14%
hI. A. RREDIG, H. R. BRONSTEIN AND WM,T. SMITH, JR. [CONTRIBUTION FROM
THE
Vol. 77
OAKRIDGENATIONAL LABORATORY, CHEMISTRY DIVISIOS]
Miscibility of Liquid Metals with Salts. 11. The Potassium-Potassium Fluoride and Cesium-Cesium Halide Systems1 BY M. A. BREDIG,H. R. BRONSTEIN AND Ww. T. SMITH, JR.* RECEXVED OCTOBER 14, 1954 Miscibility in all proportions of cesium metal with cesium halides a t and above the melting points of the pure salts, and of potassium metal with potassium fluoride fifty degrees above the melting point of the salt was found. Vapor pressure measurements on some of these mixtures confirmed their true solution nature. The temperatureconcentration range of coexistence of two liquid phases decreases in going from sodium to potassium systems and disappears altogether for the cesium systems. The solubility of the solid halides in the corresponding liquid alkali metals, a t a given temperature, increases greatly with increase of atomic number of the metal. This may be explained by the decrease of cohesive forces in both the metal and its salts as the atomic number of the metal is increased. Partial heats of solution of the solid salts, a t saturation, in the liquid metals, also correlate well with the cohesive energies, or heats of vaporization and sublimation, AH, and A H a , of the liquid metals and the solid salts, respectively: CsF, with its low heats of solution and sublimation, 10 and 37 kcal., represents one extreme case. NaF, with the values 26 and 61 kcal., respectively, represents another extreme case, the high values resulting from the small size of both Na’ and F-. The low heat values for CsF may be connected with the polarization of the large Cs+ by the small F-. The heat of solution of sodium halides in sodium metal increases from the chloride (20 kcal.) to the iodide (28 kcal.), even though the heat of sublimation of the salt decreases from 48 t o 43 kcal. More energy should be required to “substitute” anions of increasing size for electrons in a metal of comparatively high cohesive energy such as sodium (small atoms or ions, AH, = 23 kcal.). The “substitution” of different anions for electrons in cesium, a metal of much lower cohesive energy ( A H , = 16.3 kcal.) should require less energy, and the energy requirement should not be very much different for different anions. The cohesive energy of the solid cesium salt which decreases slightly from chloride to iodide ( A H . = 39 and 38 kcal., respectively), should then be a more important factor in determining the heat of solution. In accordance with this notion, the heat of solution was found to decrease slightly from 22 f 2 for CsCl to 18 f 1 kcal. for CsI.
Introduction Experimental Methods used for the determination of equilibrium comNew techniques were recently developed for the determination of phase equilibria in systems com- positions of the liquid phases and for obtaining cooling were the same as previously reported.’ In fact, the posed of liquid alkali metals and their halides.a curves ball-check valve capsule was developed in studying the It was found necessary to isolate a sample of the KF-K system. For the cesium systems the decantation equilibrated liquid phases a t the temperature of method was used exclusively since no two-liquid phase reequilibration, because these phases changed rap- gion was found. Because of the high cost of materials, especially the cesium metal, the diameter of the capsule was idly in composition even on fast cooling. Approxi- reduced from 0.75 in., employed with the sodium and pomate minimum temperatures of miscibility in all tassium systems, to 0.25 in. Since it was necessary in these proportions (consolute temperatures) of liquid determinations to know whether if any solid salt remained sodium metal with its fused halides were thus ob- under equilibration conditions as well as whether the liquid of the system was homogeneous, a further slight tained by extrapolation from measurements a t portion modification of the decantation type of capsule was made. The high solubilities slightly lower temperatures. The solid salt was placed above a stainless steel wire gauze of the metal in the fused salts were seemingly in in the upper part of the tube. After equilibration, the was tilted so as to drain the liquid from any remaincontradiction to a theory4 based on measurements furnace ing salt and then tilted to another position in order to sepawith multivalent metal-halide systems. This ap- rate the liquid into two portions by decantation. In cases parent contradiction was explained by considering where no solid phase was found undissolved above the wire the high fugacity of the alkali