The Setting of Portland Cement Chemical Reactions of Seasoning

The Setting of Portland Cement - Chemical Reactions and the Role of Calcium Sulfate. Industrial & Engineering Chemistry. Roller. 1934 26 (6), pp 669â€...
0 downloads 0 Views 1MB Size
INDUSTRIAL AND ENGINEERING CHEMISTRY

1076

RATEOF ADDITIONOF PERSULFATE. On the theory that too rapid addition of the persulfate to the soap might result in a loss of some of the available oxidizing strength, runs were made in which the rate of addition of the persulfate was varied. In one run all the persulfate was added at once, in another it was added gradually over a period of 15 minutes, and in another it was added over a period of 30 minutes. The soap was bleached to the same color in all three cases. This result indicates that the decolorization reaction utilizes all the available oxygen of the persulfate as fast as it is liberated. Since an addition period of 30 minutes had been used in prior experiments, it was continued for the sake of unif ormity. ADDITIONOF STABILIZERSAND ACCELERATORS. Compounds such as thymol, hydroquinone, and stannic phosphates are known to be stabilizers for hydrogen peroxide. Titanium and iron salts are known to catalyze its decomposition. There was a possibility that these substances might have a similar effect with potassium persulfate. Runs were therefore made with the addition of these materials in concentrations of 0.1 and 0.01 per cent by weight of the soap bleached on a dry basis. Positive results were obtained only with hydroquinone, which decreased the bleaching efficiency. This apparent effect may have been due to the formation of colored compounds of hydroquinone. CONCENTRATION OF SOAP. To handle more concentrated soaps, they were heated in a glycerol bath and stirred by hand, to avoid foaming. As contrasted with over 90 per cent color removal a t 20 per cent soap concentration, 0.5 per cent of persulfate removes only 85 per cent of the color a t 50 per cent soap concentration. This degree of color removal leaves a distinguishable yellow tinge in the bleached soap. Variations in the method and time of adding persulfate at this higher soap concentration were without effect. The optimum effect a t a lower concentration is therefore indicated. AFTERDARKENING

A limited number of observations were made on this property. As typical ones, soaps originally showing a color of 92.5 stood in the dark for 2 months and again read 92.5. The same soaps in the light bleached to 95 in that time. After 5 months the results were the same. A soap having a value of 82‘/2 bleached in 33 days in the light to 90. One sample having ten times the amount of alkali to be equivalent to the persulfate was the only case of afterdarkening noted. In that case the effect was ascribed to excess of alkali rather than to persulfate which, by test, was absent a t the conclusion of all bleaching experiments. COMPARISON WITH OTHERBLEACHES Palm oil is commercially bleached by agitation with chromic and sulfuric acids or by exposure to sunlight. The soaps from palm oil can be bleached with hydrogen peroxide or with sodium perborate. The latter two would be directly comparable in technic to the use of persulfate.

Vol. 26, No. 10

Using the determined optimum conditions and following the standard procedure outlined, a series of runs was made using hydrogen peroxide or sodium perborate in place of the potassium persulfate. In some runs the amount of peroxide or perborate was the same as the optimum amount found for the persulfatethat is, 0.5 per cent by weight of the anhydrous soap. In other runs higher concentrations of peroxide were used. The results obtained are shown in Table I. These results shorn positively that persulfate is superior to either peroxide or perborate for bleaching palm oil soap under the conditions used.

EFFECT ON OTHERSOAPS The standard procedure arrived a t in experiments with palm oil soap was also applied to a number of other kinds of soap. With a tallow soap made from a low-grade house grease only a part of the color, not mechanically removable in the saponification process, was bleached. The effect was similar to that of peroxide and of perborate in the same concentration on this soap. When applied to a boiled-down cottonseed oil foot soap of high color and rank odor, persulfate bleaching had relatively little effect. This soap appears to contain a brown pigment and a yellow-to-orange pigment. The first can be removed by the process but the second cannot. Persulfate gave a measurable bleaching effect on a soap from good-grade olive oil, which was relatively light in color. Owing to this original light color the effect of the bleaching was scarcely distinguishable. Hydrogen peroxide and perborate did not give measurable bleaching. With an olive-oil foot soap a change of color rather than a bleaching effect resulted. The original soap was dark green, which was lightened to some extent by 0.1 per cent persulfate. Increase of the persulfate to 0.5 per cent gave a greenish yellow color, and further increase produced an orange color. One possible explanation is that a mixture of green and yellow pigments is present in this soap, of which only the green is removed by the bleach. In that case presumably the yellow is carotinoid pigment which is deepened in color by the persulfate. Alternatively the green pigment may be changed by oxidation to yellow or orange. The literature on the subject of oxidation of such pigments provides no satisfactory explanation of this behavior. Peroxide and perborate show only the slight degree of bleaching given by 0.1 per cent of persulfate. No bleaching effect could be obtained in treatment of COCOnut oil soap. Corn oil soap showed only a slight reduction of color with 1 per cent of persulfate. ACKNOWLEDGMENT These results are published by permission of the Buffalo Electrochemical Company, Buffalo, N. Y. Beatrice F. Grey assisted in preparation of the data for publication. LITERATURE CITED

(1) “Ala,” Seifenensieder-Ztg., 49, 538 (1922). (2) Braun, K., Ibid., 51, 8045 (1924). (3) Braun, K., and Nast, H., Ibid., 53, 431-3,450-1 (1926). (4) Braun. K.. and Waber. H.. Ibid., 54, 430-1 (1927). TABLEI. COMPARATIVE BLEACHINQ EFFICIENCY OF PER(5) Buffalo Electrochemical Co., private communication. SULFATE, PEROXIDE, AND PERBORATE (6) Nast, H., Seifensieder-Ztg., 52, 149 (1925). C O N C NO .F SOAPT O DEQREE OF BE BLEACHED ON BLEACHINQ (7) Ibid., 52, 493-4 (1925). DRYBASIS OBTAINED (8) Ibid., 52, 559 (1925). B L E A C H I NAQENT G (9) Neumann, F., Ibid., 52, 43-4 (1925). % % b y ut. 92.5 0.5 (10) Schotte, E., Soap, 1, 19-20 (1926). Potassium persulfate 40.0 Hydrogen peroxjdea 0.5 (11) Vereinigte chemische Werke A.-G., German Patent 200,684 40.0 0.5 Hydrogen peroxidea (1906) ; French Patent 377,900 (1907) ; British Patent 15,900 60.0 1.0 Hydrogen peroxide 2.0 80.0 (1907) ; U. S. Patent 968,438 (1910). Hydrogen peroxide Hydrogen peroxide Sodium perborate

5.0

0.5

No alkali added. b Alkali equivalent t o t h a t in standard procedure added.

