5 
 
 \ 
 
 STATE OF ILLINOIS 
 
 DWIGHT H. GREEN, Governor 
 
 DEPARTMENT OP REGISTRATION AND EDUCATION 
 
 FRANK G. THOMPSON, Director 
 
 DIVISION OF THE 
 
 STATE GEOLOGICAL SURVEY 
 
 M. M. LEIGHTON, Chief 
 URBANA 
 
 REPORT OF INVESTIGATIONS— NO. 117 
 
 SOUTHERN ILLINOIS NOVACULITE AND NOVACULITE 
 GRAVEL FOR MAKING SILICA REFRACTORIES 
 
 C. W. Parmelee and C. G. Harman 
 
 Department of Ceramic Engineering 
 
 University of Illinois 
 
 ILLINOIS GEOLOGICAL 
 SURVEY LIBRARY 
 
 SEP 21 1946 
 
 PRINTED BY AUTHORITY OF THE STATE OF ILLINOIS 
 
 URBANA, ILLINOIS 
 1946 
 
ILLINOIS STATE GEOLOGICAL SURVEY 
 
 3 3051 00005 7343 
 
STATE OF ILLINOIS 
 
 DWIGHT H. GREEN, Governor 
 
 DEPARTMENT OF REGISTRATION AND EDUCATION 
 
 FRANK G. THLOMPSON, Director 
 
 DIVISION OF THE 
 
 STATE GEOLOGICAL SURVEY 
 
 M. M. LEIGHTON, Chief 
 URBANA 
 
 REPORT OF INVESTIGATIONS— NO. 117 
 
 SOUTHERN ILLINOIS NOVACULITE AND NOVACULITE 
 GRAVEL FOR MAKING SILICA REFRACTORIES 
 
 C. W. Parmelee and C. G. Harman 
 
 Department of Ceramic Engineering 
 
 University of Illinois 
 
 PRINTED BY AUTHORITY OF THE STATE OF ILLINOIS 
 
 URBANA, ILLINOIS 
 1946 
 
ORGANIZATION 
 
 STATE OF ILLINOIS 
 
 HON. DWIGHT H. GREEN, Governor 
 
 DEPARTMENT OF REGISTRATION AND EDUCATION 
 
 HON. FRANK G. THOMPSON, Director 
 
 BOARD OF NATURAL RESOURCES AND CONSERVATION 
 
 HON. FRANK G. THOMPSON, Chairman 
 NORMAN L. BOWEN, D.Sc, LL.D., Geology 
 ROGER ADAMS, Ph.D., D.Sc, Chemistry 
 LOUIS R. HOWSON, C.E., Engineering 
 CARL G. HARTMAN, Ph.D., Biology 
 EZRA JACOB KRAUS, Ph.D., D.Sc, Forestry 
 ARTHUR CUTTS WILLARD, D.Engr., L.L.D. 
 President of the University of Illinois 
 
 GEOLOGICAL SURVEY DIVISION 
 
 M. M. LEIGHTON, Chief 
 
 (14239—2500—3-46) 
 

 SCIENTIFIC AND TECHNICAL STAFF OF THE 
 STATE GEOLOGICAL SURVEY DIVISION 
 
 100 Natural Resources Building, Urbana 
 
 M. M. LEIGHTON, Ph.D. 
 Enid Townley, M.S., Assistant to the Chief 
 Velda A. Millard, Junior Asst. to the Chief 
 
 Chief 
 Helen E. McMorris, Secretary to the Chief 
 Effie Hetishee, B.S., Geological Assistant 
 
 GEOLOGICAL RESOURCES 
 
 Ralph E. Grim, Ph.D., Petrographer and 
 Principal Geologist in Charge 
 
 Coal 
 
 G. H. Cady, Ph.D., Senior Geologist and Head 
 R. J. Helfinstine, M.S., Mech. Engineer 
 Charles C. Boley, M.S., Assoc. Mining Eng. 
 Robert M. Kosanke, M.A., Asst. Geologist 
 Robert W. Ellingwood, B.S., Asst. Geologist 
 Jack A. Simon, B.A., Asst. Geologist 
 Arnold Eddings, B.A., Asst. Geologist 
 Raymond Siever, B.S., Research Assistant (on leave) 
 John A. Harrison, B.S., Research Assistant (on leave) 
 Mary E. Barnes, B.S., Research Assistant 
 Margaret Parker, B.S., Research Assistant 
 
 Oil and Gas 
 
 A. H. Bell, Ph.D., Geologist and Head 
 Frederick Squires, B.S., Petroleum Engineer 
 David H. Swann, Ph.D., Assoc. Geologist 
 Virginia Kline, Ph.D., Assoc. Geologist 
 Paul G. Luckhardt, M.S., Asst. Geologist 
 Wayne F. Meents, Asst. Geologist 
 James S. Yolton, M.S., Asst. Geologist 
 Sue R. Anderson. B.S., Research Assistant 
 
 Industrial Minerals 
 
 J. E. Lamar, B.S., Geologist and Head 
 Robert M. Grogan, Ph.D., Assoc. Geologist 
 Robert R. Reynolds, M.S., Asst. Geologist 
 Mary R. Hill, A.B., Research Assistant 
 
 Clay Resources and Clay Mineral Technology 
 
 Ralph E. Grim, Ph.D., Petrographer and Head 
 Henry M. Putman, B.A.Sc, Asst. Geologist 
 William A. White, B.S., Research Assistant 
 
 Groundwater Geology and Geophysical Exploration 
 
 Carl A. Bays, Ph.D., Geologist and Engineer, and 
 
 Head 
 Robert R. Storm, A.B. 
 Arnold C. Mason, B.S. 
 
 (on leave) 
 
 Merlyn B. Buhle, M.S., Asst. Geologist 
 M. W. Pullen, Jr., M.S., Asst. Geologist 
 Charles G. Johnson, A.B., Asst. Geologist (on leave) 
 Margaret J. Castle, Asst. Geologic Draftsman 
 Robert N. M. Urash, B.S., Research Assistant 
 George H. Davis, Research Assistant 
 
 Engineering Geology and Topographic Mapping 
 
 George E. Ekblaw, Ph.D., Geologist and Head 
 Richard F. Fisher, M.S., Asst. Geologist 
 
 Areal Geology and Paleontology 
 
 H. B. Willman, Ph.D., Geologist and Head 
 Chalmer L. Cooper, Ph.D., Geologist 
 C. Leland Horberg, Ph.D., Assoc. Geologist 
 Heinz A. Lowenstam, Ph.D., Assoc. Geologist 
 
 Subsurface Geology 
 
 L. E. Workman, M.S., Geologist and Head 
 Paul Herbert, Jr., B.S., Asst. Geologist 
 Marvin P. Meyer, B.S., Asst. Geologist 
 Elizabeth Pretzer, M.A., Research Assistant 
 
 Assoc. Geologist 
 Assoc. Geologist 
 
 Physics 
 
 R. J. Piersol, Ph.D., Physicist 
 
 Mineral Resources Records 
 
 Vivian Gordon, Head 
 
 Ruth E. Warden, B.S., Research Assistant 
 
 GEOCHEMISTRY 
 
 Frank H. Reed, Ph.D., Chief Chemist 
 Carol J. Adams, B.S., Research Assistant 
 
 Chemist and Head 
 
 Coal 
 
 G. R. Yohe, Ph.D. 
 
 Industrial Minerals 
 
 J. S. Machin, Ph.D., Chemist and Head 
 Tin Boo Yee, M.S., Research Assistant 
 
 Fluorspar 
 
 G. C. Finger, Ph.D., Chemist and Head 
 Oren F. Williams, B.Engr., Asst. Chemist 
 
 Chemical Engineering 
 
 H. W. Jackman, M.S.E., Chemical Engineer and Head 
 P. W. Henline, M.S., Assoc. Chemical Engineer 
 James C. McCullough, Research Associate 
 James H. Hanes, B.S., Research Assistant (on leave) 
 Leroy S. Miller, B.S., Research Assistant (on leave) 
 
 X-ray and Spectrography 
 
 W. F. Bradley, Ph.D., Chemist and Head 
 
 Analytical Chemistry 
 
 O. W. Rees, Ph.D., Chemist and Head 
 L. D. McVicker, B.S., Chemist 
 Howard S. Clark, A.B., Assoc. Chemist 
 Cameron, D. Lewis, M.A., Asst. Chemist 
 William T. Abel, B.A., Research Assistant 
 John C. Gogley, B.S., Research Assistant 
 Phyllis K. Brown, B.A., Research Assistant 
 
 MINERAL ECONOMICS 
 
 W. H. Voskuil, Ph.D., Mineral Economist 
 Douglas F. Stevens, M.E., Research Associate 
 Nina Hamrick, A.B., Research Assistant 
 Ethel M. King, Research Assistant 
 
 LIBRARY 
 
 Regina Lewis, B.A., B.L.S., Librarian 
 Ruby D. Frison, Technical Assistant 
 
 PUBLICATIONS 
 
 Dorothy E. Rose, B.S., Technical Editor 
 Meredith M. Calkins, Geologic Draftsman 
 Beulah F. Hopper, B.F.A., Asst. Geologic 
 
 Draftsman 
 Willis L. Busch, Principal Technical Assistant 
 Leslie D. Vaughan, Asst. Photographer 
 
 Consultants: Ceramics, Cullen W. Parmelee, M.S., D.Sc, and Ralph K. Hursh, B.S., University of Illinois 
 
 Mechanical Engineering, Seichi Konzo, M.S., University of Illinois 
 Topographic Mapping in Cooperation with the United States Geological Survey. 
 
 May 10, 1946 
 
CONTENTS 
 
 PAGE 
 
 Abstract 7 
 
 Part I— PROPERTIES OF NOVACULITE AND NOVACULITE GRAVEL 
 
 Introduction 9 
 
 Resources and samples 9 
 
 Previous work on silica refractories 9 
 
 Raw materials for silica refractories 10 
 
 Program of work 10 
 
 Acknowledgments 1 1 
 
 Crushing characteristics 1 1 
 
 Screen analysis of novaculite reduced in laboratory smooth rolls 11 
 
 Screen analysis of novaculite reduced in a wet pan 13 
 
 Experiments with packing crushed novaculite 13 
 
 Packing characteristics of three grain-size mixtures 13 
 
 Methods 13 
 
 Results 13 
 
 Packing of many sizes under various pressures 14 
 
 Inversions of silica 20 
 
 Rates of inversion of novaculite 21 
 
 Experimental details 21 
 
 Method 21 
 
 Specimen 21 
 
 Expansion measurements 23 
 
 Measurements on specimens after heating 23 
 
 Results 23 
 
 Method of representing results 23 
 
 Comparison of data from interferometer and differential apparatus 25 
 
 Influence of grain size 27 
 
 Effect of various fluxes 29 
 
 Comparison of volume changes in specimens with and without bond 29 
 
 Particle grading used 30 
 
 Influence of the amount of bond 31 
 
 Calcium phosphate 32 
 
 Sodium tungstate 32 
 
 Calcium borate 34 
 
 Borax 34 
 
 Iron oxide 36 
 
 Mixed fluxes 36 
 
 Expansion-temperature curves of unfired novaculite bodies 39 
 
 Preliminary small-scale tests on fabricated samples 40 
 
 Preparation of test samples 40 
 
 Grain size 40 
 
 Mixing of the bond with silica 41 
 
 Forming the test pieces 41 
 
 Bonding agents used 41 
 
 Firing the samples 41 
 
 Testing the fired samples 41 
 
 Compressive strength tests 41 
 
 Density measurements 41 
 
 Results of firing tests 43 
 
 Summary 44 
 
Part II— PREPARATION AND PROPERTIES OF SILICA BRICK MADE FROM 
 ILLINOIS NOVACULITE AND QUARTZITIC SANDSTONE 
 
 PAGE 
 
 Introduction 45 
 
 Preparation of material for testing 45 
 
 Novaculite gravel 45 
 
 Novaculite at the pit 45 
 
 Purification of novaculite gravel by washing 45 
 
 Sampling for tests 45 
 
 Proper grain-size distribution 45 
 
 Preparation of the material 45 
 
 Additional packing experiments 46 
 
 Method 46 
 
 Grain-sizes studied 46 
 
 Results of packing tests 46 
 
 Characteristics of unfired novaculite briquets 47 
 
 Effect of amount of tempering water on the modulus of rupture 47 
 
 Grain-size gradings 47 
 
 Testing procedure 47 
 
 Effect of particle grading on the modulus of rupture 47 
 
 Testing procedure 47 
 
 Grain sizes 47 
 
 Results of modulus of rupture tests 47 
 
 Rate of inversion of novaculite and other silica materials 49 
 
 Method of study 49 
 
 Results 49 
 
 Experiments with standard 9-inch shapes 50 
 
 General method 50 
 
 Details of procedure 50 
 
 Preparation of the bonding agents 50 
 
 Lime bonds 50 
 
 Special bond 50 
 
 Preparation of specimens 50 
 
 Grinding and mixing 50 
 
 Making the brick 51 
 
 Properties of the test brick 51 
 
 Modulus of rupture 51 
 
 Deformation under load at high temperatures 53 
 
 Porosity changes during firing 53 
 
 Microstructure 54 
 
 Thermal expansion 55 
 
 Conclusions 55 
 
 TABLES 
 
 TABLE PAGE 
 
 1. Screen analysis of novaculite reduced in smooth rolls 11 
 
 2. Screen analysis of mine-run novaculite gravel ground dry in a wet pan 13 
 
 3. Packing of different percentages of three grain sizes 14 
 
 4. Grain-size distributions used in pressure packing tests 16 
 
 5. Results of pressure packing 18 
 
 6. Densities and percentage porosities of samples 1 to 5 and 10 27 
 
 7. Percentage porosities and densities of expansion samples 29 
 
 8. Compresive strengths of novaculite briquets 134 by 134 by V/% inches 41 
 
 9. True densities of fired briquets 43 
 
 10, Screen analysis of novaculite gravel, "as received" 45 
 
TABLE PAGE 
 
 11. Screen analysis of "as received" novaculite gravel after washing through 20-mesh and 28- 
 
 mesh screens 45 
 
 12. Screen analysis of the ground washed mine-run novaculite gravel divided into three fractions 47 
 
 13. Grain-size distribution of the minimum void mixture determined by packing experiments. . 49 
 
 14. Modulus of rupture of unfired silica brick 49 
 
 15. Kind of raw material, grain sizes, and particle grading of experimental brick 51 
 
 16. Data on 9-inch novaculite brick fired in commercial kiln 52 
 
 ILLUSTRATIONS 
 
 FIGURE PAGE 
 
 1. Low-temperature inversions of silica: C, cristobalite; T, tridymite; Q, quartz 10 
 
 2. Rate of grinding of novaculite; 125-pound charge in 5-foot wet pan. Curves represent per- 
 
 centage of material on the sieve indicated 12 
 
 3. Packing of novaculite grains of three sizes 15 
 
 4. Graph illustrating the type of grain-size variation for packing experiments 17 
 
 5. Packing of crushed mine-run novaculite by pressure 19 
 
 6. Packing of crushed washed novaculite by pressure 20 
 
 7. Thermal expansion apparatus 22 
 
 8. Mold for pressing samples 23 
 
 9. Graph illustrating the effect of the apparatus constant on the expansion-time curve of 
 
 novaculite 24 
 
 10. Comparison of data from interferometer and differential apparatus 25 
 
 11. Effect of grain sizes on the rate of inversion of novaculite 26 
 
 12. Effect of the use of bond on the rate of inversion of novaculite 28 
 
 13. Effect of the use of tripoli on the rate of inversion of novaculite 30 
 
 14. Effect of varying the amount of lime bond on the rate of inversion of novaculite 31 
 
 15. Effect of calcium phosphate on the rate of inversion of novaculite 32 
 
 16. Effect of use of sodium tungstate on the rate of inversion of novaculite 33 
 
 17. Effect of use of calcium borate on the rate of inversion of novaculite 34 
 
 18. Effect of the use of borax on the rate of inversion of novaculite 35 
 
 19. Effect of the use of ferric oxide on the rate of inversion of novaculite 36 
 
 20. Effect of the use of mixed fluxes on the rate of inversion of novaculite 37 
 
 21 . Effect of other fluxes on the rate of inversion of novaculite 38 
 
 22. Graphs of expansion-temperature changes during firing of novaculite bodies 39 
 
 23. Distribution of particle size on grinding mine-run novaculite. 140 pounds ground 40 min- 
 
 utes in wet pan (dry); tailings (on 6-mesh) 8.7 percent 40 
 
 24. Heating schedules for firing novaculite brick samples 42 
 
 25. Packing of many grain sizes 46 
 
 26. Effect of amount of tempering water used on the modulus of rupture of unfired novaculite 
 
 briquets, CaO constant 48 
 
 27. Firing schedule and rates of inversion 48 
 
 28. Porosity of unfired brick versus modulus of rupture of fired brick 52 
 
 29. Thermal expansion, percent length versus temperature • 54 
 
SOUTHERN ILLINOIS NOVACULITE AND NOVACULITE 
 GRAVEL FOR MAKING SILICA REFRACTORIES 
 
 BY 
 C. VV. Parmelee and C. G. Harman 
 
 Department of Ceramic Engineering 
 University of Illinois 
 
 ABSTRACT 
 
 Extensive deposits of massive chert 
 known as novaculite and of novaculite 
 gravel are found in Alexander and Union 
 counties in southern Illinois. The novacu- 
 lite is a cryptocrystalline material, mine- 
 run samples of which contain approximately 
 97 percent silica; selected samples have a 
 slightly higher silica content. The novacu 
 lite gravel contains a small amount of clay. 
 Other highly siliceous materials found in 
 the general area are ganister, a naturally 
 granular material, and tripoli or silica, a 
 material consisting mostly of minute parti- 
 cles. 
 
