Liquid-Liquid Phase Separation in AI kali Metal-Ammonia Solutions.
I.
Lithium, Potassium, Rubidium, with New Data on Sodium
by Paul D. Schettler, Jr., and Andrew Patterson, Jr. C'ontiihiition .Yo. 1724 fTom the Sferliw Chemistry Laboratory, Yale University, h'eu Haven, Connecticut (Kl&reri
July so, 1963)
l'hastl diagrams are presented for the liquid-liquid phase srparation of lithium and potassolutions in liquid amriionia. Solutions of lithium separate into two phases a t teinperaturcs helow - 63.5" in t,hc range of conccntration between approximately 0.02 and 0.09 g.-atotn of lithiuin/mole of ammonia. separation of potassium solutions is observed below -70" in the range of concentration including approximately 0.02 and 0.07 g.-atom of potassiuri1,'rnole of ammonia. Solutions of rubidium in liquid aninionia have not been observed to separate into two phases a t tenipcratures above the freezing point of arnmonia and a t coriccntrations including those a t which lithium, sodium, and potassium do so. Visual observations of the phenomcnon of phase separation are reported which explain previous anomalous results in solutions of this type. Data on phase separation in sodiumammonia solutions a t previously unreported temperatures and concentrations are included as wcll. siutii
I t is cotrinionly stated that one of the characteristics of alkali 1ii(>ta1liquid atnnionia solutions is to exhibit separation into two inirniscible liquid phases. l'his stateinc>nt appears to bc based principally on the behavior of sodiuni- anmonia solutions, for which phase diagrams have been reported and on which extensive studies have born made. - 3 On more careful examination of the litcraturc, however, one finds the evidence for phase separation among solutions of the other alkali metals uncertain and contradictory. Thus, Hodgins and Flood rcxport visual observations and resistivity nicxasurcmicnts as evidence for phase separation i n lithium solution^,^ pointing out that Kraus and ,Johnson5 had already given convincing evidence of the separation from vapor pressure-concentration rneasurenients, and amending the statement of Birch and AIacDonald,G who stated that lithium-ammonia solutions do not show phase scparation, to read "liquidliquid phase separation above -78" is clxhibited by all alkali metal solutions except cesium Hodgins' estcmive work on cesium7 and brief tests on l i t h i ~ n i . ~Howcver, it is not possible to construct any niorc than a guess a t a phase diagram from the data available on lithium solution^.^^^^* The data on potas-
sium are even more fragmentary, there being no report of visual observation of phase separation, although vapor pressure and warming curve data6 suggest that it may occur. Observations of phase separation have been made in potassium solutions to which potassium iodide has been added, in a temperature range of -31 to -35°.9 I n spite of what is iniplicd by Hodgins and Flood's statement quoted above,4 there is no report of any deterriiination on rubidium. (1) C. A . Kraus and W. W. Lucwsse, J . A m . Chem S o c . , 44, 1949
(1922). (2) 0. Ruff and J. Zedner, Rer., 41, 1948 (1908). (3) R. A. Ogg. Report t o the Office of Naval Research, Physics Branch, 1947, included results for lithium and potassium. T h e work has not twen published. Also, t h e dissertation of I). E. Loeffer, Stanford I'niversity. 1949, includes sirnilnr material. M. J. Sienko has disclosed these results a t Lille. France, Colloque Weyl. June, 1963. (4) J. W. Hodgins and 5:. A. Flood, Can. J. Res., 27B,874 (1949). (5) C. A. Kraus and W. C. Johnson, J . A m . Chem. Soc., 47, 731 (1925). ( 6 ) A. J. Birch and D. K. C. M a c n o n d d , Trans. Faraday Soc., 44, 735 (1948). (7) J. W. Hodgins, Can. J. Res.. 27B,861 (1949). (8) P. It. .Marshall and 13. H u n t , J . Phys. Chem., 6 0 , 732 (1956) (9) D. D. Cuhicciott.i, ibid.. 53, 1302 (1949).
