Solubility of Ca(OH),and CaS04.2H20in Dilute Alkali Solutions APPLICATION TO PORTLAND CEMENT PASTES W.C. HANSEiYI AND E. E. PRESSLER Portland Cement Association Research Laboratory, Chicago, I l l .
\vhich the concentrations of eithcr K20 or Sa20 or both ranged from 0 to 0.25 mole per liter of solution.
T h e solubilities of Ca(OH)2 and CaSO4.2HzO in solutions of KaOH and KOH have been determined at 25" and 30" C. for concentrations of the alkalies up to 0.25 mole per liter. These data were used to prepare curves from which the degree of supersaturation of the liquid phase of a cement paste can be estimated. Examples of the type of information obtained from such studies are gil en.
EXPERIMESTAL PROCEDURE
Khen cement reacts with water, a t least three reactions tend tci form solid Ca(OH)2: CaSOa
I
N EXPLAINING the setting and hardening of portland ce-
3Ca0 SiO,
ment pastes, Le Chatelier (4) pointed out that the anhl-drouq cement minerals dissolve in the mixing water and produce solutions that are supersaturated with respect to the hydration products. It seems probable that the formation of supersaturated solutions plays an important part in the setting and hardening of portland cements, and that further information as to the extent of supersaturation in the liquid phase may be useful in explaining the behavior of cements during the early periods of hydration. In previous studies of the hydration of 3Ca0.SiOz, Lerch and Bogue (6)and Flint and Wells (1)observed that the solutions were supersatuiated with respect to Ca(OH)2; and it is generally believed that a similar condition occurs in the hydration of portland cements. However, from the solubility data previouslj available, it has not been possible to calculate the degree of supersaturation in portland cement pastes. The data given i n this report on the solubility of Ca(OH)2 in solutions of KOH, and X'aOH saturated with gypsum provide the information requireii for such calculations. Studies such as those of Kalousek, Juniper, and Tregoning (21 show that, in general, the principal components of the liquid phase of portland cement pastes at early ages are CaO, Ii20, Sa20, and SO3, and that the minor components are SiOz,A1,03, and in some cases Cr203. K i t h the twelve clinkers used in their studies, thc K20 ranged from 0.05 to 1.32y0 and the 9 a 2 0 from 0.02 to 0.87%. Very few, if any, of the cements manufactured a t present contain more than 1.37, K1O, but a few of them contain between 0.9 and 1.3% Ka20. These investigators prepared cements from the twelve clinkers by blending the powdered clinker with gypsum and determined the compositions of the solutions from pastes of the cements with water-cement ratios of 0.35 by weight at the ages of 7 minutes and 2 hours. The results of their stud! show that a t these early ages the concentrations of Sa20 in these solutions ranged from 0.0008 to 0.0527 mole per liter and that those of K20 ranged from 0.0002 to 0.2053 mole per liter. Hence the cement technologist is interested primarily in the portion of the system C~O-KZO-X;~~O-SO~-H~O in which the concentrations of the alkalies range from 0 to about 0.25 mole per liter2 Accordingly, except in a few cases, the investigation covered in this report was limited to a determination of the solubilities of Ca(OH)2 and CaSOa.2H20 in solutions of KOH and NaOH in Present address, Universal Atlas Cement Conyany, Buffington, Ind. This applies to the pastes during the first few hours during which the gypsum is being converted t o calcium sulfoaluminate. Folloning this period the concentrations of t h e alkalies may exceed 0 25 mole per liter, but the liquid phase then is primarily a solution of Ca(OH)p, NaOH, and KOH 1
a
+ 2(i\'a, K)OH = ( S a , K),SO, + Ca(OH)z + water
= zCaO.SiO2.hq.
