,
.'
Heat of Hydration of Calcium Sulfates J
J. C. SOUTHARD Pacific Experiment Station, U. S. Bureau of Mines, Berkeley, Calif.
K
NOWLEDGE of the energy relations among
The heats of the following reactions involving water and the various forms of calcium sulfates have been determined calorimetrically by solution in hydrochloric acid and in part by direct hydration:
the various forms of calcium sulfate and its hydrates has depended on contradictory CaS04.l/&O(stable ov.ystalline) */zHzO = CaS04.2Hz0(~0iid) though extensive data. A review of the literature aHZss= -4100 * 20 cal./g. f. W. on the subject is given in a recent paper by Newc ~ ~ ~ ~ . ; / , 4H*/rH@ ~ ~= CaSOdHzO(sO1id) ( ~ ~ ~ ~ ~ ~ man and Wells (9). The present investigation AH208 = -4600 * 20 cal./g. f. w. of the heat of hydration of calcium sulfates is 2H20 = C a S 0 ~ . 2 H ~ 0 ( . ~ i i d ) CaSOi(ao1ubleanhydrite) part of a program to clarify these uncertainties AH^^^ = -7210 * 10 cal./g. f. w. and establish a self-consistent set of thermal data. CaS%natural anhydrite) 2Hz0 = CaSO4.2Hzo(~0lid) The heats of hydration of soluble anhydrite AH^^^ -4040 20 cal./g. f. W. (CaSOJ of metastable hydrates containing water corresponding to CaS04.l/zHzO, CaS04.1/bHz0, Heat-of-solution measurements indicate that commerCaS04.1/loHz0, and of an unstable variety of c ~ S ~ ~ have . ~ been / ~ determined H ~ ~ by direct hycial hard wall plaster is a mixture of the stable and dration in a calorimeter. Other forms could not metastable hemihydrates, and that casting plaster is be made to hydrate in a reasonable time, and it was composed almost entirely of the stable crystalline form. necessarv to resort to the indirect method of determining heats of solution in hydrochloric acid. days. The excess water was then blown o f f and the residual Measurements by this method were made on natural anhyhemihydrate dried by blowing air through the autoclave drite, crystalline calcium sulfate hemihydrate, selenite from Gerlach, Nev., gypsum from the Perkins deposit in Alaska, while still a t 115" C. It was found to contain 6.23 per cent artificially prepared insoluble anhydrite, building wall and water. The natural anhydrite came from Ludwig, Nev. It was casting plasters, and set plaster, as well as on soluble anhyground to pass a 100-mesh screen, washed with hydrochloric drite and the metastable hemihydrates. acid, and dried before use. Analysis showed it to be free from water, carbonates, R203,and chlorides, but to contain 1.15 Materials per cent insoluble material which was impervious to attack All materials except the natural anhydrite and building by hydrochloric acid solution. plasters were prepared by the dehydration of carefully selected The gypsum from the Perkins deposit, Alaska, was a white large crystals of selenite from Gerlach, Nev. The selenite microcrystalline material. It was used as received after was ground in a porcelain ball mill to pass a 100-mesh screen. grinding to pass a 100-mesh screen. Its analysis was: H20, After grinding it contained 20.88 per cent water (theoretical, 20.75 per cent; CaC03, 0.28; R203,0.05; insoluble, 0.17. 20.92 per cent), was completely soluble in water, and showed The hard wall plaster was made by calcination of gypsum a negative test for carbonates. Dehydration to the desired from San Marcos Island, Mexico, in a plant kettle. I t conwater content was usually accomplished by heating the tained 0.97 per cent SiOz, 0.21 R203, 0.69 CaC03, 92.29 ground selenite in a n evacuated flask (1 mm. of mercury) at CaSOA,and 5.86 HzO. The casting plaster contained 0.43 90" to 100" C. for 10 to 12 hours. This treatment yielded per cent S O z , 0.07 R203, 1.04 CaCOa, 92.81 CaSOa, and 5.72 a n extremely reactive but unstable variety of hydrates. On HzO. They had a normal consistency of 75 and 60 and a further heating a t 100" for several days, a metastable, though setting time of 45 and 30 minutes, respectively. The "set" completely reproducible, variety of hydrates was formed. building plaster contained 77.63 CaSOd, 20.36 H20, 0.80 Samples containing water corresponding to CaS04.1/2Hz0, CaC03, 0.23 &os, and 0.97 insoluble. CaSOd.l/aHzO, and CaS04.1/10H20 were prepared in this manner. Soluble anhydrite was prepared by further dehydraApparatus tion for several days at 90" to 100" C. in a vacuum of 10-4 The direct determinations of the heat of hydration were mm. of mercury. Samples prepared in this manner still conmade in a tantalum calorimeter described by Maier (7). It tained 0.16 to 0.27 per cent water. The water could be rewas altered only by replacing the ethyl bromide boiler, which moved completely only at somewhat higher temperatures, had been used for adjusting the initial temperature, with a which resulted in a considerable change in rehydration vesimple copper cooling coil through which cold water could be locity that may have been due to partial conversion into incirculated. soluble anhydrite. The sample of insoluble anhydrite was The heats of solution in hydrochloric acid were made in a made by dehydrating ground selenite by heating in a muffle calorimeter built for the purpose (Figure 1). Preliminary a t 890" C. for 4 hours. calculations indicated that temperature changes of 0.02' Crystalline CaS04.1/2H20was prepared by heating ground to 0.2" C. were all that could be expected. Hence, high selenite and water in a stirred autoclave a t 115" C. for several
