CEMENT UNDER WATER BY DR. W. MICHAELIS, SR. at Berlin. Germany, March 9, 1909. PRICE, 50 CENTS Published by CEMENT & ENGINEERING NEWS CHICAGO, ILL. TRANSLATED BY DR. W. MICHAELIS, JR. L- O C) Z Z [r] C) CYZ <ſ T [+] In: ſ- ; , ; ; ; ; , . . . ; : s \ , , . . . . # * * * * * * : . § 3 ; i ; | | | ! ; $3;{################ # \ #§§ j # ################ ;: # # ºjº t; #####: #} # ; º : } i. t # i ºrr: * . } ~ s ; i i { # ; # ; i i i ; # # # i # ; i : }. i # : ; ; !; #; :: # º i; # f***** ; f ;” ...” x- ~4 º# } # & # ; §: ; ; : i ; ** | § * # # THE HARDENING OF CEMENT UNDER i | § i # ; : # i | ; {{ i * ::::$3 tº #5 ºtti: ####### # ################| # : # § # • * !}}}: #3; ty #. !: ###### # § # # §§ # # ; : §: - - - * ######## BY § # § # # i i #, #: ;::::: #: ! } #: # {t} - - *** # §§ ...!!!! #: {{:}; ââ ######### } # }}}} *: § k 3.# i ### DR. W. MICHAELIS, SR. # - § :::::###### §§ : # *. # i º # #!!! #### {3|{{#3. -- zz st § ####### # f s § # ? i i # i # f i É ; } # i i § ; § § ; i ; # t } : : # i i -# § ; ; } ; a. # Paper read before the German Portland Cement Manufacturers at Berlin. Germany, March 9, 1909. # # i i i § gºf i y } # i -§> ; i : : # # # ; #- i |- ; ; # g $ ; :- ;- } f ; ; # }§ { i # # --- § #; ##### §§ º #; * § ########### #fff; 3 ** t #### #; § i i § § ; # # # : i } : ; 't ;#i #: TRANSLATED BY DR. W. MICHAELIS, JR. # # # -3 ; §i i- i ; : ; § j i ; º % ; §§§ § # §: §§§ {1}{ §: §§§ §§ - #x334}}}} 24 §# # **** #15 it tº 3. # #####, º #§ it;# ###### ######## ... ????º º º } :*.iº. §§§ #sºtº ### ...?? is * f # §§ §§§º. * # {{ # ; : w # º - *** Mr**:A; * §: # :#rºss; # ### * # ; i # : # ! #t i i *4 # $ºs ; * t A. - *::::::::: *-ºxº :::::: ;: gº §: #: *:::::: - #}; #####: * ºft § # # # }}} * § Published by CEMENT & ENGINEERING NEWS CHICAGO, ILL. £iº - ######### ###### # :::::# *; # # : # - ###### § ***** º: - * * * * : *; §§§ §§ §§§§§: - # : t :::::::: # - # - º #: # # §:::::: {} § t # # :::::::::::::: §§§ # #: # §§§ - : # # # # §§§ ####### # § #:§ # §§§ :::::::: ### ######### º #########| # ########### # § #: § # . ss The Hardening Process of the Calcareous Hydraulic Cements. BY DR. W. MICHAELIS, SR. For about 150 years men of science endeavored in vain to explain the hardening process of the calcareous hydraulic cements. The cause of the futility of their efforts is to be found in the first place in the fact that the reactions, which these calcined products undergo, upon coming in contact with water, do not go on in well-defined proportions, but in uncertain and varying ratios; furthermore, it is due to the phenomenon that the part of the reactions most essential to the harden- ing offers such obstacles to the completion of the pro- cess, that the reaction can practically never come to a finish, so that the hardened product invariably consists of a mixture of decomposed material and of material which remained unaffected. This explains the difficulty or rather impossibility of ascertaining by chemical ana- lysis definite products of the reactions. Sixteen years ago I stated (Chemiker-Zeitung 1893 Vol. XVII, p. 982) “that the calcareous hydraulic cements owe their hardening mainly to the formation of colloidal calcium hydro-silicate, because calcium silicate is absolutely insoluble in water.” This definition I wish to correct and to say “because calcium hydro- silicate is difficult to dissolve and because it possesses too great a velocity of formation, that is to say, because it forms too rapidly out of its component parts.” If pure silica sole (a clear solution of silica in water) is poured into lime-water of a dilution of 1:120,000 (1 part by weight of calcium hydrate to 120,000 parts by weight of water), the mixture of these clear solu; tions becomes opalescent. If the silica sole is poured into lime-water of a concentration of 1: 60,000, a suspen- sion immediately becomes visible upon agitation of the liquid. In lime-water of a concentration of 1:30,000 a flocculent precipitate, a hydrogel, is formed at once. Whenever two bodies in solution come together and unite, no matter whether directly or after previous chemical exchange, the following may happen, in ac- cordance with the general rules established by the Rus- - 3 resee- C +& 2," sian Scientist P. P. von Weimarn as the result of his investigations of the Laws Governing Chemical Reac- tions: 1. The formation of the solid state goes on very slowly in a slightly over-saturated solution. In this case well-formed, large crystals are obtained, which Sometimes require many years for their formation. The more slowly crystallisation progresses, the larger the crystals grow as a rule. 2. The formation of a solid body proceeds quickly from an over-saturated solution: the product, however, is not very difficult to dissolve. In this event, precipi- tation results usually in clusters of needle-shaped crystals. These latter, therefore, are of common occur- rence. The crystallisation of gypsum, for instance, is a good illustration. If gypsum is allowed to crystallize slowly in a slightly Over-saturated solution, it forms large crystals; however, crystallizing quickly from solu- tions excessively over-saturated, it forms small needle- crystals. | 3. The formation of the solid body may be brought on quickly from an excessively over-saturated solu- tion and the product may be difficult to dissolve. Then we obtain coagulation, that is to say jelly-like forma- tions or hydrogels, briefly colloids, which means bodies of the consistency and of the physical properties of glue. - In this last described condition the compound of silica and calcium hydrate is always obtained at low temperatures, that is to say as amorphous calcium hydro-silicate, which is called a hydrogel in the wet state and a gel in the dry state, after loss of water by evaporation or by absorption. At 90 degrees Celsius and above, crypto-crystalline meta-silicate is formed (for instance plombierite, described by Daubree, sand- lime products hardened under high pressure, by Mich- aelis, meta-silicate, described by Dr. Ludwig). Chemical composition CaSiO2+2 aq. At from 700 to 900 Cel- sius macro-crystals of the composition CaF2 SiO4+4 aq. (as described by Robert Bunsen) are formed. At the Stockholm Congress of the International Society For Testing Building Materials in 1897 I ex- hibited finely powdered crystal of quartz transformed into hydrogel by limewater, t 100 degrees Celsius. What I attained by high temperatures at that time, I accomplished in the meantime by increased fineness of the powdered crystals of quartz, namely by using a powder the grains of which were of a diameter of con- siderably less than one micro-millimeter (the twenty-five thousandth part of an inch). This fineness is obtained by grinding a very small quantity of a fine powder in a clean, polished agate mortar for a sufficient length of time; for instance by grinding 20 milligrams for an hour. Such preparations are laborious of course, but are essential for experiments like the decomposing of hydraulic cements by large amounts of water or lime- water after the process practiced by me for many years. If powdered quartz of this fineness is agitated with distilled water, a portion of the very finest powder remains in suspension, because the grains increase their volumes by absorption of water to such an extent that they float. If this milky suspension is boiled for sev- eral days, whereby the evaporated water has to be replaced from time to time, silica hydrogel is obtained.” If lime-water is added to the suspension of finely pow- dered quartz in water in such amounts that 1 part by weight of calcium oxide is present to 3,000 parts by weight of water, coagulation sets in at once under forma- tion of a bulky flocculent precipitate of calcium silicate hydrogel. | - That this coagulation is not merely brought about by surface-attraction can be proven by treating the hy- drogel with dilute hydrochloric acid and subsequent filt- ering. Then the unchanged cores of the minute grains which constituted the quartz powder remain on the filter, while the clear filtrate contains soluble silica, that is to say, silicic acid or better silicium dioxide. The three so-called hydraulic factors of all calcareous hydraulic cements, namely silica, alumina and iron oxide, are al- ways acids, because they act as strong acids at high *The plasticity of clays is principally, if not entirely, based upon their contents of colloidal silica. The process of putrefaction, to which clays are frequently subjected, effects an increase of the percentage of colloidal silica by hydration of the finest quartz powder, especially in the presence of Small amounts of alkalies, which dissolve silica and form “soles,” or with the assistance of ammonia, which is always formed in the course of this process. 