£S5 I LOCKHEED MARTIN ENERGY RESEA l | [T 4 445k OHaA?L?L b ORNL-35%94 Contract No. W~7405-eng-26 Reactor Chemistry Division MOLTEN~SALT SOLVENTS FOR FLUORIDE VOLATILITY PROCESSING OF ALUMINUM~-MATRIX NUCLEAR FUEL ELEMENTS R. E. Thoma B. J. Sturm E. H. Guinn AUGUST 1964 OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee operated by UNION CARBIDE CORPCRATION for the U.S5. ATOMIC ENERGY COMMISSION 1ii CONTENTS -A?D st mc‘t L . & " e ” » - . . - - . - ® Introduction .« « « o ¢ o ¢ o o o o o o Choice of Constituent Fluorides as AlFi3 Solvents Survey of Potential Solvent Systems Procedures . « & & » ¢ o o o ¢ s 5 o Materials e s & 6 s e s s e o o s e o Results and Discussion . ¢« « « o « o & AlF3 Melting Point . . « ¢« « .« + .« Systems Based on LiF-KF . . . . . . The System LiF-NaF~AlF3 « ¢« o o o The System NaF-KF-AlF3 + o o « « & The System KF-ZxrFs~=AlF3 . « o « o & Conclusions e s s s e s e e & s e v . References « o o o o o s o o o o o o & Page oW M 18] O 0 =3 10 11 12 34 MOLTEN-SALT SOLVENTS FOR FLUORIDE VOLATILITY PROCESSING OF ALUMINUM~-MATRIYX NUCLEAR FUEL ELEMENTS R. E. Thoma, B. J. Sturm, E. H. Guinn ABSTRACT The results of a search for molten-salt solvents for use in fluoride wvolatility processing of aluminum-matrixz fuel ele- ments are presented. The solubility of aluminum fluoride in various mixtures of fluorides was determined in order to esti- mate the feasibility and cost of processing methods. OSufficient data were accumulated to construct equilibrium phase diagrams of the solution systems, IiF-NaF-AlFj3, LiF-KF-AlF3, LiF~K3AlFg- MFo {(where MF, is CaFp, SrFp, or Zans, and XKF-ZrFs~Al1F3. New and revised phase diagrams were determined for the limiting binary systems of the alkali fluorides with AlFa by use of a new visual method for determining the occurrence of liquidus transitions. This method provided several advantages not available in classical methods of obtaining liquidus data. For example, it was observed for the first time that immisci- ble ligquids are formed at high temperatures in Al¥s-based systems. The temperatures at which such liguids form are, however, higher than is feasible for adoption in most current chemical technologies. Of the various materials evaluated as solvents for the volatility process, the greatest potential for application was displayed by the KF-ZrF;-AlF3 system. High solubility and good dissolution rates are afforded by the inexpensive solvent salt KzZrFs. At operating tempera- tures, approximately 600°C, the AlF3 capacity of the solvent is in excess of 25 mole %. INTRODUCTION Development of the fluoride volatility process is sought as a useful and effective alternative to conventional aqueous processing for recovery of uranium from spent nuclear fuels. The process depends on dissolving fuel elements in a suitable molten fluoride solvent by passage of HF gas followed by fluorination of the resulting melt to volatilize uranium as UFa.l The UFs is purified by selective sorption on solid NaF or other fluorides. Molten~salt Tluoride wolatility processing of nuclear rfuels offers advantages not available with other chewmical reprocessing methods: (1) the process is sjmyle,z requiring only a hydrofluorination step and a fluorination step instead of the many steps--dejacketing, acid dissolu~ tion, precipitation, filtrafiion,3m5 gte.~~characteristic of the usual aqueous procedures; (2) the uranium is recovered as UPFg, the form reguired for isolope separaticn,e and (3) disposal is made of essentially all fis- sion products as water-insoluble solids. It is also one of the Tew methods that can be used Tor processing certain ceramic fuels.{ The process has been previously applied successfully to zirconium- 1,8,9 matrix fuels. Currently, it is considersd for alwminum-matrix fuels 10,11 and because of the need because of theilr use in numercus reactors anticipated for processing large quantities of these fuels. One of the most important aspectis of volatility process development in adapting the process to a particular kind of fu=l element is the selec- tion of a suitable molten~salt solvent into which to pass the HF gas during dissolution. A preliminary swrvey of prospective AlF3 solvents reported by Boles and Thoma,l2 showed that a Bel;~-LiF solvent provides moderately good solubility for AlFs;, bul,; because of expense and toxicity, an slter- native solvent system is vpreferred. They considered KF-Zr¥, solvents to be of potential use. The preliminary data obtained at that time were too limited for a critical selection of optimal solvents for use in volatility processing. Accordingly, a more intensive search for sclvents was dniti- ated using newer methods which permit rapid accuwmlation of large numbers of liguid~solid transition data. Desirabple characteristics of the solvent for the process include the following: 1. TIow cost. 2. Tigquidus temperatures below 600°C. 3. Substantial solubility of Al¥F3 at 500 to 600°C. 4. Low vapor pressure. 