 The ganister has been used successfully 
 in the manufacture of refractories, but 
 efforts to use the novaculite for this pur- 
 pose have encountered difficulties that have 
 raised doubts as to its suitability. The 
 object of this investigation was to determine 
 whether or not the novaculite and the novac- 
 ulite gravel have value for the manufacture 
 of silica bricks and shapes for lining fur- 
 naces that are operated at high temperatures. 
 The very large tonnages of silica refracto- 
 ries manufactured in Illinois from ganister 
 shipped in from other states indicates the 
 desirability of finding in Illinois a supply 
 of an equally useful material. 
 
 Because of their importance in the manu- 
 facture and use of silica refractories, the fol- 
 lowing properties of novaculite and novac- 
 ulite gravel and of silica refractories made 
 from them were given particular attention : 
 (1) Nature and rate of inversion; and (2) 
 mechanical strength in relation to particle 
 size, packing, and fluxes. 
 
 Both the novaculite and the novaculite 
 gravel, which contained a small amount of 
 clay, were separately crushed and screened 
 
 to grade them according to size. Trial mix- 
 tures of different proportions of each size 
 grade were made. The minimum amount 
 of voids obtained in the crushed novaculite 
 subjected to a packing pressure of 7000 
 pounds per square inch was 8.2 percent. 
 Under similar conditions the novaculite 
 gravel tests gave a minimum of 2.5 percent 
 voids. Washed crushed novaculite mixed in 
 the proportions : 35 percent coarse fraction, 
 15 percent medium fraction, and 50 percent 
 fine fraction, when moistened with 4 percent 
 water and packed under a pressure of 3000 
 pounds per square inch, gave a product hav- 
 ing a porosity of 19 percent. 
 
 In the study of the influence of various 
 amounts of several kinds of fluxes on the 
 southern Illinois materials, calcium oxide, 
 which is the flux usually employed in the 
 manufacture of silica brick, was used as a 
 standard of reference. Thirteen other 
 oxides, compounds, and mixtures were tried 
 as a flux and accelerator. Sodium tung- 
 state proved to be the most vigorous accel- 
 erator and borax was a close second. In- 
 teresting results were observed with iron 
 oxide, both alone and in a combination with 
 other fluxes. The calcium oxide was some- 
 what less effective than these three. 
 
 The work on the effect of particle size of 
 novaculite and novaculite gravel on inver- 
 sion in the presence of lime showed that the 
 finer sizes of the material reached a stable 
 state of inversion more quickly than the 
 coarser material, and that the coarse mate- 
 rial showed greater swelling and cracking. 
 
 Cristobalite w T as the product of the inver- 
 sion. A comparison of the inversion rates 
 of the novaculite, tripoli, and ganister from 
 Baraboo, Wisconsin (a quartzite largely 
 
 [7] 
 
SOUTHERN ILLINOIS NOVACULITE 
 
 used in the manufacture of silica brick) 
 proved the Illinois materials to be identical 
 in conduct and to have much faster rates of 
 inversion than the Baraboo quartzite and 
 to have a significant development of tridy- 
 mite at a lower temperature, which is a 
 distinct advantage. 
 
 The foregoing data were applied to the 
 preparation and forming of standard size 
 9-inch brick from the crushed sized novacu- 
 lite. The results confirmed the earlier 
 findings that the grain size determines the 
 percentage of voids, hence the strength. 
 The strength is also influenced by the 
 amount and kind of bond and the manner 
 of firing. Illinois novaculite — crushed, 
 graded, and bonded with lime — gave a 
 product correspondingly close in its charac- 
 ter and properties to the commercial silica 
 
 brick. These brick were fired in the kilns 
 of the Department of Ceramic Engineer- 
 ing at the University of Illinois and in a 
 commercial kiln. 
 
 It was further found that better brick 
 can be made by mixing with the Illinois 
 novaculite a certain proportion of the 
 extremely fine-grained silica or tripoli 
 obtainable in the same area of southern 
 Illinois as the novaculite and novaculite 
 gravel. 
 
 In summary, the results of this investiga- 
 tion indicate that Illinois novaculite, when 
 crushed, properly graded, bonded, and fired 
 furnishes a silica refractory similar in 
 appearance and properties to the commer- 
 cial silica brick made from quartzite gan- 
 ister. 
 
PART 1— PROPERTIES OF NOVACULITE AND 
 NOVACULITE GRAVEL 
 
 Introduction 
 resources and samples 
 
 A variety of high-silica materials, among 
 them novaculite, novaculite gravel, ganis- 
 ter, and tripoli, also known as "amorphous" 
 silica, are found in Alexander and Union 
 counties of extreme southern Illinois. 
 Although the gravel is used in some quan- 
 tity as road metal, the other materials find 
 only limited commercial use. The first two 
 materials are chert, the term novaculite 
 being commonly applied in southern Illinois 
 to deposits that consist of solid layers of 
 chert, and the term novaculite gravel to a 
 gravel that consists of angular chert frag- 
 ments with which is associated usually a 
 small amount of red or yellow clay and 
 varying amounts of tripoli. The ganister 
 of southern Illinois is a highly siliceous 
 material which has naturally a loose tex- 
 ture somewhat similar to that of cornmeal. 
 The tripoli or silica consists principally 
 of a microcrystalline form of silica so small 
 that the material has been referred to as 
 amorphous or cryptocrystalline. The natural 
 deposits have varying degrees of coherency 
 and the material is ground and sized for use. 
 
 There are large deposits of each of these 
 materials, although the size and number of 
 ganister deposits is believed to be less than 
 that of novaculite and novaculite gravel. 
 The novaculite gravel deposits are known 
 to be as much as 120 feet thick, the novac- 
 ulite deposits are known to be at least 50 
 feet thick, and the ganister deposits have 
 been worked ordinarily to a thickness of 10 
 to 15 feet but the total thickness may be 
 considerably greater. The ganister ordi- 
 narily is mined underground whereas the 
 novaculite and novaculite gravel are 
 obtained from open pits. 1 
 
 The samples used in this study were all 
 obtained from fresh exposures in deposits 
 
 x For a discussion of the distribution and character of the 
 cherty formations of Alexander and Union counties, see 
 Illinois Geol. Survey Rept. Inv. No. 70. 
 
 which were being operated or had been 
 operated recently. Because of the amount 
 of work involved in testing, only a limited 
 number of samples could be included in this 
 investigation. The samples examined do 
 not represent necessarily all the variation? 
 existing in the novaculite, novaculite gravel, 
 and ganister deposits in southern Illinois. 
 However, it is believed that the results of 
 the tests do provide a basic picture of the 
 nature and possible use of these materials 
 in the making of silica brick, and that from 
 the data the effects of variations in the 
 character of the materials can be evaluated. 
 
 The samples studied were as follows: 
 
 Novaculite — from a 7-foot exposure of 
 white novaculite with a few scattered light 
 yellow spots, in the NE. \ NW. \ sec. 36, 
 T. 14 S., R. 1 W., near Tamms, Illinois. 
 
 Novaculite gravel — pit-run gravel from 
 a 120-foot exposure of reddish gravel in the 
 NE. \ NW. i sec. 36, T. 14 S., R. 1 W., 
 near Tamms, Illinois. 
 
 Washed novaculite gravel — same as sam- 
 ple above but washed in the laboratory to 
 remove clay and particles finer than 28 
 mesh. 
 
 Illinois ganister — mine-run material from 
 a 12-foot exposure in a mine at the W. \ 
 NW. i NE. i sec. 18, T. 14 S., R. 1 W., 
 near Elco, Illinois. 
 
 Ganister — standard commercial material 
 already ground, from Baraboo, Wisconsin. 
 Used for purposes of comparison. 
 
 Quartzitic sandstone — pit-run material 
 from a 40-foot exposure in a pit in the SW. 
 J sec. 12, T. 14 S., R. 6 E., near Tansil, 
 Illinois. 
 
 Tripoli — a commercial product of the 
 Illinois Mineral Company at Elco, Illinois. 
 
 PREVIOUS WORK ON SILICA REFRACTORIES 
 
 Many reports on silica and its use in man- 
 ufacturing refractories have been published. 
 The earliest extensive study of the tempo- 
 rary and permanent volume changes which 
 silica undergoes upon heating (see fig. 1) 
 
 [9] 
 
10 
 
 PROPERTIES OF NOVACULITE 
 
 was made by Fenner. 2 The significance to 
 the industry of the fundamental research by 
 LeChatelier, Sosman, McDowell, and oth- 
 ers appears in any serious discussion of the 
 material and its uses at high temperatures. 
 The manner of the changes has been dis- 
 closed but the rates of change are not so 
 well known, although we recognize that 
 they are dependent upon the specific raw 
 material used, its grain size, and the nature 
 and the amount of impurities or additions 
 of chemical agents used. 
 
 500 1000 
 
 TEMPERATURE- DEGREES C 
 
 Fig. 1. — Low temperature inversions of silica: 
 C, cristobalite ; T, tridymite ; Q, quartz. 
 
 RAW MATERIALS FOR SILICA REFRACTORIES 
 
 In the choice of raw materials for silica 
 refractories, the main points considered are 
 chemical purity, size of grain obtainable, 
 and the magnitude of increase in volume 
 upon heating. 
 
 For the manufacture of silica brick, the 
 natural rock should contain at least 97 per- 
 cent of silica and should not yield too fine 
 a powder upon crushing. According to many 
 writers, the order of merit of the natural 
 forms of silica is given as chalcedony, old 
 quartzites, and vein quartz. Quartz schists, 
 sandstone, and sand are considered unsuit- 
 able, the first two on account of their struc- 
 ture and the presence of many impurities in 
 the form of inclusions, and the latter two on 
 account of their variability in composition 
 and their excessive fineness after grinding. 
 
 The Illinois novaculite, which was the 
 material under investigation, is a type of 
 rock composed of a form of silica some- 
 times described as chalcedony. It has a 
 cryptocrystalline structure and the individ- 
 ual crystals have mean diameters of 0.05 to 
 0.10 microns. 3 It is possible that a small 
 amount of opaline is present. Minute 
 quartz crystals are admixed with the chal- 
 cedony. It is well known to technologists 
 familiar with the silica minerals that the 
 cryptocrystalline variety, when heated at 
 high temperature, is readily transformed, 
 that is inverted, into the more useful cris- 
 tobalite or tridymite. The Findling's 
 "quartzite" of continental Europe, which 
 has been a very important raw material for 
 the silica brick industry of that area, is a 
 cryptocrystalline silica. Ross 4 found that 
 chert from Indiana, when strongly heated, 
 showed density changes due to inversion 
 which were regarded by him as favorable 
 indications of the value of the material for 
 silica brick. 
 
 Program of Work 
 
 The objective of this investigation was 
 to discover whether Illinois novaculite could 
 be used satisfactorily for the manufacture 
 of silica brick and similar refractories. As 
 the chemical and petrographic characteris- 
 tics indicated that it should be suitable, and 
 possibly superior to the siliceous rock ordi- 
 narily employed, it was necessary to study 
 certain physical properties such as the crush- 
 ing characteristics, the packing of the 
 crushed rock, and the rates and complete- 
 ness of the inversions which accompanied the 
 use of several bonding agents or fluxes. 
 
 The crushing characteristics are impor- 
 tant because they influence the mechanical 
 strength and porosity of the finished brick 
 or other forms. Silica refractories are unlike 
 the clay refractories in several particulars. 
 Clay undergoes progressive vitrification and 
 therefore the porosity decreases with in- 
 creased firing temperature, but silica brick 
 does not show such a change until it has 
 reached and passed its maximum safe-firing 
 
 2 Fenner, C. N., Stability relations of the silica minerals: 
 Am. Jour. Sci., vol. 36, p. 339, 1913. 
 
 3 The micron is 10 -4 millimeter or 0.000,001 m. 
 
 4 Ross, Donald W., Silica refractories: U. S. Bur. Stand- 
 ards Tech. Paper No. 116, 1919. 
 
CRUSHING CHARACTERISTICS 
 
 11 
 
 temperature, then it collapses rather quickly. 
 The maximum mechanical strength and 
 minimum porosity of a refractory depend 
 largely upon the relative proportions of the 
 several sizes and shapes of the crushed 
 rock used. 
 
 The crushed rock was screened into 
 "coarse", "medium", and "fine" fractions, 
 and experiments were made to determine 
 what proportions of the various sizes or 
 fractions should be used to obtain a mixture 
 that would yield brick having the maximum 
 mechanical strength and minimum porosity. 
 The "packing" experiments are described 
 in detail on pages 13-20. 
 
 The bonding agent also referred to as 
 flux, is more important for its reaction in 
 promoting the inversion of the chert to the 
 most useful type of crystalline silica rather 
 than for its actual bonding effect, and the 
 amount is strictly limited to the minimum 
 percentage required, usually two or three 
 percent. Larger amounts react to diminish 
 the refractoriness of the product. Clay, 
 which is a most useful bonding material, 
 cannot be used for this reason. The few 
 agents which are practicable as accelerators 
 are chosen because of their low cost and 
 efficiency. 
 
 The research having defined the condi- 
 tions of crushing to yield the desired grain 
 sizes, the most suitable proportions of these 
 sizes to furnish the minimum voids when 
 
 the grains were packed together, the 
 mechanical strengths of such mixtures, and 
 the effects of various bonds or accelerators, 
 the final phase of the work consisted of 
 making samples of standard 9-inch brick, 
 firing them, and comparing their physical 
 properties with those of commercial silica 
 brick. 
 
 ACKNOWLEDGMENTS 
 
 The investigation was done under the 
 supervision of Professor Cullen W. Parme- 
 lee of the Department of Ceramic Engineer- 
 ing of the University of Illinois, with the 
 assistance of Dr. Cameron G. Harman of 
 the Survey staff. 
 
 Crushing Characteristics 
 
 screen analysis of novaculite reduced 
 in laboratory smooth rolls 
 
 The rock was reduced to 1]4 inches and 
 finer in a jaw crusher and then put through 
 small laboratory smooth rolls, adjusted as 
 follows : 
 
 (g-2) — rolls about 1/8 inch apart. 
 (g-3) — rolls touching. 
 (g-4) — rolls about 3/32 inch apart. 
 (g-5) — rolls about 1/16 inch apart. 
 (g-6) — midway between g-4 and g-5. 
 
 The screen analysis of these grinds is 
 shown in table 1. 
 
 Table 1. — Screen Analysis of Novaculite Reduced in Smooth Rolls 
 
 
 Grind Number 
 
 Sieve 
 mesh a 
 
 g- 
 
 -2 
 
 g- 
 
 -3 
 
 g-4 
 
 g- 
 
 -5 
 
 g-6 
 
 
 % 
 
 Cum. % 
 
 % 
 
 Cum. % 
 
 % 
 
 Cum. % 
 
 % 
 
 Cum. % 
 
 % 
 
 Cum. % 
 
 Plus 8 . . . . 
 
 40.6 
 
 40.6 
 
 4.4 
 
 4.4 
 
 37.9 
 
 37.9 
 
 
 
 
 
 24.9 
 
 29.4 
 
 8 to 10. . 
 
 8.0 
 
 48.6 
 
 2.5 
 
 6.9 
 
 9.7 
 
 47.6 
 
 4.5 
 
 4.5 
 
 13.0 
 
 42.4 
 
 10 to 14. . 
 
 18.2 
 
 66.8 
 
 16.8 
 
 23.7 
 
 20.9 
 
 68.5 
 
 17.3 
 
 21.8 
 
 21.5 
 
 63.9 
 
 14 to 28.. 
 
 13.6 
 
 80.4 
 
 33.1 
 
 56.8 
 
 14.4 
 
 82.9 
 
 36.7 
 
 58.5 
 
 16.1 
 
 80.0 
 
 28 to 48.. 
 
 6.5 
 
 86.9 
 
 16.7 
 
 73.5 
 
 6.3 
 
 89.2 
 
 11.9 
 
 70.4 
 
 6.8 
 
 86.8 
 
 48 to 100. 
 
 3.0 
 
 89.9 
 
 8.9 
 
 82.4 
 
 3.3 
 
 92.5 
 
 10.4 
 
 80.8 
 
 3.6 
 
 90.4 
 
 100 to 200. 
 
 2.8 
 
 92.7 
 
 5.6 
 
 88.0 
 
 2.3 
 
 94.7 
 
 6.7 
 
 87.5 
 
 4.0 
 
 94.4 
 
 Minus 200. 
 
 9.3 
 
 100.0 
 
 12.0 
 
 100.0 
 
 5.2 
 
 99.9 
 
 12.6 
 
 100.1 
 
 6.0 
 
 100.4 
 
 a All screening analyses were made with Tyler Standard Screen Scale Sieves. 
 
12 
 
 PROPERTIES OF NOVACULITE 
 
 15 20 25 30 
 
 TIME OF GRINDING - MINUTES 
 
 Fig. 2. — Rate of grinding of novaculite ; 125-lb. charge in 5-foot wet pan. Curves represent per- 
 centage of material on the sieve indicated. 
 
