THE COMPRESSIBILITY COEFFICIENTS OF SOLUTIONS OF EIGHT ALKALI HALIDES ARTHUR F. SCOTT, VICTOR M. OBENHAUS,
AND
RALPH W. WILSON
Department of Chemistry, The Rice Institute, Houston, Texas Received November 9, 1933
Although the compressibility coefficients of aqueous solutions have been determined by several investigators,l there exist at present no comparable measurements of this property for a series of solutions of related substances. As a preliminary step to fill this need, the compressibility coefficients of a number of solutions for each of eight alkali halides have been determined at 35°C. over the pressure range 100 to 300 metric atmospheres (megabars). The solutions measured were prepared from purified materials; their composition and density were also measured with considerable precision. PURIFI CATIOI’; O F MATERIALS
The general reagents used in the purification of the salts may be described briefly. Water was doubly distilled, once froin alkaline permanganate and then once from dilute sulfuric acid with the use of tin condensers. Constant boiling hydrochloric acid and hydrobromic acid were obtained by distillation through a quartz condenser. The starting material for the preparation of lithium chloride and bromide was lithium carbonate of C.P. grade. The carbonate lyas first washed repeatedly with water until there was no test for halogen, and was then converted to the chloride or bromide by solution in the halogen acid. In order to precipitate any basic impurities a slight excess of carbonate was left a t the end of the solution. After boiling the solution the residual carbonate was filtered off. The solution containing the halogen salt was then evaporated until a crop of crystals formed on cooling the solution in ice. These crystals were collected and were drained with the aid of a centrifuge. This crystallization operation was repeated until about twothirds of the salt in solution was obtained in the form of crystals. The sample of sodium chloride used was some left over from an atomic weight determination and was of a very high degree of purity. Pure sodium bromide was obtained by a single crystallization of the C.P. salt 1 Literature references t o work before 1923 are given in the International Critical Tables, Vol. 111, p. 439. Measurements reported since 1923 are given in references 1, 2, and 6 .
931
932
A . F. SCOTT, V. M. OBENHAUS AND R. W. WILSON
from water with centrifugal drainage. Sodium iodide which had been synthesized from pure sodium oxalate and iodine was available from a previous research. Potassium chloride was precipitated once from aqueous solution by conducting hydrogen chloride gas to the surface of the solution. Centrifugal drainage was used to separate the salt from the mother liquor. The potassium bromide and potassium iodide were purified in exactly the same manner as sodium bromide. PREPARATIOK O F SOLUTIONS AND THEIR DENSITY DETERMINATIONS
One stock solution of lithium chloride and two of lithium bromide were prepared by dissolving the salts in an almost minimum amount of water.
WEIQHT OF
W E I G R T OF SAMPLE
SOLUTION
1
LiCI.. . . . . . . . . . . . . . . . . . . . . . . . . . .
LiBr (solution I) . . . . . . . . . . . . . . . . . .
.[I
LiBr (sol6tion 11).. . . . . . . . . . . . . . .
AgX
W E I G H T P E R CENT O F LiX IN SOLUTION
grams
grams
3.10328 3 16457
4,41553 4.50251
42.086 42.084 Av. = 42.085
3.36461 3.76652
4,39467 4 91968
60.410 60.409 Av. = 60.410
4.66161 4.64256
54.764 54.759 Av. = 54.761
3.93692 3.92118
I
The halogen content of these solutions was found by precipitation with silver nitrate and weighing as the silver halide. The results of these analyses, which were carried out with great care, are given in table 1. The other solutions of these salts were prepared by dilution of weighed portions of these stock solutions. The solutions of the sodium and potassium salts were prepared by dissolving a weighed amount of the salt in water and then determining the weight of the resulting solution. This procedure was adopted because, according to Baxter and Wallace (3), the sodium and potassium salts can be thoroughly dried without danger of decomposition. The two chlorides were fused in platinum vessels previous to weighing. The bromides and iodides, however, were heated in platinum only to 250°C. After two hours of heating these salts were powdered by gentle grinding in an agate mortar
COlCIPRESSIBILITT COEFFICIEKTS O F ALKALI HALlDES
933
and the heating continued for another two-hour period. After a second grinding the salts were subjected to a final heating of a t least three hours prior to weighing. It may be noted here that a few of the dilute solutions of these salts were prepared by dilution of weighed samples of the more concentrated solutions. These cases will be indicated in the table of results (table 3). Immediately after their preparation all the solutions were transferred to Pyrex glass-stoppered bottles. To prevent loss by evaporation during storage, each bottle was covered with a bottle cap held by a rubber collar which fitted tightly around the neck of the bottle. The density flasks were ordinary 100-cc. graduated flasks, the necks of which had been constricted to about 3 mm. a t the point of graduation to secure greater accuracy in setting the meniscus. The tolumes of these flasks a t 35°C. were found by dividing the vacuum weight of their water content by the density of water a t that temperature (0.99406). To determine the density of a salt solution, the weight in vacuo of that amount of solution which filled a flask to the graduation was found and divided by the volume of the flask. The