5. Low viscosity. 6 . Noncorrosive with respect to the INOR-8 container. In reprocessing aluminum-matrix fuel elements, the temperature must be kept below the melting point of alumimm (600°C)13 to avoid forming liguid metal, which is corrosive to the container alloy, INOR~-8. The sus- ceptibility of INOR-8 to corrosion by liguid aluminum is offset by its excellent structural properties at high temperatures and its resistance to corrosion by fluorine and HF. It is desirable to prevent the maximum operating temperature from exceeding a limit of at least 50°C lower than the melting point of aluminum; consequently, 600°C has been tentatively chosen as the maximum temperature for the process. The economic feasibil- ity of the process is highly dependent upon the cost of the solvent. Solvents in which the saturating concentrations for AlFi3 are as low as 8 mole % may be useful if the solvent costs less than 50 cents a pound. I, however, higher-priced solvent components are required in order to obtain the desirable solvent characteristics, the economic feasibility of the process may then require higher saturating concentrations of AlFj3. Choice of Constituent Fluorides as AlF3 Solvents The cholce of possible solvent constituents can be rapidly narroved to one group: cheap fluorides which are stable in the presence of gaseous HE and fluorine. Cationic constituents should either have only one va- lence state as the flucride, or both the lower fluoride, existing during dissolution, and the higher fluoride, formed during fluorination, should be noncorrosive, possess sufficiently low vapor pressure, and have suit- able melting characteristics for use in the solvent. Previous to the ORNL work on the volatility process there was little published information concerning the attack of aluminum metal by molten fluorides, and none regarding the relationship of HF or other dissolved oxidant gases to this attack. It is known that aluminum reacts with molten alkali metal fluorides and with molten cryolite to form free alkali metal 14 The metal also reacts with molten and the appropriate fluorocaluminate. KaThFg, forming z Th-Al alloy and a potassium fluorcaluminate. Aluminum metal reacts with solid CeFs at 1000°C to form cerium metal and AlF3z. With KaTiFe¢, aluminum reacts to form KpAlFg, free titanium metal, and an opaque phase believed to be a complex of TiFg.ls Published free-energy valuest® favor the formation of TiFs3 by the reaction: Al + Zw, = 3TiFs3 + AlF3 . The rree~energy values suggest that the alkalil metal {luorides ThFgz, Cels, AlFa, and TiFz may serve as potential sclvent constituents. Mclien Snka and NizHF2 both attack aluminum.m&%&il7m19 but have disadvantages which preclude thelir use as poteptial volatility solvents. The former is very corrosive to structural metals;zo the latter presents a vapor prcblem and decomposes during fluorination. Aluminum reacts vigorously when heated 14 Adumima reacts with molten alkali with fluorides of Ni, Cao, Fe, or Os. metal fluorcoborates and fluorosilicates to form, respectively, aluminum boride and silicon or silicides. The use of fluoroborates and fluorosili- cates in the volatility process presumably should be avoided because vorides and silicides are s0 inert that they are likely to remain in the processing solvent in the form of an annoying sludge. Relatively few Tluorides possess the properties reguired for their use as major constituents of a solvent. Fluorides of nonmetals, semimetals, inert gases, and the platinum metals either have too high a vapor pressure or are too corrosive to be considered. (See Tsble 1 for boiling points.) These objections apply as well to fluorides of Cu; Mo, Ag, W, Au, Hg, Nb, Ta, V, Cr, Ma, Co, T1l, Pb, and Sn. Scarcliiy would