+
+
+
442
-
j=
~
~
MARCH, 1940
INDUSTRIAL AND ENGINEERING CHEMISTRY
thermometer sensitivity and well-controlled heat-exchange conditions were required if the desired degree of accuracy was to be obtained. The calorimeter vessel consisted of a narrow-necked 2-quart Dewar flask ( A , Figure 1) completely immerse! in an oil bath thermostatically controlled at 25' C. * 0.01 . Temperature measurements were made with a modification of the transpoeed bridge thermometer previously described by Maier (8) and indicated in Figure 1 bv B . This type of the';mometkr consists of a fixed bridge arrangement made of four resistances, alt e r n a t e l y m a n g a n i n and copper. Such an arrangement is in balance at one temperature only. The degree of unbalance is a measure of the displacement of the temperature from this point and can be measured potentiometrically. I n t h e present work the thermometer showed zero potential at about 24" C. Each of the coils had a resistance of about 275 ohms, and a fixed current of 0.002 ampere was maintained. The resulting temperature coefficient was nearly 1000 pv per degree, which could be readily measured to 0.1 pv. The copper thermometer coils were wound of NO. 42 enameled wire in a single layer on a thin-walled (0.01inch or 0.254-mm.) copper tube which had been carefully insulated with a thin layer of Bakelite varnish. The manganin thermometer coils were wound on the outside of the co per coils. At the upper e n f of the copper tube was wound a 100-ohm manganin heating coil which was used in adding known amounts of energy. This assembly of resistances was covered with another thinwalled copper t u b e a n d soldered at the ends. I t was suspended from top cover C by three '/(-inch (3.2-mm.) FIGURE 1. CALORIMETER FOR thin-walled copper-nic kel HEATSOF SOLUTION IN HYalloy tubes, D, of low heat DROCHLORIC ACID c o n d u c t i v i t y . The thermometer was never calibrated in an absolute sense but was used as an instrument for comparing the change in temperature caused by a known and an unknown amount of heat. The actual temperature of the calorimeter a t the end of a run was determined with a 0. l o C. mercury-in-glass thermometer that had been tested at the National Bureau of Standards. Stirring was provided by a multivaned impeller carried on the end of a 3/s-inch (9.5-mm.) Monel metal tube shaft that had been turned down to less than 0.01 inch wall thickness. This shaft was supported in two ball bearings outside of the calorimeter. The axes of the stirrer shaft and the copper thermometer tube were coincident. The level of the impeller blades was about 1 cm. below the top of this tube, and the shaft rotated in such a way that a continuous stream of liquid was forced down the center of the thermometer. This subjected any material lying on the bottom of the calorimeter to continual washing and thus hastened solution. -4constant stirring rate of about 500 r. p. m. was provided by a */6-h. p. electric motor through a suitable arrangement of pulleys and belts. Such vigorous stirring contributed heat at a measurable rate (about 0.2 calorie per minute) but was necessary t o accomplish in a reasonable time the solution of such inert material as natural anhydrite. All of the metal parts within the calorimeter were made acid resistant by a coating of Bakelite lacquer which had been thoroughly baked. It was found necessary to provide additional protection to the soldered seams with a layer of beeswax and rosin. The sample was introduced during the assembling of the calorimeter in a thin-walled glass bulb, E, sealed onto the end of a glass
443
tube which ran up through the hollow stirring shaft to the outside. At the desired time the bulb was broken by jerking it up against three prongs that projected downward from the impeller. The electromotive force developed by the calorimeter thermometer was measured on a White 100,000 pv double potentiometer, and the thermometer current was adjusted to a fixed value by balancing the potential drop across a fked manganin resistance against a fixed dial setting of the same potentiometer. The energy input during calibration runs was made with the same instrument in combination with Rosa-type standard resistances calibrated by the National Bureau of Standards. Time was measured with a stop watch which was calibrated twice during the course of this investigation by comparison with a standard chronometer of the Astronomy Department of the University of California. The electromotive force of the standard cell was also checked on two occasions by comparison with three standard cells which are used only for reference and whose electromotive force relative to each other had not changed by one part in a million. The factor 4.1833 was used to convert International joules to defined calories. The determinations were carried out in such a manner that the final temperature was within 0.1' of 25" C. in nearly all instances, whence corrections to this temperature are negligible. The actual calorimetric precision was better than one part per thousand, but somewhat less accuracy is claimed because of the possible incompletion of certain hydration reactions and because the heat of hydration in some instances depended on the difference in two determinations.