PACKING CHARACTERISTICS 
 
 13 
 
 Table 2. — Screen Analysis of Mine-run Novaculite Gravel Ground Dry in a Wet Pan 
 
 
 
 
 
 
 Grindir 
 
 ig Time 
 
 
 
 
 
 Sieve 
 mesh 
 
 10 Min. 
 
 15 Min. 
 
 20 Min. 
 
 27 Min. 
 
 35 Min. 
 
 
 % 
 
 Cum. % 
 
 % 
 
 Cum. % 
 
 % 
 
 Cum. % 
 
 % 
 
 Cum. % 
 
 % 
 
 Cum. % 
 
 Plus 6 . . . . 
 
 6 to 8... 
 
 8 to 10.. 
 
 10 to 14.. 
 
 14 to 28.. 
 
 28 to 48.. 
 
 48 to 100. 
 
 100 to 200. 
 
 Minus 200. 
 
 39.6 
 10.5 
 26.6 
 
 5.5 
 5.0 
 2.7 
 1.8 
 1.8 
 
 6.5 
 
 39.6 
 50.1 
 
 76.7 
 82.2 
 87.2 
 89.9 
 91.7 
 93.5 
 
 100.0 
 
 35.1 
 9.4 
 3.9 
 10.7 
 10.3 
 6.4 
 4.7 
 8.7 
 
 10.5 
 
 35.1 
 44.5 
 48.7 
 59.1 
 69.4 
 75.8 
 80.5 
 89.2 
 
 99.7 
 
 26.5 
 
 10.2 
 
 3.7 
 
 11.7 
 
 11.7 
 
 7.7 
 
 6.2 
 
 111 
 
 11.3 
 
 26.5 
 36.7 
 40.4 
 52.1 
 68.3 
 71.5 
 77.7 
 88.8 
 
 100.1 
 
 19.0 
 11.8 
 
 4.3 
 
 11.4 
 
 12.8 
 
 8.8 
 
 7.3 
 
 16.0 
 
 8.1 
 
 19.0 
 30.8 
 35.6 
 47.0 
 59.8 
 68.6 
 75.9 
 91.9 
 
 100.0 
 
 15.7 
 
 10.1 
 
 5.3 
 
 11.3 
 
 12.9 
 9.9 
 8.5 
 
 16.6 
 
 9.8 
 
 15.7 
 25.8 
 31.1 
 
 42.4 
 55.3 
 65.2 
 73.7 
 90.3 
 
 100.1 
 
 SCREEN ANALYSIS OF NOVACULITE 
 REDUCED IN A WET PAN 
 
 Novaculite was broken in a large jaw 
 crusher, then 125-pound charges were put 
 into a 5-foot wet pan and ground dry. 
 Samples were taken for screen analysis at 
 the end of 10 minutes, 15 minutes, 20 min- 
 utes, 27 minutes, and 35 minutes. The 
 results of this test are shown in table 2 and 
 figure 2. 
 
 EXPERIMENTS WITH PACKING CRUSHED 
 NOVACULITE 
 
 The particle-size distribution has a very 
 marked influence upon the physical proper- 
 ties of the brick. In general, the denser the 
 packing, the stronger the final product. For 
 maximum resistance to spalling, the balance 
 between elasticity and mechanical strength 
 must be correct, and these properties are 
 both largely dependent upon the particle- 
 size grading. It was therefore important 
 to learn the manner in which the packed 
 density varied with grain-size distribution. 
 
 Packing Characteristics of Three 
 Grain-Size Mixtures 
 
 method 
 
 A preliminary study of the packing char- 
 acteristics of crushed novaculite was deter- 
 mined by a method similar to that used by 
 Westman and Hugill. 5 Briefly it consisted 
 of a capsule for holding a known weight of 
 the sized grains whose volume could be cal- 
 
 culated at any time during the test from the 
 known internal dimensions of the container 
 and the measured height of the sample there- 
 in. This capsule containing the sample was 
 placed in an apparatus where it could be sub- 
 jected to any desired number of impacts of a 
 controlled magnitude. Measurements of the 
 volume of the grains in the capsule were 
 taken after each series of 500 impacts, and 
 the minimum volume and closest packing 
 were considered to be reached when there 
 was no further reduction on repeated shak- 
 ing. 
 
 RESULTS 
 
 The results of packing these fractions are 
 shown in table 3 and figure 3 where the 
 values plotted are the ratios of the bulk 
 volumes to the true volumes. The lines on 
 the diagram represent mixtures which pack 
 with the same volume of voids. From the 
 plotted data it may be seen that the sample 
 containing 50 percent coarse fraction, 25 
 percent medium fraction and 25 percent fine 
 fraction represents the densest packing mix- 
 ture of this system. 
 
 In silica brick, particle-size distributions 
 must be made which yield the most satisfac- 
 tory balance of properties. This may be 
 done only by trial. These grain-size studies 
 were made in order to simplify practice for 
 studies where control of grain size was nec- 
 essary. 
 
 5 Westman, A. E. R., and Hugill, H. R., The packing of 
 particles: Jour. Am. Ceramic Soc, vol. 13. pp. 767-779, 
 1930. 
 
14 
 
 PROPERTIES OF NOVACULITE 
 
 Tabi 
 
 -Packing of Different Precentages of Three Grain Size: 
 
 Grain size — Screen mesh 
 
 
 
 
 (percent) 
 
 
 Packed Vol. 
 
 Percent 
 
 
 True Vol. 
 
 
 
 
 
 voids 
 
 Through 
 
 Through 
 
 Through 
 
 
 
 8 on 10 
 
 28 on 48 
 
 200 
 
 
 
 50 
 
 
 50 
 
 1.43 
 
 37.5 
 
 25 
 
 
 75 
 
 1.49 
 
 32.8 
 
 75 
 
 
 25 
 
 1.40 
 
 28.2 
 
 
 25 
 
 75 
 
 1.54 
 
 35.2 
 
 •• 
 
 75 
 
 25 
 
 1.52 
 
 33.3 
 
 
 50 
 
 50 
 
 1.42 
 
 29.5 
 
 75 
 
 25 
 
 
 1.62 
 
 38.2 
 
 
 50 
 
 50 
 
 1.60 
 
 37.5 
 
 
 25 
 
 75 
 
 1.63 
 
 38.9 
 
 100 
 
 
 
 1.77 
 
 43.7 
 
 
 100 
 
 
 1.77 
 
 43.8 
 
 
 
 100 
 
 1.63 
 
 38.8 
 
 70 
 
 15 
 
 15 
 
 1.44 
 
 30.5 
 
 50 
 
 25 
 
 25 
 
 1.29 
 
 22.5 
 
 40 
 
 30 
 
 30 
 
 1.34 
 
 25.6 
 
 60 
 
 20 
 
 20 
 
 1.39 
 
 27.9 
 
 50 
 
 20 
 
 30 
 
 1.32 
 
 24.3 
 
 50 
 
 15 
 
 35 
 
 1.33 
 
 24.8 
 
 50 
 
 30 
 
 20 
 
 1.33 
 
 24.7 
 
 45 
 
 30 
 
 25 
 
 1.31 
 
 23.9 
 
 45 
 
 25 
 
 30 
 
 1.30 
 
 23.1 
 
 55 
 
 20 
 
 25 
 
 1.33 
 
 24.8 
 
 55 
 
 25 
 
 20 
 
 1.34 
 
 25.7 
 
 Packing of Many Sizes Under 
 Various Pressures 
 
 Data obtained from packing of these 
 sizes gives information for proportioning 
 grains to yield well packed bodies, but in 
 practice such "gap gradings"" are not desir- 
 able. Instead "continuous" gradings are 
 preferred. In addition the brick are formed 
 by applying force, in some manner, to the 
 mass being molded, which may influence the 
 choice of grading. The packing of three 
 sizes just described is useful by way of fur- 
 nishing a starting point for a study of more 
 detailed distributions. 
 
 To make these tests, samples of well 
 mixed mine-run novaculite gravel and 
 selected novaculite w r ere placed in a mold 
 and pressed. The mold was the same one 
 used in preparing thermal expansion speci- 
 mens, and is described below. (See fig. 8.) 
 The volumes of the specimens were deter- 
 mined by measuring the height of the pis- 
 ton with a 1/1000 inch gage after applica- 
 tion of pressure. The grain-size gradings 
 studied are shown in table 4. 
 
 The percentage of voids in the samples 
 after packing are tabulated in table 5. These 
 samples were packed with 2\/i cc. water per 
 25-gram sample. 
 
PACKING OF MANY SIZES 
 
 15 
 
 FINE (1.63; 
 
 COARSE - THROUGH 8 ON 10 
 MESH 
 
 MEDIUM - THROUGH 28 ON 
 
 48 MESH 
 FINE - THROUGH 200 
 
 MESH 
 
 FIGURES ON CHART ARE 
 THE RATIO OF BULK 
 VOLUME TO TRUE 
 VOLUME . 
 
 .3Z\/\30\/L34\ 
 
 COARSE (1.77) 
 
 MEDIUM (1.77 ) 
 
 Fig. 3. — Packing of novaculite grains of three sizes. 
 
 P-l is the mix that packed to maximum 
 density in the three-sieve size mixture as 
 shown in figure 3. Another test was made 
 in which the novaculite finer than 200-mesh 
 was replaced by an equal quantity of com- 
 mercial tripoli. This tripoli is much finer 
 in average grain size than the novaculite, 
 averaging 200-mesh or finer in size. The 
 results are given under P-2 in table 5. At 
 
 1000 pounds per square inch, P-l packed 
 to 28.4 percent voids, while at the same 
 pressure, P-2 packed to 18.6 percent voids. 
 At pressures of 2000 to 2500 pounds per 
 square inch there was little difference in the 
 degree to which these two mixtures packed. 
 Further increase of pressure increased the 
 density of P-2 at a greater rate than that 
 of P-l. 
 
16 
 
 PROPERTIES OF NOVACULITE 
 
 Table 4. — Grain-size Distributions Used in Pressure Packing Tests 
 In weight percent 
 
 Made from Mine-run Novaculite Gravel 
 
 Screen mesh 
 
 
 Weight percent 
 
 by mix number 
 
 
 P-3 
 
 P-4 
 
 P-5 
 
 P-6 
 
 P-7 
 
 P-9 
 
 6 to 8 
 
 23 
 7 
 7 
 6 
 7 
 
 13 
 13 
 14 
 
 10 
 
 23 
 7 
 7 
 6 
 7 
 
 13 
 13 
 14 
 
 io 
 
 15 
 
 7 
 7 
 6 
 7 
 
 13 
 13 
 14 
 
 10 
 
 8 
 
 7 
 7 
 7 
 6 
 7 
 
 13 
 13 
 14 
 
 10 
 16 
 
 "l 
 
 7 
 
 6 
 
 7 
 
 13 
 
 13 
 
 14 
 
 10 
 
 23 
 
 27.6 
 
 8 to 10 
 
 
 10 to 14 
 
 10.6 
 
 14 to 20 
 
 11.8 
 
 20 to 28 
 
 9 
 
 28 to 48 
 
 10.5 
 
 48 to 100 
 
 5.5 
 
 100 to 200 
 
 3.5 
 
 Minus 200 (novaculite) 
 
 7.5 
 
 Minus 200 (tripoli) 
 
 15.0 
 
 
 
 
 P-10 
 
 P-ll 
 
 P-12 
 
 P-14 
 
 P-15 
 
 P-16 
 
 6 to 8 
 
 41.8 
 10.6 
 
 7.1 
 14.1 
 
 7.1 
 
 3.8 
 2.3 
 
 5.1 
 17.4 
 
 4.4 
 
 2.5 
 
 16.8 
 
 33.1 
 
 16.1 
 
 8.9 
 5.6 
 
 12.0 
 
 3.9 
 
 2.2 
 14.8 
 21.2 
 
 16~3 
 8.7 
 5.5 
 
 11.7 
 
 7.8 
 
 13.1 
 
 8.8 
 17.9 
 16.2 
 
 14.1 
 
 7.8- 
 5.4 
 
 11.6 
 
 5.3 
 
 30.2 
 
 19.8 
 14.8 
 
 7.0 
 
 3.3 
 3.1 
 
 7.4 
 14.4 
 
 28.1 
 
 8 to 10 
 
 
 10 to 14 
 
 14 to 20 
 
 20 to 48 
 
 28 to 48 
 
 48 to 100 
 
 100 to 200 
 
 21.9 
 15.1 
 
 6.5 
 
 3.4 
 
 2.3 
 
 Minus 200 (novaculite) 
 
 5.5 
 
 Minus 200 (tripoli) 
 
 17.2 
 
 
 
 Made from Selected Novaculite 
 
 
 
 
 
 P-l 
 
 P-2 
 
 P-17 
 
 P-19 
 
 P-20 
 
 P-21 
 
 6 to 8 
 
 50 
 
 25 
 
 25 
 
 50 
 
 25 
 
 25 
 
 23 
 7 
 7 
 6 
 7 
 
 13 
 13 
 14 
 
 10 
 
 15 
 
 7 
 
 7 
 
 6 
 
 7 
 
 13 
 
 13 
 
 14 
 
 10 
 
 8 
 
 7 
 
 7 
 
 7 
 
 6 
 
 7 
 
 13 
 
 13 
 
 14 
 
 10 
 16 
 
 
 8 to 10 
 
 7 
 
 10 to 14 
 
 7 
 
 14 to 20 
 
 20 to 28 
 
 28 to 48 
 
 48 to 100 
 
 100 to 200 
 
 6 
 
 7 
 
 13 
 
 13 
 
 14 
 
 Minus 200 (novaculite) 
 
 10 
 
 Minus 200 (tripoli) 
 
 23 
 
 
 
 The mixtures shown in table 4 as P-9 to 
 P-16 are of rocks crushed in the rolls as 
 described above. P-ll (tables 4 and 5) is 
 the same as g-3 in table 1. P-10 was pro- 
 duced from this by adding enough coarse 
 to bring this fraction to about 50 percent 
 and by adding sufficient tripoli to bring the 
 fines to 25 percent. The result was a 
 decided decrease in the volume of the voids. 
 At 1000 pounds per square inch, P-ll had 
 
 24.8 percent voids, and at the same pressure 
 the adjusted mix, P-10, had 16.0 percent 
 voids. Other mixes, P-9 to P-16, showed 
 the variations that might be expected from 
 a variety of different grain-size mixtures. 
 
 Novaculite may be ground in a wet pan 
 so that the cumulative percent of particles 
 coarser than any screen is a linear function 
 of the logarithm of the width of the screen 
 opening (fig. 4). In order to test the 
 
PACKING OF MANY SIZES 
 
 17 
 
 120 
 
 Fig. 4. — Graph illustrating the type of grain-size variation for packing experiments. 
 
 effect of the finest and coarsest fractions of 
 grain sizes for this type of distribution, 
 mixes P-3, P-4, P-5, P-6, and P-7 were 
 made. The distribution of the sizes used 
 are shown in table 4 and figure 4. These 
 mixes were made from mine-run novacu- 
 lite gravel which contains about 2 percent 
 clay. P-3 may be considered as the base 
 mix, in which the 6- to 8-mesh material was 
 replaced by Illinois tripoli to produce P-5, 
 P-6 and P-7. The results of packing this 
 material are shown in table 5 and figure 5. 
 These data showed that for this type of par- 
 
 ticle-size distribution, the conditions for 
 minimum voids at pressures of less than 4000 
 pounds per square inch was about 19 per- 
 cent through 200-mesh. This also implied 
 about 14 percent 6- to 8-mesh material. 
 
 The series P-17, P-19, P-20, and P-21 
 were made of clear white novaculite which 
 was free from clay. The purpose of this 
 series was to determine the effect of clay 
 upon the wet packing of grains. This series 
 had the same screen analysis as the P-3 to 
 P-7 series. The data on this series of 
 mixes has been plotted in figure 6. A com- 
 
18 
 
 PROPERTIES OF NOVACULITE 
 
 
 Ih 
 
 a 
 
 c 
 
 X 
 
 CN 
 
 Oh 
 
 ^o om» ih «vo co co n ^- 1 os 
 
 
 oovinno^oor^'vO-* 
 
 
 tL 
 
 oor^cOCN— iOOsOsl^>sO 
 
 
 Oh 
 
 ^O'O'OMrt^O^H 
 
 
 to i-H 00 1^. \D >o to CO rt O 
 
 
 
 cs 
 
 O'HOOCNXin^cNOI^ 
 
 COON^"tCOCNrHOK) 
 
 (MCSrHrtHHriHH 
 
 
 OS 
 
 ^CJOoovOwrtco-HO 
 
 CNtNCNMrtHHrHHrt 
 
 
 
 os^OTfco~HOsr^*>o-'fcN 
 
 rSCNtNCNCNrHrtrHrHrt 
 
 GO 
 
 
 oL 
 
 osi-Hi^cNOsosoor-Hr-~r^ 
 
 >^1 •* CN rH o\ 00 r^ ^0 ■+ CO 
 
 _3 
 
 "o 
 
 > 
 
 
 Osr-CN\0-<f •Os^OvDN 
 
 "3 
 O 
 
 **< 
 
 Oh 
 
 ot^r^ONooco't'too 
 
 c 
 
 <u 
 
 
 cu 
 
 Oh 
 1 
 
 CO 
 
 "o 
 
 CN 
 
 Oh 
 
 CO VO CN CN tN CN CJ\ O >JO ^ 
 
 't-tONN^i'CNr- 1 OS OO 
 
 Oh 
 
 OO • CN CN uo OS vO CN •<*< OO 
 ■<*< •OO^O'^CN.-hOS^OIO 
 
 > 
 
 O 
 
 1 
 
 O • CN ■* O ^O ^ CN O VO 
 vo ■Tt'CN^Ooor^vO'o 
 
 
 os 
 
 OS •.— iVDCOi-HOOVOOCO 
 
 
 1— 1 ^UTtCOcN-HOC^ 
 CN ■ t— 1 r-t »— 1 t-i i-H 1— 1 1— 1 
 
 
 Oh 
 
 CO CN ■* x t^ ^ ■* rH CO ^ 
 
 
 ^osr^^-^cocN^HOoo 
 
 C^,—!,— 1,-Ht— 1,-H,— 1,— 1,— 1 
 
 
 ol 
 
 Os^DOOCOOOOs^OvOvO 
 
 r-u->cocN— iooor^^Dio 
 
 
 Oh 
 
 O^oovoosososr^osr— 
 iocNOoor^\ow->TticocN 
 
 
 Oh 
 
 LOCXSOSO-fCNvOTfcor^ 
 
 ooN»o'*cNr- iosoor^vo 
 
 
 CO 
 
 Oh 
 
 CN CO CO •* CN r^- ■* C^ On CN 
 OO v O-^CNi-HOSOOW~>TfTfri 
 
 
 c 
 <u 
 
 (LI C 
 
 c i 
 
 z K 
 
 3 . 
 