5 temperatures and as feeble acids even at low tem- peratures in the presence of calcium oxide which has strong basic properties at all temperatures. All three of them, however, have amphoteric character, which is known for some time so far as aluminum oxide is con- cerned, but is almost unknown of the silicium dioxide. For this reason I propose to distinguish, according to the chemical properties, between silicic acid and silicium dioxide, aluminum acid and aluminum oxide, iron acid and iron oxide (this iron acid Fe2O3 must not be con- founded with ferric acid FeC3). Silicic acid and alumi- num acid become stronger acids in proportion to the in- crease of temperature; at the highest temperatures aluminum acid is even a stronger acid than silicic acid. The latter remark is well worthy of consideration for mineralogists, who believe, in cel tain minerals, chem- ical compounds to exist in which silica is an acid and alumina a base. I consider such “compounds,” at least in most cases, to be mixtures of two fused acids, namely silicic acid and aluminum acid, which solidified upon cooling. - - In order to determine the influence of varying phys- ical conditions and of the size of the grains of pieces of quartz or of crushed or finely ground quartz with regard to the reaction of lime-water or pure water upon it, the following tests were made: • . a) A transparent crystal of quartz weighing 9.024 gram with highly polished faces was placed in a stiff paste of chemically pure calcium hydrate contained in a large platinum crucible and heated in an autoclave at a temperature of 180 Celsius for 48 hours. Upon cooling, the lime-paste was removed and the crystal washed with dilute hot hydrochloric acid, then with a hot solution of soda in water and once more with dilute hydrochloric acid and water. Now it was dried anºl the weight found to be 9.0157 gram. The faces were just as highly polished as before the experiment. Ac- cording to the loss of weight the amount of the dis- solved silica was 0.0083 gram, while the direct de- termination of the soluble silica showed 0.009 gram, that is to say 0.1% of the quartz used for the experi- ment. - - - - b) Another crystal of quartz, likewise chemically pure, but with a ground surface, which, when magnified 50 6 times under the miscroscope, did not show grooves, was treated in the same manner. In this case 12.7945 gram showed a loss of weight of 0.0313 gram, that is to say of 0.25%. c) Fused quartz weighing 8.0917 gram from the firm of W. C. Heraeus with a perfectly smooth surface, which, however, had a clouded appearance rather than a high luster, showed, upon a similar treatment, a dull surface and lost in weight 0.0132 gram, hence 0.16%. d) A pure transparent crystal of quartz was trans- formed into a powder which left no residue on the 100 mesh sieve (100 meshes per lineal inch) and 30% on the 200 mesh sieve. This was mixed with lime paste in such proportions, that the mixture contained 92 parts by weight of powdered quartz to 8 parts by weight of calcium oxide. This plastic dough yielded, upon heating for 24 hours at 180 Celsius in an autoclave, a petrified mass, which chemical analysis proved to contain 11.433 parts by weight of soluble silica. Thus 12.4% of the quartz used for this experiment were converted into soluble silica. y e) The same quartz crystal was pulverized so finely (only 20 milligrams were ground at a time for one hour) that the largest grains measured less than one micro-millimeter. This powder, stirred up with 6 quarts of distilled water, resulted in a milky liquid in which, owing to direct water absorption, part of the quartz powder was kept in suspension for many days. After being boiled for several days, whereby the evaporated water had to be replaced from time to time, gelatinous silica was the result. Quartz silica in so fine a state of subdivision, therefore, combines directly with water and yields colloidal silica even at low temperatures, more readily of course, and more bulky flakes result, at the boiling point of water. It is well-known that a definite separation of quartz silica from amorphous silica is impossible. For in- stance, a solution of soda slightly dissolves finely pul- verized quartz. At the thirty-first annual meeting of the Association of German Portland Cement Manufacturers (in 1908), Dr. Cabolet of Hemmoor stated that in his experiments. in which commercial Portland cement was acted upon 7 by large amounts of water, for instance 1 part by weight of cement to 500 or 1,000 parts of water, swell- ing of the grains of the cement powder, that is to say formation of hydrogel, commenced only, when the lime- solution, ensuing from the decomposition of the cement by the water, showed a concentration of 1:3300, in other words that a certain amount of calcium hydrate had to be dissolved in the water before colloids were formed. It was, therefore, of great interest to ascertain the smallest amount of calcium hydrate the water was re- quired to contain in the case of silica, that is to say, to determine what minimum concentration formed the limit for the formation of colloidal calcium hydro-sili- cate. This object was attained in the following way: Of the commercial silica half-hydrate (a preparation known as silica via humida parata sicca, containing from 12 to 13 per cent of water)* an amount equal to 1 gram silicium dioxide was weighed off. This light, finely divided powder was agitated several times a day with about one liter of lime-water of five different concen- trations, namely with lime-water containing 1 part by weight of calcium oxide to 5230, 3860, 1630, 1340 and 866 parts of water. Each of these solutions was al- lowed to act upon the silica for 90 days. The total amount of calcium oxide adsorbed in the course of this treatment was 1.0253 gram. But even then the adsorb- ing power was not completely exhausted, as during the last two weeks still 0.01613 gram of calcium oxide were adsorbed out of the last, almost saturated, lime-solution. Hence a hydro-silicate was formed which contained 10 molecules of silica to 11 molecules of calcium oxide, that is to say slightly more lime than required for a mono- calcium hydro-silicate. During the first treatment, with lime-water of a con- centration of 1:5230, the calcium oxide adsorbed by the silica amounted to 0.072 gram. No visible swelling occurred in this case. The solution showed, after ad- sorption had taken place, a concentration of 1:12.000. After the second and third treatment 0.612 gram of calcium oxide was found to be adsorbed, whereby the *As hydrated silica, upon drying out, shows a continu- ous loss of water, genuine hydrates of it, do not seem to exist. 8 fine particles of the powdered silica had considerably increased in volume by swelling. The remaining lime- solutions had a concentration of 1:9160 an 1:9870. By the fourth treatment 0.325 gram was adsorbed; the lime- solution which remained was 1:1350 and the remaining solution after the fifth treatment showed a ratio of 1: 878. Thus the limit at which swelling occurs in the treatment of powdered silica with lime-solution lies be- tween the concentrations of 1:10.000 and 1:12,000, while, as we saw in the beginning, formation of hy- drogel in the case of silica in solution took place at a concentration as low as 1:60.000. If one gram of finely pulverized Portland cement, of which one cubic centimeter weighs about 0.8 gram, is agitated with two liters of distilled water, boiled im- mediately before use, the free sulphuric acid of whicl: must be neutralized with a cubic centimeter of satur- ated lime-water (the commercial aqua distillata contains as a rule a small amount of chlorine, calcium oxide and Sulphuric acid and has a slight acidous reaction), water is adsorbed by the cement: however, no swelling becomes visible, no hydrogel is formed. 0.336 gram of calcium oxide are found in solution, that is to say 0.168 gram per liter. This concentration, 1:6000, is not sufficient for the formation of hydrogel. Upon frequent agitation for 6 or 8 weeks of one gram of cement with twice the amount of water, 4 liters, whereby likewise great care was taken to prevent ad- hering and sticking together of the minute grains, 0.402 gram of calcium oxide was found to have gone into solution hence 0.1005 gram per liter. This repre- sented a lime-solution of 1:10.000, in which swelling could not take place. In the former case 53 per cent, in the latter 64 per cent of the total amount of calcium oxide contained in the cement went into Solution. Consequently nearly mono-calcium compounds remain, which are difficult to decompose by water and more so if the water contains calcium hydrate in solution. The lower the amount of lime in the calcium hydro-silicate, calcium hydro-aluminate and calcium hydro-ferrite, the more firmly the lime is retained by the acids. If now the finely powdered cement of the last two experiments is allowed to settle, after prolonged treatment with a 9 ; : :”. . .” . . . * * -s * * gº º 'º * º ... * * * Q large amount of water, which decomposed it, but did not cause formation of hydrogel on account of the lime-solution being too weak, and if the clear lime-solu- tion is poured off and replaced by a more concentrated Solution containing the same amount of calcium oxide, for