e UC"4 Chemlstry T T '~:=TI D-4500 (20th ed., Rev ) G T IATIONAL I.ABORA'I‘ORY . "operated ';:i_By' : T UNION@_CARB!DE CORPORATION o OM'IC"*ENERGY COMMISSION 'ln USA. Pnce' $‘ 50 Avmloble from the _' 70ff|ce of Technicul Services L [ nor the Commrssnon, ‘hor uny person ochng on behulf of ?he Commissmn. i, .,C e v, - 1 A Mukes .any’, warranfy or: represenfahon, !xpressed ar ‘imphad with Tespec! fn the accumcy, S x comp!ereness -or . usefu!ness of the mformofion confumed in this raport, ‘or fhot ‘the “use of any |nformanon, oppurafus, rnehod or procass dusclosad in fhESr report may not mfrmge . privately owned nghfs, ot - C : : B, ‘AsSt)mes any hablhhes wflh respect to the use of or. for dumcges resulhng fram fha use of . . any, mformahon, upparutus method or process disclosed in flus report. ) o As used ' ‘in the .. ubov;e, person uchng on behulf of the Commlssmn mciudes any emp!oyee” 'controctor of 1be Commnssmn, or omp!oyee of. such contructor, to fhe _extent fhaf such employee F “or confroctor af the Commnssron, or amployee of : such confructor prepares,_dussemmo&es, or provldes access to, ‘any. m!ormchnn ',pursuanf fo h|s amploymerlt or contruct wnh fhe Commrssson, 7 oF h1s employment wnh such confractor. : e : L P‘ e 4 . flj, ; ¥ - »i ORNL-3391 Contract No. W-7405-eng-26 CHEMISTRY DIVISION MIXTURES OF METALS WITH MOLTEN SALTS M. A. Bredig DATE ISSUED BUG 15 1363 OAK RIDGE NATIONAI, LABORATORY Oak Ridge, Tennessee operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION . y s l“ “) v ¥ ® ) iii CONTENTS Abstractl‘.l...I.........OQ-.....I'C....l‘......l..C.'I. IntroduCtion......‘O.....I.......I..l....'..."C'.'.-l.....O‘.I... "Metallic" Metal SolutionS.eec... T eesetesessessetsnessetseebense e A]kali‘Metal_Alkali-Halide SystemS. ® 5 4 0 00BN LR ORI COIBETEONEPIETSIO RS Phase Diagrams..........flfl.l Physical Properties..... S 9 8 48 48 %9 S e OS2 LI I I B BN TN B B B K N RE BN BN BN Alkaline-Farth-Metal—Halide Systems.eceeceiecsnccsverercncnee Phase Diagrams...‘...l....-..I..I...........'I..l...‘.l. EleCtriCall Conductalnce....‘......‘.l...l.........l..l..l Rare-Farth-Metal—-Trihalide Systems... Hlase Diagrams....-.............'I.................. Electrical Conductivityeeseesccoacsnss o 6 8 08 s e 80P llNOnmetallic“ Metaal SolutionSO..iI.......l‘..CG.“.O'G...'I....-.. TranSition Metals..'......l.....ll....I.l‘..l..l..ll...ll'... Posttransition MetalSeeeecesosesesesosasocccscsossssacossncacsssssae SUNMAryYeeoesse ® o 0% 60 P e &t e e Pe e R ReferenCGS.....‘...-........‘.............. & " @ o8 s et S e e b Page NN G oo oWWw M~ e 30 32 37 41 43 bds 51 52 " ) T MIXTURES OF METALS WITH MOLTEN SALTS M. A. Bredig ABSTRACT A review is presented of various types of solutions of metals in molten salts, especially in their own molten ha- lides. With relatively little reference to the older liter- ature, the progress made in the last 20 years is discussed. Roughly, the solutions are classified into two groups: The metal may retain, to some degree, its metallic properties in the solution, or it may lose them through strong interaction with the salt solvent. The alkali-metal systems are typical examples of the former type, while solutions of cadmium or bismuth represent the second. Equilibrium phase diagram data are presented in detail for many metal-salt systems. These include critical solution temperatures, that is, tem- peratures above which metal and salt are miscible in all proportions. Electrical conductivity is singled out as a most significant physical property from which conclusions on the state of the electron in the solution may be drawn. In the electronically conducting solutions, notably of the al- kali metals, the electrons may be thought to resemble F centers in color-center colored crystals. In solutions where electronic conductance is absent, monomeric, dimeric, and even more highly polymerized species of the solute metal in a low valence state must be assumed to occur. INTRODUCTION The history of the subject of interactions between metals and their salts in the molten state, though not well known to most chemists, is at least 150 years old.* It includes observations made by Davy in 1807 on the production of deeply colored melts and on difficulties in recovering metal at the cathode upon electrdlysis of molten alkali-metal hydroxides. Similar phenomena were encountered by chemists who, either in small-scale laboratory research or in larger-scalé industrial production, have been or are concerned with the recovery of certain metals by molten~salt elec- trolysis.? Colored metal "fogs" were recognized by various investigators? as one important cause of low current yield in the electrolysis of molten salts. Various theories ranging from colloidal suspensions to "subhalides" 1 were propoéed to explain the observations. One also finds in the liter- kfi; ature the statement by Nernst:4 "For not a single metal do we know a nonmetallic solvent in which the metal would dissolve without chemical interaction and from which the pure metal could be recovered simply by recrystallization ... ." We know today that many such molten-salt sol- vents do exist. A point of view which appears to be exactly opposite that Of Nernst, toward which one group of authors”~7 seems to be inclined, stresses the notion that metal atoms may indeed be dissolved as such in molten salts, especially when the interaction with the molten solvent is not a strong one. Neither view, if taken in the extreme, appears to be ‘:T): consistent with experimental observation. _ The present discussion may profitably be restricted to cases in which a metal dissolves in one of its own molten halides. Salts contain- ing anions such as nitrates, sulfates, or phosphates are likely to de- compose in reactions with the metsl. Little of a precise nature is as yet known about solutions of one metal in the molten salt of another, but a few specific cases will be briefly mentioned. Facts will be presented which demonstrate that the variety of true solutions of metals in fused salts is considerable. This includes, at one extreme, dissolu- tion with relatively little solvent interaction, which might be described by the concept of electrons substituted for anions in cavities, or that of solvated metallic electrons. At the other extreme, dissolution with chemical reaction between solvent and solute occurs, as in the formation = of "subhalides,” where unusually low ionic valence states of the metallic | element are formed. Contrary to the older literature® as well as to more . recent statements,® no evidence whatsoever seems to exist for the occur- rence of the colloidal state of metals in fused salts (with the exception of highly viscous systems such as silicates). It is convenient to divide solutions of metals in their molten ha- lides into two main categories: | 1. The metal imparts partially metallic character to its solution in the salt. The properties of the solution reflect the presence of mobile electrons. These are the true metal solutions. They are of par; ticular interest because they represent a novel state of metallic matter. ‘ 0 o} w) b i» They are somewhat similar to solutions of metals in liquid ammonia and similar solvents, but are distinguished from these largely by the ionic nature of the solvent (the salt), which, it is to be especially noted, usually contains one of the constituents of the metal which it dissolves, namely the cation. Semiconducting solutions‘(i.e., those having positive temperature coefficients of the electronic part of the conductance) are included and may, in fact, represent the majority of examples in this class. 2. Strong interaction (i.e., chemical reaction, oxidation-reduction) occurs between metal and salt. The metal assumes a valence higher than zero but lower than normal. This class might be designated "subhalide" solutions. One might distinguish two subgroups, depending on the nature of the second, that is, the metal-rich phase (solid or liquid) that, at saturation, is in equilibrium with the molten salt-rich phase: (a) The reaction goes so far as to lead to the crystallization of a solid "subhalide." ‘ (b) The metallic element itself (with a small amount of salt dissolved) forms the second phase. In this group the lower valence state of the metal is stable only in the molten solution. The division between 1 and 2 is not clear cut, and both mechanisms of dissolution may describe a single system. The distinction is between relatively mobile electrons and electrons which attach themselves to, and become part of, ions to produce a lower valence state. There are likely to be intermediate cases not clearly defined, and the attachment of elec- trons (or "subhalide" formation) may or may not be a matter of degree only, rather than a matter of a statistical equilibrium between two distinetly different states of the electron, attached and unattached. Also, this may vary with temperature and composition. The variety in the nature of metal solutions will be illustrated in the following by a systematic consideration of a number of examples. As we are dealing with mixtures of metals with salts in the liquid state, we shall also have occasion to consider metal-rich phases and the solubility of salts in liquid metals, which in many cases is greater than that of the metal in the molten salt. The discussion will be based largely on measurements of the phase behavior and the electrical behavior of these systems. "METALIIC" METAL SOLUTIONS The metal-~metal halide systems which might be termed "metallic" metal solutions are mainly those of the metals of the first two main groups of the periodic system and of some of the rare-earth metals. Farlier literature, that of the last century and the beginning of the present one, has been reviewed by Cu.bicciotti,1 and a rather complete review has been published by Ukshe and Bukun.? Many of the earlier re- sults must be considered as of rather limited value, except for the fact that they indicated the existence of stable mixtures of metals with salts, both solid and molten. With a few exceptions, we shall not deal in any detail with these older data, many of which were relatively in- accurate, or were given erroneous interpretations. We shall confine our- selves to the more recent work, which is beginning to lead to a far more satisfactory explanation of these solutions than was available 20 years ago. Alkali-Metal—Alkali-Halide Systems Phase Diagrams A study of alkaline-earth-metal—alkaline-earth-halide systems under- taken under the auspices of the Atomic Fnergy Research Program in the United States during World War II (the "Manhattan Project')® and the subsequent investigations of the alkali-metal—alkali~hslide systems begun in the early 1950's at the Oak Ridge National ILaboratory seem to have initiated a new period of considerable interest in metal-salt solutions. Except for rather demanding experimental conditions, the situation was believedll to be particularly simple with the elements of the first main group of the periodic system, where complications related to formation of a lower valence state of the metal were not anticipated. Indeed, the phase diagrams, in which the alkali metal is one component and one of its halides the other, are relatively simple (Figs. 1-6). The prinéipal data of these diagrams are given in Table 1. The temperature range of C UNCLASSIFIED ORNL-LR-DWG, 21894 A UNCLASSIFIED 69 OLE FRACTION MX | 1200 CRNL-LR-DWG. 38872 1200 |8 7 6 5 4 3 2 .1 1200 | i I [ i [ | | | © COOLING CURVES, THIS WORK X EQUILIBRATION, MAB JWIWTS 1480° HIGH TEMPERATURE RANGE & EQUILIBRATION, HRE, MB OF 1100 1080° —1100 - lggfii-lfa'-%&%- 214618 SODIUM METAL-SODIUM HALIDE aAQ4° 0,0, . COOLING CURVES SYSTEMS 900 EQUILIBRATION AND SAMPLING _| 14150 — — 1000 TWO LIQUIDS —11000 G °C NaCl - Na 900 900 ool | — 800 800 ! I o0 { x 7 SOLID SALT + LIQUID METAL ,l NaCl- Na 00 700 'l CsF-Cs 1050 [—/ — 600 600 ! ! 