 CO _Q 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 0000000000 
 0000000000 
 
 O^O^O^OOOO 
 i-h 1— 1 CN CN CO CO •* i^ vO t^ 
 
PACKING OF MANY SIZES 
 
 19 
 
 4 8 12 16 20 24 
 
 PERCENT- 200 MESH TRIPOLI REPLACING 6 TO 8 MESH NOVACULITE 
 Fig. 5. — Packing of crushed mine-run novaculite by pressure. 
 
 parison of figures 5 and 6 shows two out- 
 standing differences: The clay-free grains 
 did not pack nearly so closely. P-3 packed 
 to 18.2 percent voids at 1000 pounds per 
 square inch pressure, whereas P-17 at the 
 same pressure packed to 29.5 percent voids. 
 The minimum voids for the clay-free grains 
 was 8.2 percent at 7000 pounds per square 
 inch, whereas the mine-run mixture had a 
 
 minimum of 2.5 percent voids at the same 
 pressure. The clay-free grains also required 
 more fines to bring about the minimum void 
 composition. The clay-free grains had a 
 minimum at about 13 percent of the tripoli, 
 compared with 9 percent for the mine-run 
 samples. It appears that the clay acts as a 
 lubricating medium during packing. 
 
20 
 
 PROPERTIES OF NOVACULITE 
 
 4 
 
 PERCENT -200 MESH TRIPOLI 
 
 12 16 20 24 
 
 REPLACING 6 TO 8 MESH NOVACULITE 
 
 Fig. 6. — Packing of crushed washed novaculite by pressure. 
 
 Inversions of Silica 
 
 Reference has already been made to the 
 fact that silica exists in several crystalline 
 forms, each of which has its characteristic 
 temperature range of stability. When the 
 temperature to which any particular form 
 is subjected is higher or lower than the sta- 
 bility range of a particular form, it inverts, 
 or changes into the crystalline form peculiar 
 to the new temperature conditions. These 
 changes are of practical importance because 
 each inversion is accompanied by a change 
 
 Inversions 
 
 *Low to high-quartz 
 
 High-quartz to upper high-tridymite. . . 
 High-tridymite to high-cristobalite 
 
 of volume, expansion or contraction, and 
 some are rapid and of considerable magni- 
 tude, others are sluggish and slight. A 
 familiar example of the nature and effect 
 of a rapid inversion of a form of crystalline 
 silica is the shattering of a quartz pebble 
 when strongly heated or the lack of strength 
 in very sandy clays after they have been 
 fired. Numerous other illustrations might 
 be cited. The silica inversions which are 
 of interest in this study are: 
 
 Temperature Rate 
 
 573 d= 1°C(1063°F.) Rapid 
 
 870 ± 10° C (1598° F.) Sluggish 
 
 1470 ± 10° C (2678° F.) Sluggish 
 
 *Low-quartz is known as alpha-quartz, high-quartz as beta-quartz; upper-high-tridymite is also beta-tridymite; high- 
 cristobalite is also beta-cristobalite. 
 
INVERSION OF NOVACULITE 
 
 21 
 
 The temperatures and rate at which the 
 inversions take place in commercial opera- 
 tions are influenced by the fineness of the 
 silica, the presence and the nature of impuri- 
 ties in the material, and the kind of flux or 
 bonding agent employed, and the duration 
 of the heating. The densities of the prin- 
 cipal forms of silica are : 
 
 Low-quartz 2 . 65 
 
 High-tridymite 2 . 26 
 
 High-cristobalite 2.32 
 
 When the low-quartz inverts to high- 
 tridymite there is a large volume increase ; 
 the inversion of the high-tridymite to the 
 high-cristobalite is accompanied by an addi- 
 tional increase of volume. Quartz at 1200° 
 C. has a density of about 2.55, while cris- 
 tobalite at the same temperature has a 
 density of about 2.2. An inversion then at, 
 say 1200° C, will cause an increase in vol- 
 ume of about 15.9 percent, but the change 
 from cristobalite to tridymite involves a 
 volume change of less than 1 percent. 
 
 These are the volume changes that occur 
 during heating. During cooling, if there 
 is any untransformed quartz at 537° C, it 
 will contract the same amount that it 
 expanded. Over the range of about 200° 
 C. to 250° C, cristobalite changes, in cool- 
 ing, from a density of 2.225 to 2.29, a vol- 
 ume change of about 3 percent. Tridymite 
 has two low-temperature inversion points, 
 117° C, and 163° C. The volume changes 
 here are a fraction of one percent. 
 
 Silica brick expands permanently during 
 firing, due to the change of quartz to cris- 
 tobalite and tridymite, and the volume 
 change may amount to as much as 15 per- 
 cent but rarely reaches that figure. There- 
 fore, as would be expected, the specific grav- 
 ities of silica brick are good criteria by 
 which to judge whether the brick has had 
 proper heat treatment. As only 70 to 80 
 percent of the quartz is converted to cris- 
 tobalite and 5 to 10 percent to tridymite in 
 a commercial burn, a brick with a specific 
 gravity of less than 2.40 is considered satis- 
 factory. Much of the ware on the market 
 has specific gravities of 2.30 to 2.35 and is 
 decidedly superior to the ware of higher 
 specific gravity. 
 
 Silica refractories are useful because of 
 their rigidity at temperatures at which ordi- 
 nary fireclay brick soften, and because they 
 expand slightly when heated instead of con- 
 tracting. 
 
 Rates of Inversion of Novaculite 
 experimental details 
 
 Method. — The study of the inversion 
 rates was made by observing linear expan- 
 sions at constant temperature. For this pur- 
 pose an apparatus described by Chesters and 
 Parmelee 5 was used. For the convenience 
 of the reader a brief description of the 
 apparatus is given here. 
 
 A fixed member, a 1.2-inch diameter 
 alundum tube, OT, carried the entire 
 expansion system. Slots were cut on either 
 side of this tube so that the alundum shelf, 
 B, could be inserted into the tube to support 
 the specimen S. The specimen was sepa- 
 rated from the shelf and from the moving 
 push rod, PR, by platinum washers. The 
 gage holder, GH, was a heavy brass head 
 which was clamped to the outside tube OT, 
 by screws, S. The push rod, PR, was 
 capped with a stainless steel cap, C, which 
 had been machined to a sliding fit in GH. 
 A 1/10,000 inch Ames Dial, G, was attached 
 to GH, as shown. A counter-weight of 300 
 grams was used to take the load off the 
 tube. The specimens were made so that 
 the thermocouple, T, could be inserted at 
 their center. The furnace was a molybde- 
 num wire wound furnace which has been 
 described by Chesters and Parmelee. 6 
 
 Specimen. — Thirty grams of novaculite 
 of proper grain size was weighed out, then 
 mixed with 2yz cc. of water and the desired 
 amount of flux. This was then mixed for 
 about 15 minutes with the fingers and placed 
 in a mold (fig. 8). A uniform pressure of 
 2500 pounds per sq. in. was used in all 
 tests. The test piece was 0.9 inch in diam- 
 eter and ll/2 inches in height. It had a 
 hole in its center to accommodate a ther- 
 
 5 Chesters, J. H. and Parmelee, C. W., The measurement 
 of reaction rates at high temperatures: Jour. Am. Ceramic 
 Soc. vol. 17, No. 3, pp. 50-59, 1934. 
 
 6 Chesters, J. H. and Parmelee, C. W., The burning of 
 magnesite brick: Trans. Ceramic Soc. (Eng.), vol. 32, 
 No. 7, pp. 349-70, 1933. 
 
22 
 
 PROPERTIES OF NOVACULITE 
 
 Fig. 7. — Thermal expansion apparatus. 
 
RATES OF INVERSION 
 
 23 
 
 mocouple, as shown in figure 7. After press- 
 ing, the samples were dried and stored in a 
 dessicator. The test pieces were measured 
 carefully with a 1/100 millimeter platform 
 Ames dial device. 
 
 Fig. 8. — Mold for pressing samples. 
 
 Expansion measurements. — The appara- 
 tus was set up as shown in figure 7, and 
 the furnace was well swept with hydrogen. 
 When conditions were safe, the tempera- 
 ture was raised at an approximately linear 
 rate of 23° C. per minute. When the 
 desired temperature was reached, it was 
 held constant at that figure until the com- 
 pletion of the test. Temperature and dial 
 
 readings were taken every two minutes 
 until changes occurred more slowly, when 
 readings were taken at less frequent inter- 
 vals. 
 
 Measurements on specimens after firing. 
 — The porosity of the fired test pieces was 
 measured by the suspended weight method, 
 the samples having been saturated with 
 water in a vacuum. Densities were deter- 
 mined by means of pycnometers. Water 
 was added to the samples in the pycnome- 
 ters in vacuo. 
 
 results 
 
 Method of representing results. — In the 
 graphs which follow, expansion has been 
 plotted against time. During the heating 
 period, the temperature was proportional 
 to the elapsed time, the heating rate being 
 23° C. per minute. The volume changes 
 occurring during this period were not accu- 
 rately determined, due to this fast rate of 
 heating. The chief concern was the expan- 
 sion at constant temperature. In every case 
 the inversion was from quartz to cristobal- 
 ite, and the expansion at constant tempera- 
 ture was caused by the inversion, together 
 with other changes of physical structure. 
 
 The initial rapid expansion below 27 min- 
 utes (about 600° C.) was due to the ther- 
 mal expansion of low quartz. The nega- 
 tive expansion which then occurred was too 
 great to be due entirely to the negative 
 thermal expansion of high quartz. The 
 negative expansion has been found to con- 
 tinue until cristobalite is formed in appre- 
 ciable quantities. The point at which con- 
 stant temperature started is marked on each 
 curve. Figure 9 has been prepared as a 
 typical example of the method of represent- 
 ing the data. 
 
24 
 
 PROPERTIES OF NOVACVLITE 
 
 250 
 
 3 00 
 
 150 200 
 
 TIME- MINUTES 
 Fig. 9. — Graph illustrating the effect of the apparatus constant on the expansion-time curve 
 
 of novaculite. 
 
 3 50 
 
RATES OF INVERSION 
 
 25 
 
 1.4 
 
 1.2 
 
 _) 
 < 
 
 a. 
 
 IV 0.4 
 
 X 
 
 00 
 
 
 
 
 
 \ ^^ 1 
 
 
 
 
 
 
 V x 
 
 «. 
 
 
 
 
 
 \d 
 
 J- 
 
 
 
 o o o 
 
 BY INTERFEROMETER 
 
 
 
 
 • • • • 
 
 BY DIFFERENTIAL METHOD 
 NTERFEROMETER SPECIMEN 
 
 
 
 
 
 
 
 SOLID PIECE NOVACULITE 
 
 
 
 D 
 
 DIFFERENTIAL SPECIMEN 
 PRESSED POWDERED 
 NOVACULITE WITH 2%CQ0 
 
 _ct 
 
 
 (SAM PL 
 
 E -1-2 ) 
 
 
 200 
 
 1000 
 
 1200 
 
 400 600 ©00 
 
 TEMPERATURE - DEGREES C 
 Fig. 10. — Comparison of data from interferometer and differential apparatus. 
 
 Comparison of data from interferometer 
 and differential apparatus. — The interfero- 
 meter is well known as an instrument of 
 great precision for the measurements of 
 thermal expansion. It was for this reason 
 that it was used here as check on the differ- 
 ential apparatus. It was particularly desir- 
 able to make sure of the negative expansion 
 above 600° C. For the interferometer test, 
 samples were ground from clear solid novac- 
 ulite. The rate of heating was 2° C. per 
 minute. The results are shown in figure 10, 
 curve I. The shrinkage above 700° C. was 
 
 greater than that of high quartz. The 
 cause of this shrinkage was not known for 
 certain, but we have postulated that it is 
 due to a reaction between the silica and its 
 impurities. The chemical analysis shows a 
 loss on ignition (probably combined water) 
 of 0.62 percent which might account, in 
 part at least for this shrinkage. The other 
 impurities are MgO, 0.12 percent, Al 2 O s 
 1.16 percent, alkalies 0.11 percent, and 
 Fe 2 3 0.33 percent. Any sintering would 
 have to be caused by one or more of these 
 constituents. 
 
26 
 
 PROPERTIES OF NOVACULITE 
 
 
 
 
 
 
 
 
 
 
 
 
 rv^ 
 
 ^ 
 
 -05 
 
 a 10 
 
 
 
 
 
 g 
 
 ^- • 2 
 
 
 
 
 
 
 
 
 -0—0-00 | 
 
 
 
 
 
 
 
 
 
 
 1 100% FINES 
 
 2 50% FINE, 50% MEDIUM 
 
 3 50 % FINE', 25% MEDIUM, 25% CC 
 
 4 50% FINE, 50% COARSE 
 
 5 / 3 FINE , y 3 MEDIUM , |/ 3 COARSE 
 10 y A FINE, % MEDIUM , '/ 2 COARSE 
 
 ARSE 
 
 1 5 
 
 
 
 
 w 
 
 0-» TEfv 
 
 IP. CONSTA 
 ABOVE 
 
 r 
 
 NT AT I4C 
 . HERE 
 
 o°c 
 
 
 
 0. 5 
 
 I 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 50 
 
 100 
 
 250 
 
 300 
 
 Fig. l: 
 
 150 200 
 
 TIME - MINUTES 
 -Effect of grain sizes on the rate of inversion of novaculite. 
 
RATES OF INVERSION 
 
 27 
 
 Curve D was determined with the differ- 
 ential apparatus on a sample of 200-mesh 
 novaculite bonded with 2.0 percent CaO. 
 Curves I and D are in substantial agree- 
 ment, except that curve D shows a much 
 greater shrinkage above 700° C. In this 
 case positive expansion was again resumed 
 at 1040° C. which indicates the inversion 
 of quartz to cristobalite. This point would 
 vary considerably, depending upon the rate 
 of heating of the specimens. 
 
 Influence of grain size. — The following 
 samples were made up of selected novacu- 
 lite of different particle-size grading, bonded 
 with 2.0 percent CaO. The grading was 
 as follows : 
 
 No. 1. 100 percent fines. 
 
 No. 2. Y2 fines, Yi medium. 
 
 No. 3. x /i fines, 34 medium, 34 coarse. 
 
 No. 4. 3^2 fines, Y2 coarse. 
 
 No. 5. l /z fines, Y medium, Y coarse. 
 
 No. 10. 34 fines, 34 medium, Y coarse. 
 
 The sieve size of the fines, medium, and 
 coarse fractions were: through 200-mesh, 
 through 28 on 48-mesh, and through 8 on 
 10-mesh, respectively. The results of the 
 expansion tests have been plotted in figure 
 11. The initial and final porosities and the 
 final densities are tabulated in table 6. 
 
 Table 6. — Densities and Percentage Porosities 
 of Samples 
 
 1 to 5 and 10 
 
 Sample 
 
 Density 
 
 Porosity 
 
 Porosity 
 
 Number 
 
 20° C 
 
 raw 
 
 fired 
 
 
 
 (percent) 
 
 (percent) 
 
 1 
 
 2.35 
 
 38.4 
 
 38.4 
 
 2 
 
 2.36 
 
 32.0 
 
 30.4 
 
 3 
 
 2.36 
 
 25.9 
 
 28.2 
 
 4 
 
 2.36 
 
 26.9 
 
 29.6 
 
 5 
 
 2.36 
 
 25.9 
 
 26.9 
 
 10 
 
 2.36 
 
 25.3 
 
 26.1 
 
 None of these expansions was great 
 enough to account for the density changes 
 observed. The specific gravities of the speci- 
 mens after heating showed conclusively that 
 the material had inverted largely to cristo- 
 balite, presumably about 85 to 90 percent. 
 The expansion system was repeatedly 
 checked to assure accurate results. The 
 material undoubtedly shrank at the same 
 time that the expansion due to inversion 
 
 was taking place. The porosity values 
 observed were not helpful in explaining the 
 cause. The volume changes were the results 
 of two or more phenomena. In most cases 
 the effect has been due largely to the inver- 
 sion of high-quartz to cristobalite. 
 
 Because of the lack of accurate informa- 
 tion concerning all the volume changes that 
 have taken place, the total amount of expan- 
 sion cannot be interpreted specifically, but 
 the shapes of the curves are important be- 
 cause they tell the rates at which these 
 changes proceed. 
 