Volume 68, Sumber 10 October, 1,Wh
The presmt paper reports ni(~asur(wents on solutions of lithiuni, potassium, and rubidium in liquid an~nioiiia, and some additional data on sodiuiri.I0 Iikrenie care has bcen taken to prcparc solutions free of riiaterials other than the alkali nietal and ainmonia, both to assure thc stability and long life of the solutions and to exclude additional components which have a profound effect on the temperature and concentration ranges in which phase separation is observed. lo
Experimental Two rather different means of determining the phase diagram have been uscld. The niethod triost often used by other invcstigators to ascertain whcthrr phase separation has occurred is to iiieasiire the conductance of a solution as the teniperature is changed and to await a jump in the measured quantity.' With the intention only of making some preliminary measurements on potassium, we undertook to use a straightforward visual obscrvatiori of phase separation. Froin these visual observations, we concluded that (for potassium at least) uncertainties in conductance nieasurenients may well have been due to the peculiar surface tension behavior of these solutions and especially to the manner of phase separation. The phase separation studies on potassium have, therefore, been made cntirely visually, as will be described. I n the ease of solutions of lithium, a inore refined measurement technique which separates the two phases and permits analysis of both phases has been ernployed.'O In all casos, the distillation of ammonia and alkali nietals has been coriducted in a high vacuum system from which undesired contaminants have bem rigorously excluded and in which, apart from the component being distilled, a vacuuni in excess of torr is maintained. All glassware which came in contact with the nietal, liquid aniinonia, or the finished solution was cleaned prior to use by successive treatment in alcoholic potassium hydroxide solution, fuming nitric acid-concentrated sulfuric acid niixturti, and distilled water followed by drying in an oven. For the nicasureinents on potassium the apparatus in Fig. 1 was cniploycd. The tube A was evacuated to torr, heated at 300" overnight, arid previously prepared potassium metal (AIallinckrodt 6692) was triple distilled from a measured length of capillary through a succession of sausage-shaped tubes into tube A. Triple-distilled ammonia, dried over sodium, was distilled into tube A, following which the tube was scaled and separated from the vacuum system. The separated tube was keptj at - 196" until it was needed, and then inswted in a short length of rubber tubing which was attached to a longer length of Pyrex tubing The Journal of Physical Chemistru
1 VACl
L II
I
-4c/, 5P INCH
of the same diaiiictcr as tubc -1 for a handle. The sealed tube and handle werc iniincrsed i n a tht~niostat made from a dewar flask containing 4 1. of cold tlthanol stirred by a nieclianical stirrer. The temperatures were measured by short-range alcohol thcrnionieters calibrated to better than 0.1O against a platinum resistance thermonieter. The tubc was shaken periodically, and times up to 10 niin. allowed for attainirient of equilibrium before the tube was removed moniclntarily from thc bath for inspection to ser if one or two phases existed. If two phases were found, a higher tempc>rature was tried; if only one, a lower tcniperature. When the approximate tcmpcrature of phase separation mas found, it was fixed mor(' precisely by picking temperaturcs alternately slightly above it and then below it. Thc nnaximuni temperature of phase separation was judged to have been found when it was boxed in as described by at least two points 0.1" apart and with several points no more than a few tenths of a degree above and below this temperature such that all those above showed only one phase, and all those below, two phases. To eliniiriate errors caused by the cxperirnenter subjectively anticipating the temperature reading, the thermometer was not read until after the one- or two-phase observation was made. Of about 350 individual readings, only three mere inconsistent. This lends assurance to the visual observation of phase separation. Temperature control was achieved by adding to the alcohol bath small amounts of powdered Dry Ice through a wire strainer a t about 30-sec. intervals, varying the amount of Dry Ice added and the intervals between addition as dictated by the rate of deviation of the thermometer from the desired temperature. (10) P. D. Schettler. J r . , and A. Patterson, *Jr., .J. Phys. Chem., 68, 2870 (1964).
2867
LIQUID-LIQUID PHASESEPARATION IN ALKALI METAL--AMMONIA SOLUTIONS
Temperatures could be kept constant to within k0.05" for any necesE,ary length of time. After measurements on one tube were completed, it was weighed and frlozen in liquid nitrogen. Reaction of the potassium witlrl the ammonia could then conveniently be checked by using a Oudin coil leak tester; in nearly every tube prepared, and for all those reported here, the characteristic pale, dirty green electrical discharge in ammonia vapor was still clearly visible, indicating that very little, if any, decomposition had taken place, with attendant hydrogen evolution. The tubes were then opened, the ammonia allowed to evaporate into a stream of nitrogen gas, the contents oxidized in air, and the residue titrated to the carbonate end point with standard acid. The ammonia was determined by difference in weight between the full and empty tube. The reproducibility of temperature measurement was such that the most probable cause of scatter of the points (see Results) is difficulty with the analysis. Further, nearly half