92.5

40.0

RECEIVED August 6, 1934. Presented before the Division of Industrial and Engineering Chemistry a t t h e 88th Meeting of the American Chemical Society, Cleveland, Ohio, September 10 t o 15, 1934.

The Setting of Portland Cement Chemical Reactions of Seasoning, Reversion, and Restoration' PAULS. ROLLER,Nonmetallic Minerals Experiment Station, U. S. Bureau of Mines, New Brunswick, N. J.

A

LTHOUGH the effects gypsum on the set given in The changes in setting and in the amount of the atmosphere on Table I1 of the previous paper and composition of the liquid phase of six refers to the addition of one per the setting of Portland different normally gaged clinkers have been cent by weight of 2.5-micron cement are recognized, the full investigated in relation to exposure of the clinkers CaS04.2H20.) significance of these effects apto water vapor and carbon dioxide. The clinkers The o b s e r v a t i o n s made on parently has not been appreclinker A which all tests have ciated. For example, it has perbehaved in the same way qualitatively and in some shown to be a well-burnt normal haps never before been comrespects quantitatively. pletely realized that, as ascerclinker are shown in Table I. Direct evidence exists for activation by abtained in the present work, suitIn column 1 is given the time of sorbed wuter vapor both of tricalcium silicate able exposure (seasoning) of a exposure of 18 kg. of the ground and tricalcium aluminate; deactivation by abclinker always retards its set clinker contained in a closed but even though no gypsum is added. not air-tight container to the air sorbed carbon dioxide is assumed o n reasonable The present study has further of the laboratory a t 20 to 70 per grounds. The present experimental evidence shown that a clinker which has cent relative humidity (average, indicates the existence of such activation and become slow-setting on season45 per cent). During the ninth deactivation of tricalcium silicate and tricalcium ing, may revert to quick-setting, to eleventh month of active testaluminate in Portland cement. Taking the then be restored to slow-setting ing, the room was frequently again, and so on. It has been artificially humidified to about effects into account, the action of absorbed water readily a s c e r t a i n e d that the 60 per cent, and the clinker vapor and of carbon dioxide on the setting of absorption of moisture is restirred up for sampling so that Portland cement is satisfactorily explained. sponsible for the slowing, and conditions were specially favorthe absomtion of carbon diable for absomtion of water oxide for quickening of the set. A similar reciprocal re- vapor by the clinker at this time. The stkady increase in tardation by water vapor and acceleration by carbon dioxide R, or ratio of the loss on heating a t 500' C. to the have frequently been noted for Portland c e m e n t i . e., for difference between the loss a t 1000" and a t 500" C. (colclinker containing added gypsum (compare especially cita- umn 4), shows that moisture was absorbed in excess of tions 5 and 16). It has been observed also (.5) that sealed carbon dioxide. Simultaneously, as the rest of the table storage and dry air free of carbon dioxide are without material shows, a striking retardation of the set, a corresponding effect. In practice the repeated and often much-feared decrease (as determined from the amount of liquid phase) in changes in setting appear sometimes to occur with surprising mixing water retained by the setting clinker, or an indicated decreased hydration (IS), and marked changes in composiabruptness (4A). I n the present work the changes both in setting and in tion of the liquid phase, which will be referred to later, took amount and composition of the liquid phase have been in- place. vestigated in relation to the absorption of moisture and of Seasoning similar to that for clinker A in Table I could be carbon dioxide. Uncertainties caused by the presence of effected in a few days instead of one or two years by more gypsum have been avoided by working with clinkers instead effective exposure a t a higher humidity. Seven kilograms of of cement. The experimental method has been previously f-clinkers B and C were placed in a shallow pan in a moist described (IS), the liquid phase being extracted, as before, cabinet a t room temperature and a t a relative humidity of 75 15 minutes after mixing to or close to normal consistency. per cent. The clinkers were thoroughly stirred up once a The reaction mechanism of set previously deduced ( I S ) , day. This treatment will be referred to as moist seasoning. with one added condition pointed to by the present experiThe results on moist seasoning for 7 days are shown in mental results-namely, that absorbed water vapor and car- Table 11. The change from quick to slow set and the correbon dioxide change the chemical activity of tricalcium sili- sponding change in water retained and in composition of liquid cate and of tricalcium aluminate-explains satisfactorily the phase are seen to be similar to those for clinker A after 21 effects of absorption on setting. Direct evidence exists, as months of ordinary seasoning. However, the absorption will be seen, for the action of absorbed water vapor on tri- ratio R, or moisture absorption in excess of carbon dioxide, calcium silicate ((2,s)and tricalcium aluminate (C,il), while is higher than for t-clinker A, and other differences exist which the reciprocal action of absorbed carbon dioxide has been will be referred to later; this makes it desirable to designate postulated on reasonable grounds and is indicated by the the moist-seasoned clinkers as overseasoned, or o-clinkers. present results. Clinkers D, E, and F were stored uninterruptedly in the laboratory in a manner similar to that of clinker A. 'At the SEASONING OF A CLINKER end of 21 months the originally quick-setting clinkers showed A description of the state of seasoning of a powdered clinker a slow set and other properties similar to t-clinker A, leading due to its absorption of water vapor in excess of carbon di- to their designation also as t-clinkers. Data on the setting, oxide has previously been given (13). (The effect of added water retained, and liquid-phase composition of t-clinkers 1 Thin paper is the second in a series from the Nonmetallic MineralsExperiD, E, and F are given in Table 111, together nTith the values ment Station connected with the problem of the utilization of natural of R for the same clinkers in the f-state. For all the t-clinkanhydrite a6 a retarder (of the set of Portland cement. For the first paper ers, R averages 0.9 * 0.1. in this series eee literature reference IS. 1077

INDUSTRIAL AND ENGINEERING CHEMISTRY

1078

Vol. 26, No. 10

CONCENTRATION OF CALCIUM SELFATEIN CLINKER LIQUID calculated from its solubility a t room temperature, which is PHASE To determine accurately the concentration of calcium sulfate in the presence of other ions, the same procedure may be followed as was used previously for calcium hydroxide. In other words, a t a given ionic strength of the solution and for I

I

I

heated

.os .Ol .I .z .3 .5 .7 LO L5 IONIC STRENGTH, moles/liter, log scale FIGURE1. CONCENTRATION OF CALCIUM SULFATE IN THE LIQUIDPHASE