instance by 300 ccm of saturated lime-water, forma- tion of hydrogel in large flakes results immediately upon agitation of the previously decomposed cement. A cer- tain, but small amount of calcium oxide is adsorbed by the decomposed cement out of the concentrated lime- Solution. This amount, however, is only a small per- centage of the material subjected to the experiment. Of the from 33% to 40% of calcium oxide extracted from the cement by the treatment with water, only from 3% to 5% are absorbed again. This shows that for the formation of colloids only a very few per cent of calcium oxide are required, a fact which must be well borne in mind for a clear understanding of the hardening pro- CéSS. - Opponents to my theory argued that Portland cement treated with so large an amount of water no longer re- mained Portland cement. To this I will reply that a cement is decomposed in any event, hence does not re- main the original product, no matter whether acted upon by an equal amount or by a thousand times larger amount of water. Only by the virtue of this decomposing process it is able to harden and to become a cement, that is to say a binding substance, a “mineral glue.” Of course, the small amount of water used in practice for gauging cement decomposes it only partially; only part of the cement powder, the very finest particles and the surface of the larger grains are acted upon at all. The larger the amount of water, the more completely a ce- ment is decomposed, if care is taken that the grains do not stick together and become enveloped in the hydro- gel formed at once by the decomposition of the smallest particles. Even if a thousand times as much water as ce- ment is used, it can be observed that the inside of min- ute cement grains remains unaffected, because it is pro- tected from the water by the hydrogel surrounding it, which latter results from the decomposition of the out- side of the grains. For this reason it is well-nigh im- possible to determine by chemical analysis the composi- tion of the hydrates formed. 4 - e.” * * 19 " * ~ * * * ºr ar If the decomposed Portland cement of the last des- cribed two experiments, after settling at the bottom of the flask, and after decantation of the solution contain- ing the extracted lime, instead of being agitated with a Saturated lime-solution, is allowed to remain at the bot- tom and if the concentrated lime-water is carefully poured upon it without disturbing the sediment, the grains of the latter do not swell to the extent of forming large flakes, but are cemented together owing to the adsorption of lime. The brown color of the sediment turns lighter and lighter and finally resembles pure cal- cium hydrate, that is to say it becomes as white as ST1CW. The hydrates of alumina and iron oxide with 3 mole- cules of water (R2O3, 3H2O), if agitated with saturated lime-water, yield crystalline precipitates as a rule, be- cause the compounds which result, namely tri-calcium hydro-aluminate and tri-calcium hydro-ferrite, form slowly. They are minute thin hexagonal plates. In the case of excessively over-saturated solutions, also hydro- gel may result, that is to say colloidal calcium hydro- aluminate and calcium hydro-ferrite. The tri-calcium hy- dro-aluminate as well as the double compound of the lat- ter with 3 molecules of calcium sulphate are almost in- soluble in concentrated lime-water. Only the highest hyd- rates of alumina and iron oxides are capable of harden- ing with lime-paste. The de-hydrated oxides, the mono- hydrates and di-hydrates do not harden with it, or har- den so slowly that practically no advantage can be de- rived from this process. In filtering the clear diluted lime-solutions in the course of the experiments above described, namely after treatment of Portland cement with a large quantity of water, a person can not fail to observe, on account of the slow filtering, that the solu- tion passing through the filter is not pure lime-water, that is to say not a genuine solution, but a sole. This must be ascribed to the fact that owing to the small amount of alkalies in Portland cement (on an average 0.8 per cent), alkali-silicate goes into solution at the same time with the lime. This, of course, happens only in the event of lime-solutions so diluted that the limit, at which formation of hydrogel begins, is not reached. However, upon addition of a sufficient amount of con- centrated lime-water to the clear filtrate until half-satur- 11 ated lime-water is obtained, opalescence sets in and, up- On further concentration of the lime-water, the amount of Silica carried into solution, although very small, is precipitated in flocculent form as colloidal calcium hydro- silicate. Thus, through the decomposition of Portland cement by so large an excess of water, calcium oxide is extracted from the cement, which, in the calcined state, represents an Over-Saturated solution of lime in a mixture of vari- ous fused compounds that solidified upon cooling; there- by the amount of the acid constituents, that is to say of the anions, or rather formers of anions, has been in- creased in the insoluble part, so that we may consider this hydrated residue the electro-negative part and the calcium oxide the electro-positive ion. Upon sufficient concentration of the lime-solution, the calcium oxide, the cation, travels to the anions, or both ions travel one to the other. As they combine very rapidly and as, furthermore, the compound formed is almost insoluble, formation of colloids is the result in accordance with the rules of von Weimarn referred to in the beginning. That the concentration of the lime-water in this case must be 1: 3300 before formation of hydrogel takes place, whereas for silica in solution a concentration of 1: 60,- 000 and for very finely pulverized silica 1: 10,000 were found to be sufficient, is due to the counteracting influ- ence of the alkalies. These act in opposition to the lime, because they form soles, that means carry the silica into solution, if the water contains a certain small amount of alkalies. The calcium hydrate, on the other hand, is the most effective means of precipitating silica and of converting it into hydrogel and, in the course of time, into a gel. Moreover, the hydrolysis has to be overcome in the case of the reaction of 1000 parts of water upon 1 part of cement. - A Roman cement of excellent quality, the natural ce- ment from Peissenberg, which contains 23.9% of silicic acid, 9.7% of aluminum acid, 4% of iron acid and 47.2% of calcium oxide, besides a small amount of gyp- sum and other ingredients, was agitated with water in the same way as the Portland cements formerly des- cribed. After a prolonged treatment of one gram of this cement with 3000 ccm of water, 8.4% of calcium 12 oxide, that is to say 0.084 gram, were found to have gone into Solution. After this decomposition had taken p'ace, the insoluble residue was agitated with a lime-solution containing 0.28 gram of calcium oxide per liter. Yet, no formation of hydrogel became noticeable and no lime was adsorbed. However, upon being agitated with lime- water, containing 0.4816 gram calcium oxide per liter, considerable swelling occurred accompanied by adsorp- tion of 0.0504 gram of calcium oxide. Hence, only an amount of lime equal to 5% of the original amount of cement was adsorbed. In the case of a Hydraulic Lime, Chaux Lafarge 1, which was treated in the same way in the proportions of 1 gram to 3 liters of boiled distilled water, previously carefully neutralized by a small addition of lime-water, 0.3612 gram calcium oxide went into solution. No swell- ing occurred, because the solution contained only 0.1204 gram calcium oxide per liter. Then the concentration of the lime-water was increased to 0.2311 gram calcium oxide per liter; also in this case no hydrogel was formed and no calcium oxide adsorbed. Only after the contents of lime was increased to 0.439 gram calcium oxide per liter, swelling set in. The amount of lime adsorbed was 0.0319 gram calcium oxide, that is to say only about 3 per cent of the quantity of hydraulic lime used for the . experiment. ...-- From this can be seen that, after hydrolysis has tak- en place, in all cases approximately mono-calcium com- pounds remain, which appear to be very stable. Here- by the contents of anions or formers of anions in the residue increases. The insoluble residue, therefore, is charged electro-negative after the cations, calcium ox- ide, potassium oxide and sodium oxide have been ex- tracted partially or entirely. As soon as the concentra- tion of the lime-water has reached a certain degree, the calcium cations travel to the anions, about in the same manner as iron filings are drawn to a magnet, or the ions are attracted one by the other. Calcium hydro-silicate forms rapidly as almost insoluble compound, which is invariably precipitated as hydrogel, whereas calcium hy- dro-aluminate and calcium hydro-ferrite, which are slightly more soluble, form more slowly and take the shape partly of minute crystalline bodies and partly of hydrogel. 