eliminate consideration of most rare earths, all transuranium elements, also Sc, Y, Re, Hf, Te, Fr, Ra, Ac, and Pa. Fluorides of Zn, Ga, In, and Cd do not qualify because their reduetion by aluminum would form corrosive liguid metal. Uranium fluoride of natural isotopic camposition is objectionable because its use would alter the isotoplic composition of the fuel being processed. Thus the list of possible solvent constituents is therefore narrowed to the following fluorides: Li¥ BeFa Al¥3 TiFs FeFo NaF MgFos LaFs ZxFe NiF'as KF Cal» Cel's Thi'y RbF Sr¥2 CsF BaFp Most of the fluorides in this group are suitable only as minor constitu-~ ents of the solvent for the following reasons: 1. EKbF, CsF, LaF3, and ThF4 are moderately expensive. n 2. TFeFz and NiFz in high concentrations may present corrosion problems during fluorination. 3. MgFp, CaFa, SrFa, BaFz, and CeF3 have very high melting points (see Table 1). 4, TiFs in high concentrations may exert somewhat excessive vapor pressure. These compounds were, therefore, considered not as major solvent constit- uents but merely as possible additives for possibly depressing the liquidus of a promising solvent. The remaining six compounds--LiF, NaF, KF, BeFp, AlF3, and ZrFi--were given the principal consideration as solvent compo- nents. survey of Potential Solvent Systems Except for BeFp, which is too viscous, none of the promising fluoride constituents individually has the necessary low melting point (below 600°C) to be used directly as the solvent (see Table 1). A mumber of binary mix- tures of these fluorides do, however, form adeguately low melting eutectics (see Table 2). The only Al¥s binary system which provides sufficiently low melting mixtures for possible use in the process is the KF-AlF3 sys- tem; its capacity for additional AlFj3 at the process temperature is, however, limited to about 5 mole %, too low for process use. In order to find low-melting solvent systems suitable for the proc- ess, phase relationships were studied in the ternary systems formed by dissolving Al¥3 in a molten binary solvent. Little concern was given to more complex solvent systems because the study of polycomponent systems delineating the phase reactions occurring as AlF3 dissolves in such sol- vents iz too inveolved to permit adequate characterization in a reasonable length of time. In addition, the probability of diagnosing the cause of off-performance difficulties in engineering tests by identification of crystallized solids is remote for multicomponent salt systems unless de- tailed investigation of the phase behavior has been made. Accordingly, when a fourth component was considered, it was usually only as @& minor addition to a promising ternary system, e.g., A~B~AlF3, included in oxrder to lower the liguidus temperatures enough to meet process requirements. Although the fluoride velatility process is intended for reprocessing o fuels conteaining both uranium and alumimen, consideration was given only to the solubility of the resulting Al¥F3z in evaluating 2 solvent. The uranium content of these fuels, generally less than 1 at. %, yields too low a UFg concenvration in the solvent to affect the ligquidus temperature significsently. PROCEDURES 2,21 . . - - l ~a The initial phase studies of the solvent systems were performed ,._J primarily by classical procedures wiich proved to be generally inadeguate for systems containing Al¥i., Because AlFz and alkali metal fluoroaluwni- nates are frequently not microscopically distinguishable from each other, these systems were not amenable to studies employing the guenching tech- niqus. Alsc; the thermal changs at the liouidus temperature was often too small to be jesdily deteected by thermal analysis. Even when a thermal change was detected,; its interpretation, without visual observaltion or accompanying quench data, was equivoeal.. Most of these difficultlies were overcome Dy melting the mixtures in an inert-atmosphere glove box (Figs. 1 and 2) and observing phase changss through a window. GLigquidus temperaturcs were determined accurately (nsu- ally * 2°C) by noting the temperature at which the first crystals were obgerved in a cooling melt. The melts were stivred manually to prevent supercooling and to ensure uniTormity of composition and temperature. Usually aboub & mole of sall contalned in a hydrogen~fired; polished nickel crucible was used for the study. Intense illumination was provided by a Zirconarce photomicrographic lamp (Fish-Schurman Corporavion, New York). The lignt Trom this instrument overrides the near infrared hackground radiation from the melt to temperatures of aboul 1200°9C and thus facili- tates determination of liguidus under conditions where other methods would be difficult or impossible. Atmosplicre control was obtalined by evacuating the glove box to 30 u and refilling with helium purified by passage through 9 activated charcozal cooled with liquid nitrogen. 