the beneficiation of ball clays and kaolins. The rewet- ting properties of this clay in comparison with one partially de- watered by centrifuging may be of interest. In each case, final removal of water was accomplished by drying at 70" C. When rewetted, as in the formation of casting slips, there was no differ- ence in gel strength and casting rate. A difference did show up in the DTA curves : a deflection was observed at 380° C for the electrophoresized ball clay. No difference, however, was found by chemical or X-ray analyses. It is suspected that the results had something to do with the electron charge. R. C. Mielenz: There are several examples of cracking in clay formations, sometimes below the water table, in which drying cannot explain the degree of cracking present, and which might be explained by syneresis. I have not been able to find a satisfactory exi>lanation of syneresis as far as it relates to clays. This process or property has been studied extensively in the organic and inorganic gels and dyestuffs, but little work has been done with clays. Syneresis is a spontaneous separation of a homogeneous colloidal system into a coherent gel and a liquid. In a study of the geologic significance of syneresis Jiingst (1934) describes many features Part V] Clay TErnxoi-oov ix Son. Mkciiaxics 251 whirl] liavf lipon cNiihiiiicd ;i.s r;iiii(ln>p prints. iliyiiiK n large Buchner funnels to a certain depth %vith a certain he.id of water. The amazing observation at the time was that after a few days of standing with the head of water on them, cracks began to develop, some of them big enough to let one's finger go through. Later on, some of these bentonites were actually used to seal Ibis reservoir. They wcu-ked very w(dl for a while, but later cracks began to develop in the benlonite lining of the wall, so that il bad to be done all over again. T. F. Biiehrer: .\rr Ilierc .iliy other tests be.siranes of essentially natural benlonite or admixed with local soils. With hydration swelling of the bentonite seals off the pores and reduces the iierme.ability of the lining. For a quick field test, the free-swell test is recommended. This test involves granulation of the dried sample through No. 30 sieve, a bulk volume of 10 cc of the Xo. 30 to No. 50 sample is sifted into distilled water; and the ajiparent bulk volume of the wetted material is observed after 24 hours. It the bulk volume does not e.\ceed (! times the original volume, the material is con- sidered In be inferior or unsui(abl(! for use. Material pa.ssing this te-^t i< \\'ortii_\ of further testing. W. T. Cardwell, Jr.: In Ihi> oil industry, we too lire inlerested in the rehilion between clay swelling ability and se;iling ability. In a drilling fluid it is ilesir.ible to Use a clay that will form only a thin filter cake on the wall of the well, a cake whd .Mexauder l>am. a leUable struc- ture : Engineering News-Record, v. 104. p. TO.*?. Anonymous, lO.'iOa, Hydraulic-filled dam of fine volcanic iish fails disastrously : Engineering-News Kecord. v. 104, pp. S(')!)-S71. Anonymous, 1941, A study of methods used in measurement and analysis of settlement loads in streams : Rept. 4, Joint program of the Tennessee Valley Authority, Corps of Engineers, Dept. Agriculture, Geol. Survey, Bur. Reclamation, Indian Service, and Iowa Inst. Hydraulic Res., Univ. Iowa, 203 pp. Anonymous. 1047, Diagnosis and improvement of saline :ind alkali soils : U. S. Dept. Agriculture. V. H. Regional Salinity Laboratory, Riverside, California, VXt'.', pji. Anonymous. 1!)4!), American Society for Testing Materials Standards. T.»40 Book. Ft. III. pp. 1177-1170. Anonymous, California Research Corporation. 10.")0. Particle- size data on the reference clay minerals: Am. Petroleum Inst., Proj. 49. Prelim. Rept. 7, pp. 129-133. Abdun-Nur. E. A., lOoO, A standard classification of soils as proposed bv the Bureau of Reclamation : Am. Soc. Testing JIat., Sp. Tech. Pub. no. 113, pp. 2."")-41. Ackermaun. Ernst. 194S (tr. 19.")0l. Thixotropic und Fliesseigen- schaften feinkorniger Boden : (Jeohigische Rundschau, v. 3(), 194S ; translati'd by II. .\. C. Nathan, Nat. Res. Council of Canada, Trans. TT i.'iO. 21 pp. Allen, Harold, 1942, Classific-itinn of soils and control of pro- cedures used in construction of endianUments : Puldic Roads, v. 22, pp. 263-284. Austin. C. R., Numes, .1, L., and Sullivan, .1. D., 1942. Basic factors involved in bloating of clays: Am. Inst. Min. Met. Eng. Tech. Pub. 148(5: Min. Technology, v. 6, no. 4, 11 pp. Barshad, I., 1948, Vermicidite in its relation to biotite ;is re- vealed by base exchange reactions. X-ray analysis, differential thermal curves, and water content: Am. Mineralogist, v. 33, pp. ('>.">-678. Bates, T. F., Hildebrand, F. A., and Swineford, Ada, 1950. Morphology and structure of eudellite and halloysite : Am. Miner- alogist, V, 35, pp. 463-484. Baver, L. D., 1940, Soil physics: New York, ,Iohn Wiley and Sons, Inc., pp. 90-97. Baver, L, D., and Winterkorn, H. F., 19.35, Sorption of liquids by soil colloids, II : Soil Sci., v. 40, pp. 403-419. Beskow, (iunnar, 1935 (tr. 1947), Soil freezing and frost heav- ing with special application to roads and railroads : Swedish Geol. Soc. ser. C, no 375, 26th Year Book, no. 3 ; translated by .1. O. Osterberg, Techncd. Inst., Northwestern I'niv., Evanston, Illinois. 252 Clays and Clay Technology [Bull. 169 Beskow, Gunnar, 1948, Scandinavian soil frost research of the past decade : Highway Res. Board Proc. v. 27, pp. 372-382. Bodman, G. B., 1941, Di.sciission [of Lee, 1941] : Am. S?oc. Civil Eng., Trans., v. 106, pp. 577-607. Bonnell, D. G. R., 1932, Studies of gels. II. Effect of hydrogen ion concentration on the syncresis of silica gel : Faraday Soc, Trans., v. 28, pt. I. pp. 12-19. Bonnell, D. G. R., 19.33, Gels. IV, Swelling of silica gel : Fara- day Soc. Trans., v. 29, pp. 1217-1220. Bonnell, D. G. R., 1933a, Gels. V, Effect of neutral electrolytes on the syneresis of silica gels : Faraday Soc. Trans., v. 29, pp. 1221-1226. Bosazza, V. L., 1944, The adsorption of some organic dyes by clays and clay minerals : Am. Mineralogist, v. 29, pp. 23.5-241. Bouyoucous, G. .T., 1915, Effect of temperature on .some of the most important physical processes in soil : Michigan Agri. Expt. Sta., Tech. Bull. 22. Bradley, W. F., 1940, The structural scheme of attapulgite : Am. Mineralogist, v. 25, pp. 405-410. Bradley, W. F., 1945, Molecular associations between mont- morillonite and some polyfunctional organic liquids : Am. Chem. Soc. Jour., T. 67, pp. 975-981. Bradley, W. F., 1946, Diagnostic criteria for clay minerals : Am. Mineralogist, v. 30, pp. 704-713. Bradley, W. F., and Grim, R. E., 1951, High temperature ther- mal effects of clav and related materials : Am. Mineralogist, v. 36, pp. 182-201. Bradley, W. F., Grim. R. E., and Clark, W. I., 1937, A study of the behavior of montmorillonite upon wetting : Zeitschr. Kris- tallographie, v. 97, pp. 216-222. Braune, H., and Richter, I., 1949, (abst. 1950) The thixotropy of bentonite suspensions : Kolloid-Zeitschr., v. 113, pp. 20-29 ; (abst.) Chem. Abst., v. 44, p. 5184. Brindley, G. W., 1951, X-ray identification and structure of clay minerals : Loudon, Minereralog. Soc, 345 pp. Burmister, D. M., 1949, Principles and techniques of soil identi- fication; Highway Res. Board, Proc. v. 29, pp. 402-4.33. Burnett, G. E., 1936, Bentonite and cement for use as grout in fine sand formations : U. S. Bur. Reclamation, Denver, Colorado, 7 pp., mimeo. Burwell, E. B., Jr., and Moneymaker, B. C., 1950, Geology in dam construction : Geol. Soc. America. Berkey Volume, pp. 11-43. Carey, A. S., 1949, Natural clay grout in open work gravel : (abst.) Geol. Soc. America Bull., v. fiO. pp. 1878-1879. Casagrande, A., 1932, The structure of clay and its importance in foundation engineering : Boston Soc. Civil Eng., v. 19, pp. 72-126. Casagrande, Arthur, 1948, Classification and identification of soils: Am. Soc. Civil Eng., Trans., v. 113, pp. 901-991. Casagrande, L., 1937, Grossversuch zur Erhiihung der Traffa- higheit von schwebenden Pfahlgrundungen durch electro-chemische Behandlung : Die Bautechnik, v. 15, pp. 14-16. Casagrande, L., 1939, Die elektrochemische Boden Verfestlgung : Die Bautechnik, v. 17, pp. 228-230. Casagrande, L., 1947, The application of electro-osmosis to prac- tical problems in foundations and earth works : Great Britain Dept. Sei. Indus. Res., Building Res., Tech. Paper no. 30. Casagrande, L., 1949, Electro-osmosis in soils : Geotechnique, v. 1, pp. 159-177. Conley, J. E., Wilson, H., Klinefelter, T. A., et al., 1948, Pro- duction of lightweight concrete aggregates from clays, shales, slates and other materials : U. S. Bur. Mines Rept. Inv. 4401, 121 pp. Cordon, W. A., 1944, Grouting with bentonite mixed in products of petroleum : U. S. Bur. Reclamation Laboratory Rept. C-229. Cordon, W. A., and Hickey, M. E.. 1948. Properties of concrete made with typical lightweight aggregates : V. S. Bur. Reclamation, Materials Laboratory Rept. C-385, 19 pp. Cox, J. B., 1929, Construction control on Hawaiian hydraulic fill dam based on physical chemistry : Hvdraulic Eng., v. 5, pp. 16-20 and 22. Cummings, A. E., 1947, The foundation problem in Mexico City : Southern Texas Conf. on Soil Mechanics and Foundation Eng. Proc, 9 pp. Davidson, Donald T., 1949, Exploratory evaluation of some or- ganic cations as soil stabilizing agents : Highway Res. Board, Proc, V. 29, pp. 531-537. Davidson, D. T., and Glab, J. E., 1949, An organic compound as a stabilizing agent for two soil-aggregate mixtures : Highway Res. Board Proc, v. 29, pp. 537-543. Dawson, R. F., 1953 (in press). The design of building footings on expansive clay .soils : Am. Soc. Civil Eng., Proc. Dean, L. A., 1947, Differential thermal analysis of Hawaiian soils: Soil Sci., v. 63, pp. 95-105. Dean, L. A., and Rubins, E. J., 1947, Anion exchange in soils. Part I Exchangeable phosphorus and the anion exchange capacity : Soil Sci., v. 63, p. 377. DeLapparent, J., 1935, An essential constituent of fuller's earth : Comptes Rendus Acad. Sci., v. 201, pp. 481-483. Denisov, N. J., 1946 (tr. 1951), Settlement properties of loessial soils : Govt. Pub. Office Soviet Sci., abst., translated by K. P. Kar- poff and H. J. Gihbs, U. S. Bur. Reclamation, Eng. Laboratories, 15 pp. (mimeo.) Dyal, R. S., and Hendricks, S. B., 1950, Total surface of clays in polar liquids as a characteristic index : Soil Sci., v. 69, pp, 421-432. Endell, Kurd, 1935, Beitrag zur chemischen Erforschung und Behandlung von Tonboden : Die Bautechnik, v. 13, no. 18. Endell, Kurd, (tr. 1941), The swelling capacity of clays in the construction foundation and its technical significance : Die Bau- technik, V. 19, p. 220 ; tr. by U. S. Corps of Engineers. Endell, Kurd, and Hofmann, U., 1936, Electrochemical harden- ing of clay soils : First Internat. Conf. Soils Mechanics and Foundations Eng. (Harvard) v. 1, pp. 273-275. Erlenbach, Lutz, 1936, Anwendung de Elektro-Chemischen Ver- festigund auf schwimmende Pfahlgrundungen : Die Bautechnik, v. 14, pp. 257-259. Fairbairn, H. W., 1949, Structural petrology of deformed rocks: Cambridge, Mass., Addison-Wesley Press, Inc., 344 pp. Felt, E. J., 19.50, Field trials to locate and eliminate potential wave areas prior to construction of concrete pavement on soils developed from Taylor marl: Portland Cement Assoc, (Chicago) 146 pp. Felt, E. J., 1953, Influence of soil volume change and vegetation on highway engineering : Univ. Colorado Eng. Exp. Sta. Circ, Highway series no. 26. pp. 52-76. Ferguson, .!., and Applebey, M. P., 1930, The syneresis of silica- gel : Faraday Soc. Trans, v. 26, Part II. pp. 642-655. Fersman, A., 1908, On the palygorskite group : Acad. Imperial Sci. St. Petersburg Bull., v. 2, pp. 255-274. Fischer, E. K., and Gans, D. M., 1946, Dispersions of finely divided solids in liquid media : Colloid Chemistry, v. VI. New York, Reinhold Publishing Corp., pp. 286-327. Eraser, H. .1., 1935, Experimental study of the porosity and per- meability of clastic sediments : Jour. Geology, v. 43, pp. 785-909. Freundlich, H., 1930. Kapillarchemie, v. 1, 4th ed., Leipzig. Freundlich, H., 1935, Thixotropy, Paris, Hermann and Co. Freundlich, H., and Juliusburger, F., 1935, Quicksand as a thixotropic system : Faraday Soc. Trans., v. 31, p. 769. Frye, J. C., and Swineford, Ada, 1946, Silicified rock in the Ogallala formation : Kansas Geol. Survey Bull. 64, Part 2, pp. 33-76. Gallaway, B. M., and Buchanan, S. J., 1951. Lime stabilization of clay soil : Texas Agri. and Mech. Coll., 5th ser., v. 7, Texas Expt. Sta. Bull. 124, 57 pp. Gapon, E. N., 1930. (abst. 1930), The theory of syneresis: Ukrainskii Khem. Zhur., Sci. Pt., 5, pp. 59-61 ; abst. Chem. Ab- stracts, V. 24, p. 5203. Giesecke, S. E., 1922, Columns and walls lifted by swelling clay under floor : Engineering News-Record, v. 88, pp. 192-193. Gieseking, J. E., 19.39, The mechanism of cation exchange in the montmorillonite-beidellite-nontronite type of clay minerals : Soil Sci., V. 47, pp. 1-14. Glasstoue, Samuel, 1946. Textbook of physical chemistry, 2nd ed.. New York, Van Nostrand Company, Inc., 1.320 pp. Goldberg, Irving, and Klein. Alexander. 19.53, Some effects of treating expansive clays with calcium hydroxide : Am. Soc. Test- ing Mat., Special. Tech. Pub. 142, pp. 53-67. Graton, L. C, and Fraser. H. J.. 19.35, Systematic packing of spheres — with particular relation to porosity and permeability : Jour. Geology, v. 43, pp. 785-909. Green, Henry, and Weltmann, R. N., 1946, Thixotropy : Colloid Chemistry, v. 6, pp. 328-347. Grim, R. E., 1949, Mineralogical composition in relation to the properties of certain soils : Geotechnique, v. 1, pp. 139-147. Grim, R. E., 1942, Modern .concepts of clav minerals: Jour. Geology, v. 50, pp. 225-275. Grim, R. E., AUaway. W. H., and Cuthbert, F. L., 1947, Reac- tions of different clav minerals with some organic cations : Am. Cer. Soc Jour., v. 30^ pp. 137-142. Part VI Clay Tkiiinology ix Soil Mechanics 253 (Jruni'i-. J. \V., 1933, The crystal structure of nacrite and a comparison of certain optical proi)erties of the kaolin group with structure: Zeitschr. KristalloKraphie. v. sr>, pp. 34r)-3r)4. Gruner, .1. W., 1934, The structures of verniiculites and their collapse by dehydration : Am. Jlineralogist, v. 19, pp. 557-575. Hanna. W. C., 1947, Unfavorable chemical reactions of aggre- gates in concrete and a suggested corrective : Am. Soc. Testing Mat. I'roc, V. 47, p. 986. Ilarmau. C. (J., and Fraulini, Feli.x, 1940, Properties of kaolin- ite as a function of its particle size: Am. Cer. Soc. Jour., v. 23, pp. 252-259. Hauser, E., and LeBeau, D. S., 1941, Studies in colloidal clays, 1 1 : .Jour. Phys. Chemistry, v. 45, pp. 54-(i4. Ilauser. E. A., and lycBeau, D. S., 1940. Colloid chemistry of clay minerals and chiy films: Colloid Cheniislry, v. (i, pp. 191-213. ilavens, .T. II., Young, .T. L., Jr., and P.aker, R. F., 194S, Sepa- ration, fractionation, and mineralogy of clays in soils : Highway lies. Board i'roc, v. 28, pp. 409-480. Ilavens. J. II.. Young. J. T... Jr., and I>rake, W. B., 1949, Some chemical, physical, and mineralogical features of soil colloids: Highway Res. Board I'roc., v. 29, pp. 507-577. Heller. Wilfried. 1937 (abst. 1937), Syncresis: Comptes Rendus Acad. Sci., v. 204, pp. 43-45; (abst.) Chem. Abstracts v. 31, p. 1679. Hendricks, S. B., 1936, Concerning the crystal structure of kaolinite, Al:03-2Si02-2n:0, and the composition of anau.xite : Zeitschr. Kristallographie. v. 95, pp. 247-2.52. Hendricks, S. B., 1941, Base exchange of the clay mineral mont- morillonite for organic cations and its dependence upon adsorption due to van der Waals' forces : Jour. Phys. Chemistry, v. 45, pp. 65-81. Hendricks, S. B., and Jefferson, M. E., 1938, Crystal structure of vermiculites and mi.xed vermiculite-chlorites : Am. Mineralogist, V. 23, pp. 851-862. Hendricks, S. B.. Nelson, R. A., and Alexander, L. T., 1940, Hydration mechanism of the clay mineral monlmorillonite satu- rated with various cations : Am. Chem. Soc. Jour., v. 62, Ft. I, pp. 1457-1404. Hirashima, K. B., 1948, Highway experience with thixotroi)ic volcanic clay : Highway Res. Board I'roc, v. 28, pp. 481-494. Ilnfmann. U. and Bilke, AV.. 1930, tMier die innerkrystalline Qtiollung und das Basenaustauschvermogen der Montmorillonits : Kolloid Zeitschr., v. 77. pp. 239-251. Hofmann. U., Endell, K., and Wilm, P.. 1933, Kristallstruktur und Quellung von Jlontmorillonit : Zeitschr. Kristallographie, v. 86A, pp. 340-347. Hofmann, U.. Endell, K., and Wilni, D.. 1933a. The crystal structure and swelling of montmorillonite : Zeitschr. Kristallogra- phie. v. 80, pp. 340.348. Hofmann, V.. Endell, K.. and Wilm, D., 1934, RiJentgenogra- phische und kolloid chemische L'ntersuchungen liber Ton : Ange- wandte Cbemie, v. 47, pp. .5.39.547. Holtz, W. G., and Gibbs, H. J., 19.53 (in press). Engineering properties of expansive clays : Am. Soc. Civil Eng. Hoskings, J. S., 1940, The soil clay mineralogy of some Aus- tralian soils developed on granitic and basaltic parent material : Australian Council of Sci. and Ind. Res., v. 13, pp. 206-216. Hostetter, J. C, and Roberts, H. S., 1921, Notes on the dissoci- ation of ferric oxide dissolved in glass and its relation to the color of iron-bearing glasses : Am. C*'ramio Soc .lour., v. 4. jip. 927-938. Houwink, R., and Burgers. W. G., 19.39, Elasticity, plasticity, and structure of matter: Cambridge Univ. Press, 376 pp. Hvorslev, M. J. 1939, Uber die Festigheitseigenschaften ge- storter bindiger Boden : Ingeniorvidenskabelige Skrifter, A. 45, Copenhagen. Jackson, F. G., 1924, Oxidation of ceramic ware during firing : II, Decomposition of various compounds of iron with sulfur under simulated kiln conditions : Am. Ceramic Soc. Jour., v. 7, pp. 223- 237. Jennings, J. E., 19.50, Foundations for buildings in the Orange Free State gold fields : South African Inst. Civil Eng., Jour., v. 49, pp. 87-112. Jenny, H., 1932, Studies on the mechanism of ionic exchange in colloidal aluminum silicates : Jour. Phvs. Chemistry, v. 36, pp. 2217-2258. Jenny, H., and Gieseking, J. E., 1936, Behavior of polyvalent cations in base exchange : Soil Sci., v. 42, p. 273. Jenny, H., and Reitemeier, R. F., 19.34, Ionic exchange in relation to the stabilitv of colloidal systems: Jour. Phvs. Chemistry, v. .39, pp. 593-604. Johnson, R. H., 1920, The cementation process in sandstone : Am. Assoc. Petroleum Geologists Bull., v. 4, pp. 33-35. Johnson, A. L., and Davidson, D. T., 1947, Clay technology and its application to soil stabilization : Highway Res. Board Proc, v. 27, pp. 418-430. Jiingst, H., 1934, Zur geologischen Bedeutung der Synarese : Geol. Rundschau, v. 25, pp. 312-325. Karpoff, K. P., 19.53, Stabilization of fine-grained soils hy elec- tro-osmotic and electrochemical methods : Highway Res. Board Proc, V. 32, pp. 526-.538. Kerr, P. F., and Kulp, J. L., 1949. Reference clay localities : Am. Petroleum lust., Proj. 49, Prelim. Rept. 2, 103 pp. Kerr, P. P., Kulp, J.L., and Hamilton. P. K., 1949, Differential thermal analyses of reference clay mineral specimens : Am. Petro- leum Inst., Proj. 49, Prelim. Rept. 3, 48 pp. Knight, B. H., 1949, Aggregates for concrete work : Civil Eng. and Public Works Rev., v. 44, pp. 448-450. Kolbuszewski, .T., 19.52, A study of the electro-chemical harden- ing of clay : Civil Eng. and Public Works Rev., v. 47, pp. 5.56.5.58. Kruge, R. W., Sparks, M. M., and Tunma, E. C, 1949, Eight- weight aggregate concrete : Am. Concrete Inst., Proc. v. 45, pp. 625-642. Kubiena, W. L., 1938, Micrnpedology, Ames, Iowa, Collegiate Press, Inc., 423 pp. Ladd, G. E., 1934, Bank slide in deep cut caused by drought : Engineering News-Record, v. 112, pp. .324.326. Lambe, T. W., 1954a, The improvement of soil proi)crties with dispersants : Boston Soc. Civil Eng. Jour., v. 41, pp. 189-207. Lambe, T. W., 19.54b, The permeability of fine-grained soils: Am. Soc. Testing Mat. Proc, v. 59 (in press). Eambe, T. W., and Martin, R. T., 1953, Composition and engi- neering properties of soil : Highway Res. Board I'roc, v. .32, pp. 576-590. Lambe, T. W.. and Michaels, A. S., 1954, Altering soil proper- ties with chemicals : Chem. Eng. News, v. 32, pp. 488-492. Larsen, D. H., 1946, Colloidal features of drilling fluids: Col- loid Chemistry, v. 6, pp. 509-530. Laws, W. I)., and Page, J. B., 1946, Silicate of soda as a soil stabilizing agent : Highway Res. Board Bull. 1, 21 pp. Lea, F. M., 1938, The cheniisti-y of pozzolans : Symposium on the chemistry of cement, [Stockholml, Proc, pp. 400-490. Lee, C. H., 1941, Sealing the lagoon lining on Treasure Island with salt: Am. Soc. Civil Eng. Trans., v. 106, i)p. 577-607. Lewis, D. R., 19.50, Base exchange data on the reference clay minerals : Am. Petroleum Inst. Proj. 49, Prelim. Rept. 7, pp. 91-133. Longchambon, M. H., 19.37, Caract^ristique de la sfpiolite d'am- pandandrava et la formule des s^piolites : Soc. Fran<;aise Min(5ra- logie Bull., V. 60, pp. 232-276. Loughlin, G. F., 1928, Usefulness of petrology in the selection of limestone : Rock Products, v 31, p. 50. MacEwan, D. M. C, 1944, Halloysite and metahalloysite : Na- ture, V. 1.54, pp. 577-578. MacLean, D. J., and (Jwatkin, P. M., 1946, Moisture movements occurring in soil due to the existeuce of a thermal gradient: Road Res. Lab., Eng., Note. Mack.soud, A. M., 1939. The nature and origin of mountain leather: Columbia Univ., Master's Thesis, pp. 1-29. Marshall, C. E., 1949. The colloid chemistry of the silicate minerals : New York, Academic Press, Inc., 195 pp. Maryott, A. A., and Smith E. R., 1951, Table of dielectric con- stants of pure liquids : U. S. Dept. Commerce, Nat. Bur. Standards, Circ 514, 44 pp. Mattson, S., 1929, The laws of soil colloidal behavior I. : Soil Sci., v. 28, p. 179. Mielenz, R. C, and King, M. E., 1951, Identification of clay minerals bv staining tests : Am. Soc. Testing Mat. Proc, v. 51, pp. 1213-12.33. Mielenz, R. C, and Okeson, C. J., 1946, Foundation displace- ments along the Malheur River siphon as effected by swelling shales : Ecotj. Geology, v. 41, pp. 266-281. Mielenz, R. C, Greene, K. T., and Schieltz, N. C, 1951, Natural pozzolans for concrete : Econ. Geology, v. 46, pp. 311-.328. Mielenz, R. C, Witte, L. P., and Glantz, O. J., 19.50, Effect of calcination on natural pozzolans : Am. Soc Testing Mat. Spec Tech. Pub. 99, pp. 43-91. Nelson, R. A., and Hendricks, S. B., 1943, Specific surface of some clay minerals, soils, and soil colloids : Soil Sci., v. 56, pp. 285-296. Nutting, P. G., 1943, Some standard dehydration curves of minerals: U. S. Geol. Survey, Prof. Paper 197-E. p. 197. 254 Clays axd Clay Technology' tBtill. Kill Oiton, E., and Stale.v, H. F., 1909, Status of C, Fe, and S in clays duriiif,' various stages of luirniuK : Xnt. rSriok Mfgs. Assoc, 3ni Rept., Indianapolis, Indiana. Parmelee. C. W., and Rodriquez, A. R., I'.M'J. Catalytic mulliti- zation of kaolinite metallic oxide : Am. Cenunic Soo. Jour., v. 25, p. 1. Pauling, L., 1930, The structure ot niicns nml related minerals: Natl. Acad. Sci. Proc, v. 16, pp. 12:M2!i. Pennington, R. P., and .laokson, M. !<., 1947, Segregation of the clay minerals ot polycomponent soil clays : Soil. Sci. America Proc, V. 12, pp. 4.")2-4.57. Petersen, T. H., 19.">0, Lightweight aggregate concrete's: Housing and Home Finance Agency, U. S. Govt. Printing Off.. 28 pp. Petti.iohn, F. .1., 1940. Sedimentary rocks. New York, Harper Brothers, pp. 46-49. Phelps, D. S., 19.^2, Removing clay from gravel : Rock Products, V. ">.'), p. 64. Pl.'ink. C .!.. and Brake, L. C, 1947, Differences between silica and silioa-alnmina gels. II, A proposed mechanism for the gelation and syneresis of these gels : Colloid Sci. ,Tour., v. 2, pp. 41.S-427. Prakash, Sutya, 1933 (abst. 1933), Continuity of phases during sol-gel transfornmtions : Kolloid-Zeitschr., v. 64, pp. 293-.300 ; (ahst.) Chem. Abstracts, v. 27, p. 5613. Preece, K. F., 1947, Geotechnics and geotechnical research : Highway Res. Board Proc, v. 27, pp. 384-416. Price, W. H., and Cordon, W. A., 1940. Tests of lightweight aggregate concrete designed for monolithic construction : Am. Con- crete Inst. Proc. V. 45, pp. 581-600. Rhoades, Roger, and Mielenz, R. C. 1948. Petrographic and mineralogic characteristics of aggregates : Am. Soc Testing Mat. Spec. Tech. Pub. S3, pp. 20-48. Riley, C JI., 1951. Relation of chemical properties to the Idoat- ing of clays : Am. Ceramic Soc. .Tour., v. .34. pp. 121-128. Roseuqnist. I. Q.. 1946. Om icires kvikkagtighet : Statens Veg- vcsen Veglaboratoriet. Medd. 4, Oslo. Ross, C. S., and Hendricks, S. B., 1945, Minerals of the mont- morillonite group : U. S. Geol. Survey Prof. Paper 205-B, 70 pp. Ro.ssi, Giuseppi, and Marescotti. Angelo, 1936 (abst. 1937), Syneresis II. Gazz. Chim. ital., v. 66, pp. 223-227; (abst.) Chem. Abstracts, v. 31, p. 926. Russell. .T. C. Olmstead. L. B.. and Hendrickson. B. H., 1929. Form of soil structure : Am. Soil Surv. Assoc Bull. 10. pp. 120-13.3. Samuels. S. G.. 1950, The effect of base exchange on the en- gineering properties of soils: Pept. Sci. and Indus. Res., Bldg. Res. Sta.. Rept. C176. 16 pp. Schaad. W.. and Haefeli. R.. 1946. Die Anweudnng der Eleck- trizitiit zur Entwasserung und Verbessernug feinkorniger Boden arten : Strasse and Verkehr. no. 23/24. pp. 311-316. Schaad. W.. and Haefeli, R., 1947, Electrokinetic phenomena and their application in soil mechanics: Schweizerische P.auzeitung. v. 65, pp. 216-217. 223-226, 235-238. Schachtschabel, P., 1940. Untersuchuugen fiber die Sorption der Tonmineralien und organischen Bodeukolloide : Kolloid-Beihefte. v. 51. pp. ]99-27«!. Shorey. R. C. 1914, Presence of some benzene derivatives in .soil : .Tour. Agr. Res., v. 1, pp. 357-363. Simpson, W. E., 19.39. Foundation experiences with clav in Texas : Civil Eng., v. 4, no. 11. Smith, W. O., 19.39. Thermal conductivities in moist soils: Soil Sci. America Proc, v. 4, pp. .32-40. Soveri, TIrpu, 1950, Differential thermal analy.ses of some Quaternary Clays of Fennoscandia : Sucnnalaisen Tiedeakatemian Toimituksia, Annales Academiae Scientiarum Fennicae, Ser. A.. Ill Geologica-Geographiea 23, 103 pp. Spangler. M. G., 1949, Electrical hardening of clays ad.iaceut to aluminum friction piles : Highway Res. Board Proc, v. 29, pp 589-599. Spiel, S., 1940, Effect of adsorbed electrolytes on properties of monodisper.sed clay-water systems: Am. Ceramic Soc. Jour., v. 23, pp. .33.38. Steinfeld, Karl, 1951, Die Entwasserung von Feinboden : Die Bautechnik, v. 28. pp. 269-274. Stout, P. R., 19.39. Alterations in the crystal structure of clay minerals as a result of phosphate fixation : Soil Sci. Soc. America Proc, v. 4. pp. 177-182. Taber, Stephen, 1929, Frost lic:iviiig: Jour, (leologv v 37 pn 428-461. Taber, Stephen, 1930, The mccliauics of frost heaving : Jour. Geology, v. .38, pp. 303-317. Talmud. T., and Sncliovolskaya. S.. 1931 (abst. 1931), Models of lyophilic colloids: Kolloiil-Zeitsclir., v. .55, pp. 48-64; (abst.) Cliem. Alistracts. v. 25, iip. :'.540-:',541 . Terzagiii, Karl, 19.50, Jlechanics of land-slides: Geol. Soc. Amer- ica, Bcrkey Volume, pp. 83-123. Terzaghi, Karl, and Peck, R. P.., 1948, Soils mechanics in engi- neering practice : New York, .John Wiley and Sons, .566 pp. Trask. P. D., 19.39, Organic content of Recent marine sediments, ill Trask. P. D., ed.. Recent marine sediments. Am. Assoc. Petro- leum Geologists, p. 429. Trejo. .1. H.. 1946. Water niovenuMit in soils under thermic poten- tials : Princeton Univ.. School of Engineering, Master's Thesis. Tuthill, L. H., 1945, Concrete operations in the concrete ship I)rogram : Am. Concrete Inst. Proc, v. 41. pp. 137-177. U. S. Bur. Reclamation. 1949, Concrete manual : V. S. Bur. Reclamation, 5th ed., 489 pp. Vey, D.. 1949. The mechanics of soil consolidation by electro- osmosis: Highway Res. Board Proc, v. 29, pp. 578-589. Waldschmidt, W. A., 1941, Cementing materials in sandstones and their probable influence on migration and accumulation of oil and gas: Am. Assoc. Petroleum Geologists Bull. v. 25, pp. 1839- 1879. Walker, G. F., 1949, Distinction of vermieulite, chli>rite, and montnuu'illonite clays : Nature, v. 164. p. 577. Walker. G. F.. 19.50, Verniicuiitc-organic complexes: Nature, V. 166, pp. 695-696. Weil-Malherbe. H., and Weiss, .T., 1948, Colour reactions and adsor]ition of some aluminosilicates : Chem. Soc Jour., London, pp. 2164-2169. White. W. A.. 1949, Atterbei-g plastic limits of clav minerals: Am. Miner.-ilogist, v. 34, pp. 508-512. Winkler, H. G. F., 1938, Thixotropie von mineralpulvern mikroskopischer Griisse : Kolloid-Beihefte, v. 48. Winkler. H. G. F.. 1943, tjl)er die Thixotropie des Jlontmoril- lonits: Kolloid-Zeitschr., v. 105, pp. 29-38. Winkler, II. G. F., 1949, Clays and their projiertie.s — an inter- pretation: Research, v. 2, pp. 175-183. Winterkorn. H. F., 1940, Physical-chemical testing of soils ami .ipiilicalion of results in practice: Highway Res. Board Proc. V. 20. pp. 798-806. Winterkorn, H. F., 1946, Principles and practice of soil stabili- zation : Colloid Chemistry, v. 6. pp. 459-492. • Winterkorn. H. P.. 1947, Fundamental similarities between electro-osmotic und thermo-osmotic phenomena : Highw.ay Res. Board Proc, v. 27, pp. 443-455. Winterkorn. H. F., 1948, A fundami'ntal approach to the sta- bilization of cohesive soils: Ilighwav Res. Board Proc, v. 28, pp. 415-422. Winterkorn. H. F.. 1953, JcjIi exiierience with exchange phe- nomena involving inorganic and organic ions: Am. Soc. Testing Mat. Special Tech. I>ub. 142, pi). 29-4;i. Winterkorn, H. F., 1953, Surface-chemical properties of clay minerals and soils: Am. Soc. Testing Mat. Special Tech. Pub. 142, l)p. 44-.52. Winterkorn, H. F., and Shoudlniry, A. N. D., 1949, Importance of volume relationships in soil stabilization : Highway Res. Board Proc. \-. 29, pp. 5.53-.5aver (1929), a soil scientist, showed the effect of in- creasing amounts of lithium, sodium, potassium, niag- iiesinm, and calcium h.vdi'oxide on the pll, electro- kinetic jiotential, electrical conductivity, and viscosity of dilute suspensions of an electrodialized colloidal frac- tion of Putnam silt loam. IJaver added (piantities up to several times beyond the exciiange capacity of his mate- i-ial, but naturally did not investigate many of the fac- tors in which the ceramist would be interested. A num- ber of ceramic investigators have contributed portions of a similar picture of the effect of the amount and kind of exchangeable ions on kaolinitc; comi)aratively less lias been (lone on illite. .10 . 9 J i 1 1 1 1 1 1 1 .» 1 1 1 1 / 1 / I 7 / / / / \H \^^0^ / / z .6 J/ T o ^'-"""^ L (n ^^-•"^ /^ z y^ / In NaCl 0. /^ / to / / SonrnoN D .5 / / to / / / / o r / / / y 0. _4 _3 / ,- '/ A FLORIDA KAOLIN ^-'' ',^ _2 E. N. 4 1 a 2 4 6 6 1 . 1 • HCl NaOH ME PES 100 s. Figure 2. Titration curves of electrodial.vzed Florida liaolin in distilled water and in N/2 NaCl .solution, showing exchange acidity, exchange alkalinit.v, and exchange neutralit.v (E.X.). Endell. Fendius and llofmann (1934), Graham and Sullivan (1938; Sullivan and Graham 1940), Speil (1940), Siefert and Henry (1947), Harman and Frau- lini (1940), Grim (1942) and others have discussed the ceramic effects of ion exchanges, mostly from the stand- point of the cations in the system. The effects reported by Barker and Truog (1938, 1939) on additions of .sodium carbonate ]ii-obnt, is essential in solving pr(thlems of drying losses resulting from cracking. The impoi-tance of the water films on and between clay particles cannot be overestimated. In certain instances, kaolinitic clay pieces while drying shrink to dimensions smaller tlian the corresponding leugtiis when dry: apparently the clay flakes undergo a relaxation as the last portions of water are removed (Norton, 1933). Figure 3 shows the relationshii) between water content and drying shrinkage for electrodialyzed (II) and so- dium-saturated (Na) Tennessee ball clay (Henry and Siefert, 1941). ■ ■/ / j''' UJ / / o . P., 19.36. The microbiology of aging clays: Am. Ceramic Soc. ,Tonr., v. 19. no. 6. pp. 169-175. Click. P. P., 1936a, The effects of various treatments upon the aging of a ceramic bodv : Am. Ceramic Soc. .Tour., v. 19, no. 9, pp. 240-242. Click. P. P., 1938, The relation of microbiological activity to the aging of some ceramic materials: Doctorate Dissertation, The Ohio State TTniversity Press. Cr.