13 After this preliminary study of the decomposition of cements by water and of the action of calcium oxide upon the decomposed residue, let me try to explain the setting and hardening process of Portland cement. As an illustration we may use a group of three grains of calcined cement which are represented in the cuts as spheres magnified many thousand times, although they are in reality irregular, sharp-cornered, more or less porous little grains. To these three grains of Portland Hig. 1. cement we add, in accordance with actual practice, the customary 2 per cent of gypsum represented in Fig. 1 by a small sphere. We have now to consider the effect of one-third of its weight of water or of about the same volume of water upon this cement: As said before, the calcined cement represents a “solid solution” over-satur- ated with lime, that is to say a mixture of fused com- pounds of silicic acid, aluminum acid and iron acid with calcium oxide, in which a certain percentage of uncom- 14 bined calcium oxide is partly dissolved and partly en- closed; this Solution became solid upon cooling. The water with which the cement is gauged immediately dis- Solves calcium oxide, calcium aluminate, calcium sul- phate, alkali-silicate and possibly also a small amount of calcium ferrite; thereby an over-saturated solution is formed which contains calcium oxide, tri-calcium aluminate, tri-calcium ferrite and calcium sulphate. The last named salt combines with the tri-calcium alum- inate and forms calcium sulpho-aluminate as long as gypsum is available, since every per cent of alumina requires 5 per cent of gypsum for the formation of this double-compound. If alite happens to be in a cement clinker, a crystal formation occurring in slowly cooled clinker, while my fused and quickly cooled Portland cement did not contain any alite crystals at all, this alite is decomposed by the water at once. Calcium oxide and calcium aluminate go into Solution, whereas silica remains insoluble in the form of hydrate or hydro-silicate low in lime, because it is absolutely insoluble in lime- solutions, even if the latter are very diluted. Thus the Small amount of water with which a cement is gauged is immediately transformed into an excessively over-satur- ated Solution of various salts which react one upon an- other. The resulting compounds, however, owing to their insolubility in concentrated lime-water, form imperfectly built clusters of needle-shaped crystals radiating from various centers. Owing to the slow formation of the calcium hydro-aluminate and of the calcium hydro-fer- rite these latter occur also in hexagonal minute plates. This is as far as the process of crystallization goes which is identical with the setting and hardening pro- cess of gypsum (see Fig. 2). By this process setting and hardening of a cement may possibly be effected, but by no means can a hydraulic cement, that is to Say a water- resisting mortar, be formed in this way, because the water surrounding a mortar made up merely of inter- locking crystals, as in the case of hardened gypsum, is able to penetrate the hardened cement at all times. The water, therefore, would be in a position to soften, dis- solve, and completely destroy the mortar. The absolute proof for the fact that this crystalliga- tion is not the hydraulic hardening process proper is 15 to be found in the observation that the silica never takes a part in this process of crystalligation. We know from experience, however, that for the formation of a a 35%, Fig. 2. hydraulic mortar no other elements are required, but silica and calcium oxide. The same silica, via humida parata sicca, with about 13 per cent of water, or the mono-hydrate containing 23 per cent of water, which are transformed into hydrogel, into a thick gelatinous mass, when in contact with lime-water, result in a hardened colloid or gel, if mixed with lime-paste in the same man- ner as a puzzuolana. The silica resulting from chemical analyses, that is to say silica precipitated in the wet state and calcined over a blast-lamp at 1400 Celsius, at the same temperature at which Portland cement is burnt, if thoroughly mixed with lime-paste and dried be- tween towels or filter paper in order to remove the sur- plus of water, yields as hard a gel as neat cement mor- tar does. By this has been proven long ago that col- loidal calcium hydro-silicate, calcium silicate gel, is able to attain great hardness and strength. - . ... 16 We know that, in the presence of a sufficiently concer- trated lime-solution, calcium hydro-silicate can only be formed as colloid, as hydrogel, on account of the rapid- ity with which it forms and because of its insolubility. Furthermore, we know that also calcium hydro-alumin- ate and calcium hydro-ferrite are precipitated as hydro- gels from excessively over-saturated solutions Owing to the difficulty with which they dissolve in concentrated lime-water. The faster over-saturated solutions are created, namely with increased fineness of the cement and by using a minimum of water for gauging, and the Sooner the calcium compounds of silicic acid, aluminum acid and iron acid form, which are all nearly insoluble in lime-water, the more colloids and the less crystalloids are formed. All these points support my theory that the formation of colloids is the main factor in the hardening of hydraulic cements. After the decomposing and dissolving influence of the water upon the freshly gauged cement has lasted for Some time and after the previously described com- pounds have been formed, a point is reached very soon at which, all of a sudden, the over-saturated solution surrounding the cement grains coagulates. This is the point of time, when the water disappears on the cement pats and when the setting cement attains a dull surface, briefly the time when the setting proper has taken place. The hydrogel formed is very low in lime in the begin- ning and acts in a similar way as the silica hydrogel; by loss of water it dries out and forms a solid, opales- cent mass, a dense colloid. In time, more and more cal- cium hydrate enters this colloid by adsorption, because the hydrogel permits only the water to percolate, but retains the lime (semi-permeability); with the grow- ing adsorption of lime the hydrogel becomes more and more dense, unplastic and solid. - In Fig. 3, the formation of hydrogel is given by the dotted lines. The crystals are now imbedded in the colloidal mass in the same way as splinters of wood or bristles of a brush would appear interspersed in a stiff glue. In the beginning the hydrogel contains a large amount of water and, therefore, is soft. The inside of the grains, however, extracts the water from the hy- drogel very readily and hydrates in the course of time. 17 The water, therefore, enters into the interior of the grains, whereby the hydrogel on their surface solidi- fies and becomes a perfectly impervious solid colloid. (see Fig. 4). The same observation can be made with * * * e * * & *s • * * * * * g * * * . . . . - ... ... • * * * * • * * * * * * * * w * e * * * * * * * * * * • * > . . . .es • * * *es e º z_s v. * * clay mud, which becomes an absolutely waterproof mass, if dried out on a water-absorbing plate or pad. This water-absorption from within explains the hardening un- der water and makes possible the steady increase in Strength of the mortar. To what degree the calcium hydro-silicate gel hardens by loss of water has best been proven by Frederick Ransome's process of manti- facturing artificial sandstone, by which 90 parts by weight of pure quartz sand, consisting of grains of vari- ous size, are mixed with concentrated Sodium water- glass in such quantities that 5 parts by weight of Solu- ble silica are contained in it. This mass is moulded and then immersed in a correspondingly strong Solution of calcium chloride, in which it hardens by the formation 18 of colloidal calcium hydro-silicate originating from the reaction between sodium silicate and calcium chloride, The sodium chloride likewise resulting from this re- action is washed out by running water or in “rain chambers,” and the artificial stone is dried in the air. This process yields an excellent sandstone of a com- pression strength of from 7,500 lbs to 9,000 lbs. per square inch. A mortar made in the proportions of 1 cement to 9 sand from the best Portland cement will never show more than a small fraction of this strength. • e s = ** - p • s • * º ** - * ee * -.' . º *...* * * * ** * - - * - ** ~. ^. * º - .*.*. */A •, & Sº..."-- *N. * * e • * ~ * • •'. .: th sa §s ". º e w * .7% w S Ş. ". * * * * * • * Nº s = ". • * * * * e * t tº a º "... F. * **, *. * * * Aº .* * & $ e ** Y **'. * & . . ** * !\! * * | º * } #. * º - : B ;: §§ Fºš: ; ; R} { ; : ; : , ; * e * RSX º * t : , t * * * º g §§§ J.E. a º V A ...' . . *s • ,” .” ,' * * * d & ,’ * a s º s * & *, * *: X g ºr & .." • * - º & & º e * • * • e $ - * © * º st:Erºs se” W * , " " ºf A //y///ff/F---... : : 2:- .