the tubes exploded at some point during the air oxidation, so the method was discarded for a more satisfactory one described below. For the measurements on lithium, a different piece of glassware was used, except for measurements at the critical point, where the method used above was employed, though with a more elaborate thermostat. The glass apparatus, Fig. 1, ref. 10, has also been used in phase separation studies of sodium-liquid ammonia with and without added sodium iodide.'O It allowedl the separation and sampling of the two phases in am inert atmosphere anld at a precisely controlled and measured temperature. Lithium (K and K Labora-. tories, 99.9% pure) wire was measured off and the con-. taminating coating removed with a paint scraper under heptane which had previously been dried with sodium and swept free of nitrogen by passing argon through it. The bright, freshly scraped, and still wet piece of lithium was then quickly transferred to the preparation bulb1 which had already been filled with argon and the remaining heptane removed in a stream of argon. The apparatus, already attached to the vacuum system, was immediately sealed and evacuated for an extended period. Ammonia, previously purified, was then added in the amount desired while the preparation bulb was cooled in Dry Icealcohol, followed by the addition of purified helium a t about 650 torr pressure. The apparatus was then sealed from the vacuum system and removed to a low temperature thermostat controlled to 0.01" or better. The detailed operation of the apparatus is described elsewhere.l0 Briefly, by inverting the apparatus and lowering a float, separate samples of the two phases were
obtained. Since the two phases could be separated a t a measured temperature, duplicate determinations are possible for the points lying below the consolute temperature. The separated tubes were frozen in liquid nitrogen, opened, the ammonia evaporated by a stream of carbon dioxide-free helium, and the lithium decomposed in a stream of water vapor. Conductivity water was then added, a protective atmosphere of butane placed in the titration flask, and a straightforward acid-base titration conducted. The end point was determined with a differential electrode or acid-base indicator. The entire procedure was much more satisfactory than that used for potassium. The amount of ammonia was determined, as before, from the weight of the sealed tube before and after opening. For the preparation of rubidium metal, rubidium chloride (K and K Laboratories, 99.8%) was reduced by calcium metal filings in the side tube shown in Fig. 1. At temperatures near the softening point of Pyrex glass, reaction of an intimate mixture of powdered calcium and rubidium chloride took place in less than an hour. Flame photometric analyses of the residue indicated about 94% yield of the rubidium. The mixture had to be degassed with great caution in advance of the distillation, this being facilitated by placing the entire apparatus (similar to Fig. 1) inside a controlled oven. I n one determination a Faraday tube arrangement was used in place of tube A so that a range of concentrations and temperatures could quickly be surveyed with one preparation of solution.
Results The determinations are shown in Table I and are plotted, together with data for sodium,' in Fig. 2. To the data on sodium we have added points which result from other studies on sodium-ammonia solutions as noted in the legend. The two points at -75' are extrapolated a slight distance from a measured concentration of sodium iodide to zero sodium iodide.'O The two points a t -56.5' are direct determinations.'O For lithium the consolute temperature is -63.5 f 0.2" and the consolate composition is 0.0451 0.005 g.-atom of lithium/mole of ammonia. For potassium the consolute temperature is -70.0 f 0.4" and the consolute composition is 0.0455 =k 0.005 g.-atom of potassium/mole of ammonia. I n our determinations on rubidium, a tube was prepared and analyzed to contain 0.0545 g.-atom of rubidium/mole of ammonia ; although this concentration is close to that for phase separation in lithium, sodium, and potassium, we were unable to detect liquid-liquid phase separation at any temperature above that a t which the ammonia froze, approximately -78". In the preparation made with a
*
Volume 58,Number 10
October, 1954
2868
PAUL11. SCHETTLER, JR.,A N D ANDREWPATTERSON, ,JLt.
-'r
~
Table I : Temperature of Initial Separation of Concentration
POTASSIUM SODIUM-1(RWS'_POINTS A SODIUM-FRAPPES WlNTS @ SODIUM-SCHETTLER X S0DlUM.EXTRAWLATEDSCHETTLER t LITHIUM 0
LLS a
Function
e
. o \o
T e m p . of initial separation
.c
,. .
~ ) "
-.
n,
Mole ratio
m
Lithium in liquid ammonia
a'o
\
-75.00 -75.00 -70 00 -67.00 -67.00 - 63 ,51 - 63.31" -67.00 -67.00 -70.00 -70.00 -75.00 -75.00
\
\
0 02337 0 02611 0.02786 0.02979 0,02966 0 0 0 0 0 0 0
04509 06580 06611 07472 07497 OX576 08614
1 372 1 ,533 1 635 1 749 1 741
2 3 3 4 4
G48 868 882 3875 402 5 036 5 058
Potassiurn in liquid ammonia
-801
'
.01
1
.02
I
I
I
I
I
03 0 4 .OS .O 07 .OB 09 MOLES ALKALI METAL BYOLE AMMONIA ALKALI MFTAL - AMMONIA PHASE DIAGRAMS
I
010
I 0.11
Figure 2. Plot of phaRe data for lithium and potassium, with Kraus's data for sodium (small circles). The data of Frappe ("Phenomenes de transports electriques danv les solutions sodium-ammoniac liquide," I). E. S. Lille, 1958) are shown as triangles. Data of Schettler (ref. 10) shown as X (-75") are extrapolated from finite to zero concentration of sodium iodide; those shown as concentric circles (-56.5') are actual determinations on sodium. Lithium data shown as extended crosses are double points indistinguishable on the scale of the graph.