-iham. R. P., and Sullivan. .T. P.. 1938, Critical study of methods of determining exchangeable bases in clays: .\m. Ceramic Soc. Jour., V. 21. pp. 176-1,«;3. Graham, R. P., and Sullivan, J. P., 19.39. Workability of clays: Am. Ceramic Soc. .Tour., v. 22, no. 5, pp. 152-1.56. Grim. R. E., 1942, Modern concept of clay materials: .Tour. Geology, v. .50, no. 3, pp. 225-275. Grim. R. E., and Cuthbert, F. L.. 1945. The bonding action of clays. Part I. Clays in green molding sands: Illinois Geol. Sur- vey. Itept. Inv. 102. . . . T'niv. Illinois Exper. Sta. Bull. 357. Grim. R. E.. and Cuthbert. F. T... 1946, The bonding action of clays. Part II. Clays in dry molding sands: Illinois Ceol. Survey Rept. Inv. 110. .. . TTniv. Illinois Eng. Exper. Sta. Bull. .362. Grim, R. E.. and Rowland. R. A.. 1942, Pifferential thermal analyses of clay minerals and other hydrous materials : Am. Mineralogist, v. 27. no. 11. pp. 746-761. . . v. 27. no. 12, pp. 801-818. Grim. R. E., and Rowland, R. A.. 1944, Pifferential thermal analysis of clays and shales, a control and prospecting method : Am. Ceramic Soc. .Tour., v. 27. no. 3. pp. 65-76. Grupclli, Logan, 1944. Can the ceramist use emulsions?: Am. Ceramic Soc. Bull., v. 23. no. 8, pp. 288-290. Gruver, R. M., 1948. Precision method of thermal analysis: Am. Ceramic Soc. .Tour., v. 31, no. 12, pp. 323-.328. Hardy. W. B., 1026. Friction, surface energy, and lubrication, in Alexander. .Terome. Colloid chemistry theoretical ami applied, pp. 288-305. Xew York, The Chemical Catalog Co., Inc. Harman, C. C and Fraulini, F.. 1940. Properties of kaolinite as a function of its particle size: .\m. Ceramic Soc. .Tour., v. 2.3, no. 9, pp. 2.52-2.58. Harman. C. C Schaffer, C. F., Blanchard, M. IC, and .Tohnson, H. C, 1944, Study of the factors involved in glaze — slip control, I-IV: Am. Ceramic Soc. Jour., v. 27, no. 7, pp. 202-220. Hartley, G. S., 1935, General discussion on colloidal electrolytes : Faraday Soc. Trans., v. 31. p. 68. Hauser, E. A., 1941, Colloid chemistry in ceramii s ; Am. Ce- ramic Soc. .Tour., V. 24. no. 6, pp. 179-189. Hauser, E. A., and Johnson. .V. I... 1942. Plasticity of clays : Am. Ceramic Soc. .Tour., v. 25, no. 0. pp. 223-227. Hauth, W. E. Jr., 1951. Crystal chemistry in ceramics, I-VIII : Am. Ceramic Soc. .Tour., v. 30, no. 1, pp. 5-7. to v. 30, no. 6, pp. 203-205. Henry, E. C, 1942. Plasticity and workability of liall clays: Am. Ceramic Soc. Bull., v. 21, no. 11, pp. 269-271. Henry, E. C, 1943. Measurement of workability of ceramic bodies for plastic molding processes : Am. (_Vr;imic Soc. .Tour., v. 26, no. 1, pp. 37-39. Henry, E. C, and Sicfert, A. C., 1941, Phistic and drying prop- erties of certain clays as influenced by electrolyte content : Am. Ceramic Soc. Jour., v. 24, no. 9, pp. 281-285. Henry, E. C, and Taylor, X. W., 19.38, Acid- and base-binding capacities and viscosity relations in certain whiteware days: Am. Ceramic Soc. Jour., v. 21, no. 5, pp. 165-17,5. Hofmann, U., and Bilke, W., 19.36, Uber die inncrkristalline Quclhing unr(n'ed methods for deterniin;iti4)n : Am. Ceramic Soc. Jour., V. 27, no. 4, pp. 97-113. Russell, R., Jr., Mohr, W. C, and Rice. II. II., 1949. Casting characteristics of clays: II. Influence of slip preparation on prop- erties of cast ware : Am. Ceramic Soc. Jour., v. .32, pp. 105-11.3. Saunders, L. R.. Enright. D. P.. and Weyl. W. A., 10.50, \Vettability, a function of the polarizabilitv of the surface ions: Jour. Applied Physics, v. 21. no. 4, pp. .3.3.8-344. Schwartz, Bernard, 1052, A note on the effect of surface tension of water on the plasticity of clay : Am. Ceramic Soc. Jour., v. 35, no. 2, pp. 41-43. Shaw, M. C. 1037. Clay refining by flotation methods: Am. Ceramic Soc. Bnl!.. vol. 16. no. 7, pp. 291-294. Siefert. A. C, and Henry. E. C. 1047. Effect of exchangeable cations on hydrophilic nature of kaolin and bentonite: Am. Ceramic Soc. Jour., v. 30, no. 2. pp. 37-48. Smith, .T. B., 1051, Effect of water quality on the manufacture of ceramic ware and porcelain enamel : Am. Ceramic Soc. Bull., v. 30, no. 9, pp. 301-303. Speil, S., 1940, Effect of electrolytes on monodisperse clay-water systems : Am. Ceramic Soc. Jour., v. 23, no. 2, pp. 33-38. Speil. Sidney, Berkelhamer, L. H., Pask, J. A., and Davies, Ben. 1945, Differential thermal analyses — its application to clays and other aluminous materials : V. S. Bur. Mines Tech. Paper d, and the like, canimt be adecpuitely handled on the basis of studies made .so far. SEALING PROPERTIES Clay was first used in mud to prevent loss of the mud fluid itself; however, some of the mud is lost as it filters though the cake that seals permeable formations. The filter cake on the wall of the hole thus increases in thick- ness at a rate which depends upon many factors, includ- ing the thickness of the cake already formed, the viscosity of the fluid phase of the mud (generally a function of the prevailing temperature), the pressure differ- ential across the cake, and, most important of all, the permeability of the cake itself. Cake permeability in turn depends upon the nature of the solids in the mud, their state of flocculation, and to some extent the pres- sure differential across the cake, some cakes being rela- tively compressible. In view of the obvious disadvantage of having a filter cake build up against a permeable for- mation, reducing the diameter of the hole, it seems strange that so long a time elapsed between the intro- duction of the use of clay slurries for rotary drilling and the first study of caking properties of clays. A study of the jnoneering work of Kimpp (1923), in the middle twenties, followed by that of Jones and Babson (1936) and Jones (1938) will enable the reader to trace the development of cake permeability testing, generally known as filtration testing or wall-building testing. Sev- eral years ago the industry adopted a method of evalu- ating the filtration behavior of all types of muds. A full description is given in the Code for the Field Test- ing of Drilling Fluids published by the American Petro- leum Institute. In the test, which determines what is probably the most important characteristic of mud, mud is filtered under pressure through a filter paper under specified conditions of time, temperature, and pressure, so that the lo.ss of filtrate, rate of cake growth, and the physical nature of the cake can be determined. Methods of evaluating the data as well as variations in test pro- cedures are given in publications of Larsen (1938), Byck (1940), Williams and Cannon (1939), Proko]). and others. Table 1 shows the A.P.I, filter loss, or "water loss," as it is commonly termed, for fifteen different clays made up into slurries of fifteen centipoises viscosity, A. P. I., in distilled water. Measurements of cake weight and solids content, made at the same time, permitted the 272 Clays and Clay Technology Table 1. Filter loss of clay. I Bull. 169 Base- exchange capacity Y'ield in bbls. 15 cp mud per ton Solids in filter cake (Percent by weight) A. P. I. water loss at 15 cp Filter cake permeability microdarcys pH 1. Hectorite. pure (California) . 70 160 6.5 7 0.85 8.6 77 125 10 11 1.8 8.2 56 71 16 15 2.1 8.7 67 18 50 11 1.5 7.5 5 Tllitp fFithian Illinois) - 16 13 67 57 38. 7.4 3 14 70 190 285. 7.0 7 AtfcaDulffite (Florida) :. . 33 105 23 105 68. 7.1 16 17 60 35 15. 7.7 Q Plav CFrazier Mt California) - 51 28 37 22 5.5 8.9 10 Clav CEl Paso Countv Texas) 32 15 58 20 3.7 7.4 39 14 64 27 6.5 8.9 12. Clay (Rogers' Dry Lake, California, upper layer) 40 15 63 21 6.6 9.6 13. Clay (Rogers' Dry Lake, California, lower layer) 37 9} 2 66 8 1.3 9.2 28 16 55 43 19. 9.4 15. Clay (IVIcKittrick, California) 12 26 53 28 5.4 7.6 Note: Clays 1-15, inclusive, do not necessarily corresponil to clays now marketed for drilling mud use. Tlie names of clays 9-15, inclusive, are geographical localities, and not trade names. calculation of filter-cake permeabilities, and these are shown in microdarcys in the next column of table 1. It ma.v be seen that a roii,ti'h parallelism obtains between water loss and cake permeability, but it is the latter which is more important, so that it should be computed whenever possible as a matter of routine from the re- sults of the A.P.I, test.* The ojreat variation in filter- cake permeability among- the different clays shown is striking, and makes it clear why relatively few clays are suitable for drilling mud use. Many clay deposits which were extensively used two and three decades ago as sources of drilling mud have now been abandoned, since it was found that these gave muds having poor filtration characteristics. On the other hand, careful prospecting in known clay deposits in many cases has yielded bodies of clay with excellent fil- tration behavior, and it is common for processors of bentonite and other clay to make careful filtration tests as one means of product control. Of course a clay ex- • For the standard A.P.I, te.st conditions, the filter cake permeabil- ity is approximatel.v 0.7 x 10"' x (water loss in cc) x (cake thickness in 32nds inch). Table 2. Clay A (Run-of-mine Hector clay, 5 percent in water) liibiting good filtration behavior does so only under fa- vorable conditions ; aiw environmental conditions tend- ing to bring about flocculatioii of the clay and other solids in the mud will in general increase the filter-cake permeability of the mud. A filter cake is actually a sediment formed under pres- sure, so that for a slurry of a given clay, the sedimenta- tion volume should determine filter-cake permeability, other factors remaining the same. Flocculatioii and de- flocculation determine sedimentation volume, however. This is illustrated in table 2 (1946), which shows how deflocculation and flocculatioii of a typical drilling mud reduce and increase, respectively, the filter-cake permea- bility, the "sedimentation volume" of the filter cake, Treatment of suspension A. P. I. viscosity Perme- ability of filter cake Specific volume of filter cake! Specific volume of sediment^ Thinned with Graham's Salt. — No treatment Thickened with MgO - - . 15 25 115 0.7 1.2 8.1 6.6 7.4 11.6 1.6 2.4 25. I 1 1 1 1 1 1 ' ^ © - ® Q ® %^^ ® „< - ® ® - ® - 1 1 1 1 1 1 1 1 1 Microdarcys, calculated from A.P.I, filtration test. 2 Cc of cake per gram of clay, A.P.T. filtration test. 3 Cc of sediment per gram of clay, sediment Cfiilrifuged from 10:1 dilution of the clay suspension with distilled water. Figure 3. i'art \11J Clay Tkciinology in the Petroleum Industry 273 and the sedimentation volume of the clay as determined by centrifufrinp: portions of the diluted mud. Miscellaneous Functions of Clay in Mud It is of some interest to note that Speller ( 1936 ; 1938) several years aflocculating and degelling agents commonl.v us(h1 with ordinary cla.vs such as the polyj)hosphates, tanstiiffs, and humic acids. Attapulgite finds extensive use, therefore, in areas in which rock salt is drilled through, and in which the mud thereupon be- comes saturated, or nearly .so, with salt. In such a solu- tion, ordinary clays fail to swell and comi)lete loss of gel strength is the usual result. Attapulgite is added to such muds to give the gel strength necessary and desirable for cuttings removal, weighting material suspension, and the like. The acieular nature of attapulgite particles causes them to behave essentially as filter aids, much in the fashion of asbestos or diatomaceous earth. One result of till' fibrous structure is that muds made up with attajiul- gite have a very high filter-cake permeability, even though i)ure water has been used. Where it is necessary to have low cake permeabilit.v with such a mud, organic colloids that are so completely hydrophilic that they are unaffected by the presence of salt are generall.v added. Such colloids ma.v be gelatinized starch, generally added in a thin-boiling, pregelatiuized form (Larsen. 1947) ; gums such as gum trauacanth fllarth. 1935) and gum karaya (Kennedy and Teplitz, 1943) ; or cellulose deriva- tives, such as sodium carboxymethylcellulose (Wagner, 1947) and hydroxyetliyl cellulose (Himel and Lee, 1951). Illite. Much clay of the illite type becomes incorpo- i-ated in drilling muds from shales encountered during drilling, as has been noted. The only relatively pure deposit of an illite-type clay mined and sold for drilling mud use which has come to the writer's attention is the cla.v from Grundy County. Illinois, sold as grundite. Although this deposit is inactive at the present time, its properties have been described in a paper by Grim (1939). The properties of an illite by itself, as repre- sented bv clay from a type deposit, cla.v 5 of table 1, are uot favorable for a mud because of the fairly high cake permeabilit.v. Halloysite. Some attempt was made about ten years ago to market for drilling mud purposes a halloysite clay occurring in Colorado. No other halloysite, or even a clay rich in halloysite, is known to the writer as cur- rently used in muds. Kaolinite. There is no a priori reason to expect that either hallo.vsite or kaolinite would be well adapted for drilling mud use in relatively pure form, an expectation borne out bv table 1. A Tennessee ball clay is currently being marketed for use in mud for water-well drilling and for drilling shot holes for seismographie exjilora- tion : the demands on such a mud are not generall.y severe. CHEMICAL TREATMENT OF CLAY MUDS Addition of Degelling Agents In the ordinary course of drilling, clay -water muds tend to increase in consistency, which results is an increase in gel strength. The change in consistency may be the result of drilling through mud-making formations, or the result of the addition of soluble salts such as sodium 276 Clays and Clay Techxology I Bull. 169 chloride, calcium sulphate, and the like. In the early days of drilling with rotary mud, notably in the 1920 's, attempts were made to treat sucli thickened muds with the materials found useful in ceramic slips, particularly sodium hydroxide, sodium carbonate, and sodium silicate. Such attempts were largely unsuccessful ; the mechanism of the production of the gel is still controversial, bxit it may arise from quite different causes than are opera- tive in china clays. For example, the addition of a small amount of sodium hydroxide to a slip of kaolin in water will bring about marked thinning, whereas the addition of a like amount of caustic soda to a bentonite suspen- sion will cause thickening. Most claj'-water drilling muds tend toward behavior of the latter type, and until the early 1930 's the problem of degelling drilling muds by chemical treatment was largely unsolved. Today the chemical agents used for this purpose fall largely into two groups : inoi'ganic complex phosphates, and organic weak acids of generally high molecular weight such as tanstutfs and humic acids. An early anonymous publica- tion (1932) of the Shell Laboratories in Amsterdam lists many organic compounds useful for thinning muds. The introduction of the molecularly dehydrated phos- phates as drilling-mud thinners was made by Wayne (1942) in the early 1930's. At the present time, sodium pyrophosphate, both in the tetrasodium and the diso- dium dihydrogen forms ; sodium tripolyphosphate ; and the glassy sodium phosphates of obscure structure but ranging in composition from the so-called tetraphos- phate to the hexametaphos]ihate are widely used. Anal- ogous compounds such as tlie stannates (Wayne. 1951') and vanadates have been proposed but are not commer- cially used at the present time because of their scarcit.y and high price. Orthophosphates are used scarcely at all, and other simple sodium compounds which remove cal- cium, such as sodium carbonate and sodium bicarbonate, are used only for special treatments and generally in combination with other materials. The mechanism of degelling by complex phosphates is no less obscure than the mechanism of the gelling of the muds in the first place. Grim (1947) has suggested that gelling may be ascribable to the building up of layer i^pon layer of hexagonally disposed water molecules on the surface of the clay particles, aiul that degelling b.v complex phosphates may result from the fit of polyphos- y)hate chains and rings onto the same surface, thereby interfering with the building up of oriented water layers. Van Olphen (1951) has advanced the hypothesis that the edges of montmorillonite flakelets are at least in parts positively charged, that the gelation of benton- ite in aqueous si;spensions therefor comes about from an edge to face orientation which would naturally give a gel of low solids content, and tliat accordingly the action of the polyphosphates is essentially one of adsorp- tion at the loci of positive charges with the consequent neutralization or even reversal of charge at these points.* These are both reasonable proposals, and the most that can be said at this time is that more work must be done on the problem. The complete solution will no doubt involve full consideration of the structure of polyphos- phates as compared with those of orthophosphates: the •.I. EndeU (1950) has recently pubUshed an electron micrograph showing particles of a gold sol (presumably negatively charged) adhering to the edges of kaolinite platelets. existence of P-O-P bonds with tetragonal coordination of oxygen ions about the phosphorus ions, analogous to silicates; the relatively high ionization constants of the polyphosphates (as evidenced, for example, by the feeble alkalinity of tetrasodium polyphosphate and the actually neutral pH of many metaphosphates, including Graham's salt — the so-called sodium hexametaphos- phate), which would enable neutralization of the postu- lated positive charge of the montmorillonite fiake by one portion of the polyphosphate molecule and the pres- entation of a negative charge by another portion; prob- ably also the ability of certain polyphosphates to sequester calcium ion ; and finally the ability of polyphos- phates to prevent growth of certain crystals where it is possible for them to fit on the lattice of the growing crystal, such as on calcite. Cyclic trimetaphosphate neither thins mud initially, nor i^revents the growth of calcite crystals, but after hydrolysis has taken place and the ring is split, it becomes an excellent thinner. This is in complete accord with the findings of Raistriek (1949), who found that nucleation of calcite crystals was not prevented by tri- and tetrametaphosphate at first, but was after hydrolysis of these ring compounds. It is significant that the principal effects of the addition of polyjiliosphates to a drilling mud is reduction in shear strengtli rather than increase in Bingham fluidity (or reduction of differential viscositv). (Melrose and Lilien- thal, 1951 ; van Olphen, 1951.) The organic compounds that have jiroved most useful in thinning drilling muds are the several tanstuft's, of which c|uebracho extract is the most widely used, and liumic acid, generally derived from certain lignites. Tlie fact that these materials thin best at an alkaline pll, and indeed at pH's high enough so that fairly complete ionization of the acids can be assumed, suggests that the mechanism may be related to that which gives the poly- phosphates their effectivness. Indeed, at pH's of 12 or higher, gelatinized starch acts as a thinner for day- water muds in a way very like the tanstutfs, and the higher pH involved is in accordance M'ith the much lower ionization constant of the alcoholic hydroxjd groups in starch as compared with the phenolic and even carboxylic groups in tannic and humic acids. A discus- sion of some aspects of the behavior of tanstutfs and polyphospliates in thinning muds is given in several articles including which may be mentioned those by Loomis, Ford and Fidiam (1941), Garrison (1940), and van Olplien (1950 b, 1951). A special usage of tanstuffs and humic acids, aside from simple thinning in normal, moderately alkaline pH range is discussed below. Lime Muds. The term "lime muds" as currently used needs some explanation. In the early days of drilling, limestone was often added to mud, occasionally as ground massive limestone and .sometimes as beet- sugar-factory waste lime, and such muds were occasion- ally called "lime muds." Also, calcium hydroxide, as ordinary slaked lime, formerly was added to mud when a thickening effect was desired, but the adverse effect upon filter-cake permeability led to its abandonment. Within the last decade, the new and almost revolution- ary technique has been developed of using slaked lime in clay-water drilling muds at a high pll and in the presence of a protective colloid, which may he of the Part VII 1 Clay Technology k\ the Petroleum Industry 277 class of laiistuffs and liuinic acid derivativps, or gelatin- ized stareh. The term "protective colloid" has been deliberately used, because the system is quite completely flocculated without the organic material, and it has only been througii knowledge of the proper use of these organic materials in conjunction with lime that the latter has been usable at all. The pIT i-ange of such muds is high, generally between 12 and ]3, and is achieved generally by the joint use of caustic soda and hydrated lime. The concentration of the organic material is like- wise high, ranging upward from 4 or 5 pounds of or- ganic material per barrel of mud. When clay is added to such a mud, either as a drilling-mud clay deliberately added or as ground-up cuttings encountered during drilling, even though the clay w'oidd normally readily become dispersed in water, dispersion in the lime type of mud is at a minimum, and only a slight contribution is made to the consistency of the mud. This effect is of enormous advantage in drilling through long intervals of mud-making shale, and indeed, with jjroper mud con- trol, there seems to be no limit to tlie footage of normallx' dispersible shales wliich can be so drilled. Such miuls are characterized principally by very low gel strength both after agitation and after a period of quiescence. and by quite low filter cake permeability. Substantial contamination by salt can occur without harming the mud. particidarly when gelatinized starch is present. Muds of this type are generally made from a mud al- ready in use, and the conversion of ordinary muds into high pTI lime muds calls for a great deal of skill on the part of the mud engineer, since the initial tlocculation which takes place results in great thickening of the mud, which, however, is temporary. The usual explana- tion given for the behavior of such a mud is that the presence of the calcium ion prevents swelling of any normally swelling clays, as well as collapsing clays of that type present in a mud undergoing conversion, that the oi-ganic thinner, together with any starch which may also be present, prevents adhesion between the particles so treated by the calcium ion. Field technology of such muds is at tlie present time still ranch more of an art than a science, and although fairly ]irecise con- trol techniques have been developed, involving for the most part the reaction of the mud and the filtrate to chemical titrations of several kinds, control and remedial treatments are largely empirical. However, it appears that too high a concentration of calcium ion is unde- sirable, and in practice the calciinn-ion concentration is carefully controlled, possibly at a level somewhat under that which would correspond to complete calcification of the clays pi-esent. This is suggested by the ability of bentonite, when added to such a mud, to behave in a nearlj- normal manner as regai-ds filter-cake-permeability reduction ; although the recent work of Gruner and Vogel (1950) on the swelling of calcium bentonite at high pH's suggests a possible mechanism. The quantities of lime consumed during the course of drilling through shales are substantial, and obviously considerable base exchange takes place during the use of such a mud. Lime muds of this type have been the subject of several re- cent reviews (Goins, 1950; Lancaster and Mitchell, 1949 ;McCray, 1949). A recent development in lime-base muds of particular interest to clav mineralogists is the discovery that cer- tain of these muds, when used for drilling at great depths where unusually high bottom-hole temperatures are encountered, undergo a solidification to an extent occasionally so great that the mud resembles set port- land cement. A laboratory method of reproducing the various solidification phenomena on a small scale has been described by Graj', Cramer and Litman (Gray, Cramer, and Litman, 1951) ; and an investigation from the stajidpoint of the mineralogical changes involved has recently been published by Gray, Neznayko and Gilke- son (1952), who found that calcium silicates, among which analcite and xonotlite could be identified, were formed, apparently from the clay and quartz originally present. It was found that retardation of the solidification coidd be brought about by the use of organic dispersing agents of the type which ai'e effective as portland-cement re- tarders. Salinification of Muds Interstitial clays occuring in oil sands, either as clays laid down with the sediment itself or possibly formed in place, have recently received much attention from the standpoint of their interaction with fluids intruding from the well bore (Darley, 1947; Griffiths, 1946; Grim, 1947; Johnston and Beeson, 1945; Suter, 1948). Prior to disturbing the reservoir by drilling, such clays are of course in equilibrium with any interstitial water present, which almost universally is substantially saline in char- acter, and (piite generally contains ions of alkaline earth metals as well as of alkali metals. I'nless a drilling mud is of a special type, the filtrate therefrom is likely to have a much smaller ionic strength than that of inter- stitial reservoir waters, an eft"eet which indeed has been recognized from the earliest days of electrical w^ell- logging. One miist. for a variety of reasons, prevent intrusion of drilling-mud filtrates in the producing for- mations as much as possible, not only for the reason dis- cussed here, but also because of the possibility of reduc- ing the pei-meability of the sands to oil from purely capillary effects. However, with practically all muds having an aqueous base, some lo.ss of filtrate is nearly unavoidable, and severe reduction in permeability of oil-bearing strata to the flow of fluids of any kind has been shown to be the result of intrusion of relatively fresh water filtrates, particularly when the sands contain appreciable amounts of clay. Even where the clay is not of a swelling type, undoubtedly a reduction in adhesion of the claj' or other minute particles to the surface of the sand grains can be brought about by the intrusion of fresh water in accordance with the iirinciples of ad- hesion capability investigated by von Buzagh (1930). Table 3, taken from von Buzagh (1937), shows the reduction in adhesion of quartz particles to a quartz plate, in the presence of distilled water and of increasing concentrations of NaCl and CaCL. The sine of the angle of tilt necessary to dislodge the quartz particles by gravitational action is shown for different concentrations of the salts. The great increase in adhesion for even quite dilute electrolyte concentrations as compared with fresh water is clearly seen, as well as the difference between sodium and calcium chlorides at the same concentrations. The possibilit.y of the dislodgement of fine particles normally adhering to the surface of sand grains by the interposition of fresh water or weak brines of alkali- 278 Clays axd Clay Technology [Bull. 169 Table 3. Concentration in millimols per liter NaCl CaCh Percent sins Percent sina 0.058 0.117 0.233 0.095 0.165 0.342 0.43 0.055 0.111 0.222 095 5 317 10 20 573 40 metal salts alone, whereby such very small particles can move to block the fine openings in the capillaries, is evident from table '.i, so that swelling- clay or even clay of any kind does not have to be present to explain many of the results obtained in Huid-permeability in- vestigations. Investigation of the swelling- of micro- samples of the fine fraction obtained from oil-sand cores subject to permeability reduction by fresh-water fil- tration by means of the Enslin (1933) apparatus, carried out in 1947 by Foster (Poster, 1947), showed a relatively feeble swelling, yet the core was from a formation i)ar- ticularly subject to permeability reduction from fresh- water intrusion. (It is not intended here to minimize the im|M)rtance of clay swelling in formation permeability, but merely to call attention to another factor which has perhaps been overlooked in the past, and may be found to be of some importance.) Recent work by the research group of the Union Oil Company of California has in- cluded the investigation from a very fundamental and very thorough standpoint of the clay-mineral content of the fine fractions of California oil-sand cores (Xahin, et al., 1951; No\wik and Kruger, 1951). The treatment of drilling muds so that their filtrates will fail to have a swelling or dislodging effect or both upon interstitial clays and other fine particles has been given much attention in recent years. The usual expedi- ent, as one might deduce either from clay-swelling data or adhesion data of the von Buzagh type, is to salinify the mud, particularly with salts, of di- and trivalent cations (Cross and Cro-ss, 1937; Darlev. 