CaO
(1
f (3 - z)Ca(OEI)2 (2)
+ H20 = Ca(OH)2
(3 i
Preliminary tests showed that in very dilute solutions of KOH reaction 1did not saturate the solution with Ca(0H):. IIowever. in cement pastes reaction 2 furnishes more than sufficient Ca(OH)? to saturate the solution. Accordingly, regardless of the alkali contcnt of the cement, the liquid phase of a cement paste tends to become supersaturated with respect to Ca(OH)2, and during the first few hours while gypsum is a component of the solid phase the liquid phase tends to establish an equilibrium with respcct to both CaSO4.2Hz0 and Ca(OH)2. During this peiiod in which Ca(OH)Z is precipitating as fine crystals, there probably is not an opportunity for the establishment of a true equilibrium because the solubility of Ca(OH'12 (7') decreases as the size of the crystals in the solid phase increases. Hence studies such as those to be described can show only approximately the degree to which the liquid phase is supersaturated with respect to Ca(OH)p at anv given period. In accordance with reaction 1, mistures of gypsum and alkali hydroxides react to form alkali sulfates and Ca(OH)2, and mixtures of Ca(OH)2 and alkali sulfates react to form gypsum and alkali hydroxides. I n thjs investigation the reaction mixtures were prepared by adding an excess of both finely powdered gypsum and CaO to standardized solutions of the alkali hydroxides. The CaO was added to ensure saturation with Ca(OH)z in all of the mixtures and to assist in the formation of nuclei for the crystallization of the Ca(OH)2. The reaction mixtures, in Erlenmeyer flasks, were agitated for XI to 48 hours in a water bath a t either 30 =t0.1 or 25 + 0.1 O C. This period of shaking was adopted after a few preliminary trials showed that longer periods of shaking did not have a marked effect upon the quantities of CaO and SO3in the liquid phase. For euample, n-hen the concentration of KzO was 0.0050 mole per liter, the concentrations of CaO were 0.0298 and 0.0295 mole per liter, respectively, for shaking periods of 46 and 144 hours, and the concentrations of SO1 were identical for the two periods of shaking. At the end of the shaking period the reaction mixture was poured into a Buchner funnel fitted with a dense filter paper, and sufficient liquid for the analyses was drawn off by suction. This n as transferred immediately to a stoppered flask from which the samples were measured. The alkalinity of one sample was determined a t once by titration with sJandardized HC1 (approximately 0.1 X). The CaO Tyas determined by precipitating as the oxalate and titrating with standardized KMn04. The SO, was precipitated and weighed as BaSOIt and the Na20, which was determined for only a few of the mixtures, was precipitated and weighed as sodium uranyl zinc acetate. Table I gives the results of this study.
1280
INDUSTRIAL AND ENGINEERING CHEMISTRY
October 1947
TABLE I. SOLUBILITY OF Ca(0II): S01,VTlOKS O F Compn. of Original Soln , hIole/Liter KzO S a l 0 KzO+SazO Sone 0.0050 0.0025 0.0150 0.0100 0.0251 0.0125 0.0150 0.0251 0.0150 0.0301 0,0401 0,0200 0.0150 0.0501 0.0251 0.0251 0,0050 0,0500 0.0601 0.0301 0.0150 0.0501 0.1002 0.1002 0,1002 0.1504 None 0,1504 0,1504 0.2005 Sone 0.2506 0.1253 0.3750 None 0,5012 I . 0023
Sone h-one 0 0025 0 0050 0 0101 Kone 0 0125 0 0151 0 0050 0 0101 Sone Sone 0 0202 0 0252 None 0 0252 0 0252 0 0500 0 0050 Sone 0.0302 0,0504 0 0252 Xone 0 0050 0.0252 Sone 0 1448 0 0050 0 0252 None 0 1930 None 0 1250 None 0 3750 Sone Sone
h'one 0.0050 0.0050 0,0200 0.0201 0.0251 0,0250 0,0301 0,0301 0.0301 0,0301 0.0401 0,0402 0.0402 0,0501 0.0503 0.0503 0,0550 0.0560 0.0601 0.0603 0.0654 0.0753 0.1002 0,1052 0.1254 0,1504 0,1448 0,1554 0,1756 0.2005 0.1930 0,2506 0.2503 0,3750 0.3750 0.5012 1.0023
KOH
ASD
SaOH
~ S DCaSO4.2H20 IS AT 25 AiYD 30 ' C.