* * * * * * * º * º 4'- s: * * * * º 2 * º - - e. e *:: ... • * * • * a • * - * ~~ * * * * sº • * * * * 423 º e w & ** ~ * 4 e º 'º' • * Cl - - Fig. 4 18. During the hardening process of cements, the strength obtained by the hardening colloids is further increased by crystallization of calcium hydrate, calcium hydro- aluminate, calcium hydro-ferrite and also calcium sulpho- aluminate, in case enough gypsum is available in the ce- ment or is contained in the surrounding water; even 19 large crystals of calcium sulphate are sometimes found. All these crystals are formed slowly from slightly over- Saturated solutions. In hardened Portland cement only tri-calcium hydro- aluminate and tri-calcium hydro-ferrite are to be found. In hydraulic cements lower in lime, for instance in Roman or natural cements also sesqui- and di-calcium compounds are liable to occur. The tri-calcium hydro- aluminate I obtained 30 years ago by dissolving in water any fused compound of aluminum acid and calcium oxide containing less calcium oxide than tri-calcium aluminate would contain and by subsequent admixture of an ex- cess of concentrated lime-water; furthermore, it was ob- tained by violent agitation of precipitated and thoroughly filtered colloidal alumina hydrate with a large excess of lime-water. In both cases invariably within 24 hours hexagonal plates are formed which become noticeable, even if they are very small, through the peculiar reflec- tion of the light resembling satin, if the solution is agi- tated. The tri-calcium hydro-ferrite I obtained in the same way from iron oxide hydrogel precipitated cold and thoroughly filtered. The formation of this calcium hyd- ro-ferrite is the cause of the disappearance of the rust from iron imbedded in lime mortar or cement mortar, which after a short time looks as if it was sand-papered or cleaned with an acid. About 40 years ago I burnt as the first, as far as I know, barium cements by replacing the carbonate of lime by witherite, barium carbonate. The result was a cement much stronger than Portland cement on account of a higher specific gravity. A few experiments, however, convinced me in a short time that this barium cement was not a hydraulic cement; it was too soluble and was softened by water, because it owed its hardening mainly to a process of crystallization in the same way as gyp- S11111. * The barium cement, therefore, is the best proof of the fact that hardening can take place also without swelling or formation of hydrogel. But it proves, fur- thermore, that without formation of colloids IIO harden- ing under water is conceivable. - 20 - Another feature characteristic for both, barium hyd- rate and strontium hydrate, which distinguishes them from calcium hydrate, is their action upon silica. Solu- tions of barium hydrate or strontium hydrate do not transform silica into a colloidal hydro-silicate, but into crystallized compounds. Moreover, no formation of hydrogel takes place upon agitating Portland cement previously decomposed by large amounts of water, that is to say the puzzuolana-like insoluble residue, with Solutions of barium hydrate or strontium hydrate. The experiments described in the foregoing are a fur- ther confirmation of what I declared in my paper read at the thirtieth annual meeting of the Association of Ger- man Portland Cement Manufacturers (in 1907), namely that the formation of colloidal calcium hydro-silicate, and to a lesser degree also the formation of colloidal calcium hydro-aluminate and calcium hydro-ferrite, is the only essential and important feature in the hardening process of all known calcareous hydraulic cements. The basis of all hardening under water is the formation of colloids, which originate from the reaction of a sufficiently con- centrated lime-solution upon a puzzuolana. The water with which the cement is gauged decomposes the Sur- face of the cement grain and extracts calcium oxide. This leaves the outside of the grain overloaded with formers of anions, especially with silica, and thus brings on formation of hydrogel. In the case of puzzuolana proper, trass, Santorin earth, Roman and Neapolitan puzzuolana, the drying out of the hydrogel takes place by the hydration of the puzzuolana. If, for instance, trass is thoroughly mixed with lime-paste So that each grain of trass is sur- rounded on all sides by lime-paste, that is to say that it is imbedded in a hydrogel, namely lime-paste, the lime- solution of the lime-paste reacts upon the anions of the powdered puzzuolana; for puzzuolana is a volcanic slag granulated by water and steam, a product rich in anion- formers and of electro-negative character in the presence of the calcium hydrate-cations. This volcanic slag, furthermore, contains hydro-silicates, that is to say products formed after granulation by steam and water had taken place, which can easily be decomposed, for instance analcime, natrolithe, phillipsite. All these 21 harden readily with lime-paste, the calcium hydrate taking the place of the alkali. Through the absorption of the water of the lime-paste by the gradually hydrat- ing puzzuolana the lime-paste becomes more and more dense and, in time, a solid colloid, while on the surface of the puzzuolana grains the hydrogel solidifies, which originates from the reaction of the lime-solution upon the electro-negative elements of the puzzuolana, in the same way as we observed in the case of the Portland cement grains. The conclusion to be drawn from this with regard to the application of puzzuolana mortars is the advisability of using lime-paste in the manufacture of puzzuolana mortars in preference to the dry powder of calcium hydrate. The use of the dry calcium hydrate must be considered as detrimental, because it sacrifices one of the main advantages of the puzzuolana mortar, namely its imperviousness to water before Setting takes place, a property that it owes to the colloidal calcium hydrate and which prevents it from being softened and washed away upon immersion in water. For about 100 years, since Vicat's time, it is known that alumina-Silicates (clays and shales used by Polhem in the construction of the locks of the Goeta Canal in 1700), if de-hydrated at beginning red heat, yield very valuable artificial puzzuolanas. Upon expulsion of the water dissociation takes place; silicic acid and aluminum acid are liberated and each of them becomes capable of combining with bases. The aluminum acid, if previously heated at a low temperature is capable of combining with calcium hydrate and of forming tri-cal- cium hydro-aluminate which crystallizes in hexagonal plates. The silicic acid forms in contact with lime-solu- tion colloidal calcium hydro-silicate of varying composi- tion. As many as 3 molecules of calcium hydrate can combine with 2 molecules of silica. For instance, pure white kaolin is a good illustration; if de-hydrated at from 600 to 700 Celsius, pulverized very finely and separafed from the coarse by careful levigation and then agitated with a surplus of lime-water (for instance 1 gram de-hydrated kaolin with 3 liters of concentrated lime-water frequently agitated in the be- ginning in order to prevent sticking together of the 22 grains), in the course of one or two years, almost the total amount of aluminum acid can be transformed into the aluminate A12O3, 3CaO,8H2O (Michaelis) or A12 O3,3CaO, 10H2O (Dr. Gino Gallo). By heating at higher temperatures the ability of the aluminum acid of combining with lime in aqueous solutions is made more difficult and ultimately becomes impossible altogether. At the highest temperatures aluminum acid and silicic acid combine, so that even the latter does no longer com- bine with calcium hydrate. For the calcium hydrate it was demonstrated in the foregoing that a certain amount had to be present in the solution in order to overcome the hydrolysis and the dissolving influence of the alkalies, which latter work towards the formation of soles, and to bring on the formation of colloids, principally of calcium hydro- silicate. For magnesium hydrate, which dissolves only in 55,000 parts of water at ordinary temperatures, this required concentration can never occur in water under atmospheric pressure. Consequently no one has ever observed puzzuolana hardening, swelling and formation of hydrogel caused by magnesia, and the statement that dolomite is more effective than pure lime in the manu- facture of puzzuolana mortars is, therefore, erroneous: on the contrary the magnesia in it is merely ballast; it can form almost insoluble magnesia hydrate and finally magnesium carbonate, which latter is inferior to cal- cium carbonate, because it is more soluble. Only at high temperatures, in water under high pressure, can colloidal magnesium hydro-silicate be formed, or by chemical ex- change between soluble magnesium salts and soluble silicates, for instance sodium silicate,water glass. If the assumption is correct, that formation of col- loids takes place only, if lime-water of a sufficient con- centration acts upon silica in solution or upon com- pounds of soluble silica containing