1947; Griffiths, 1946; Sherborne, 1951; Temple, 1951)." Calcium chloride has been used in concentrations of several percent by weight of the fluid phase of the mud, and even soluble aluminum salts have been used. Naturally, the effect intended for the interstitial clays is not without a like effect on the clays in the drilling muds themselves, and unless special measures are taken the mud properties will be adversely affected by the addition of such salts. Fortunately, gelatinized starch is fully effective to ob- tain low filter-cake permeability in such cases, and is generally used with such a mud. The filtrate from a lime mud such as has been discussed in the previous section falls somewhat short of ideal performance in this con- nection, partly no doubt because of its very high pll. Reversal of Charge Some have suggested not merely converting a mud to a calcium-base clay, but reversing the charge to a posi- tive micelle. Perhaps tlie most elaborate work along these lines is that of Bond (1944), who used such materials as methylene blue for producing- a positively charged bentonite for u.se in drilling muds. With such a mud anv swelling or even dispersion of normally hydratable clay encountered during the course of drilling is of course out of the question. Muds of this particular type do not appear to have been used commercially, however, prob- ably because of their relatively high cost and also be- cause of the intervention of the higli-pll-lime technique, which has solved most of the problems in connection with heaving-shale drilling. USE OF CLAYS IN MUDS CONTAINING OILS Emulsion Muds AYithin the past decade the use of emulsions as drill- ing fluids has become fairly widespread. Both types of emulsion, that is, oil-in-water and water-in-oil. have been used, but the former is much more common. Substan- tially all of the oil-in-water type of emulsion drilling fluids presently used contain clay, and indeed are most generally made up b.v tlie addition of oil to a clay-water mud which has already been in use. The ability of colloidal clays, particularly bentonitic clays, to act as emulsifying agents for oil-in-water emulsions is so well known that little need be said here. In fact, it has been found that some clays are .such good emulsifying agents in themselves that no additional emulsifiers need be added, but this property is seldom relied upon in actual practice, since the use of organic emulsifying agents in a clay-water mud can readily be made to give great stability to the emulsion. One function of clays in such muds which is generally overlooked, but which neverthe- less is most important, is that of providing a preferen- tially water-wettable filter cake for the mud to filter against. The fact that the walls of the pores jn such a clay filter cake are hydrophilic prevents the passage of the minute droplets of emulsified oil, which presumably are then enabled to block off the pores and to reduce the over-all permeability of the cake. The filtrate from such a mud. when the emulsion is properly stabilized, is substantially 100 percent water, and all of the oil is retained by the cake, which is in accordance with the picture just presented, as is also the very considerable reduction in over-all fluid loss resulting from the emulsi- fication of oil into the clay-water base mud. An interaction of the clay with the emulsifying agent used in a particular emulsion mud has been studied by Doscher (1949) ; the particular mud proposed is of in- terest as it makes use of calcium ions for tying the emulsifying agent to the clay inirticle as well as to pre- vent swelling of clays both in the emulsion and such as may be contacted by filtrate from the emulsion. Oil Base Muds The attempt to keep intruded water entirely away from oil-bearing formations has led to the development of drilling fluids made up with oil instead of water, the so-called oil-base muds. For the most part, such drilling fluids do not have clay added, and since they are not ordinarilj' used for "making hole" as distinguished from drilling into an oil-bearing formation, clays en- countered during drilling with such a mud are generally of little consequence, aside from the fact that they are generally not dispersible in the fluid in the first place. Clays are of importance, however, in particular oil-base muds which have been used. Rolshausen and Bishkin I'art N'lli Clay TECHNOi.oiiv in the Petroleum Industry 279 (1937) proposed the incorporation of beatonite in an oil-base mud in order to take up any free water en- countered during drillinfr. A recent article by Lummus and Dunn (19521 describes a recently developed type of oil-base drilling fluid made up from oil and a dry con- centrate, which is stated to consist of "a chemically treated mixture of clay and weighing material." Another and most interesting type of oil-base nuid in which mortmorillonitie clays play an all -important part is that patented by ITauser (1950), wliich makes use of the swelling properties in oil of reaction products of long chain amines and hentonite. also discovered by him fl95(); and Jardan. 1952). Such olcopliilic bentonites, wliich are substantially completely coated by the hydro- phiihie organic cation through a base-exchange reaction, impart both gel strength and, because of their platy character, low fluid loss to oil-base muds, and also enable the cmidsification of any water encoiuitered during u.se into the nnid as a stable watcr-in-oil emulsion. A par- ticidar advantage is that the organophilic clay retains its gelling ability even at very high temperatures, tlius avoiding a difficulty often encountered with other types. DISCUSSION H. F. Coffer: I would liI;o lo have furtlici' disriission on the reactions of chemicals in thinning muds. W. T. Cardwell, Jr.: The work of I.oomis. Fonl. and Fidiani (1!U1) several years .-igo is still liasic. Their work showed a correlation hetween the calcid.ited surface areas of the ela.v particles and the amount of p.vroiihosphate required in a suspension to attain minimum viscos- ity. This amount was just sufficient to form a monomolecular layer of the pyrophosjihate on all the surfaces of the clay in the sus- pension. The ad.sorption of pyrophosphate on the clay surfaces presumal>ly prevents the formation of an extended water struc- ture. Formation of this extended water structure is thought to account for the viscosit.v. H. van Olphen: Actual adsori>tion studies indicate that adsorption of the poly- phosph.ite is not complete. It is, however, sufTicient to be explained as lateral surface adsorption. A minimum in the '"viscosity" curve occurs owing to a balance hetween the peptization by the pol.y- phosphate anion and the fioccul.ition by the Xa* ions; at high concentrations the Xa-polyphosiiliate finally flocculates the clay completely (van Olphen, IH.'ila). As I see it the phosphate-anion adsorption charges the particle edges and prevents their edgewise association. Hence no volumi- nous clay-particle network is built up and the yield stress of the system is reiluced. ( The so-called "viscosity" rediictii>n of the drill- ing fluid is primarily yield-stre.ss reduction. I 30 minutes • yield stress yield stress dynes/sq.c zero minutes yield stress 3 NoCI cone me/I Figure 4. The resent in the iutermicellar licpiid may charge the broken silica sheets jiosi- lively as they do with a quartz sol. Therefore the i)re.sence of a positive edge charge would be in line with general colloid chemical experience. Turning to the reduction of the yield stress of the pure gel by addition of about .3 me./l of NaOl, Na^SO., XaOH, or CaOb, this ma,y be seen as the result of the compression of the double layer on both flat surface and lateral surface, thus reducing the mutual attraction of the two. On further addition of these salts, however, both double layer potentials will be so far reduced that edge-to- edge or flat-surface-to-flat-surface association is no longer pre- vented and the system gels again owing to dominant van der Waals' attraction. However, this gel is thixotropic and has a completely different chara<'ter. This is tiie condition of a virgin drilling fluid when ijrepared from impure cla.v and water. The reduction of gela- tion b.v treating chemicals is now achieved by adsorption of anions to the positive edges, which gives the edges a protective negative charge, edge-to-edge association is prevented, and the main contri- bution to the formation of a voluminous network is eliminated. It may be mentioned that the ma.ior rheological effect of treating agents on the gelled clay suspension is a reduction of the Bingham yield stress of the system, the differential viscosity remaining es- sentially the same. E. A. Hauser: The development of organic-reacted bentonites has completely revised the old theory of thixotropy for we are now dealing with a system where electrokinetie forces are no longer involved. A thixotropic gel can be formed by reactions such that the indi- vidual particles develop a balance between the solvated layer and the van der Waals' attraction forces. The moment this balance is reached, the system sets up as a thixotropic gel. 280 Clays and Clay Technology [Bull. 169 SELECTED REFERENCES Ambrose, H. A., anl Loomis, A. G., 1935, Fluiilities of thixo- tropic gels : bentonite suspensions : Ph.vsics, v. 4, pp. 265-273. Anonymous, 1932, Die Bebandhing K. Delft. van Olplien, II., 1051a, liec. tra\. cliini. pays bas, v. 09, p. 130S, D.'lft [Thesis]. I'iKott. K. ,T. S., 1942, Mud tinw in drilUuR : Am. Petroleum Inst., Drilling and Production Practice 1941, pp. 91-10?,. Proliop, r. L., 1!)52, Radial filtration of drilling mud: .Tour. I'ctroletim Technolog.v, Petridenni Trans., v. 105, pp. u-10. Haistrick, 15., 1040, The influence of foreign ions on crystal growth from solution. I. The staliilization of the supersaturation of calcium carlionate solutions hy anions possessing 0-P-O-P-O chains: Faraday Soc, Discussions, v. 5, pp. 284-237. Kolshauscn. F. W., and Hishkin, S. Ia. 10;!7. Oil liase hydrat- ahle drilling fluid : V. S. Patent 2,009,825, of Nov. 2;5, 1037. Ro.ss, C. S., 1041, Sedimentary analcity : Am. Mineralogist, V. 20, pp. 027-029. Roth, Robert S., 1951, The structure of montmorillonite in relation to the occurrence and properties of certain bentonites : Univ. Microfilms, Pub. 2741, Microfilm Abstracts, v. 11, pp. 091-092. Shaw. B. T., and Hund)ert, R. P., 1941, Electron micrographs of clay minerals: Snil Sci. Soc. America Proc, v. Ti, pp. 140-149. Sherborne. .1. K.. 1951. Rotary drilling fluids: U. S. Patent 2,551,708, of May 8, 1951. Speller, F. N., et al., 10,30, Report of special subcommittee on corrosion fatigue of drill pipe: Am. Petroleum Inst. Proc., sec. 4, V. 17, pp. 07-00. Speller, F. N., 19.3S, Prevention of corrosion fatigue failures: U. S. Patent 2,132,580, of October 11, 1938. Suter, H. II.. 1948. Mineralogie und Pctrographie in der Krdcil- industrie: Schweiz. Mineralog. und Petrograph. Mitt. Band 28, Heft 1. pp. 12-3.5. Temple, Scott, 1951, Composition and treatment of drilling fluids: U. S. Patent 2,571,003, of October 10, 1051. Yerwey, E. .T. AV., and Overbeek, .1. Tb. G., 1048, Theory of the stability of lyophobic colloids : Elsevier Pub. Co., Inc. AV.'igner, C. R., 1047, Drilling fluids and methods of use: V. S. Patent 2,425,7()8, of August 10, 1047. Wayne, T. 15., 1042, Trealment of drilling fluids: U. S. Patent 2.204,877, of September 1, 1942. Wavne, T. 15., 1951, Treatment of drilling fluids: U. S. Patent 2,.5.53,224, of May 15, 1051. Weber, E., 1000, The liquefaction of clays by alkalis, and the use of fluid clay casting in the ceramic industry : Ceramic Soc. [Eug.], Trans., "v. 8, pp. 11-22. Williams, C. E., Jr., and Bruce, (I. H., 1051, Carrying capacity of drilling muam. tion in the permeable formation to the more dilute solu- tion in the borehole. The current flow is not large ; in typical instances, it is of the order of milliamperes. The current density in the shale is very small. The shale cell written in conventional fashion is Mud//Interstitial water in the permeable rock/ Shale/Mud If the mud and interstitial water are both assumed to be sodium chloride solutions differing only in ionic strength, and if the shale potential is considered to fol- low the Xernst equation irrespective of the ionic strength of the solutions separated, the cell may be formally written NaCl/'/NaCl /Xan^lectrode XaCl a-2 ai aa (2) The expression for the potential of a cell of this na- t\ire involving as it does both the equivalent of a sodium electrode potential and a liquid junction potential is KT fli E = 2t, /« — F a-. (3) where f_ is the transference number of the chloride ion and ff] and ^2 the mean ionic activities of the two sodium chloride solutions. For practical use equation (3) may be readily ca.st in a form in which the potential E is related to the activity ratio (ai/02) and temperature T. For this purpose it is only necessary to assume some reasonable average so- dium chloride concentration for which the value of the transference number as a function of temperature can be found. If E is no-w identified with the maximum or static S.P. on a log and the activity of the mud, a^. is measured, it is possible to arrive at a value of (7i, the activity of the interstitial water. Knowing this activity the salinity and hence resistivity of the interstitial wa- ter is easily found. Simple charts to accomplish these steps have been publi-shed (Wyllie 1951). Equation (3) thus forms the basis of a method of computing interstitial water resistivity. The method has been used with success by the author (W^yllie 19-t9a) and is now standard practice in Gulf operations. A similar method was evolved independently by Tixier (1949) from a consideration of the S.P. phenomena and interstitial water resistivities observed by him in the Rockj' Mountain area of the United States. Tixier 's method, based purely on log data, so closely approxi- mates the results of the writer's theoretical analysis that it may be construed as lending support to the analy- sis made. "While the method is used with success within the Gulf organization, others have not found it equally use- ful.* A major difficult}' is certainly that of obtaining a fully developed S.P. deflection w-hen geometrical ef- fects, e.g., thin beds or excessive filtrate invasion, are conspiring to reduce the maximum S.P. deflection. The formulae evolved by Doll (1949) for correcting the S.P. for such geometrical effects are apparently too idealized to be invariabl.v successful in practice. However, static S.P. logging will do much to overcome these difficulties, * Dunlap, H. F., Personal communication, 1952. 28S Clays and Clay TEcnxoLOfiY [Bull. 160 while the interpretation of eon vent ional S.P. logs may be greatly improved if the maximnm consistent S.P. deflection above and below (for preference) or in the vicinity of the bed of interest, is selected. However, some judgment is required in the use of this technique, for in some areas the salinity of intei-stitial waters may change abruptly from bed to bed while in others the change is gradual. Thus on the Gulf Coast the technique may be reliable, but in California it is hazardous. It is the experience of the writer and his associates that in the United States the method may be used to calculate a water resistivity sufficiently reliable for a log interpretation in about 70 percent of cases. In an effort to improve this situation and to determine to what extent the results obtained have been fortuitous, atten- tion has been given to the basic assumptions made in the formulation of equation (.3) and the degree to which these assumptions may be relied upon in practical in- .stances. This program is still continuing, but has de- volved into two basic lines of inquiry. These are an in- vestigation of the existence and magnitude of potentials other than electrochemical potentials which are recorded on the S.P. log, and the mechanism of the electrochemical potential. The latter problem has involved both the theory of charged membrane electrodes and particularly the problem of the bi-ionic potential. Elecirokinetic Potentials. In drilling operations using rotary rigs, the borehole contains mud fluid. This fluid is designed to remove cuttings produced at the drill-bit and to seal otf permeable beds while drilling proceeds. Tlie density of the mud is nuiintained at a figure which is sufficient to ensure that tlie hydrostatic pressure in the mud column opposite any permeable bed exceeds the pressure of the fluids in tlie bed. While this pressure difference ensures that no bed yields its fluids into the borehole during drilling, it inevitably leads to the steady infiltration of drilling mud into all permeable beds. To minimize this infiltration nuuls are designed to build on tlie faces of permeable beds a filter cake having a per- meability as low as the ingenuity of the mud engineer can devise. Nevertheless, no filter cake is impermeable and in all wells there is a steady flow of mud filtrate from the borehole through the filter cake into permeable beds. The pressure differential causing the flow of filtrate is a function of the mud density, the formation pressure and the location of the bed in the borehole. While all three factors vary, it may be said that in the depth ranges now of interest in petroleum production the pres- sure differential is always of the order of hundreds of pounds per square inch and in deep wells it may be several thousand pounds per square inch. There is also a range in resistivity of modern drilling muds, but the present tendency seems to be for them to be progressively lowered. Resistivities of the order of 0.5 ohm-meters to 1..5 ohm-meters at a temperature of about 70° F are now common, although figures both above and below these limits are frequently encountered. These resistivities implv ionic .strengths of the order of 0.05 to 0.20 molal. Such ionic strengths would not normally be considered as liable to give rise to electrokinetic potentials of any magnitude, but experiments have shown (Wyllie, 1951) that this assiunption is not justified in well-logging. Although the high ionic strengths of the filtrates which are forced through the filter cakes sheathing permeable beds are not themselves conducive to high streaming potentials, they are more than outweighed in total effect by the high zeta potentials of the filter-cake materials and the very large pressure diftVrentials causing fluid flow. The filter cakes are largely composed of montmoril- lonite clay, the exchangeable ions generally being so- dium; but probably some calcium also is present, jiartic- ularly in the case of modern lime-base muds (liergman, 1952). The cakes also contain weighting material in the form of barytes, along with silica and other solids arising from the drilling operations. Oil in the filter cake is characteristic when oil-emulsion drilling muds are em- ployed. It has been found that since it is deformable the permeability of a filter cake is itself a function of the pressure differential across it ; thus the streaming poten- tial across a filter cake is not the usual linear function of the pressure differential inducing fluid flow. The streaming potential appears to follow a relationship of the form E, = kP>' w Here E^ is streaming potential in nn-. k a con.staut which depends primarily on the mud resistivity (and thus on the mud filter-cake resistivity since these are interde- pendent) and J/ is a constant for any particular mud at a particular temperature. The value of y seems to depend upon the deformation ability of the filter cake ; an avei-age value is about 0.75. In figure 4 are shown some average values for the .streaming potential-pressure relationships of drilling muds of different resistivities. The resistivities are expressed at a temperature of 25° C. The fact that the constant k in equation (4) is approx- imately proportional to the resistivity of the mud from which a filter cake is derived is explicable if it is con- sidered that the total non-conductive solid content of a filter cake is largely independent of resistivity. The solid content is probably related to the mud weight. Hence k would be expected to be proportional to the mud-filtrate resistivity if the surface density of the charge on the filter-cake particles were either constant or filtrate-resistivity dependent. Data ((uoted by Mar- shall (1949) seem to show that the surface-charge density of kaolinite is practically concentration inde- pendent when the clay is in the sodium form, and the same may be true of montmorillonite. Thus in the range of ionic strensths characteristic of drilling muds the sur- face-charge density of the montmorillonite (bentonite) which is invariably used may well be considered con- stant. If this is so, the comparatively systematic varia- tion with resistivity of the electrokiuetic-potential char- acteristics of muds of widely dift'ering types becomes explicable. Nevertheless, the eft'ect of temperature on the charge density is not clear (Wyllie, 19511 and remains to be investigated further. The charge on the filter cake is such that the filtrate which penetrates a filter cake is positive with respect to the mud from which it is filtered. This means that when the electrochemical S.P. is negative, the S.P. de- flection is numerically increased by the streaming po- tential. The maximum S.P. or Static S.P., which is I'j.rt VTT] Cl.AY TECHXOLOfiV IN THE PETROLEUM InDUSTRY 287 10 100 I00< PRESSURE DIFFERENTIAL (P.SI.) FiGURK 4. Averase slreaiiiinj: iiutciitial-iliffcrciitial pressure relationships fttr aijueuus drilling; iniids. asKUiiK'd ill tlie practical application of equation (3) to he wiiolly clectrocheinical in orif>in, is in fact the aljiTcbraic sum of an electrochemical potential and an eh>ctrokinetic potential. The majrnitude of the electro- kinetic component of the total S.P. cannot always be considered negligible as the data of figure 4 reveal. For liigh nmd resistivities in particular, and even for com- paratively low mild resistivities if the pressnre differen- tial is liigh. the electrokinetic comi)onent may be far from nc!ili^ibh'. Thus in deep wells, jiai-tieularly, it would seem that allowance must be made \'in- the elec- trokinetic jiotential. From a purely practical standpoint, assnmiiifi' that as an apprii.xiuiation no correction for the electrokinetic po- tential is hi'ini: made, it is interesting to note that the error in tiie computation of an activity ratio from erpia- tion (3) will not be a function of the ionic strength of the interstitial water. This fact is shown grapliically in figure .5. This figure reveals, for example, that the ratio of appai'ent activity of interstitial water to true activity of interstitial water is 1.65 whether the true activity of the interstitial water is either twice or one hundred fold greater than the mud activity.* It also shows that to use etpiation (3) empirically, by arbitrarily adjusting the value of 2f_ NT F to allow for the effects of elec- trokinetic potentials, is not practicable since the em- pirical constant would be a parameter contingent ujion • However, the error in the calculation of inter.stitial water resistiv- ity is less for high absolute interstitial-water activities. Both the shape of the resi.stivity-salinity curve at high salinities and the nature of the activity-salinity relationship at high activities con- tribute to this fact. the absolute activity of the interstitial water. This pro- cedure is mentioned since it appears to have been at- tempted in field practice. The only practical method of reducing errors resulting from electrokinetic potentials, without making a specific correction for tiie electrokinetic component of the S.P., is to lower the mud resistivity. This has been discussed previously (Wyllie, 1951). Another method of reducing the streaming potential is to adjust the exchangeable ion on the montmorillonite in the filter cake, possibl.v by substituting a pol.yvalent ion for the predominant sodium ion. However, this seems impracticable both because of the expense involved and because the large zeta potential itself contributes de- sirable properties to the mud as a plastering agent and as a susi)ending medium for drill cuttings. Also, as will be discussed below, the presence of any substantial quantities of polyvalent ions in the drilling mud would seriously complicate the interpretation of the electro- chemical component of the S.P. curve. Investigations of the electrokinetic component in actual field operations are now being carried on using a device ( Wyllie, 1951b) which makes use of the hydro- static pressure in the borehole to build a filter cake on the exterior of a porous container. The container volume is sufficiently large that its internal pressure is not seri- ously increased by the accumulation within it of mud filtrate. As the device is lowered down the borehole, the streaming j)otential between a reversible electrode within the container and a similar electrode outside is auto- maticall.v recorded as a function of depth. Since the internal pressure of the container is almost constant, the streaming potential across the filter cake is recortled as a function of the external h.vdrostatic borehole pres- sure. Using this device the electrokinetic potential char- acteristics of muds can be logged in the borehole. This procedure eliminates the effects of mud aging which may occur when mud samples arc transferred to the labora- 1 UJ S 1- in Q S Z u. o ^ t- > > K o < 1 I6S 100 -- — -- ■- -- ■- — ■ — — — - — — / < / / / y o,^ / y f \ ^ / V !/ Y 9 t f / nr /I c?/ 3.3 ELECTRO POTEN KINE TIAL no / A .f 20 1 / y 1 / / ^1 1 1 J — 10 20 30 40 50 60 70 80 90 100 110 120 130 140 rSO 160 TOTAL SP (MILLIVOLTS) FiGUKE 5. Effect of an uncnrrected electrokinetic potential com- ponent on the calculation of activity ratios from the S.P. curve. 288 Clays and Clay Techxology' I Bull. 169 tory for testing. Also it is possible to attain readily differential pressures of the order of thousands of pounds per square inch. For technical reasons such pressure differentials are not readily obtained in the laboratorj'. Tests of this device appear to confirm the accuracy of equation (4). They may also lead to further knowledge concerning the range of pressure differential over which the exponent .(/ in the relationship Eg ^= A-P" may be considered constant ("W.vllie, 1951). In exceptional cases where pressure dift'erentials are very high and the interstitial water activity is lower than that of the mud, it appears that the streaming potential may be sufficiently large to effect an entire reversal of the S.P. curve. Thus in figure 6 the S.P. of the sand at 10,215 to 10,155 feet is —38 mv. Water was actually produced from this formation and had a salinity of 2,083 mg/1 of which 022 nig/1 were sodium ions, 5.8 mg/1, magnesium and 23.2 mg/1. calcium. The measured pressure differential across the filter cake was 1950 psi. For the mud resistivity used, 1.55 ohm-meters at 25° C (mud filtrate resistivity 0.66 ohm-meters), figure 4 indicates a streaming potential of about — 60 mv. This implies an electrochemical potential of -|-22 mv, which for a formation temperature of 194° F leads to a salinity, based on equation (3) and the curves given in W.yllie (1951), of about 3,000 gm/l. The agreement here is satisfactory, but it may be noted that no allow- ance has been made for the possible effect of tempera- ture on the electrokinetic characteristics of the drilling mud filter cake. AVhile preliminary experiments appear to indicate that such a temperature effect is not large, it is not yet possible to state whether or not this is gen- erally triie. In fine, however, it appears that streaming potentials cannot be neglected if equation (3) is to be used for the purposes of practical computation. Thus methods of logging the streaming potential in the borehole or of otherwise allowing for it should be further investigated. This is particularly true for regions such as Venezuela and parts of California where the low natural salinity of interstitial waters contributes to the use of relatively high resistivity natural drilling muds. Where drilling mud resisitivities at 25° C are 0.50 ohm-meters or less. Figure 6. A reversal of electrooliemioal S.P. which results from high electroliinetic S.P. (electro- chemical S.P. at 10125-101.5."!' is positive). it is probably permissible to forego any streaming po- tential correction unless pressure differentials are known to exceed about 1,000 psi. Nevertheless, the practical use of equation (3), assum- ing that a true electrochemical potential can be obtained, is contingent upon its validity. The question of the va- lidity of equation (3) is discussed below. Shale as a Memhrane Electrode. Wyllie and Patnode (1950) suggested that the electrochemical properties of natural shales could be rationalized if it were considered that these materials represented natural embodiments of charged membranes. They also presented a brief dis- cussion of what they believed was the essential electro- chemical structure of shale membranes, drawing an anal- ogy between the structure of shales and the structure of heterogenous membranes prepared by bonding arti- ficial cation-exchange materials with inert and insulating plastics. It is proposed here to amplify these concepts. It may be noted that there has been no alternative to the structure tentatively proposed other than one by Wil- liams (Wyllie, 1949), who in a discussion of an early paper by the writer, appeared to suggest that the electro- chemical energy of the shale cell was less a function of the properties of the solutions separated by the mem- brane than a property of the membrane itself. In Williams" view the shale is in equilibrium with the saline water contained in its interstices. He suggested that the chloride ion is more strongly adsorbed than the sodium ion, so that the former is constrained by adsorbtive forces while the sodium constitutes the counter ion. When in contact with a dilute salt solution, the sodium ions are conceived as diffusing away from the clay par- ticles in the shale before the corresponding chloride ions are desorbed. He suggested that the "leading"' of the counter ions gave rise to the potential effects observed. This view accorded with his belief that the shale poten- tial was relatively transient. It also served to explain his observations that the potential was in large measure dependent on the nature of the shale membrane used. While certain aspects of Williams' theory are in broad accord w'ith prevailing concepts of the action of mem- brane electrodes, the statement that the potential is transient is not in accordance with other experimental observations. Provided no physical leaks or cracks in a shale occur, the observed potential is maintained if the activities of the solutions separated by a shale membrane are maintained constant. If, of course, the potential is measured on closed circuit, the flow of current itself tends to equalize the concentrations of the two solutions and iience diminish the observed potential. Nor did Williams appear to take cognizance of the quantitative agreement between the potentials observed and those demanded by the Nernst ecjuation. This agreement, the potentials observed using sandy shale specimens, and the behavior of shale membranes when in contact with solu- tions containing cations other than sodium, all suggest that a shale barrier behaves as a typical charged-mem- brane electrode. The Electrochemical Properties of Charged Mem- branes. In any discussion of the properties of charged membranes or, as they are sometimes termed, membranes of high ionic selectivity, it would be invidious to neglect I'art VII Clay Technology ix the Petroleim Industry 289 to mention the pioneer work of Leoiior ^liehaelis. Por- tnnately, a recent review hy Sollner (lOoO). who has himself made a great eontributiou to tiie subject, gives adequate references to work carried out prior to the last two years. Although imperfect in several res))ei'ts. the best theory now extant covering the pro|ierties of cliarged mem- branes is the so-called ileyer-Sievers-Teorell theory (M.S.T.) suggested indciieiulenlh' b\' Mcver and Sievers (1936) and Teorell (1935). The membranes to be considered may be conceived as being porous, the size of the pores being of the order of a few Aniistriim units. In their internal structure the l)ores may be visualized as being geometrically sinular, for exanii)le, except in their size, to the pores iu any unconsolidated or consolidated porous medium. However, along the pore walls are irregularly arranged fixed chai-ges. These charges may be either positive or nega- tive and are electrically neutralized by appropriate anion or cation counter ions. Thus a charged membrane may be expected to exhibit cation exchange, although the magnitude of this exchange capacity need not necessarily be large. Whether the charge ou the membrane is posi- tive or negative docs not affect its fundamental behavior. For convenience of presentation, since shale membranes appear to be negatively charged, attention will here be given to membranes of this type. e e e e ® © Figure Cross-section through pores of a negatively charged membrane of high cationie selectivitv. In figure 7 is shown schematically a cross-section of two pores in a membrane of high ionic selectivity im- mersed in an aqueous solution containing positive and negative ions. The fixed negative charges shown may be a residual electrostatic charge (aualagous to the charges on clays) or may be a negative group such as a sulphonic acid group. The latter is characteristic of many synthetic cation-exchange materials. Each negative group is bal- anced by an adjacent positive ion. although these ions may be conceived as possessing reasonable mobility. The charge balancing is dynamic and statistical rather than static and exact. Around each fixed negative charge a zone may be conceived into which, due to electrostatic repulsion effects, an anion cannot effect entry. If this 10—91001 zone of I'epidsion effectively fills a pore (i.e., fills it to an extent which does not permit a particular size of anion to fiiul unrestricted passage between the periphery of the repulsion zone and the pore w-all), that pore will be effectively blocked to anions. Clearly then, three factors affect this situation; the magnitude of the negative charge, the effective size of the anion and the numner in which the negative charge is disposed iu relation to the internal structure of the pore. These are the so-called sterie-geometrical effects discussed by Sollner (1945). If complete pore blocking is achieved, the membrane be- comes impervious to anions and it will conduct only by the passage of cations. This means that the cation-trans- ference number is unity, the aniou-trausference niunber zero. Such a membrane will give potentials which obey the Xernst equation exactly when separating two solu- tions having different activities hut with a cation in common. Such a potential will also be stable since no eoutiuuous pa.ssage of cations through the membrane can take place. This follows becau.se the separation of charge which would be involved would be too jirodigious to be permissible.