'The procsdure is ra As many as sixteen composibions in & given system can be studied in an nt 8=nhr period by sequential additions of prewsighed specimens. Since the melt may be observed as phase changes occur, the apparatus 1s also useful for obtaining interpretable thermal analysis data. To enswre accuracy, the temperature recorder was periodically standardized against LiF melt- ing at 848 * 100.22"24 The precision of temperature measurements obtainable in routine use of Chromel-Alumel thermocouples is generally believed to be *59C. Occa~ sional calibrations with pure salts have shown that the accuracy of thermal transition temperatures reported here is within the #*5°C precision limits. Correspondingly, the accuracy of visual transition data reported here is within *3°C. A similar visual procedure called "visual polythermal method"” has been used by Russian investigators,25 but thelr procedure seems inferior to that used here in that it apparently does not permit (1) agitation of the observed melt, (2) addition of salt during the run to alter the com- position, or (3) use of vacuum to control the atmosphere. A sharp increase in viscosity 1s often displayed by molbten salts as they cool to temperatures approaching the liguidus. This effect was noted in most of the systems discussed in this report and was useful in signal- ing the onset of crystallization. Such a sharp change in viscosity has been observed also by Velyukov and Sipriya26 for NaszAlFs and L13A1Fg and by 8111527 for “nCla, A sharp change in the electrical conductance at phase~-trnasition temperatures was also noted. Preliminary measurement of electrical conductance using an ohlmmeter (Model 630, Triplett Electrical Instrument Company, Bluffton, Ohio) indicated that it may also provide a procedure useful for studying AlF3 systems. MATERIALS Purity of the fluorides used in the phase studies was very important. The molten~-salt systems were studied primerily by a visual procedure which, in determination of a liquidus temperature, depended on the appearance of precipitate, oftten in such a form that 1t clouded the melt. Accordingly, any impurity which clouded the melt interfered with the study. Because they reacted to form very sparingly soluble phases, hydroxides and water vapor proved Lo e especizlly objectionable and in some cases initially veyy difficult to remove or avold in the preparation of thess [luorides. Hydroxides and moisture were additionally objectionable because they attacked the metal comtalner, contaminating the melt with highly colored nickel ion. Often such melts gave rise to lrreproducible Liquidus Tenper - atures due, presumably. Lo a progressive increase in oxide concenbration as the hydroxide reacted. ree methods were found to e useful for preparing flusrides of low oxygern content: 1. Vacuum sublimaiion: Applicable to purdifying cafimer*ial AlF3 and ZrFs, as the corrasponding oxides have exbremely low volatility. ZrFs was obtained with as 1little as 250 ppm oxygen using the apcaratus shown in Flg. 3. 4 2. Vacuunm distilletion after vrecipitation of oxide: Applicable to KF. The vapor preasure of KOM =t 850 to 1000°C¢ is high enough to preclude reduction of oxygen impurity to less than 1200 ta 1500 v Dy distillatifin. However, in moliten potassium fluoride, KOH reacts with various m;tal fluorides to prc:10¢tafe metal oxiuea as Tollows: +OXHET XKF ZMF + xKOH — 2ZMO X x/2] and purification of KF is possible by vacuum distile lation from the remaining wmolten mixture The use of Fel: and reFa to Pfflfilp¢uatu oxlde pirod iiced\crystals of KF which combained 900 ppm oxygen: the use af 2.3 mole % UF, gave a product containing only 300 PP 3. Ammonium bifluoride Tusion: Fasion of alkali metal fluorides with hydrated AlF; in the presence of molten N W2 proved to bhe useful for preparing ILi:A1Fg, NezhlFg, KallFsg. KA1F,, and CszA1Fg. Slowv cooling of the melts to promote crystal growith and selection of the better-crystellized portion served to provide additional puritication. Fusion of NHyHF,; with hy- drated AlFa formed (NH4)3§ ..... 