1" 1026° 1033° 10001 | — 1000 |- —1000 J__ OC x X DC TWO LIQUIDS [/ Two Liauibs 900} 4 —{900 * NaBr - Na { \r 950~ x — 800 Nal - Na 800 N 740° \ ¥ & 700+ {700 , | Ll g SOLID SALT + LIQUID METAL { Lx 656° | 10 20 30 40 50 60 70 80 90 l S0LID ¢ SALT+ LIQUID METAL i K 900 L L1 b 1 1 1 | sool L1 1 1 11 i Tty Y eoo mole % 00 20 30 40 50 60 70 B8O 909 4 23456.78.9 1.2.3 45 e 7 s 9 MOLE % SODIUM METAL MOLE FRAGTION OF METAL—> MOLE FRACTION OF METAL —> Fig. 1. Potassium Metal-Potassium Halide Systems [J. W. Johnson and M. Fig. 2. Sodium Metal-Sodium Halide Systems, Fig. 3. Sodium Metal—~Sodium Halide Systems [M. A. Bredig, J. Pbys. Chem. 62, 606 (1958) High-Temperature Range [adapted from M. A. A. Bredig and H. R. Bronstein, . Phys. Chem. 64, (reprinted by permission of the copyright Bredig and H. R. Bronstein, ]. Phys. Chem. 65 {1960) (reprinted by permission of the copyright owner, the American Chemical Society)l. 64, 64 (1960} ]. owner, the American Chemical Society)]. UNCLASSIFIED ORNL-LR-DWG, 310678 I ! I 1300 f— — LiF - Li 12C0 2 HOO 1000 | E c | 900 {i— : \, ’ \ 800i— - RbF - Rb = 700 _ CsF - Cs l I | 8005 25 50 75 100 MOLE % METAL Fig. 4. Alkali Metal-Alkali Metal Fluoride Systems [A. S. Dworkin, H R. Bronstein, and M. A. Bredig, J. Phys. Chem. 66, 572 (1962) (reprinted by permission of the copyright owner, the American Chemical Society)]. UNCLASS FIED ORNL-LR-DWG. 37011 T 1 T 700 €50 600 °C 550 500 CESIUM METAL - HALIDE SYSTEMS 450 ' ' L 0 25 50 75 100 MOLE % METAL Fig. 5. Cesium Metal-Cesium Halide Systems [adopted from M. A. Bredig, H. R. Bronstein, and W. T. Smith, Jr., J. Am. Chem. Soc. 77, 1454 (1955)]. UNCLASSIFIED ORNL-LR-DWG, 32909A sool o] | | ] -X-. i\ N\ / RbF- Rb ~ 0 T N a 773 \\ 750 — —1 706° 700 }— E‘ -Rb — °C X 696° —. \\'\x Rb8Sr-Rb \1 —— 650’( T —] -—x—-x ,\.,,’N /—;bl Rb \ . /x‘ X SIS' 600 — — 550 I | | 25 50 75 MOLE % Rb METAL Fig. 6. Rubidium Metal—Rubidium Halide Systems [M. A. Bredig and J. W. Johnson, J. Pbys. Chem. 64, 1900 (1960) (reprinted by permission of the copyright owner, the American Chemical Society)]. | ) Al ) Table 1. Principal Data for Alkali-Metal—-Alkali-Haelide Phase Diagrams Monotectic Salt- Consolute Metal Salt Tesp Phase Comp. (mole % M) Eutectic Metal System mp ey Tem Com Comp™ mp MX-M (°x) Salt-Rich Metal- ° § 1 i y (mole % MX) (°k) (MX) Rich X (mo e M Solid ILiquid ILiquid LiF 1121 1120 1b 3 1603 40 452 Licl 883 ggob 0.5b 452 LiBr 823 822b 452 LiT 742 74P 1b 452 NaF 1268 1263 3b 83 1453 28 10™9 370 NaCl 1073 1068 0.15 2.1 97.7 1353 50 370 NaBr 1020 1013 0.15 2.9 96 .6 1299 52 370 Nal 933 930 1.6 98.6 1306 59 370 KF 1131 1122 %49 51L.7 1177 20 3 x 104 337 KC1 1043 1024 0.04¢ 10.5 75,0 1063 39 109 337 KBr 1007 o981 0.03¢ 19.0 69.2 1001 4ty 10™° 337 KI 954, 931 13.5 82.5 990 50 10™° 337 ROF 1068 1046 9b 40b 1063 21 312 RbCl 995 969 18 57 979 37 312 RbBr 965 a a a d 910b,e 44b ye 312 RbI 920 888 22 73 907 51 312 CsF 976 d d d a a d 103 302 CsCl 918 a d a d el a 108 302 CsBr 909 d,a d,a d,a d,a d,a d,a 10™7 302 CsI 899 a d d d a a 107 302 b cEstimated. By extrapolation from ref 13. e SEstimate by extrapolation. 0 miscibility gap. Unstable. special interest is that near the melting points of the salts and above. The melting points of the alkali metals are very much lower than those of their salts and are not far above room temperature. The solubility of the salts in the liquid metals, which is considersble in the higher temperature range, decreases rapidly with decreasing temperature, so that the composition of the eutectic liquids in these systems is that of almost pure alkali metal. The mole fraction of the salt in these eutec- tics is, with few exceptions, less than 1077, as estimated by extrapole~ fion of the liquidus curves. Although a eutectic of this nature has sometimes been known under the term "monotectic," this usage will not be followed here. We shall use this term? to designate the equilibrium between two liquids and one solid of a composition not intermediate be- tween those of the two liquids. Thus, in most alkali-metal—alkali-halide systems the melting point of the salt is lowered by the addition of metal until the monotectic temperature and composition are reached. Above the "monotectic horizontal" we £ind the region of coexistence of one solid and one liquid phase, and the region of the coexistence of two liquids, one richer in salt, the other richer in metal. In the alkall-metal sys- tems the compositions of the two liquid phases approach each other monotonically with increasing temperature until, at the critical solution or consolute temperature, they equal each other. At and above this tem- perature, only one liquid phase exists under equilibrium conditions. Below the monotectic temperature, solid sslt, containing in solid solu- tion very small amounts of metal (which decrease rapidly with decreasing temperature),’? is in equilibrium with a solution of the salt in liquid metal. The general déscription of the phase diagram given here fits the 1ithium,*% sodium,'+1% and potassium'® metal systems with each of their halides, all of which, with the exception of the chloride, bromide, and iodide systems of lithium, are known in detail. Among these three alkali metals the potassium systems (Fig. 1) exhibit the highest degree of misci- bility of the salt with the metal in the liquid state. The sodium sys- - tems (Figs. 2 and 3) are intermediate, and lithium metal (Fig. 4) shows the least tendency of all alkali-metal systems to mix with its halides * w? 'y in the liquid phase. The critical solution temperature in these systems represents a qualitative measure of the relative miscibility and of the relative deviations from "ideal" solution behavior. This consolute tem- perature is 1330°C in the lithium fluoride system, ranges from 1028 to 1180°C in the sodium systems, and ranges from 717 to 904°C in the potas- sium systems (Teble 1). The temperature range in which two liquids coexist is about 530°C in the lithium fluoride system, about 200 to 400°C for sodium systems, and only 20 to 60°C for the potassium systems. No precise data appear as yet to be available for ternary alkali-metal systems embodying two or more cations with one anion (besides the elec- tron), or several anions with one cation. It is evident that the cesium—cesium-halide systems!? are qualita- tively different from the majority of the alkali-metal—alkali-halide sys- tems, in that the liquids are miscible in all proportions (i.e., the consolute temperature lies below the liquidus line) (Fig. 5). The liqui- dus curve, depicting the temperatures for the solidification of the ce- sium halide from the liquid mixture, descends without discontinuity (except that there ought to be a slight kink at the transformation tem- perature of CsCl) from the melting point of the pure salt to the eutectic point, which is almost identical in melting temperature (30°C) and compo- sition with the pure metal.” The four rubidium phase diasgrams!® show, as might be expected, a behavior intermediate between the potassium and the cesium systems (Fig. 6). The temperature range of only partial miscibil- ity is less than 20°C in the fluoride, chloride, and iodide systems, and the miscibility gap is absent in the bromide system. It is interesting to note (Fig. 7) that the trend, with increasing atomic number, size, or polarizability of the halide ion, toward greater miscibility as expressed by the value of the critical solution temperature [T, = 1180, 1080, and 1026°C for Na(F, Cl, Br); 904, 790, and 728°C for K(F, Cl, Br); and 790, 706, and < 655°C for Rb(F, Cl, Br) respectively] is, in the iodide systems ¥Delimarskii and Markov (ref 18), in Table 29, page 211, "Solubility of Metals in Fused Salts," misstate the case of the cesium systems, by erroneously taking the metal concentration of the solution in equilibrium with solid salt for a metal concentration of a salt-rich phase in equilib- rium with liquid cesium metal. Many other metal solubilities in this table are much outdated and quite in error. b 4 e b 10 Fig. 7. Critical Solution Temperature in Alkali Metal—Alkali Metal Halide Systems vs Molar Refrac- tion of the Gaseous Halide lon. Tes cs UNCLASSIFIED ORNL-LR-DWG. 67077 1200 ¥ I T l T l Y - 1100 | z o Ci 1000} Br 900 - 8ok K Ci T700— Br 600 2 - 3 -1 Rx (cm3 mole™) UNCLASSIFIED ORNL-LR-DWG, 70124 5000 I S I | [ I I 4000 | ~ 3000 |- — 2 5 E & : KB,r’/ L~ / / //' Csl / / 7/ // 1000 {- . — 4 | 2 | | l | | | | | | 01 02 03 0.4 05 2 *u 0.6 0.7 08 09 ({0 Fig. 8. Excess Free Energy of Mixing as Function of Volume Fraction of Metal. O i i i % i i i ! i i | " oy “_gfl, 11 of these three alkali metals, either reversed [Na(I):1033°] or consider- ably diminished [K(I):708° and Rb(I):634°], with data lacking for the lithium systems and for Cs-CsBr. Although it seems reasonable that the solution behavior and T, might be related to the size and polarizability of the anions, no simple quantitative relationship to these anion proper- ties has been deduced. Besides the value of the minimm temperature of complete miscibility (Te), the ("eritical") composition at the consolute temperature is of interest. The metal content of the "eritical" solution is seen; in Table 1, to increase from the fluoride to the iodide for the sodium, po- tassium, and rubidium systems; the same is true for the compositions of the inflection on the solidus curve of the cesium systems, which is some- what representative of the "submerged" (i.e., unstable) critical solu- tions (28, 47, and 65 mole % cesium metal for CsF, CsCl, and CsI solution with cesium respectively). It has been pointed out that the correspond- ing "eritical" volume fractions approach the value 0.50 for all systems, indicating the quite plausible significance of the molar volume in the solution behavior.t® The activity coefficients of the salts, calculated from the liquidus curves for the precipitation of solid salts from the melts, increase from unity with increasing metal concentration, corresponding to positive devi- ations of the salt component from Raoult's law. This is usual in systems exhibiting (or having a tendency to exhibit) liquid-liquid miscibility gaps, when no strong ("chemical') interactions such as in the cadmium— cadmium halide systems (cf. below) take place. In Fig. 8 are plotted values of RT Iny_ ., Vs ¢§etal for the systems K-KBr, Rb-RbBr, and Cs-CsI, where ¢metal is the volume fraction of the metal. The temperature dependence of the solubility of the solid salts in the liquid metals is another feature of interest. An unusually low value of the partial molar heat of solution is derived from the temperature depehdence of the solubility of the solid salts in the liquid metals for the fluorides of cesium (10 keal/mole) and potaessium (13 kecal/mole) in comparison with the other halides of these metals. 12 No explanation of these phenomena which is free of arbitrary assump- tions has been proposed, although rather interesting speculationsils20 lead to a crude but instructive picture of the solution behavibr of these systems. Pitzer?l suggests that the excess free energy of mixing is associated with the conversion of the metallic state of binding of the metal elec- trons to an ionic type of binding and that the mixing of electrons with halide ions occurs with little excess free energy, if any. He obtains fair agreement between experimental excess free energies derived from the phase diagrams and a theoretical calculation of the difference in energy of a metallic and ionic model of an alkali metal. An instructive view msy be obtained according to Blander?? by dividing the process for the dissolution of metal into the two steps: M(liquid) — M(gas) , M(gas) — M(in dilute solution) . The first step is simply the vaporization of the metal for which data are available.?? The thermodynamics related to the second step, which is the dissolution of the gas molecules, is discussed elsewhere.?* The influ- ence of the first step may be illustrated by the relative values of RT 1n p given in Table 2, where p is the partial pressure of the metal mono- mer in the vapor in atmospheres. Aside from differences in the second process, the influence of the vaporization step would tend toward an in- creasing relative solubility of metal in salts in the order of increasing vapor pressure, Li < Na < K < Rb < Cs. Where Henry's law applies, the second process is best described in terms of the Henry's law constant, K,, Ke = csol/cgas ’ where Cool is the concentration of metal in solution in moles/cm® and cgas is the concentration of metal in the gas phase in the same units. The free energy of solution (M, = —RT 1n K,) is the sum of the free energy of forming a hole the size of the metal atom (sbout 10 * 5 kcal/ mole) and the free energy of interaction of the metal atom with the salt. 9 i1 ag u‘l) 13 Table 2. Miscibility and Metal Vaporization Data for the Alkali-Metal—-Alksli-Halide Systems RT In p (monamer) Metal bp Range of Values of Tg (kcal mole) (OK) _ (oK) (at 1000°K) L - 13.8 1604 1603 Na -3.6 1163 1303-1453 K 0.8 1039 981~1177 Rb +0.3 97 9071063 s +0.6 958 (<850 to <950 "submerged") As an example, one might consider the dissolution of sodium in a hypo- thetical sodium salt having a molar volume of about 50 cm? at the boiling point (1163°K) to form a 1 mole % solution. These parameters are not very different from those for real salts. The free-energy change, M., for this process, if Henry's law holds even approximately, is ~7 kcal/mole, and the interaction energy of a metal atom with the solution would have to be about =17 * 5 kcal/mole. Since the solubilities of sodium are higher than 1 mole %, the postulate of the dissolution of metal atoms with weak interaction is seen to be improbable. That the dissolution of alkali metals in molten salts is Influenced greatly by the same factors influ- encing the vaporization is shown by the comparison in Table 2 of the boil- ing points of the metals and the consolute temperatures. This is, of course, only suggestive, and much more data and a more complete quantita- - tive analysis are necessary for any definite conclusions. Physical Properties Optical Observations. — Aside from the early qualitative findings about the deep