metal a t tempera- gauze, both of these portions were analyzed and identity of composition was taken to indicate the presence of only one tures where the multivalent metals have low liquid phase. fugacities. The potassium metal contained 0.27% sodium, by specThree principal topics are discussed in the present trographic analysis, and 0.3y0 oxygen, determinedKusing report : (1) the direct experimental demonstration the Wurtz reaction. Anhydrous reagent grade potassium fluoride and C.P. cesium chloride were treated in the same of miscibility in all proportions of an alkali metal manner as described previously* for sodium salts. Cesium with its salts in the liquid state, ( 2 ) the effect upon iodide was purchased in large, optically and chemically pure miscibility of variation of the cation in systems con- crystal fragments. Cesium fluoride was prepared by treating cesium chloride with aqueous silver fluoride obtained by taining the same anion, as in the series NaF-Na, silver carbonate in hydrofluoric acid. DehydraKF-K and CsF-Cs, and (3) a comparison of the ef- dissolving tion and fusion of cesium fluoride and potassium fluoride fect of variation of the halide ion in systems con- were carried out in platinum containers. taining the same cation, when this cation is a large The cesium metal was pure except for traces (0.01%) of one such as Csf in the series CsF-Cs, CsC1-Cs rubidium and potassium. Its transfer from the glass amcontaining two grams, to the test capsule was accomand CsI-Cs, or a small one such as Na+ as in the poule, plished in the glass apparatus shown in Fig. 1. The amseries of the systems of sodium metal with its hal- poule was opened under hexane and placed in the upper part ides studied previously. of the assembly with an atmosphere of argon above it. (1) Work performed for the U. S. Atomic Energy Commission. Presented in part before t h e Division of Physical and Inorganic Chemistry of the American Chemical Society, Kansas City, Mo., March 23-April 1, 1954. ( 2 ) Consultant from the Department of Chemistry, University of Tennessee. (3) RI. A. Bredig, J . W. Johnson and Wm. T. Smith, Jr., THIS JOURNAL, 77, 307 (1955) (4) D. Cubicciotti and C. Thurmond, rbid., 71, 2149 (1949); D. Cnbicciotti. i b i d . , 7 1 , 4119 (1949); i b i d . , 74, 1198 (1952).
After pumping off the hexane the cesium metal, again under argon, was allowed to melt and drop into the stainless steel capsule containing the desired quantity of salt. The capsule was then connected through a rubber stopper and a glass tube to a vacuum system to remove the argon. After evacuation, the glass tube was sealed off and the upper part of the stainless steel capsule was crimped in a vise, sawed (5) J. C. White, (1954).
as, 210
W,J. Ross and Robert Rowan, Jr., Ann!. Chem.,
March 20, 1955
MISCIBILITY OF LIQUIDMETALS WITH SALTS
off above the crimp and Heliarc-welded to complete the loading operation.
4!00-
1455
KF-K ONE L l O U l O P H A S E
4000Li
g
,9!0"C
900
c 3
600
20
40 60 80 MOLE PERCENT POTASSIUM M E T A L
Fig. 2.-The
'ic-
CESIUM AMPULE
potassium fluoride-potassium metal system.
ing point of the salt. In the experiments run above 910' and listed in Table I, the composition of the samples turned out to be identical with that of the charge. Thus, the existence of only one liquid phase in this temperature range was directly demonstrated. TABLE I EQUILIBRIUM PHASECOMPOSITIONS IN
Phases present Solid salt and one S O L I D SALT STAINLESS STEEL METAL SCREEN
Fig. 1.-Equilibration tube and loading apparatus for use with cesium metal.
Results and Discussion Tables I and I1 contain the results of solubility determinations in the systems KF-K, CsF-Cs, CsCl-Cs and CsI-Cs. In Fig. 2 the results for the KF-K system are plotted. Figure 3 affords a comparison of the three fluoride systems, NaF-Na, KF-K and CsF-Cs. Figure 4 is a plot of the logarithm of salt solubility in liquid metal DS. the reciprocal of the absolute temperature for all of the nine alkali metal-alkali halide systems investigated thus far.6
I n contrast to the sodium systems, it was possible with these systems to measure experimentally the solubilities up to complete miscibility. This is largely due to the lower temperatures required here. For instance, the consolute temperature in the KFK system, i.e., the temperature a t which the solubility curve of the metal in the fused salt meets the solubility curve of the salt in the liquid metal, was found to be 910' which is only 50' above the melt(6) Five experimental points on t h e KCI-K system have been taken from measurements by J. W. Johnson of this Laboratory, which are t o he reported in another connection later.