Cornpn of S o h in Contact with Solid Ca(i3H)z and Solid Cas06 2 1 1 ~ 0 ;\Iole/I,iter , -___
KzO
CaO
D A T AF O R 0.0319 0,0298 0.0294 0.0256 0.0236 0,0228 0,0226 0.0226 0.0231 0.0216 0.0214 0.0211 0.0198 0.0195 0.0185 0,0188 0.0194 0,0184 0,0190 0.0187 0.0180 0.0175 0.0174 0.0162 0.0167 0.0164 0.0162 0.0150 0.0162 0.0162 0.0164 0.0149 0.0161 0.0169 0,0094 0.0051 0.0093 0,0034
SO:
OH
30° C. 0.0125 0.0141 0 0141 0.0191 0.0196 0,0223 0,0219 0,0236 0.0241 0.0252 0,0253 0,0299 0.0303 0.0303 0,0373 0.0376 0.0363 0.0395 0,0398 0.0430 0.0425 0,0468 0,0544 0.0710 0.0783 0.0961 0.1167 0.1151 0.1211 0.1382 0.1626 0.1586 0.2055 0.2040 0.2121 0.2629 0,2083 0,3662
0.0387 0,0419 0.0393 0,0529 0.0477 0.0623 0.0514 0.0685 O.OZ37 0.0536 0.0524 0.0626 0,0602 0,0689 0.0618 0.0633 0.0672 0.0698 0.0689 0.0710 0 0711 0.0734 0.0771 0.0806 0,0884 0,0940 0.0972 0,1024 0.1011 0.1048 0.1062 0.1146 0.1136 0,1188 0.1456 0.2279 0.1386 0,2752
YazO
.... 0,0052 ..;. .... ( 1 . .
....
0.0052
.... ,...
....
.... .... .... 0,0253
0.0251
0.0504 0,0052
....
....
0.0502 0.0256 ,...
0,0056 0,0259
.... ...
... . . I
1
.... ,... .... ....
... ...
....
+
SazOa
paled.
0 0001 0,0053 0.0044 0.0199 0,0199 0.0256 0.0250 0.0302 0.0303 0.0304 0.0301 0.0399 0.0406 0.0401 0.0497 0.0502 0,0505 0.0561 0.0552 0.0598 0.0599 0,0659 0,0756 0.0951 0.1058 0.1257 0.1491 0.1513 0,1653 0.1744 0,1991 0.2010 0.24626 0.2465b 0.2565b 0.3718 0.26836 0.5000b
DATAFOR 25O C . Sone Sone None 0.0331 0,0292 0.0100 None 0.0100 0.0200 0.0240 0.0200 Sone 0,0401 0.0211 0,0401 None 0,0183 None 0.0601 0.0601 0.1002 0.0161 0.1002 None None 0.2005 0.0165 0.2005 a Calculated from the determined vali b P n S 0 , K,SC)A H.0 nrnhnhiv formed
0.0123 0.0151 0.0190 0,0289 0.0426 0,0722 0.1562
0.0414 0.0488 0.0502 0.0649 0,0726 0.0888 0.1164
0.0001 0,0103 0,0201 0.0403 0.0606 0 . I005 0.19796
1281
crystals in the >olid phase. For this rcwon thp data obtained in this study are not considered to reuresent conditions of true etruilibrium. IIowever, it is belicvcrf that the curves of Figure 1 furnish a means of estimating fairly accurattzly the degree of supeIsaturation of the liquid phase of a cement paste. This use of the data is tlemonstrated in t h r discussion which follows. APPLICATIO\S TO CEMENT PASTES
The data uqed in this discussion for cement 1 art taken from the work of Kalousek, Jumper, and Tregoning ( 2 ) and those for cements A, B, and C are from studies made in this laboratory. Table I1 gives the chemical compositions of the cements. Kalousek and eo-workers determined the eompositions of the solutions taken from cement pastes at the end of 7 minutes. The paHtes had a watercement ratio of 0.35 by weight, and were made from cements prepared by blending powdered clinker and poTvdered gypsum. In this laboratory the solutions were removed at the end of 3 minutes from pastes having a water-cement ratio of 0.50 by weight. The cements were commercial products in which the clinker and gypsum were interground. CEMENT A. This cement contained 0.0470 NazO and 0.19yo KzO and the composition3 of the liquid phase a t 3 minutes was 0.0412 CaO, 0.0151 SO1, 0.0006 Na20, and 0.0022 KzO. The sum of Sa20 and KzO was 0.0028. Figure I indicates that a KzO) requires solution containing 0.003 (Na20 0.031 CaO and 0.013 SO3 for saturation with respect to CaSO4.2HZ0and Ca(OH),. Hence, the liquid phase from the paste of cement A would have to precipitate 0.0412 - 0.0310, or 0.0102 CaO, and 0.0151 - 0.0130. or 0.0021 SO,for its comDosition