coloration of molten alkali halides containing excess metal, there appears to be, thus far,'only one brief study of a semiquan- titative nature. Mollwo2? measured the visible and near-infrared absorp- tion spectra of melts of alkali-halide crystals colored by exposure to 14 alkali-metal vapor. He found broad sbsorption maxima at 790 mp for all three sodium salts studied (NaCl, NeBr, and NaIl) and at 980 mu for the corresponding potassium salts. In contrast to this, the position of the sbsorption maximum in the solid salts varies with the anion, or the lat- tice parameter. Mollwo believed that the absorption bands of the melts could be considered as resonance lines of the metal atoms, displaced and broadened by intermolecular fields. However, it seems entirely possible to ascribe the absorption spectrum to an electron whose wave function is not confined to one metal core (M) but spread out over several M' ions [e“(Nfi)X] as in an F center, but without the decisive influence, upon its energy, of a rigid crystal lattice. Electrical Properties. — In a very few rather crude experiments, Mollwo?? found'no directional transport of the colored cloud from a melting, F-center colored crystal. He considered this as a confirmation of the optical finding, which indicated the presence of uncharged (i.e., nondissociated) metal atoms.2® This conclusion does not seem to be con- firmed by later observations. Bronstein and Bredig?7»28 measured the conductance of alkali-metal— alkali-metal-salt mixtures and demonstrated very considerable transport of electricity by electrons in solution. The stainless steel apparatus for the determination of the specific electrical conductivity, which they describe in detail, included a synthetic-sapphire capillary dip cell which was found to be entirely inert to most alkali-metal-halide mixtures. A1l of the seven sodium and potassium systems investigated (the tempera- tures of the Na~NaF system were not accessible) had in common a rise in the specific conductivity of the salt melt on addition of metal.27»28 This rise, however, varied greatly in extent and nature from system to system. TFigures 9, 10, and 11 are typical examples. In the sodium sys- tems the rapid initial rise in conductivity is characteristically slowed, ~ and only in cases of sufficient solubility (Na-NaBr and Na-Nal at higher temperatures) is it finally accelerated as the solubility limit in con- centration is approached. In the potassium solutions, on the other hand, the rise is monotonically accelerated with increasing metal concentra=- tion. In solutions of potassium in KI, a specific conductivity of 600 ohm™% em™t, that is, 400 times that of the pure molten salt, was measured n - A 15 at 42 mole % potassium, indicating an electronic contribution to the con- ductivity of this solution of more than 99.5%. The potassium fluoride solutions are distinguished from those in the iodide by a far slower initial rise in conductivity, with the chloride and bromide melts being intermediate. Bronstéin and Bredig define a characteristic or apparent W &) equivalent conductance Ny of the dissolved metal by the equation A - (L —-DN_JA _ “soln ( M)sfli My = ’ Ny ORNL-CR- DWa. 230484 10 L ‘I—fil T I T I T T ! T { T T T - I T 30 9 . 8 28 L 890° . 5 B I'_- 7 —26 ™~ - a 845 24 L Na - NaCl 1 2 122 ) —{20 . [760" 118 KB 5700 —10 SPEGIFIC CONDUCTANCE (chmi! cm™) ] > SPECIFIC CONDUCTANGE (ohmtem™) Q° ] 82 ]K-KCI 8 860° ] 9 16 — 4 5 Na-NaBr 4, 1 4L 1 L 1 | .l I 1 1 I 1 J L I 1 l i O 0 2 4 6 8 0 2 4 6 8 10 SODIUM METAL (mole %) POTASSIUM METAL (mole %) Fig. 9. Specific Conductivity of Alkali Metal Solutions in Alkali Halides [H. R. Bronstein and M. A. Bredig, J. Am. Chem. Soc. 80, 2080 (1958) (re- printed by permission of the copyright owner, the American Chemical Society)]. x, SPECIFIC CONDUCTIVITY, ohm™! cm™! UNCLASSIFIED ORNL-LR - DWG. 54649 16 Fig. 10. Specific Conductivity of Solutions of Sodium Metal in Sodium lodide [H. R. Bronsteif and M. A. Bredig, J. Pbys. Chem. 65, 1221 (1961) (re- printed by permission of the copyright owner, the 16 a T T 1 T T 14( * . 12| - 900°C 10} 8_ 800°C . Ameri ica i . NoI- Na merican Chemical Society)] 2} ~ | | | | | 1 0 2 4 6 8 10 12 14 MOLE PER CENT SODIUM METAL 20 UNCLASSIFIED ORNL-LR-DWG. 54650 18 - -~ - N H o x , SPECIFIC CONDUCTIVITY, ohm !cm! o _ 1 l [ [ { 4 6 8 10 12 MOLE PER CENT POTASSIUM METAL Fig. 11. Specific Conductivity of Solutions of Potassium Metal in Potassium lodide or Fluoride [H. R. Bronstein and M. A. Bredig, J. Pbys. Chem. 65, 1221 (1961) (reprinted by permission of the copy- right owner, the American Chemical Society)]. ) ok %) 17 where AL oin and Ajppy BT the equivalent conductances of the solution and salt, and Ny is the equivalent fraction of metal. The term Ay repre- sents the change of the conductance caused by the addition of one equiva~ lent of metal to an amount of salt conteined in the solution of the given composition which is between two electrodé plates 1 cm apart. Figures 12 and 13 show, in a more striking form than Figs. 9, 10, and 11, the typical difference in the behavior of sodium and potassium. While in the potassium systems the apparent equivalent conductance of the metal solute, fys Tises monotonically and with increased rate with inéreasing metal content, it drops rapidly in the sodium solutions toward a minimum, which obviously would be followed by a rise if greater solUbility would permit higher concentrations of metal to be dissolved. These characteristic changes of Ay, with metal concentration mey be interpreted as follows: At infinite dilution of the metal, in both the sodium and potassium solutions, electrons are in a state in which they can contribute to the carrying of current. Rice,?? assuming a "random walk" of the electrons, has performed theoretical calculations for this state at infinite dilution which are correct within an order of magnitude of the observed conductivity. For the potassium solutions, Rice's theory predicts an increase of Ay with metal concentration. Bronstein and Bredig suggest that electron orbital overlap plays a role in this increase, and doubtlessly this becomes important at higher metal concentrations. How- ever, in the sodium solutions, another important factor appears to be - present that actually removes electrons from the conduction process with .dincreasing metal concentration.';The trapping of electrons in pairs to form diatomic molecules of Na, waS'éuggested27 to rationalize this dif- ference. The fact that Nap is relatively more steble in the vapor than Kz (the heats of dissociation are 17.5 and 11.8 keal/mole respectively) suggests that this is reasonable. Measurements at higherrtemperature in the sodiumppontaining systems, 1if this hypothesis were true, would lead to behavior more like therpotassiumpcontaining systems because of the greater dissociation of Nas. In the lithium systems a still greater de- gree of pairing should occur. This,'rather than a short relaxation time,3° might be the reason for the absence of electron spin resonance in solutions of lithium in molten lithium iodide. EQUIVALENT CONDUCTANCE (ohm™! em?2 equivix 1073) UNGLASSIFIED 18 ORNL-LR-DWG. 23047A METAL (mole %) Fig. 13. Equivalent Conductance of Alkali Metal Dissolved in Alkali Halides [H. R. Bronstein and M. A. Bredig, J. Phys. Chem. 65, 1223 (1961) (reprinted by permission of the copyright owner, * the American Chemical Society)l. A, , METAL EQUIVALENT CONDUCTANCE, ohmicm? equivix 1073 n o @ o > i n o 0 L) o N Fig. 12. Equivalent Conductance of Alkali Metal Dissolved in Alkali Halides [H. R. Bronstein and M. A. Bredig, J. Am. Chem. Soc. 80, 2080 (1958) (re- printed by permission of the copyright owner, the American Chemical Society)]. UNCLASSIFIED ORNL-LR-DWG. 54651A i | 1 | ] ] - 2 4 6 8 {0 12 14 - MOLE PER CENT METAL (:) o 9 N 19 Another interesting aspect is the variation of Afl (i.e., Ay extrapo- lated to infinite dilution) with variation of either the anion or the metal ion. Table 3 gives a summary of the equivalent molar conductances, AE, of sodium and potassium metals in infinitely dilute solution in most of their molten halides. These values were obtained by a short extrapo- lation, to zero metal concentration, of the curves for fy vs metal con- centration. . Two definite trends may be seen in these data: First, the contribu- tion to the conductivity by the metal solute increases greatly in going from the fluoride to the iodide systems, that is, with atomic number, or size, of the halide ion. Second, it decreases in going from sodium to potassium (i.e., apparently, with increasing atomic numbef, or size, of the metal ion). The former trend has been attributed?7+28 to the increase in the polarizability of the halide ion with increasing size, which is thought to facilitate transmigsion of an electron from one cation, or group of cations, to another. This is reminiscent of the findings by Taube and others that the rate constant for exchange of an electron be- tween a complexed cation such as [Cr(NH3)sX]?t or [Cr(H,0)sX]?* and hy- drated Ccret ions is very highly dependent on the nature of the ligand halide ion, X~, presumably on its polarizability. However, while the ef- fect in aqueous solution shows a very high power dependence on the polari- zebility of X~ (the bimolecular rate constant for electron exchange be- tween [Cr(NH;)s5X]2t and Cret changes from 2.7 x 10™% to 5.1 X 10™2 to 0.32 to 5.5 for X = F, Cl, Br, and I respectively), in the metal-salt melts the apparent dependence of Afi'on polarizability of X is of a lower power. Quite significantly, the iodide systems exhibit a weakening of the trend established by the fluoride, chloride, and bromide systems. This is remi- niscent of a similar change in the trend_for the critical solution temper- atures mentioned sbove, & similarity which‘may not be merely coincidental. Of considerable interest would be a discussion of the temperature dependence of the electronic part of the conduction in infinitely dilute solution. Unfortunately, the data are, by their very nature (i.e., small differences of two large numbers divided by a small concentration value), not nearly accurate enough for this purpose, except to show that the tem- perature dependence at infinite dilution is very small. At higher metal 20 Teble 3. Molar Conductance A;Z of Sodium and Potassium in Infinitely Dilute Solution in Their Molten Halides at 900°C (ohm™ cm? mole™') Crystal Radius of Cation (A) F ¢l Br 1 Ne, 0.95 6000 12,000 (16,000)® X 1.33 800 2800 6,000 8,100 (820°c) (870°c) Molar refraction 2.5 9 12.7 19 of gaseous anion (em® mole™?) Cube of crystal 2452 5.94 7.43 9.95 radius (A?) a‘Es’t:.:i.mzad:,eél. UNCLASSIFIED 50 103 ORNL-LR-DWG, 70744 gls aofznec e o oNaBr o r o B % o :NuI eKI .RbIr S 3 20 f 3 / = 10 ? V4 x| -3t e 35 4.0 45 4 5 6 4 5 6 7 Fig. 14. Electron Mobility in Alkali Halide Crystals [A. Smakula, Nachr. Ges. Wiss. Géttingen, U 1(4), 55 (1934)]. REDUCED TEMPERATURE, 7/ o} » " 13} 2 21 concentrations the temperature coefficient (positive) of the metal solute conductivity appears to depend both on metal concentration and temperature (cf. Fig. 13). This might be attributed??+28 to the equilibrium between associated species such as Nap, and Ko and electrons, 2Kt + 2™ = K,. At still higher metal contents, where the liquid is essentially a solution of salt in metal, the temperature coefficient appears to become negative, as expected for metals. The very small temperature dependence of the metal solute conductivity at infinite dilution, A;, significantly distinguishes the melts from the F-center colored crystals studied by Smakula.3! In the crystals the rela- tively large activation energy of conductance by F-center electrons is directly related to the thermal motions of the ions (i.e., the character- istic temperature of the crystal as derived from the specific heat). Here, the large differences in the electron mobility for different halide crys- tals of the same alkali metal are wiped out by the use of a reduced tem- perature scale, Tabs/e (Fig. 14). A similar relationship using, for ex- ample, the melting point of the salt obviously (cf. Fig. 13) does not exist for A&,‘which seems nearly constant with temperature and corresponds to an electron mobility several orders of magnitude greater than that in the crystal. It appears, then, that when the thermal motion of the ioms and the electron mobility are as high as they are in the melt, the re maining individual differences in Afi may be expected to be due to other aspects of the conduction mechanism, such as the polarizability of the - anions which act as negative potential barriers, as discussed above., In both the crystals and the melts, the electrons in the sodium sys- tems exhibit higher mobility than in the potassium systems. This does not seem to be well understood at present, although a connection with the smaller interionic distances in the sodium case may exist. '~ We have seen that it is possiblé to prepare binary mixtures of many alkali metals with their molten halides in all proportions of the two components, salt and metal. Compared with the study'of the con&ucfivity of the salt-rich solutions, only a start has been made in measuring the conductivity of metal-rich solutions. Figure 15 shows the whole concen- tration range for several potassium—potassium-halide systems and demon- strates the large deviation from additive behavior. Bronstein, Dworkin, 22 UNCLASSIFIED 17 — — ORNL-LR-DWG, 56263 I T 1T 1 6 . 