* A similarly perfect membrane sejiartiug two solutions iu which the cations are not the same would also give rise to a potential, the so-called bi-iouic potential (B.I. P.). Although repi'oducible B.I.P.'s are easily de- termined, this t.ype of potential is inherently unstable, since the two cations will tend to diffuse through the membrane until their mixed concentrations are identical on both sides. There is no electrostatic barrier to this mutual diffusion. The reproducible B.l.P. which is initially measured (when no diffusion has occurred) ap- pears to be a function of the activities of the cations separated by the membrane and their transference num- bers and activity coefficients within the membrane. Membranes have been made in the laboi-atory by the pi'ocess of compacting a powdered artificial or natural cation-exchange matei'ial (clay) under high pressure and bonding the compacted particles hy filling the inter- stices between them with an insulating plastic. The best of these membranes appear to be almost jxu-fcct. Such membranes have a cation-transference number which is very close to unity, i.e.. they have only a very small anion "leak." In general, however, most membi'anes are not perfect and in such cases the passage of anions into the membrane pores becomes possible. The distribution of anions and cations within and without the membrane is then controlled by a Donnan di.stributiou. In such a distribution the product of the cation and anion activi- ties inside the membrane is equal to the same product iu the solution iu which the membrane is immersed. Owing to the presence of fixed negative charges within the pores, it follows that the cation activity within the membrane is higher and the anion activity lower than in the external solution. The ionic strength is now also a factor of considerable importance. The higher the ionic • It is also ot interest to note that if such a membrane has a finite hydraulic permeability, water forced through the membrane will not contain salt. This has been experimentally verified. Thus a perfecUy ion-selective membrane would be a perfect sieve to re- move salt from water or to concentrate salt on one side. Tt may be noted that if natural shale membranes are substantially im- per\'ious to anions (as their electrochemical performance in- dicates) they would affect the saturation of salts in water expressed from them during compaction. This would serve to ex- plain thermodynamically a property of shales that was postu- lated as a necessity by de Sitter (1947) in his theory of the diagenesis of oil-field brines. 290 Clays and Clay Technology [Bull. 169 strength of the external solution, the greater the anion activitj- (and concentration) within tlie membrane. If a membrane of imperfect ionic selectivity is used to separate, for example, two sodium chloride solutions, then each face of the membrane adjusts itself so as to be in Donnan equilibrium with the solution in contact -with it. Within the membranes is formed a liquid junction potential between the different concentrations of anions and cations characteristic of these two Donnan distri- butions. If the ionic activity of the fixed charges within the membrane, expressed as gram moles per 1,000 grams of water within the pores of the membrane, is A. it is possible to compute the potential across the membrane as a function of A, the concentrations of the external solutions and the mobilities of the anions and cations within the membrane. This calculation forms the basis of the M.S.T. theory of membrane behavior. The total potential is composed of three separate components; a Donnan potential at each membrane face and a liquid junction potential within the membrane. For NaCl solutions this may be represented as -Membrane- Solution 1 Xa* = oi CI- = a, -A + V A' + 4n,' -.1 + V A= + 4a=- cr = 1 CI- = \a* = CI- + A I Xu* = CI- + .1 .1 + V .l-' + -Inr .1 + V .1" 4- ■»"=' I Sohitiou 2 Xa* = «j CI- = n» Donnan I'otential #1 Liquid .Tnnction I'utential Ponnan I'utential Donnan Potential #1 ^ In Liquid Junction Potenti: ''' / A + VA^ + 4ar \ RT ■ V In PV A- + 4(ii= + VA'~[ F 1_V A.- + 402= + UAJ RT Donnan Potential #2 = In ■ Here V = ■ TT <- Xn F, ^ /a + VA= — 4or\ 01 where Z/ns and Cci are the sodium and chloride ion mobilities within the membrane. The total potential is thus A +\/ A' + 4o:= RT a, • E = In — F 02 A + V AM^"4o? RT Vln FV A" + 4ar + VaI Lv A' + 4or + LA J _\/ A^ + 4o2= + UA_ (5) If A » fli or (1-2 equation (5) reduces to the Nernst equation. RT F In ai / Oo. On the other hand if A « a^ or ff2 equation (5) reduces to the ordinary liquid-junc- tion potential E ■= RT/F Vln Oi/oo. These potentials thus represent the upper and lower limits possible. Equation (5) shows that when ai or 0,2 are small by comparison with A the potential developed across a charged membrane may be expected to approximate closeij^ to the Nernst potential. The larger the absolute 4731 THEORETICAL POTENTIAL -0.5 0.0 0.5 1.0 -LOG NaCI NORMALITY Figure S. Potentials across sandy shales separating O.OIN NaCl solutions from NaCl solutions of increasing concentrations. value of A, the higher will be the permissible activities (concentrations) which may be used before the potential developed seriously diverges from the Nernst potential. Thus it may be seen that, qualitatively at least, the data tabulated in table 1 are indicative of two charged mem- branes of Avhich one, the Woodford shale, has a higher effective charge, A, than the Conemaugh shale. WTiile both shales tend to give potentials obeying the Nernst equation when the activities Oj and Oo are small, the Conemaugh shale diverges more from the Nernst equa- tion than does the Woodford when these activities be- come larger. A more convenient way of showing this effect is to maintain one activity, a-i, at a constant and small figure and progressive^ to increase the activity Oi. Then a plot of potential on a linear scale against activity, ai, on a logarithmic scale will give a straight line having a poten- Part VII] Clay Tkciixologv in the PETROLEu>r Inihstky 291 tial chan-re (-(lual to (RT/F) or 59.17 mv at 25° C for each tt'iitold increase in Oi if the Nernst equation is obeyed. Fifriire 8 shows some t.vpieal results obtained on a series of soniowliat sandy sliales taken fi-07n different depths ill tlie same well. All but one of these shales L;i\e [)oteiitials whieh follow the Xernst equation when tlie activities are small. The potentials diverj^e progres- sively from the Nern.st equation as the activity of the one solution is progressively increased. The potentials at high activities are alwavs lower than the Nernst jiotenfial as would be expected if an increasing propor- tion of current were being carried by anions. (,'learly the shales which gave rise to the data in figure 8 are verv far from being ideal membrane elec- trodes. Although superior to these in their behavior, the Woodford and the Conemaugh shales used to obtain the data of table 1 are likewise not ideal. The writer and his associates have now tested the electrochemical performance of large numbers of shale samples obtained from wells in all parts of the world, and it can be said that onl.v a few of these si)ecimens were superior to the Woodford shale specimen used to obtain the data of table 1. The great ma.iority of shale specimens gave ])otentials, when separating solutions of high ionic strength, whieh were considerably below- tlie theoretical Nei'ust potential. The data obtained indicate that when measured in the laborator.v most shale specimens do not act as ideal membranes through whieh electric current is eifectively carried only by cations. The data and the jn-evious dis- cussion also serve to explain the observation of Williams noted above. Unless a systematic investigation is made, the potentials given by diiferent shale specimens when separating identical pairs of solntions ajipear quite un- predictable and entirel.v a function of the nature of the shales emplo.ved. And indeed these potentials are a func- tion of the shales, or at least of their structure, since the shale comjiosition probably controls the extent of anion leak. Alternatively, the physical and chemical nature of the shales, following the lines of the M.S.T. theory, ma.v he said to control the magnitude of the charge, A, and tlie cation anion intramembrane mobilit.v ratio. It follows then, if all shale specimens examined in the laboratory are found to deviate to greater or lesser ex- tents from the ideal behavior quantitatively expressed by the Nernst equation, that ecpiation (3), which is based on ideal behavior, is not generally ajijilicable. Specificall.v, apart from any other disabilities from which it might suifer, such as the assumption that the critical ions affecting the shale are sodium ions only, equation (3) would be expected to hold fairly generally only if interstitial waters were of low^ ionic strength. From a practical standpoint equation (3) would then be worthless. Clearly the question of the ideality of shales as mem- brane electrodes when in situ in the earth represents, from a practical standpoint, the crux of the entire problem. The Ideality of Shale Beds in Situ as Menibrane Elec- trodes. To settle the question whether shale beds in situ can be considered as substantiallv perfect negatively charged membi'anes is not eas.v. Several approaches are possible. Certain of these are direct, others involve purely deductive reasoning. None are putirel.v satis- factor.v. The direct approach is clearl.v to check eqnation (3) hv actual measui-ement. That is. to com|)are salinities computed using equation (•'!) with those actually meas- ured. To achieve this satisfactorily and with the degree of accuracy required for a truly valid test is not easy. It is imperative that the total S.P. be accurately measured opposite a very thick clean formation if a standard S.P. log is used, or preferably by using a static S.P. log. The total S.P. must then be corrected for an electrokinetic potential conqjonent if one exists. The activity of the mild must be measured. This ma.v be done using a .s.vn- thetic membrane electrode of verv high ionic selectivit.v (W.vllie and Patnode, 1950). While there are indica- tions that a rather good activit.y may be derived from the resistivit.y of the mud filtrate bv considering the filtrate as a pure solution of sodium chloride, sufficient experience has not .vet been accumulated to suggest that this is invariably true. Some of the data obtained to date are given in table 2. Theoreticall.v, however, unless the filtrate is composed onl.y of soluble sodium .salts, the properties of the membrane must influence somewhat the activity measured. That is, the activit.y with respect to the measuring electrode is theoretically not exactl.v its activit.v with resj^ect to shales unless the filtrate solution is monocationic. The reason for this will be showni below. However, unless there is a considerable percentage of ions other than sodium dissolved in the mud the error is likelv to be negligible by comparison with other errors. Table 2. Ttejutionship hctirern mud aciivity direcihj mensured uainfi (in Amhfrlitt' 1 1\-IIKI/ poli/sti/rctic eJeftrodc (mud iifftiinut standard A'ff'"'/ solution) and the activity computed from the mud filtrate resistirity assuming the filtrate to be a sodium chloride solution. Nature of mud Mud resis- tivity and temperature ohm-meters Mud filtrate resistivity and temperature ohm-meters Measured activnty molality .\ctivity from fil- trate re- sistivity molality Caustic-quebracho Bentonite Bentonite Bentonite 1.7at.50°F 1.8 at 70"^ 0.9at70''F 1.5 at 60°F 7.9at76''F 2.0at76°F 0.93 at .50^ 0.97at70°F 0.56at70'F 0.98 at 60°F 5.9 at 76°F 2.0at76°F 0.1045 0.088 0.145 0.099 0.0137 0.053 0.105 0.081 0.135 0.092 Natural Natural 0.0105 0.038 Trials fulfilling these rigid conditions have been few. The data from one. held in a well in Oklahoma, are given in table 3. In table 3 the calculations of the electrochem- ical S.P. were based on sodium chloride solutions of con- centrations equal to the total dissolved solid contents of the waters produced. The streaming potentials were derived from laboratory measurements made on mud samples taken at the time of logging. These samples were heated during the measurements to the temperatures whieh obtained in the hole opposite the formations. Mud activities were measured with Amberlite IR-100/Pol.v- styrene electrodes. Pressure differentials across the filter cakes were determined by direct measurement of the formation pressures and the hydrostatic pressure of the mud column. 292 Clays and ('lay Teciixolooy Table 3, Comparison between theoretical electrochemical and elect rokinetic t<.P. and measured .S'./*. IliuU. 169 Section Interstitial water activity ac Mud activity am ac/am Theoretical electrochemical emf Theoretical electrokinetic emf Total theoretical emf Maximum measured S. P. Trial number Feet mv mv mv mv 1254-1300 1254-1300 1254-1300 1254-1300.. 0.36 0.36 0.36 0.36 0.36 1.65 1.65 1.65 1.63 0.0688 0.0688 0.0879 0.0983 0.1185 0.0688 0.0879 0.0983 0.1185 5.24 5.24 4.10 3.66 3.04 24.0 18-8 16.8 13.9 52.0 52.0 44.5 40.5 35.0 100.5 93.0 89.0 83.5 6.2 7.0 6.0 5.5 8.5 8.0 7.0 6.5 10.0 58.2 59.0 50.5 46.0 43.3 108.5 100.0 95.5 93.5 55.5 55.0 46.5 45.3 43.0 111.0 103.0 86.5 108.5 1 2 3 4 1254-1300. -- 2358-2375 2358-2375 2358-2375 2358-2375 5 2 ■ 3 4 5 Water analyses. Ions (mg/1) 1254-1300 ft. 2358-2375 ft. SO4 --- CI 1,140 18.364 381 11,691 446 210 nil 96.780 HCOj - Na Ca.. Mg 24 48.982 8.310 2.254 34,050 156,350 The five trials recorded were made at intervals during the course of about one month during- which the hole was being deepened. The formations were cored and were found to be relatively free of shale or clay contamination. Considering the assumptions made, table 3 reflects rea- sonably good agreement between potentials based on equation (3) and those actually measured. It may be noted that the ionic strength of the 2358-2375 feet for- mation water is large. An analysis made by the writer (1949a) showed that in a number of different regions in Kentucky, Oklahoma and Illinois the agreement found between the S.P. mea- sured and that based on equation (3) was good. In these regions the salinities of the interstitial waters ranged as high as 172,350 mg/1. Nevertheless, in this examination no correction was made for possible streaming poten- tials, and mud activities were not actually measured. It has been noted previously that the total S.P. is widely and rather successfully used in the routine em- ployment of equation (3). While perhaps suggestive of the basic accuracy of equation (3), this fact cannot be considered a direct substantiation of its validity since electrokinetic potential corrections are not normally made nor accurate activities used. The possibility that the agreement is dependent upon a fortuitous compensa- tion of error has already been noted (Wyllie, 1951). Indirect approaches to the basic problem, whether or not shales * can be considered perfect membranes, may now be considered. • The writer is aware that a shale is difficult to define rigidly. At what point does a shale becon-te a sandy shale and the latter a shaiy sand? In this paper shale, unqualified, is deemed to mean an argiliaceous material which, if free of geometrical effects, would give on a log a S.P. shale baseline. The classification is admittedly arbitrary and could even be unrealistic geologically, since a section of sandstone in which every pore was entirely filled with shale could, under certain circumstances, give a per- fect shale baseline. If a shale is not a perfect membrane then any shale M'hich is sandwiched between two horizontal permeable beds Avhich contain different salinities of interstitial water would not only develop a potential between the two permeable beds but would permit current to flow from one bed to the other in a perpendicular direction. The current would flow in such a direction as eventually to equalize the salinities of the waters in the two per- meable beds. The rapidity with which this equalization would take place would depend upon the thickness of the intervening shale bed and the difference in the salinities of the waters in the two permeable beds. The greater the salinity difference and the thinner the shale bed, the higher the potential gradient across the shale and the more rapid the rate of ionic migration. However, for thick beds and for relatively small differences in interstitial water salinities the rate of equalization would not be high and even geologic time would probably be insufiicient to effect substantiall.y complete equalization. Thus the fact that shale beds are known to separate per- meable beds which have interstitial waters of different salinities is not valid evidence of their complete anion impermeability. Indeed the fact that in many regions the salinity of interstitial waters changes very slowl.v with depth might tend to indicate rather that the contrary were true and that shales were relatively inefficient as anion impermeable membranes. Perhaps the most suggestive fact which bears on the question of the efficiency of shale as membranes is the observed constancy of shale baselines on logs made in every part of the world. On the great majority of logs, it is possible to draw a consistent shale baseline over many thousands of feet of log. This baseline on the usual log is drawn through the most positive excursions of the S.P. curve, and it is defined in most reasonably shale-rich sections by numerous shale beds. When the fact is considered that any change in the contact poten- tial of the borehole electrode with the mud fluid affects the baseline, it may be said that each of these shale beds possess an electrochemical property which, within a few millivolts, is identical. Here, then, is a measure of con- sistency which in any one well is possesed by shales laid down at what must often have been widely differing geologic times. Nor are these shales physically identical. They generally differ in the amounts of clays, organic materials and silts which they contain and the extent to which they have been lithified. Part VTIl Cl.AY TEfllXOI.OfiY IN THE PETROI.l:rM Tntiistky 293 For simplicity a clean, slialc-free permeable bed separatiii«r two shale beds may be considered. It will be assimifd. as seems generally to be the case, that within the iieriiieahle bed the salinity of the interstitial water is uniform. Then the S.P. deflection at the top of the per- meable bed results from a chemical interaction between the mud and interstitial water with the shale lying above the permeable bed. This interaction leads to the production of an emf which in turn causes the S.P. curve to deflect a definite amount. If the permeable bed is sufficiently thick, the ohmic potential changes in the borehole which affect the boreiiole electrode j)otential do not persist over the entire bed thickness. The S.P. de- flection climbs to a plateau and the boreiiole potential does not alter until the electrode is subjected to the effects of current circulation which result from the elee- trocbenncal interaction of the lower shale bed with the interstitial water and mud. The potential change is now in a sense opposite to that exiierieiieed by the electrode at the top of the bed and the S.P. deflects from its l)lateaii opposite the permeable bed. finally attaining a fresh value opposite the lower .shale bed. If the poten- tial opposite the lower shale bed is identical with that (ipjiosite the upper, that is, if there is a constant shale baseline, it follows that the potential given by the elee- trochemical interaction of the lower shale bed is identical with that given by the upper. By the same token, if the shale baseline is maintained over thousands of feet, it follows that each shale which separates a permeable bed behaves in an identical fashion* eleetroehemically. This point may be further clarified if it is noted, as pointed out by Doll (1949), that the S.P. curve is sym- metrical. A constant "sand line," that is a line drawn through the maximum negative excursions of a conven- tional log, is not obtained unless all interstitial waters in permeable beds are identical. Such sand lines are ob- tained (fig. 2) but generally over small distances com- pared with shale baselines. Thus, as noted above, the existence of a constant shale baseline extending over an entire S.P. log implies remarkable electrochemical ho- mogeneity amongst all the shales in the borehole. Now the electrochemical potential developed by each shale above and below a jienneable bed is controlled by the same two fluids, the mud and the same interstitial water. If the two shales are imperfect, this implies that each shale possesses an identical anionic leak. On the basis of the M.S.T. theory each possesses the same charge, A. and the same ratio of cation and anion mobilities. Figure 8 shows that lealn- mendn-anes do not show the same electrochemical potentials when separating identi- cal solutions except in the range of C(uiceutration in which they follow the Xernst (Hpiation. Figiire 8 may be considered typical of results obtained with imperfect membranes. Even when membranes are made from iden- tical percentages of identical cation exchange material and plastic and are monlded in an identical manner, it is extremely rare for two membranees to give identical potentials in the concentration range in which the Nernst potential is not followed. It is, however, pertinent to note that in the range where the Nernst equation is obeyed CONNATE WATER • The fact that certain thin shale beds do not return to tlie base- line obtained in conventional S.P. logging is purely a geometri- cal effect and does not affect the argument developed. Doll (1949) has adequately treated this point. ^Ezzzzzzzzza) HYDROUS MICA <@ KAOLIN o SILICA Figure 9. p;ieftr()clicniical structure of .shale. the physical nature of a non-homogeneous electrode, or the amount or type of cation exchange material it con- tains are of no importance. This is to be expected the- oreticallv and has been demonstrated experimentally (Wyllie'and Patrode, 1950). The weight of existing evidence seems to indicate that shales in situ in the earth do act as perfect membrane electrodes. In this event equations of the type of eciua- tion (3) may be expected to be applicable to the electro- chemical potentials developed when all conditions for their application are fulfilled. Nevertheless, it must be pointed out that if further research on the nature of shales and on the structure of charged membranes should disclose that the value of A and the cation /anion mobil- ity ratio in all shales could reasonably be identical, this conclusion could no longer be sustained. In this jiaper the fact that shales in situ do obey the Nernst ('(piation will be assumed and the consequences of that assumption will be examined. A Tentative Electrochemical SInicturc for Shales. In figure 9 is given a tentative two-dimensional physical picture of the essential structure of shales from the elec- trocheniical standpoint. For this purpose the principaj constituents of shale are assumed to be silica, hydrous mica, and kaolin. These components, for the marine shales which are of patricular importance in petroleum exploration, have been found by the writer's colleagues to be typical. These components appear also to be in essential agreement with the results reported by others (Millot. 19.72). In most shales, the silica particles vary in size, many being similar in size to the clay particles. The plate-like structure of the clays makes them tend to orientate themselves parallel to the bedding planes of the shale. In figure 9 the particles of hydrous mica are drawn surrounded bj' a water sheath in which are located the exchangeable ions. In this same sheath there will, therefore, exist a considerable negative electrostatic charge, a force repelling anions. Between the particles of clay and silica there is water containing (lissolved salts. In the figure the line of demarcation between the water sheaths surrounding the claj' particles and the inter-particle water has been drawn for convenience of presentation. Actually this is artificial and a swarm of ions graduTilh- changing in ccmiposition is to be expected. Under the enormous pressure of compaction resulting from the weight of overlying strata, the particles of all kinds are forced tightly together. If the pressure is such that all continuous paths for ions through the shale involve at some point passage through the water sheath surrounding a clay particle, it will mean that within the shale will exist a surface across which electric current 294 Clays and Clay Technology [Bull. 169 SPONTANEOUS-POTENTIAL millivolts m -D X RESISTIVITY -ohms, m'/m RESISTIVITY -ohms, m'/m 20 ^ - m + l6",S-4>" NORMALS ? .0 2o' 8" LATERAL ? 10 Q flMPI 1 NTiRM ? 1 Q H IRH 1 <;r.AI F 100 n HIAH ^9 ^r.fliP IQQ- ^1 r; L3I 1 1 1 1 1 1 T^l 1 1 1 1 1 1 ^. S -' < 5 j o K ^ , \ ^ -^ :> ~? <( c < :3> V. ■ — ■^ i> < ' r^ >. — <:: s 5 S^ "~< =*^ "^ ■;=. i < - -^ 4 -^ c 1 r 0> i 7 ^~ — -, o ^ \ - \.. -- -- — - f^ ^ ■"Y- "■ \ - ) c <- ^S ) > 1 1 .c ^ c> F i ." - / > S A ^ 5 1 i v N. L ^ : ^ t > . o / >___ << — '>-j -- — / 1 ]'■ ■''-= 1 ./ <; -— r J / <-. > > i « t . ■ ^ — < I K^ ^ { * ^ 5 } 5 y ^ ^'■=' : ^ • -= ^ ^ \ /* I •> o / ^ i ^ ) i- "^ f ^ f F V \ 1 ,/' i 4 > - \ / C" \ ^ » 37 go <* I ■ >■-» ^^- — 3 ^ -^ o , r 1 1 ^ *" ' — > If 1 > _J - 1 s ^ y , H '^ — (^ Figure 10. Shale baseline shift of 60 mv apparently resulfing from a gradation of interstitial water salinity. I'arl \J1| Clay Technology ix the Petroleuji Industry 295 eau only be carried by cations; i.e., tlie anion transfer- ence will be zero. From a thermodynamic standpoint, if there is snch a surface in the shale, the shale will func- tion as a perfect negatively charged membrane. It is, perhaps, not innnediately obvious that the con- i-cntration of the connate water in the shale, which hirgely controls its electrical resistivity, does not affect the potential developed. Thus, marine shales frequently show a lower resistivity tlian non-marine shales (Clau- det. 1950) although electrochemically both shale types may behave similarly. Tliis fact may be demonstrated if a system is considered comiiosed of a number of per- I'cct nii'ndjranes separated by solutions of different con- centration. For convenience all the solutions will be con- sidered to consist of NaCl, but the result can be readily generalized. Consider: NaCl NaCl XaCl XaCl NaCl 3Iciiibranes- Tlicn the potential is RT a, RT an RT a, RT n, E = -In \ In 1 ]-n 1 In — F a-i F «:, F a4 F ar. RT ai - — 1« — F a. (6) That is, only the exterior solutions affect the over-all po- tential developed. The electrical resistivity of the system is a function of the resistivities of the membranes and tlie solutions. If the conducting paths in the membranes are small by comparison with the patlis through the solutions, the latter will be dominant. This appears to be the physical situation in shales. If overburden pressure is removed from a shale, a certain degree of elastic expansion is to be anticipated. This alone will tend to give rise to leak paths available to anions, for the general effect will be to decrease the effective charge, A, of the membrane. For while the charges on the clays are not altered, the effective volume in which they are disposed will be increased, leading to a diminution in the molal activity, A. In laboratory jiractice tliis effect is fi'cquently reinforced by unavoid- able di'ying and cracking of the shale whicli gives rise 1o even larger leak paths. Hence it is not surprising that the electrochemical perfornuinces of shale samples when tested in the laboratory arc often inferior to their apparent performances in situ. Experimental verification has been found of the postu- lated effect of pressure. Thus membranes formed by moulding jiowdered shale in Incite under a pressure of 5,000 psi liave been found to possess an electrochemical l)erformauce superior to that of the shale samples before powdering (Wyllie and Patnode, 1950). An analogous effect appears to be obtained when using synthetic sul- phonated phenol formaldehvde cation exchange resins (Wyllie, 1952). ^hale Bajieline Shifts. An abrupt shift of the shale liasi'liue is sometimes observed. Although the phenome- non is comparatively rare it unquestionably exists. Wlien considering .such shifts, it is wise to concentrate on those which are characteristic of a partii'ular environment and which are reproduced on all logs run in a particular field. Isolated shale baseline shifts on logs are more often the consequence of an instrumental fault in the logging equipment than the manifestation of a genuine plu-- nomenon. In figure 10 is rej)roduced a log which shows a well- defined shale baseline shift. This shift is characteristic of all logs in the area. From figure 10 it may be seen to amount to -(-60 mv. On the log there is a good shale baseline down to 1760 feet and another below 1875 feet. The maximum S.P. deflection of the permeable beds above 17()0 feet is rather constant and amounts to — 37 mv. The max- imum deflection of the beds at 1835-1860 feet with refer- ence to the lower shale baseline is — 97 mv. The mud activity is 0.033 g moles/1000 g water. Application of equation (3) gives an interstitial water activity of about 0.064 corresponding to — 37 mv and 0.74 corre- sponding to — 97 mv. These activities in turn correspond to NaCl solutions with salinities of about 5,000 mg/1 and 66,000 mg/1 respectively. At the formation tempera- ture of 86° F, solutions with these salinities would have resistivities of about 1 .0 ohm-meter and 0.09 ohm-meters. It has been established that in the section 1680-1780 feet the formation factor (1942) of the permeable sec- tions is rather constant. Thus the slope of the resistivity curves between these depths appears to reflect a rather steady change in interstitial water resistivity. The re- sistivity charge is from about 1 ohm-meter to 10 ohm- meters, a ratio of about 10 to 1. This ratio agrees closely with the ratio of the calculated water resistivities, 1 ohm- meter and 0.09 ohm-meters. Figure 10 appears to be an excellent example of a shale baseline shift which results because the salinity