1Fg. 1us thermal decampo- sition at 600°C in a helium stream yielded anhydrous AlFa which was comparable in purcity to the sublime=d produect. Alkalins earth fluorides were also purified by NHzIF; treatment. RESULTS AND DISCUGSION In the search for high~capacity solvent systems for use in the fluo- ride volatility process, equilibrium solubility data for aluminum fluoride in seven fluoride systems were obtained. The systems examined included LiF-KF-AlF3, K3AlFe-1iF-CaF,, K3Al1Fg-LiF-SrFy, K3AlFg~LiF-ZnF,, LiF-NaF- AlF3, NaF-KF-AlFj3, and KF-ZrF4-~AlFs. The phase diagrams constructed fram the data obtained in these examinations show that only one of these systems, KF-Zr¥Fas-AlFi;, can be expected to have practical application as a solvent. New data were obtained for the limiting binary systems LiF-AlF3, NaF-Al¥Fs, KF-AlF3, and KF-ZrF¥4. New phase diagrams of each of these systems are shown in Figs. 8 to 11. Experimental data for the systems reported here were collected simultaneously for several systems, thus making it possible to curtail the efforts on any one system as the development in another system showed promise. AlF3 Melting Point Previously reported experimental values for the melting point of Al¥3 range from 98612 to 104000.28 All of our visual observation data indicate that these values are low and that the melting point is higher than the reported sublimation temperature, 127000,29 though not as high (19200C) as was regarded by Steunenberg and ngel.BO Because high liquidus temperatures and vapdr pressures prevented visual study of mixtures con~ taining over 56 mole % AlFj3, too little data were obtained to permit a good extrapolation of its melting point. Since AlF3 is of similar struc- ture to CngBl and of comparable size relationship,32 we surmise that its equilibrium melting point at 1 atm is close to that for CrFs, 1404°C. Systems Based on LiF-KI¥ Mixtures of the lightest alksli fluorides, ILiF, NeF, and KF, are not so low melting as those obtainable with EbF or CsF; the cost of these latter two materials, however, precludes their economic use in process development. The binary mixture of the cheaper fluorides which affords the lowest~-melting solvent is the cguimolar LAF-KF eutectic wixvure which melts at 5007°C (sec Fig. 4). ‘The phase diasranm of the system LAF-XF-AlFa, consbructed on the basis of the datva shown 1n Table 3, is given in Fig. 5. Ilovariant equilib- ia are listed in Tsble 4. The diagvan shows clearly that LiF-KF mixtures cannot provide useful solvents because off the extent to which the primars; phase fields of the high-melting caonpounds K3Allg and KzLiAl¥e approach the limiting binary system L[iF-Al¥Fa3. The iternary system 1s canprised of the subsystems KE-LiF-Kz81Fs, KaAlFe~-11i2AlFg~-1aF, and KaAlF,-TiA1F5. Both of the composition seclions KallFg-LialAlFs and K3Alls-TaiT are apvarently quasi- binary. The minimum ligquidus temperature zlong the composition section L K:fi..’fi.l.st“Lj..P; is I?QOOC; \.«l-a— Jrl‘i t}A\. Seb L. L\.J.LL I /\ < §> - ,.J b = N & high liguidus proiiie for the system 1i fram possiblis use as a sclvent. 'The possibility that the liguidus Tor some 3 o Y S £ —— - L e o S e - inexpenglive Tour~-compoient cambinations mi for LiF-KF-A1F3 gsve impebtus to an investigatlon of the effect of the addi- tives Ja¥y, Srify, and “nlb's. Accordingly, an investigation wag made of thc extent to which scie of the Group 1l fluorides depressed the ternaxy lia- uidus. Resulbs of these experiments are given in Taebles 5 to 7. Ths minor benefits of adding thess components to the leitnsry mixbtures were insufflfi- cient to sugzes! thet sxiensive igvestigation of the mullicomponent systems was practical. At AlFa conceutrations sbove 50 mole % in the LiF-KF~-AlFa syatem we cbsarved imniscible liquids. Thelr phase relationships are not yet ade- quately explzined; cither twoe tiue (i.e,, igsotronic, liauids or one true » s 33-34 - liguid and one liguid rstalline phase (resoPhase ’77) ecould be present. The System LiF-Nap-AlFa An examinabtion was made of The liguldus surface