15 |- 14 % sl ";x 12~ § . e 1 Fig. 15. Specific Conductivity of Solutions of °~1ol— - Potassium Halides in Potassium Metal [H. R. ool i Bronstein, A. S. Dworkin, and M. A. Bredig, J. é Chem. Phys. 37, 677 (1962) (reprinted by permission |- - i of the copyright owner, the American Institute of g 7 | . Physies)]. 2 of 1 s 1 n 4% . . / 3'— ,I — / | _ i ”’,/ - " 1 1 | 1 1 1 0 40 20 30 40 50 & 7O 80 90 100 MOLE PER CENT POTASSIUM and Bredig3? found that up to 20 mole % KI the resistivity of potassium metal increased linearly with increasing salt concentration. Subsequent work>? showed that the proportionality factor & in the equation for the specific resistivity, p (uohm-cm) = N, + b, increases with the size of the anion, from 420 for KF (700°), to 740 for KBr (740°), to 920 pohm-cm for KI (700°). An appropriate explenation appears to be the increase in the cross section of the larger anions for scattering moving electrons. An alternative explanation, that of the exclusion of conducting volume by the introduction of insulator (salt) molecules, is entirely unsatisfactory: It is true that the equivalent conductivities, that is, the conductivities of 1 mole of solution of salt in metal, of KI, KBr, or KF in potassium, fall essentially on the same curve when plotted against volume fraction. However, the conductivity drops far more rapidly, initially by a factor of 5 to 10, than the exclusidn of conducting volume would require, namely, at 700°, from approximately 990,000 ohm™ cm? mole™! for pure potassium metal to 620,000 at a volume - (.& I v i 1 23 fraction of potassium of 0.95, that is, (990,000 — 620,000)/0.05 x 990,000 = 370,000/49,500 or 7.5 times faster. It is not unexpected that the macroscopic picture of exclusion of conducting volume does not work in this situation where the mean free path of the electrons is large com- pared with the excluding particle size, which makes the scattering process the all-important feature. Alkaline-Earth-Metal—-Halide Systems Phase Diagrams The older literature's> contains a rather bewildering variety of claims for the existence of extensive solubility of alkaline-earth metals in their molten halides and the existence of subhalides of the formula MX or MpX, (similar to HgpX,). The assumption of solid alkaline-earth subhalides, for example, the claim by WShler and Rodewald’?“ regarding the existence of a s0lid CaCl, based on chemical analysis of deep-red crystals prepared by fusing together calcium and CaCls, has not been borne out by later investigations of Eastman, Cubicciotti, and Thurmond©s»33 and of Bredig and Johnson.2% (Considerable solubility of the metals in their salts and some solubility of the salts in the liquid metal were found, but no indication of solid reaction products, subhalides (MK).' A more recent suggestion®7+38 of a layer~type crystal structure for a preparation pre- sumed to be "CaCl" turned out soon thereafter>? to be erroneous; the crys- tals investigated were those of the ternary compound CaCl,*CaH,, or CaClH. Only & start has asctually been made to obtain truly accurate phase diagrams of the alkaline~earth-metal-~halide systems. Schifer and Niklas4© in 1952 very briefly reported a study of the system Ba-BaCl,, which is characterized by & solubility of 15 mole % barium in molten BaCl, and of 5 mole % BaCl, in liquid barium metal at the monotectic temperature of 878°C, and by a relatively low consolute temperature of 1010°C, only 50° above the melting_point of the salt. These results, confirmed by Peterson and Hinkebein*l»4? (Fig. 16), differed greatly from those obtained ear- lier by Eastman, Cubicciotti, and Thurmond}®»35 which gave little, if any, 2 UNCLASSIFIED ORNL-LR-DWG. 70760 1050 T ! T 1T 1 | 1 T T - o THERMAL ANALYSIS L X QUENCHED SAMPLE 1000 950 3 a\; J —~— 1 o 900§ X 2 | w ° | E gs50f D = 1 < 1 s i o BOOR | - = | w { - 1 750]- atl, :~ ¥ | n.o D o O / 700 - a+p a+ly ] 1 i 1 1 ] ! 1 1 0 10 20 30 40 50 60 T0 80 90 100 BaCl,p MOLE PER CENT Ba Ba Fig. 16. Barium Metal-Barium Chloride System (J. A. Hinkebein, Ph.D. Thesis, lowa State College, 1958). indication of complete miscibility at higher temperatures. The difference is due to faults of the earlier technique in determining the temperature- concentration area of the coexistence of two liquids by analysis of quenched melts. As emphasized by Bredig and co-workersil»15:43 gng py others,4® it is in general impossible to preserve, by quenching, the equilibrium concentrastion of a liquid phase saturated with respect to another liquid phase, because of rapid precipitation and segregation of the second liquid phase. What one may obtain, instead, is a phase re- sembling the monotectic liquid in composition. Another occasional source of error is the contraction of the salt phase on freezing, which produces a sump hole and/or cracks which fill with excess liquid metal. To avoid these errors it is necessary to resort to entirely different techniques such as thermal or differential thermal analysis, that is, the taking of cooling curves, as Schi8fer and Niklas did, or to & process of separating a sample of each liquid phase at the equilibration temperature as de- scribed by Bronstein and Bredig.?7? A still different method, which relies on diffusion of the metal into the salt contained in a separate compart- ment of a capsule,%* is subject to some doubt, as the wetting character- - istics of the salt and metal and the creeping of liquid metal along the i 1 § E g b Y b 25 inside walls of the metal containers may also lead to distorted results, aside from the inconvenience of long periods of time for equilibration. Besides the measurements on the Ba-BaCl, system, the results are believed to be reasonably accurate for the Ca~CaF2 system reported by Rogers, Tomlinson, and Richardson;45 for the Ca-CaCl, system reported by Hinkebein and Peterson“l:42 and by others;%4»%% and for the higher tem- perature range of the Ca-CaBr, and Ca-Cal, reported by Staffansson, Tomlinson, and Richardson.# A low solubility of magnesium in molten MgCl, was observed, 0.90 (Zhurin*7) and 0.30 mole % (Rogers, Tomlinson, and Richardson%’), both at 800°C. A comparison (cf. Tables 1 and 4) of the solubility of the various alkaline-ecarth metals in their molten chlorides near the melting points of the salts reveals a trend somewhat similar to that found in the alkali-metal systems, that is, a rapid increase in solubility with atomic number of the metal; Mg-MgClp, < 0.5; Ca-CaClpy, 2.7; Sr-SrCl,, 5.5; Ba=BaCls, 20 mole %. In the Ca-CaF,; system a rather high miscibility, that is, a rather small temperature range (30°) for coexistence of two liquids, was found,45 with a critical solution temperature of 1322°C, almost 100° below the melting point of the salt (Fig. 17). The critical solution temperatures in the other three calcium-halide systems are very similar, yet the tem- perature range of only partial miscibility in these systems covers 500°. This, of course, is caused by the relatively much lower melting points of CaCl,, CaBr,, and Cal, (772, 742, and 780°C respectively), as compared with 1418° for CaF, with an entirely different crystal structure. From a comparison of the melting-point depréssion of CaF, produced by calcium with that produced by NaF, Rogers, Tomlinson, and Richardson“® concluded that the metal dissolved either as calcium atoms or as Ca22+ molecule ions, nemely, according to either Ca — Ca® or Ca + Ca?+ — (Cap?*, both giving a cryoscoplc number, n, of unlty, or approx1mately unity, in the Raoult—van?t Hoff formula AT = (T, - 1T) = (Rmz/m%)-nng , where N, 1s the mole fraction of calcium metal. In_both cases, solid solution was as- sumed not to oceur, and it was concluded that the reaction Ca + Cat — 2Cat did not take place. However, in this case, the AH in the formula Teble 4. Principal Data for Some Alkaline-Earth-Metal—Halide Systems , Monotectic Salt- - g;::im Sz;t Phase Comp (mole % M) Eutectic Consolute Mzgal }&2{ j (°K) ?3%) Salt-Rich M;ii-,z%l_ ?f;l)’ (moggm% ) 'J(?g%a (mogzmi . (°x) Solid Liquid ILiquid MgCl, 987 ~987 ? 0.2 ~923 ~100 923 CaFa 1691 1563 & 25.5 67 1094 98.6 1595 % 5 45 1110 caCl, 1045 1093 2.70 99,5 1033 2.0 1610 + 5 62 1110 o CeBr, 1015 1100 2.3 99.6 1000 3.02 1610+ 5 64 1110 Cals 1053 1104 3.8P 99,7 1033 2.0 7 1650 £ 5 74 1110 srcl, 1145 1112 8 5.5 ? ? ? ? ? 1044 BaCl, 1235 1163 c 15.0 95 985 ~99 1290 50 1002 a yeonsiderable metal solubility in (fluorite type of) solid is likely. See ref 46, Cconsidereble metal solubility in high-temperature crystal form of BaCl, (fluorite type ?) is likely. ¥ b - N 1 27 UNCLASSIFIED ORNL- LR-DWG. 70746 T ¥ T — 1300 S 1200 Fig. 17. Calcium Metal-Calcium Fluoride System § (P. S. Rogers, J. W. Tomlinson, and F. D. Richard- E 1100 son, p 919 in Physical Chemistry of Process Metal- & . = o DIFFERENTIAL lurgy, Interscience, New York, 1961). M 000\~ THERMAL ANALYSIS 0 SOLUBILITY 200 |- gool o 100 COMPOSITION (mole % Ca) would become approximately 10 kcal, quite different from the calorimetric value of 7.1 (Naylor48), which is hardly in doubt. It seems, then, more reasonable to assume solid solubility actually to occur both with NaF and Ce (as CaF??). Mollwo,4° in measurements on F centers, found the solu- bility of caleium in solid CaF, at 1300°C to be approximately 30 mole %, which might well be too large by as much as a factor of 10. With the value of 7.1 kecal for A , the modified formula A7 = B2 1[N, (11q) — N, (solid)] ANy - | would leave the possibility open for a cryoscopic n larger than 1, possibly 2; at low metal concentration; This cryoscopic n = 2 would be in agreement with the findings in the éorresponding chloride system, CaCl,-Ca, where Dworkin, Bronstein, and Bredig3® found n also to be only slightly less than 2. Here,'howevér, no solid solution was assumed, the crystal structure being'quite different, with sixfold rather than eight- fold coordination of snions around cations. ~In both cases the reaction would then seem to be Ca — Ca?¥ + 2¢~, as also suggested by the elec- trical conductivity (discussed below), with a possible, though much less 28 plausible, additional dissolution mechanism, Ca + Ca?t — 2¢a*, not ex- cluded at this stage, and with increasing formation of (Cap)?* at higher metal concentration. In the Sr-SrCl, system, Staffanssont* found n closely equal to 1, assuming a heat of fusion of 4.1 + 0.6 kcal. Dworkin, Bronstein, and Bredig®? questioned this latter heat value as possibly being too low. However, a recent calorimetric determination (Dworkin and Bredig®*) essen- tially confirmed the low value, which actually gives fair agreement in the rather low entropies of fusion of CaF, and SrCl, (4.2 and 3.4 e.u.) of similar, fluorite-type crystal structure. Mollwo found the solubility of strontium in solid SrCl, to be high, similar to that of calcium in CaF,.4° As in the case of CaFjy-Ca, the existence of solid solution in the fluorite type of structure could lead, then, to an interpretation of the melting- point depression in terms of n ® 2, corresponding to Sr — Srét — 2e” (and/or, less likely, Sr + Sr3t — 2sr¥), Solid solution has also been reported*®s4* for barium in the high- temperature form of BaCl,, most probably also possessing a fluorite type of structure. For $m-— T AT 43 = ~ = 1100°C (observed experimentally4l) 3 No(£) — Na(e) AN, 0.04 JA'S) AT 4 x 1100 m ~ n= X ~ N1.8% 2. RT AN, 2 x 1233 Again, a solution mechanism corresponding either to Ba — Balt + 2e™ and/or (less likely) Ba + Ba?t — 2Bat seems indicated by these experi- mental data. On the basis of activity measurements carried out by equi- libration of MgCl, with MgAl alloys of known activity, Rogers, Tomlinson, and Richardson*’ suggest for Mg-MgCl, the solution mechanism Mg + Mgt — (Mg2)2+. The precision attained does not exclude the alternate mecha- nisms Mg — Mg?