100
Mole '7c K F Liq Liq. saltmetalTemp., rich rich OC. Charge phase phase
658 54.7 697 29.3 750 47.9 803 48.7 832 51 0 858 100.0 100.0 849 81.4 (97.5)a
I
7.6 12,3 17,7 30.4
THE
SYSTEM KF-K
Experimental method Ball-check valve capsule Decantation Ball-check valve capsule Decantation Ball-check valve capsule
liquid Solid salt a n d two 1 luids
853 862 862 863 865 865 865 865 Two 865 liquids 865 877 877 881 886 886 894 894 90.5 Consolute (910) 948 One 948 liquid 975 980 1060
56.7 75.8 94.0 71.1 94.8 71.3 93.7 56.3 92.6 57.3 90.3 55.5 50.2 51.6 59.4 67.5 85.6 71.6 85.2 67.8 84 4 67.6 82.0 68.4 67.0 78.3 57.8 64.7 69.5 76.0 74.7 60.0 30.0 60.0
76.7 73.5 59.9 30.0 58.5
51.9
Ball-check valve capsule
32.01 58.2
Values in parentheses obtained by interpolation.
There is considerably greater scattering of experimental points in the KF-K system for the less dense metal-rich liquid phase than for the more dense salt-rich liquid phase (Fig. 2 ) . It is possible that due to a comparatively small difference in density of the phases in this case, and to a possibly considerable interfacial tension, droplets of the heavier salt-rich phase were sometimes floating on top of the metal phase.
M. A. BREDIG,H.R. BRONSTEIN AND Wu. T.SMITH, JR.
1456
VOl. 77
The data illustrate a striking effect on the soluTABLE I1 EQUILIBRIUM PHASE COMPOSITIONS IX THE CESIUM HALIDE- bility of the solid salt in the liquid metal produced by the variation of the cation in both the metal and CESIUMSYSTEMS the halide: increasing the atomic number of the Mole % salt Decantate from filtrate 3 9
metallic element from sodium to potassium and cesium leads to very large increases in solubility of Phases the solid salt in the liquid metal. For instance, at Exptl. method System present 550” the solubilities of the halides of sodium, poDecant.: no filtration csI;-cs tassium and cesium in the metals range from 0.1 to 1 1 4 S o decant; no filtr. 0.2, 1 to 2 , and 15 to 25 mole %. respectively 22.1 Solid salt 36.3 (Tables I and I1 and Fig. 4). Largely as a conseand 50 1 Decant.; no filtration quence of this trend, the concentration-temperaone 63 3 liauid ture area of the diagram in which two liquid phases 74.6 are in equilibrium shrinks rapidly with increasing Cooling curve atomic number of the metal. Indeed, the experi5.9 CsCI-cs mental results show that two liquid phases no longer Solid salt 8.6 coexist in the cesium systems as they still do in the and 13.5 KF-K system. Only solid cesium salts and one 32.4 one liquid solution were ever found in equilibrium. liquid 56.1 78 0 Decantation from filThus, increasing additions of cesium metal to the 67 2 trate salt lower the melting point continuously. An uni2.8 One broken, “S” shaped liquidus curve (Fig. 3 ) conliquid 86 1 nects the melting point of the salt with the eu83.7 76.2 tectic point which can be shown by extrapolation of Solid and the log s us. 1/T plot to practically coincide with the Cooling curve salt melting point of the metal. For CsF, e.g., the soluCsI-cs 1 7 bility a t 30” is less than 0.001 mole yo. Solid salt 4.3 and 1 9 . 0 Decantation from filThe solid phases, though essentially pure salt, 36 6 trate one contained a small amount of cesium metal in solid liquid solution. This was indicated by the deep purple72.5 blue-black color of the undissolved, quenched salt Cooling curve crystals which had been equilibrated with the metal 42.0 Decantation from filOne solutions. The color, on dissolving these crystals, liquid 73.4 trate was found to have penetrated them. Tn the case of 58 0 the KF-K system the color of the solid salt phase is Solid and Cooling curve salt light yellow. (’ Since 1 1 0 decaritate was obtained, this value was acPartial heats of solution, a t saturation, of the cepted as the solubility value. Incomplete removal of salts in the liquid metals were calculated from the hydrogen fluoride may possibly have accounted for the somewhat lower values 680 and 684” obtained for the melt- slopes of the logarithm of the solubility (mole %) iiig point of this salt by F. M. Jaeger, 2. anorg. Chem., 101, ZS. 1/T plots. This relation may be assumed to be 193 (1917), and H. Wartenberg and H. Schulz, Z . Elcktroapproximately valid in the region where the compochem., 27, 568 (1921). The value 703’ reported here was sition of the solid is almost that of the pure salt obtained on samples from which hydrogen fluoride had been and where the plotted data give a straight line. removed by evacuation and flushing with argon at 600” until no acidity was detectable in the gas stream. This In this manner the heat of solution of cesium fluovalue is in good agreement with 705’ reported by 0. Schmitz- ride was found to be 10 f 0.5 kcal. compared with Ilumont and co-workers, 2. anorg. Chem., 252, 329 (1944); 22 f 2 and 18 i 1 for the chloride and iodide, reDetermined accurately by analysis. 