Direct Measurement of Heat of Hydration The method of direct hydration in the calorimeter could be applied only to metastable forms. The more densely crystalline stable hemihydrate and natural anhydrite could not be made to hydrate completely in any suitable time. The results obtained by this method are shown in Table I. The time required for the completion of the hydration reaction ranged from about 40 minutes for the extremely reactive unstable unannealed type of CaS04.1/2H20to 2 hours for soluble anhydrite. I n each instance '/4 to gram formula weight (9. f. w.) was used. The calorimeter proper contained 5 liters of distilled water, and the tantalum reaction vessel 500 cc. of a saturated calcium sulfate solution. At least two calibration runs were made in connection with each heat-ofsetting determination and the heat-loss corrections based on the straight-line relation between the rate of change of temperature and the temperature, which was ascertained in every instance. TABLE I. HEAT OF HYDRATION OF CALCIUMSULFATES TO CASO4.2HzO AS DETERMIXED BY DIRECT HYDRATION I N A TANTALUM CALORIMETER Hz0
Material
HIO
%
Soluble anhydrite Soluble anhydrite Soluble anhydrite
6.17 6.41 6.14 6.41 6.21 2.40 1.13 0.45 0.45 0.16 0.16
per Mole CaSO4
Mole
... ...
o:5is
0.500 0.186 0.086 0.034 0.034 0.012 0.012
H e a t of Hydration Cal./g. 2. w. Cas04 5200 5182 4957 4514 4566 6178 6752 7009 7013 7149 7188
Measurement of Heat of Solution in Hydrochloric Acid These measurements were necessary to be able to calculate
the heats of hydration of natural anhydrite and the stable crystalline hemihydrate; at the same time it afforded a n op-
VOL. 32, NO. 3
INDUSTRIAL AND ENGINEERING CHEMISTRY
444
TABLE 11. HEATOF SOLVTION IX CALORIES PER GRAM-FORMULA-WEIGHT OF CALCIUM SULFATEIN 53.43 KG. OF 2.031 N HYDROCHLORIC ACIDSOLUTION Gerlach Selenite CaSO4.2Hz0 +5672 5668 5666 5663 5673 Mean Cor. for diln. of HClsoln. Cor. for CaCOs
Alaska Gypsum, CaSO4.2Hz0 +5636 5607 5630
5668 32
.. -
Cor. mean
+5700
Metastable Hemihydrate, CaSO4.*/zHzO (Prep. 2) +lo91 1089
Metastable Hemihydrate CaSOd I/zHzO (Prep. 3) 4-1112
.. ..
.. ....
.. .. *. ..
5628 32 38
1090 8
1112 8
.. ~
+5698
+lo98
.. +1120
portunity t o ascertain heats of hydration on the other forms by an independent method. The weight of sample in each instance was that equivalent to 6.000 grams of gypsum. Each sample was dissolved in 1862 grams (vacuum) of 2.03 N hydrochloric acid solution. The time required for complete solution ranged from 20 minutes for gypsum to 2 hours for metastable hemihydrate. The latter tended t o set into a solid cake before more than fractional solution had taken place. However, complete solution was finally obtained in each instance, as was shown by both visual inspection and analysis of an aliquot portion of the calorimeter contents. The results of the heat-of-solution determinations are shown in Table 11. A correction corresponding to 16.0 calories per mole of water was added for the heat of dilution of the solution by the water contained in the hydrated calcium sulfates. Thus for CaS04.2H20 the average observed heat of solution was 5668 calories and the corrected value, 5700 calories. This correction was determined by measuring the heat of dilution by a known amount of water of the solution obtained on dissolving anhydrite in the hydrochloric acid in the calorimeter in the course of a regular determination. This correction corresponds closely to that (15.8 calories per g. f . w. water) given by Rossini (12) for pure hydrochloric acid of the same concentration.