1.5 times that of gypsum (14) and from the fact that the curve for the natural anhydrite must lie nearly parallel to the curve for gypsum. (The curves for gypsum and natural anhydrite converge slightly because of the decrease in aqueous tension with increase in concentration.) A small portion of the curve for plaster is also shown. Points for the calcium sulfate concentration in the liquid phase of t-clinkers A, D, E, and F have been plotted in Figure 1 from the data of Tables I and 111. The points fall on or near the gypsum (or syngenite) curve. For t-clinkers, therefore, the liquid phase is saturated with respect to gypsum. Since analysis was made 15 minutes after mixing, and in view of the rapidity of the solution reactions (IS),it may be concluded that gypsum saturation of the liquid phase of t-clinkers is attained practically with the onset of setting. Obviously the origin of the dissolved calcium sulfate is sulfur trioxide in the clinker itself. The sulfur trioxide content usually is between 0.2 and 0.5 per cent. The concentration of calcium sulfate in the liquid phase varies with seasoning of a clinker. For quick-setting f- or s-clinkers it is very low. Points plotted in Figure 1 from t h e data of Tables I and I1 for f-clinkers B and C and for sclinker A show that for these clinkers the concentration averages about one-tenth gypsum saturation. For slowsetting m-clinker A the concentration has increased to about one-third saturation. For thoroughly seasoned, slow-setting t-clinkers, gypsum saturation of the clinker liquid phase has been attained.

CHEMICAL REACTIONS OF SEASONING

Mean ion product os. ionic strength for potassium sulfate solutions saturated with gypsum or syngenite and for liquid phase.

similar ion types the mean ion product (Ca++ X S04--)1/2 is taken as a measure of the effective concentration of calcium sulfate in solution. In Figure 1 is plotted from available data (16) the mean ion product (Ca++ X Sob--) 1 / 2 , measuring the concentration of calcium sulfate, against the ionic strength of solutions of potassium sulfate saturated with gypsum. The curve is a straight line until shortly before the invariant point a t which gypsum and syngenite (CaSO4.K2SO,.HZO) are solid phases. Beyond the invariant point the curve is extrapolated as a dotted line for gypsum as metastable solid phase and is drawn as a solid line (7) for syngenite as stable phase. The curve for natural anhydrite as a metastable phase is

The retarding action of absorbed water vapor is to be understood from its effect on the chemical reactions of setting. The effect may be inferred from the change in liquid-phase composition as seasoning of a clinker proceeds. The following reactions have been deduced as occurring during setting (IS): 3CaO.Al203

+ 3CaS04 e 3Ca0.Al2O3.3Ca-(2)

(s)

SO4.XH20 (s) { 3Ca0~A1208~XH~0 (s) ) + Ca(OH)2'E 4CaO.AI23CaO.Al2O3(s) { 3CaO~.41203~XH~0 (s) ) 03.XHzO ~ C ~ O . A ~ Z O ~ (s) .XH + ~3CaS04 O e 3Ca0.A1208.3CaS04.XH20(s) + Ca(0H); (9)

(4)

(7)

Equations 2 and 4 differ slightly from those previously given in that hydrated C3A, as well as anhydrous crystalline

TABLEI. CHANGES IN ABSORPTION RATIO,SETTING BEHAVIOR, AND LIQCID-PHASE COMPOSITION DURIKG SEASONING OF GROUND CLINHER A ABSORPTION WATER RESEASONINQ Loas ON RATIO, MIXINQ T I M E OF S E T Time StateD IQNITION R WATBR Initial Final TAINED Hr. Min. H r . Min. Cc./lOO 8 . Months % % Quick setting 0 f 0.70 0.37 18:3 Quick setting 3010 S 6 17.3 Quick setting 28.0 0154 0:97 S 9 13.7 3 30 1 03 20.9 m 10 13.9 5 00 11 m 1:05 0:67 20.9 2 26 13.5 6 30 21 t 1.28 0.94 19.8 2 52 a f , freshly seasoned; s, slightly seasoned; m, moderately seasoned; t, thoroughly aeasoned.

OH-

0 : 232 0.286 0.195 0.176 0.107

LIQUID-PHASE COMPOSITION-A1201 l/iSOd-. '/&a++ MQ./CC .... .... 0 006 0.0037 0.0060 0.0040 0.006 0.0047 0.0020 0.065 .... 0.0015 0.075 0.0076 0.0015 0.172 0.0329

M . ea. p e r cc.

RATIO,SETTING BEHAVIOR, AND LIQUID-PHASE COMPOSITION TABLE11. CHANQESIN ABSORPTION B AND c STATE'

OF

LOSS ON

S ~ A S O N I N Q IQNITION

ABSORPTION RATIO,

R

%

(Exposed 7 day8 in a moist cabinet with stirring once a der) WATER RETIMEOF SET MIXINQ OHTAINED Final WATER Initial % H r . Min. Hr. Min. Cc./lOO 8 .

--

CLINKER B

f 0

0.81 1.45

0.47 1.13

22.0 20.6

3

0.21 1.21

0.31 3.16

34.0 20.6

0 35

0

10 25

50 00

1 6

ON

EXPOSURE O F f-CLINKERS

LIQTXD-PEASE CWPOSITION 1/&a + + '/is04 - -

M . eq. p e r cc.

Ah03 Mg./cc.

17.4 14.2

0.2638 0.1155

0.0162 0.1112

0.0063 0.0182

0.0058 0.0010

23.8 14.2

0.3860 0.1790

0.0150 0.2750

0,0019 0.0116

0,0250 0.001c-

CLINKER C

f 0

a

2 40

No final set due to large amount of water for normal consistency.

(I

4 30

October, 1931

INDUSTRIAL AND ENGINEERING CHEMISTRY

CIA, is indicated as a component of reaction. The hydrated C3A is doubtless the more significant as regards the reverse reaction of Equations 2 and 4. The concentration of calcium hydroxide in the above equations has been written Ca(OH)’2 instead of Ca(OH)2to show that ordinarily only calcium hydroxide a t a high concentration above a threshold value near lime saturation can react to form tetracalcium aluminate. Ca(0H)’z refers to the fraction of calcium hydroxide above the threshold concentration for reaction, and is designated reactive calcium hydroxide. [According to Assarsson (3) and Wells ( I 7 ) , tetracalcium aluminate (CJ) at an extremely low concentration of alumina as in the cement or clinker liquid phase, is stable

signaled by an increase in concentration of calcium sulfate in the liquid phase. For t-clinkers, the small amount of sulfur trioxide in the clinker rapidly saturates the liquid phase with respect to gypsum, and gypsum may crystallize out. Sulfoaluminate, it may be concluded, is not formed during setting of thoroughly seasoned slow-setting clinkers. As judged by the steady increase in calcium sulfate concentration with seasoning, sulfoaluminate formation, while existent for quick-setting clinkers, decreases as seasoning proceeds until it is nil for thoroughly seasoned clinkers. Reaction 7 indicates that the underlying cause of this transition is an increased formation, as seasoning proceeds, of reactive calcium hydroxide in solution.