an excess of electro- negative elements, swelling, formation of hydrogel, ought to be out of the question, if concentrated lime- water is allowed to react upon lime-saturated or even with lime over-saturated compounds, for instance Port- land cement. Experiments carried on in this direction showed in fact that the same Portland cement which, * , , - 23 tº • * • ‘ • * * * * , * - * * , * 9 * : * * * --w * . ^ - is º * : * , ** upon decomposition and extraction of its surplus of lime by a large amount of water, exhibited considerable Swelling, formation of hydrogel, when brought in con- tact with a half-concentrated lime-solution, did not pro- duce any signs of swelling, even if agitated for many months with saturated lime-water, 1 part of calcium hydrate to 1000 parts of water, or showed only a trace of hydrogel formation, if the cement contained a small percentage of alkali-silicate. - The hydraulic lime from Teil, Chaux Lafarge 1, and the Roman or natural cement from Trifail, upon being agitated with saturated lime-solutions, did not show any signs of formation of hydrogel. Finely pulverized sin- tered di-calcium ferrite, after having been agitated with concentrated lime-water for several months, was found to be hydrated, but did not form hydrogel, nor did it adsorb any lime; it, therefore, represented a compound saturated with lime. However, upon decomposition of the di-calcium ferrite with 1000 times its weight of distilled water, which extracted a certain amount of calcium oxide, and upon subsequent agitation of the residue with concentrated lime-water, swelling and for- mation of hydrogel set in. Likewise fused and disin- tegrated di-calcium silicate showed no swelling, no ad- sorption of lime, when agitated with concentrated lime- water. Only after decomposition with 1000 parts of distilled water and extraction of some of the lime from the surface of the minute grains, which therefore possessed a high percentage of silica, did formation of hydrogel occur upon addition of concentrated lime- water. The assumption that only electro-negative free ions of a certain minimum concentration produce Swell- ing, that is to say formation of hydrogel, is proven by all these experiments. In opposition to this it may be said, that, if this were true, Portland cement could not harden with lime-paste, whereas such a mixture hard- ens very well as daily experience shows. The answer to this possible objection is that in the case of a Port- land cement-lime mortar only hardening of the cement by formation of crystals enters into play and that the colloidal lime-paste offers the resistance to water, which otherwise the colloidal calcium hydro-silicate would do. t ºf . ** • 's * • . . . . " - 24 - º * t t." ºt A mixed mortar of this kind, however, is much more easily decomposed by the water, because the colloidal calcium hydrate alone is soluble in less than 800 parts of water, that is to say by far more soluble than the calcium hydro-silicate. Yet, the case is not as serious as it appears, because formation of calcium hydro-sili- cate sets in, if the colloidal calcium hydrate should be dissolved, namely as soon as lime-water acts upon the cement grains the surface of which was previously de- composed by water, whereby the contents of silicic acid, aluminum acid and iron acid increased. Of course, the mortar would become porous by the washing out of the calcium hydrate or in other words it would be less im- pervious to water. - That the foregoing explanations of the hardening pro- cess, as illustrated by the accompanying cuts, are by no means solely theoretical deliberations, has been proven by microscopical study.* However, the formation of colloids does not become visible in the first stages of the hardening, because the thin films representing the cell-walls of the gelatinous foam-like mass are perfectly transparent in the beginning. In the course of time these walls collapse, dry out and accept irreg- ular forms. Only then the colloid becomes opaque and can be clearly distinguished. According to the amount of water used in gauging the cement, the hydro- gel which forms contains more or less water and, as a consequence, is more or less dense. Its hardening, and at the same time the hardening of the hydraulic mortar as a whole, is solely effected by absorption of water from within, if the mortar is kept under water, that is to say by extraction of the water by the layers of the cement grain next to the surface which gradually become hydrated. The hydrogel covering the surface of the cement grain becomes denser and more solid by *H. Le Chatelier: Annales des Mines, 1887. II 345. A. Martens: Mitteilungen der Koeniglichen Versuchsan- stalt, 1897, p. 89. . A. E. Toernebohm. Ueber die Petrographie des Portland- zements, Stockholm, 1897. E. Stern: Chemiker Zeitung, 1908, No. 47 and No. 85. H. Ambronn. Tonindustrie-Zeitung, 1909, p. 270. 25 this process of water-absorption until a degree of den- sity is reached where it becomes perfectly impervious to water. As an illustration of this process I men- tioned in the foregoing the imperviousness of clay mud dried out on a water-absorbing plate. The hydrogel is finally transformed into a solid gel and thereby the inside of the cement grain is enveloped in an air-tight and water-tight coating, which excludes further action of the water or of the carbonic acid of the atmosphere upon it. The unaffected core of the cement grain, therefore, represents an inert mass imbedded in the mortar. This explains the phenomenon, to which I called attention 30 years ago, that a neat cement mortar, which was kept under water for several years, if once more finely pulverized, can harden again and obtain the higher a strength the coarser the cement was when gauged with water in the first place. Through the re- grinding of the previously hardened cement the inside of the grains, which was not accessible to the water and therefore remained undecomposed became hydrated when gauged with water for the second time and thus could supply new cementing substance. In my paper on the “Relative Value of Portland Ce- ment” published in 1876 I demonstrated that Portland cement, which had hardened under water for four years, after being pulverized and made into briquettes for a second time, showed a tensile strength of 75 1bs, per square inch after immersion in water for 7 days, of 120 lbs. after 30 days and of 133 lbs. after 90 days. - - These briquettes were finely pulverized, after having hardened under water for several years, and showed, upon being gauged with water for a third time, still a considerable strength. Before I introduced the 200 mesh sieve into the ce- ment industry, at a time when the commercial cement left a residue of at least 30 per cent on the 100 mesh sieve, the amount of the cement which did not take an active part in the hardening was very much larger than at the present time of course. At that time hardly 33 per cent of the cement was utilized, and yet my demand for fine grinding was bitterly opposed. I advocated for 26 a mortar of 1 cement to 3 sand a tensile strength of at least 75 lbs. per square inch after immersion in water for 7 days and 120 lbs. after 28 days. This was found unreasonable at that time. If a cement hardens in air, the hardening of the hydrogel is brought about by absorption of the water by the inside of the cement grains and by evaporation of water into the atmosphere. This loss of water in- creases the strength by complete transformation of the hydrogel into Solid gel; however, rapid evaporation of water is accompanied by shrinking which manifests it- self in all cases in which mortars exposed to the air do not contain enough sand or are gauged with a surplus Of water. - - At places, where the hydrogel is not dense enough, for instance in the hollow places in Fig. 3 and 4, or where by evaporation pores and interstices have been formed, water and salt solutions can enter and exert their dis- solving and decomposing influence. Through the inter- stices, therefore, which were filled with hydrogel of too high a percentage of water, water and salt solutions (sea water) can further penetrate. Calcium oxide and, furthermore, aluminum acid and silicic acid can be car- ried into solution. These form, in the case of dormant water large crystals of calcium hydrate, calcium hydro- aluminate, calcium sulpho-aluminate and sometimes also of calcium sulphate. Or in active waters, if a continu- ous current of water percolates through the mortar, all these compounds are washed out, whereby the mortar becomes honey-combed. Finally even the dense coating protecting the cement grains, the Solid gel itself, may be decomposed and dissolved, so that the cores of the grains, which so far remained intact, are unprotected and are litrewise exposed to the decomposing and dis- solving influence of the surrounding liquid. The result may be that at last the whole cement is decomposed in- to its cornponent parts, that all calcium oxide is washed away and that only a flocculent mass of colloidal silica, alumina and iron oxide remains, which may still be held in place by the pressure exerted by the Overlying load and by the outer crust of the construction which became solid by the formation of calcium carbonate. 