* + 2¢~, or Mg + Mgt — 2Mg*, and the data appear to fit best a mixed mechanism, or in other words, an equilibrium 2~ + 2Mg*t+ = (Mgs )%+, or 2Mg* = (Mgy)?t, discussed more generally below. b2l 4 29 Electrical Conductance Dworkin, Bronstein, and Bredig®?:52:53 found it impossible to apply the method used for measuring conductivity in the alkali-metal systems to the alkaline-earth systems. While the components, molten salt or liquid metal, were found singly not to react extensively, if at all, with the synthetic sapphire cell, their mixtures did attack the latter readily. Recourse was taken to an all-molybdenum-metal apparatus embodying an as- sembly of two parallel rods as electrodes which were immersed into the melt from gbove to a varying, accurately determined depth. With pure molten salts the low resistance of this cell led to large effects due to electrochemical polarization of the electrodes and resulted in a large frequency dependence of the measured resistance which could not be ex- trapolated to infinitely high frequency with any accuracy at all. How- ever, the addition of even small amounts of metal, acting as a depolar- izer, removed the frequency dependence. Thus, it was possible to use the apparatus with the metal solutions and to calibrate it with solutions of cadmium in cadmium chloride. The specific conductivity of these solu- tions is known from the early measurements of Aten in 1910 (ref 54) and was confirmed by new measurements with the sapphire cell,3? Only results for solutions of calcium and strontium in their diha- lides, excepting the fluoride, have been reported thus far.46:30 It was concluded, as far as the limited data permit, that calcium behaves some- what like sodium in that the rate of increase in specific conductivity with metal concentration decreases (Fig. 18). The same is true for strontium in SrBr, at the relatively low temperature of 700°C, but in the dichloride, at 910°C, the behavior of strontium, giving a linear increase of conductance with metal concentration, lies somewhere between that of sodium and potaessium, with barium expected to resemble the latter. The decreasing equivalent conductance of the calcium-metal solute was attri- buted to the formation, not of diatomic molecules, Ca,, as with sodium, but of hypothetical molecule ions, (Ca2)2+, in which the trap for elec- trons is the single two-electron bond similar to that in Nap, or in Cdy?* and Hg,?t. A1 AR R et A el et e e bttt B Y b 1t 30 UNCLASSIFIED ORNL-LR-DWG.70122 3.4 ] 3.2 3.0 o ™ o > Fig. 18. Specific Conductivity of Alka- line-Earth Metal Solutions in Alkaline- Earth Halides (partially unpublished data). o b n M D © ” // Sr-SrBr, 700° SPECIFIC CONDUCTIVITY, ohm™cm’ © 7’ 7% + / +/D/ / /O)/Cu-cuz 830° _ o4 1.6 ) l l 0 1 2 3 MOLE PER CENT METAL From both the melting-point depression, above, and the decrease in metal solute conductance from 800 to 450 ohm™' cm? equiv™', it would seem to follow that at saturation (approximately 3 mole %) somewhat less than half of the dissolved calcium metal is associated to Cay?t. More information is certainly required to put these ideas about the constitution of the solutions of the alkaline-earth metals in their di- halides on a firm basis, and especially to exclude with certainty the existence of Mt alkaline-earth ions in limited concentrations. Rare-EFarth-Metal-Trihalide Systems The systems comprising rare-earth metals with their trihalides (see Table 5) represent an especially interesting and more complex group of metal-metal-halide systems, as they occupy an intermediate position. They are farther than the alkaline-earth systems along the line leading from the extreme of the alkali-metal solutions, with metallic properties and with no solid subhalide formation, to the systems of the transition L7 Table 5. Principsal Data for Some Rare-Earth-Metal—Halide Systems = MX3; mp Depression ‘Butectic "Subhalides” mp (°K) SOlubIEjty 10 Metal mp o MX3 Melting Asm(a) . Ja) () ) (Congr.:C; incongr.:i) (°k) Point (dT dn 0 n Conp Temp 1 1 1t " " n n n Mole % Temp (°x) (eu) w (wole % M) (%K) Fo.us)' Mpaz" MKy 5" MK poin" (ex) Pwre HEK LaCls = 1131 1.5 470 2.4 2.0 1099 None None None None 12.0 1293 1193 1187 LaBrs(d) 2061 (11.8)(e) 375 (2.4) 14.5 1001 None None None None 15.5 1173 1193 Lals 1052 (12.0) 490 (2.8) 8.2 1007 10231 None None 1103¢ 33 1173 1193 1173 CeCls 1090 1.7 480 2.6 9.0 1050 None None None None 9 .73 1068 ceBrp'd) 1005 (11.8) 345 (2.0) 12.0 960 None None None Nome 14 1173 1068 Cels 1034 12.0 460 2.7 8.8 o288 10044 None None 10811 32 1173 1068 1064 PrCls 1059 11l.4 565 3.0 17.0 919 None 9321 None None 19 1073 1208 prar, (3 966 (12.0) 545 3.4 16.0 852 8744 None None Nome 18 1023 1208 Pri, 1011 12.6 500 3.1 11.9 939 946C None None 10311 29 1073 1208 1180 N4AC1, 1032 - 11.6 590 3.3 14.0 913 None 9531 9751 11141 31 1173 1297 NdBrs 955 (12.0) 1297 NAI; 1060 9.2 740 3.2 26.5 764 None None None 835¢ 37 1073 1297 car, () 1204 (12 7) 555 (2.8 2) 14.0 1098 None None None 11041 14 1173 1585 YI, 1270 (10 ?) 390 (1.5 22) 12.0 1221 None None None None 15 1423 1782 (a)A. S. Dworkin and M. A, Bredig, J. Phys. Chem., March 1963. (v) {e) _ n = (ar,/aN,), (45 /RT ). (gee ree 113. e)Figures in parentheses are estimstes. Liquidus slope at infinite dilution, mostly from work by Corbett et al. ¢ 4] 32 and posttransition elements, where metal-salt miscibility in genersl be- kfifi comes more limited, where electrical conductivity of the solutions gives no indication of metallic character, and where solid "subhalides" do occur. Phase Diagrams An early phase diagram by Cubicciotti®? of the system Ce-CeCl; in~ dicating a solubility in CeClas as high as 33 mole % cerium metal was later’® found to be in error: A solubility value of 9 mole % cerium, apparently rather constent between 780 and 880°C, was obtained by Mellors and Senderoff,’® and confirmed by measurements of Bronstein, Dworkin, and Bredig53 at 855°, On the metal side, the phase diagram proposed by Mellors and Senderoff seems to require further investigation, as both the true melting point of cerium metal (817°C) and & transition occurring in the solid state (730°C) were not taken into account. The phase diagram of the corresponding lanthanum system, as deter- mined by Keneshea and Cubicciotti®? (Fig. 19), gives a solubility of the metal in the salt very similar to that of cerium. No intermediate solid phases, that is, no solid compounds containing cerium or lanthanum in a valence state lower than 3, were observed., In this respect, then, these two systems resemble the alkaline-earth systems. Also, in both cases, it appears from the melting-point depressions of the trichlorides that n, the apparent number of new particles produced on dissolving one atom of - metal, is approximately 3 or slightly less, perhaps 2.5. This signifi- cantly rules out the smaller values, n = 1 for the dissolution of these metals as atoms and n = 1.5 for Mt ions®® formed according to 2Ce + CeCl; — 3CeCl or 2Ce + 4CeCls — 3Ce(CeCl,), unless no or few (CeCl,)™ - complex ions were present in pure CeCls, so that (CeCl,)™ ions also would act as new particles, in which case n would be approximately 3. No com- pounds of monovalent rare-earth elements seem ever to have been reported, even though some of these elements have long been known in the divalent state, notably Yb2+, Eu?*, and sm?+, with 14, 7, and 6 electrons in the 4f shell, that is, with the 4f shell full, half-filled, or nearly half- ) filled respectively. We shall, however, see below that an equation Q-J ¥ 33 UNCLASSIFIED ORNL-LR-DWG. 70745 1000 T : T T I T I T L, Lo+ L, Ly S147 — fsigopo-aatk] o © 9001 e % ° 4 2z L+7-Lo(sS) | w ° i 860 E st o o 1 x 103 ohm™?! cmfl), unigue among halides, except perhaps for Ag,F. solid NdI,,gs, on the other hand, has a different crystal structure, is 34 UNCLASSIFIED ORNL~LR-DWG. 56845 500 l | T 1 1 ® ] = u.fi-t ST > ! 800}~ / ¢ — o 158' + L T‘C. \‘ * » ,.’ @ L ] ] 102 7m_ Q’.Tu’—-_— p— 680%*1 S —pr— e | - * ' ” <+ oft‘ ’l “dClu S i & €00~ Nd ! Nd Gl o | ! ' & | | i 14 | 1 ) Nd CI3 5 o 15 20 25 30 as MOLE % Nd N NdCls Fig. 20. Neodymium Metal~Neodymium Trichloride System (thermal analysis: +, equilibration: @) [L. F. Druding and J. D. Corbett, J. Am. Chem. Soc. 83, 2462 (1961) (reprinted by per- mission of the copyright owner, the American Chemical Society)]. UNCLASSIFIED ORNL- LR-DWG, 56816 T l v ‘ ¥ j’ -7 I v l 800 1 « 787 750 . J Fig. 21. Neodymium Metal-Neo- Too— dymium Triiodide System (thermal 4 analysis: +, equilibration: o) (L. F. 50— . - Druding and J. D. Corbett, J. Am. | Chem. Soc. 83, 2462 (1961)(reprinted TE.’gP' 600 " . by permission of the copyright owner, 1 (562 [Nd the American Chemical Society)]. 550|— "] L 4 . 500~ 49 . Yy A - 1." T T¥ e+ T T¥Y ¥V Nl 450 Ndll; o5 o ey i NdL, 8 16 24 3z 40 MOLE% Nd IN NdI, 1] " 35 UNCLASSIFIED ORNL-LR-DWG, 70759 I | [ I | | | I | 880 — + THERMAL ANALYSIS ;Tr 860~ o EQUILIBRATION 1 7 saol x LITERATURE oo *—*‘i T 1 ~ 820 + t o 9 / o ~— ++ 0 800}~ _ x J - 2 780 y - x £ Lk £ AN o o 760 *, — = +, - \ X)Atq-‘_—‘_-lfi-i-_—_ 740 |- + + .4 -] —*—%*fl . 720 — ° I“1'12.42 L"Iz.oo | | | | | | | | | | 2l 7000 4 8 12 16 20 24 28 32 36 ‘96 100 L0I3 MOLE PER CENT La IN LuI:5 La Fig. 22. Lonthanum Metal-Lanthanum Triiodide System [J. D. Corbett et al.,, Discussions Faraday Soc. 32, 81 (1961) (re- printed by permission of the copyright owner, the Faraday Society)]. UNCLASSIFIED ORNL-LR-DWG. 70758 A v 820 — | | | | ] [ I ! | ] 8001~ + THERMAL ANALYSIS © EQUILIBRATION 780 x LITERATURE 760} S 7404 Lt \ T 720 * + - = * / w 700t ' o — s N\ F i T, - 680 \ g / . | . - : :t:: + + —-—_‘.'+'h#—+";+"-# #H‘T"Z—";T + ¥ + 6601 +* W — 6401 Pris s Priz.0, _ | | | II I 1 | | ] | | , o 4 8 12 16 20 24 28 32 36 40 100 PrI, MOLE PER GENT Pr IN Prl, Pr Fig. 23. Praseodymi.um Metal ~Praseodymium Triiodide System {J. D. Corbett et al., Discussions Faraday Soc. 32, 81 (1961) (reprinted by permission of the copyright owner, the Faraday Society)]. 36 seltlike, nonconducting, and melts 225° below NdIj (Fig. 21). The prop- erties of the metal-like diiodides were interpreted as being due to the presence of "metallic" lattice electrons plus tripositive metal cations, M3*te~(I~),, while the saltlike NdCl, is simply NG**+(C17),. This inter- - pretation is further supported by the relative weakness of the paramag- netism of solid LaI, with a susceptibility of 220 x 10™° emu/mole, in- ‘dicating the absence of La?t ions. The solubility of the metal in the triiodide at 830°, that is, above the melting point of any diiodide, decreases from 37 mole % in NdI; (Fig. 21) to 29 in Pris (Fig. 23) and 30 in Cels, but rises again to 33 mole % in Lal; (Fig. 22). A similar deviation from the trend is shown by lanthanum in LaCls; with a solubility of 10.3 mole % metal at 900° (Fig. 19), much higher then expected in the series: 30 mole % in NaC1l; (Fig. 20), 20 in PrCls;,5t»5% and 9 in CeCl;.’>?»?® The reason for this "anomaly" with lanthanum is not quite clear; it may, however, be remembered that La?¥, like Sc3* and Ce*t, possesses, in contrast to the other rare-earth ions, a noble-gas type of electronic shell. Examination of the liquidus curves for the metal-like diiodides (Figs. 22 and 23) confirms the high degree of dissociation of molten Lal, into La?t ions: The depression of the melting point by the addition of Lal; (Fig. 22) is only of the order of 0,50° per mole % MX3, corre- sponding, with an estimated heat of fusion of 10 kcal/mole, to & cryo- scopic n, that is, number of particles per triiodide molecule dissolved, of only 0.2. For CeI,®% these numbers are larger, namely, 1.16° per mole % Cel; and n = 0.5, and are still slightly larger for PrI, (Fig. 23): 1.4° per mole % PrI; and n = 0.6, indicating less dissociation ac- cording to M** — M?* + €=, The apparent alignment of Cel, with PrI, rather than with Lal, again is of interest and not fully explained. The increase in the stability of the M2t rare-earth ion with in- creasing atomic number leads to a sharp maximum with Eu2+, which contains seven 4f electrons, that is, half of the full 4f shell, and is more stable than Eu?¥ even in aqueous systems. This stability, which is very low with gadolinium, perhaps gradually rises again in going toward the well-established Yb?¥, which contains the complete set of 14 4f electrons. 1]} e s e i h n 37 Electrical Conductivity The first measurements of the electrical conductance of solutions of cerium in cerium trichloride, by Mellors and Senderoff,58 indicating a rapid rise in conductivity up to 0.65 mole % metal followed suddenly by a very slow rise above that concentration, were shown by Bronstein, Dworkin, and Bredig®?:?3 to be in error because of reaction of the solu- tions with ceramic crucible material. (The results of a study of emf and their interpretation in terms of a Ce¥ ion® must have been equally af=- fected.) Actually, the conductivity in the Ce-CeCl; system, as in La- LaCls,”® rises with a monotonically increasing rate to a value, in the saturated metal solution, approximately five times that of the salt. In the neodymium system this ratio amounts to only 1.7, and in the praseo- dymium system it is intermediate, 2.5.%® A similar relationship holds for the equivalent conductance of these solutions (Fig. 24) obtained with UNCLASSIFIED ORNL-LR -DWG. 70125 B | T T r ] 1 | ] | i 10} | _ fi,otos . 9 |K'8 — IE , S | B e 8 i s i L Fig. 24. Specific Conductivity of Rare Earth - 7I- ; . Metal-Rare Earth Metal Chloride Systems. S 'l Lo, 910° 5 e ! ] =2 O & — O Q T . 2 o ———— m —r Pr, 830° , Nd, 856° ] ] ] | ] 5 10 15 20 25 30 MOLE PER CENT METAL 38 estimates of the equivalent volumes on the basis of the density measure- \EJ ments of Mellors and Senderoff in the CeCli-Ce system.’