AND LIQUID-PHASE COMPOSITION O F t-CLINKERS D, E, T ~ B L11 E1. akBSORPTION RATIO,SETTIKG BEHAVIOR, (Seasoned 21 months as for clinker A , Table I) LO8S ON

IC-

CLINKER

NITION

I~BSORPTION

RATIO, R

70 t-D t-E t-F

2 30 1 89 0.85

0.78 0.75 1.02

f-CLINKER Absorption MIXING ratio, WATER TIMEOF SET ignition R Initial Final % % H r . Min. H r . Min. 2 41 8 00 0.78? 20.8 1.66 2 08 7 00 0.39 23.5 1.40 1 50 4 30 0.50 19.9 0.37 Lose on

1079

WATER RETAINED

Cc./lOOg. 14.2 15.7 13.1

AND

??

~LIQt?ID-PHASE COMPOSITION OH1/lSOd-l/rCa++ M. eq. per cc. 0.1040 0.1175 0.0327 0.1885 0.3245 0.0182 0.2066 0.6340 0.0200

--

AlzOa

MO./CC.

0:OOiO 0.0015

OF t-CLINKERS TABLE Iv. ABSORPTIONR.ITIO,SETTING BEHAVIOR.AND LIQUID-PHSSECOMPOSITION ON EXPOSURE

(Exposed 7 days in a moist cabinet with etirring once a day) WATER Loas ON RATIO; MIXINQ TIMEO F SET RE--LIQUID-PHASE COMPOSITION CLINKER IGNITION R WATER Initial Final TAINED OH 1/2so~-l/rCa++ 7” 9,% H r . Min. Hr. Min. Cc./lOOa. M. ea. .- -v e r cc. 0.1045 0.1504 6 00 14.5 0.0134 19.75 2 40 1.33 t-A 2.31 0.1040 0.0917 7 30 15.1 0.0230 0.83 21.5O 2 12 t-D 3.41 0.1500 0.3400 25.0 2 21 8 30 0.0237 16.6 1.38 t-E 2 83 0.1515 0.6280 14.0 0.0178 2 06 5 30 2.42 20.5b t-F 1.88 a About 2% below normal consistency. b About 1% below normal consistency.

.

.~BBORPTION

+onlya t high concentrations of calcium hydroxide approaching saturation. To cause the formation of CIA, a certain high concentration of calcium hydroxide in excess probably .of the equilibrium value, would obviously be necessary. The threshold concentration of calcium hydroxide necessary for reaction, and the equilibrium value, are from this point of view related.] In view of the vanishingly small concentration of alumina in the liquid phase, formation of tetracalcium aluminate (CIA) and sulfoaluminate according to Equations 2, 4,and 7 will be restricted largely to the surface, or near to the surface, .of C3A in cement. Liquid-phase composition shows tetracalcium aluminate to be a stable phase in setting clinker or .cement; the tetracalcium compound has recently also been detected in setting cement. [Crystalline tetracalcium aluminate has been reported in set Portland cement by Assarsson (2) who has thoroughly investigated (3) the system calcium .oxide-alumina-water. The tetracalcium aluminate, though the stable phase, is reported as being present only sparingly. ‘This is in line with the conclusion of this study that the C& is formed only in small catalytic amounts, the CIA in cement hydrating mainly to a metastable tricalcium aluminate hydrate.] Formation of stable C4A a t the surface of CIA in cement, it has been postulated ( I S ) , hinders the rapid metastable hydration of residual CIA to hydrated C3A and thus slows the setting. The direction and extent of reactions 2, 4, and 7 will depend on the closeness of the composition of the liquid phase to that required for stability of the solid phases. In other words, solid phases will form or will be decomposed depending on the concentrations in the liquid phase. In the presence of a sustained high concentration of calcium hydroxide, sulfoaluminate, if formed by reaction 4,will in turn be decomposed in accordance with reaction 7 . Such decomposition (in effect, decreased formation of sulfoaluminate) according to Equation 7 and as would be expected is

All01

M o .,/ c- -e-. s

0.0010 0.0010 0.0010

0.OOOb

A direct indication of an increase in reactive calcium hydroxide with seasoning is seen in the steady decrease in the amount of alumina dissolved in the clinker liquid phase (Tables I and 11). With decreased sulfoaluminate formation, alumina in solution can have diminished only by increased reaction with calcium hydroxide. The increased reaction, in turn, can have come about only through increased formation of calcium hydroxide while a t a concentration in the neighborhood of lime saturation and high enough to react with the minute amount of alumina in solution. To account for the apparent increase in reactive calcium hydroxide on seasoning, attention must be directed to tricalcium silicate (CIS), which is the source of reactive calcium hydroxide in the cement liquid phase. C3S is acted upon by water vapor. This is shown physically in an etching of the surface of the pure crystals ( I ) . In studying the action of liquid water on C3S and CZS, Beckmann (4) has stressed the observation that absorbed water vapor causes the C3S (and also cy-C2S) to react more readily with the liquid water. The activation of C3Sdue to absorption of water vapor was shown in several different ways, among which was a speeding up of the hydrolysis of this compound with consequent increased rate of liberation of calcium hydroxide in solution. It may be concluded that water vapor absorbed on seasoning activates CIS in Portland cement with respect to its hydrolysis and results thus in increased formation of reactive calcium hydroxide. The amount of stable CIA that is formed will therefore increase (Equation 2), and a progressive slowing of the set should be effected with increase in seasoning as is actually observed. Seasoning of the clinkers has resulted in changes in the setting, in water retained, and in composition of the liquid phase similar to those previously observed ( I S ) on adding gypsum to s-clinker A. This agreement is to be anticipated from the standpoint that in both cases the cause is the samenamely, increased formation of Ca(OH)’2.