27 This state of affairs has frequently been observed at large blocks immersed in sea water. Some violent shock is sufficient in a similar case to make the entire con- struction collapse at a single blow. Or it may happen that the solutions penetrating in- to the hydraulic mortar react in the same manner as the magnesium sulphate of the sea water does, which is the cause of compounds that considerably increase in Volume upon crystallization and thereby destroy the co- hesion of the mortar by formation of cracks or by complete disintegration. If, for instance, only 5 per cent of the aluminum acid contained in a Portland ce- ment is transformed into the double-salt consisting of 1 molecule tricalcium aluminate, 3 molecules calcium sui- phate and 30 molecules of water, an increase of volume of 15 per cent occurs, which makes it obvious that the mortar is transformed into a mass without any cohesion whatever. The formation of colloids in the case of all calcareous hydraulic cements, originating from the reaction of suf- ficiently concentrated lime-water upon the anions silicic acid, aluminum acid and iron acid, explains the other- wise surprising fact, that a mortar which, in the begin- ning, admits water becomes waterproof by the water entering its interior, although this water decomposes the cement and thereby may be expected to make the mortar porous. The percolating water extracts lime from the surface of the cement grains, thereby increases the electro-negative portion of the elements and, upon sufficiently high concentration of the lime-solution, closes the pores through formation of colloids. In my paper on the “Constitution of Hydraulic Ce- ments,” read in 1906 before the Association of German Portland Cement Manufacturers, I still regarded these colloids as true chemical compounds and gave them formulas and named them vicatite, Smeatonite and so forth, although I pointed out at that time that these colloidal compounds do not occur in stoechiometrical proportions as well defined as crystalloids. To-day I am of a different opinion, as a further study of the col- loids has convinced me that they are heterogeneous com- pounds. 28 In the same way as in the dyeing process concentra- tion of the dye-solution and duration of the process are the factors governing the amount of dye taken up by the fabric, or as in the leather tanning the same factors determine how much tannine is taken up by the prepared skin, thus the colloids behave which form out of silicic acid, aluminum acid and iron acid with calcium hydrate, as I explained 16 years ago. Even observation with the naked eye shows that gradually more and more calcium hydrate enters the gelatinous mass or precipi- tates on its surface. The formerly clear hydrogel changes, the cell-walls become less transparent and col- lapse in time. Chemical analysis reveals that the con- tents of lime of the hydrogel increases until finally the Outside of it represents almost pure colloidal calcium hydrate (semipermeability). This experience excludes any further thoughts of for- mation of chemical compounds in well-defined propor- tions. From these explanations the conclusion may be drawn that in hardened hydraulic cements calcium hydrate can occur only as colloid, while it is well-known that, for instance in hardened Portland cement, a consid- erable amount of crystallized calcium hydrate is to be found. This is to be explained by the fact that only a certain amount of calcium hydrate can be precipitated and adsorbed by the silicic acid, aluminum acid and iron acid. The surplus of lime, the element with which Port- land cement is over-loaded, can, in the course of the hardening process, gradually dissolve and crystallize from slightly over-saturated solutions. Thus the large crys- tals of calcium hydrate are formed which can fre- quently be observed in hardened Portland cement. Now let us replace in our illustration the three Port- land cement grains by three grains of glass-like, that is to say quickly cooled, slag which may contain on an average 27% of silicic acid, 15.3% of aluminum acid and 44.8% of calcium oxide, hence 45 molecules of silicic acid to 15 molecules of aluminum acid to 80 molecules of calcium oxide. In a similar slag the anions supersede. It is electro-negative, viewed from the electro-chemical standpoint, briefly it is a puzzuolana. Water is appar- ently without influence upon it, as it represents an al- most insoluble calcium glass, a solid solution of calcium 29 oxide, silicic acid and aluminum acid. Ground very finely and agitated with 1000 times its weight of water, hardly 1 per cent of calcium oxide went into solution. Solutions of sulphate of lime and of sulphate of mag- nesia did not act upon it. However, lime-water of suf- ficient concentration immediately starts formation of hy- drogel. The resultant compounds are colloidal calcium hydro-silicate as the principal substance, furthermore calcium hydro-aluminate and a small amount of calcium Sulpho-aluminate, because only little calcium sulphate is present. The amount of calcium hydrate adsorbed by the glassy slag out of the lime-water is rather small, in One instance it amounted to 6 per cent. The surface of the grains of the powdered slag, therefore must not be imagined to be composed of the former compounds of calcium oxide and silica, but silica hydrate or calcium hydro-silicate very low in lime, which is able to form colloidal calcium hydro-silicate with sufficiently con- centrated lime-water. In the case of slag dust, spontaneously disintegrated Slag, on the other hand, from 4 to 5 per cent of cal- cium hydrate go into solution in 1000 parts of water, for such disintegrated slag contains free calcium oxide besides stable calcium compounds as calcium Ortho- Silicate and calcium meta-silicate, akermanite and anor- thite. Agitated with concentrated lime-water, it adsorbs only a very small amount of calcium hydrate without formation of hydrogel. It does not yield a hydraulic mortar. Finally we have to consider foamy slag, obtained by incomplete granulation or by a specially directed pro- cess of granulation, whereby, in addition to glassy slag, also decomposed slag is formed. We can best form an idea of this slag by imagining one of the three cement grains of our illustration to be composed of the glassy modification and the two others of the decomposed kind which contains free lime. Distilled water extracts from 5 to 9 per cent of lime from this kind of slag. The decomposed residue adsorbs from 1 to 2 per cent of calcium hydrate from concentrated lime-water. This adsorption is accompanied by a slight Swelling and formation of hydrogel. - $0 If mixed with lime-paste in the manner customary for the manufacture of puzzuolana mortars, the glassy slag hardens very well, even if stored for many years, the slag dust yields next to no strength, and the foamy slag hardens very slowly and develops but little strength. The “Hercynia” cement, which is granulated under a Special process and is claimed to consist solely of slag, was agitated with 1000 parts of distilled water. The amount of calcium oxide extracted by the water was 5.5 per cent. The residue placed in concentrated lime- water adsorbed 7.25 per cent of calcium oxide accom- panied by formation of hydrogel. This showed that this cement was able to give off to the water less lime than the glassy slag in it was able to adsorb. . It has been tried to explain the hardening of the slags, especially that of the puzzuolanas, by the contents of alkalies dissolved in the water used for gauging. This explanation, however, can not be accepted for every kind of puzzuolana hardening. The only decisive factor is the concentration of the lime-solution formed in the course of the hardening process. The alkalies rather counteract the hydraulic hardening because they form “soles” with silicic acid and aluminum acid, that is to say they dissolve these colloids. For the hydraulic har- dening, however, only the formation of colloids is of importance, and this can only be brought on by calcium hydrate. The alkalies can contribute to the hardening only indirectly, as I explained in my paper read in 1907, namely by decomposing calcium compounds and by pre- cipitating from them calcium hydrate, as for instance by the decomposition of the calcium sulphide contained in slags, or by transforming silica into the soluble state. The hydraulic hardening, in my opinion, is a com- bined process of chemical energy, electric and surface energy, to which the former, as I believe to have de- monstrated to every one's satisfaction, contributes the greater share. Chemical affinity between elements of opposite electrolytic character undoubtedly enters into play. Those, therefore, who conceived the hydraulic hardening process as a puzzuolana