of the LiF-Nab-AlFa t Al¥3 concentrations between 0 and 35 mole %. As in the TiF-KF- 1¥3 system, the composition area at wnich liguidus temperstures arc bhelow 500°C is much too small for the systen to be of practical value in the volatility process. The vhase disgram, shown in Fig. 6, is dominated by the cryolite phase NazdlFg, whieh erystallizes {ram LiF-Nap-AlFi3 ligul as o high~-melting phase Tor much of the lower AlFj3 part of the syslewm. Data obtained Por the system are given in Table &. The System NaF-KF-A1F4 Preliminary investigation of the system NaF-KF-AlFj3 made by Barton et gi.35 indicated that aluminum fluoride solubility was negligible in NaF-KF mixtures except at temperatures above the NaF-KF eutectic. This inference was corroborated by additional experiments conducted as part of the present investigation. The System KF-ZrFgz-Al1F4 Binary systems of the alkali fluorides with Zx¥s; afford low-melting solvent mixtures for the heavy-metal fluorides UFy; and ThFg and can be expected to provide useful solvents for AlF3 as well. Othef solvents would be preferred because of the high cost of ZrF, and because of the volatility of Zr¥Fgi-rich ligquids at high temperatures. Nevertheless, the liguidus temperatures at concentrations of 35 to 45 mole % ZrFs; in the KF-Z2rFs system and the availability of KpZrFe¢ as an inexpensive reagent suggested the use of the reagent in the preliminary evaluation of the aluminum solvent systecms.l2 The experimental data obtained in this inves- tigation are shown in Table 9, The phase diagram of the system is shown in Fig. 7. Crystallization reactions within the system KF-ZrF4-AlF3 have been characterized in detail except for those involving AlFj3 and ZrF,. Both of these coamponents are high melting and volatile. Their phase reac- tions are extremely difficult to exsmine at high temperatures because of this volatility end their low heat of fusion, which preclude most dynamic methods for obtaining phase data. Their crystalliization reactions in the ternary mixtures suggest that the only interaction cccurring between them at high temperatures is the formation of a eutectiec. For these reasons, we have cmitted investigation of the limiting binary system AlF3-ZrF,;. At 600°C, the maximm acceptable temperature for the process, a solvent, KF- zrFs (63-37 mole %), was found to have 15 mole % AlF3 capacity. Idguidus temperatures In the binary system KF-ZrFs exclude the use of solvents richer in X¥F. It can be seen from the phase diagram of the system KF-ZxrFs- AlF3 (Fig. 7), constructed in this study, that by a single addition of KF after partial dissolution of the fuel element the solubility of AlF3 is increased from 15 to 26 mole %. 172 Two dnmisaible ligulds or & liguid and a liquid ecrystalline phase o — o o g e B —— T i - QDA were found in the KF-Al¥a binary systam above 53 mole % Al¥s at 980°C. The two-ligquid region apparently cxtends iaho the KE-AlF3-7Z1xFs teprnasy systerm hut not Lo compositions curreally of joterest as volatllity sol- VauLe. CONCTUSTONS The resulis of the investigstions reported here togelther with the L - - results of dissolulion rate and corrosion rate tests made by Chemdcal £ Technology Trivision persomnel indicais conclusively that the syshom KE- 1 fuels. They alss shov that thic essential ceriteria necessarily imposed in sciecting a solvent systan, il.e.; maximm eguilibrium solubility, mexionug rates of dissolution, minimal rsies of container vessel corrosion, and minimus solvent costs, are not met (competitively) by sny of the other systems considersd in preliminary or cwrrent studies. Accordingly, more caniplete dava have been obtained Tor the system KP-Zrke-Al¥s thae Tor ang of the other systems reported here. It wss observed for the first tims that dvmiscible liquids are formed at high temperatures in AlFs3-based systems. The Lemperatures at which such liquids form ave; however, higher A in most current chemical technclogies. g: o = M 0 Hy ¢ w L v [e) D h 2 =3 o 5 < o }- . O £ On evaluating the merits of possible AlFa solvent silxtures with respect to phase; corrosion, snd cost data,; the binary mixture KF-Zx¥Fg by \ - - (63wd? mole %) was found to satisfy best the composit 9 eriteria. OCn the basis of this evaluation, this mixture is vecommendsed by Reactor Chamistiy and. Unemical Technology Division personnel as the most