of the water in a series of rather shaly permeable beds changes, so that the water in eon- tact with the shale which constitutes a part of the upper shale baseline is very different from the water in contact with shale which constitutes a part of the lower shale baseline. However, both the upper and lower shales ap- pareutlj' have electrochemically identical properties. The change in salinity which is shown in figure 10 may readily be discerned on the log. Hence the conclu- sions drawn from the S.P. curve can be easily checked against the data of the resistivity curves. Were this change to have taken place within a very thin permeable bed, as may be possible under suitable conditions of flushing from an outcrop, conclusions based on the form of the S.P. curve would be almost impossible to check on the resistivity curves and the shift would be corre- spondingly more inexplicable. It is difficult to say how many baseline shifts are the result of salinity changes in permeable beds and how many result from genuine differences in shale nature. It does not seem uni-easonable to expect a shale baseline shift at a facies change, par- ticularly from marine to nonmarine, and it is believed that these Have been observed. Such effects appear to be bound up with the theory of the bi-ionic potential. It has been assumed in the foregoing discussion and calculations that interstitial waters and miid fluids can be treated as sodium chloride solutions. This is an over- simplification and must inti'oduce errors. The errors in- troduced may be divided into two parts. Those arising from the effect of cations other than sodium on the 296 Clays and Clay Technology [Bull. 169 potential across the shale and those arising from the effeet of both cations and anions, other than sodium and chloride, on the liquid-junction potential which is formed between the intei'stitial water and the mud. When a formation is shale-free the latter may easily be a.ssessed if liquid-junction potentials between solu- tions having compositions akin to those actualh' en- countered in tlie field are computed. These potentials may be compared witli the liquid-junction potentials computed for sodium chloride solutions of similar ionic strength. The latter is the assumption generally made as a practical convenience to permit the solution of equa- tion (3). The error involved is generally minor. If the formation is not substantially shale-free, the error becomes more difficult to assess. Theoretically, however, a shaly permeable bed is merely a shale witli large anion leak paths. This concept is given formal treatment below. The problem of formulating equations to define the B.I.P. has been considered by both :Marshall (1948) and Sollner (19-19). The work of SoUner is perhaps the more extensive. Sollner considers that in a membrane in which cation transference is unity the sign and magni- tude of the B.I.P. are controlled by the ratio of the adsorbabilities of the two ions within the membrane and by the ratio of their intramembrane diffusion velocities. Steric hindrance effects witliin the membrane may also play a part. For examjile, if certain pores in a mem- brane are so small that tliey will admit only one of the two ions being considered, the adsorbability of the smaller ions in such pores will be much greater than their adsorbability elsewhere in the membrane. Basi- cally, however, Sollner appears to consider that the distribution of the two ions within the membrane is everywhere the same. IMarshall has formally treated bi- ionic potentials as a liquid junction potential to which the Henderson equation may be applied. Anion mobil- ities are considered to be zei-o. This treatment is cpiite permissible thermodynamically and its validity is not dependent wpon any details of the electrode mechanisms except for the implicit assumption that the distribution of the ions to which the membrane is selective is identi- cal on the membrane faces in contact with the solutions and in the solutions themselves. This assumption does not seem to involve any numerical errors when mono- valent ion mixtures are considered. However, the concept does not seem to be theoretically sound when mixtures of monovalent and divalent ions are considered, for in such cases it is well known that the divalent ion is preferentially adsorbed by an ion-exchange material when the ionic strength of the mixture of monovalent and divalent ions is small. In general the distribution of ions on the ion-exchange material is not identical with the distribution of the same ions in a solution in contact with the ion-exchange material but is a function of the ratio of concentrations of the two ions in the solution and its ionic sti-ength. The potential across a perfectly cation selective mem- brane which separates tw'o monovalent ion solutions, for example NaCl and KCl, may then be considered as the sum of two Donnau potentials at the faces of the mem- brane in contact with the solutions and a licjuid junction potential formed inside tlie membrane. This liquid junc- tion j)otential is a function of the relative mobilities of the ions Xa* and K* within the membrane. This con- cept differs from that of Sollner, since the distribution of ions in a membrane when a B.I.P. is measured is not considered to be uniform. In fact it is considered that the membrane grades from a condition in which all its exchange positions are filled by K* ions at its interface with the KCl solution to a condition where all its ex- change positions are occupied by Na* ions. The latter condition exists at the interface of the membrane with the NaCl solution. Moi-e generally for mixed solutions of two ions of whatever valency, the distribution of ions on each membrane face is calculable if the distribu- tion coefficients of the cation exchange material of the membrane is known for the ions considered. The liquid junction potential is formed within the membrane be- tween the ions on each face of the membrane. These considerations give rise to equations analagous to those of the JM.S.T. theory if the membrane is not considered to be perfectly selective to one type of ion, e.g., cations. However, for ideally cation selective mem- branes the equations simplify, since the anions then play no part except for their effect on the activities of the cations in the solutions. The following equations may be derived (see WvUie, 193-1, and W\"llie and Kanaan, 1954). I. Manovalint — Monovalent e.g. NaCl versus KCl KT ttK Vk E = 1)1 Here (1^, (ly,, are the ionic activities of the sodium and potassium ions separated by the membrane, TjK.'Vxa is their ionic mobility ratio within the membrane (ecjual to the ratio of their transference numbers). II. Monovalenf — Monovalent -\- Monovalent e.g. NaCl versus NaCl + KCl BT a'j,, + aK{VK/V^ra) E = In F a".va ■ Here aV,, is the activity of the sodium ion in the mixed sodium and potassium solution, a"}ia is the sodium ion activity in the pure sodium chloride solution. I I I. Monorahnt — Divalent e.g. XaCl versus CaClo RT a^^a U^a/Uca £■ = In ■ h 2F ara Aya RT 2F A^aUxa RT fl-.v„ : In 2F acn RT 2F AyaU-Xa 2Aca t'' Ca Ih-a/Vca A \-a Uxa In 2Ac Tic if Aca — iAxa Axa Uxa Vca VxJVca—'^ + Pa \n- rt \\l Clay Technology in the Petroleum Industry 207 Here the symbols are as before witli the addition that Aca is the activity of calcium ions within the membrane when the membrane is wholly in the calcium form, and Axa the eorrespondiny: sodium activity when the mem- brane is wholly in the sodium form. The assiuniition that Ara = iAxa implies that the activity coefficients of the uak'ium ions in the calcium form of the membrane is identical with the activity coefficient of sodium ions in the sodium form of the membrane. While quantitative information concerning such activity coefficients is lack- inor, this assumption may be in considerable error. Never- theless the activity ratio, Aya/Aca, is constant and the approximate relationship given above based on a ratio of unity may be useful for certain qualitative applications of the relationship. The ratio Aya^'2'Aca is probably greater than unity. TV. Monovalent -\- Divalent — Monovalent e.g. NaCl + CaClo versus NaCl " RT RT a'ya -4.VO E = In • F a"xa ayaR F (Uya] f {aNaR)--a.Ca ayaR I {Vc + 1 I a^Na^ -Ay^ Vs. I Ca , OyaR fFv„l {2{ayaRy--aca It/caJ I a^sa' -Aya Ui. Uc In- Uy„ 2{ayaR)- ■ aca OyaR H iyo- Vy„ Uca Here the symbols are as before. In addition Oyji is the activity of the sodium ions on the membrane when it is in contact with a mixed solution of NaCl and CaCU which has a sodium ion activity in the mixed solution of a'ya and a calcium ion activity of Oca- This complex eipuition may also be written in terms of the exchange coefficient of the sodium-calcium exchange. A similar but not identical equation pertains to the system monovalent -f divalent versus monovalent. A'. Monovalent + Divalent — Monovalent + Divalent e.g. NaCl + CaClo versus NaCI + CaCl. This may conveniently be written as RT a'ya RT x"yaR RT E In F a"ya Uca yca C^ATo 2yca Vca 7.Va_ '' x'yaR F 'Uy. 2-/Ca'l 2yca -f yA'o J ■ 7-V(i x'ya In Uc, •"^'^0 'Uya 2yca' 2yca + ywff JJca y.Va. 11—91001 Here a'ya and a"ya are the activities of the sodium ions in the mixed sodium and calcium chloride solutions sepa- rated by the membrane ; x'yaR is the fraction of the ex- change sites occupied by sodium ions on the membrane face in contact with the solution containing sodium ions of activity rti, and x"yaR is the corresponding fraction on the other membrane face. The ratio yca'yNa is the ratio of the activity coefficients of the calcium and sodium ions when the membrane is in the calcium or sodium form respeetivelj'. Although still a simplification, the only cations in interstitial waters in permeable formations may be con- sidered to be sodium and calcium. In general this is true, but occasionally the magnesium ion content is not negligi- ble. As a divalent ion, however, magnesium can probably be lumped with calcium without the introduction of any serious error. In most, but not all. interstitial waters, the number of equivalents of sodium ion greatly exceeds the number of equivalents of calcium ion. In the mud fluid the jirincipal soluble cations are also sodium and calcium, with again the sodiiuu ion i)redomi- uating. Nevertheless, in modern lime-base muds the amount of soluble calcium can he appi'cciable. This has been shown by Bergman (1952). A shale in situ may thus be considered to separate two solutions which differ in ionic strength but both of which contain sodium and calcium ions. The equation derived as case V above should then be applicable. Before the equation may be used it is necessary to consider Avhat values x'ygR, x"y„R are likely to have in natural shales. Should these be identical it is clear that the equation of case V reduces to RT a'ya E = In F a"y„ Thus if the fraction of exchange sites occupied by sodium ions on the shales in the borehole is identical witJi the fraction of exchange sites occupied by sodium on the shale face in contact with the interstitial water in the permeable formation, the shale will obey the Nernst equation for the sodium ions in the interstitial water and the mud. In this case the only eifect of the divalent calcium ion will be to increase the ionic strengths of the solutions and otherwise to affect the sodium ion activ- ities. Now the magnitude of x'yaR, the fraction of exchange sites occupied by .sodium on the shale face in equilibrium with the interstitial water, undoubtedly varies both with the ratio of sodium to calcium ions in the interstitial water and with the ab.solute concentrations of these ions in the water. Work by Case (1933) and Taylor (1929), shown in tables 4 and 5, suggest that although this fraction is variable it has an average of something over 0.7 in marine shales. Limited data obtained by the writer's colleagues tend to confirm this figure. By and large the ratio will tend to be highest, for any Na/Ca ratio in an interstitial water, if the interstitial water is highly saline, and lowest in waters which are compara- tively fresh. In the extreme case of .shales in the zone of surface waters, it appears from table V that x'yaR approximates zero. In the borehole the situation is more difficult to assess. Although the usual Na/Ca ratio in the mud will be 298 Clays and Clay Technology Table -',. Ihihi of E. M. Taylor (1029). (Na and Ca as me./lOO g. ) [Bull. 169 Company Location Depth in feet Horizon Na Ca Percent Na Percent Ca Gypsy- - Gypsy Hughes, Okla.. .. 3,300 3,303 2,048 2,584 5,317 5,112 2,985 2.S90 3,612 3,613 3.595 3,597 3,598 2,393 6,386 Top Cromwell Top Prue Chattanooga Simpson Over Woodbine Over Woodbine 12.4 7.2 5.7 7.7 16.8 19.0 13.8 11.2 6.3 6.3 8.7 7.3 5.8 17.6 27.7 29.6 5.6 0.8 4.0 4.2 2.3 6.6 6.5 4.1 4.2 4.4 1.8 3.4 2.2 6.6 9.9 7.1 69 90 59 65 88 74 68 73 60 59 83 68 83 73 74 81 10 Tvilsa, Okla... . Osage. Okla. 35 Grant, Okla Smith. Okla. 12 Amerada , _- -. , Smith, Okla 26 Atlantic - - - 32 W. G. W. G. O.- Kern Calif. Miocene Miocene. 40 W.G.O - Kern, Calif 41 W. G. 0. - 17 W. G. 0. 32 W.G.O Kern, Calif. 17 Gulf Caddo, La - Tokio 27 Std. of Calif. - - 7 26 Std. of Calif.7 19 73 27 much higher than in the interstitial water, the equilib- rium calcium fraction of the clays on the faces of shale formations in the borehole may not greatly diflt'er from the fraction on the faces of the same shales in contact with interstitial water. This would be the case because the low ionic strength of the usual drilling mud would lead to preferential divalent ion absorbtion. Hence in most cases the difference between x'naR and x'\r„R may be extremely small. The ratio of x'^Ji to x"ifaB would then approximate unity. For a baseline shift to occur in crossing a permeable bed in which the interstitial water is uniform, the ratio of X NnR x"NaR for the upper shale bed would have to be different from that for the lower shale bed. If the type of clay in these beds is similar, as seems to be the case, it seems possible that the exchange constants of the two .'•'.va/? beds are similar and therefore the ratio for the two beds are virtually identical. Thus no baseline shift Vila yca would occur unless the ratios and differed I'ca yxa greatly. This would fit observations on logs. If, however, two shales differed in their exchange x'xaR constants thev could give rise to different ratios .r".y„B when separatini:- tlif same two solutions containing r.v„ sodium and calcium ions. I'robablv in such cases yca and would also differ. Such an effect may be pos- sible. Further work on the properties of cla.ys in shales and their variation from shale bed to shale bed in a geologic column is necessary before the probability can be assessed. If, however, as stated by Millot {19.j2) kaoliiiitic minerals are characteristic of fluviolacustrine continental faeies and illite of saline lagoonal and basic continental faeies, it is not unreasonable to expect a baseline shift when a borehole passes from a non-marine to a marine sequence. Within each faeies the baseline should be essentially constant. As previously noted, if RT a\^a x'a-„K = x"k,R. E = In . F «.".■„ In this event, the potential across the shale and therefore the entire S.P. measured will be less than that given by equation (3), which assumes that the total salt content of the interstitial water behaves as NaCl, not merely the actual NaCl content of the interstitial water. If, how- ever, x"}jaR is not equal to .T'jaT? the potential is given by the equation of case V. Assuming x'^aR = 0.7 and that when using high sodium /calcium ion ratio drilling muds the clays of the shale face in the borehole will be enriched with respect to sodium ions, the potential change may be calculated if certain assumptions are made. It will be assumed that x"x„R = 1.0, i.e. all divalent ions on the shale in the borehole are replaced by sodium, the worst possible case. Experiments with shales indicate that may be about Uca 5, although other experiments with synthetic membranes now in progress indicate that pressure may increase this yca ratio considerably. The ratio is difficult to estimate YA-a and for convenience will be considered to be unity. I'ai-t VII I Clay Teciixoi.ocv ix the Phtkolkim Industry TaUe 5. Data of L. C. Case (1933). 299 Well Depth in feet Horizon — remarks Ca Percent Na Percent Ca Roxana No. 1 Gypsy core drill . . . Blackwell No. 1 _ Gypsy Petros No. 4 Gypsy Toge No. 4 - Pine TiKcr No. 1 . Gypsy ToRe No. 3. Sinclair No. 2 3,63.'; 646 2.923 2,510 2.810 3.890 2.837 2.677 Tonkawa series Lower Permian red shale Lower part of shale 20' above Gilcrease sand Above shaly .sand Woodford, just above Hunton limestone In contact with oil sand Between two streaks of Cromwell sand Tulsa outcrop. 6' below ba.se of Checkerboard limestone Tulsa outcrop sample, 25' below Checkerboard limestone-. Tulsa outcrop, underclay of coal 12' below Checkerboard 2-6" below coal Tulsa outcrop, as above hut 0-2" below coal Other Tulsa outcrop samples 4.3 U.5 8.2 11.8 15.5 4.2 19.5 15.6 0.0005 0.0013 0.0064 0.0059 0.0045 0.0079 3.7 4.5 .\verage 5.6 8.1 12.3 10.8 6.8 5.12 54 72 73 76 79 72 74 81 73 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 46 28 27 24 21 28 26 19 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 ( Xii :iik1 Ca in me./lOO ^'. ) Tlicn In = +9.2 mv and the o.\i)rossion, 1>T Uno/Ucu — yca/fNa Vx„ yca yca x'xaR Uca y.v« y.Va In = — G.8 iiiv. ■r"s„l\ U,\„ yc„ yca Uca y.vo yxa The iiici-casi- ill the S.P. (aetually an increase in the negative S.P. recorded) would be 2.4 niv over that obtain- inu- if only the sodiiiiu ion activities in the two solu- tions were afTectins;- the shale potential. For x'xaR =^ 0.5 a similar calculation gives the increase in negative S.P. as 5.6 niv. It is seen that these effects are not large unless .r'y„R is very small by comparison with a-".v„K. In muds which decrease x"xaR below x'^aR (as may occur with some lime-base muds) the negative S.P. will be somewhat RT aVa reduced below the In figure. It ma\' be noted F a'\-a tiiat these potentials are not dependent upon the actual interstitial water salinity except inasmuch as this salinity aflfects the \alue of x'xaR- Thus their effects are greater percentagewise when the ratio O' Xa is small than when a"xa it is large, and may be expected to be at a maximum in the ease of relatively fresh interstitial waters. In general this analysis indicates that the electro- chemical S.P. is likely to be more a function of the sodium ion content of the drilling mud and interstitial water than of the total saline contents. Thus equation (3) may be expected to give values for connate water salinity which are somewhat too small. It will serve best when applied to waters having a high primary salinity.* The Electrochemistry of Shaly Sands. The effects of a laminated sequence of thin shale and clean sand beds on the shape and magnitude of a S.P. curve have been adequately discussed by Doll (1950a). In the treatment used by Doll it is assumed that the thin shale lamina- tions are eleetrochemically perfect, i.e. are perfectly cation selective. Since these shales are subject to the full pressure of their overburden the writer believes this assumption to be justified. An experimental verification of Doll's equations using log data from the Corpus Christi area of Texas has recently been made by Poupon (1951). The agreement calculations based on log data and on Doll's equations was good and lead to the eon- 2t_ RT elusion that the value of F in equation (3) was about 80 mv. This value seems to indicate that the thin shale laminae were cation selective. AMiether Poupon 's isolated data are indicative of the general truth of Doll's assumptions is not certain. The influence of interstitial shale or clay in a per- meable bed is more difficult to assess quantitatively. On the assumption that each individual shale or clay particle acts as an individual shale cell when it separates mud fluid from interstitial water, Doll made approximate calculations to show that the S.P. opposite a shaly sand would be less than the value it would have opposite a clean .sand containing identical interstitial water. Doll ♦It has recently been found (Wyllie. 1952) that with many artifi- cial membranes Ca-+ ion appear.^ to behave as a monovalent ion, possibly a"fe Ca (OH)*. The data of MclJean, Barber and Mar- .shall (1951) given in Table I of their paper for potassium against mixed solutions of potassium and calcium ions are also explicable if the calcium ion behaves as a monovalent ion with a constant mobility with respect to potassium of about 1.01!. The membrane used by these workers was a 600° C Putnam clay membrane. However, shale specimens so far tested give a nor- mal divalent reaction with calcium ions. W^ere calcium to behave as a monovalent ion in the earth, the arguments given above for baseline shifts would remain substantiaUy similar. However the potential across the shale would then tend to exceed that given by equation (3) particularly for interstitial water rich in cal- cium ion. The equations of case II are applicable. 300 Clats and Clay Technology [Bull. 169 demonstrated also that the S.P. would be further diminished if the shaly sand contained hydrocarbons. The correct interpretation of beds containing inter- stitial shale particles is peculiarly difficult since the S.P. is diminished but not necessarily altered in shape. A shape effect is often observed in the S.P. opposite inter- laminated shale and sand sequences unless the laminae are extremely thin. When the S.P. curve indicates lami- nations the log interpreter is at least aware of a probable diminution in the S.P. deflection. The effect of shal.y sands on the S.P. is more insidious and this is the more unfortunate since such sands are frequently of economic importance, t A mathematical approach to the problem of a shaly sand containing interstitial clay material may be based on the M.S.T. theory. In this analysis the structure of a shaly sand will be analagous to that shown on figure 9 for a shale.* However, in a shaly sand not all con- ducting passages may be considered to be blocked by the effects of electrostatic repulsion. Rather the clays may be considered to exist in the pores and on the grain siirfaces in the manner suggested by Grim and Cuthbert (1945). Their effect will be to introduce in the pores a charge as defined in the M.S.T. theory. If this average effective charge is fairly large the shaly sands will be- have as inefficient membrane electrodes. The potential they give when separating two solutions will lie some- where between that for a perfectly anion-impermeable membrane and the ordinary liquid junction potential between the two solutions. The term average effective charge, A, is used because of the steric hindrance effects which must always be con- sidered in membrane phenomena. Thus a clay particle located in a pore constriction in the permeable bed will have a rather large effective value of A, since the effec- tive volume in which the negative charges on the clay are disposed will be the volume of the pore neck rather than the entire pore volume. Conversely the same clay particle located in the pore itself would have a relatively small effective charge A. Thus the amount of clay in a permeable bed in a shaly sand is far less important from an electrochemical standpoint than is its disposition. In all cases, however, if the amount of water in the pores is diminished the effective charge A will be increased, since A is expressed as a molality and is dependent to some extent on the water content of the permeable bed. Thus the addition of oil to the pores will serve to in- crease the effective value of A and thus to increase the efficiency of the shaly sand as a membrane. Since the total electrochemical S.P. is the algebraic sum of the shale potential and the potential between the mud and interstitial water in a permeable bed, an increase in the membrane efficiency of a shaly sand implies a de- crease in the total S.P. In the earth the sign of the ordi- nary liquid junction potential is such that this potential increases the S.P. If, however, the efiiciency of the shaly sand as a membrane becomes sufficient to reverse the t Note added in proof : An investigation of the S.P. and resistivity plienomena shown by dirty sands lias recently been published by Wyllie and Southwick, 1954. This investigation obtained data that provide answers to many of the problems that were un- answered in 1952. * This approach will be found similar to that used bv deWMtte (1950), who apparently was unaware of the M.S.T. theory or the work of Sollner. However, deWitte found no concentration de- pendence of the potential of the type shown in figure 11. This is not in accord witli all other work on membrane electrodes. GO 90 NERNST POTENTIAL y^^ AA / 30 d / ' 01/ / ACTIVITY RATIO /■K) / / / -10 LIQUID JUNCTION POTENTIAL 0.8 1.0 '/a. lao 18.0 FiGUKE 11. The form of the variation of potential with .abso- lute concentrations for an imperfectly selective negatively charged membrane separating NaCl solutions of constant activity ratio. liquid junction potential, this potential will actually oppose the shale potential. The decrease in the total S.P. may thus be very great. Figure 11 shows a typical curve for a shaly sand separating NaCl solutions having an activity ratio of 10: 1. The two limiting potentials are the Nernst poten- tial of about 58.5 mv at the temperature considered, and the liquid junction potential of — 11.5 mv. Here the positive sign is applied to the solution of lesser activit.y. When both solutions are very dilute, i.e. ag is small, the potential developed approximates to 58.5 mv. When both concentrations are large, the curve is asymp- totic to — 11.5 mv. For the optimum application of equation (3) the po- tential should be • — 11.5 mv. Figure 11 suggests that shaly sands will involve least error in the use of equa- tion (3) when interstitial water salinities are high, and that in all eases the error will be decreased if the mud salinity is maintained at a high figure. In the case of very shaly sands allied to very low interstitial salinities it would appear desirable to increase the mud activity to a figure greater than the interstitial water activity and to record positive S.P. deflections. Low mud resistivities, that is high mud activities, seem extremely desirable. They should improve the ac- curacy with which the electrochemical S.P. approxi- mates to equation (3), they serve to diminish the elee- Irokinetic potential correction, and they assist in the interpretation of focnssed current resistivity logs. When running conventional S.P. logs these advantages may be partly offset by the short-circuiting effect of the con- ductive mud, which serves to reduce the S.P. This mav be Part VII Clay Teciixology ix the Petroleum Industuy 301 OYercome by usingr a Static S.P. lo0, Xew methods of correlation by resistivity value of electrical logs : Am. Assoc. Petroleum Geologists Bull., V. 34. pp. 2027-2060. de Sitter. L. TJ., 1947, Diagenesis of oil-field brines : Am. Assoc. Petroleum Geologists Bull., v. 31, pp. 20.30-2046. dc Witte, L., 1950, Experimental studies of the characteristics of the electrochemical potentials encountered in drill holes: Am. Inst. Min. >Iet. Eng., Paper 12G, Oct. 13, 1950 meeting, Los An- geles, f'aliforiiia [unpublished]. de Witte, L., 1950a, Relations between resistivities and fluid contents of porous rocks : Oil and Gas Jour., v. 49, no. 10, pp. 120-1.32. Doll, H. G., 1949, The S.P. log — theoretical analysis and princi- ples of interpretations: Am. Inst. Min. Met. Eng. Trans., v. 179, pp. 146-185. Doll, H. G., 1950, Selective S.P. logging: Am. Inst. Min. Met. Eng. Trans., v. 189, pp. 129-142. Doll, H. G., 1950a, The S.P. log in shaly sands : Am. Inst. Min. Met. Eng. Trans., v. 189, pp. 205-214. Grim, R. E., and Cuthbert, F. L., 1945, The bonding action of clays. Part 1 — Clays in green molding sands : Illinois Geol., Sur- vey Rept. Inv. 102, 55 pp. Harned, H. S., and Owen, B. B., 19.50, The physical chemistry of electrolytic solutions, Xew York, Reinhold Pubiishing Corp. Kressman, T. R. E., 1952, in Duncan. J. P., Theory and prac- tice of ion exchange : X'ature. v. 169, pp. 22-24. Marshall, C. E., 1948, The electro-chemical properties of mineral membranes. Till. The theory of selective membrane behavior: Jour. Phys. Colloid Chemistry, v. ,52, pp. 1284-1295. Marshall, C. E., 1949, The colloid chemistry of the silicate min- erals, 195 pp., Xew York, Academic Press, Inc. McLean, E. O., Barber. S. A., and Marshall, C. E., 1951, Ioni- zation of soils and soil colloids. I. Methods for simultaneous deter- mination of two cationic activities : Soil Sci., v. 72, p. 315 Meyer, K. H., and Sievers, J.-F.. 1936, La permeability des membranes. I. Th«^orie de la permeability ionique: Helvetica Chi mica Acta, v. 19, pp. 649-6G4 .... II. Essais avec des membranes selectives artificielles : Helvetica Chimica Acta, v. 19, pp. 665-667 .... III. La permfabilitS ionique dc couches liquides non- aquenses : Helvetica Chimica Acta. v. 19, pp. 948-962. Millot, G., 1952, in Afackonzie, R. C, Geological aspects of clay mineralogy : Xature, v. 169, p. 6.56. Mounce, W. D., and Rust, W. M., 1945, Natural potentials in well logging: Am. Inst. Min. Met. Eng. Trans., v. 164, pp. 288-294. Patnode. H. W.. and Wyllie. M. R. J.. 19.50, The presence of conductive solids in reservoir rocks as a factor in electric log interpretation : Am. Inst. Min. Met. Eng. Trans., v. 189, pp. 47-52. Schlumbergor, C, Schlumberger. M.. and Leonardon, E. G., 1934, Electrical coring, a method of determining bottom-hole data by elec- trical measurements: Am. Inst. Min. Met. Eng. Trans., v. 110, pp. 2:;7 272. Schlumberger. (,'.. Schlumberger. JL. and Leonardon. E. G., 1934a. A new contribution to subsurface studies by means of elec- trical measurements in drill holes : Am. Inst. Min. Met. Eng. Trans., v. 110, pp. 27.3-289. Sollncr, Karl. 1945. The physical chemistry of membranes with particular reference to the electrical behavior of membranes of porous character : Jour. Phys. Colloid Cbemistrv, v. 49, pp. 171- 191. SoUner, Karl, 1949, The origin of bi-ionic potentials .icross po- rous membranes of high ionic selectivity : .Tour. Phvs. Colloid Chemistry, v. .53, pp. 1211-12.39. SoUuer, Karl, 19.50, Recent advances in the electrochemistry of membranes of high ionic selectivity : Electrochem. Soc. Jour., v. 97, pp. 139C-151C. Taylor, E. McK.. 1929, The replaceable bases in the clays and shales overlying petroliferous strata : Inst. Petroleum Technology Jour., v. 1.5. pp. 207-210. Taylor, E. McK., 19.30. An examination of clays associated with oil-bearing strata in the United States : Inst. Petroleum Tech- nology Jour., v. 16, pp. 6S1-0S3. Teorell, T., 1935, An attempt to formulate a quantitative theory of membrane permeability : Soc. Exper. Biol. Medicine Proc, v. 33, pp. 2S2-2S5. Teorell, T., 1935a, Studies of the diffusion effect upon ionic dis- tribution. I. Some theoretical considerations : X'at. Acad. Sci. Proc, V. 21, pp. 1.52-162. Tixicr, M. P., 1949, Electric-log analysis : Oil and Gas Jour., V. 48. no. 7. pp. 143-14.S. 217-219. Wyllie. M. R. .!.. 1948. Some electi-odiemical properties of shales: Science, v. lOS. pp. 684-680. Wyllie. M. R. .1.. 1949. A quantitative analysis of the electro- chemical component of the S.P. curve : Am. Inst. Min. Met. Eng. Trans., v. 186. pp. 17-26. Wyllie, JL R. J., 1949a, Statistical study of accuracy of some connate-water resistivity determinations made from self-potential log data : .\m. Assoc. Petroleum Geologists Bull., v. 33, pp. 1,892- 1900. Wyllie, M. R. ,1., 1951. An investigation of the electrokinetic comiionent of the self potential curve: Am. Inst. Min. Met. Eng. Trans., v. 192, pp. 1-18. ^Vyllie, M. R. .!.. 