following observations will complete the foregoing description of the preparation of the solutions and their density determinations. All weighings were made with calibrated weights on the same balance. Whenever flasks were involved, suitable counterpoises were employed. Precautions were always taken to insure that the surfaces of the flask and counterpoise mere subjected to the same treatment. All meighings were reduced io vacuum standard by the usual formula. The thermostat employed in the density measurements was the same one used in the compressibility experiments and will be described in more detail later. To bring the density flask and its contents to the desired temperature the flaskwas immersed in the thermostat and allowed to remain there for a t least two hours before the final adjustment of the meniscus. This adjustment was accomplished in two ways. If the composition of the solution had already been established, the adjustment was made by means of fine capillaries. I n some instances, however, the weight of the solution in the density flask served also as the final weight in the determination of the composition of the solution by dilution. In these cases the meniscus was adjusted either by addition of water or by evaporation by passage of a current of dry filtered air over the surface of the meniscus. Since the weight of a solution in the density flask was always greater than 100 g. and was determined to the nearest milligram, the precision of the density measurements is a t least one part in 50,000. Check measurements were not made as a rule, but the occasional ones, using different flasks, agreed well within the above estimate. The flasks were again calibrated after being in use for a year and showed no appreciable change. The precision of the weight per cent composition of the solutions varies some-
934
A. F. SCOTT, V. M. OBENHAUS AND R. W. WILSON
what with the method of preparation and diminished with decreasing concentration. The uncertainty in this quantity is probably less than one part in 10,000. T H E DETERMINATION OF T H E COMPRESSIBILITY COEFFICIENTS
These measurements were made by the method in use in the Harvard Laboratory. Since this method has been described fully in other papers (7), it will be necessary to give here only the more important details bearing on the present experiments. The compression pump, which was manufactured by A. L. Henderer's Sons, was connected with the compression cylinder and the pressure gage. This gage was of the absolute type, in which pressures are measured by an oscillating piston weighted with carefully standardized weights. A piston with an accurately known area (0.31619 cmez)was loaned to us for this work. The piezometer was the one used by Dr. B. J. Mair and Miss E. H. Lanman in their experiments (sa). It was constructed of glass and was provided with a solid glass stopcock which was removed in order to fill the vessel. The piezometer had a volume of about 34 cc. and in an experiment it contained, as a rule, approximately 31 cc. of solution and 3 cc. of mercury. The compression cylinder was mounted in a large thermostat which has already been described (9). It was possible to regulate the temperature of the bath so that the maximum variation was less than 0.005"C. The temperature of the bath was 35.OO0C., established by means of a mercury thermometer certified by the Bureau of Standards. The compressibility coefficients were calculated by use of the usual equation
where
W is the weight of solution contained in the piezometer; w1 is the weight of mercury needed to compensate for the volume change produced by changing the pressure from 100 to 300 megabars when the piezometer is filled with mercury; w 2 is the weight of mercury needed to compensate for the volume change produced by changing the pressure from 100 to 300 megabars when the solution is filled with W grams of solution and mercury; D is the density of the solution a t 35°C. and a t atmospheric pressure; d is the density of mercury at 35°C. under 300 metric atmospheres pressure (13.525); @' is the compressibility coefficient of the reference substance, mercury, and and was assumed to be 4.00 X
93 5
COMPRESSIBILITY COEFFICIENTS O F ALKALI HALIDES
0 is the average compressibility coefficient over the pressure range 100 to 300 metric atmospheres of the substance in the piezometer. In determining the values of w1and w2 the following procedure suggested by Mair and Lanman was adopted. The amount of mercury in the piezometer was adjusted a t the outset so that the electrical contact through the platinum wire-mercury interface was broken when the applied pressure was very close to 300 megabars. After the exact pressure was measured, mercury was removed from the piezometer until the pressure necessary t o break the contact was approximately 100 megabars. Following exact determination of this second pressure, the mercury, which had been withTABLE 2 D a t a and calculations for the compressibility coejicient of s o d i u m iodide solution N o . 4
1
w 2 (grams). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . w 1 (grams) .................................. p X lo6.. .............................................. Accepted value of p
x
106 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
lXPERIMENT I
3.0358 0.1500 38.54
1
1
E X P E R I M E N T I1
134 324 100 015 34 309 34.341 1,10998 2 8964 292 09 101.71 292 11 292 10 3.0426 0.1500 38.56
38.55
weight on gage x acceleration of gravity where area of piston x lo6 weight on gage includes weight on platform plus weight of platform; acceleration of gravity is 979.272, and area of piston was 0.31619 cm2.