 Figure 11 shows that the finer grained 
 mixes reached stability much earlier than 
 the coarser grained mixes. Sample 1, com- 
 posed of novaculite finer than 200-mesh, 
 showed no further expansion after one hour 
 at 1400° C. The total expansion was also 
 very low, indicating that much sintering 
 and shrinkage had taken place. This was 
 also the strongest specimen of the lot. Sam- 
 ple 2, of equal parts of 200-mesh and 28 on 
 48-mesh material was similar to sample 1 
 but was still expanding slightly at the end 
 of the test. The expansion was greater 
 than in sample 1, but less than in the other 
 samples. In general, an increase in size of 
 grain slowed the inversion and increased 
 the expansion. The final densities of the 
 samples were all very much the same as 
 shown in table 6. If the porosities had been 
 properly measured, taking all factors into 
 account, we should be able to interpret these 
 changes more fully. Finer grains invert 
 more rapidly, as may be seen from curves 1 
 and 2. It should be a safe assumption that 
 most of the shrinkage and sintering was 
 associated with the finest fractions. From 
 the figure we may conclude that these 
 changes were largely completed very 
 soon after the maximum temperature had 
 been reached. The continued expansion of 
 samples 3, 4, 5, and 10 may then be assumed 
 to be due to the inversion of the larger 
 grains to cristobalite. Independent obser- 
 vation has shown that the larger grains 
 after rapid inversion to cristobalite were 
 soft and rather chalky and were no doubt 
 full of many small pores, which suggested 
 a shattering effect. 
 
28 
 
 PROPERTIES OF NOVACVLITE 
 
 x 3 .o 
 
 o 
 
 z 
 
 2.5 
 
 Z 
 
 o 
 
 z 
 < 
 
 LLl 
 
 
 
 
 
 
 
 1- A 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1- 5 
 
 
 
 
 
 
 
 
 
 
 
 l-A 100 % FINES, NO BOND 
 1-5 100 % FINES, 2%CQO 
 
 
 
 
 
 - 1400° C 
 
 
 
 
 
 
 
 
 
 - 1400° C 
 
 
 
 
 
 
 
 j 
 
 
 
 
 
 
 
 50 
 
 100 
 
 250 
 
 3 00 
 
 150 200 
 
 TIME - MINUTES 
 Fig. 12. — Effect of the use of bond on the rate of inversion of novaculite. 
 
EFFECT OF VARIOUS FLUXES 
 
 29 
 
 Table 7. — Percentage Porosities and Densities of Expansion Samples 
 
 Number 
 
 Flux 
 
 Porosity 
 raw 
 
 (percent) 
 
 Porosity 
 burned 
 
 (percent) 
 
 Density 
 26° C 
 
 F-l 
 
 F-2 
 
 F-4 
 
 F-4A 
 
 F-8 
 
 F-8A 
 
 F-8B 
 
 F-l 3 
 
 F-21 
 
 F-22 
 
 F-23 
 
 F-24 
 
 F-25 
 
 F-26 
 
 3% CaO. 
 
 3% Calcium phosphate 
 
 3%Fe 2 3 
 
 3% Fe 2 3 as Fe(OH) 3 
 
 3% borax 
 
 3% fused borax 80 mesh 
 
 3% borax 
 
 Sodium tungstate 
 
 \y 2 % borax+0.6% Fe 2 O 3 +0.6% CaO 
 
 W 2 % BaO+1^% B 2 3 
 
 \y 2 % BaO+\y 2 % Fe 2 3 
 
 2% CaO+1^% Fe 2 3 +1^% Na 2 0. . 
 1% CaO+0.9% Fe 2 O 3 +0.9% Na 2 0. . . 
 1.2% BaO+0.9% Fe 2 O 3 +0.9% Na 2 Q. 
 
 26.3 
 24.5 
 26.7 
 26.7 
 27.4 
 26.7 
 33.5 
 24.8 
 27.1 
 26.8 
 24.7 
 22.6 
 20.4 
 20.2 
 
 28.1 
 23.5 
 32.0 
 29.7 
 33.7 
 32.1 
 34.0 
 28.3 
 31.4 
 30.1 
 27.0 
 27.8 
 31.0 
 30.4 
 
 2.35 
 
 2.36 
 
 2.36 
 2.30 
 2.29 
 2.32 
 2.34 
 
 2.3S 
 
 2.35 
 2.35 
 2.34 
 2.34 
 
 Effect of Various Fluxes 
 
 comparison of volume changes in 
 specimens with and without bond 
 
 The differences in volume changes be- 
 tween bonded and unbonded specimens is 
 shown in figure 12. These were made of 
 novaculite finer than 200 mesh, 1-5 was 
 bonded with 2 percent CaO and 1-A was 
 unbonded. Both samples inverted almost 
 entirely to cristobalite. The lime-bonded 
 sample reached a constant length after 
 
 about 1 hour at 1400° C, whereas the 
 unbonded sample was still changing after 
 4j/£ hours at the same temperature. The 
 expansion of the lime-bonded specimen was 
 much less than that of the unbounded sam- 
 ple. This was taken to mean that the sin- 
 tering and consolidation processes were 
 much greater in the presence of lime. Both 
 samples showed the characteristic shrinkage 
 above 600° C. The expansion, hence the 
 inversion to cristobalite, began at a lower 
 temperature in the presence of the lime. 
 
30 
 
 PROPERTIES OF NOVACULITE 
 
 3 .5 
 
 3 
 
 2.5 
 
 S 2 . 
 
 Cd 
 LU 
 
 0_ 
 
 I 
 
 z 
 
 1.0 
 
 2% CaO GRAIN SIZES : 
 50% COARSE 
 25% MEDIUM 
 25% FINE 
 THE FINES OF T- 5- I ARE 
 
 COMMERCIAL TRIPOLI 
 THE FINES OF 5 ARE ALL 
 NOVACULITE 
 
 50 
 
 100 
 
 250 
 
 Fig. 13. 
 
 150 200 
 
 TIME - MINUTES 
 -Effect of the use of tripoli on the rate of inversion of novaculite. 
 
 PARTICLE GRADING USED 
 
 The mixture of grain sizes used in all 
 experiments given below in this section 
 were 50 percent 8 to 10-mesh, 25 percent 
 14 to 28-mesh, and 25 percent 200-mesh. 
 The finest fraction was composed of Illinois 
 
 tripoli instead of powdered novaculite. The 
 effect of the tripoli as compared with novac- 
 ulite was very slight, as shown in figure 13. 
 The body containing the tripoli as the fine 
 fraction inverted somewhat more slowly 
 than the all-novaculite body. 
 
EFFECT OF VARIOUS FLUXES 
 
 31 
 
 4.0 
 
 3.5 
 
 O 3.0 
 
 Z 
 
 2 5 
 
 2.0 
 
 ! .5 
 
 0.5 
 
 
 
 
 
 ; — • 
 
 F-l _^ 
 
 
 
 
 
 
 
 T-5-l_< 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 <5>£— 1400 
 
 °c 
 
 
 
 
 
 
 Ip— 1400 
 
 • c 
 
 
 
 
 
 
 
 
 F-l 3<? 
 T-5-1 2 
 
 £ CaO 
 % CaO 
 
 
 
 
 
 
 
 
 
 
 50 
 
 150 
 
 TIME 
 
 200 
 
 MINUTES 
 
 300 
 
 350 
 
 Fig. 14. — Effect of varying the amount of lime bonds on the rate of inversion of novaculite. 
 
 INFLUENCE OF THE AMOUNT OF BOND 
 
 The difference in rate of inversion be- 
 tween samples containing 2 percent and 3 
 percent CaO is shown in figure 14. It is 
 evident that F-l with 3 percent CaO in- 
 verted more rapidly than sample T-5-1, 
 
 which was identical except that only 2 per- 
 cent CaO was added. The difference in 
 rate of inversion was only moderate. The 
 final densities of the two samples were: F-l, 
 2.35 and T-5-1, 2.35. The inversion also 
 started earlier in the presence of 3 percent 
 CaO than in the presence of 2 percent CaO. 
 
32 
 
 PROPERTIES OF NOVACULITE 
 
 3.0 
 
 2.5 
 
 Z 
 id 
 U 
 
 a: i.5 
 
 z 
 o 
 to 
 
 Z 10 
 < 
 D_ 
 X 
 
 0.5 
 
 
 
 
 
 
 1 
 
 F-2 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 d* — 1400 
 
 °c 
 
 
 
 
 
 
 
 F-2, 3< 
 
 ^? FUSED 
 
 MONOCALC 
 
 IUM PHO' 
 
 dPHATE 
 
 
 
 
 
 
 
 
 50 
 
 100 
 
 250 
 
 300 
 
 150 200 
 
 TIME - MINUTES 
 Fig. 15. — Effect of calcium phosphate on the rate of inversion of novaculite. 
 
 350 
 
 CALCIUM PHOSPHATE 
 
 Three percent fused monocalcium phos- 
 phate produced the result shown in figure 
 15. The rate of inversion is not as great 
 as that produced by 2 percent CaO. There 
 was a significant difference between the 
 conduct of the two fluxes as shown by the 
 forms of the graphs representing changes 
 at about 600° C. Some action was at work 
 in this sample which prevented the shrink- 
 age in the high-quartz region characteristic 
 of most of the other samples. Calcium phos- 
 phate cannot be said to be a good accelera- 
 tor for this material. 
 
 SODIUM TUNGSTATE 
 
 Three percent sodium tungstate produced 
 very rapid inversion as may be seen from 
 figure 16. A considerable amount of the 
 expansion observed on sample F-13 was 
 thought to be due to the swelling of the flux 
 during fusion. After about one hour at 
 1400° C, the temperature was raised to 
 1500° C, and no further expansion was 
 noted. This was taken to mean that the 
 inversion had become almost complete. 
 Sample F-13MD was a sample of mine-run 
 novaculite gravel containing 3 percent 
 sodium tungstate which had been heated at 
 
EFFECT OF VARIOUS FLUXES 
 
 33 
 
 o 
 
 z 
 
 LU 3.5 
 
 2-5 
 
 20 
 
 5 
 
 
 
 
 Q^^^-O-^ 
 
 F-13 
 
 
 
 &y* — 14-00 
 
 15 
 
 C 
 
 T 
 
 oo°c 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 > 
 
 
 
 
 
 
 
 r -l3 MD 
 
 (1 
 
 
 
 
 1280° C 
 
 
 
 p— 1100° < 
 
 
 BOTH 5 
 
 3% S 
 
 F- 13 Rl 
 
 PECIMENS 
 ODIUM TU 
 
 [PRESENT. 
 
 CONTAIN 
 NGSTATE 
 
 5 THE 
 
 
 
 
 FIRST FIRING 
 
 F-13 MD HAS BEEN PRE- 
 VIOUSLY FIRED 12 HOURS 
 AT 920 ° C 
 
 r 
 
 ■> 
 
 
 
 
 
 
 50 
 
 250 
 
 3 0-. 
 
 100 150 200 
 
 TIME - MINUTES 
 Fig. 16. — Effect of the use of sodium tungstate on the rate of inversion of novaculite. 
 
 920° C. for 12 hours. A similar sample 
 after the same heat treatment had a density 
 of 2.52, showing that some inversion had 
 already taken place. This curve shows 
 how the inversion decreases with time of 
 
 heating at a given temperature. The sam- 
 ple was held at 1 100° C. for about 125 min- 
 utes, then raised to 1280° C. At each tem- 
 perature the expansion was rapid at first 
 and gradually became progressively slower. 
 
34 
 
 PROPERTIES OF NOVACULITE 
 
 I .5 
 
 I 
 
 O 
 
 z 
 
 U 0.5 
 
 z 
 
 2-0.5 
 
 z 
 < 
 
 0_ 
 X 
 
 -I .0 
 
 I .5 
 
 A 
 
 
 
 
 
 
 
 A 
 
 
 
 
 
 
 
 r 
 
 i 
 
 
 
 
 F-3 
 
 
 
 
 
 
 
 
 
 
 
 
 
 f-3, : 
 
 3% CALCIL 
 
 JM BORAT 
 
 E 
 
 
 1400° 
 
 C 
 
 
 
 
 
 50 
 
 100 
 
 150 200 
 
 TIME - MINUTES 
 
 350 
 
 Fig. 17.— Effect of use of calcium borate on the rate of inversion of novaculite. 
 
 CALCIUM BORATE 
 
 When the flux used was 3 percent crys- 
 tallised calcium borate an excessive shrink- 
 age occurred above 700° C, an example of 
 which is shown in figure 17. This speci- 
 men showed a net increase in length during 
 firing of less than 0.4 percent. 
 
 BORAX 
 
 The effects of 3 percent borax on the rate 
 of inversion to cristobalite at 1400° C, at 
 1300° C, and at 1200° C. is shown in fig- 
 
 ure 18. The borax was added to sample 
 F-8 in solution. F-8A contained 80-mesh 
 fused borax, and sample F-8B, 240-mesh 
 fused borax. A comparison of the curves 
 in figure 18 with those in figure 16 show 
 that borax produced greater acceleration 
 of the inversion than sodium tungstate. 
 Sample F-8 was almost completely inverted 
 before the temperature had reached 1400° 
 C. Afterwards at a constant temperature 
 of 1400° a small continuous shrinkage 
 occurred. F-8B was held at a constant tem- 
 perature of 1300° C, and the inversion 
 proceeded about as rapidly as at 1400° C. 
 
EFFECT OF VARIOUS FLUXES 
 
 35 
 
 5.5 
 
 Fig. 1 
 
 150 200 250 
 
 TIME - MINUTES 
 Effect of the use of borax on the rate of inversion of novaculite. 
 
 300 
 
 There was also a small shrinkage at con- 
 stant temperature. Sample F-8A was held 
 at 1200° C. and the specimen ceased to 
 expand at the end of about one hour at this 
 temperature. The inversion must have been 
 
 complete, as no further expansion was 
 observed when the temperature was raised 
 to 1400° C. The density of this sample 
 was found to be 2.29, showing lower density 
 than any of the other samples in table 8. 
 
36 
 
 PROPERTIES OF NOVACULITE 
 
 150 200 
 
 TIME - MINUTES 
 Fig. 19. — Effect of the use of ferric oxide on the rate of inversion of novaculite. 
 
 IRON OXIDE 
 
 Figure 19 shows the effect of the iron 
 oxide upon the rate of inversion. Sample 
 F-4 contained 3 percent pulverized c.p. 
 Fe 2 3 . Sample F-4A was prepared by add- 
 ing the 3 percent Fe 2 0, as precipitated 
 Fe(OH) 3 . The precipitated iron was 
 much more effective. 
 
 MIXED FLUXES 
 
 Sample F-21 (fig. 20) was composed of 
 1.5 percent fused borax, 0.6 percent CaO 
 as Ca(OH) 2 in solution, and 0.6 percent 
 Fe.,0 3 as c.p. iron oxide, making a total of 
 2.7 percent fluxes. The rate of inversion 
 was very great but not equal to that when 
 borax alone was used (fig. 18). 
 
EFFECT OF VARIOUS FLUXES 
 
 37 
 
 150 200 
 
 TIME - MINUTES 
 Fig. 20. — Effect of the use of mixed fluxes on the rate of inversion of novaculite. 
 
 350 
 
 The fluxes in samples F-24 and F-25 were 
 prepared according to Salmang. 7 Sample 
 F-24 contained 2 percent CaO as Ca ( OH ) 2 
 + 1.5 percent Fe 2 O s as precipitated Fe 
 (OH) 3 + 1.5 percent Na 2 as Na 2 CO a . 
 
 Salmang added the flux in an insoluble 
 form, which is to be preferred. F-25 was 
 the same as F-24, except the amount of the 
 flux was 3 percent instead of 5 percent. 
 
 'Salmang, H. and Wentz, B. f Ber. Deut. Keram. Ges., 
 12, 1, 1, 1931. 
 
38 
 
 PROPERTIES OF NOVACULITE 
 
 4 .0 
 
 3 .0 
 
 Z 
 
 z 
 o 
 
 z 
 
 X 
 
 Ld 
 
 I 
 
 5 
 
 
 
 
 
 
 
 
 f- ; 
 
 -JL~-*r—~—< 
 
 
 
 F-23 
 
 
 
 
 
 
 
 F-26 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1 
 
 1 
 
 
 
 
 
 
 
 
 J' 
 
 — 1150 < 
 >-H400 ( 
 
 
 
 
 
 
 
 
 F-22 
 F- 23 
 
 F- 26 
 
 3% FUSED (1^2. Ba +lj/ 2 B 2 3 
 
 1 l / z % Fe z 3 AS PRECIPITATED 
 Fe (OH) 3 + l>2% BaO AS 
 PRECIPITATED Ba(0H) 2 
 
 1/4% BaO AS BaCO 3 +09% 
 Fe z 3 AS PRECIPITATED 
 F<?(OH) 3 + 0.9% Na ? AS 
 
 ) 
 
 
 . 
 
 
 
 
 
 NaOH 
 
 
 
 
 100 
 
 250 
 
 300 
 
 Fig. 21. 
 
 150 200 
 
 TIME - MINUTES 
 ■Effect of other fluxes on the rate of inversion of novaculite. 
 
 The expansion caused by F-24 was very 
 rapid, and almost equal to that produced 
 by borax. Continued shrinkage was ob- 
 served at 1400° C, in samples F-24 and 
 F-25. 
 