1951a. Theoretical considerations involved in the determination of petroleum reservoir parameters from electric log data: 3d World Petroleum Cong. Proc, sec. II, pp. 378-393, Leiden, the Netherlands, E. J. Brill. Wyllie, M. R. J., 1951b, U. S. Patent No. 2,569,625, October 2, 1951. AV.vlIie, M. R. .L, 1952, Gordon research conference on ion ex- change : Am. Assoc. Advancement of Sci., ,Tul.v 1952. Wyllie, M. R. J., 1952a, Procedures for the direct employment of neutron log data in electric log Interpretation : Geophysics, v. 17, p. 790. Wyllie, 51. R. .!., and Morgan, F., 1951, Comparison of electric log and core analysis data for (5ulf's Frank No. 1, Velma pool, Stephens County, Oklahoma, in llfteenth technical conference on petroleum production : Pennsylvania State College, Mineral In- dustries Exper. Sta., Bull. .59. pp. 111-127. Wyllie, M. R. J., and Patnode, H. W., 19.50. The development of membranes' prepared from artificial cation-exchange materials with particular reference to the determination of sodium-ion activities : Jour. Phys. Colloid Chemistry, v. 54, pp. 204-227. Wyllie, M. R. .1.. 19.54. Ion exchange membranes. I : .Tour. Phys. Chem.. v. .58. pp. 67-73. Wyllie, M. R. J., and Kanaan, S. L., 1954, Ion exchange mem- branes. II : Jour. Phys. Chem.. v. 58, pp. 73-80. Wyllie, M. R. J., and Southwick, P. F., 19.54, An experimental investigation of the S.P. and resistivity phenomena in dirty sands : Jour. Petroleum Tech., v. 6, No. 2, pp. 44-57. ROLE OF CLAY IN OIL RESERVOIRS By NoRRia Johnston * Occurrence. The followiug observations apply to the saudstoue oil pools of the United States. Carbonate reservoirs seldom, if ever, contain appreciable quantities of clay. Sandy sediments, on the other hand, frequently contain clays of many types, in quantities ranging from practically zero to a very high percentage. Hughes and Pfister (1947) have stated: "Few producing oil sections are free from clay." This is contrary to the prevailing opinion 6 or 7 years ago, \yhen the remark was fre- quently heard that "California sands may be dirty (argillaceous), but east of the Rockies, clay in oil sands is a rarity." It may be true that very clean sands are more frequently found in central plains and eastern areas than in California, and even that the average condition of all sands east of the Rockies may be classified as much cleaner than that of Pacific Coast sands. But the fact remains that most oil and gas sands do contain certain quantities of many types of clay. The types %vhich have been found and identified in- clude chiefly montmorillonite, illite, and kaolinite, witli many less commonly mentioned forms, such as attapul- gite, serieite, chlorite, anauxite. and so on. For the specific purposes of tlie petroleum engineer, it has been customary to group all clays as belonging to, or at least as being similar in behavior to, the three main types, montmorillonite, illite, and kaolinite. In this order, the clays show properties ranging from extreme surface activity, strong swelling tendency, and great ease of base exchange, to a set of opposite characteristics : sur- face inactivity, absence of swelling tendency, and lack of base-exchange activity. It is obvious from these prop- erties that the effects on the accumulation and produc- tion of oil should be greatest in the case of the active montmorillonites, and least in the ease of the compara- tively inactive kaolinites. Identification of the clays is carried out by the usual means. After mechanical dis- integration of the sands, finer constituent particles, which normally include most of the clays, are separated by elutriation. X-ray diffraction, differential thermal analysis, infra red analysis, microscopic examination, and other means are employed to obtain specific identi- fication. Clays are important in oil sands almost entirely be- cause they occur in small particles adjacent to the sand grains, in a region whicli is normally occupied by the interstitial water almost always found associated with oil sands. The reason this association is so important is that clays exhibit their activity almost exclusively with respect to water and its dissolved ions. Clays dis- persed in oil, or in contact with it, in the absence of water would in general behave as inert particles. So the surface-active clays would be largely indistinguish- able in behavior from tlie inactive clays. The particles of clay normally are attached to the surfaces of the sand grains in whatever distribution chance and the forces of water or air dictated at the time of deposition of the sand grains, or during the time of formation of the clay from the sand. It is generally believed that clay particles are either deposited at the same time as tlie sand grains * President, Petroleum Technologists, Inc., Monteliello, California. or are formed in situ, mainlj- from feldspars, after deposition of the sand, rather than that the.v migrate into position through the pore channels of a previously sedimented, clean, porous medium. Thus it is statistically likely tliat quite often, two adjacent sand grains may make contact oulj- tliroi;gh one or more clay particles. Figure 1 is a sketch to illustrate various sand and claj' configurations with reference to the oil and water con- tent of a pore in an oil sand. Three sand grains form the pore, which is lined with interstitial water in which are submerged several claj' particles. At the sand grain contact A, there are shown three clay particles in hydrated form, separating the grains by a finite and appreciable distance. At B are several clay particles which hydration has swelled and weakened to such an extent that portions have become dislodged and are free to migrate with the fluids in the pore channels. At C are shown some portions of cla^- particles extending fartlier than usual from the sand surfaces and causing some excess volume of interstitial water to be held in the pore space. At D are shown some hydrated clay par- ticles which have attracted from the ionic atmosphere of the surrounding brine certain positive ions, which tend to neutralize the inherent negative charge of the clay particles. Effects of Clay on Oil-Sand Behavior. The chief characteristic of clays that makes a dirty sand dift'erent in its behavior from a clean sand is the extensive hydra- tion, which results in swelling and consequently in a partial loss of permeability of the sand. Another result of the forces that accompany swelling is the increased compressibility of a dirty sand as compared with a clean A. CLAr HYDRAT/OfJ EXPANDS MEDIUM B. HYDRATIOf^ StVELLING LOOSENS R^RTICLES C. HYDRATION INCREASES INTERSTITIAL WATER D. IONIC LAYER - CONDUCTING SOLIDS Figure 1. ( 30C ) I'lirt VTTl Ci>AY Tixnxoi.oGY IX THE Petrolkim Txm-STKY 307 saiiii. Interstitial water tends to be more abundant in an argrillaeeous than in a elean sand. Tlie tendenoy for a clayey sand to be water-wet is grreater than for a elean sand, as clays are normally strongrly hydrophyllie. The stronpr tendency for ion adsorption makes a clayey sand give (piite a different electric-lo 'YPiCAL Dif?rY ,Na CLEAN s/mr> ^/'ts.ai 6<: ' S73 S7B FlGtTRE 3. solutions contribute somewhat to hydration, or possibly to microscopic rearrangment of clay particles. It was Johnston and Beeson 's view that, except for the appre- ciable effects of overburden pressure and formation temperature and possible adsorption of some of the oil constituents on some of the solid surfaces, the brine permeability was a closer approach to the true single- phase reservoir permeability than is the commonly IO,O00 • ^ k "^ ■\ \ ^ \ A \ 1 \^ \ \\ yJOU \\\ 1 \\ Q. p. ZRMEAt vun vs SALINh r ^ \ \ \ \ \ ' \ AIR 300 SALINITY £00 100 GPG CHLORIDE Figure 4. I'ait VII Clay Technology in the Petroleum Industry 309 Fff£SH M^reft Fe£iJ^ HWygJP BKM£ pepMEA&iurr _ Afct FlGUBE 5. incasurod air permeability. Since that time one major oil company has measured more than 15,000 brine and fi-esli-\vater permeabilities, thus learning a great deal about the distribution of hj-dratable clays in oilfields over a wide area. Johnston and Beeson pointed out certain sources of error, such as the toluene extraction followed by oven drying. However, to obtain the single-phase permeability of a reservoir sand wet with its interstitial brine, there appeared to be no less damaging way than to extract the oil and vacuum saturate with the brine. They showed the effect of i-edueed permeability around a well bore on the resulting well prodvictivity, in terms of both radial distance of the plugging effect of fresh water in- trusion, and also the degree of susceptibility of the sand. Figure 2 shows the drastic effect on well productivity of severe hydration plugging for only a short distance into the sand, and conversely the relative difficulty of im- proving well productivity by increasing sand permea- bility, even 10-fold, for a considerable distance from the well bore. Figure 3 shows on a log scale some typical permeabilities of sand samples to air, brine, and fresh water. The upper five samples would be considered argillaceous or "dirty"; the lower five are "clean." Figure 4 shows the effect on a few samples of succes- sively lower brine concentrations, from ?, percent to fresh water. The log scale somewhat masks the drastic nature of the reduction, but allows actual values to be shown. The differently sliaped curves suggest the great variety in type, quantity, and distribution of clays in the pore spaces of oil sands. The fact that permeability declines relatively little as the salinity is reduced several- fold suggests that where necessary, fresh water may be rather freelj- admixed with appropriate brines, for sub- surface in.jection, as in secondary recovery operations, in sands that would refuse to accept fresh water itself at any eeo)iomic rate. Figure 5 shows how clearly fresh water can dift'erentiate between the strata of laminar sediments as regards hydratable clay content, as con- trasted with the relative constancy of permeability char- acteristics in a massive sediment. The ratio of fresh water to brine permeabilities varies from to 100 per- cent for the laminar sand, and 51 to 81 percent for the P€fiMEABILrTy MD Figure 6. massive sand. Figure 6 shows the sharpness with which fresh-water permeability can indicate a sudden change from a clean to a dirty sand, whereas air and brine were ineffective in the differentiation. The samples were taken every 18 inches over a vertical distance of 23 feet. Many other illustrations could be cited of the presence of clays as shown by water permeabilities. Clay content can be determined by quantitative measurements, and the min- eralogical classification can be determined by the more elaborate physical measurements such as X-ray diffrac- tion, infra-red, and differential thermal anal.yses. Fre- quently the result desired is a determination of the effect of the clay content on permeability, so the simplest approach is the single-phase measurement of the water permeability. Yuster (1945) favored the use of brines over fresh water for flooding, although the Bradford operators con- tinued to use fresh water as it was more readily avail- able and laboratory tests showed no greater recovery with brine. In 1946 Kersten showed about a 2 to 1 im- pi-ovement in initial productivity index of wells com- pleted with oil-base mud as compared Avith fresh-water clay-base mud. His graphic presentation made tliis deduction inescapable. He stated, too, that the higher index for oil-base completions was apparently main- tained through much of the well history. The same year Kelley, Ham, and Dooley (1946) arrived at several pertinent i-ecommendations for obtaining maximum M'ell productivity: minimize water entry into the sand, add salts to whatever water may be allowed to enter, keep the time of contact of drilling muds against the produc- ing sand down to a minimum, use low-weight mud, and select the mud-treating agents so they will have a mini- mum hydration effect in the oil sand. Here we have an interesting and troublesome conflict : the chemical ad- ditions to drilling mud should disperse the mud clays thoroughly, but immediately on entering the oil sand they should change their nature to that of flocculating agents ! Wade (1947) has presented a monumental study of the contributions of many factors to the productivity of an oil well. One of his conclusions is that the prodnc- tivit3- index of a well is higher the shorter the time the 310 Clays and Clay Techxology [Bull. 169 drilling mud has stood against the formation, and the less the time allowed for this action, the better, indicat- ing that filtrate action on the sand near the well bore is unfavorable to well productivity. This action may include both water block by hydration of clays, and the intrusion of drilling-mud clays into the sand insofar as this is possible. One other conclusion of Wade is that individual pool correlations show oil-base mud to give a higher well productivity than water base, by factors varying from 1.5 to 2.0. Short time in the zone shows advantages as great as 246 percent in completion effec- tiveness. Further, "Wade states that the average inter- stitial water of dirty sands at 100 md may exceed the value for clean sands by as much as 27 percent, which also contributes to lower productivity for dirty sands. At 10 md, the interstitial water values differ by 60 per- cent, which helps to explain the wider divergence of pro- duction effectiveness from the theoretical value at lower sand permeabilities (Johnston and Sherborne 1943). Radford (1947) discussed the improvement in mud not only to avoid drilling trouble, but to avoid damage to the "productivity of the sand. He also stressed the deleterious effect of long times of completion. One ex- ample showed the performance of two quite comparable wells, one which had been completed in 2.5 weeks, the other in 5 months. The rapidly completed well flowed 80 percent of its ultimate recovery in 12 mouths and was abandoned in 4.5 years. The well completed slowly did not flow at all, and in 11.5 years produced only 40 per- cent as much ultimate recovery per acre foot. The choice IjPtween oil and oil-base mud was inconclusive, but the choice between either of these and water-base mud was decisive. In commenting on the work of Radford, Dan Johnston (Radford 1947) showed an advantage of 2.5 to 1 in well productivity for special low-water-loss mud versus ordinary water-base mud. These observations all ])oint to the damage to the permeability of the sand near the well bore by fresh-water filtrate, the damage being greater as more water invades the region. This is a function of time, pressure difference, mud-cake perme- ability (water-loss characteristic of the mud), oil vis- cosity, number of round trips of the bits and reamers, and of course, the argillaceous character of the sand. Moyer (1927) discussed analytically the effect on economic ultimate recovery of a decrease in sand per- meability around the well bore. He indicated that as fresh-water invasion lowered sand permeability by clay swelling, it also, and conse(iuently, increased the inter- stitial or immobilized water. For an arbitrary and rather mild case of permeability damage, involving a loss from 57 md to 10 md, he showed analytically a decrease in final productivity index of 70 percent and a loss of ulti- mate recovery of 4 percent for an invaded zone only 15 inches deep around the well bore. Also the time required to recover the oil was greatly increased. A more drastic reduction of permeability would have been well within the limits of actual California practice in many in- stances, and would have resulted in still greater losses of productivity index and economic ultimate recovery, and still greater lengthening of the time required to obtain the ultimate recover}'. Hughes and Pfister (1947) made a most comprehensive study of the relation between injection well indices and laboratory data as affected by the saline content of the water. This work was carried out for the benefit of secondarj'-recovery operations, and specifically compared well rates using subsurface brines with well rates using fresh water or fresh water with such additives as sodium carbonate and sodium bicarbonate. Early trials in Brad- ford using these two additives in fresh water, supposedly to strip oil off the sand grains, were disappointing, as the high pH and prevalence of sodium ion tended to base-exchange the clays to the sodium form, which then hydrated more completely and lowered the well index. They conclude that injection water must be such as to keep anj' formation clays flocculated, and, in general, the best water for this purpose is the subsurface brine origi- nating in the sand. Use of this brine at once solves the disposal problem, maintains reservoir pressure, maintains well index, and tends to increase economic ultimate recovery with a reduction in the time required for re- covery. Thus the produced brines begin to take on the quality of a natural resource rather than a nuisance. Some of the observations leading to the above conclusions are rather striking. A group of 16 Bradford core samples tested in the laboratory showed subsurface brine per- meability was 280 percent of fresh-water permeability througli the sands. Another group, from I^>artlesville, showed an average brine permeability of 400 jjercent of the fresh-water permeability. Well-injection rates re- sponded immediately to brine replacing fresh water, to the extent of 20 percent, with 10 percent lower injec- tion pressure at the sand face, and the reaction was found to be reversible. Most of the laboratory data were taken on .sands containing mainly illite ; where montmo- rillonite jiredominates, the effects would be more striking. Commenting on the survey by Hughes and Pfister (1947), K. B. Nowels cited the case of the Woodson field, Throckmorton County, Texas. In 1941 the injection of clear, fresh water plugged an injection well from an intake rate of 23 barrels per day to 1 barrel per day in 13 days. Another dropped from 24 barrels per day to zei-o in that time, and the average of 6 wells decreased from 22 barrels to 1.6 barrels per day during the test. Because it was felt that one of the lime feeders had failed to operate correctly and had fed too much lime, the wells were treated with 50 gallons of 65 percent hydrochloric acid and the Avater injection continued im- mediately. The result was astonishing; injection rates recovered from 1,6 barrels to an average of 52 barrels per day, nearly 2.5 times their original rate. This new injection capacity- was maintained, or improved in some cases. It is now felt that what actually caused the recovery of sand permeability was the result of base exchanging the clays to the relatively nou-hydrating hydrogen form. The reaction of fresh water on the acid form of the clays is not serious, and as long as no appreciable quantities of sodium salts are bi'ought into contact with the clays, future fresh-water hydration will be slight. This type of observation is what leads the present writer to believe that there may be better injec- tion waters than the indigenous water itself, but cer- tainly the latter is usually the most satisfactory of readily available waters. Sherborne and Fischer (1949) showed that oil-base mud completions average about twice the productivity index provided by water-base mud completions. This gives c|uicker recovery and higher idtimate recovery. I'arl Vll C'l.AY Technology ix the PETROLEnr Inditstry ■Ul They further stated that any use of water in a well is to be diseourajred, but oeeasionally the need for reliable electric logs overshadows tlic iiii|)ortanci' of well produe- tivity. Ilujrhes (1950) has iJi-cscntcd a thor(inj;h summary of the many concepts relatinjr the behavior of oil reservoirs to their clay content, lie quotes Kryninc (1945) as elassifyiufj sandstones as ortliorpiartzite, graywacke, or arkose in nature. The first is clean sand ; the second con- tains angular grains of (piartz. feldspar, and chert, set in a matrix of clay; the third comprises "ashy, light gray (or red) dirty sands with much feldspar," fre- quently cemented by clay. "The 'fines' of the ina.jority of producing oil sands consist of particles of shales, silts, and clays, and impart the characteristic dirty ajipear- ance. " Hughes states that clays aid compaction and shrinkage and help production by promoting subsidence on release of reservoir pressure. Though mud techniques are highly developed, the i)roblcms of keeping mud fil- trate and mud particles out of the sand are less thor- oughly solved, and in many ways are more serious than mud problems connected with drilling. There is an unsolved conflict between the best mud for electric log- giug and that for optimum well-completion effectiveness. It was pointed out that in well washing and clean-out techniques great care should be exercised to use only those chemical substances that woidd assure floeeulation of formation clays. Somerton (1949) has suggested that tlie lack of an effective water drive in several California reservoirs may be attributed to the high clay content of the reservoir sands. Hughes further urges caution in the use in injection water of certain organic materials of a surface-active nature, as they may tend to disperse clays. All authors agree that, whatever is done to the oil sand around a well boi-e. the days must be and must remain flocculated. Nahin (1951) and co-workers conclude for the specific eases of the Stevens sand in Paloma and the Gatchell sand in East Coalinga that the clay content identified is in harmonv' with the observed sensitivity of the Stevens and insensitivity of the Gatchell to fresli-water hydra- tion plugging. The Stevens sand contains a])])rcciable quantities of hydratable montmorillonites, while the Gatchell sand contains clays mainly of the kaolin type. AV. T. Cardwell, Jr. (Nahin et al., 1951) comments favor- ably on the citing by Nahin of a minimum of unsup- ported speculation, and agrees that "perhaps a more important factor than the amount of potentially swell- able clay present is the manner in wliich such a clay component is interlaminated with the non-swelling min- erals in the reservoir matrix." Cardwell further states that to be liighlj^ swelling, a clay must be a montmoril- lonite, but that even this may be insufficient, as some montmorillonites do not swell. Actual swelling tests on the fines must be made before the presence of hydratable clay can be proved. Compressihility of Sand. Just as shale is compress- ible inider overburden pressures, so also is sand contain- ing an appreciable quantity of elaj', more particularly when many of the clay particles occur as the bond be- tween adjacent sand grains. Clay and shale are subject to extensive compaction, probably largely by reduction of tlu'ir water of hydration by virtue of pressure. Power * showed experimentally imbibition pressures of shales ranged from about 3,000 to 19,000 psi. Clay particles supporting overburden pressures at sand-grain contacts would certainly ])e subjected to pressure intensities of this magnitude, i)articularlj' as the pressure in the fluid phases is reduced by production of the fluids. This com- paction of a dirty sand reduces the pore .space, and adds to the production of oil or gas per psi pressure decline over the comparable amounts recovered from a rigid sand. Also, the compaction apparently r(>sults in over- burden subsidence. Subsidence of the ground level fre- (piently takes j)lace. as at Maracaibo and Long Beach, when extensive oil and gas withdrawals have been made from arkosic sand. In some hca\y oil fields, an apjireci- able fraction of the entire iiltimate recovery is made available by virtue of pay zone compaction. Another evidence of compaction is the shi-inking of dirty core samples when they are cleaned and dried in the laboratory. Sometimes this is so extreme that the sum of the water and oil extracted from a sample is greater than the entire pore space after drying the extracted sample, and apparent saturations exceeding 100 jjcrcent of the pore space are the result. In eases of extreme contraction, it becomes desirable to measure the pore space by the difference in weight of the dried and vacuum-saturated core samples, so as to have a measure of the "wet pore space." An alternative method is to measure the bulk volume, dry and wet, and add the difference to the falsely small dry-pore space. Interstitial Water. The amount of interstitial water held by a sand at capillary equilibrium is dependent on the size of pore spaces and channels, the number of grain contacts, and the area of rock surface exposed to wetting by interstitial water. Obviously a fine sand will have finer pores, greater surface area, and more grain con- tacts, and therefore will hold more interstitial water than a coarse sand. A sand with greater distribution of sand-grain sizes will likewise hold more water. The pres- ence of hydratable clay is not only a contributing factor to grain-size distribution, but the tendency of clays to bind water adds to the other factors to in.sure that a dirty sand will contain more interstitial water than a clean sand having the same characteristics of the nou- swelling matrix. Figure 1 shows a po.ssible physical pic- ture of the wa.y in which greater quantities of inter- stitial water may be held by the clay content of a dirty sand. Hydrophilic Wetiabilitf). Because clays actively ab- sorb or attract water, their presence in quantity on the surface of sand grains may tend to offset the oil-wet tendency of certain sand-grain surfaces and make a sand more nearly water-wet than would be the case for the same sand in the absence of the clay. Conducting Solids. The conducting solids in surface- active clays are of special interest to those interpreting electric logs. The.y owe their existence to the fact that clay particles are usually negatively charged in water, so that a layer of positive charges is attracted to the surface. This layer of ionic charge is closelj' allied to the 1 Personal communication. 312 Clays and Clay Technology [Bull. 169 surface of the clay, but the charges are not immobilized as in the case of au insulating material. Thus they be- come available for electrical conduction. This type of conduction is ascribed to the solid itself, as it is entii'el.y different from the electrolytic conduction in the pore channels. Also, the percentage of ions in the interstitial water that is adsorbed on the clay surfaces varies with the concentration of the electi'olytes present. At low salinity, a greater fraction of the ions available is ad- sorbed than at high salinity. Thus there is a tendency for the formation factor to be more influenced by a given clay content of the sand when the interstitial water is of low salinity than when the brine is concentrated. Control of Clay for Better Oil Recovery. The most successful way to combat hydration swelling of clays in oil sands is to keep extraneous water entirely out of the zone. If water must be used, it is best to use water carrying a fair concentration of soluble calcium salts, provided this is known to be compatible with the for- mation water. Another, usually more powerful corrective for clays that have been swelled by liydration is treat- ment with concentrated hydrochloric acid. The shorter the time any mud is left against the pay sand, the higher are the resulting well-productivity in- dices. Damage with time is worst in the ease of dirty sands and ordinary clay-water mud. Least damage oc- curs with oil-base mud, or oil, against a clean sand. Those sands which show a great decrease in permea- bility from air to brine, and no great further decrease with fresh water, probably are showing the result of loosening of particles, rather than hydration. Not much treatment can be developed for such sands, other than to di-aw ou them at low rates so that the loose particles will not all migrate to constrictions, which they are capable of plugging. Subsidence may be avoided and a menacing situation turned into a promising one by replacing oil and gas withdrawals with injected water. If the injection pro- gram is carried out intelligently, a quite appreciably greater ultimate recovery is almost certain to result. A pool that has been producing great quantities of water will not necessarily yield much higher ultimate recovery with water injection, but most pools not having a suffi- cient natural water drive will be benefitted by secondarj^ methods, and the tendency toward subsidence will be greatly reduced or entirely abated. Any injection water should be carefully studied with respect to compatability with both the formation and the indigenous water. If an injected water maintains as high a sand permeabilitj" as that measured with the formation water itself, the injection water is certainly of desired quality. It is thought that some slight im- provement over formation water permeability may be possible where the formation water is of low salinity, although a brief study of figure 2 will show that no extensive effort along this line is warranted, unless and until severe damage has been inflicted. The main recom- mendation for all cases is that the injection water should be so treated that formation clays are flocculated. Problems for Research. A great many interesting problems exist in this field, and successful research would \inearth information of untold value. The fol- lowing problems are some of the more important and worthwhile : 1. Compressibility of sands under reservoir condi- tions for sands of varying type, amount, and clay- content distribution. This study should also encom- pass the effect of fluid pressure within the pore system. 2. Means of adjusting drilling-mud filtrate so that it disperses clays in the mud, j-et flocculates clays in the oil sand. 3. Means of economically recuperating the sand per- meability after hydration damage. 4. Measurement of true reservoir porosity. 5. Measurement of true and altered reservoir per- meability. 