I N D U S T R I A L A N D E N G I N E E R I S G C H E 1L1 I S T R Y

1080

OVERSE ASOMNQ To see if the retarding effect of absorbed water vapor could be increased beyond the maximum effect observed on ordinary seasoning, the slow-setting t-clinkers A, D, E, and F were exposed in a moist cabinet as described above for f-clinkers B and C. The relative humidity was now 75 per cent, whereas the clinkers originally had been seasoned to the tstate a t a relative humidity averaging 45 per cent. Table IV shows the results, which should be compared with those of Tables I and 111. An absorption both of moisture and carbon dioxide has occurred, but the absorption of moisture was greater, as seen by the increase in R from an average of 0.9 for the t-clinkers to an average of about 1.5 for the moist-seasoned clinkers. Although after the moist exposure the clinkers remained as slow setting as originally when in the t-state, nevertheles other effects supervened which led to the belief that the clinkers had been undesirably overseasoned. The clinkers therefore will be referred to as o-clinkers. The latter required about 1.5 per cent more water in the percentage for normal consistency than the original t-clinkers; also the pats were very voluminous, suggesting low strength on hardening. The objectionable effect of the increased water vapor absorption may be explained by activation of tricalcium aluminate by absorbed moisture. Such an activation has been observed directly by Pulfrich and Linck ( I I ) , who write that “a long storage of the melt (of CsA) leads to devitrification, to decomposition, and to an increase in the hydration power.” The percentage hydration of the o-clinkers, as measured by the water retained ( l a ) ,is one per cent greater than that of the original t-clinkers. The apparent increased hydration of the o-clinkers would correspond to the presence of an activated CSA. .lo

.a5

o

.I

A, D, E, and F. Activation of the C3h in o-clinkers due to excess absorbed water vapor is thus indicated. Similarly, corresponding to a greater tendency to sulfoaluminate (Equation 4) as well as C4A formation by an active C3A, the concentration of calcium sulfate for the oclinkers falls below the gypsum-saturation curve (Figure 1) and, except for clinker E, below the values for the corresponding t-clinkers A, D, and F. An interesting result, which may be traced to an activation of C3A by absorbed water vapor, is found in the following experiment on comparative retardation with plaster and with gypsum. Table V shows that, for m-clinker A, plaster is more effective than gypsum, but is less effective for o-clinkers B and C. The explanation is based on the capacity of sulfoaluminate formation to cause quick set (1.9). Added plaster, giving a higher calcium sulfate concentration than gypsum, is, relative to the gypsum, less effective as a retarder with o-clinkers B and C than with m-clinker A because of the greater activity of the CIA in the overseasoned clinkers. TABLEV. COMPARISON OF TIMEOF SETOF m- AND 0-CLINKERS WITH ONE PER CENTEACHOF GYPSUM AND PLASTER TIMEOF SET GYPSUM

CLINKERMIXING WATER Initial Final Hr. Min. Hr. Min. 20.9 2 10 m-A 5 30 3 15 7 00 o-B 20.6 3 00 6 30 0-c 20.6

PLASTER

Initial Final Hr. Min. Hr. Min. 3 40 7 00 2 25 6 00 2 10 6 00

The action of absorbed water vapor, which results in seasoning of a clinker, may be summarized as follows: Both Cas and G A are activated but show a tendency to produce opposite effects. Apparently activation of C3S predominates, so a striking retardation of the set with a great decrease in the mixing-water requirement is obtained. If the absorption leads to a significant activation of CaA, due, for example, to excess&e exposure a t too high ‘ humidity, a deleterious overseasoning takes place. At the humidifications of the present experiments the deleterious effect consisted in an increase in water for normal consistency, in an increase in indicated hydration, in a voluminous paste (suggesting low strength) but not, probably in part because of increased activation of CSS, in quickened setting.

t-dnkv

.Ol

Vol. 26, No. 10

.L

.3

5 .

.7

1.0

1.5

ABSORPTION OF CARBONDIOXIDE Carbon dioxide, it is reasonable to assume, may on its FIGURE2. THE absorption unite with potential reactive calcium hydroxide contained in the original C&3XHzO complex high in Mean ion product os. ionic strength in relation to line of saturation with calcium hydroxide. lime. In this way, absorbed carbon dioxide could nullify the effect of absorbed water vapor in accelerating the h i Changes in the activity of CIA may be correlated with drolytic splitting off of calcium hydroxide from CIS. The rechanges in concentration of calcium hydroxide and calcium sultant loss in reactive calcium hydroxide will account for the sulfate in the liquid phase. The changes in concentration action, commonly observed for Portland cement and obdue to C3A should, on the basis of the reactions described, be served in this work on clinkers, of absorbed carbon dioxide in opposite in direction to those due to CIS. Since they would accelerating the set. In the present experiments carbon dioxide, humidified b y affect both CIS and CIA, slight differences in setting temperature, in the plasticity of the cement paste, etc., which accom- bubbling through water or aqueous solutions, was passed pany changes in the activity of the compounds, are assumed over a few hundred grams of ground clinker in a rotating to have a negligible effect on the concentrations in the liquid Mason jar. The time of contact, unless otherwise specified, phase in comparison to the effect due to the changes in ac- was one hour. The clinker was tested shortly after exposure. The action of carbon dioxide in relation to the state of tivity themselves. The C3A in an active state should be able to react (Equa- seasoning is seen on comparing the results for clinker A in tion 2) with calcium hydroxide a t a lower threshold concen- the m- and in the t-state (Table VI). As judged by ratio R tration in solution than C3A in a normal or less active state. after exposure compared with its value (given in Table I) Despite the probable increased activation of CIS by part of for the original clinker A, the net absorption of carbon dioxthe excess moisture absorbed during the moist seasoning of ide, both by m- and t-clinker A, has been relatively small. the t-clinkers, Figure 2 shows that the effective concentration Nevertheless, for m-clinker A, the indicated percentage hyof calcium hydroxide for the o-clinkers falls below the lime- dration of the setting clinker has increased by 1.3 per cent saturation line and below the values for the original t-clinkers and reversion of the m-clinker to quick set has taken place.

103 s a l e CONCENTRATION OF CALCIUM HYDROXIDE IN LIQUIDPHASE IONIC STRENGTH, moles/

liter,

October, 1934 TABLE VI. STATE

INDUSTRIAL

.iBSORPTIOU

R LTIO,

RELATIVE Loss

OF HCMIDITY O N ~ E A s O N I N G O F GAS IGNITION

T"

vn

100 100

1.09

,I

m t

t

65

TABLE VII.

,I

1.54 1.37

SETTIKG

4ND ENGINEERING

BEHAVIOR, AND

ABSORPTION RATIO, MIXING MATER R % ,0.63 20.9 1.00 19.8 0.93 19.8

RELATIVE ABSORPTION HUMIDITY Loss ON RATIO, MIXINQ CLINKER O F G A S IGNITION R WATER % % % 0-B 100 2.01 1.01 20.6 0-c 100 1.91 1.45 21.6 t-E 85 3.24 0.56 23.5 t-F 100 1.29 1.05 19.9 t-F 65 1.10 1.04 19.9

~ A I Q C I D - P H A S EC O M P O S I T I O X OF

CARBON DIOXIDE

TIMEOF Initial Hr. Min. 0 22 2 34 1 28

SETTING B E H l V I O R AND LIQUID-PHASE

CHEMISTRT

SET Final Hr. M i n . 3 00 6 30 5 00

WATER RB-

13:s

t- A N D CARBON DIOXIDE

TIMEOF SET Initial Final Hr. Min. Hr. Min. 2 00 5 00 0 35 4 00 0 03 0 20 1 40 5 00 1 12 3 30