 Sample F-22 (fig. 21) contained a flux 
 composed of 1.5 percent BaO and 1.5 per- 
 cent B 2 3 , which had been fused and pul- 
 verized. Sample F-23 contained a flux 
 
 composed of 1.5 percent BaO as Ba(OH) 2 
 and 1.5 percent Fe 2 O s as precipitated Fe 
 (OH) 3 There was not much difference 
 between the behavior of these fluxes. The 
 flux in sample F-26 was composed of 1.5 
 percent BaO as BaCO s + 0-9 percent 
 Fe 2 3 as precipitated Fe(OH) 3 + 0.9 per- 
 cent Na 2 as NaOH. This sample in- 
 verted almost completely to cristobalite at 
 1150° C. in the time shown in figure 21. 
 
EFFECT OF VARIOUS FLUXES 
 
 39 
 
 2.0 
 
 1.4 
 
 1.2 
 
 Z 
 
 O 0.8 
 
 (/) 
 
 z 
 < 
 
 0_ 
 X 
 
 LU 
 
 
 
 
 
 
 
 
 
 
 
 
 
 F 
 
 M -1 T 
 
 
 
 
 ^^o< 
 
 k*-^^ 
 
 
 
 
 
 
 c// m 
 
 
 
 
 
 
 
 
 
 
 FBBIO 
 
 
 
 
 
 
 y 
 
 /N \ 
 
 
 
 
 
 
 
 I 
 
 
 
 
 
 
 
 
 
 
 
 FM-I 3X CaO 
 FBBIO 3% FLUX: BaC0 3 + 
 H 3 B0 3 -I-Fe203 
 
 
 
 
 
 
 
 
 
 200 400 600 800 1000 1200 
 
 TEMPERATURE -DEGREES C 
 Fig. 22. — Graphs of expansion-temperature changes during firing of novaculite bodies. 
 
 K00 
 
 EXPANSION TEMPERATURE CURVES OF 
 UNFIRED NOVACULITE BODIES 
 
 Thermal expansion studies were made 
 of two samples, employing a very slow 
 heating rate. The curves are shown in fig- 
 
 ure 22. The samples were made of the 
 same grain-size grading that was used for 
 the compression tests described on page 41. 
 Sample FM-1 was 97 percent novaculite 
 and 3 percent CaO. Cristobalite began to be 
 formed rapidly at 1150° C. Sample FBB-10 
 
40 
 
 PROPERTIES OF NOVACULITE 
 
 contained 3 percent bond that was a fusion 
 of the following mixture: 130 grams Ba 
 CO a , 177 grams H 3 BO s and 100 grams 
 Fe 2 3 . This shrinkage in the high-quartz 
 region was greater than with the lime- 
 bonded sample, and was continuous except 
 for a slight expansion between 1180° C. 
 and 1250° C. due to cristobalite. The speci- 
 men was heated to 1500°, but the shrink- 
 age still continued. This presumably was 
 
 the differential of a simultaneous expansion 
 due to cristobalite formation and a shrink- 
 age due to the flux. It is conceivable that 
 a proper blend could be made such that the 
 specimen would show little volume change 
 above 700° C. The specimen was very hard 
 and dense after heating. It probably soft- 
 ened at the high temperatures. If this bond 
 reduces the refractoriness of the silica, 
 smaller percentages of it should be used. 
 
 
 
 
 
 
 
 
 
 
 
 Q 
 UJ 
 Z 
 < 84 
 
 UJ 
 
 O 
 UJ 
 
 Z 
 UJ 
 
 u 
 cr 
 
 UJ 48 
 0_ 
 
 Ul 
 
 > 
 
 _l 
 
 D 
 
 
 
 
 
 S- 
 
 = 4.oe -c 
 
 . 04 4 P 
 
 
 
 
 S= SCREEN OPENING IN MM 
 P= CUMULATIVE PERCENT 
 
 COARSER THAN S / 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ■)/ 
 
 
 
 
 
 
 
 
 o / 
 
 
 50% 6-32MESH 
 25% 32-100 MESH 
 
 
 O 24 
 
 
 / 
 
 > 
 
 
 
 25% 
 
 - IUU 1\ 
 
 icon 
 
 
 O 
 
 to 
 
 
 
 
 
 
 
 
 
 8 10 14 28 48 100 2 00 325 
 
 SCREEN OPENING - MM. (TYLER MESH) 
 
 Fig. 23. — Distribution of particle size on grinding mine-run novaculite. 140 lbs. ground 40 
 minutes in wet pan (dry) ; tailings (on 6-mesh) 8.7 percent. 
 
 Preliminary Small-Scale Tests 
 on Fabricated Samples 
 
 preparation of test samples 
 
 Grain size. — The rock was ground in a 
 wet pan until the distribution of sizes was 
 that shown in figure 23. The same grad- 
 ing was obtained from both the mine-run 
 
 novaculite gravel and selected novaculite. 
 Fine clay was present in the mine-run novac- 
 ulite gravel but not in the selected novacu- 
 lite. This grain sizing was chosen because 
 it produced a good dense brick and because 
 it could be easily duplicated. The correct 
 grading for commercial use would have to 
 be obtained by trial, in additional experi- 
 ments. 
 
PRELIMINARY TESTS 
 
 41 
 
 Table 8. — Compressive Strengths of Novaculite 
 Briquets \}/i by 134 BY ^-Vi Inches 
 
 Number 
 
 Material 
 
 Bond 
 
 Compressive strengths — lb. per sq. in. 
 
 Test 
 1 
 
 Test 
 2 
 
 Fired in 
 No. 1 re- 
 fired in 
 No. 2 
 
 Test 
 3 
 
 OM 
 
 OP 
 
 P-ll 
 
 P-12 
 
 P-l 
 
 1-3M 
 
 M-4 
 
 P-5 
 
 M-6 
 
 P-7 
 
 P-8 
 
 M-15 
 M-16 
 
 P-l 7 
 M-19 
 X 
 
 Mine-run**. 
 Selected f. . . 
 Selected. . . . 
 Selected. . . . 
 Selected. . . . 
 Mine-run. . . 
 Mine-run. . . 
 
 Selected. . 
 Mine-run. 
 Selected. . 
 Selected. . 
 
 Mine-run. 
 
 Mine-run 
 
 Selected 
 
 Mine-run 
 
 From Commercial 
 brick 
 
 None 
 
 None 
 
 l%CaO 
 
 2%CaO 
 
 3%CaO 
 
 3% CaO 
 
 2M% CaO+^ 8 % BaO+^% 
 
 Fe 2 3 
 
 \y 2 % BaO+13^% Fe 2 3 
 
 2%CaO+l%FB10* 
 
 2%CaO+l%FB10* 
 
 2% CaO+13^% Fe 2 Oa+lM% 
 
 Na 2 
 
 3% Feldspar... 
 
 3% Anna Kaolin 
 
 l%CaO+l%FB10* 
 
 !K%BaO+l^%B 2 3 
 
 1100 
 590 
 
 1260 
 1570 
 
 2020 
 1750 
 1890 
 1970 
 
 1770 
 
 1360 
 
 870 
 
 1030 
 
 2600 
 
 1300 
 610 
 
 1840 
 2040 
 
 2680 
 2650 
 2980 
 3450 
 
 2500 
 1930 
 
 1420 
 
 1280 
 640 
 
 1900 
 2100 
 
 2660 
 2230 
 2700 
 3570 
 
 2380 
 1000 
 
 1400 
 
 2710 
 
 1960 
 820 
 950 
 1200 
 1600 
 1120 
 
 2860 
 
 2350 
 1930 
 
 2020 
 
 1670 
 1750 
 
 1910 
 
 *FB10 was prepared by fusing the following mixture: 130 grams BaC03, 177 grams H3BO3, and 100 grams Fe 2 03. 
 
 **Mine-run novaculite gravel. 
 
 fNovaculite. 
 
 Mixing of the bond with silica. — Ten 
 pounds of batch was put in a bucket and 
 the bond (in aqueous suspension) was 
 poured on the same. The whole was made 
 to a good slop-mold consistency by stirring 
 for twenty minutes. 
 
 Forming the test pieces. — A wooden mold 
 was used in which six I14 by l 1 /^ by 21/2- 
 inch test specimens could be formed by 
 throwing the wet mix into the mold with 
 considerable force. The excess was then 
 struck off with a spatula and the briquets 
 were removed by inverting the mold and 
 taking it apart. 
 
 Bonding agents used. — Various bonds, 
 shown in table 8, were used in order to 
 determine their effect on the cold compres- 
 sive strength of the fired briquets. Each 
 bonding material (table 8) was mixed 
 as a slurry and added in the proper quan- 
 tity. Since it was not believed possible to 
 obtain as good mixing in the small quanti- 
 ties used as would be obtained in a wet pan, 
 3 percent of the flux was used instead of the 
 customary 2 percent. 
 
 Firing the samples. — The samples were 
 fired in the laboratory test kilns according 
 to heating schedules shown in figure 24. 
 
 From inspection of samples after ther- 
 mal expansion tests it was evident that the 
 greatest cause of weakness in novaculite 
 bricks was the too rapid inversion to 
 cristobalite. Trials were made to deter- 
 mine the temperature limits at which this 
 inversion should take place. The rather 
 rapid increases in temperature to cone 14 
 in firings Nos. 1 and 2 were made with the 
 idea that if inversion were substantially 
 complete, it would not weaken the brick. 
 If much uninverted quartz remained, the 
 brick would have been "punky" and weak. 
 
 TESTING THE FIRED SAMPLES 
 
 Compressive strength tests. — Cold com- 
 pressive strengths were determined on the 
 1 14 by 1 14 by 2j/£ inch samples. The ends 
 were set in plaster and the samples were 
 broken in a hydraulic press. 
 
 Density measurements. — The densities of 
 many of the samples were determined in 
 order to obtain an index of the amount of 
 
42 
 
 PROPERTIES OF N OVA CU LITE 
 
 
 
 
 
 \\ 
 
 
 
 
 
 
 
 1/ 
 
 
 
 
 
 
 
 
 
 
 sj 
 
 
 
 
 
 
 
 
 LL 
 
 Z 
 
 c 
 
 1 
 > 
 
 
 
 
 
 
 
 
 
 (f)\ 
 
 
 / 
 
 
 
 
 <s 
 
 r M 
 
 j ^ 
 
 
 
 
 
 
 
 C 
 
 r 
 ) 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 / N.,^^^^ 
 
 
 
 
 
 
 o -" 
 
 Z £ 
 
 — c 
 
 o e 
 
 o I u 
 
 <45 
 
 D 933803Q - 3yniVa3dlAI31 
 
PRELIMINARY TESTS 
 
 43 
 
 inversion to cristobalite that had taken 
 place. These measurements were made 
 carefully with pycnometers, using very 
 finely ground samples which had been wetted 
 under a vacuum. 
 
 RESULTS OF FIRING TESTS 
 
 Table 8 shows the effects of the compo- 
 sition of the bond and the heat treatment on 
 the compressive strengths. The heat treat- 
 ment of these samples is shown in figure 24. 
 The figures in the third vertical column of 
 table 8 show the effect of refiring samples 
 from test No. 1 in test No. 2. In every 
 case the strength was substantially increased 
 by this treatment. The samples from firing 
 No. 2 were in every case stronger than those 
 from firing No. 1. Some of the refired 
 brick were stronger after the second firing 
 than those of No. 2, but in general, the 
 strengths were substantially the same. 
 Those stronger after this treatment were 
 P-7 and M-15 ; P-5 was weaker. With one 
 exception, test No. 2 gave stronger brick 
 than No. 1 because of the longer, more 
 gradual fire, and firing No. 3 produced the 
 weakest brick because the temperature was 
 not high enough to develop the bond prop- 
 erly. The mine-run novaculite gravel 
 without bond and containing its associated 
 clay had maximum strength at the low tem- 
 perature fire. The sample bonded with 
 feldspar also had a good strength when 
 fired to cone 7. 
 
 Aside from the heating-up period, which 
 is too rapid for commercial size brick, the 
 desired rate of heating novaculite for silica 
 brick would be between firing Nos. 2 and 3. 
 Firing No. 3 did not go to nearly high 
 enough temperature to develop the bond and 
 the tridymite, but the inversion was slow 
 enough at that rate to prevent damage to a 
 full-size brick. The inversion of the silica 
 in fire No. 2 was too rapid and full-size 
 brick were "punky." In these tests no 
 attempt was made to produce the desirable 
 network of tridymite crystals which develop 
 in time at a high temperature. It is prob- 
 able that there was some tridymite in some 
 of the specimens, although not in sufficient 
 quantities. The rate of heating should be 
 
 Table 9. — True Densities of Fired Briquettes 
 
 
 Test 
 
 Test 
 
 Fired in Test No. 1 
 
 
 No. 1 
 
 No. 2 
 
 Refired in Test No. 2 
 
 OM 
 
 
 
 2.29 
 
 OP 
 
 
 2.36 
 
 
 M-19 
 
 2.34 
 
 
 
 P-l 
 
 
 2.32 
 
 2.32 
 
 1-3M 
 
 
 
 2.32 
 
 M-4 
 
 
 
 2.34 
 
 P-5 
 
 2.32 
 
 2.31 
 
 
 M-6 
 
 2.31 
 
 
 2.33 
 
 P-7 
 
 
 
 2.32 
 
 M-15 
 
 
 
 2.29 
 
 M-16 
 
 
 
 2.30 
 
 very slow when inversion first begins and 
 the rate may be gradually increased after 
 that. 
 
 The composition of the bond was an 
 important factor in the development of cold 
 compressive strength. Pieces cut from 
 unfired commercial silica brick and then 
 fired in the same kiln were used for compari- 
 son. The lime-bonded novaculite brick 
 were not as strong as the commercial silica 
 brick. The cold strength of some experi- 
 mental brick made with other bonds was 
 much greater than the commercial brick 
 which was lime bonded. The strongest 
 brick was P-7, which was selected novacu- 
 lite bonded with 2 percent CaO + 1 per- 
 cent of a fusion of the following mixture: 
 130 grams BaCO s , 177 grams H 3 B0 3 + 
 100 grams Fe 2 2 . The average compres- 
 sive strength of briquets of this composition 
 was 3570 pounds per square inch. 
 
 Another very good quality brick was M-4 
 which was bonded with 2j4 percent CaO 
 + }i percent Fe 2 3 + y% percent BaO. 
 The frits used in samples P-8 and M-19 
 were too soluble, which prevented their sat- 
 isfactory test. On the whole, the use of 
 BaO or BaO + Fe 2 3 + B 2 O s gave a dis- 
 tinct advantage in raising the mechanical 
 strength of the brick. 
 
 It is known that Al 2 O s is harmful to the 
 refractoriness of a silica brick. Sosman* 
 says this oxide should not exceed 2j/£ P er " 
 cent of the total and it should probably be 
 
 8 Sosman, R. B., The properties of silica: Chem. Cat. 
 Co., 1927. 
 
44 
 
 PROPERTIES OF NOVACULITE 
 
 lower than that. The Al 2 O s in all these 
 samples was much lower than this except 
 M-15 and M-16. Other tests should be 
 made on brick having these novel bonds be- 
 fore their use can be fully evaluated. 
 
 The specific gravities of some of the speci- 
 mens are shown in table 9. They indicate 
 a very high degree of inversion. 
 
 SUMMARY 
 
 Novaculite may be ground to give the 
 grain sizes required to produce the desired 
 structure of a brick. Tripoli may be added 
 to increase the amount of fines. The total 
 volume of voids accompanying the packing 
 of novaculite grains decreases with decreas- 
 ing grain size. 
 
 The grain-size distribution affects the 
 rate of inversion to cristobalite and the 
 amount of total expansion during firing. 
 The finest grained mixtures invert the most 
 rapidly and expand the least. In firing a 
 novaculite brick, a shrinkage occurs above 
 600° C. which continues until cristobalite 
 
 begins to be formed. This is assumed to be 
 due to sintering and solid state reactions. 
 
 The bond serves a double purpose, that of 
 drawing the grains into a compact mass and 
 that of accelerating the inversion. The com- 
 position of the bond affects the rate of inver- 
 sion and the temperature at which it begins. 
 
 Borax accelerates the inversion to cristo- 
 balite more than any other flux used. A 
 little BaO replacing part of the CaO has 
 a beneficial effect. Fe 2 3 when added in a 
 finely divided form is a good fluxing agent. 
 The amount of total shrinkage is depend- 
 ent to a large degree upon the kind of flux 
 and the form in which it is added. 
 
 Novaculite briquets may be made which 
 have superior compressive strengths. The 
 replacement of some of the CaO with BaO, 
 Fe 2 3 , or B 2 O s greatly increases the cold 
 crushing strength of novaculite bricks. 
 
 In firing novaculite, the temperature 
 should increase 20° or 30° C. per day from 
 1000° C. for about three days, after which 
 the rate should be increased slightly. 
 
PART II— PREPARATION AND PROPERTIES OF SILICA 
 
 BRICK MADE FROM ILLINOIS NOVACULITE 
 
 AND QUARTZITIC SANDSTONE 
 
 Introduction 
 
 The purpose of this part of the work was 
 to study the characteristics of novaculite 
 brick with respect to ( 1 ) the preparation 
 of the raw material, (2) fabrication, and 
 (3) the finished product. In addition to 
 novaculite, some tests have been made on a 
 quartzitic sandstone from Tansil, Illinois. 
 Where necessary, the properties of the Illi- 
 nois material have been compared with those 
 of commercial products. 
 
 Preparation of Material for Testing 
 
 NOVACULITE GRAVEL 
 
 Novaculite at the pit. — Overlying the 
 hard compact massive novaculite are many 
 feet of novaculite gravel containing a small 
 amount of red clay. The gravel fragments 
 are not rounded, as true gravel, but are 
 angular and sharp edged. The screen analy- 
 sis of the novaculite as received is shown in 
 table 10. 
 