+
I
RESULTS AT TWO TEhlPERATURES
DATAAT 30" C. The data of Table I are plotted in Figure 1 t u show the manner in which the concentrations of CaO and SO,vary with changes in the concentrations of the alkalies, when the solid phases are Ca(0H)Z and CaSOa.2Hz0. In a number of eases, from two to four values were obtained for one concentration of KaO IUazO, but only one of the values was plotted when the spread brtween the values was not sufficient to be shown by the scale used in the figure. I t appears from this plot that the solubilities of Ca(0H)z and CaSOa.2HzO are nearly the same in the solutions of KOH plus NaOH as they are in the solutions of KOH. Since the cement technologist is interested mainly in solutions containing both K 2 0and Sa20 in which the concentrations of KZO are generally much greater than those of NazO, only a f e r studies were made ivith solutions containing SazOalone. The results with KazO alone indicate that the solubilities of Ca(0H)z and CaS04.2Ha0are slightly different in the solutions of NatO from those in the solutions of the two alkalies or of K20 alone. DATAAT 25" C. The data obtained a t this temperature are given in Table I. A comparison of these results with t,he data a t . 30" C. s h o w that this change in temperature did not have a marked effect upon the solubilities of either Ca(0H)Z or CaSOI.2H20 in the solutions of the alkali hydroxides. Hence, in the applicationof theresultsof thisstudytocement past,es, dataobt,ained a t normal laboratory temperatures may be used without serious error. It was pointed out in the discussion of the experimental procedure that the solubility of Ca(OH), varies with the size of the
+
phase of this particular paste a t 3 minutes was supersaturated with respect to both Ca(OH), arid CaSO4.2H2O. CEMEXT 1. The clinker from m-hich this cement was prepared contained 0.02% NazO and 0.1170 KzO, and the composition of the liquid phase a t 7 minutes was 0.0417 CaO, 0.0127 SOa, 0.0008 KazO, and 0.0018 KzO. The sum of NazO and IGO was 0.0026, which is very close to the value of 0.0028 found for the liquid phase of cement A. Hence, according to Figure 1 the liquid phase from the paste of cement 1Should have nearly the same composition as that from the paste of cement A-that is, 0.0310 CaO and 0.0130 so^. The degree of supersaturation is, therefore, 0.0417 - 0.0310, or 0.0107 CaO, and 0.0127 - 0.0130, or - 0.0003 SO3. From these values it appears that the liquid phase of the paste of cement 1 was not supersaturated with respect to gypsum but was supersaturated to the extent of 0.0107 mole per liter with respect to Ca(OH21. 3 In all cases in this discussion the compositions of the solutions are given in moles per liter of solution.
~ 11, ~cHEbIICAL ~ ~, O ~ ~ ~ \ I P O S I T IOF O xcEAIENTS, IN pERcENT
T
BY
WEIGHT
Ignition Fezoa ca' 'Igo Lc'ss 22.0 22.2 5 . 71 2 ., 15 ".' 65.8 1 . 1 O1.62 .O1 1.82 22.4 5.0 3.0 63.1 2.5 1 . 7 3 0.92 C 25.3 3 . 9 3.1 63.6 1 . 1 1.87 0.81 a Composition of clinker used i n preparation of cerlent Kalousek and oo-workers (3) as clinker S o . 3. Cement
'"
'.'