® The results obtained with the corresponding iodide system367 are not very different; however, the increases from the pure salts to solutions containing, for ex- ample, 10 mole % metal, correspond to factors as high as 15, 19, 6, and 2 for the La, Ce, Pr, and Nd systems respectively. These findings may be ex- plained in terms of the systematics of the stability of the divalent rare- earth cation, M*¥, as observed in the phase diagram studies above. Where no solid compound of the composition MXp; is found, or where such a solid exhibits metallic character,?®»60 conductivity of the molten solution is ] high, indicating the equilibrium M?**+ = M3* + e~ to be far to the right. Praseodymium, forming no PrX, in the chloride system, but & solid phase - PrClp, s of mixed valency which is stable in a very limited temperature range, % and forming, in the iodide system, a diiodide, not saltlike but metallic in character,®? also occupies an intermediate position be- tween cerium and neodymium as far as conductivity is concerned. UNGLASSIFIED ORNL~LR-DWG. 70423 TT 1T ] I I T 2 | T — | ;' " — i - } 10 ! - ] N - N ) E I L ! e s ~ Fig. 25. Specific Conductivity in Rare - | 2 A ,’ | Earth Metal—-Rare Earth Metal lodide Systems | & ,’goo% [R. A. Sallach et al., J. Pbys. Chem., in press ‘ £l A (1963) (reprinted by permission of the copyright 3 ! owner, the American Chemical Society)]. Ss | ol /b . o 840° 780° = / o 4 | -~ W / ) ! 3l - i ’ - 2;" ' Nd - 820° = — . ] ——././ g |1 Loy | - 0 5 10 15 20 25 30 35 (‘i') MOLE PER CENT METAL " u 39 Of special interest is the maximum in the curve of conductivity vs concentration for solutions of neodymium in NdIs at approximately 23 mole % neodymium, which corresponds to a maximum deviation from addi- tivity of the conductivity of mixtures of NdIs and NdI, at 16.7 mole % neodymium metal, corresponding to 50 mole % NAI, (Fig. 25). Bredig et al.%7 interpret this maximum in terms of an electron exchange between Nd** and Nd** for which the probability has a meximum at a concentration ratio of NdI,:NdIs = l:l, since both these salts as pure liquids conduct only ionically. The same effect is thought to cause the curvature of the conductivity~concentration curve in the NACl3-NdCl, system,50 but, probably because of the smaller polarizability of C1™ compared with I, the rate of the electron exchange i1s lower and a maximum is not produced. In any comparison of the conductance observed in the rare-earth- metal—halide systems, with both the alkaline-earth and alkali-metal systems, it is reasonable to meke the correlation for the equivalent volume, that is, a volume of solution containing the same ratio of the number of obstacles affecting electronic conduction, namely, the number of anions to the number of metal electrons. However, the number of cations in such a volume decreases in going from the alkali-metal halides to the rare-earth-metal halides by a factor of 3. Now, according to Bredig,68 the electronic conductivity, that is, the contribution of the metal in the dilute state, may be thought of as being proportional both to the total number of cations and to the number of valence electrons introduced by the addition of metai, that is, to the concentration of the two components making up a metal. The product of these two quanti- ties 1s, then, thought to be a significant parameter. It equals the product of the equivalent or mole fraction of the metal, NM’ representing the concentration of electrons from M = M?* 4+ ze™, and the reciprocal of the valency of M, 1/z, representing the concentration of MZt ions. Figures 26 and 27 show the contribution of the metal solute per equivalent of mixture, AM'NM,= Ayorn — (lf— NM)ABalt’ plotted against NM/Z, for sev- eral chloride systems. It is clear that this simple correlation can be expected to be significant only for those few MXZwM systems in which all M atoms react according to M — M?* + ze™, or, in other words, where there is no or little complication by the trapping of electrons in pairs such UNCLASSIFIED ORNL-LR-DWG, 66930 20 400 T T 1 T 18 16 300 - 14 T, S ‘T - > o 2> z 12 *5 N . 5 £ KCI-K T 10 ° 200 . $ T8 % - '4: s LoCI3-Lu < g ) 100 CeCIB-Ce i 4 SrCl,-Sr 2 LaCly-La . 2 e - CeCiz-Ce —SrC12SY, | 1 L ! 004 008 042 046 001 002 003 004 005 N, [e1x(M*] = N, /2 Fig. 26. Electronic Conductance in Some Metal-Metal Chloride Systems. Left: Equivalent Conductance of Metal vs Mole Fraction of Metal. Right: Electronic Contribution to Equivalent Conductance of Solution vs Product of Elec- tron and Cation Concentrations [M. A. Brédig, J. Chem. Phys. 37, 914 (1962) (reprinted by permission of the copy- right owner, the American Institute of Physics)]. UNCLASSIFIED ORNL-LR-DWG, 66931 400 T T 77 300 - T.Z - g 2 o 3 Q o T £ £ 5 7, 200 - 7 g < ?: KI-K Loly-La . = 100 Ce13—0e — 4l Cels-Ce Lolx-La 2 P 1 o Lt o | 1 004 008 O{2 016 001 002 003 004 Ny leXM*1= m,, /2 Fig. 27. Electrenic Conductance in Ce, La, and K Metal—Metal lodide Systems. Left: Equivalent Conductance of Metal vs Mole Fraction of Metal. Right: Electronic Con- fribution to Equivalent Conductance of Solution vs Product of Electron and Cation Concentrations (M. A. Bredig). " »n 41 as in molecules (e.g., Na,), in molecule ions (Ca,?*), or singly in lower valence states such as Pr?t, guitable systems appear to be KCl-K, SrCl,- Sr, and La-LaCl; or Ce-CeClz. Figure 26 shows La-LaCls; and Ce-CeCls to have indeed a dependence of AM-NM on NM/z rather similar to that of K-KCl. It is perhaps possible to ascribe the fact that the curve for Sr-SrCl, is considerably below the other ones to the trapping of electrons in (Sr2)2+ molecule ions. However, the attempt to interpret these finer details is not really satisfactory, as is illustrated by Fig. 27, where AM'NM for CeI3~Ce first deviates negatively and thereafter positively from AM'NM for KI-K. An explanation does not seem available, Similarly, the fact that the rise in fyy with the metal concentration is much faster for cerium in CeIs; (ref 69) than in CeCl; remains unex- plained, except that it possibly has something to do with the stability of the metallically conducting solid Cel, and the nonexistence of a similar solid CeCls. "NONMETALLIC" METAL SOLUTIONS The term used here to classify a second large group of solutions of metals, essentially the heavy metals, in their molten halides is based on the observation, made on a number of such systems, that the electrical conductivity of the melt is not increased, if altered at all, on addition of metal. In contrast to this viewpoint about the nommetallic character of these solutions, Cubicciotti’ has pointed to the metallic luster which deeply colored (black) solutions of bismuth in molten BiCl; reportedly exhibit. The significance of this apparent discrepancy is not clear, and we shall proceed with the descriptlon of this second class of solutions, bearing in mind that further study may tend to blur the sharp line of separation found to be a convenience at this time. There is, however, one more criterion, which rather sharply sepa- rates many of the systems to be discussed helow from the solutions of the electropositive metals dealt with on the preceding pages: This is the formation of electrically nonconducting, stoichiometric solid halides of the metal in a low valence state. Corbett et al. have suggested that the solubility of the posttransition metals in their molten normal halides parallels or, rather, depends on their ability to form a cationic species of lower than normal charge. We shall see that this tendency increases Table 6. Solubility of Some Transition and Posttransition Metals in Their Molten Halides Solubility Temp Solid _ Solubility Temp Solid M-mz (mOle % M) ( ° C ) "Subhali deS " Rers M wcz (mole % M) ( o c ) "S’llbhalides " RefS Ni-NicCl, 9.1 977 None 70 In-InCl InCl, In;InClg 97 "InClz 1 (InAlCL,) Ag-AgCl 0.03 490 None 71, T1~-T1CL 0.09 550 T1Cl 71 0.06 700 0.09 650 (T1A1C1,7) Zn-7nCl, 0.18 500 None 80 Sn-SnCl, 0.0032 500 71 1.64 600 Zn-7Znlis 0.28 500 None 80 Sn-SnBro 0.068 500 71 1.65 670 0d-Cacl, 14..0 550 None 54,79 Pb=-PbCl, 0.020 600 71 21..0 800 0.052 700 30.0 1000 0.123 800 ~ Ca-CdBr, 14.0 550 None 79 Pb=FbI, 0,024 440 ® 20.0 700 0.15 600 28.0 900 0.41 700 Cd~-Cal, 2.5 400 None 79 Sb~SbC13 0.018 270 80 -';2-8 ggg Sb-SbIs 1.69 200 SboTy, 80, 102 ’ 3.5 300 Heg~HeCl, 7.0 280 HgoClo 73 5.8 400 13‘8 ggg Bi-BiCl, 28.0 202 107 : ' 4640 320 "pic1" Hg-HeTp 25.0 230 HeoIo 76 28.0 550 [Bis(AlC1,)5] 112 '35.0 280 100.0 >780 Ga=-Ga,Cl, 3.7 180 GaCl, GaGaCl, 71 Bi-BiBrj; 21.0 205 - 108 (GaAlcl,) 57.0 294 "BiBr" 45,0 440 Ga-GasBr, 14.0 180 GaBr, GaGaBr, 95 100.0 5538 (G&AJ.BI‘4 ) 80 * 100.0 >458 "BiI" ¥) 43 rapidly within each group with increasing atomic weight (see Table 6). It also depends on the halide ion, and Corbett et al. attribute this latter trend to the relative tendency toward stabilization of the higher oxida- tion state by complexing with the halide ion, normally in the sequence Cl > Br > I, but reversed for Cd and Hg. Transition Metals The occurrence in this group of elements of more than one stable valence state is a well established general feature of their chemistry. However, very little seems to be known about the interaction of these metals with their molten salts. Solid salts of a monovalent state of the metal seem to be almost entirely missing, the bivalent state being the lowest in which the great majority of stable solid or liquid halides of the transition elements are known to occur. There is a notable exception [besides the compound KNi(CN)s] in the study of the nickel-nickel chloride systems by Johnson, Cubicciotti, and Kelly,’® in which a solubility as high as 9 mole % nickel metal was found at the eutectic temperature of 980°C. The authors consider the formation of Nit ions from Ni + Ni?t — 2Nit as being indicated by the freezing-point depression in conjunction with the heat of fusion, but do not rule out the possibility of an alter- nate, "physical," interpretation. However, in the latter case an asymptotic approach of the activity of NiCl, to the curve of ideal activity vs concentration near afiiClg =1 would be expected but is not found. A third possibility which was not discussed by the authors, and the odds of which are hard to evaluate in this case, is the splitting-off of two electrons, according to Ni — Nilt + 2e~, similar to the dissolution of electropositive metals in their molten halides discussed in the preceding sections. Electrical conductance meas- urements will have to be made to settle this question. Such measurements are likely to show either very little change in conductance, indicating Nit ions, or some small rise resembling that in the NACl;-NACl, system, above, entailing a similar interprétatidn in terms of electron exchange between metal ions of different valency, Ni%t and Ni?*, A large rise in conductivity, corresponding to the introduction of truly mobile electrons, 4l that is, of electrons in shallow, F-center-like traps, is considered the third, least likely possibility. Posttransition Metals Tt appears that in the first B subgroup of the periodic system, only the solubility of silver in silver chloride has been determined and found by Corbett and Winbush?* to be as low as 0.06 mole % at 700°C. The only solid subhalide known in this group is Ag,F, prepared in aqueous solution and readily decomposed at temperatures above 90°C.7? In the second B subgroup the products of interaction of mercury metal with its dihalides, that is, the "subhalides" Hg,X, of mercury, have been well known for a long time and are stable, even in contact with water. The phase diagram of the (anhydrous) HgCl,-Hg system (Fig. 28) as deter- mined by Yosim and Mayer’? shows solid HgoCl, to be stable up to the "syntectic" temperature of 525°, where it decomposes into a salt-rich melt of almost the same composition and a small amount of a liquid mercury phase containing 6.8 mole % HgCl, in solution. With the use of a calori- metric value of the heat of fusion of HgCl,,’% the depression of its UNCLASSIFIED ORNL-LR -DWG, 56700 T sk 3 2 600 550 500} 4501 Hg,Cl,+L, wezs RUFF 8 SCHNEDER 0O SMITH & MENZIES TEMPERATURE °C THIS WORK © THERMAL DATA A VISUAL DATA ngmzagm_ o 10 20 30 40 50 60 70 80 90 100 MOLE PERCENT Hg Fig. 28. Mercury—Mercuric Chloride System [S. J. Yosim and S. W. Mayer, J. Phys. Chem. 64, 909 (1960} (reprinted by permission of the copyright owner, the American Chemical Society)]. o+ " ¥ 45 freezing point on addition of mercury metal was interpreted in terms of the dissolution either as mercury atoms or as undissociated Hg,Cl, mole- cules from Hg + HgCl, — HgpClp. According to a thermodynamic analysis of the liquidus curve for HgpCl, by Yosim and Meyer,?2 Hg,Cl, is a more likely solute species than the mercury atom in that concentration region. Since HgoCl, does not impart electrical conductivity-to its solution in molten HgCl,,”” the former must be undissociated, as is the latter. The deepening in color of these solutions either with increasing temperature or mercury-metal content merits further investigation. A study of the HgI,-Hg system by PElaebon and Laude, 1929,7° seems to be too incomplete to allow definite conclusions to be drawn. The cadmium halide systems have been’” and still are under quite extensive investigation. With one exception,78 most probably in error, if for no other reason but that the reported phase diagram for CdCl,-Cd violates the phase rule, no solid subhalides Cd,X, have been observed. However, considerable solubility of metal in the molten halides exists, and it rises with temperature.79 Like that in the alkali-metal halide systems, the solubllity seems to be largest in the bromide and lowest in the iodide system. A very low value of 1.5 mole % at 600° given for the iodide system,go however, seems to be in error, éompared with 10 mole %, according to Topol and Landis.”’