 Table 10. — Screen Analysis of Novaculite 
 Gravel, "as received" 
 
 Table 11. — Screen Analysis of "as received' 
 Novaculite Gravel after Washing through 
 20-mesh and 28-mesh screens 
 
 Particle size, sieve mesh 
 
 Wt. percent 
 
 Plus 1-inch 
 
 7.8 
 
 1 to ^8-inch 
 
 29.2 
 
 % to 3^-inch 
 
 22.2 
 
 3^ to 6 mesh 
 
 19.9 
 
 6 to 10 mesh 
 
 3.3 
 
 10 to 20 mesh 
 
 Minus 20 mesh 
 
 2.6 
 15.0 
 
 
 
 Total 
 
 100.0 
 
 
 
 Washing of novaculite gravel. — In the 
 laboratory the gravel was placed on a 20- 
 mesh screen and the clay and minus 20- 
 mesh material was washed away by a stream 
 of water. Twelve percent of material was 
 washed through a 28-mesh screen. Table 
 11 shows the screen analysis of the mine- 
 run gravel after washing. 
 
 Particle size, 
 
 sieve mesh 
 
 Wt. percent 
 
 Plus 1-inch 
 
 8 8 
 
 1 to 3^-inch 
 
 33 2 
 
 % to M-inch 
 
 25 2 
 
 3^-inch to 6 mesh 
 
 22 6 
 
 6 to 10 mesh 
 
 3 8 
 
 10 to 20 mesh 
 
 Minus 20 mesh 
 
 28 mesh 
 
 3.0 
 
 3.4 
 12 
 
 
 
 SAMPLING FOR TESTS 
 
 Three hundred pounds of the novaculite 
 gravel was washed, dried, thoroughly mixed, 
 and stored for later use. Four hundred 
 pounds of the novaculite was crushed to 
 lumps one inch in diameter or less, mixed, 
 and stored. It did not require washing. 
 The quartzitic sandstone occurs in thick 
 deposits of massive rock. Three hundred 
 pounds were crushed to lumps one inch in 
 diameter or smaller, mixed well, and stored 
 without any further treatment. 
 
 PROPER GRAIN-SIZE DISTRIBUTION 
 
 Preparation of the material. — Three 125- 
 pound batches of washed novaculite gravel 
 were ground separately for 10 minutes each 
 in a wet pan. Each lot of crushed material 
 was then divided, by screening, into three 
 grain-size groups. These groups were: 
 
 Coarse — on 14-mesh 
 
 Medium — Through 14-mesh on 100-mesh 
 
 Fine — Through 100-mesh 
 
 The relative amounts of each fraction 
 obtained were : coarse 29.8 percent, medium 
 34.7 percent, and fine 35.5 percent. 
 
 [45] 
 
46 
 
 PROPERTIES OF SILICA BRICK 
 
 FINE 
 
 SIZES OF GRAINS 
 
 
 
 
 
 
 
 
 Tyler Weight percent 
 
 
 
 Mesh Coarse Med Fine 
 
 THROUGH 6 17.5 
 
 
 
 Figures on curves refer to 
 the ratio of bulk volume 
 to true volume 
 
 6 - 8 40.8 
 
 8 - 10 33 7 42-8 
 
 
 
 
 10- 20 7.6 23.6 4.4 
 
 20-26 29 2 8. 9 
 
 48-100 3 1 1 7. 5 
 
 1 00 - 150 1 .4 1 5. 3 
 
 1 50 - 200 1 0, 8 
 
 200 AND FINER . 9 42. 6 
 
 
 
 
 PACKED UNDER 300 POUNDS PER 
 SQUARE INCH PRESSURE WITH 
 
 
 
 
 
 4% MOI STURE . 
 
 
 
 
 
 
 
 
 \\V 
 
 
 
 AAAA 
 
 MODULUS OF RUPTURE A 
 
 
 'W 
 
 
 
 NO. M R. ^v" / \j 
 
 
 ^x\ 
 
 
 
 / V / 
 
 
 
 8 - 18 9 -^A / \J 
 
 \ / \ 
 
 
 v^v^A 
 
 
 
 X - 179 - / 7\ 
 12 - 157 ~"y \ / \ , 
 
 
 
 
 
 
 / \ / \ / \ 
 
 
 
 
 
 
 f<A/V\ 
 
 
 
 
 
 
 
 
 
 \ 7 
 
 
 
 . AAAA 
 
 COARSE 
 
 
 
 
 
 
 
 MEDIUM 
 
 Fig. 25. — Packing of many grain sizes. 
 
 ADDITIONAL PACKING EXPERIMENTS 
 
 Method. — A mold of uniform circular 
 cross-section was charged with 30 grams of 
 grain mixtures, which had been moistened 
 with 2 cc. of water. The charge was then 
 packed by a piston with the application of a 
 pressure of 3000 pounds per square inch. 
 The volume of packed grains was calcu- 
 lated from the height of the piston after 
 packing. 
 
 Grain sizes studied. — Various mixtures 
 of the coarse (on 14-mesh), fine (through 
 100-mesh), and medium (through 14-mesh 
 on 100-mesh) fractions, were packed. The 
 screen analyses of these fractions are shown 
 in table 12. 
 
 Results of packing tests. — The results of 
 these tests are shown in figure 25. The 
 values plotted were the ratios of the bulk 
 volumes to the true volumes. The lines on 
 the diagram represent mixtures which pack 
 to the same density. From the plotted data 
 it may be seen that 35 percent of the coarse, 
 15 percent medium and 50 percent fine 
 (fractions identified in table 12) is repre- 
 sentative of the densest packing mixture of 
 this system. The grain-size distribution of 
 this mixture, as calculated from the end 
 members in the system, is shown in table 13. 
 When this mixture was moistened with 4 
 percent water and pressed under a pressure 
 of 3000 pounds per square inch, the result- 
 ing porosity was 19 percent. 
 
PREPARATION OF MATERIAL 
 
 47 
 
 Table 12. — Screen Analysis of the Ground Washed Mine-run Novaculite Gravel 
 
 Divided into Three Fractions 
 
 In percent 
 
 Sieve mesh 
 
 Coarse 
 
 Medium 
 
 Fine 
 
 Plus 6 
 
 17.5 
 
 40.2 
 
 33.7 
 
 7.6 
 
 42'0 
 
 23.6 
 
 29.2 
 
 3.1 
 
 1.4 
 
 0.9 
 
 
 6 to 8 
 
 
 8 to 10 
 
 
 10 to 20 
 
 
 20 to 28 
 
 4.4 
 
 28 to 48 
 
 8.9 
 
 48 to 100 
 
 17.5 
 
 100 to 150 
 
 15.3 
 
 150 to 200 
 
 10.8 
 
 Minus 200 
 
 42.6 
 
 
 
 CHARACTERISTICS OF UNFIRED NOVACU- 
 LITE BRIQUETS 
 
 EFFECT OF AMOUNT OF TEMPERING WATER 
 ON THE MODULUS OF RUPTURE 
 
 Grain-size gradings. — The two grain- 
 size gradings used were designated as "A" 
 and "B." The "A" grading was that natu- 
 rally obtained when 120 pounds of rock was 
 ground 15 minutes in a wet pan. The "B" 
 grading was the distribution computed 
 from the equation : 9 
 
 d = P 2 D x 10~ 4 
 
 Testing procedure. — Each batch was 
 large enough to form 10 specimens, and 
 after adding 2 percent CaO as limewater, 
 it was mixed as thoroughly as possible by 
 hand, then tamped into a 1 by 1 by 7 inch 
 mold. The tamping force may be described 
 as medium. After drying, the samples were 
 tested for transverse strength. The data 
 are shown in figure 26. These curves show 
 maximum strength to be at a certain opti- 
 mum water content. Both the maximum 
 strength and the optimum water content 
 varied with the grain-size grading. The 
 maximum strength and optimum water con- 
 tent will also be found to depend upon the 
 ramming pressure. 
 
 ETFECT OF PARTICLE GRADING ON THE MODULUS 
 OF RUPTURE 
 
 Testing procedure. — It was noted in pre- 
 vious experiments that the optimum water 
 content could be judged fairly well from 
 the "feel" of the wet mixture. This pro- 
 cedure was used in the preparation of the 
 
 samples for the tests described in this sec- 
 tion. The batches were prepared with 3 
 percent CaO (added as lime water), and 
 the amount of water most appropriate for 
 the mix. Briquets were formed by ram- 
 ming the wet batch into a 1 by 1 by 7 inch 
 steel mold. 
 
 Two ramming pressures were used for 
 this series of tests. In the lower pressure 
 ram, the mix was tamped into the mold by 
 ramming with the end of a 10-inch spatula 
 handle which was held vertically in the 
 hand and brought down into the mix with 
 as much force as possible. The higher ram- 
 ming pressure was obtained by fitting a 
 wooden plug into the mold and driving it 
 with a wooden mallet. This procedure 
 could be used to fill the mold about two- 
 thirds full. To complete the job, a large 
 excess of batch was piled on top of the mold 
 and then beaten down into the mold by 
 heavy blows from the mallet. 
 
 Grain sizes. — The particle gradings are 
 shown in figure 25. Mixtures were chosen 
 so as to cross the minimum void area in this 
 figure. 
 
 Results of modulus of rupture tests. — 
 The compositions are shown in figure 25 
 and the results of these tests are summar- 
 ized in table 14 and figure 25. The speci- 
 ments formed by the high ramming pressure 
 gave fairly uniform results, while the low 
 ramming pressure produced specimens with 
 
 » Taylor and Thompson. Concrete — Plane and Rein- 
 forced, p. 775. 
 
 D= diameter of largest grain. 
 
 d =any chosen diameter. 
 
 P = per cent of mixture smaller than any given diameter. 
 
48 
 
 PROPERTIES OF SILICA BRICK 
 
 z 
 
 a 
 m 
 
 cr. 
 
 LJ 
 
 Q_ 50 
 
 
 
 
 
 
 
 
 
 
 V ^^^ 
 
 
 
 
 
 
 
 
 
 
 ^ 
 
 ^"-\^ 
 
 Y^^^ 
 
 
 
 
 
 
 / c 
 
 / 
 
 ) 
 
 
 
 ^\*-— PARTICLE GRADING A 
 
 
 
 
 
 
 G B 
 
 
 \ 
 
 
 
 
 
 
 
 
 
 
 
 15 16 17 18 19 
 
 TEMPERING WATER - PERCENT OF DRY SOLIDS 
 
 Fig. 26. — Effect of amount of tempering water used on the modulus of rupture of unfired 
 novaculite briquets, CaO constant. 
 
 30 40 50 
 
 TIME OF HEATING - HOURS 
 Fig. 27. — Firing schedule and rates of inversion. 
 
RATE OF INVERSION 
 
 49 
 
 wide variation in strength. Specimens 
 formed by high ramming pressure varied 
 from the mean by a maximum of but 6.8 
 percent, but those made by the low ram- 
 ming pressure showed a 43 percent varia- 
 tion. In general the low ramming pressure 
 produced bars having 30 to 80 percent of 
 the strength of the high ramming pressure 
 specimens. 
 
 Reference to figure 25 shows that the 
 maximum strength was possessed by those 
 specimens whose grain sizes fell within the 
 densest packing area. The grain-size dis- 
 tribution is indicated by the coordinates on 
 the graph, and the transverse strengths of 
 the briquets are indicated by the arrows. 
 The strengths are shown in a little more 
 detail in table 14. These data show that 
 the strength is a function of the density of 
 packing of the grains in the brick. As the 
 forming pressure increases, the porosity 
 decreases and the strength increases. This 
 change with pressure becomes very small 
 at higher pressures. The most uniform 
 results are obtained in this upper range of 
 pressures. 
 
 Table 13. — Grain-size Distribution of the 
 
 Minimum Void Mixture Determined by 
 
 Packing Experiments 
 
 Sieve mesh 
 
 Wt. percent 
 
 6 to 8 
 
 6.6 
 14 3 
 
 8 to 10 
 
 11.8 
 
 10 to 20 
 
 9.1 
 
 20 to 28 
 
 5.7 
 
 28 to 48 
 
 8.8 
 
 48 to 100 
 
 9 1 
 
 100 to 150 
 
 7.8 
 
 150 to 200 
 
 5.4 
 
 Minus 200 
 
 21.4 
 
 
 
 Rate of Inversion 
 novaculite and other silica materials 
 
 Method of study. — The materials chosen 
 for study were Illinois tripoli, novaculite, 
 Baraboo quartzite, Illinois ganister, and an 
 Illinois quartzite-sandstone. The same 
 grain-size distribution was used in all the 
 specimens, that is, the optimum grain-size 
 found by the packing experiments, and 
 shown in table 13. Each material was mixed 
 with freshly slaked milk of lime, so that the 
 CaO content of each specimen was 2.0 per- 
 cent. The specimens were formed by ram- 
 ming tightly in a 1 by 1 by 4 inch mold. 
 
 The molded pieces were put in refractory 
 trays as an aid in making draw trials from 
 a hot furnace. Each tray contained one 
 specimen of each type of silica. These were 
 fired in a gas kiln according to the heating 
 schedule shown in figure 27. Samples were 
 drawn from the furnace after treatments as 
 indicated at A, B, C, and D. As soon as 
 drawn, they were thrown into a crucible 
 partially filled with sand and placed in a 
 furnace heated at a temperature of about 
 800° C. After covering the samples with 
 hot sand they were allowed to cool slowly 
 to room temperature. The cold samples 
 were pulverized to pass 200 mesh, and the 
 density of the powders wetted in vacuo was 
 determined in pycnometers. 
 
 Results. — The results are shown in fig- 
 ure 27. The Baraboo quartzite showed a 
 steady decrease in specific gravity, and 
 after the entire heat treatment the average 
 density of the material was 2.345, indicat- 
 ing the presence of some uninverted quartz. 
 There was some tridymite present, but the 
 predominating portion was cristobalite. 
 
 Table 14. — Modulus of Rupture of Unfired Silica Brick. 
 In pounds per square inch 
 
 High... 
 Low. . . . 
 Average 
 
 No. 5 
 Ramming 
 pressure 
 
 High Low 
 
 192 110 
 169 64 
 180 87 
 
 No. 8 
 Ramming 
 pressure 
 
 High Low 
 
 193 148 
 181 62 
 189 107 
 
 No. 10 
 Ramming 
 pressure 
 
 High Low 
 
 184 109 
 169 46 
 179 81 
 
 No. 12 
 Ramming 
 pressure 
 
 High Low 
 
 163 129 
 151 70 
 157 90 
 
 No. R 
 Ramming 
 pressure 
 
 Med. 
 
 80 
 50 
 70 
 
50 
 
 PROPERTIES OE SILICA BRICK 
 
 Microscopic examinations revealed the 
 quartz to be the cores of large grains. 
 
 The curve for the Illinois quartzite-sand- 
 stone was very similar to that of Baraboo 
 quartzite. The quartzitic sandstone was 
 high in fluxing materials, which increased 
 the specific gravity somewhat. There was 
 very little uninverted quartz in the sample 
 after the full heat treatment, yet its density- 
 was 2.37, due to the presence of heavier 
 impurities. 
 
 The curves for Illinois ganister, tripoli 
 and novaculite are identical and show a 
 much greater rate of inversion than the 
 other silica samples. The minimum density 
 shown at the C draw contained considerable 
 tridymite, as the higher temperature 
 (above 1470° C.) showed an increase from 
 2.30 to 2.33 in average density. Aside 
 from impurities and the bonding agent, the 
 Illinois ganister and the novaculite after 
 the D treatment were pure cristobalite, 
 and no quartz was present. An appreciable 
 amount of tridymite had been formed by the 
 B treatment. The effect of the B and C 
 treatment was to increase the tridymite. 
 
 Experiments with Standard 
 9-inch Shapes 
 
 General method. — Standard 9-inch brick 
 were made in the laboratory by hand mold- 
 ing and were fired in a regular commercial 
 kiln. The following properties were ex- 
 amined : 
 
 a. Modulus of rupture. 
 
 b. Deformation under load at high tempera- 
 
 tures. 
 
 c. True density. 
 
 d. Porosity. 
 
 e. Exterior volume changes. 
 
 f. Over-all changes in length. 
 
 g. Microstructure. 
 
 h. Thermal expansion. 
 
 Details of procedure. — Raw materials 
 included: 
 
 a. Selected white novaculite (clean). 
 
 b. Novaculite gravel and associated red clay. 
 
 c. Washed novaculite gravel. 
 
 d. Quartzitic sandstone. 
 
 All these materials contained the iron 
 from the crushing operations. 
 
 Preparation of Bonding Agents 
 
 lime bonds 
 
 High quality lime was slaked in an ex- 
 cess of distilled water, screened through a 
 120-mesh sieve, and put in a 10-liter glass 
 carboy. The CaO content per cc of sus- 
 pension was determined by analysis. 
 
 special bond 
 
 A special bond was prepared by fusing 
 the following mixture: 130 parts BaC0 3 , 
 177 parts H3BO3, and 100 parts Fe 2 3 . 
 This fusion, designated as FB-10, was pul- 
 verized to pass a 200-mesh screen and 
 mixed with sufficient lime water to form 
 two parts CaO to one part of the fused 
 mixture. 
 
 Preparation of Specimens 
 grinding and mixing 
 
 The mine-run novaculite gravel was put 
 directly into a wet pan together with the 
 lime water and was ground for 15 minutes. 
 The screen analysis of this mix is shown 
 in table 15. 
 