Faraday tube, we were unable to detect phase separation within a range of concentration including 0.02 to 0.10 g.-atom of rubidium/mole of ammonia and temperatures above the freezing point of the ammonia. From these experiments we conclude that phase separation for rubidium in all probability does not exist, and that a miscibility gap, if any, is too small and too close to the freezing point of the solution to be detected by visual means, which as we shall describe below is the most sensitive test we have been able to devise.
Discussion When we attempted to repeat some of the observations reported by other workers we soon found ,that phase separation is much more easily observed in a tube of diameter relatively small compared to the depth of solution in it, while in a spherical bulb it is quite commonly not visible at all. Kraus' relates that his method for sodium was restricted to certain The Journal of Physical Chemistry
-77.5 -74.8 -72.7 -71.2 -70.9 -70.6 -70.6 -71.4 -72.6 -73.2 -75.5
0 0 0 0 0 0 0 0 0 0 0
0239 0284 0308 0389 0370 0365 0465 0598 0620 0614 0667
1 403 1 668 1 8086 2 284 2 173 2 143 2 730 3 511 3 641 3 605 3 917
Sodium in liquid ammonia -56.5 -56.5 -56.5 -56.5 -75.0* -75.0b
0.08212 0,08180 0,01734 0,01698 0,008 0.11
4 4 1 0 0 6
822 803 018 997 4 4
a These samples were determined by the visual method m with potwsium. These data were extrapolated from finite concentration of sodium iodide to zero sodium iodide. See ref. 10.
ranges of concentration and temperature. In particular, he obtained small irregularities around the consolute temperature and large deviations at temperatures more than 20" below the consolute point. Our observations suggest that the reason for these difficulties lies in the surface-wetting and surface tension characteristics of the solutions. It thus appears desirable to recount some of our qualitative observations. In what follows, the comments are equally applicable to lithium, sodium, and potassium. The less dense, concentrated metallic phase has a bronze luster with a reddish violet appearance, more so when the solution
LIQCID-LIQUII) I’IIASE SEPARATION IN ALKALI METAL-AMMONIA SOLUTIONS
is concentrated or saturated and when the solution is viewed directly on its surface rather than in contact with the glass tube. The more dense, dilute phase has the typical inky blue color. The contrast between the two phases is greatest a t lower temperatures, and becomes progressively less sure as one nioves toward the consolute temperature. When the composition is near this temperature, but two phases are present, the blue phase,is observed to adhere to the glass and to “wet” it more effectively than does the concentrated phmc. When a tube was withdrawn from the cold bath a two-phase solution would look as if it were composed cornpletcly of dilute phase, but within a few seconds the layer of dilute solution would slide down the glass sides revealing two phases. KO doubt this phenomenon is the source of reports that there is no visible difference between phases, and one can anticipate it would cause trouble in resistance measurements to determine the presence of two phases. With solutions high in total metal content, the first droplet of dilute phase would form, not a t the bottom of the tube, but as a ring just under the surface of the concentrated phase where it met the glass tube, a t the edge of the meniscus. When the solutions are frozen t o liquid nitrogen temperature, the lithiurn solution retains its bronzy appearance, while tho others become blue-gray crystalline masses.
2869
The behavior of rubidium requires special comment. I n the Faraday tube arrangement, the concentration could be varied conveniently over a range of fivefold, from 0.02 to 0.1 g.-atom of rubidium/mole of ammonia. The temperature was varied from that of Dry Ice in alcohol upward several degrees through use of an intermediate thermostat of liquid ammonia. The more concentrated solutions were bronzy, the more dilute blue. At no time was an interface observed which signified phase separation. Distilled portions of the ammonia solvent were prone to freeze on the walls of the tube mixed with the blue, dilute solution and then to melt and run down suddenly on warming a fraction of a degree, thus giving an appearance not unlike that described for phase separation in the other solutions a t higher temperatures. This behavior never gave rise to a clear-cut phase boundary. On this evidence, we have concluded that with the most sensitive means of observation a t our disposal, phase separation is not seen in rubidium-ammonia solutions above the temperature a t which the ammonia freezes.
Acknowledgment. We are particularly indebted t o M. le Chanoine GBrard Lepoutre, who assisted in much of this work, criticized it, and suggested the visual method for determining the potassium diagram. We thank also the National Science Foundation for their support.
Volume 68,.Vumber 10 October, 1964