Simultaneously, the amount of dissolved calcium sulfate has decreased considerably (increased sulfoaluminate formation), and the concentration of alumina has increased. These changes in setting, in water retained, and in composition of the liquid phase are just opposite to those on seasoning. They indicate that on absorption of carbon dioxide the formation of reactive calcium hydroxide by hydrolysis of CIS has decreased, and the hydration of C A correspondingly increased, with resultant quickened setting. The t-clinker A, however, having absorbed more water vapor, was apparently more resistant to the action of carbon dioxide than the same clinker in the m-state of seasoning. The time of set of the t-clinker, even after 2.5 hours of the same exposure as for the m-clinker, was practically unaffected. However, when the relative humidity of the gas was decreased to 65 per cent, a small decrease in time of set and a small corresponding change in water retained and in composition of the liquid phase was manifested after one hour of exposure (Table VI). As shown in Table VII, compared with Tables I1 and 111, both o-B and t-F, just as t-clinker A, are resistant to the action of carbon dioxide, absorbing little carbon dioxide in excess of water vapor and showing little change in setting or in liquid-phase composition. (Qualitative test indicated t-clinker D also to be resistant to carbon dioxide.) With o-clinker C and t-clinker E the result, however, was different. As seen from the prominent decrease in R from 3.16 to 1.45 for 0-C, and from 0.75 to 0.56 for t-E, both of the clinkers have absorbed a considerable quantity of carbon dioxide in excess of water vapor. Simultaneously, a sort of false set has been produced for o-clinker C, and pronounced quick set for t-clinker E. The change to false set of o-clinker C, for which R is still as high as 1.45 aft,er exposure to the humidified carbon dioxide, and in which therefore Cas is probably still moderately active, is accompanied by little change in indicated hydration or in composition of the liquid phase. The change to quick set of t-clinker E is accompanied, however, by changes in liquid phase, which are similar to those for m-clinker A treated with carbon dioxide and opposite to those of seasoning. The indicated percentage hydration of the setting clinker has increased by 3 per cent, the calcium sulfate concentration has decreased from gypsum saturation for the original t-clinker E: to one-sixth that of gypsum saturation (Figure I), and the alumina concentration has increased from 0.0020 to 0.0035 mg. per cc. The difference in reaction of t-clinker E to carbon dioxide from that of the other t-clinkers was in a way presaged by its difference in behavior as a t-clinker. The water for normal consistency, 23.5 per cent, was strikingly higher than that of the other t-clinkers. The t-clinker E also m-as susceptible to heat, reverting to quick set after heating a t 215' C. whereas simultaneously t-clinker F was practically unaffected. The

AND t - C L I U K E R

-LIQUID-PHASE OH 1/2S04-M . ea. ver cc. .. 0.206 0.033

TAINED

Cc./lOO . .a . 15.1

COMPOSITION OF

m-

1081

0:099

0-CLINKERS

WATER RE-

0:i57

B,

COXPOSITION1/2Ca AlrO: Me./ec. 0.009 0.0040 + +

....

0.0235

....

0.0015

c, E, AKD F AFTER EXPOSURE TO

15.2 14.5 18.7

-LIQIJID-PHAS~ OH1/zso4-M . eq. per cc. 0.076 0.117 0.1035 0.241 0.459 0.121

13:7

0.174

TAINED

Cc./lOO 8 .

.i -4FTEH E X P O S U R E TO

....

-

COMPOSITION l/zCa + + 4hOa Mg./cc. 0.0206 0.0010 0.0163 0.0010 0.0023 0.0035

0:638

....

....

0.0143

0.0015

t-clinker E (like in-clinker A) also showed a distinct tendency to reversion in set on storage during the dry winter months, whereas the other t-clinkers were under the same conditions unaffected. [The continued absorption of carbon dioxide from the atmosphere in excess of water vapor in spite of the slight concentration of the former is understandable on the basis of Rodt's observation (1%)that the absorption of water vapor by Portland cement approaches a maximum a t a given relative humidity. Hence a ground clinker or cement which is placed in an atmosphere of decreased humidity may continuously absorb carbon dioxide but no water vapor. The effect of such gradual absorption of carbon dioxide will of course be opposite to that of the slow absorption of water vapor in excess of carbon dioxide.] SUSCEPTIBILITY TO CARBONDIOXIDE I N RELATION TO CLINKERPROPERTIES Since clinkers C and E were already well seasoned when exposed to carbon dioxide, the reversion in set of these clinkers caused by the exposure to carbon dioxide is due evidently to an intrinsic property or properties of the clinkers which is apart from the degree of seasoning as noted for clinker A. In the following discussion an attempt is made to differentiate the susceptible clinkers from the stable ones according to properties. The susceptible clinkers C and E showed a relatively high absorption power. Conditions being the same, the absorption (ignition loss) is appreciably greater for C and E than for the other clinkers both in passing from the f- to the tor o-state, and after exposure to carbon dioxide in the t- or ostate as seen from Tables I, 11, and 111. TABLEVIII. COMPOSITION RATIOSFOR CLINKERS CLINKER D B F

6

E

CaO - FREECaO Si02 AlrOa FezOa 2.09 2.06 2.02 1.96 1.91 1.94

+

+

Si02 -

AlzOa 4.90 4.17 3.18 3.02 2.75 2.91

SiO: FerOa ALOa 2.77 2.46 2.24 2.05 1.92 1.83

+

The susceptible clinkers can further be distinguished from the others by the chemical composition. Table VI11 shows that the magnitude of the ratios (calcium oxide minus free calcium oxide) to (silica plus alumina plus ferric oxide), silica to alumina, and silica to (alumina plus ferric oxide) is least for the susceptible clinkers C and E. A low ratio is considered generally to correspond to a tendency to quick set. The ratios silica to alumina, and silica to (alumina plus ferric oxide) show the greatest difference between C, E, and the other clinkers. The ratio silica to (alumina plus ferric oxide) is of special interest as it alone gives the right order for clinkers C and E ; that is, the ratio is lower for the more susceptible clinker E than for C. The ratio silica to (alumina plus ferric oxide) is only 1.92 for C and 1.83 for E, or below the

INDUSTRIAL AND E N G I N E E R I N G CHEMISTRY

1082

value 2.0 which is sometimes considered (IO) as a lower safe limit. The distinction between the clinkers was not so clear when compared on the basis of calculated compound composition (Table I of the previous paper, IS). Though clinker C appears high in CIA (15 per cent) and low in ratio of Cas to C3A (2.1), the computed composition of clinker E, on the other hand, appears close to normal, with 12 per cent CIA and a ratio of CIS to CIA of 3.5 which actually is greater than for C and for the stable clinkers A and E. A third distinction which exists is with respect to alkali content. That of clinkers C and E, about 1.3 per cent (potassium oxide plus sodium oxide), is appreciably higher than for the other clinkers for which the alkali is less than 1.0 per cent. An apparent exception is the stable clinker F for which the alkali is 1.41 per cent. However, it was ascertained that during the production of this clinker, electrically precipitated dust was returned to the raw mix, a procedure that was not applied in the case of the other clinkers. Alkali in precipitated dust appears to be combined mainly as alkali sulfate (9). The presence in clinker F of much of the alkali as sulfate is indicated by the abnormally high sulfur trioxide content, 1.02 per cent, for this clinker.