 The selected novaculite rock was re- 
 duced in a jaw crusher, then put in a wet 
 pan with the required amount of bond and 
 ground for 15 minutes. The same treat- 
 ment was given the quartzitic sandstone. 
 
 The washed novaculite gravel was 
 ground dry in a wet pan and divided into 
 three fractions as described on page 13. 
 The bond and water were added to these 
 batches, then mixed carefully by hand. The 
 grain sizes of the different mixes are in- 
 dicated under the proper heading in table 
 15. 
 
MODULUS OF RUPTURE 
 
 51 
 
 Table 15. — Kind of Raw Material, Grain Sizes, and Particle Grading of 
 kxperimental brick 
 
 
 Novaculitc, 
 selected 
 
 Novaculite, 
 
 mine-run 
 
 gravel 
 
 Quartzite 
 sandstone 
 
 Novaculitc gravel, washed 
 
 Screen mesh 
 
 Brick No. 
 
 CP-2,CP-3, 
 CP-7 
 
 CM-0, CM-2 
 
 QS 
 
 WN-I 
 WN-II 
 
 WN-I 1 1 
 
 485 
 
 
 Percentage on screen 
 
 Plus 6 
 
 13.3 
 
 in a 
 
 23*0 
 13.0 
 11.6 
 
 ii'.2 
 
 9.6 
 
 ii'.o 
 
 6.8 
 
 24.2 
 15.8 
 
 17.2 
 
 14~4 
 10.0 
 
 20.1 
 6.7 
 
 4.7 
 
 1A 
 
 3.9 
 30.9 
 
 '3.9 
 
 22.3 
 
 6.6 
 14.3 
 11.8 
 
 9.1 
 
 5.7 
 8.8 
 9.1 
 7.8 
 5.4 
 
 21.4 
 
 None 
 
 17.3 
 
 14.5 
 
 9.1 
 
 5.7 
 8.8 
 9.1 
 7.8 
 5.4 
 
 21.4 
 
 None 
 
 6 to 8 
 
 None 
 
 8 to 10 
 
 None 
 
 10 to 20 
 
 18.0 
 
 20 to 28 
 
 15.9 
 
 28 to 48 
 
 20.2 
 
 48 to 100 
 
 8.8 
 
 100 to 150 
 
 8.6 
 
 150 to 200 
 
 4.8 
 
 Minus 200 
 
 
 Minus 200 to 325 
 
 Minus 325 
 
 8.7 
 15.0 
 
 
 
 Making the Brick 
 
 The wet batches were pounded firmly 
 into a mold and the bricks thus formed 
 were removed and allowed to dry a few 
 days in air before going to the dryer. The 
 kind of material used, the kind and amount 
 of bond, and the designation numbers of the 
 brick are shown in table 16. 
 
 The gradings of the CP series and the 
 CM series are the natural gradings obtained 
 from the respective materials when ground 
 in the laboratory wet pan. The same is 
 true of the quartzitic sandstone brick. The 
 grading used in the WN-I and the WN-11 
 brick was that which was shown in previous 
 tests to be the optimum grain-size distribu- 
 tion. The size distribution of batch No. 
 WN-I 1 1 was the same as WN-I and 
 WN-II except that the coarsest grains were 
 through 6-mesh on 8-mesh. Number 485 
 had a maximum grain diameter of through 
 10-mesh on 20-mesh. This requires the 
 addition of 15 percent of silica finer than 
 325-mesh. For this fraction, minus 325- 
 mesh tripoli was used. 
 
 The brick were fired with commercial 
 silica brick in periodic kilns. 
 
 Properties of the Test Brick 
 modulus of rupture 
 
 In the cross-breaking test, silica brick 
 always fail in tension by tearing apart 
 below the neutral axis. For this reason the 
 test can be considered as a satisfactory 
 measure of the degree to which brick are 
 bonded together. As the strength and dura- 
 bility of silica refractories are limited 
 largely by the degree to which fragments 
 are bonded, this is a rather important test. 
 
 The cross-breaking strength data of some 
 novaculite experimental brick are shown in 
 table 16. The kind of raw material, par- 
 ticle grading, and brick numbers will be 
 found in table 15. Brick No. CM-2 was 
 the weakest of all. Too much water was 
 used in the batch and the result was a high 
 porosity (33 percent), and a weak bond. 
 The individual grains in this brick were 
 somewhat shattered. This sugggests that 
 good brick cannot be made from novaculite 
 unless the grains are forced tightly together 
 during the forming process. Brick No. 485 
 had the highest modulus of rupture of any 
 brick tested, and also the lowest porosity. 
 The other brick, being intermediate be- 
 
52 
 
 PROPERTIES OF SILICA BRICK 
 
 Q 
 Id 
 <*■ 800 
 
 °^ 
 
 ' > \_ o 
 
 o ^"v. 
 
 o \ 
 
 28 29 30 31 
 
 POROSITY OF UNFIRED BRICK - PERCENT 
 
 Fig. 28. — Porosity of unfired brick versus modulus of rupture of fired brick. 
 
 'I able 16.— Data on 9-inch Novaculite Brick Fired in Commercial Kiln 
 
 
 
 
 Change during firing 
 
 Porosity 
 
 
 Modulus 
 
 
 Brick 
 No. 
 
 Bond 
 
 
 
 
 
 Density 
 after 
 
 of 
 
 Material 
 
 
 
 
 
 rupture 
 
 
 
 Length 
 
 Volume 
 
 Unfired 
 
 Fired 
 
 firing 
 
 lb. per 
 
 
 
 
 % 
 
 % 
 
 % 
 
 % 
 
 
 sq. in. 
 
 Novaculite 
 
 
 
 
 
 
 
 
 
 crude gravel 
 
 CM-0 
 
 None 
 
 4.1 
 
 13.2 
 
 25.6 
 
 
 
 339 
 
 Novaculite 
 
 
 
 
 
 
 
 
 
 crude gravel 
 
 CM-2 
 
 2.0% CaO 
 
 2.7 
 
 
 33.5 
 
 
 2.33 
 
 271 
 
 Massive novaculite. . 
 
 CP-2 
 
 2.0% CaO 
 
 3.1 
 
 9.4 
 
 28.3 
 
 27.2 
 
 2.32 
 
 779 
 
 Massive novaculite. . 
 
 CP-3 
 
 3.0% CaO 
 
 3.5 
 
 9.5 
 
 28.0 
 
 26.9 
 
 2.30 
 
 754 
 
 Massive novaculite. . 
 
 CP-7 
 
 2.0% CaO 
 + 1% FB 
 
 3.1 
 
 9.1 
 
 
 
 
 
 
 
 10 
 
 
 
 27.6 
 
 27.6 
 
 2.30 
 
 829 
 
 Quartzite sandstone. 
 
 QS 
 
 3% CaO 
 
 5.0 
 
 15.0 
 
 
 
 
 724 
 
 Washed novaculite 
 
 
 
 
 
 
 
 
 
 gravel 
 
 WN-I 
 
 2% CaO 
 
 2.6 
 
 8.0 
 
 26.5 
 
 24.5 
 
 
 839 
 
 Washed novaculite 
 
 
 
 
 
 
 
 
 
 gravel 
 
 WN-II 
 
 3% CaO 
 
 2.8 
 
 8.4 
 
 26.9 
 
 25.8 
 
 
 732 
 
 Washed novaculite 
 
 
 
 
 
 
 
 
 
 gravel 
 
 WN-III 
 
 3% CaO 
 
 3.0 
 
 9.0 
 
 26.4 
 
 25.3 
 
 
 664 
 
 Washed novaculite 
 
 
 
 
 
 
 
 
 
 gravel 
 
 485 
 
 3% CaO 
 
 1.6 
 
 4.9 
 
 25.1 
 
 24.8 
 
 2.29 
 
 1037 
 
 Commercial brick 
 
 
 
 
 
 
 
 
 
 made from quartzite 
 
 
 
 
 
 
 
 
 791 
 
POROSITY CHANGES 
 
 53 
 
 tween these two extremes in porosity, are 
 likewise intermediate in their modulus of 
 rupture values. 
 
 There is a rough relationship between 
 the modulus of rupture and the correspond- 
 ing porosities (unfired) of all the 9-inch 
 experimental bonded novaculite brick. The 
 porosity of the unfired brick has been plot- 
 ted against the modulus of rupture of the 
 fired brick in figure 28. In these brick, the 
 CaO varied between 2.0 and 3.0 percent. 
 They were all given the same heat treat- 
 ment during firing. These data show that 
 the raw porosity, or factors which influence 
 the porosity of unfired silica brick, will, 
 under the influence of the same firing treat- 
 ment, exert a predominating influence on 
 the strength of the brick. The effect of the 
 amount of CaO used as a bond may be ob- 
 served by comparing CP-2 with CP-3 and 
 WN-I with W-II. The only difference 
 between CP-2 and CP-3 and between 
 WN-I and WN-I I, in the raw state, was 
 the amount of CaO, as may be seen from 
 tables 15 and 16. CP-2 with 2 percent CaO 
 was stronger than CP-3 which had 3 
 percent CaO, and WN-I with 2 percent 
 CaO was stronger than WN-II containing 
 3 percent CaO. It is evident from these 
 data that increasing the lime from 2 to 3 
 percent decreased the strength under the 
 given firing conditions. A similar compar- 
 ison may be made between brick WN-II 
 and brick WN-III. The only difference 
 between these brick is that the maximum 
 grain of brick was 8-mesh, whereas in brick 
 WN-II it was 6-mesh. The lime content 
 was 3 percent in each case. Under the same 
 firing conditions, the effect of increasing the 
 flux content was the same as reducing the 
 maximum grain size. The rate of inversion 
 of silica was hastened by increasing the 
 flux content and also increased by decreasing 
 the grain size. Too fast inversion of silica 
 caused weakening of the structure of the 
 brick. This suggests that brick such as 
 CP-3, WN-II and WN-III, could be im- 
 proved by a slower rate of temperature in- 
 crease in the early part of the inversion 
 period (about 1100° to 1150° C). 
 
 Brick No. 485 apparently does not follow 
 these rules, since it is the finest grained 
 
 mixture and is also composed of 3 percent 
 CaO. No batch of this grain size grading 
 was made using 2 percent CaO, so it cannot 
 be said whether the higher strength in this 
 case would result from the 2 percent or 3 
 percent lime bond. It probably would be 
 decreased by decreasing the amount of 
 bond, because the great decrease in the 
 grain size had no adverse influence on the 
 strength of the brick. In former experi- 
 ments it was noted that rapid inversion 
 caused pronounced swelling and cracking 
 in specimens which contained coarse grains, 
 but this did not happen when the speci- 
 mens contained no coarse grains. 
 
 DEFORMATION UNDER LOAD AT 
 HIGH TEMPERATURES 
 
 Bricks CP-2 and CP-3 were heated to 
 1500°C under a load of 25 pounds per 
 square inch, according to the A.S.T.M. 
 method. After this treatment the brick had 
 increased in length 0.03 percent. 
 
 POROSITY CHANGES DURING FIRING 
 
 The porosity, both in the fired and un- 
 fired brick, was computed from the true 
 densities and the exterior volumes of the 
 specimens. In every case there was a 
 decrease in porosity due to firing. The 
 porosity was usually about one percent less 
 atter firing than before. 
 
 The difference in bulk volumes before and 
 after firing was less than theoretical. The 
 volume changes of the various specimens 
 varied between 8 and 9 percent of the per- 
 manent exterior volume expansion from un- 
 fired to fired condition. The true-volume 
 expansion was about 15 percent. This 
 means that there was some sort of shrink- 
 age taking place which counteracted part 
 of the expansion due to inversion. 
 
 It is interesting to compare the volume 
 changes shown in table 16 of brick WN-I, 
 WN-II, and WN-III. No. WN-I with 
 2.0 percent CaO, had a bulk volume expan- 
 sion of 8.0 percent and a transverse strength 
 of 839 pounds per square inch. WN-11, 
 differing from WN-I only in its CaO con- 
 tent, had an exterior expansion of 8.4 per- 
 cent and a modulus of 732. WN-III, simi- 
 
54 
 
 PROPERTIES OF SILICA BRICK 
 
 l_ 10 
 
 o 
 
 z 
 o 
 
 ? 0.6 
 
 a. 
 x 
 
 0.2 
 
 CP-2 
 
 NOVACULI 
 COARSE GR 
 
 CM- 
 
 
 ^ 
 
 \ A 
 
 
 
 
 TE 5-*l 
 
 NOVACUL 
 MINE R 
 NO BOI 
 
 
 ^ "^ 
 
 
 x COMMERCIAL BAR 
 
 ABOO 
 
 
 JN / 
 
 f" 
 
 485 
 
 X NOVACULITE 
 FINE GRAIN 
 
 
 
 \ WN- 
 NO\ 
 INT 
 " ST 
 
 - 1 
 
 'ACULITE 
 
 ERMEDIATE 
 
 ANDARD" 
 
 GRAIN SL 
 
 IE 
 
 
 
 ' 
 
 
 
 
 
 
 * 
 
 
 I 
 
 
 
 
 
 
 
 
 1 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 600 800 I0O0 
 
 TEMPERATURE - DEGREES C 
 
 Fig. 29. — Thermal expansion, percent length versus temperature. 
 
 lar to WN-II but of smaller maximum 
 grain size, had still higher volume expan- 
 sion and lower strength. This is in accord 
 with the conclusion that reduction in 
 strength is caused by rapid inversion pro- 
 moted by increasing the flux content or de- 
 creasing the grain size. 
 
 Brick 485 increased only 4.9 percent by 
 volume during firing. This small volume 
 increase remains unaccounted for, unless it 
 be assumed to be characteristic of the 
 material. 
 
 MICROSTRUCTURE 
 
 The coarse grains are composed of very 
 small crystals of cristobalite. Only rarely 
 was any quartz observed. Some quartz was 
 found in the center of grains that were 
 larger than J inch in diameter. In cross- 
 
 sections made from WN-I, WN-II, WN- 
 III, and 485, no quartz was found. The 
 coarse grains were all cristobalite, while 
 in the groundmass, where the fine silica 
 was in contact with lime, good tridymite 
 development was noted. The tridymite 
 crystals were developed much better in 
 brick 485 than in any other, and much 
 better in the WN brick than in the CP 
 brick. Novaculite brick were consistently 
 lower in quartz content than quartzite 
 brick. The CP brick were rather coarse 
 grained, with a high percentage of very 
 coarse grains and a deficiency of fines. 
 These brick were rather poor in tridymite 
 and rich in cristobalite, and almost free 
 of quartz. 
 
 A peculiarity noted in the microstructure 
 of the novaculite brick was the presence of 
 
THERMAL EXPANSION 
 
 55 
 
 fine spherical specks in the larger grains 
 which had inverted to cristobalite. These 
 specks appeared to be small pores which 
 presumably are related to the slightly 
 porous nature of the raw novaculite. 
 
 THERMAL EXPANSION 
 
 The thermal expansion of some experi- 
 mental novaculite brick and a commercial 
 quartzite brick are shown in figure 29. 
 Brick CP-2 was a coarse-grained brick 
 composed almost entirely of cristobalite. 
 CM-O was an unbonded brick made from 
 crushed mine-run novaculite gravel. Brick 
 WN-I has a thermal expansion very simi- 
 lar to that of a typical commercial brick. 
 No. 485 had a definitely lower thermal ex- 
 pansion than the commercial quartzite 
 brick used for comparison. Under the same 
 firing treatment, the thermal expansion 
 decreases with decreasing grain size. 
 
 Conclusions 
 
 The compact massive novaculite is of 
 sufficient purity to make a high grade silica 
 brick, and may be crushed to yield desirable 
 distribution of grain sizes. The novaculite 
 gravel may easily be washed to remove the 
 associated clay, after which it has proper- 
 ties similar to the massive pure rock. 
 
 The strength of fired silica brick depend 
 largely upon the manner in which the brick 
 is made. The grain-size distribution of the 
 siliceous material determines the density 
 to which the mass can be packed, has an 
 
 important influence upon the strength of 
 the resulting fired brick, and influences the 
 thermal expansion by fixing the ratio of 
 cristobalite to tridymite. The transverse 
 strengths of fired silica brick are roughly 
 proportional to their corresponding poros- 
 ites in the unfired state. Too rapid inver- 
 sion of novaculite causes a weakening of 
 the brick. This may be influenced by the 
 grain size, the amount of bond, or the 
 manner of firing. Novaculite with the same 
 particle grading and same CaO content as 
 commercial quartzite brick yield about the 
 same type of brick as the commercial quart- 
 zite. Novaculite should be fired a little 
 more slowly than the quartzite in the early 
 stages of the inversion. High quality brick 
 were obtained when novaculite was fired 
 according to standard commercial silica 
 brick practice. 
 
 Novaculite brick have a smaller exterior 
 overall expansion than quartzite brick. 
 When made with a maximum grain size 
 of 6-mesh and with 2 percent CaO, the 
 over-all expansion was 8.0 by volume. With 
 a maximum grain size of 10-mesh and 3 
 percent CaO the over-all expansion was 
 4.9 percent by volume. This small expan- 
 sion has been ascribed to a shrinkage of the 
 novaculite itself. 
 
 Better brick were made when fine grained 
 material was used. The extremely fine 
 fraction may be added as Illinois tripoli. 
 For best results with novaculite, it is 
 recommended that a mix similar to the one 
 designated as No. 485 be used. 
 
 Illinois State Geological Survey 
 
 Report of Investigations No. 117 
 
 1946