hardening, namely Chevreul and later Fremy, were right; for the formation 31 of colloids is invariably the consequence of a reaction between a lime solution and silicic acid, aluminum acid and iron acid or with compounds from which these anions are set free during the hardening process. To the theory of these scientists has now been added the carplanation for the gradual progress of the hardening by water-absorp- tion, by the extraction of the water of he hydrogel from within, which transforms the gelatinous plastic mass into a solid body, the hardened colloid, and thereby makes a water-mortar out of the cement. No matter what future investigations may prove with regard to the various physical states of a body, whether they consider colloids not fundamentally different from crystals, but merely the root of crystal formation, a stage in which the directing forces are not sufficiently pronounced or, on account of the rapidity of formation, have no time to develop, the colloidal state by all means represents a condition in which motion within the liquid is made very difficult, namely in the case of hydrogel, or in which motion ceases completely, in the case of the gel or hardened colloid. These two stages, which are the origin for semipermeabiltv and impermeability, arc the cause of the stability of the hydraulic cements un- der zorater. A hardened hydraulic cement, however, does not rep- resent the last stage of transformation. It is merely a temporary product created by the action of the water upon the calcined cement. The final state is reached only after all lime has been transformed into carbonate of lime and all anions into hydrates. Until this has been accomplished the chemical changes within a cement do not cease; and after this object is attained, the hydraulic hardening process can commence over again, as the final products of this transformation, silica hy. drate, alumina hydrate and iron hydroxide, upon being ground finely, yield again a hydraulic mortar, if mixed with calcium hydrate. This, however, would be without practical value, because the hydrates are mixed with so much carbonate of lime, formed during the hardening. and with large amounts of sand, added in the making of the mortar, that the strength obtained by a second hardening would be very small. 32 It is a well-known fact, that in the course of the reaction of the water upon hydraulic cements heat is generated, by means of which the progress of the hardening is determined. By the absorption of the water of the hydrogel on the part of the cement grain, which is used for the hydration of the inside of the grain, a further amount of heat is generated, that is to say a long time after setting has taken place. This shows that temperature measurements during the setting are not a reliable method of determining the end of this first part of the hardening process. In many instances I observed, as late at 24 hours after gauging of the cement, a renewed strong increase of the temperature. However, the maximum of the temperature curve may with sufficient accurateness be regarded as the time when the cement set. Both, setting and hardening, are caused by water going into chemical combination, the latter, therefore, is merely the continuation of the former and in fact both can not very well be regarded as separated processes. They are two terms in common use indicating nothing but the beginning and the final stages of the same process. With regard to the accelerating or retarding influ- ence of certain chemicals, of which small amounts are added to the commercial cement or dissolved in water with which the cement is gauged, my opinion is that these substances have been wrongly considered to act as catalyzers or catalysators for lack of a sufficient explanation of the manner in which they act. It was demonstrated in the beginning of this paper that the amorphous colloidal state is obtained by very rapid formation of almost insoluble substances, whereas crystals are possible only in the case of slow formation. These rules make it very apparent that all admixtures which bring on formation of crystals must necessarily retard the hardening process, and that, on the other hand, chemicals contributing to the formation of col- loids must accelerate the setting. If silica in solution, for instance waterglass, is added to the water with which the cement is gauged, the formation of colloidal calcium hydro-silicate takes place very rapidly; the water is used for the formation of the hydrogel and, 33 therefore, disappears at once. However, upon addition of gypsum or calcium chloride, crystal formation is the result. The water, utilized in this case as water of Crystallisation, is adsorbed slowly. If alkali-carbonates are used as admixtures, the concentration of the lime- water necessary for the formation of hydrogel is reached more slowly, because the amount of lime going into Solution is transformed into carbonate of lime. Furthermore, alkali hydrate is formed as a consequence of this reaction, which likewise retards the process by dissolving silica and alumina instead of precipitating them. - - Finally a few words may be said about blowing cement. It suffices to sift a sample of the cement raw mixture, whether obtained by the dry or wet process, through the 200 mesh sieve, in order to be convinced that a perfectly homogeneous mixture of the raw ma- terials can not be obtained in practical operation. The process of vitrification, of course, contributes to secur- ing a more perfect mixture. But even by the clinkering process the inside of the coarser grains of calcined lime can never be reached by the fusing silicates sur- rounding them. The powdered clinker, therefore, must contain more or less uncombined lime, according to the proportions of the raw materials used. This free lime is bound to hydrate in the course of time. Upon enter- ing of the water, extracted from the hydrogel, into the interior of the cement grain and subsequent hydration of the calcium oxide, an increase of volume, that is to say expansion takes place. If the cement grains are very small their surface consequently proportionally very large, the elasticity of the hardened crust around them will be able to overcome the pressure from within. In this case the expansion inside of the grain contributes even to the solidification of the mass and to an increase of strength. In the case of larger grains, however, the internal tension becomes so great that the coating sur- rounding the grains bursts. This is the cause of the blowing of a cement which can be observed only after the cement has set, for the reason that it can take place only after a certain cohesion and rigidity of the mortar has been obtained which is destroyed by expan- sjon from within after plasticity and elasticity of the 34 mortar mass have ceased to exist on account of water- absorption. The grains of a powdered cement result- ing from an imperfect mixture can, therefore, be com- pared to brick with lumps of lime in them... For 150 years chemists knew that the silica of many minerals, upon decomposition of the latter with acids, is obtained as gelatinous substance; but only 30 years ago* I recognized with certainty that, when silica combines with lime in contact with water at low temperatures, in- variably a gelatinous compound is formed, namely col- loidal calcium hydro-silicate. Since this was demon- strated another 30 years of experimental work were required in order to prove beyond doubt my theory published in 1907 in a paper read before the German Cement Manufacturers’ Association which pronounced the formation of colloids in the course of the hydraulic hardening process to be the only essential and character- istic feature. Through my paper of 1907 and through the present publication the hardening process has, in my opinion, been definitely explained. After a study of 45 years I feel confident to have succeeded in unveiling this mystery. *Journal Du Ceramiste & Chaufournier, Paris, March. 22, 1880, Deutsche Toepfer & Zeitung, Berlin, 1880, p. 193. 35 - * sº in a ſt's *.*, * *3, ". . . . . * * * * * * * 3 *s, * : * * * * * * * º º § { g Jº a " ; , , . .”, " . . . . . . . " , ; *...* *...*.*.*.*.*.* * ...","...º. , , , . . . . . . .'' º: , 'sº ºf: a & º, & 4. *, *. tº ºt. º' tº tº "lºº" i", "ºh, º, . ; * º, d # it." ","...', º: o * : ; ; ; tº , ; ; ; ; ſ sº.º.º.º.º.º. “. . . . 'I', ','!'; trº tºº, § *" - : tº . . . . . . . 1,'º','!'." º ; : º lºs º º i 3. . . . tº: s & i Fi ºf • * * * * * r , ; ||||||||||||||| 3 9015 00206 2175 | : . . . . . " ; , , , , , ºr , , ; ; ; ; ; ; (; . . . . . . . . $! : " : - b :º - # * - * ., $:::: . . .'; t < * , , {{< * , ; , #3 & , , § ºf tº * * * : * : * * * # ; : * * * * * * * * • #2 * : * * * * y # , , , ; ; ; ; ; ; ; ; §:::::::::: is . . . . . . st w; sº ſº; s ; } it: , , ; ; , ; ; ; § {:};}} ; p. 1 & i & 3" tº:::: - * t? : * 3, § $r? § 3 & ‘. . . . ; ; ; ; , ; ; ; ; ; a t < * * * * ; ::::: #!!}} f; ; ; ; #1 ::::: # $ 3 * 3 * * * * º!: g > * * * ; : * 4 & {}}.} ::...º. Wiedt « t jSTs.