6. Measurement of true reservoir water content. DISCUSSION W. T. Cardwell, Jr.: Where are the clay particles located with respect to the sand grains as they exist in the reservoir, and is it possible that the position of the clay particles may be changed as a consequence of coriuK and extraction of the core sample? Can the filtrate of a salt-water mud harm a formation by destcell- ing or flocculating clay in the formation; and could the flocculated clay particles be loosened in this action and be allowed to migrate and act as check valves in the flow channels? If the formation permeability had been reduced only by clay swelling would it not be possible to reverse the process by injecting an electrolyte which would return the clay to its original condi- tion? Would it not be possible that the reversal could take place during production as the contaminating electrolyte was displaced through diffusion and flow of the original interstitial water? Ralph E. Grim: Clay may occur as discrete particles mixed with the sand grains. It is very common, however, to find clay actually plastered on the surfaces of the sand grains; in that case it may not be observed by microscopic examination and may be particularly difiicult to sepa- rate from the grains. Should a dispersing agent be used, it may be- come impossible to determine the cation situation in the original clay. Norris Johnston: In answer to Cardwell's first question, I agree with Grim that more commonly clay is attached to the sand grains, rather than being in the form of discrete particles. Therefore it would seem logical that neither the coring procedure nor the extraction would involve sufficient forces to separate any appreciable portion of the attached clay. Cardwell's second question is more difficult to answer. If the salt- water mud filtrate carries chiefly sodium chloride, the deswelling or flocculating action may be accompanied by a base exchange that may promote more active swelling when the formation water returns after the invasion by the filtrate. This subsequent swelling might loosen particles of clay. The loosening would not seem so easily explained as the result of flocculating as it would as the result of swelling. If the filtrate contained an important quantity of calcium salts, the possible base exchange might favor deswelling or flocculation, which should retard rather than accelerate detach- ment of clay particles which might plug flow channels. In any case, coring, and possibly also extracting, may make .some alterations in the clay status relative to the original condition in the reservoir. One way to minimize this is to use oil or oil-base mud or oil phase emulsion for the drilling fluid, in order to avoid an aqueous filtrate. I. Barshad: In answer to the third question it appears that this is similar to the reversibility of alkali soils. If an alkali soil high in salt is saturated with a salt solution, the permeability is relatively high. If salts are leached from the soil, particularly sodium salts which produce swelling, the permeability is reduced. This can be reversed by adding another .salt such as calcium sulphate and the perme- ability is reversed. Essentially the same phenomenon probably I'iiit Vill Clay Technology ix the PETROLEr>[ Industry :?13 occurs in reservoir sands. If the reservoir sands are contacted hy a sodium salt solution, the salt will diffuse into the sands. The sodium salt can then be removed hy water and the process is revers- il)le. If a calcium salt contacts the reservoir sand, the clay present may remain flocculated and the process may not be reversible. SELECTED REFERENCES Beckstrom, R. C, and Van Tuyl, F. M., 1!VJ7, Effect of tiooding oil sands with alkaline solutions : Am. Assoc. Petroleum Geologists Bull., V. 11, pp. 223-227. Carll, John F., 1880, The geology of the oil regions of Warren, Venango, Clarion and Butler Counties : Pennsylvania Geol. Survey 2. v. 3, pp. 263-269. Case. ly. C, 1933, Base replacement studies of Oklahoma shales — critique of Taylor hypothesis: Am. Assoc. Petroleum Geologists Bull.. V. 17. pp. 66-79. Fancher. G. H., Lewis, J. A., and Barnes, K. B., 1933, Some physical characteristics of oil sands: Pennsylvania State College Min. Ind. Exper. Sta., Bull. 12, 1941. Grim, R. E., 1939, Properties of clays, in Recent marine sedi- ments : Am. Assoc. Petroleum Geologists Symposium, pp. 466-495. Grim, R. E., 1942, Modern concepts of clay materials : Jour. Geology, v. 50, no. 3, pp. 231-233. Ileek, E. T., 1941, Some theoretical aspects of the use of connate water in flooding operations : Producers Monthly, v. 6, no. 7, pp. 8-11. Ilindry, H. W., 1941, Characteristics and application of an oil base mud : Am. Inst. Min. Eng., Petroleum Technology, Tech. Pub. 1322. Hughes, R. v., 1950, The application of modern clay concepts to oilfield development: Am. Petroleum lust., Krilliug and I'roduction Practice, p. 151. Hughes, R. v., and Pfister, R. .T., 1947, Advantages of brines in secondary recovery of petroleum by water flooding: Am. Inst. Min. Eng. Petroleum Technology, Tech. Pub. 2127. .Johnston, N., and Beeson, C. M., 1945, Water permeability of reservoir sands: Am. Inst. Min. Eng.. Petroleum Technologv, Tech. Pub. 1S71. .Tohnston, X., and Sherborne, .1. E., 1943, Permeability as related to productivity index : Am. Petroleum Inst., Drilling and Produc- tion Practice, pp. 66-80. Kelley, W. P., 1939, Base exchange in relation to sediments, in Recent marine se4-46."». Kelley, W. P., Ham, T. F.. and Dooley, A. B., 1946, Review of special water base mud developments : Am. Petroleum Inst., Drill- ing and Production Practice, p. 51. Kelley, W. P., and .Jenny, H., 1936, The relation of crystal struc- ture to base exchange and its be.'iring on base exchange in soils: Soil Sci., V. 41, pp. 307-382. Kelley, W. P., and Liebig, G. F., Jr., 19:! I. Base exchange in relation to composition of clay with special reference to effect of sea water : Am. As.soc. Petroleum (ieologists Bull., v. 18, pp. 358- 367. Kerston, Glenn V.. 1946. Results and use of oil base fluids in drilling and comidpting wells: Am. Petroleum Inst., Drilling and Production Practice, p. (!!. Krynine, P. D., 1940, Petrology and genesis of the third Brad- ford sand: Pennsylvania State College Min. Ind. Exper. Sta., Bull. 29, p. 71. Krynine, Paul 1)., 1945, Sediments and the search for oil : Pro- ducers Monthly, v. 9, Jan. 1945, p. 12. Moyer, Vaughn, 1947, Some theoretical aspects of well drainage and economic ultimate recover}' : Am. Inst. Min. Eng., Petroleum Technology, Tech. Pub. 2201. iluskat, M., 1937, Flow of homogeneous fluids through porous media, p. 93, New York, McGraw Hill Book Co. Nahin, P., et al., 1951, Mineralogical studies of California oil- bearing formations : Am. Inst. Min. Eng., Petroleum Technology, Tech. Pub. 30.59, p. 151. Nutting, P. (;., 1926, Geological relations between petroleum, silica, and water : Econ. Geology, v. 21. pp. 234-242 .... Oil and Gas Jour., March 31, 1927, p. 76 May 5, 1927, p. 32 Econ. Geology, v. 23, pp. 773-777, 1928. Radford, H. E., 1947, Factors influencing the solution of mud fluid for completion of wells : Am. Petroleum Inst., Drilling and Production Practice, p. 23. Sherborne, J. IC, and Fischer, P. W., 1949, Use of improved drilling fluids in well completion : World Oil, v. 122, no. 7, p. 112. Smith, K. W.. 1942, Brines as flooding liquids : Pennsylvania State College, Min. Ind. Exper. Sta., 7th Ann. Tech. Meeting. Somerton, W. H., 1949, Water flooding as a method of increasing California oil production. Part 1 : California .Jour. JNIines and Geology, v. 45, p. 123 Part 2, v. 45, p. 363 Part 3, v. 45, p. 541. Taylor, E. M., 1928, The bearing of base exchange on the genesis of petroleum : Inst. Petroleum Technology .lour., v. 14, pp. 825-840 V. 15, pp. 207-210, 1929 v. 16, pp. 681-683, 1930. Thorn., W. T., 1926, Possible natural soda drive in the Salt Creek type of pool and its significance in terms of increased oil recoveries: Am. Inst. Min. Eng., Petroleum Devel. and Technology, 1926, pp. 210-217. Travers, W. J., .Jr., 1942, Completion practices related to well productivity : Am. Inst. Min. Eng., Petroleum Technology, Tech. Pub. 1465. " Wade, F. R., 1947. The evaluation of completion practice from productivity index and permeability data : Am. Petroleum Inst., Drilling and Production Practice, p. 186. Waldo, A. W., 1938, Petrology of the Bradford sand of the Kane district: Pennsylvania State College, Min. Ind. Exper. Sta., Bull. 24, p. 75. Yuster, S. T., 1945, Progress reports Pennsylvania State College, Min. Ind. Exper. Sta.: Producers Monthly, v. 9, no. 4, p. 11. USE OF CLAYS AS PETROLEUM CRACKING CATALYSTS By T. H. Milliken,* A. G. Oblad,** and G. A. Mills *•• Introduction. The use of catalysts in the petroleum industry in recent years has undergone a remarkable expansion. The most important of these catalysts are employed in cracking processes, first introduced in 1936 (Houdry et al., 1938). As a consequence of the growth of catalytic cracking, the manufacture of cracking cata- Ij'sts has in itself become a major industry with an esti- mated sales value of $61,000,000 in 1952. Principal em- phasis has been placed on two types of catalysts both essentialljr composed of silica and alumina, one derived from clay and the other synthesized from aluminum and silicate solutions. A total of approximatel.y 470 tons per day of cracking catalysts are manufactured and used to process about 2,000,000 barrels of crude daily. Clay cata- lysts account for about 40 percent of the total manufac- ture. Such a special use of clays lias placed requirements on these materials significantly diiferent than those met in other uses of clays. Moreover, the very particular physical and chemical properties required for good com- mercial hydrocarbon-cracking catalysts limit the eco- nomic use of clays to specific types. Until recently, only a few deposits of subbentonites were of commercial inter- est for catalyst manufacture. As in certain other uses of the surface properties of clays, it is the peculiar structural features of clays which enables them to play a special role for catalyst manu- facture. At present only two clays are used for the preparation of commercial cracking catalyst : montmorillonite and halloysite. The patent literature cites many other nat- urally occurring materials as sources of such catalysts ; kaolinite, vermiculite, and bauxite, for example. In this paper, only montmorillonite, halloysite, and kaolinite will be discussed. The general properties of cracking cata- * Assistant Director of Researcli, Houdry Process Corp., Research and Deveiopnrent L.aboratories, Marcus Hook. Penn. ** Associate Manager of Research and Development, Houdry Proc- ess Corp., Research and Development Laboratories, Marcus Hook, Penn. ••• Director of Research, Houdry Process Corp., Research and De- velopment Laboratories, Marcus Hook, Penn. lysts and raw claj's from which they are manufactured are shown in table 1. which presents comparative data for chemical, physical, and catalytic properties. The catalyst is employed in the cracking of heavy petroleum fractions. In such a process, oil, more or less in the vapor state, is passed over the catalyst at 425- 500°C, atmospheric pressure, and contact times of 6-20 seconds. Under the directive influence of the catalyst, a series of ('(miplicated reactions ( Greensfelder, 1951) takes place with the consequent products much different from those obtained in thermal cracking. Catalytic cracking is employed in the petroleum industry because of the high quality of gasoline produced and becau.se of the desirable distribution of products. Typically, some 6 percent of a methane to propane fraction, 10 percent butanes, 45 percent gasoline, and 40 percent recycle oil are pro- duced. At the same time 1 to 3 percent of the charge stock is deposited on the catalyst as a nonvolatile hydro- carbonaceous residue commonly called "coke." After a 10 to 20 minute cracking period the coke is removed from the catalyst by controlled combustion. In this burning process, the temperature is limited to perhaps 600°C, although a higher local surface temperature may prevail. After such a regeneration the catalyst is ready for reuse. During use at elevated temperatures, the cata- lyst is exposed to various organic compounds including those containing oxygen, sulfur, nitrogen, as well as lieavy metals contained in the charge stock. In addition, the catalyst must withstand water over a wide range of vapor pressures. Whereas early commercial catalytic cracking apparatuses utilized a fixed bed, present day units utilize moving beds with transfer of the catalyst from a cracking zone to a regeneration zone. (Ardern et al., 1951 ; IMurphree, 1951). A typical process is illus- trated by the flow sheet shown in figure 1. In order to maintain a "heat balance" between the endothermic vaporization and cracking of the hydrocarbons and the exothermic regeneration, a large quantity of catalyst must be circulated continuously. Depending upon re- Table 1. Properties of (new) cracking catalysts after calcination at 550'° C. Clay Sj-nthetic Montmorillonite Halloysite SiOi-.iliOa SiOi-MgO Raw .\ctivated Houdry type 1 Raw Activated Chemical SiOa Wt percent 87.5 12.5 0.1 0.0 0.1 300-600 45-50 66.3 1.6 31.8 0.2 0.2 650 45-50 63.7 24.3 5.5 2.5 2.0 2.0 5 71.8 22.1 3.9 1.5 0.3 0.2 300 40 76.7 18.2 4.4 0.1 1.0 230 40 49.1 49.7 0.1 0.6 0.2 0.2 12-20 57.1 \ljO3 - - 40.6 MgO --- - 1.7 FeiOa - - -- '- -- 0.4 CaO - 0.0 NaaO - C.l Physical 160 Catalytic 35-40 Pliysk'al and catalytic properties on fresh catalysts after calcination at about 550° C. (314 ) I'ait VI I J Clay Technology in the Petroleum Industry 315 HOUDRIFLOW CATALYST LIFT EXCESS FLUE GAS ! FldUHE 1 finery requirements, 0.5 to 1.5 tons of catalyst are eir- cnlated per barrel of crude charged. This tremendous circulation reqiiires a high degree of physical stability. In short, the catah'st must maintain cracking activity when subjected to high temperatures, chemical attack from steam and contaminants in the charge, and have good resistance to abrasion. Cracking Catalysts From Montmorillonite. There is no known chemical or physical test which will enable one to ]ircdict whether or not a particular montmorillonite clay will respond to acid treatment and produce a cata- lyst (if liigh activity. Very few montmorillonite clays are known which yield satisfactory catalysts. The situa- tion is similar to that encountered in the bleaching clay industry where it is also necessary to treat the clay sample with acid and test for decolorization activity in order to evaluate the potentialities of a particular clay. Three deposits of montmorillonite clay suitable for the manufacture of cracking catalysts have been exploited on a commercial scale. Two are in Arizona and one in Mississippi. Recently the entire supply of such clay has come first from Chito and then from Chambers, Arizona. Some catalyst has been manufactured from clay mined at Jackson, Mississippi, but this nuiterial has proved to be less desirable than that from the Arizona deposits and its manufacture has been discontinued. Certain other American deposits have been reported bj' Mills, Holmes, and Cornelius (1950) to be satisfactory for the production of cracking catalysts. They found, upon ex- amining a variety of montmorillonite clays, a wide range of susceptibility to activation and that a few clays yielded catalysts of maximum activity. These highly active catalysts proved to be comparable with synthetic silica-alumina catalysts. Suehiro (1949) has also re- ]H)rted on the activation of certain clays for the produc- tion of cracking catalysts. Acid Treatment. Acid treatment of bentonites for the removal of "basic" constituents has been studied by a number of investigators. Nutting (1933; 1935; 1937; REMOVAL .CMS /lOO GMS CLAY Figure 2. Chemical composition of clii.v remaining after acid treatment (ordi- nate), calculated on the basis of constant SiO2=100. Solid circles show Mississippi day, open circles Nevada clav. (After Mills et al., lOiiO, p. 1177.) 1943) has studied the changes in chemical composition believed to be related to decolorizing ability. Other au- thors (Burghardt, 1931; Hagner, 1939; Scliroter, 1940; Hofmann, et al., 1935; Lopez-Gonzalez, et al., 1952; Eseard, et al., 1950) have reported the chemical composi- tion of many clays acid treated to give material with optimum bleaching properties. Figure 2 shows the change in composition of Ash Meadows, Nevada, clay as a func- tion of acid treatment (Mills, et al., 1950). In this figure "R2O3" represents the "basic" constituents of the raw clay. In order to simplify the discussion on the catalj'tic properties of the treated clays, only the initial activities 4! fB'-^^~°~»'°->S*o — ^.-c 08 \ 2 4 6 8 10 12 14 "(yjj" REMOVAL, 6MS./I00 GMS. CLAY Figure 3. Catal.vtic cracking activit.v of acid-treated hentonite as a function of severity of acid treatment. Solid circles show Missis- sippi clay, open circles Nevada clay. (After Mills et al., 1950, p. 1181.) 316 Clays and Clay Technology [Bull. 169 300 1 260 ^220 n c leO ;;i40 ' 100 I 60 20 60 30 20 ^' SURFACE AREA \y /o POROSITY °e~. PELLET OENSITY 2 ■r.o." e 10 12 14 CMS/ 100 CMS CLAY Figure 4. Effect of severity of acid treatment on physical properties (meas- ured on calcined pellets) for Nevada clay. (After Mills et al., 1950, p. 1177.) as measured by the empirical Catalytic Aetivitj- Test A (Alexander, 1947) will be used. This test, referred to as CAT-A, utilizes the conversion of a light East Texas gas oil to gasoline as a measure of activity. In figure 3 the effect of "R2O3" removal on activity of two montmorillonitic ela.ys is shown. The activity rises as the "E2O3" is removed until a maximum is reached. On further removal of basic constituents, the activity begins to decline. It would appear that the two clays, aside from having different activity' maxima versus R2O3 removal, activate to different magnitudes. The activity and product distribution of these particular clay cata- lysts are similar to those obtained on testing active syn- thetic silica-alumina catalysts. The physical changes taking place during removal of the basic constituents of the clay are of interest. The change in surface area and porosity with increasing removal of the basic components of the clay is shown in figure 4 (Mills, et al., 1950). The area increases continu- ously until 14 percent R2O3 is removed. On further acid treatment, the area decreases. Lopez-Gonzalez (1952), Escard (et al., 1950), and others have also reported variation of surface area with acid treatment. The physical properties of acid-treated clay catalysts have also been extensively studied bj' Ries (1952) and by Oulton (1948). These authors reported that the gross physical form, pellet or powder, has no effect on surface area. It is further reported that a to 40 micron fraction of powdered elaj^ had the same surface area as a 40 to 100 micron fraction. This finding indicates that the frac- tion of the total surface area contributed by the mega- surface is negligible compared with the micro-surface. A typical nitrogen adsorption isotherm may be utilized as a measure of the surface as shown in figure 5. The pore-size distribution of elav catalvsts has been reported by Oulton (1948), Ries (1952), Drake (1949), and Mills, Holmes, and Cornelius (1950). Pore size dis- tribution reported by Oulton (1948) for a commercial cla.v catalyst (Piltrol) is shown in figure G. The great similarity between the nitrogen-adsorption isotherms of the aeid-activated montmorillonite clav and 0.1 0.6 P/Po STTROGEl PRESSURE Figure .5. Nitrogen isotherm for acid-activated Ash Meadows, Nevada, montmorillonite. Surface area is 340 square meters per gram. T 1 40 r =^ ^-= i rri , 1 — ; — 20 21 22 23 24 25 26 2? 28 29 30 3l 32 33 34 35 36 3? 36 39 40 RADIUS OF CAPILLARIES, ANGSTROMS FiGURK 6. Distribution of pore size versus surface area for aeid-tn-ated montmorillonite clay. Part Vll Clay Teciixologt ix the Petrolkuji Industry 317 tliose of synthetic silica-alumina ^ ^ -.-''' ;^^ - — " HjO PRESSURE IN ATMOSPHERE Figure 10. Isochore plot for alumina and activated clay. for till' L-lay saiiijilc a larjix' aniouut of the sorbet! water falls in a narrow energy range, 47-62 kilocalories per mole of water. No such narrow range exists for the alumina; the energy changes rapidly with water content in this case. Assuming that the portion of the clay iso- therm parallel to the alumina isotherm represents the water held by the "free" alumina and extrapolating the linear portion of the curve (dotted line in fig. 9), the equilibrium amount of water at 1 atmosphere of water pressure is obtained. For the 450° C isotherm this value is 0.4 percent water. Since the total water sorbed at these conditions is 2.4 percent, the difference leaves 2.0 percent water that can be attributed to residual clay structtire. The theoretical or calculated value for this type of water in the raw Ash Meadows sample of mont- morillonite is 4.93 percent HoO. Thus if 2.0 percent is the valne for the acid-treated sample, 2.0/4.93 or 40 percent of the original clay structure is left after acid treatment. Making proper allowance for the loss of basic constitu- ents, the alumina in the residual clay structure .should be 10. G percent by weight of the acid-treated clay. Since the total alumina content of the activated clay is 22.1 percent, the difference leaves 11.5 percent "free" alu- mina in a non-clay form. Using this value for the alumina, 0.4 percent for the reversible water and 330 sq. m/g. for the surface area, the value 1.1 X 10"* grams of water per square meter of alumina is obtained. This checks the pure alumina value (0.8 X lO"*) remarkably well considering the number of steps used to obtain it. This method, being extremely complex, is not proposed as a means of quantitatively determining the amount of "free" alumina in any given material. However, the agreement shown by data given in column 5 of table 1 indicates qualitatively that a certain fraction of an alumina present in the clay catalysts is behaving very similarly to gamma-alumina. On the basis of the correlation developed in the pre- \'ious paragraphs and taking note of the physical data relating to active clay catalysts, the following mechanism is proposed for the acid activation of clays: 1) all or nearly all the ions held in base exchangeable positions are removed ; 2) the smaller clay particles are completely destroyed, yielding essentiallv a mixture of silica and alumina; 3) jiart of the alumina set free by the acid treatment is dissolved by the acid. In step 2 the larger particles of clay are also attacked by the acid, particu- larly at the edges of the layers. However, in this instance the destruction of the lattice proceeds very slowly as the stability of a large particle is much greater than that of a small particle. Thus, two reactions take place: 1) ■ there is a loss of lattice structure occurring relatively rapidly, and 2) there is a dissolving of alumina from both the "free" alumina created by the destruction of the small clay particles and the alumina still held in the lattice of the large particles. If the acid to clay-weight ratio is high, the "free" alumina created by lattice de- struction is completely dissolved and the activity drops as shown in figure 3. However, the lattice destruction continues by the further removal of alumina from the large crystallites, so that more silica is released. As a re- sult of the release of silica, the surface area increases, as illustrated in figure 4. It is well known that after mild acid treatment the active clay catalyst still shows a strong montmorillonite pattern. This is consistent with the proposed mechanism in that 1) the particles of the raw clay in the size range 20-100 A (the readily destroyed clay fraction) are too small to diffract X-rays; 2) the larger particles (the stable clay fraction) are responsible for the X-ray pat- tern in both the raw clay and the acid-activated clay. The existence of monmorillonite particles in the 20-100 A range has been shown by Davis and coworkers (1950) and T; F. Bates (1952). As previously shown, there is a similarity between kaolins activated by calcination followed by acid treat- ment and acid-activated montmorillonites. Kaolins, in distinction to montmorillonites, consist substantialh' of relatively large, well-formed crystallites (Davis, et al., 1950). On the basis of the mechanism proposed, these materials should be relatively stable to acid attack. Kaolinite, as previously pointed out, shows only mod- erate activation on acid treatment. For the raw kaolinite having large stable crystallites, the clay structure is destroyed only when and where the alumina is removed from the lattice by complete dissolution. Thus for the kaolins, in contrast to the montmorillonites, a free silica structure is believed created, but little or no "free" alumina remains. The calcination of kaolins, however, apparently de- stroys the lattice structure in the case of halloysite and kaolinite. Such calcination probably creates a large num- ber of very small particles. The mechanism of acid acti- vation of these calcined clays is similar to that described for montmorillonite. ]\IacEwan (1951) has discussed the structural changes of kaolins occurring during the calcination at 550° C and has pointed out the possibility that a new structure has been created. Such a structure is not identifiable since the crj'stallites are so small that X-ray diffraction cannot be used to determine its presence. Even if a new structure were created, the stability of the new crystal- lites in the smaller than 100 A range would be much less than that of the original clay which consi.sts of particles in the 1,000 to 20,000 A range. Therefore, whether a new structure or an amorphous one is induced by calcination of these clays, subsequent acid treatment would bring about the same results; namely, a rapid 320 Clays and Clay Technology I Bull. 169 destruction of the small clay crj'stallites and a conse- quent deposition of "free" alumina in either case. That extremely small crystals are less stable than large crystals is strongly supported by many physical and chemical phenomena. One supporting piece of evidence is the variation of the endothermic peak (600°C) in the differential thermal analysis of different samples of kaolinite. This variation is known to depend on the de- gree of crystalline perfection. Another example is the shift in the endothermic peak to lower temperature on prolonged wet grinding of kaolinite. In another field, Dowden (1952) has pointed out that for finely divided metals the surface Tamman temperature is 0.3 of the absolute melting point, whereas it is known to be 0.5 for the bulk metal. Bevan and coworkers (1948) report a high surface electrical conductivity of oxide semi- conductors compared to their bulk conductivity and ascribed the phenomenon to different energy states in the bulk and surface of a particle. An extremely small particle, 20 to 100 A in diameter, of clay has a large percentage of its total ions ou its surface. In a crystal lattice the ions on the surface still retain the crystal pattern. However, such surface ions tend to be distorted (Cook, Oblad, and Pack, 1951) as a result of trying to accommodate the unused valences. The smaller a particle is, the larger the percentage of ions on the surface is. Consequently, the distortion or strain becomes larger and larger as the particles become smaller. If the terminal ion of the lattice is an oxygen, it will generally pick Tip an associated proton to satisfy its valence requirements. However, this does not relieve the distortion completely. The effect of the distortion probablj- extends into the lattice at least two or three ions deep, leading to differences in the coulombic forces around these outer ions, and weakening the structure in this region. Thus, in a 30 A diameter particle the first two or three oxygen ion laj'ers would constitute from 300 TEMOERATURE ' PiGUEE 11. Ammonia loss versus temperature. Open circles show silica alumina gel NHj- exchanged ; solid circles show Ash Meadows, Nevada, montmorillonite NH»- exchanged ; half-filled circles show natrolite NHi exchanged. 50 to 80 percent of the total oxygens in the crystallite, assuming it is roughly a cube or a sphere. In clays with their layer-like structure, the instability of a layer one unit cell thick (along the c axis) would be a function of the dimensions of the crj'stallite along the other two axes of the crystal (in the case of kaolin and mont- morillonite the a and b axes). If the dimensions 20-30 A are assumed for the a, b axes, then the oxygens at the terminal edges of the layer would constitute from 35-65 percent of the total oxygens in the particle. The study of the decomposition of ammonium-ex- changed zeolitic materials of widely varying average particle size shows a pronounced relationship between particle size and stability. Typical curves depicting loss of ammonia versus temperature are shown in figure 11, Curves are shown for an 87.5 percent silica, 12.5 percent alumina gel-type synthetic zeolite, a raw montmorillonite (Nevada Ash Meadows), and a sample of natrolite, all treated with ammonium chloride to replace the normally held ions — sodium, calcium, etc. — with ammonium ion. The fir.st two of these zeolites represent materials oc- curring in the manufacture of commercial cracking catalysts. The abscissa is the percentage of the total am- monia content of the sample lost by heating to the tem- perature shown on the ordinate and holding at this temperature in a flowing stream of dry, oxygen-free nitrogen until the sample reaches a constant weight. Comparison of the ciirves below 450°C shows that the acid-ammonium ' ' salt ' ' of the clay decomposes to a signifi- cant extent at considerably lower temperatures than the ammonium natrolite and higher temperatures than the synthetic zeolite. The crystallite or particle size of the natrolite sample was large (20-200 microns). The sjti- thetie zeolite sample avei-age particle size was 30 A. The clay size range of 30 to 10,000 A falls between these two extremes. The reactions are not truly reversible under the conditions of the experiment. Exposure of the samples, after loss of ammonia by heating, to an atmos- phere of ammonia restores only a small part of the ammonia lost. This irreversibility proves conclusively that the structure causing the "acid" is destroyed during the dissociation of the ammonia. The ionic structure responsible for the "acid" is essen- tialh' the same for the clay and the natrolite, in each case aluminum ions are isomorphously substituted for silicon ions in the lattice. Therefore, the clay "acid" should be equivalent in strength to the natrolite "acid" and the decomposition curves should be the same. Con- trary- to expectation, the two materials behave quite differently on decomposition. A likely explanation for the dissimilarity appears to lie in the very large differ- ence in the average pai-ticle size of the different ma- terials. This leads to a lower structural stability for at least part of the ammonium montmorillonite. Here again the smaller crystallites have lower stability. This is also supported by the fact that the synthetic zeolite with the lowest particle size decomposed at the lowest temper- ature. One other point of interest in this experiment is that, of the three materials, only the synthetic zeolite had a high catalytic activity after removal of the ammonia by heating. This would indicate that a relatively gross inter- mixing of small particles of silica and alumina is more favorable to catalytic activity than the ordered ionic Part VII] Clat Technology in the Petroleum Industry 321 ■ ■ 1 1 1 1 1 1 r 1 1 A / ,-- ->'' ,-' -.-^' o.-0<'-"° :o'°-'' 1 1 1 1 -.1 — 1 — 1 J 1 1 uf, i. :.., ../I, :.t...i -- U'alOM (■'•• ••' ■•<*■■ If (I •■•l.'*l tf ClICUM tlllC* ■ FiGt-RE 12. Density of silica alumina synthetic catalysts. arrangement of silica and alumina found in natural zeolites and clays. Nature of the Catalyst. In the foregoing discussion it was pointed out that clays in their natural state are not catah^sts and that the process of acid treatment and calcination form a new composition, at least in part, having manj- characteristics of silica-alumina gel catalysts. At this point it is pertinent to discuss some experi- mental work that has been done on gel catalysts and which has a bearing on the relationship between the chemical and physical properties and the structure of the material, (tblad. Milliken, and Mills (1951) have shown by various physical and chemical tests that the structure of sjnithetic silica-alumina gels after calcina- tion is a mixture of extremelj- small silica and alumina particles. The alumina present has the specifie gravity of gamma-alumina (fig. 12). The authors have pro- posed that aluminum and silicon share oxj-gens at the linear interfaces between the alumina and silica particles. The degree of oxygen sharing is enhanced by the method of preparation, wherein the hydroxides are coprecipitated or treated after precipitation with a strong base so that all or nearly all the alumina is in the four-coordinate state or "acid" form according to Pauling (1930). If the mixed gels are put in the am- monium "zeolite" form the material may be calcined, at which time it loses all the ammonia (viz. fig. 11), and yields a material with a high surface area (200 to 300 sq m/g) and cracking activity. It retains none of its original base-exchange capacity but, nevertheless, has a base-exchange potential. Utilizing strong inorganic or organic bases it was found that the activity of the catalyst formed bj' this method could be almost completely poisoned with as little as 0.06 milliequivalents of base per gram of catah'st. An accurate correlation was found between the amount of quinoline chemisorbed by a catalyst and its ability to crack hydrocarbons (fig. 13). On the basis of these and other data it was proposed that the alumina is present as gamma-alumina, which is largely six-coordinated aluminum, and, as shown in fig- ure 10 and table 2, contains reversible water, probabl.y as OH" groups. The low acidity. 