* Pressure
in megabars equals
drawn and in the meantime weighed, was now returned to the piezometer and the initial pressure checked. The values of wlor w2 corresponding to exactly 200 metric atmospheres were calculated by direct proportion from the weight of mercury removed and the pressure difference which covered very closely the range 100 to 300 megabars. Certain points with regard to the routine of the experiments should perhaps be noted here. Following the recommendation of Cohen and Schut (4),the capillary tube below the platinum contact was allowed to become moistened with water before the initial pressure application in each experiment. The practice was adopted of completing the pressure de-
936
A . F. SCOTT, V. M. OBENHAUS AND R .
SOLUTION
I
W. WILSON
TABLE 3 Table of results W E I G H T P E R CENT OF SALT
AVERAGE DEVIATIOPi
DENBITY
Lithium chloride solutions 42.031 34.898 26.021 23.010 18,969 16.279 9.6550 4.1221
1.26364 1,20970 1.14836 1.12896 1.10418 1.08776 1.04943 1.01772
20.12 22.61 26.14 27,41 29.06 30.52 34.45 38.35
0.05 0.01 0.03 0.02 0.03 0 .oo 0.01 0.00
22.37 24.13 25.69 27.09 27.70 29.46 30.94 32.92 35.50 37. i7 38.57
0.02 0 01 0.05 0.03 0.02 0 02 0.03 0 .00 0.02 0.01
Lithium bromide solutions 60,410 54.761 49,949 45.438 43.493 36,696 31.897 25,388 17 229 12,959 8,6052
1.72090 1.60856 1,52364 1.45294 1.42456 1,33423 1,27784 1.20827 1.13097 1,09397 1,05845
Sodium chloride solutions
1 (i) 2 (9 3 6) 4 (i)
24,8200 16,9863 9.1598 6,1298 2.8661
1.17942 1.11690 1.05882 1.03671 1.01388
27.10 30.99 35.56 37 I54 39,71
0.01 0.03 0.01 0.02 0.03
49 32 20 9
1 54400 1 30294 1 17443 1 07392
23 29 34 38
0 0 0 0
1.80644 1,42173 1.19758 1,10998
24.22 31.39 36.36 38.55
295 380 7701 8333
60,802 40.413 22,712 13.896
42 74 03 10
03 03 00 02
0 01
0 01 0 03 0.01
937
COMPRESSlBILITT COEFFl CIENTS O F ALKALI HALIDES
SOLUTION
I
TABLE 3-Concluded WEIGHT P E R CENT
AVERAGE
OF SALT
Potassium chloride solutions 1 17419 1 10725 1 04118
29 64 33 35 37.83
1
1 1
0 02 0 04 0 02
Potassium bromide solutions
1 36525 1 27822 1 18086 1 07209
29 51 31.72 34 68 38.54
0.01 0.15 0.04 0.04
27.74 34.22 38.44 40,04 40.87
0.16 0.12 0.17 0.02 0.04
Potassium iodide solutions 58.262 34.901 16.415 8,9594 4.4120
1,68149 1.32072 1.12576 1 ,06226 1,02648
termination fifteen minutes after the initial application of pressure. Further, in a pressure determination sufficient pressure was applied first to just break the electrical contact between the mercury in the piezometer and the platinum wire in the capillary. The pressure was then gradually reduced until contact was just made and the pressure corresponding to this point was the one accepted. The value of the piezometer constant W Iwas found to be 0.1500 g. from eight different experiments made over a period of one year. Mair and Lanman found tol to be 0.1473 g. for the same piezometer at 25°C. The 2 per cent difference is a reasonable figure for the change in the compressibility of the glass and mercury for the 10°C. change in temperature. In the table of results (table 3) will be given only the final calculated values of the compressibility coefficients for the solutions, To illustrate both the experimental procedure of the compressibility determination and the method of computation of the data, the calculations for a typical experiment are given in table 2. RESULTS
All of the results of the experiments are summarized in table 3. Along with each solution number are given symbols to indicate the method of preparation of that solution. In this code “i” means that the solution was made independently of all other solutions while, for example, “d2” indi-
938
A . F. SCOTT, V. M. OBENHAUS AND R. W. WILSON
cates that that solution was prepared by dilution of solution no. 2 of that series. With the exception of lithium bromide solution No. 5, each value of /3 is the mean of at least two independent measurements. In
FIG. 1. THE COMPRESSIBILITY COEFFICIENTS OF SOLUTIONS PLOTTED AGAINST THECONCENTRATION
order to have a rough measure of the uncertainty of the mean value of p there is given in the last column of table 3 the average deviation of the individual values of p from the given mean, expressed in the units of the compressibility coefficient. Attention is called to the fact that a number
COMPRESSIBILITY COEFFICIENTS O F ALKALI HALIDES
939