Crystalline Hemihydrate CaSO4.‘/zHzO +I605 1611 1606
.. ..
1607 8
Soluble Anhydrite, CaSOn -1404 1404
.. .. .. -1404 .. ..
.. -
+1615
-1404
Insoluble Anhydrite, CaSOn +1671 1673
Natural Anhydrite, CaSOn
+;:E;
.. .. ..
1688 1637
1672
1656
..
-.. +1672
.. *.
.. f1666
Commercial Hard Wall Plaster, 5.86%
HzO +127 ‘7 1257
.. .. ..
Commercial Casting Plaster, 5.72% HzO +1226 1228
+1366
..
.. ..
1227 7 122
5612 32 138
.. ..
1267 8 81
Set Commercial Plaster, 20.36% Hz0 +5512
+1356
..
+5682
can be calculated from the data of Bachstrom (1) on normal hydrochloric acid solution and for complete saturation with carbon dioxide. Table I11 shows the heats of hydration of the various calcium sulfates as calculated from the heat of solution measurements summarized in Table 11.
Discussion of Results Three materials with the molecular composition CaSOa.1/*H20were investigated. The most important is the stable crystalline form, which is the variety on which most of van’t Hoff’s equilibrium measurements (6) and most x-ray investigations (2-5, 10) have been made. If the heat of hydration given in Table I11 is rounded to the nearest 10 calories, a value of -4100 * 20 calories per g. f. w. is obtained, which is in agreement with that given by n’ewman and Wells (9). The_-corrected heats of hydration (Table 111) of the com-
X
B Y D/RECT HYDRATION
0 B Y SOLUTfON l N HCL
TABLE 111. HEATOF HYDRATION OF CALCIEM SULFATES CALCULATED FROM MEASUREMENTS OF HEATOF SOLUTION IN HYDROCHLORIC ACID Water Content
Material
% Soluble anhydrite Insoluble anhydrite Natural anhydrite Mstaatable hemihydrate (prep. 2 ) hletastable hemihydrate (grep. 3) S t a le crystalline hemihydrate Commercial wall plaster Commercial casting plaster ~~
0.27 0.0 0.0
HzO per Mole Cas04 Mole 0.020 0.0 0.0
H e a t of HydraH e a t of tion Cor. Hydrat o Theoretical tion Water Content Cal./g. f. w . Cas04 7104 7211 4028 4028 4044 4044
6.21
0.500
4602
4602
6.25
0.504
4580
4601
6.23 5.86
0,502 0.480
4085 4344
4097 4230
5.72
0.466
4344
4130
a0
0.2.5 0.5 MOLS W A T E RP E R MOL CASO,
FIQURE 2. HEAT OF HYDRATION TO C~S04.2H20OF METASTABLE CALCIUM SULFATES
~
A correction was also applied for the heat of solution of the small amount of calcium carbonate present in the Alaska gypsum. This was determined by measuring the heat of solution in the usual CaSO4-HC1 solution of a weight of pure calcite equivalent to that present in the gypsum. The correction amounts to about 0.6 per cent of the total heat of solution. It corresponds to 8000 calories per g. f. w. calcium carbonate which differs from the value ordinarily used for this reaction because all of the carbon dioxide remained in solution. It is rather to be compared with 8200 calories which
mercial plasters indicate that casting plaster is composed almost entirely of this form, while more than three fourths of aged hard wall plaster may be composed of it. The remaining fourth of hard wall plaster is probably largely made up of metastable hemihydrate. This is confirmed by solubility measurements made by Riddell (11). The rounded value for theheat of hydrationof themetastablehemihydrate is -4600 * 20 calories per g. f. w. No previous calorimeter measurements on this form have been reported. Debye-Scherrer diagrams of Feitknecht show it to be crystalline and of
MARCH, 1940
INDUSTRIAL AND ENGINEERIKG CHEMISTRY
virtually the same structure as the stable crystalline form. However, there can be little doubt of its higher energy content, for the heat of hydration mas determined by two entirely different calorimetric methods on three separate preparations. The unstable, highly reactive variety obtained by dissociation in a vacuum at 100" C. is probably largely a mixture of undissociated gypsum and soluble anhydrite, since its heat of hydration of about - 5200 calories approximates the -5400 calories that would be expected from such a mixture. The heat of hydration of strictly anhydrous soluble anhydrite could not be determined because it was impossible to prepare such material. Heating for days a t 100" C. in a vacuum corresponding to l o p 4 mm. of mercury left the material with a constant water content of around 0.2 per cent. Temperatures in excess of 200" C. were required for complete dehydration. Under