satisfactory sol- vent Tor dissolution of alumimun-based materials the fluorids volatiiity process, 13 Table 1. TFluoride Transition Temperature and Free Fnergy of Formation Compound Melting Point® Boiling Point™® "Fr, 2089k (oc) (o¢) (keal/mole )¢ HE ~83.36 19,46 64,7 LiF 848,02 1681.0 138.8 BeFg 545,02 115%9.0 207.5 BF'3 -128,72 -99,9 269.5 CFy 151.9 NaF 996, 02 1704.0 129.0 MgFa 1263.0% 2260.0 250.8 AlF3 1273.04 306.0 SiFy -90.3% ~95.5 360, 0 PFs -93.8 w84, 6 KF 857.0% 1502.0 127.4 CaFa 1418, 0% 500.0 SeFa 350.0 TiF, 283, 350.0 VF 3 1406.0% ~1400.0 254,0 Vs 19.5 4.3 CrFa 172.0 CrF; 1404, 0F 250, 0 Cr¥'s ~150.0 ~150.0 327.0 MnF'2 930.0 180.0 MnF 2 223.0 FeFs 950,08 1800.0 158.0 FeFa 1300.08 ~1300.08 219.0 Co¥'z ~1200.0 ~1725,0 147.0 CoF s 174.0 NiFa 1677.0 147.0 CuF'p 118.0 ZnFp 872.0 1500.0 164.0 GalFa ~ 239.0 GeFa 271.0 14 Table 1. (continued) -AF comoound Melting Point™ Boiling Point’’° £, 298%K = (oc) (oc) (keal/mole)® AsFa -5,0 58 189.0 Selg -34 6 45,9 221.8 ROF 798.0°% 1408.0 125.7 ST¥o 1400. 0 2410.0 276.7 YF3 380.0 I 910.0° ~900. 0% 424, 0 NbF 5 78.9 233.3 320.0 MoFs 17.5 35.0 383.0 RnF's 279.0 PAF 3 105.0 Agy 435,0 44,3 CdF, 1110.0 1748.0 153.3 In¥F3 234.0 SnFa 215,05 850. 05 147.0 Sni¥y, 237.0 SbF' 290.0 376.0 200. 0 SoF s 8.3 142.7 286.5 TeFg -37.8 38,9 292,0 CSF 682.0% 1251.0 124.5 BaF, 1290.0 2260.0 274.0 LaFs 403,0 CeFs 1460.0° 398.0 HEF, 413.0 TaFs 95.1 229.2 339.0 WFe g.2 17.0 ReFg 18.8 47,6 258.0 Oskg 3,4 47.3 199.0 PLF2 72.0 AuFs3 84,0 HeF s 645.0 647.,0 83.0 15 Table 1. (continued) b,c L 5F Compound Melti?gcfioint& Boili?gcfioint (kcii/i2§:§c T1F 327.0 655.0 60.0 TiF 159.0 POF, 824.0° 146.6 BiF3 850. 0 200. 0 BiFs 151.4 230.0 ThFy 1100.0 1680.0 454.0 UF4 1036.0 1417.0 421.5 “Landolt-Bdrnstein, Zahlenverte und Funktionen, Vol. 2, Eigenschgften der Materie in Ihren Aggregatzustidnden, Paxrt 4, "Kalorische Zuslandsgrdssen, opringer, Berlin, bth ed., 1961, P, Brewer, 'The Fusion and Vaporization Deta of the Holides," Paper 7, p 193-275 in The Chemistry and Metallurgy of Miscellaneous Materials: Thermodynamics, ed. by L. L. Ouill, MeGraw-Hill, New York, 1950; Metallurgical Laboratory Report CC-3455 (1946). A, Glassner, The Thermochemical Properties cf the Oxides, Fluorides, and Chlorides to 25000K, ANL-5750 (1957). dSublimaticn point. ®B. J. Sturm and C. W. Sheridan, Inorg. Syntheses, 7, g7 (1963). = s, g Sturm, Inorg. Chem. 1, 665 (1962). Elnreported melting points based on work at ORNL, sometimes only an approximate value based on preliminary experiments. 16 Table 2. Binary Fiuoride Systems of Potential Use as FProcess Solvents Eub»?tlb(gi?Pm%"tGJQ Components Congggfigifiégm{ag §e%bld‘Reference 706 IiF-AlF3 1405 a,b 689 TiF-Al¥ 3 32.5 2,0 685 NaF-AlF 3 45 a,b 570 KF=-AlF 3 40 a,b 370 BelF,~AlF 5 22 c 652 LiF-Na¥ 40 2,4 492 LiF-K¥ 50 a,d 71.0 Nak=-KF 60 a,d 355 LiF-Bela 52 &a,d,e 365 NaF-Bel's 55 a,d 340 NaeF-BeFa 43 a,d 323 KF-Bels 7.5 a,e 330 KF-HeFs, 59 a,e 507 LiF-Zrity 49 d,e 500 Nef-7xr¥Fs 40.5 d,e 430" KE -7k, 42t d,e 720 Na2AlFg~11 2A1Fe 60 a. 936 Na 3A1Fg~KA1F, 39 O %E. M. levin et al., Phase Diagrams for Ceramists, Am. Ceram. Soc. 1956, b e Tlfifififman5= 5 co~-Chenmical Constants of Binary bystems in Concentrated S0lutioiid 1. Interscience, New York, 196C. °R. L. Poles and K. E. Thomz, Volatility Process Phase Studies - A Survey of Molten Fluoride Solvent Mixtures Suitable for Dissolution of AlF3, ORNL TM-400 {oct. 22, 1962); {the binary eutectic composition and t ratur* were not actually determined butl were cotimated from prelim- Fo~-AlF 4 ternary sysoes R. E. Thoma, cd., Thase Liagrams of Nuclear Reactor Materials, ORNL-254& (Nov., 6, 1952). Levin, op. cit., Part 11, 193%. on current work. £y I . Values are base Table 3. LiF-KPF-A1F3 Liguid-Solid Trensition Data Second Liguidus Temperature (°C) Solidus Temperature (°C) Crystallization Temperature (°C) (Thermal Analysis) Composition (mole %) 1 ; +vg o] TiF %5 AiT Thermal Blectriecal Analysis Conductivity Visual Thermal Electrical Cbservation Analysis Conductivity 0.0 848 848 8§48 8.5 11.5 779 86.0 14.0 725 735 711 80.0 20.0 765, 5 766 75.2 24 8 784, 5 785 75.0 25.0 772 771 787 771 71.6 28.4 775 775 699 66.7 33.3 735 734 738 708 711 63.2 36.8 745 772 706 62.5 37.5 728 60,9 39.1 730.5 60.0 40.0 770 770 775 58.8 41,2 V47 747 710 57.2 42.8 g12 812 g24 57.1 42.9 786 55,6 by, L 802 802 709 52,7 47.2 860 50.0 50,0 1035 