® The latter determination is not com- pletely satisfactory either, because of a probably too rapid rise in solubility at low temperature which produces an entirely improbable curvature in a logarithmic plot of solubility vs l/T. Consolute tempera- tures may be estimated to be very high, in the general vicinity of 1500°, and are not verifiable because of high vapor pressure. A considerable amount of data exists for the cadmium—cadmium chloride system.77»81=83 aten’4 proved the solutions to be not colloidel in nature because of the large melting-point depression of CdCl; on addition of cad- mium metal. He also showed that the electrical conductivity does not in- crease, but rather decreases slightly. Farquarson and Heymann®% found the solutions to be diamagnetic, and this was confirmed by Grjotheim et al.8% and by Nachtrieb.8® It indicated that the metal dissolves either in the form of atoms, molecules, or molecule ions such as (Cdp)3T, analogous to (Hgy)?*, and not as ca* ions.87 Cryoscopic data83:79 to- gether with a recent calorimetric heat of fusion”4 of 7.22 kcal/mole for 46 CdCl, lead to the suggestion of ideal behavior for concentrations of CdgClg in CdCl, below approximately 2 mole % CdzCl,, but increasing nega- tive deviation from Reoult's law for CdCl, with increasing metal (Cd,Cl,) solute concentration. This negative deviation (7CdClz = 0.96 or 0.99 for a cadmium atom or a CdpClp solute, respectively) at the monotectic (15 mole % Cd,Cl,) may be explsined by the substitution of large cations (cd; )2t for smaller Cd?T ions. It is analogous to similsr deviations found in systems with other large cations such as KCl-CdCl, and may or may not be thought of in terms of the formation of complex anions such as (cac1,)?~.88 perhaps to explain the asymptotic decrease in the negative deviation with decreasing metal concentration, the formation of undis- sociated species Cd,Clg end CdsX, has been suggested,’’ but does not ap- pear plausible in a highly ionized and ionizing medium such as molten - CdClp. An alternative explanation is that in small concentrations the complex anion may act as a "common ion" with respect to some of the ions of the solvent salt, CdCl,, or, in other words, that a certain concentra- tion of complex anions such as (CdCl4)2‘ is present, even in pure molten CdClp,. Solid solution formation which would invalidate any such inter- vretations of cryoscopic data is deemed unlikely. The observed negative deviation from Raoult's law would be more dif- ficult to reconcile with the “"physical" model (cadmium atoms), although Nachtrieb®® points out that the true state of affairs may be a subtle one, perhaps corresponding to a cadmium atom "solvated" by a Cd?* ion82 or to a solute species Cdp?t with an asymmetric charge distribution. Observations by Grjotheim et 81.8° and by Hertzog and Klemm®? on the electrical transport of Cd with respect to C1~ were interpreted in terms of strong interaction between Cd and Cd?*. Other effects such as those produced by the addition of a third component upon the solubility of cadmium metal may be taken as an indication that a fairly stable {(Cd,)2+ ion exists. According to Cubicciotti®® the addition of XCl to CdCl, grad- ually lowers the solubility of cadmium in the melt. Corbett et al. explained this by assuming that the excess chloride ions stabilize the higher oxidation state of cadmium by complexing.8°:88:91 However, by adding A1Cl; (a strong Lewis acid) they prepared a solid, diamagnetic compound of the formula Cd,(AlCl,), conteining all of the cadmium in oxi- dation state I,9° In the melts the increased stability of the cadmium(I) n ” LAY 47 oxidation state when the halide ion X~ (C1™, Br~, I~) is replaced by the larger AlX,~ was considered to result from the decrease in the interaction of the Cd*¥ cation with the anion, and, in the solid, from the decrease in the difference in lattice energies of the salts in the two oxidation states. The contrast between the dark~brown color of melts containing no AlClz and the light-green color of those in which ALCl,” is present was attributed to the role, in the former case, of the (highly polarizable!) uncomplexed halide ion as a bridge in a weak association between the ions in the two different oxidation states, Cd,2* and Cd?*t. Other examples of deep color are found even in aqueous systems containing Fe, Sn, Sb, or Cu in two valence states.®3 In the zinc systems the solubility of the metal does not much exceed 1 mole %.80 Because of the strong halide complexing tendency of Zn?*, it is doubtful whether addition of AlX; would raise the solubility of zinc in ZnX, to any considerable extent by allowing the lower valence state (zn,%%?) to form in greater concentrations. In the third B subgroup of the periodic system the halides of mono- valent indium and thallium have long been known, and some of those of gal- lium have also been prepared recently.?498 "Dihalides" of these elements have been recognized as ionic salts of the monovalent metal with anionic complexes of the trivalent metal, for example, Ga(I)[Ga(III)cl,].?9=101 Again, A1Cl; was employed by Corbett et al.91+24:95:98 0 aid in the com- pletion of the reaction 2Ga + Ga’*t — 3@at, that is, to convert all of the gallium into the low oxidation state (I), as in Ga(I)ALCLl,.° The solu- bility of aluminum metal in molten AlI; was reported’t to be as high as 0.3 mole % at 423°C. In the fourth group of posttransition elements, dihalides of germa- nium as well as those of tin and lead are well known stable compounds. The solubilities of the latter metals in the molten dihalides are very small, 7> 80 and it does not seem to have been shown as yet that addition of ALCl,; incréases the solubility, although this would seem to be a test for the existence of an oxidation state lower than II. All three elements of the fifth poéttransition group, arsenic, anti=- mony, and bismuth, possess a very stable oxidation state (III). The solubility of the metal in the molten trihalides rises rapidly with in- creasing atomic number, and in the bismuth systems even leads to the for- 48 mation of solid "subhalides" of the approximate stoichiometry BiX. Vapor pressure and emf studies of the SbI3-Sb system by Corbett et al.}02,103 demonstrated the existence of a catenated I,Sb-~SbI, species. Older claimsi®4 for the existence of dihalides BiX, of bismuth were not substan- tiated later,1°5 especially not in more recent phase behavior and related studies by Sokolova, Urazov, and Kuznetsov:©® and particularly by Yosim et al.t%7 10 (rigs. 29-31). The latter study is distinguished by the ex- perimental feat of determining the consolute temperature (778°) under a BiCl; pressure of 80 atm. The latest suggestion for solid bismth "mono- chloride," "BiCl," on the basis of x-ray diffraction measurements,l'! is the stoichiometric formula BigCl, or (Big)(BiCls),(BisClg), allegedly con- taining, besides anions BiCl3?*™ and Bi,Clg?™, the remarkable complex cation Big”¥ corresponding to the unique oxidation state of 5/9. On the other hand, in the solid complex compound with A1Cls, BiAlCl, (first pre- pared by Corbett and MbMullangl), Levy gz_g}.llz propose the trimeric 800 T T T I T T T T 700 m.— S '&J 500 L, E Fig. 29. Bismuth Metal~Bismuth sc( 400 - Trichloride System [S. J. Yosim et & L o al.,, J. Pbys. Chem. 63, 230 (1959) 300 LBl (reprinted by permission of the copy- = @ "z Yy E L right owner, the American Chemical 4e ' LetBl Society)] 200 e : . ’ E (BicCl) + Bi 100 = BiCly+ (BiCl) E . o Loy 0O 10 20 30 40 50 60 70 B8O 90 100 MOLE PERCENT B/ISMUTH ® VIsUAL DECANTATION & THERMAL ANALYSIS © " " 49 form of oxidation state (I), namely, triangular (Bi;)3*, which they also find in the molten state to represent the best interpretation of radial distribution functions bbtained by x-ray diffraction. Various other studies have been made on the molten solutions of bismuth metal in its trihalides.tl3 Keneshea and Cubicciotti have meas- ured the vapor pressures of BiX; over these solutions?4~16 and their i I | [ | ' | | { THERMAL ANALYSIS -0 500 |- DECANTATION— @ VISUAL—A 400 — TEMPERATURE (°C) O s I Ly+Bi OO0 c 0 p-BiBry + BiBr BiBr +BI 200 [© | O a-BiBry + BI Br ool—4— L L 1t 1 11 1 | O 10 20 3 4 50 €60 7 80 9 00 MOLE % BISMUTH Fig. 30. Bismuth Metal-Bismuth Tribromide System [S. J. Yosim et al., J. Pbhys. Chem. 66, 28 (1961) (reprinted by permission of the copy- right owner, the American Chemical Society)]. O THERMAL ANALYSIS ® DECANTATION A VISUAL 500 500 M 400 ) . w Bily+L, - : o 300 —-emaee. - - - - - - o oo & l_:_ 300: ----- q-nu‘a’nfl---3—-:6'.-:3‘)-::8‘-&--3—0w- : < Qo C o o o O : & 1 ABitHL, a ' 5 3 - 200 Bily HBIT) i (BiD+B 4200 { § | ] 1 : : |00 i 1 i 1 1 i al 1 1 |°0 0 0 20 30 40 850 60 70 B0 90 I0O MOLE % BISMUTH Fig. 31. Bismuth Metal=Bismuth Triiodide System [S. J. Yosim et al., J. Phys. Chem. 66, 28 (1961) (reprinted by permission of the copy- right owner, the American Chemical Society)]. 50 densities.”*117,118 ymile the chloride and bromide systems show large positive deviation from Raoult's law for the solvent, BiX3, when atoms are assumed to be the solute species, Bil; seems to behave ideally up to 30 mole % bismuth. The reasons for this peculiar behavior of the iodide system are not apparent. Bredigtl? interpreted the vapor pressure meas- urements in the Bi-BiBr; system in terms of an equilibrium between Bijp molecules, (Bi,)?t ions, and BiBrs, namely, 2Bi, + 2Bi’* — 3(Bi,)?%, but an equilibrium Bi, + Bi?*t = (Bi;)?* might explain these data equally well.112 At higher temperatures the equilibrium appears to shift to Bij and its dissociation products, Bi3* and electrons. A multiple mechanism for the dissolution of bismuth in BiCls was also proposed by Msyer, Yosim, and Topol,L09»120 corpettl?l suggested an interpretation of the same vapor pressure data in terms of a tetramer, (BiCl),, possibly with Bi-Bi bonds to explain the diamagnetism of solid "BiCl."22 The diamagnetism of the molten solutions was discussed by Nechtrieb.8® On the basis of emf and polarographic measurements, Topol, Yosim, and Osteryoungl?3:124 pro- pose the existence of an equilibrium nBiX ==1Bian in both the chloride and bromide systems, with n = 4 apparently given preference. An equilibrium of this nature is also indicated by the optical absorption measurements of Boston and Smith,'2° in which the ratio of two optically distinct species, probably the monomer and a polymer of Bit, was shown to depend on the bismuth-metal concentration. Thus, the majority of authors are found to prefer the assumption of polymeric subhalide formation in solu- tions of bismuth in the trihalides to that of bismuth atoms,7»117,118 The early measurements of the electrical conductance by Atent?6:127 ghowed a decrease in the equivalent conductivity of solutions with metal concen- tration, This is much better understood in terms of subhalide and Bi, molecule formation than of dissolution of bismuth atoms. The solubility of salts of several posttransition metals in the liquid metals was dealt with by Yosim and Luchsinger.128 At sufficiently high temperatures bismuth halides are completely miscible with bismuth metal; HgoCl, was found to be soluble in mercury metal to the extent of 7 mole % at 600°C. The solubility of PbCl, in lead at 1000° is given as 1 mole %, while the salts of the remaining metals are considered insoluble in their metals. There appears to be no simple relationship governing this behavior. The question of specificity was also examined, that is, the question - of whether metels dissolve salts other than their own halides. The solu- i% v »d 51 bility of a foreign salt can be explained in terms of oxidation-reduction reactions to form the halide of the solvent metal, which then dissolves in the metal. | SUMMARY Mixtures of metals with their molten halides are not colloidal sus- pensions, but true solutions. On the basis of both thermodynamic and electrical conductance measurements, it is clear that many of the mixtures are solutions in which the electrons introduced by the metal, especially in the case of electropositive metals, are in shallow traps and conse- quently mobile, as, for example, in potassium or in lanthanum systems. The mixtures also include various other types of solutions in which the electrons are partly or wholly in traps of greater depth. Such traps are believed to be diatomic metal molecules such as Na,; diatomic molecule ions such as Hgy?