271, 347 (1953). spectively. I t seems significant that cesium fluo< 200, 1 , I I , ride stands out with its low heat of solution as it --IIIODC h o F - b o does with its low heat of sublimation, AHs = 36.8 rroot kcal., as against 39.3 and 38.4, re~pectively,~ and Th0 L ~ Q L I ~ D S that there is little difference between the chloride roo3 1 and iodide. In the sodium systems the fluoride has also been found3 to stand apart from the others, but by a high rather than a low value of the heat of solutior (30 kcal. for the fluoride compared with 20 + 3 and 28 f 2 for the chloride and iodide). Xn explanation may be based on the exceptionally high cohesive energy of the fluoride (heat of sublimation, AHs = 61 kcal.) due to the relatively small sizes of both Y a f and F- and on the increasing amount of energy required to “substitute” a larger aniono(RI- = 2.20 A. as compared with Rcl- = ,: I- L~-L-LLL-LL--LI\J ‘0 20 30 40 50 60 70 80 90 4iC 1.80 A . ) for an electron in the liquid metal (or a Remainder Charge of filemp., (ap‘(2. prox.) trate 34Y 42 452 46 527 61 599 86 640 57 662 84 671 86 94‘ 692 703* looc 486 GO 513 35 549 48 53 601 611 76 621 86 620 6!J 7 0 . 0 622 70 7 0 . 1 626 91 86 5 630 85 75 i ( i . 7 632 pid 640 100 399 37 450 34 549 49 571 61 580 67 5 6 . 1 a 90 592 93 609 580 43 43 4 580 43 41.9 597 72 73.0 598 60 j 8 2 iquid 626 looc
I
I
~
//-\
~
~
,rO
~
MLT4L CONCiN7RAT10N
Fig 3.-CoinpLtrison
(mole %)
of the three systems NaF-Na, KF-K, CsF-Cs.
(7) Leo Brewer. in “ T h e Chemistry and Metallurgy of Miscellaneous 3Iaterials. Thermodynamics,” Vo:. IV-IOB, of National Nuclear Energy Series. SIcGraw-Hill Book C o . . N e w York, N. Y . , 1950.
March 20, 1955
1457
MISCIBILITYOF LIQUIDMETALSWITH SALTS
I
I
’ x T (OK1
Fig. 4.-The
I
106
solubility of solid alkali halides in alkali metals.
larger molecule MX for an M atom). I n comparison chlorides to the iodide (AHs = 39.3 and 38.4 kcal. with the differences in “substitution energy” the respectively) is reflected in the-rather slight-dedifferences in cohesive energy of the salts (AH,: 48 crease of the heat of solution from 220* 2 to 18 (NaCl), 44 (NaBr), 43 kcal. (NaI)) appear to be 1 kcal. I n going from F- ( R = 1.33 A.) to the C1minor except in the case of sodium fluoride (61 ( R = 1.80 A.) there may be a significant contribukcal.). Applying the same principles to the cesium tion to the increase in heat of solution resulting systems we find that the exceptionally low cohesive from an increase in “substitution energy” even in energy of the fluoride (AH, = 36.8 kcal.) leads to a the case of cesium. This contribution adds to the low heat of solution. Also, a smaller difference in effect of increased cohesive energy of the chloride the energy required for substituting anions of dif- and leads to an even greater difference between the ferent sizes for electrons may be expected with ce- chloride and fluoride (22 and 10 kcal.). I n the sium, a metal of large atoms or ions (Res+ = 1.69 case of the sodium systems i t counteracts the effect A.) and of low cohesive energy (AH, = 16.3 kcal.), of the lower cohesive energy of the chloride and $ban with the more cohesive sodium ( R N ~=+0.98 therefore diminishes the difference between the A,, AHv = 23.1 kcal.). Thus in the cesium systems heats of solution of the fluoride and chloride (26 the small difference in the “substitution energy” and 20 kcal.). and the relative constancy or small decrease of the (8) No attempt was made t o determine the small solubility of t h e cohesive energy of the solid salt in going from the low-temperature form of CsCI, which is isomorphous with CsI.
*
M. A. B ~ D I GH., I