these conditions partial conversion to insoluble anhydrite always occurred. To estimate the heat of hydration of absolutely anhydrous soluble anhydrite, it was necessary to extrapolate the values obtained on samples containing various amounts of water. il plot of the heat of hydration us. moles water per mole of calcium sulfate is shown in Figure 2. A straight line drawn through these points closely represents the data of all the hydrates prepared in a gaseous medium below 100" C. The intersection of this line at the zero water-content axis gives -7210 * 10 calories per g. f . w.as the heat of hydration of soluble anhydrite. From this same line the heat of hydration of the metastable hemihydrate was taken to be -4600 calories. The heats of hydration of natural anhydrite and of insoluble anhydrite prepared a t 870" t o 900" C. were found to be -4040 * 20 and -4030 + 20 calories per g. f. w.,respectively. The virtual coincidence of these heats indicates that complete crystallization of anhydrite from the soluble to insoluble form had taken place. The heat of hydration of soluble anhydrite was estimated
445
by Sewman and Wells (9) as being not less than 6990; they state that the heat of transition of soluble anhydrite to natural anhydrite is not less than 3000 calories per g. f. TV. These values are not a t variance with the more definite ones determined in this investigation. Heats of solution in hydrochloric acid were also made on three forms of CaS04.2Hz0-namely, selenite, microcrystalline gypsum, and set plaster. The values obtained for the first two are virtually identical; that of set plaster does not vary from the others by more than can be accounted for by uncertainties in analysis and the corrections involved. It is therefore concluded that there is little if any difference in energy content in the various varieties of CaS04.2HL0.
Acknowledgment The helpful advice and encouragement of C. G. hfaier and K. K. Kelley of the U. S. Bureau of Mines and of W.C. Riddell of the Pacific Portland Cement Company is gratefully acknowledged.
Literature Cited (1) BHchstrom, H. L., J . Am. Chem. SOC.,47,2432 (1925). and Gallitelli, P., 2.Krist., 96, 376 (1937). (2) Bussem, UT., (3) Caspari, W.A., Proc. Roy. SOC.(London), A155,41 (1936). (4) Feitknecht, W., Helv. Chim. Acta, 14,85 (1931). (5) Gallitelli, P., Periodico mineral. (Rome), 4, 159 (1933). (6) Hoff, J. H. van't, Armstrong, E. F., Hinrichsen, W., Weigert, F., and Just, G., 2. physilz. Chem., 45,257 (1903). (7) Maier, C. G., J . Am. Chem. SOC.,52,2160 (1930). (8) Maier, C. G., J . Phys. Chem., 34, 2866 (1930). (9) Kewman, E. S., and Wells, L. S., J . Research Natl. Bur. Standards, 20, 825 (1938). (10) Onorato, E., Periodico mineral. (Rome), 3, 135 (1932). (11) Riddell, W. C., private communication. (12) Rossini, F. R., J . Research .\-atatl. Bur. Standards, 9, 697 (1932). PVBLISHED by permission of t h e Director, Bureau of Mines, United States Department of t h e Interior.
( N o t subject t o copyright.)
Overcoming Thermocouple Errors in HighTemperature Induction Furnaces FRANK DAY, JR., AND HURD W. SAFFORD University of Pittsburgh, Pittsburgh, Penna.
T
FIGURE 1
HIS paper is published so that individuals employing thermocouples in graphite-core tube-type induction furnaces may obtain constant readings. The furnace employed contained a copper-tube induction coil. JThen a thermocouple (22-gage platinum-platinum, 10 per cent rhodium) was placed in the porcelain combustion tube in which experiments are made and the temperature recorded automatically through a potentiometer, the curve was regular up to about 1300" F. and then became erratic (Figure 1, curve A ) . That this effect was obtained only while the induced current was operating was obvious, for when the latter was discontinued the curve again became regular. The irregularity is due to an electromotive force which is set up in the thermocouple in opposition to that normally developed. This might possibly be attributed to thermionic emission from the graphite core a t relatively high temperatures. The emission may build up a charge on the protection tube covering the thermocouple and the porcelain may become sufficiently conducting to carry the charge to the interior.
.