of improvements in both technique and apparatus were made in the course of the compressibility measurements of solutions of potassium bromide and potassium iodide and of lithium chloride solution No. 1,which were the first to be investigated. For this reason the p values of these solutions are not so precise as those of the other solutions, as is evident from the data in the last column. Figure 1is presented to show the nature of the variation of the compressibility coefficient with the composition of the solutions. It can be seen that the anion determines both the relative position and the shape of the curves drawn through the plotted points. Thus, the lines representing the chlorides are all convex upwards; the lines representing the iodides are all convex downwards; and the lines representing the intermediate bromides have been drawn straight. The solutions of the two lithium salts, however, exhibit some noteworthy anomalies. Since these anomalies are apparently associated with the molecular composition of the solutions, the number ( N ) of moles of water per mole of salt for the various solutions is given in the graph alongside the plotted points. In the case of the lithium chloride solutions there is a definite break in the variation of p with concentration a t the solution with the composition N = 10. This critical point is the same as that found from the examination of such other properties of lithium chloride solutions as the density and magnetic susceptibility. As a matter of fact the investigation (8)of this last-named property indicated that the magnetic susceptibility of lithium chloride has two definite values, one in solutions less concentrated than N = 10, and the other in solutions more concentrated than N = 6. The transition between these two distinct values takes place in the concentration region 10 > N > 6. Whether this second change in the solution state mould affect the compressibility of the solutions cannot be ascertained from the present data. The possibility of a second break in the p-concentration curve a t a point where N = 6 is not entirely out of the question, because the plotted point N = 3.25 does not fall on the curve as drawn. Lithium bromide solutions appear to be more irregular even than those of lithium chloride. In the first place the compressibility coefficients of these solutions do not follow the usual sequence Li-Na-K. In the second place there are two abrupt changes in the variation of with concentration. One of these breaks occurs in rather dilute solutions, somewhere in the concentration region 30 > N > 24. The change in the rate of variation of p with percentage composition a t this point is not very different from that found in the lithium chloride solutions. The second break in concentrated solutions 8 > iV > 6 is of an entirely different type. The rate of change of p with concentration does not appear to be affected by the change in the solution state, but the p values in the more concentrated solutions are 0.4 X units higher than they would be if there were no break in
940
A. F. SCOTT, V. M. OBENHAUS AND R. W. WILSON
the rariation of p with concentration. Furthermore, this change in the solution state must take place in a rather short concentration range which we hope to explore more carefully. In conclusion it should be noted that Huttig and Keller ( 5 ) found that the densities, refractive indices, and coefficients of extinction of the lithium halide solutions do show discontinuities with changes in concentration when N = 6 , near 30, and 75. Further discussion of these results from a different point of view will be the subject of another paper. In conclusion we wish to express our gratitude to Professor Arthur B. Lamb of Harvard University for the loan of the piston of the pressure gage used in this investigation. REFERENCES (1) ADAMS:J. Am. Chem. Soc. 63, 3769 (1931); 64, 2229 (1932). A N D GIBSOX:J. Am. Chem. SOC.64, 4520 (1932). (2) ADAMS (3) BAXTER AND WALLACE: J. Am. Chem. SOC.38,75 (1916). (4) COHEN AND SCIIUT: Piezochemie, p. 28. Akademische Verlagsgesellschaft m.b.H, Leipzig (1919). ( 5 ) HUTTIGAND KELLER:Z.Elektrochem. 31, 390 (1925). (5a) MAIR AND LANMAN: J. Am. Chem. SOC.66, 390 (1934). (6) PERMAN AND URRY:Proc. Roy. SOC.London 126, 44 (1930). (7) RICHARDS: See J. Franklin Inst. 198, 25 (1924) for a bibliography of hie work. (8) SCOTTAND BLAIR:J. Phys. Chem. 37, 481 (1933). (9) SCOTTA N D DuRH.~M:J. Phys. Chem. 34, 531 (1931).