In conclusion, clinkers C and E which are susceptible to carbon dioxide were found to be differentiated from the stable clinkers with respect to the following properties: absorption power, chemical composition, alkali content, and color of the powdered clinker. SECONDARY EFFECTO F CARBON DIOXIDE The effects of absorbed carbon dioxide that have been noted may be referred back to the action of this gas on CIS. However, CIA as well as C3S is activated by absorbed moisture, so that a deactivating effect of carbon dioxide on CIA as well as on C3S might be expected. Not much evidence, however, has been obtained which would correspond to such an effect. The lack of evidence of deactivation perhaps accords with the relatively small harmful effect of CIA in an activated state on the setting, at least at the humidifications applied in the present experiments. One result, however, was obtained with o-clinkers B and C, different from that for the t-clinkers, which suggests an effect. of carbon dioxide on C3A. For both t- and o-clinkers treated with humidified carbon dioxide, the concentration of calcium hydroxide in the liquid phase is decreased (Figure 2), a re-

TABLEIX. REVERSION PRODUCED BY HEATINQ tLoss

CLrNKnB

ABSORPTION ON

IGNITION"

RATIO, R

% t-A o-B t-D a Calculated

0.37 0.90 0.40 0.95 0.37 1.77 from loss of moiature.

MIXING WATER

% 21.0 22.0 22.5

TIMEOF SET Initial Final Ht. M i n . H r . Min. 0 08 1 00 0 os 1 00 0 15 4 30

Hence the conclusion that clinkers C and E are higher in alkali than the other clinkers holds if that alkali alone is considered which is combined in a manner other than as sulfate. Alkali sulfate in distinction to other alkali compounds hardly reacts with calcium hydroxide and thus has hardly any effect on the setting time. Though the alkali content of C and E is 1.3 per cent as against 1.0 per cent for clinkers A, B, and D, the concentration of alkali in the liquid phases of C and E is more than twice as great as for the other clinkers. Grimm (6) concluded that the presence of much alkali in cement is the cause of reversion in set on storage. He believed that carbon dioxide decomposed alkali aluminate in the cement to give the much-feared carbonate. Such a postulated chemical decomposition is hardly probable in view of the short space of time between slow set and reversion. Furthermore, the reversions often are repeated. Finally, in view of only traces of alumina in the liquid phase, reaction of alkali aluminate with calcium hydroxide in solution apparently is practically as complete as reaction of alkali carbonate would be, and so the accelerating effect on the set would be expected to be the same. If reversion in set actually is associated with much alkali in cement, the cause of the alkali action is to be sought on the one hand in the unfavorable accelerating effect on the set due to reaction with calcium hydroxide in solution (IS), and on the other hand in a possible promoter action according to which the alkalies might stimulate absorption of carbon dioxide by C3S in cement. As a fourth and final distinction, it should be mentioned that both clinkers C and E when powdered were brown in color whereas the other clinkers were gray. A brown color is variously attributed to pnderburning, underlime, and breakdown of kiln lining. A brown color may also be obtained by excess quenching of the red-hot clinker with water. The brown color often is regarded unfavorably, and that of clinkers C and E is mentioned as a possible symptom of a tendency to reversion in set.

Vol. 26. No. IO

AND

0-CLINKERS A., B. , AND D

WATER RE-

TAINED

Cc./lOO o. 10.4 17.7 17.6

OH -

270"

c.

LIQVID-PHAE~ COMPOSITION '/:804--

M.eq. 0.3100 0.2345 0.2210

AT

'/rCa

++

0.0078 0.0122 0.0062

>

Ala08

Mo./cc.

per cc.

0.0066 0.0065 0.0093

0.00560.0030, O.OO4Q

sult that would correspond merely to an active C3A in t h e presence of a deactivated Cas. For the t-clinkers, the concentration of calcium sulfate similarly was less after treatment with carbon dioxide than before (Figure l), as would also be expected for deactivation of C B . For the o-clinkers, however, the reverse was true. The slight increase in calcium sulfate concentration after carbon dioxide treatment of the o-clinkers, for which the C3A was originally in a more active state than for the t-clinkers, can be explained by assuming that the carbon dioxide has exerted a slight deactivating effect on the originally reactive C3A, as well as on C3S in the o-clinkers. REVERSION IN SET BY ABSTRACTION OF WATERVAPOR In addition to reversion in set due to absorption of carbon dioxide, reversion should be effected also by any possible abstraction of absorbed water vapor from the C3S in cement. An attempt to cause reversion by abstracting absorbed water vapor by means of dry air did not succeed. Heating the clinker to a high enough temperature, however, proved very effective. The results on heating t- and o-clinkers a t 270" C. are shown in Table IX which should be compared with Tables I, 11,and 111. The loss of absorbed water vapor is shown in the decrease in R , and the clinkers are seen to have become quick setting. The liquid phase now is similar to that before seasoning, showing high retention of water by the setting clinker, extremely low calcium sulfate concentration (Figure 1), and relatively high alumina concentration. The concentration of calcium hydroxide in the liquid phase of the heated t- and 0-clinkers is seen from Figure 2 to be greater than that for the original clinkers and to lie above the line of lime saturation, This result indicates an expected thermal deactivation of C d , as well as of Cas. Deactivation of the C3S alone would cause a decrease in the rate of hydrolysis and therefore a tendency to a decrease in the concentration of calcium hydroxide, The observed increase, however, indicates that a higher threshold concentration of calcium hydroxide was necessary for reaction with CaA, OW-

October, 1934

INDUSTRIAL AND ENGINEERING

ing presumably to thermal deactivation of the CIA. The deactivation effect, as regards the setting apparently is negligible compared to the effect of deactivation of Cas, just as the effect of activation of C3A is relatively negligible in the reverse case of seasoning, a t least with seasoning to the t-stage.