¥, Cdy?*, Ca,?t; more complex cations such as Bi,3*t, Bi,*¥, or perhaps even Big®¥; and simple monomeric ions of the metal in a lower than "normal' valence state such as Bi* and Nd?*. Electron exchange between cations of different valence states of the metal seems to occur and to contribute slight electron mobility. The electronic conduction process and the state of the electron in solutions, such as the alkali- metal solutions, in which the electrons are quite mobile, are not too well understood as yet. However, there are a few first attempts to describe theoretically the structure and electrical behavior of such solutions in terms of F(color)-center-like electrons, but the degree of delocalization of the mobile electron or, in other words, the number of metal cations with respect to which each elebtron‘fiay be considered gquantized needs further clarification. Solute metal atoms resembling gaseous atoms do not deserve serious consideration, as their very high polarizability must lead to strong interaction with the ions of the molten salt; that is, for- mation of F-center-like configurations, if not with the molecule ions or similarly deep traps Jjust mentioned above. Measurements of magnetic sus- ceptibility, of paramagnetic and nuclear magnetic resonance, and of the Hall effect may perhaps bring further enlightenment regarding this more detailed aspect of the shallow traps in which the electrons contributing considerable electronic conductivity are thought to be located. 8. 9. 10. 1 15. 16. 17, 18. 19. 20. 21. 22 23, 52 REFERENCES D. Cubicciotti, USAEC Document MDDC-1058 (1946). P. Drossbach, Elektrochemie Geschmolzener Salze, Springer, Bérlin, | 1938. R. Lorenz and W. Eitel, Akadem. Verl. Ges., Leipzig, 1926. W. Nernst, Theoretische Chemie, 8.-10. ed., p 468, Stuttgart, 1921. D. Cubicciotti, J. Chem. Educ. 37, 540 (1960). D. Cubicciottl, Transactions AIME, TP3589D, J. Metals 1953, 1106. F. J.)Keneshea, Jr., and D. Cubicciotti, J. Phys. Chem. 62, 843-47 (1958), - R. B. Ellis, Chem. Eng. News 38, 96 (1960). E. A. Ukshe and N. G. Bukun, Russ. Chem. Rev. (English Transl.) 30, 90 (1961). E. D. Eastmean, D. D. Cubicciotti, and C. D. Thurmond, "Temperature- Composition Diagrams of Metal—Metal Halide Systems" in The Chemistry and Metallurgy of Miscellaneous Materisls. Thermodynamics, ed. by L. L. Quill, National Nuclear Energy Series I1V-19B, p 6, McGraw- Hill, 1950, M. A. Bredig, Je. We JOhIlSOD., and W. T. Smi'th, JI'., J. Am. Chem. Soc. 77, 307 (1955). Metals Handbook, vol. 1, 8th Ed., p 25, American Society for Metals, 1961. H. ROgener, Ann. Physik 29, 386 (1937). A. S. Dworkin, H. R. Bronstein, and M. A. Bredig, J. Phys. Chem. 66, 572 (19625. M. A. Bredig and H. R. Bronstein, J. Phys. Chem. 64, 64 (1960). Je W. Johnson and M. A. Bredig, J. Phys. Chem. 62, 604 (1958). M. A. Bredig, H. R. Bronstein, and W. 7. Smith, Jr., J. Am. Chem. Soc. 77, L1454 (1955). Iu. XK. Delimarskii and B. F. Markov, Electrochemistry of Fused Salts (English translation), Sigma Press, Weshington, D.C., 1961. M. A. Bredig and J. W. Johnson, J. Phys. Chem. 64, 1899 (1960). K. Pitzer, "The Nature of Chemical Bonding in Fused Alkeli Halide- Metal Systems,” 18th International Congress of Pure and Applied Chemistry, Montreal, Canada, Aug. 6-12, 1961. K. Pitzer, J. Am. Chem. Soc. 84, 2025-28 (1962)}. Milton Blander, private communication. D. Re Stuff and G. L. Sinke, Thermodynamic Properties of the Ele- ments, American Chemical Society Monograph, 1956, “ty w et o 24 25 264 27 28, 29. 30. 31. 32. 33. 34, 35. 36. 37. 38. 39. 40, 4L 42 43 45. 46, 47 48, 53 M. Blander (ed.), Selected Topics in the Physicel Chemistry of Molten Salts (in preparation). E. Mollwo, Chem, Abstracts 30, 4757% (1936). E. Mollwo, Physik. Z. 36, 740 (1935). He. R. Bronstein and M. A. Bredig, J. Am. Chem. Soc. 80, 2077 (1958). H. R. Bronstein and M. A. Bredig, J. Phys. Chem. 65, 1220 (1961). S. A. Rice, Discussions Faraday Soc. 32, 181 (1961). Judith Brown, UCRL-9944 (Dec. 13, 1961). A. Smakula, Nachr. Ges. Wiss. GSttingen, II 1(4), 55-65 (1934). H. R. Bronstein, A. S. Dworkin, and M. A, Bredig, J. Chem. Phys. 34, 1843-44 (1961). H. R. Bronstein, A. S. Dworkin, and M. A. Bredig, J. Chem. Phys. 37, 677-78 (1962). L. Wohler and G. Rodewald, Z. Anorg. Allgem. Chem. 61, 54 (1909). D. D.)Cublcc1ott1 and C. D. Thurmond, J. Am. Chem. Soc. 7l 2149 (1949 M. A. Bredig and J. W. Johnson, unpublished data, 1951. P. Ehrlich and L. Gentsch, Naturwissenschaften 40, 460 (1953). . Ehrlich and L. Gentsch, Naturwissenschaften 41, 211-12 (1954). P. Ehrlich, B. Alt, and L. Gentsch, Z. Anorg. Allgem. Chem. 283, 58-73 (1956) H. Schéfer and A. Niklas, Angew. Chem. 64, 611 (1952). J. A. Hinkebein, Ph.D. Thesis, Iowa State College, 1958, D. Peterson and J. A. Hinkebein, J. Phys. Chem. 63, 1360-63 (1959). M. A. Bredig, Proceedings, High-Temperature Symposium, Berkeley, Calif., 1956. L. I. Staffansson, "The Physical Chemistry of Metals in Their Molten Halldes," Ph.D. Thesis, London, December 1959. P. S. Rogers, J. W. Tomlinson, and ¥. D. Richardson, p 909 in International Symposium on the Physical Chemistry of Process Metal- lurgy, Pittsburgh, 1959, Ed. by G. R. St. Plerre, Interscience, New York, 1961. A. 8. Dworkin, H. R. Bronstein, and M. A. Bredig, J. Phys. Chem. (to be published). A. I. Zhurin, Metallurg 10(4), 87 (1935). B. F. Naylor, J. Am. Chem. Soc. 67, 150 (1945). 49. 50, 51. 52 53. 54 . 55. 56. 57. 58. 59. 60. 6l. 62. 63, 64 o 65. 66 67. 68. 69. 70, 72. 54 E. Mollwo, Nachr. Ges. Wiss. GOttingen, II 1(6), 79-89 (1934). A. S. Dworkin, H. R. Bronstein, and M. A. Bredig, Discussions Faraday Soc. 32, 188 (1962). A. S. Dworkin and M. A. Bredig, J. Phys. Chem. 67, 697 (1963). H. R. Bronstein, A. S. Dworkin, and M. A, Bredig, J. Phys. Chem. 64, 1344 (1960). H. Re. Bronstein, A. S. Dworkin, and M. A. Bred_lg, J. Phys. Chem, 66, 44 (1962). A. H. W, A‘ten, D PhySiky Chem. E, 578-97 (1910)0 D. Cubicciotti, J. Am. Chem. Soc. 71, 4119 (1949). G. W. Mellors and S. Senderoff, J. Phys. Chem. 63, 1111 (1959). F. J. Keneshea, Jr., and D. Cubicciotti, J. Chem. Eng. Data 6, 507=9 (1961). Ge We Mellors and S. Senderoff, J. Phys. Chem. 64, 294 (1960). Je D. Corbett, F. Druding, and C. B, Lindahl, J. Inorg. Nucl. Chem. 17, 176-77 (1961). Je D. Corbett et al., Discussions Faraday Soc. 32, 79-83 (1961) L. F. Druding and J. D. Corbett, American Chemical Society Mbtg., New York, September, 1960. Abstracts page 21N (No. 50). L. F. Druding and J. D. Corbett, J. Am. Chem. Soc. 81, 5512 (1959). L. F.)Drudlng and J. D. Corbett, J. Am. Chem. Soc. 83, 246267 (1961 J« D. Corbett et al., J. Am. Chem. Soc.(to be published). S. .Senderoff and G. W. Mellors, J. Electrochem. 105, 224 (1958). A. S. Dworkin, H. R. Bronstein, and M. A. Bredig, J. Phys. Chem. 66, 12012 (1962). M. A. Bredig et al., "The Electrical Conductivity of Solutions of Metels in Their Molten Halides. VI. Lanthanum, Cerium, Praseo- dymium, and Neodymium in Their Molten Jodides ," J. Phys. Chem. (to be published in May or June 1963). M. A. Bredig, J. Chem. Phys. 37, 914-15 (1962). M. A. Bredig et _al., "The Electrical Conductivity of Solutions of Metals in Their Molten Halides. VI. Lenthanum, Cerium, Praseo- dymium, and Neodymium in Their Molten Iodides," J. Phys. Chem. (to be published in May or June 1963). Js W, gohnson, D. Cubicciotti, and C. M. Kelley, J. Phys. Chem. 62, 1107-9 (1958). Je D.)Corbe‘t'b and S. V. WlnbuSh, Je« Am. Chem. SocC.,. 77, 3964’—66 (1955 A. Hettich, Z. Anorg. Allgem. Chem. 167, 67 (1927). ") w2 ot 73 s 72 764 77 78. 79 . 80. 8l. 82, 83. 84 85. 86. 87. 88. 89. 0. o9l. 92, 93. %%, 95, 9. 97. 28. 99. 55 S. J. Yosim and S. W. Meyer, J. Phys. Chem. 64, 909 (1960). L. E. Topol and L. D. Ransom, J. Phys. Chem. 64, 1339 (1960). H. Foote and N. Martin, Am. Chem. J. 41, 451 (1908). H. Pe']).abon and Mne. Laude, Bull. Soc. Chim. France ser 4 45, 488-92 (1929). o Ge Ve )Hevesy and E. LOwenstein, Z. Anorg. Allgem. Chem. 187, 266~72 (1930). — (zr. G.)Urazov and A. S. Karnaukov, Dokl. Akad, Nauk SSSR 96, 535 1954 ). "‘ L. E. Topol and A. L. Laudis, J. Am. Chem. Soc. 82, 6291-93 (1960). Je D. Corbett, S. v. Winbush, and F. C. Albers, J. Am. Chem. Soc. 79, 3020-24 (1957). R. E. Hedger and H. Terrey, Trans. Faraday Soc. 32, 1614 (1936). 1?. E.)Heymann and E. Friedlénder, Z. Physik. Chem., Al48, 177-94 1930). E. He;)smann, J. Martin, and M. Mulcaby, J. Phys. Chem. 47, 473-84 (1943). ""' , J. Farquarson and E. Heymann, Trans, Faraday Soc. 31, 1004 (1935). K. Grjotheim, F. GrSonvold, and J. Krogh-Moe, J. Am. Chem. Soc. 77, 5824 (1955). N. H. Nachtrieb, J. Phys. Chem. 66, 1163-67 (1962). J. D. Corbett, Inorg. Chem. 1, 700 (1962). M. A. Bredig, J. Chem. Phys. 37, 451-52 (1962). W. Hertzog and A. Klemm, Z. Naturforsch. 15a, 1100-1107 (1961). D. Cubicciotti, J. Am. Chem. Soc. 74, 1198 (1952). Je D.)Corbett and R. K. McMullan, J. Am, Chem. Soc. 78, 2906~7 (1956). _ J. D. Corbett, W. J. Burkhard, and L. F. Druding, J. Am. Chem., Soc. 83, 76-80 (1961). N. Davidson et al., J. Am. Chem. Soc. 73, 1946 (1951). J. D. Corbett and R. K. McMullan, J. Am. Chem. Soc. 77, 4217 (1955). J. D. Corbett and A. Hershaft, J. Am. Chem. Soc. 80, 1530 (1958). R. Carlston, E. Griswold, and J. Kleinberg, J. Am. Chem. Soc. &0, 1532-35 (1958). . ' R. J. Clark, E. Griswold, and J. Kleinberg, J. Am. Chem. Soc. 80, 4764 (1958). R. K.)McMullan and J. D. Corbett, J. Am. Chem. Soc. 80, 476164 (1958). B L. A. Woodward, G. Carton, and H. L. Roberts, J. Chem. Soc. 1956, 3721-22. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109, 110. 1il. 112. 113, 114, 115. 116. 117. 118. 119, 120. 121. 122. 123. 124, 56 L. A. Woodward, G. Garton, and H. L. Roberts, J. Chem. Soc. 1956, 3723. G. Garton and H. M. Powell, J. Inorg. Nucl. Chem. 4, 84 (1957). ?. L. Bruner and J. D. Corbett, J. Inorg. Nucl. Chem. 20, 62-65 1961). Je. D. Corbett and F. C. Albers, J. Am. Chem. Soc. 82, 533 (1960). W. Herz and A. Guttmann, Z. Anorg. Allgem. Chem. 356, 422-32 (1908). B. Eggink, Z. Physik. Chem. 64, 449 (1908). M. A. Sokolova, G. Ge. Urazov, and V. G. Kurnetsov, Khim. Redkikh Elementov, Akad. Nauk SSSR, Inst. Obshch. i Neorgan. Khim. 1, 102-14 (1954)., S. J. Yosim et al., J. Phys. Chem. 63, 230 (1959). S. J. Yosim et al., J. Phys. Chem. 66, 28-31 (1962). S« W. Mayer, S. J. Yosim, and L. E. Topol, J. Phys. Chem. 64, 238-40 (1960). A. J. Dernell and S. J. Yosim, J. Phys. Chem. 63, 1813-15 (1959). A. Hershaft and J. D. Corbett, J. Chem. Phys. 36, 551-52 (1962). H. A. Levy et al., J. Phys. Chem. 64, 1959 (1960). G. Cleary and D. Cubiceiotti, J. Am. Chem. Soc. 74, 557 (1952). D. Cubicciotti, F. J. Keneshea, and C. M. Kelley, J. Phys. Chem. 62, 463-67 (1958) ?. Cu%1cc1otti and F. J. Keneshea, Jr., J. Phys. Chem. 62, 999-~1002 1958 B D. Cd?lCClottl and F. J. Keneshea, Jr., J. Phys. Chem. 63, 295 (1959 - F. J.;bneshea,dr., and D. Cubicciotti, J. Phys. Chem. 63, 1112-15 (1959 T F. J. Keneshea, Jr., and D. Cubicciotti, J. Phys. Chem. 63, 1472-74 (1959). M. A. Bredig, J. Phys. Chem. 63, 978 (1959). %. E.)T0pol S. W. Mayer, and L. D. Ransom, J. Phys. Chem. 64, 862 1960 J» D. Corbett, J. Phys. Chem. 62, 1149 (1958). Je. D. Corbett, J. Am. Chem. Soc. 60, 4757-60 (1958). L. E. Topol, S. J. Yosim, and R. A. Osteryoung, J. Phys. Chem. 63, 1511 (1961). L. E. Topol and R. A. Osteryoung, "Electromotive Force, Polaro- graphic, and Chronopotentiometric Studies in Molten Bismuth—Bismuth Tribromide Solutions," J. Phys. Chem. (to be published). =t 125. 126. 127. 128. 57 C. R. Boston and G. P. Smith, J. Phys. Chem. 66, 1178-81 (1962). A. H, W. Aten, Z.Physik. Chem. 66, 641 (1909). A. H. W. Aten, Z. Physik. Chem. 73, 624-37 (1910). Se Je Yosim and E. B. Luchsinger, Ann. N.Y. Acad. Sci. 79, 1079-87 (1960). & 4 - b 1. _2-3. beo 5-6. 7-41. 42. 43, blrs 45. 46 47, 48. 49-58. 59. 60. 61. 62. 63. 64 . 65. 66. 6'7. 68. 69. 70. 71. 72 59 INTERNAL. DISTRIBUTION Biology Library 73, Central Research Library T4 Reactor Division Library 75. ORNL — Y-12 Technical Library 76 Document Reference Section 77 . Laboratory Records Department 78. Laboratory Records, ORNL R.C. 79. C. F. Baes, Jr. 80. C. J. Barton gl. P. B. Bien - 82. R. E. Biggers 83. F. F. Blankenship 84 C. R. Boston 85. M. A. Bredig 86. J. H. Burns 87. W. L. Carter 88. G. H. Cartledge 89. G. W. Clark 0. W. E. Clark 91. E. L. Compere 9. T. F. Commolly 93. W. H. Cook 9%. F. L. Culler 95. J. E. Eorgan 96, Hs A. Friedman 97. R. A. Gilbert 98. G. Goldberg 99. Gerald Goldstein 100. EXTERNAL DISTRTBUTION ORNL-3391 UC-4 — Chemistry TID-4500 (20th ed., Rev.) H. Ce D. Ge Je S. C. A. C. H. W. A. G. R. Je C. M. Ge Je E. R. We A. J. T. D. I. Je W. Hoffman A. Horton W. James H. Jenks S. Johnson S. Kirslis E. Larson P. Litman H. Liu F. McDuffie J. McDowell S. Meyer E. Moore L. Moore E. Savolainen S. Shoup, Jr. Je. Skinner P. Smith A. Swartout H. Taylor E. Thoma C. Thurber M. Weinberg P. Young H. Davies (consultant) F. Hornig (consultant) Perlman (consultant) D. Roberts (consultant) 10l. Research and Development Division, AEC, ORO 102-685. Given distribution as shown in TID-4500 (20th ed., Rev.) under . Chemistry category (75 copies — OTS)