Nazo 0.04 0.06 1.00 1, from
Kzo 0.19 O.ll 1.30 0.08 data of
INDUSTRIAL AND ENGINEERING CHEMISTRY
1282
Vol. 39, No. 10
I t is known (3,8)that a part of the sulfur in cement clinkers is present in the form of alkali sulfates, and that these sulfates dissolve quickly lvhen the cement is mixed r i t h water. In order for the dissolved alkali sulfates to satisfy the equilibrium required by reaction 1, it is necessary for them to react n-ith Ca(OH), and precipitate CaSOr.2H20. dinw most cements contain minor :tinourits of uncombined CaO, the Ca(OIIj2required for this ritaction with the alkali sulfates has to be obtained through thcb hydrolysis of the other minerals, as illuHtrated in react,ion 2. 011 thcs basis of these caonsidciations, it appears that, this particular iwment may have contained a relatiwly large amount of readily wluble hlkali sulfates? and that thy rates of hydrolysis of the calcium minerals were not sufficient t o provide in 3 minutes thts ( ' a 0 rquirctl to react n-ith the ulfatcs. It is also a possibilit,?, x-: iiliratly stated undcr the iliscussion of cenicnt 1, that a iempc)rary s oI SO3 may have hwn produced in the solution due ro t~lic~ n i of~ cither hc:niihydratr or soluble anhydrite in the c i ~ n i i ~ ~ i t , e hypothcsis that sonic cemrnts contain s u f h c i d alkali siilfater to require an approciable period of time for th(' other niiritrralp to supply thc ('a(OI1 i 2 necessary to establish equilibrium hi,t n w n the sulfatrs and hydroxides of calcium, sodium, and pota3siuni may demand sunie icvisions in r in theories pertaining t i h c x role of gypsum i n preventing flash set. Roller (61 accoun1x fur tlic ability of gypsum to prevent flash set of cements protlucrxl (ram rlinkers of high alkali content, on the basis that, the. g>-psuni w i i - t .- with the alkali hytlrosides i o form alkali suli'aic~s ancl ( xa(OH)2. .lc>tually the clinlwr from which ccment 13 w i t ' pro~ l u c c drc~cluirdgypsum to lirevent flash set even though it appcai'-: 11i:tt rhc major part of the i w d i l > - soluble alkalics entcrcd t h i , as ,dfatcle. It is not thta purpose of t,tiis p a p i ' t o I~I,V~IS t h(>various theorieh on thc ttiny and liardcning of c ~ c ~ ~ i i c ~ n t , : t i i d this c!xaniplii is cited mi.rely t o indicate the possiliilitic~i( i f thi-: iiiethocl of qrudying the reactions of cement with ivttt(>r. ( 'MEST C. This cenieiit contairicd l . 0 O C Sa20 ~ ~ and 0.08' , K!O, and the cliiiker had a n SOa cciiitcrit of O.G7%. Ttic rotiipncir i i j n of tho liquid phase was 0.0481 (la(), 0.0883 SOi, C.0623 SarO. : + i ~ t l 0.0082 &O. The concentmtioiis of C:tO and SO,, i n :I stturated soliitiiiii ivitli an alkali wntent of 0.0705 i q 0.0170 ('a() iintl 0.0490 SO?. Th(1 tlcgrce of supersaturation of the liquicl piinsf, of the paste !vas thewfore 0.0481 - 0.0170, or 0.0311 C'ao. i t i d 0.0883 - 0.0490, or 0.0393 SO3. H(*rc,, aq with cement 13, the, i ~ x ~ ~ CaO . ~ s , is i m t sufficient to satisfy the cxcc:?s SOs as (:aSO,. These data for cements B and C iudicate that both L O and S a 2 0exist in cement clinkers as sulfates, and that several niinuteb arc rcquired for the hydrolysis of the other minerals to supply t,hr ulfates in the liquid ph (:a(ORi2 required to react nith th the past,es. Examples of this apparent supersaturation of tho nolurions with alkali sulfates could be cited from the work of Kalowek and eo-workers, and from the studies with other ecmcnts in this laboratory. IIowver, the cases cited appear t o be sufficient to demonstrate the type of information that may hc obtained by studying the comporitions of fikrates from cement pastw by means of Figure 1. (!
I
.O2
.06
.IO
I .I4
I I
.I8
Moles K20 and Na20 per liter
Figure 1. Coiicentrations of CaO a n d SO3 r s . tioris of KZO a n d Ya20
.22
C : o i i c ~ nI ria-
Gypsurn is knon-n t