RESTORATION OF SET Reciprocal to reversion in set is the restoration of set by a n additional absorption of water vapor. Such restoration of the set of cements is well known. From the standpoint of the present work, reversion and restoration should be able to continue as long as sufficient C3S is available for further absorption. In addition to restoration by absorption of water vapor, the possibility exists also of restoration by desorption of carbon dioxide. Meyers (8)seems to consider such a desorption possible and describes restoration to slow set by placing a cement in a sealed atmosphere over slaked lime. Unfortunately it was not proved that the aqueous tension of the atmosphere really was slight or had no effect. In the present work, restoration of set by absorption of water vapor was effected on clinkers rendered quick setting by heat and by carbon dioxide. Humidified air a t room temperature was passed for several days through 60 grams of clinker in a test tube. The time of approximate arbitrary “initial” set was measured on 20-gram samples. On heating t-clinker D to about 300” C. almost all the absorbed water was driven out, leaving a carbonated residue for which R was exceptionally small-only 0.01 (Table X). Treatment with air at 70 per cent relative humidity increased R to 0.32 and increased the time of set to 40 minutes; a t 90 per cent relative humidity R was increased further to 0.47, and the initial set was now much slower, taking place in 2 hours and 30 minutes (Table X). TABLEx. RESTORATION OF HEATEDt-CLINKER D TO SLOW SETTINQ BY EXPOSURE TO WATER VAPORFOR 3.5 DAYS RELATIVE HUMIDITY ?lo

CoLtr ol 70 90

ABSORPTIONMIXING Loss ON WATER INITIAL TIMEOF SET IGNITION RATIO,R % 01, Hr. M i n .

.-

I-

1.81 2.28“ 2.52

0.01 0.32 0.47

21.5 21.5 21.5

0 0 2

5 40 30

COz absorption on humidification assumed equal to that at 90% humidity.

CHEMISTRY

1083

phase. The increase in reactive calcium hydroxide on seasoning is explained by activation, for which direct experimental evidence exists, of tricalcium silicate by absorbed water vapor. The tricalcium silicate thereby hydrolyzes more rapidly to yield calcium hydroxide in solution, and is thus more effective as a natural retarder of the set. A n undesirable overseasoning was observed on strong absorption of water vapor. The overseasoning is attributed on the basis of direct evidence to activation by the excess absorbed water vapor of tricalcium aluminate, in addition to activation of tricalcium silicate. Changes in liquid-phase composition also point to activation of tricalcium aluminate. At the humidifications that were employed, the deleterious consequences as regards setting were not excessive (though the strength may have been more seriously affected). Artificial seasoning of a clinker, it is apparent, must be considered with regard to a possible adverse overseasoning of the clinker, Carbon dioxide absorbed in excess of water vapor accelerates the setting. It is assumed as reasonable that the carbon dioxide may react with potential calcium hydroxide in the tricalcium silicate-water vapor complex, resulting in deactivation of the CIS with respect to its hydrolysis, and hence in acceleration of the setting. A deactivation effect of carbon dioxide is by analogy postulated also for activated tricalcium aluminate, but any such deactivation of C3A by carbon dioxide appears to be very subordinate, in accord perhaps with the relatively subordinate effect on setting of activation of CIA. The effectiveness of carbon dioxide in causing reversion of set was found to depend upon the state of seasoning. Independently of the seasoning, certain clinkers showed an inherent susceptibility to quick setting to carbon dioxide. The susceptible clinkers were differentiated from the resistant clinkers on the basis of chemical composition and other properties. Reversion in set of clinkers was obtained by abstraction of absorbed water vapor-that is, by heating a t a sufficiently high temperature. Besides the expected deactivation of tricalcium silicate, evidence was obtained also of deactivation of tricalcium aluminate, but just as in the reverse case of seasoning, the effect on the C3S predominates as regards setting. Restoration to slow setting was effected by humidified air, of clinkers that have become quick setting because of the action of heat or of carbon dioxide. The restoration is accompanied by a fresh absorption of water vapor in excess of carbon dioxide. The restoration is interpreted as due to activation by the water vapor of residual tricalcium silicate in Portland cement, much as in ordinary seasoning.

The t-clinker E, rendered very quick setting with carbon dioxide (Table VII), was restored to slow set when exposed to air a t 90 per cent relative humidity but not, a t equal duration of the exposure, with the air a t only 70 per cent relative humidity (Table XI). In spite of an initial high ignition loss of nearly 3 per cent, apparently enough CaS was available LITERATURE CITED for fresh absorption of water vapor with the relative humidity . (1) Andersen and Lee, J. Wash. Acad. Sci., 23, 338 (1933). at 90 per cent to result in an increase in R from 0.55 to 0.71, (2) Assarsson, Sveriges Geol. Unders6kning (in German), Ser. C. 379 and correspondingly in a second slowing of the set of the (1933); Zement, 23, 1, 15 (1934). (3) Assarsson, 2. anorg. allgem. Chem., 200, 385 (1931); 205, 335 clinker. (1932); 214, 158 (1933). TABLE XI. RESTORATION OF CARBON DIOXIDE-TREATED t(4) Beckmann, Zement, 16, 37 (1927). CLINKER E TO SLOWSETTIXQ B Y EXPOSURE TO WATER (4A) Discussion of Beckman (4). Prot. Ver. deut. Portland Cement VAPOR FOR 3.5 DAYS Fabr.. 28, 200 (1905). ABSORPTION MIXINQ RELATIVE Loss O N WATER INITIAL TIMEOF SET HUMIDITY IQNITION RATIO.R % % .% Hr. Min. Control 2.95 0.55 24.0 0 4 70 90

4.06 4.365

0.59 0.71

24.0 24.0

0

1

7

45

a COz absorption on humidification assumed equal to that at 70% relative humidity.

CONCLUSIONS The change for the six clinkers from quick to slow set with increase in seasoning, or absorption of water vapor in excess of carbon dioxide, has been correlated with an apparent increased formation of reactive calcium hydroxide in the liquid

(5) Gadd, Brit. Port. Cement Research Assoc. Pamphlet 1 (1922). ( 6 ) Grimm, Zement, 15, 775, 796 (1926). (7) International Critical Tables, Vol. IV, p. 353, McGraw-Hill Book Co., New York, 1928. (8) Meyers, Concrete, Cement Mill Seo., 19, 128 (1921). (9) Nestell and Anderson, J. IND.ENO.CHEX.,9, 646 (1917). (10) Pitt, Chem. Eng. Mining Rev., 23, 227 (1931). (11) Pulfrich and Linck, Kolloid-Z., 34, 117 (1934). (12) Rodt, Zement, 14, 520 (1925). (13) Roller, IND.ENO.CHEM.,26, 669 (1934). (14) Roller, J. Phys. Chem., 35, 1113 (1931). (15) Schaohtschabel, Zement, 21, 643 (1932). (16) Seidell, “Solubilities,” D. Van Nostrand Co., New York, 1919. (17) Wells, Bur. Standards J . Research, 1, 951 (1928). RECEIVED June 10, 1934.