0.06 me./g, is associated with the alumina at or near the silica-alumina interface, where the strain set up bj' the bond sharing of the two dissimilar materials causes the normally six-coordinated alumina to tend to become four-coordinated or more like silica in structure. However, the influence of silica alone was believed to be insufficient to bring about the coor- dination shift for the following reasons : first, zeolitically held ammonium ions are completely lost from the cat- alyst by heat treatment and under the experimental con- ditions are not reversiblj' adsorbed; second, under the projier conditions calcined silica-alumina catalyst can be partially reconverted to a silica-alumina "zeolite." The calcined eatalj'st on treatment with solutions of ammonium acetate at various pH, said treatment being continued to approximately equilibrium by continuously LEGEtJD O DIRECT EXCHANGE ON CATALYST ©BE-EXCMANGE FROM pH INDICATED BY DOTTED LINE 25 30 35 40 % GASOLINE - CAT -A FlGtTRE 13. Quinoline chemisnrption at 316°C as a func- tion of activity for cracking light east Texas gas oO. Circle with central dot represents SiO-. — Al-Os (Houdry type S) ; circle with dash represents SiOs — 1 percent AI2O3; rectangle with central dot represents clay catalyst (Filtrol) ; plus shows position of SiOa — MgO ; V shows SiOi — ZrOj PM OF iVMONlUM ACETJiTE SOLUTION FiGUBE 14. Base-exchange capacity of calcined synthetic silica-alumina versus pH of exchange solution. 322 Clays and Clay Technology [Bull. 169 running fresh salt solution over the sample, gives the relationship shown by the curve in figure 14. In this curve the base-exchange capacity is shown as a function of pH of the exchange solution. A sample was pretreated with ammonium acetate solution of 9.5 pH. It was found to contain 0.96 me./g of dry calcined catalyst. The squares on the graph represent the base-exchange ca- pacity of the material measured subsequently at the indicated pH. This experiment appears to refute the con- cept that acids of varying strength exist in the catalyst since it indicates a new structure has been created by the pretreatment at 9.5 pH capable of holding 0.81 me. ammonium ion per gram at 4.5 pH. This is to be com- pared with the value of 0.25 me. NH4* base-exchange capacity per gram at 4.5 pH on the original calcined material. From these data it appears that the presence of both silica and a basic ion are necessary for the alumina to shift from six coordination to four coordination at pH lower than about 10. Further, it can be postulated that for the bulk of the alumina particle the structural change taking place during either the loss or the gain of a basic ion is not readily reversible and that the new structure is relatively stable at the experimental condi- tions. This bulk quality is modified the closer the alu- minum ions in the particle are to the silica-alumina inter- face, at which point even very weak bases, such as olefins, paraffins, or other hydrocarbons, can caiise the coordina- tion shift. Structural reversibility requires that bonds be broken and reformed in the alumina structure. In the presence of water and OH", as in the ammonium acetate experi- ment just discussed, this would occur readily. In the case of catalysts at normal cracking temperatures, 450-500 °C, the shift can be presumed to occur with equal facility for the following reasons : first, the 0" and OH" ions are highly mobile in the structure, particularly at 450- 500°C; and, second, the hydroxyl or water content of the alumina in this range is reversible and is held with a wide range of energies (viz. fig. 10). At the temperature of 450°C the isotope 0'* ion can be substituted for the 0*® ions in the lattice to equilibrium in less than 30 sec- onds by treating a sample of the silica-alumina catalyst with a known amount of H^.O^^. As a base approaches the alumina-silica interface the demand for an extra oxygen or hydroxyl necessary for the shift from six to four coordination can be supplied to the site of greatest strain — i.e., those sites closest to the silica — from the lower energy sites by means of the mobility of the oxy- gen ions. Actually the transfer is probably effected more by small .shifts of individual adjacent ions rather than the long range movement of a specific 0" or OH" from a low energy site to a higher energy site. The presence of water vapor will increase the number of low-energy sites available, make possible the forma- tion of more acid sites, and thus change the activity of the catalyst. That such takes place has been found by Hansford (1947) for certain hydrocarbon reactions and by Mills and Hindin (1950) for deuterium-hydrogen ex- change between various hydrocarbons and catalyst. Application of the Reversibility Concept to Clay Structures. It is obvious that the catalyst structure is a very specific one and tliat clays must be modified by par- ticular chemical and physical treatments to approxi- mate the synthetic silica-alumina catalj'st. Moreover, the application of some of the above concepts to certain char- acteristics of clays seems possible and may offer an ex- planation for some of the anomalies in clay behavior. One of the least reproducible procedures used in clay investigations is the preparation of the so-called "hydro- gen" or "acid" clays. Various investigators using simi- lar or apparently similar techniques on samples of the same montmorillonite clay report widely varying re- sults (Kelley, 1948). It is quite likely that the variations depend on the very low stability of the four coordinate aluminum ions in the "acid" state. As Kelley has pointed out, the use of electrodialysis for cation removal results in partial destriiction of the clay and an appear- ance of free silica and alumina at the electrodes. Electro- dialysis requires very long periods of time compared with the mild acid treatment iisually employed in pre- paring a hydrogen clay. The long time allows the degra- dation of the clay to proceed to a point where it is obvious that destruction has occurred. Likewise on pro- longed mild acid treatment the destruction of the clay is equally apparent (Gedroiz, 1924). Further evidence as to the instability of the "acid" form of the clay is the previously mentioned nonreversibility of the am- monium-clay, and the drop in temperature of the endo- thermic peak from 700° to 500°C. In this last instance, the acid-treated clay has a differential thermal analysis endothermic peak of 500°C while the raw montmorillo- nite has a 700° C peak. This drop represents a very pronounced change in the energy with which the alu- minum ion layer hydrosyls are held. Since the four- coordinate aluminum ions in the silica layer "are always within one oxygen distance from the hydroxyl position, any cliange in the structure of tliis aluminum would af- fect the energy with which the hydroxyl is held. The three important phenomena described in the pre- vious paragraph can be explained on the basis of struc- tural change. An "irreversible" change takes place as a consequence of removal of the cations from the base- exchange sites. The degree to which this change takes place is a function of time and temperature and par- ticle size. The inherent instability of extremely small particles leads to rapid breakdown of tlie structure with the loss of the stabilizing basic ion. The larger particles take considerably longer times or more severe treating conditions before losing the structural characteristics of the alumina. There are a number of instances where materials con- taining four-coordinate aluminum ions in the "acid" state do not yield active catalysts. Thomas (et al., 1950) has reported that mildly acid-treated, calcined "hydro- gen" montmorillonite is not catalytieally active. It has been reported here that calcined ammonium montmoril- lonite and calcined ammonium natrolite are not active cracking catalysts. The surface areas of those montmoril- lonite samples are adequate for cracking activity. In the opinion of a number of workers in this field these materials should be catalysts, for, according to their theories, a four-coordinated aluminum with an associated proton should be present in each. The present authors think that such is far from true. They consider that an isolated four-coordinate aluminum with an associated proton is insufficient for activity. Furthermore, the exist- I'MI-t \-II Clay Teciixology ix the Petroleum Industry 323 46 44 og36 >< o 32 28 0§I6 i«l.2 30B 16 t-: Tahle 3. Effeot of hydrogen sulfide trralment at 550° C. on montniorillonitc clay cnlalyst. 09 I.I 03 Q5 07 WT % FejOj Flea Id-: l.'i. Effect u( iron on selectivity of Iloudry synthetic silica-alumina; catalytic activity test-A. ence of such a four-coordiiuite aeid is disproved by the data cited. In each of tlic three cases cited the acid is iri-cvcrsil)ly destroyed by either the acid treatment or the calciiiatiiiii. In the case of syntlietic silica-alumina cata- lyst, llie resiihial structure after calcination of the am- moniiun zeolite contains gamma-alumina, a crystal form that can shift reversibly to a limited extent. The calcined "hydrop'en" montmorillonite still retains its aluminum in tile clay -like structure and thus no gannua-alumina or aiiv oilier iiU'ta-stablc furiii of iilniiiiiia is formed to LEGEND Oxygens on Plane of Paper Oitygens Below Plane of Paper Hydroxyls on Plane of Paper Hydroxyls Below Plane of Paper Aluminum Crystal Ions 1 Hydroted Lalfice 2 Defiydrafed Lattice Figure t6. Change in alumina layer structure on 5r)0-750°C. Dehydration of montmorillonite (after Bradley). O CO) O Mole percent HiS in gaa during calcination Gaaoline volume percentage Coke weight percentage Gas weight percentage Gas gravity (air=t.O) 20 38.8 24.8 3.6 6.2 6.5 tf.3 1.55 0.66 give the desired reversibility which, the authors predi- cate, is retiuircil for catalytic cracking. The destruction of the clay lattice, already alluded to, must be sufficient to give some free alumina in order to result in an active clay catalyst. Obviously, if the fore- going hypotheses relating to the structure and action of sjmthetic catalysts are correct, the alumina formed from the clay must have some oxygen linkages remaining be- tween it and the silica. In this manner the "strained" interfacial layer is created. Effect of Oxides Other Than iSilica and Alumina. The alkali and alkaline-earth metal ions are removed to a large extent by acid treatment, and if care is taken dur- ing washing their removal is complete for all practical purposes. This removal is important since these basic ions are powerful poisons for the catalyst, acting to "neu- 5 15 25 35 45 55 GASOLINE , VOL. % CATALYST SELECTIVITY , CATALYTIC TEST - A Figure 17. 324 Clays and Clay Technology [Bull. 169 tralize" the acid craekintj function which is necessary for activity. The iron, and in montmorillonites the mag- nesia, remaining in the clay after acid treatment are generally isomorphou.sly substituted for six-coordinate alumina in the clay lattice. The presence of free iron and magnesia materially alters the nature of the eataly.st. If these ions remain in their original lattice positions, they have only minor effects on the activity of the tinislied clay catalyst. Free magnesia does not appear to be basic enough to act as a poison. Instead magnesia modifies the catalysts prepared from clay containing it by shifting the distribution of products, giving somewhat less gas for the same conversion than synthetic silica-alumina catalysts (see fig. 17), thus being more like synthetic silica-magnesia. The iron in freshly prepared or "new" clay catalysts causes a slight increase in coke formation over that given by synthetic catalysts. On continued u.se, particularly when the catalysts is exposed to petroleum crudes pos- sessing a high content of organic sulfur compounds, the iron tends to leave its lattice positions and as a "free" or non-isomorphonsly substituted iron causes large changes in the performance of the catalyst. Figure 15 illustrates tlie effect of small amounts of iron added to silica-alumina synthetic catalyst. Table 3 shows the effect to be similar when a clay catalyst containing iron is treated with hydrogen sulfide. In both instances the de- crease in gasoline j^ield and the increase in coke and gas indicate the presence of "free" iron. At present there are three expedients for avoiding or preventing the damaging effect of iron. Two are now used commercially, the third has been extensively tested on pilot plant scale. The three expedients will be de- scribed in the following paragraphs. The first expedient is represented by Filtrol Corpo- ration's "SR" catalyst. This catalyst is prepared from a halloysite with a fairly low iron content. The com- mercial results, using this eataly.st with a charge stock having a high sulfur content, appear to be satisfactory insofar as resistance to the effects of sulfur is concerned (E. J. Thomas, 1950). The second expedient used in many commercial cracking plants is specific for catalysts prepared from iron-containing montmorillonite. It was found that pre- treatment of the clay catalyst with steam before it enters the cracking zone will, to a large extent, prevent the harmful activation of the iron. This phenomenon is probably related to the reversible water content of the clay in the following manner. As pointed out earlier in this paper, a large part of the reversible water content can be associated with the residual montmoi-illonite structure. This water, as has been postulated by Grim and Bradley (1948), comes from the hydroxyl groups on the ahimina layer occurring in the hexagonal ring of silica tetrahedra. Bradley, on the strength of X-ray data, has proposed that on the loss of this OH" group the ox.ygen and aluminum ions in the lattice shift to a new arrangement shown in figure 16. The new ar- rangement requires a slightly larger oxygen-to-oxygen distance than is normal in a closely packed oxygen structure, and results in a strain that is relieved by readsorption of water and reformation of the original lattice with the hydroxyl g-roups replaced. A strained structure is more easily destroyed than an unstrained structure. The situation is similar in a way to the strain induced by the large surface to bulk ratio of extremely small particles discussed previously. Thus the reaction between the iron ions in the strained lattice and the sulfur ions in the hydrocarbons or hydrogen sulfide on the surface of the lattice is enhanced and takes place under conditions at which little or no reaction would occur with an unstrained lattice. The clay catalyst dur- ing the regeneration portion of the cracking c^'cle loses the clay lattice OIP groups since the temperature in regeneration is 620 °C and the partial pressure of water vapor seldom exceeds 0.05 atmospheres (fig. 10). If this material is introduced to the reactor with a high-sulfur- content crude oil without presteaming, the lattice is in a strained state and the iron reacts with the sulfur in the hydrocarbon charge. If the catalyst before reaching the cracking zone is presteamed at the proper tempera- ture and steam pressure to insure replacement of the hydroxyl groups, the lattice is not strained and the iron-sulfur reaction proceeds more slowly or is pre- vented entirely. The third means of solving the iron problem is to remove it selectively from the clay (Shabaker et al.). The treatment about to be described yields a catalyst of low iron content. This catalyst is known as Hoiidry Type 1. For montmorillonite the technique involves acid activation of the clay followed by treatment at 750° C with a dry inert gas, such as nitrogen, containing from 1 to 25 mole percent hydrogen sulfide. Since at these conditions the hydroxyl water is lost from the clay structure and the lattice is in the strained .state de- scribed above, the iron reacts readily with the hydrogen sulfide. The color of the clay, originally light tan or off-white, turns black on sulfidation. The sulfided clay on oxidation (500°C) becomes a bright brick red, indi- cating the presence of free iron oxide. The clay with the iron in the sulfide form is leached with cold, dilute acid to remove the iron and minor amounts of alumina. During the course of this treatment the clay structure is destroyed to an even greater degree than in normal acid activation. However, some residual lattice struc- ture can be seen on X-raj' diffraction of samples of this material. If the clay catalyst prepared in this fashion contains magnesia in the 1.0 to 6.0 percent range, it gives a distribution of cracked products on testing dif- ferent from those obtained with the original clay catalyst or synthetic silica-alumina catalyst. The relationship be- tween the gas and gasoline yields for a number of dif- ferent catalysts is shown in figure 17. The two curves for commercial clay catalyst and "modified" or iron- free Type 1 clay catalyst were obtained by testing Filtrol catalysts and Houdr.y Type 1 catalysts prepared from Filtrol elaj'. Catalysts of different activity levels were obtained by steam treatment. It can be seen that the Type 1 tends to be more like a synthetic silica- magnesia catalyst than the original clay eataly.st before modification. These data indicate that the magnesia (4.9 percent) in the clay lattice remaining after acid activation has been activated by the destruction or par- tial destruction of the lattice diiring the iron removal. When montmorillonites containing little or no magnesia are activated and modified, using the procedure de- Part VII] Clay Technology ix the Petroleum Industry 325 scribed, the values obtainecl for the gasoline and gas yields fall on the synthetic silica-alumina catalyst curve. Thus the properties of the Type 1 catalyst can be like those of a synthetic silica-alumina or tend towards those of a silica-magnesia by selection of clays of different magnesia contents, '^^'1len iron and magnesia are not present in the final catalyst prepared from montmoril- lonite-type clays, the cracking cliaractcristics are very similar to .synthetic silica-alumina gel-type catalyst. This finding further supports the proposal that the new form created when clay is activated is essentially a mi.xtnre of amorphous silica and alumina. The residual clay lat- tice of the modified clay catalyst disappears on heating to 785°C and the structure appears to be amorphous since no X-ray diffraction pattern is observed. The ther- mal stability of iron-free Type 1 gives it an important advantage over ordinary acid-activated clay catalyst which is deactivated at about 75°C lower temperature. Using the Type 1 technique, catalysts can be prepared from both montmorillonites and kaolins with excellent sulfur stability. In the case of montmorillonites con- taining magnesia, the new catalyst structure has specific cracking characteristics that are desirable in many applications. Summary. The changes taking place in a raontmoril- lonite-, kaolinite-. or halloysite-clay structure on acid treatment and calcination can be interpreted as involv- ing the creation of a structure with physical and chemi- cal characteristics and catalytic activity similar to syn- thetic silica-alumina gel catalyst. Such clay properties as high temperature deh.ydration behavior oulj^ reflect the degree to which a minor fraction of the clay struc- ture has remained unattacked. When magnesia is absent and when iron is removed sclectivel.y bj' further treat- ment, there results a catalyst having essentially identical characteristics as synthetic silica-alumina. The original structure and chemical composition of the raw clay play a large part in determining the degree and type of de- struction taking place on acid and heat treatment. In spite of the work that has been done in the field of clay catalysts, one must still resort to empirical testing of any given clay in order to establish its suitability as a raw material for catalyst manufacture. Consideration of experimental data presented for re- versible hydration at high temperatures, loss of ammonia upon heating the ammonium zeolite form, and titration experiments by base leads to a concept of reversible coordination shift from six to four of the aluminum ion. This concept has been utilized to explain behavior of clays in electrodialysis. Acknowledgment. Permission by the Houdry Process Corporation to publish this work is acknowledged with appreciation. Contributions of other members of this laboratory are also acknowledged, particularly those of E. B. Cornelius, ,T. J. Donovan, and S. G. Hindin. DISCUSSION M. W. Tamele: Milliken's idea of considering clay to be a reservoir of silica and alumina is interesting and ma.v resolve some of the discrepancies noted in the past. Various ideas have been recorded in the litera- ture as to how much RiOa must be removed to achieve optimum activity, and the material in this paper reconciles the situation. Will Millikcn explain the presence of pores of 25 A radius (fig. 6). I cannot see how such pores can be formed by the removal of the inner layer of the montmorillouite lattice, but rather it would indicate to me that another phase is being introduced. T. H. Milliken: The 25 to 26 A radius pore size was calculated assuming the pores to be capillaries and essentially round. This is obviously not the case and for interlayer spacing, different methods of calcula- tion should be used. We have no correlation of this spacing with X-ray diffraction powder spectra. Although we cannot picU up spacings of this magnitude with our instrument ; second-, third-, and fourth-order reflections should show up. There is a good chance of a new phase being introduced. These longer spacings may be due to the presence of particles of silica between the originally swollen layers preventing them from coming together again. Do MacEwan's techniques show these? D. M. C. MacEwan: Spacings up to 120 A have been measured in our laboratories (by K. Xorrish),and I do not think there is any doubt that spacing of the reported pore size in the activated material could be ob- served if they exist as definite spacings. I have examined activated montmorillonite with filtered radiation and no such long spacings were seen. On the other hand there was evidence to indicate that there was some material between the layers which prevented their collapsing completely but the order of magnitude of the pores that would result would be about 10 A. Those particular observations might have missed a spacing of about 30 A had it been present. Isaac Barshad: Does not the acid treatment l)reak up the lattice structure by removing units throughout the particle and giving it a sieve-like structure? This ma.v explain anomalies noted in work on the total exchange capacity of activated materials. T. H. Milliken: The possible sieve-structure in the silica sheets might be ex- plained on the basis of the fourfold Al positions being points of acid attack. SELECTED REFERENCES Alexander, J., 1947, Am. Petroleum Inst. Proc, v. 27, p. .51. Ardern, D. B., Dart, J. C and Lassiat, R. C, 1951, Catalytic cracking in fixed- and moving-bed processes, in Progress in pe- troleum technology, pp. 13-29, Washington, D. C, American Chemical Society. Bevan, D. J. M., Shelton. J. P., and Anderson, J. S., 1948, Properties of some simple oxides and spinels at high temperatures : Chem. Soc. (London) Jour., v. 1948, pp. 1729-1741. Burghardt, O., 1931, Activated bleaching clays : Ind. and Eng. Chemistry, v. 23, pp. 800-802. Cook, M. A., Pack, D. H., and Oblad, A. G., 1951, Structural model of low pressure "physical" adsorption : Jour. Chem. Physics, V. 19, pp. 307-376. David.son. R. C. 1947, Petroleum Refiner, v. 26, p. 663. Davis. D. W., et al., 1950, Electron micrographs of reference day minerals: Am. Petroleum Inst., Proj. 49, Prelim. Rept. 6, 17 pp.. New York, Columbia University. Dowden, D. A., 1952, Catalytic activity of nickel, theoretical aspects : Ind. and Eng. Chemistry, v. 44, pp. 977-985. Drake. L. C, 1949, Pore-size distribution in porous materials, application of high-pressure mercury porosimeter to cracking catalysts : Ind. and Eng. Chemistry, v. 41, pp. 780-785. Escard. J., M6ring, J., and Perrin-Bonnet, I., 1950, Activation de la montmorillonite : Jour. Chimie Physique, tome 47, pp. 234- 237. Gedroiz, K. K., 1924, Pouvy, Nenasyshchennye osno¥ani.vami. lletodika o<5redeleniya v pochvakh vodorodnogoiona, nakhod.yas- chegosya v pogloshchennom sostoyanii. Potrebnost' pochvy v iz- vesti, kak neytralizatore nenasyshchennosti : Zhurnal Opytnoy Agronomii, Imeni p.s. Kossoviua. Otdepl. Opiginal'nye Stat'i. [Russisches Journal fiir experimentelle Landwirtschaft, Adschnitt 1, Originalarbeiten, v. 22, pp. 3-28. (In Russian with summary in German) ]. Greensfelder, B. S., 1951, The mechanism of catalytic cracking, in Progress in petroleum technology, pp. 3-12, Washington, D. C, American Chemical Society. 326 Clays and Clay Technology [Bull. 169 Greenall. Alexander, 1948, Montmorillonite cracking catalyst, X-ray diffraction : Ind. and Eng. Chemistry, v. 40, pp. 2148-2151. (Ireenall. Alexander, W-t'.K Montmorillonite cracking catalyst, demonstration of presence of hydrogen ion in heated Filtrol clay catalysts : Ind. and Eng. Chemistry, v. 41, pp. 1485-1480. Grim, R. E., and Bradley, W. F.. 1948, Rehydration and dehy- dration of the clay minerals : Am. Mineralogist, v. 33, pp. 50-59. Hagner, A. F., 1939, Adsorptive clays of the Texas Gulf coast : Am. Mineralogist, v. 24, pp. 67-108. Hansford, R. C, 1947, A mechanism of catalytic cracking : Ind. and Eng. Chemistry, v. 39, pp. 849-852. Hofmanu, U., and Endell, K., 1935, Mitteilung'uer die alkivie- rung der rohen bleicherde : Angewandte Chemie, v. 48. pp. 187-191. Holmes, J., and Mills, G. A., 1951, Aging of a bentonite crack- ing catalyst in air or steam: Jour. Phys. Colloid Chemistry, v. 55, pp. 1.302-1320. Houdry, E., Burt, W. F., Pew, A. E. Jr., and Peters, W. A. Jr., 1938. Catalytic processing by the Houdry process : Nat. Petroleum News, V. 30, no. 48, pp. R5'70-R580. Houdry, E., Bnrt, W. F., Pew, A. E. Jr., and Peters. W. A. Jr., 1938a, Catalytic processing of petroleum hydrocarbons by the Houdry process: Refiner and Natural Gasoline Manufacturer, v. 17, pp'. 574.582, 619. Kelley, W. P., 1948, Cation exchange in .soils : Am. Chem. Soc. Mon. Ser., No. 109, 144 pp., New York, Reinhold Publishing Corp. Lopez-Gonzalez, J. de D., and Deitz, V. R., 19.52, Surface changes in an original and actiyated bentonite : Nat. Bur. Standards Jour. Research, v. 48, pp. 325-333. MacEwan, D. M. C, 1951. The montmorillonite minerals (mont- morillonoids) , in Brindley, G. W., Editor, X-ray identification and crystal structures of clay minerals, p. 120, London, Mineralog. Soc, Clay Minerals Group. Mills, G. A., U. S. Patent 2,485,626 to Houdry Process Corpora- tion (kaolin catalysts) . Mills, G. A., and Hindin, S. G., 19.50, Chemical characteriza- tion of catalysts. II. Oxygen exchange between water and cracking agents : Am. Chem. Soc. Jour., v. 72, pp. 5.549-5.554. Mills, G. A., Holmes, J., and Cornelius, E. B., 19.50, Acid activa- tion of some bentonite clays : Jour. Phys. Colloid Chemistry, v. 54, pp. 1170-1185. Murphree, E. V., 1951, Fluid catalytic cracking process, in Progress in petroleum technology, pp. .30-38, Washington, D. C, American Chemical Society. Nutting, P. G., 19.33, The bleaching clays: U. S. Geol. Survey, Circ. 3, 51 pp. Nutting, P. G.. 1935. Technical basis of bleaching clay industry : Am. Assoc. Petroleum (Jcologists Bull., v. 19. pp. 1043-i052. Nutting, P. G., 1937, A study of bleach clay solubility : Franklin Inst. Jour., v. 224, pp. 3.39.362. Nutting, P. G., 1943, Adsorbent elavs : U. S. Geol. Survey Bull. 928-C, p. 127. Oblad, A. G., Milliken, T. H., and Mills, G. A. 1951, Chemical characteristics and structure of cracking catalysts, in Frankenburg, W. G., et al.. Advances in catalysis and related subjects, v. 3, pp. 199-247, New York, Academic Press, Inc. Oulton. T. D., 1948, The pre size-surface area distribution of a cracking catalyst : .Tour. Phys. Colloid Chemistry, v. .52, pp. 1296- 1314. Pauling. Linus, 1930, The structure of some sodium and calcium aluminosilicates : Nat. Acad. Sci. Proc, v. 16, pp. 4.53-4.59. Richardson, H. M., 1951, Phase changes which occur on heating kaolin clays, in Brindley, G. W., Editor, X-ray identification and crystal structures of clay minerals, pp. 76-85, London, Mineralog. Soc, Clay Minerals Group. Ries, H. E., 1952, Structure and sintering properties of crack- ing catalysts and related materials, in Frankenburg, W. G., et al.. Advances in catalysis and related subjects, v. 4, pp. 87-149, New York, Academic Press, Inc. Ritter, H. L., and Drake, L. C, 1949, Pore-size distribution in porous materials: Ind. and Eng. Chemistry, Anal. Ed. (Anal. Chemistry), v. 17, pp. 782-786. Schroter, G. A., and Campbell, I., 1940, Geological features of some deposits of bleaching clay : Min. Technology, v. 4, pp. 1-31. Shabaker, H. A., Mills, G. A., and Denison, R. C, U. S. Patents 2,466,046 to 2,466,0.52 and 2..561,422 to Houdry Process Corpora- tion. Suehiro, Yoshiyuki. 1949, Preparation of activated clays and property of its tabulett in use for catalytic cracking : Chem. Soc. Japan Jour., Ind. Chem. Section, y. 52, p. 16-17. Thomas, C. L., Hickey. J., and Stecker, G., 1950, Chemistry of clay cracking catalysts: Ind. and Eng. Chemistry, v. 42, pp. 866-871. Thomas, E. J., 1950, Fluid catalytic cracking of high-sulfur stock with natural catalysts : Oil and Gas .Jour., v. 48, \io. 46, pp. 221, 224, 228. ^Valthall, J. H., Miller, P., and Striplin, M. M., 1945, Develop- ment of a sulfuric acid process for production of alumina from clay ; Am. Inst. Chem. Eng. Trans., v. 41, pp. 53-140. AVAILABLE DIVISION OF MINES PUBLICATIONS ON CLAY AND RELATED SUBJECTS Dictricli, ^Yal(lenlal•. F., 1928, The day resources and ceraniir industry of Ciilitornia : California Div. Mines and Jliuiurr ]>ull. 99. .'i8.'? pp. Tndiviilual i/lay deposits throug-hout the state are deserihed in detail and the physical, eheniieal, and ceramic chai-acteristics of the clay are listed. Price !|;2.{)(). Sutiierhmd. -1. C. 1935, Geolojiica! irivesti.i;ati()ns of the clays of Riverside and Orange Counties, southern Califor- nia: California Div. Mines Kept. :>1. I'art A, jip. 51-87. Price 60('. Johnson. F. 'P., and Hickcr. Spangler, 1949, lone-Carbondale clays, Amador County, California: California Jour. Mines and (ieology v. 45, no. 3 (July 1949), pp. 491-498. Gives the results of a ll. S. Bureau of Mines drillinfi profii-am foi- hipli-alumina clays, including' chemical analyses of the core samples. Price 75(*. Jalins. K. 11.. and l.ance, J. P., 1950, Geology of the San Dieguitu pyropliyllite area, San Diego County, Califiu'uia: ('alif(irnia Div. Mines, Special Kept. 4, 32 pp.. Price 50^. i'age, B. M., 1951, Tale deposits of steatite grade, Inyo County. California: Calii'm-nia Div. Mines. S])eeial Kept. 8, 35 pp. Price 85^. Bull. 154, 1951, Geolog:ic Guidebook of the San Francisco Bay Counties, containing "Clay aiul the ceramic industry of the S. F. Bay Counties," bv M. D. Turner and "Ceramic education and industry in the S. F. Bay area," by J. A. Pask. Price$2.50. Wright, L. A., 1952, Geology