75.0 25.0 996 995 | 995 LT Table 3. (continued) conosition (aote 1) —prigude Tomerature Co) orysiallinanson Sgpidus Tememate (o LiF X5 AlF3 goervation Analysis Conductivity (Efmperature (o) Analysis Conductivity Thermel fnalysis) 57.2 42.8 570 559 54,6 45.4 569 565 50.0 50.0 642 575 42,8 57.2 710 5.9 70.6 23.5 971 714 1.1 66.7 22.2 951 20.0 60.0 20.0 917 27.3 54,5 18.2 889 33.3 50.0 16.7 §65 41,7 43,7 14,6 839.5 45.5 40,9 13.5 821 50.0 37.5 12.5 800 55,6 33.3 11.1 769 718 62.5 28.1 9.4 727 64.1 26.9 9.0 722.5 722 722 714 21.5 7.1 749.5 718 83.4 12.5 4.1 791 67.5 7.5 25.0 731.5 730 647 61.4 13.6 25.C £90 56.25 18.75 25.0 649.5 649 8T Table 3. (continued) Liguidus Temperature (°C) Second Crystallization Soiidus Temperature (OC) Temperature (°0) Thermal Electrical (Thérma} Anslysis) Analysis Conductivity Composition (mole %) TiT =5 AiTs Visual Thermal Electrical Ubservation Analysis Conductivity 50.0 25.0 25.0 696 645 45.0 30.0 25.0 727 40.9 34,1 25.0 749 749 645 37.5 37.5 25,0 767 762 645 30.0 45,0 25.0 802 778 25.0 50.0 25.0 843 779 33,3 16.7 50.0 976 25.0 25.0 50.0 893 20.0 30.0 50.0 858 16.7 33.3 50.0 824 100.0 854 852 g52 15.0 40.0 45.0 593 589 567 14.3 38.0 47.7 623 622 587 565 13.0 34.8 52,2 970 593 564 12,0 32.0 56.0 1098 592 563 15.4 30.8 53.8 1043 597 561 21.4 28.6 50.0 813 805 605 26.7 26,7 46.6 690 685 25,0 25.0 50.0 858 608 560 61 20 Table 4. Invariant Equilibria in the System LiF-KF-AlFj Comosition (nole §) Temperalure py. of pui1ibriu 50.0 50.0 492 Butectic 93.0 7.0 850 Eutectie 56.0 44,0 560 Butectic 50.4 49,6 575 Peritectic 85.5 14.5 711 Eutectic 4.0 36,0 710 Futectic 53,0 47.0 890 Peritectic (2) 56,0 19.0 25.0 648 Butectic® 33.0 42.0 25.0 778.5 Peritectic™ 28.1 62.5 9.4 722.5 Butectic’ 6.0 48.0 46.0 200 Futectic 45,5 53.0 1.5 490 Futectic %In subsystem KablFg-Td 3AlFg. bIn subsystem K3AlFg-IiF. 21 Table 5, K3AlFg-LiF-CaFs Liquid-Solid Transition Deta Comoattion (one §) Iy Thesal tesiyts Do 6 at2 Tigquidus Liquidus lization Temp. Solidus 100,0 993.5 992 992 80.0 20.0 971.5 %9 66.7 33.3 955 950 937 57.2 42.8 973 Oy 50,0 50.0 1011 95 40,0 20.0 40D.0 959. 5 905 33.3 33,3 33.3 908 875 682 28.6 42.8 28,6 869,5 858 682 25.0 50,0 25.0 839.5 20.0 40.0 40.0 961 18.2 36.4 45,4 996. 5 13.3 53,3 33.3 912 11.7 58,9 29.4 876 850 8.7 69.6 21.7 781 697 688 8.0 64.0 28.0 848 695 688 7.6 61.6 30.8 870 695 18.9 54,1 27.0 827.5 715 680 16.3 60.5 23.2 799.5 14.3 65.3 20.4 779 759 675 12,7 69.1 18.2 761 10.8 73.8 15.4 730.5 714 680 9.3 77.4 13,3 719 710 680 24,1 69.0 6,9 831 720 22.6 64.5 12,9 830.5 714 680 21.2 60.6 18.2 826 20.0 57.2 22.8 825 712 675 22 Table 6. K3AlFg~Sr¥y Idquid-Solid Transition Data - Ligquidus Liguidus lization Temp. Solidus 20,0 50.0 1048 757 60.0 40.0 270 967 66,7 33.3 913 75.0 25.0 832 80.0 20.0 74 83.3 16,7 780 778 767 7.7 76.9 15.4 139 ‘703 692 14.3 71l.4 14,3 776 705 628 20.0 66,7 13.3 829 25.0 62.5 12.5 853 848 703 695 66,7 16,7 16,7 9635.5 964 20,0 25.0 25.0 939 230 686 40.0 30.0 30.0 930 918 690 33,3 33.3 33.3 935 903 690 23 Table 7. K3AlFg-liF-/nFs Liquid-Solid Transition Data Campostton (mole B peciime T Bseond Ol Tiquidus Ligquidus lization Tenp. Solidus 1.2 87.8 722.5 722 722 11.8 84.0 4.2 720 720 655 11.3 80.6 8.1 713 712 663 0.9 77.5 11.6 705 705 668 10.4 74.6 14.9 701 701 670 9.7 69.4 20.8 690 689 670 9.0 65.0 26.0 675 675 668 563 8.5 61.0 30.5 668 668 562 8.0 57.5 34.5 G'73 562 7.5 54,4 38.1 666 7.0 50.0 43.0 667 24 Teble 8. LiF-NaF-AlF3 Liquid-Solid Transition Data Composition (mole #) pZindy T ford Gt - Tdquidus liguidus lization Tewp. Solidus 75.0 25,0 1007 1005 55.6 444 720 622 50.0 50.0 853 47.7 52.3 1030 45.4 54,6 1083 g.3 41.7 50.0 858 o174 15.4 38.4 46,2 729 660 599 21,4 35.8 42.8 623 645 600 26.7 33.3 40.0 662 638 601 35.3 22.4 35.3 636 627 603 11.1 66,7 22,2 254 785 20.0 60.0 20.0 919 Thts 092 27.3 54.5 18.2 882 627 694 33.3 50.0 16.7 849 722 694 25 Teble 9. The System KF-AlF3~ZrFs cogortsion (mje ) VI7EY Mermal alyere pred Llquidus Liguidus lization Temp. Solidus 75.0 25.0 920 * 4 920 920 75.0 25.0 935.5 %+ 4 932 932 50.0 50.0 604 t 4 600 448P, 480 45,4 54.6 580 * 6 580 440 40.0 60.0 573 + 4 33.3 66,7 650 * 4 25.0 75.0 760 * 4 50.0 50.0 648 + 3 bl bh4 11.1 975 t 3 540 40,0 40.0 20.0 1030 % 4 520 408 33.3 33.3 33.3 >1030 £ 4 585,442 410 66.7 22.2 11.1 890 * 3 499 480 70.0 10.0 20.0 920 + 3 915 72.7 9.1 18.2 938 * 3 934 72.7 9.1 18.2 942 % 3 G34 76,9 7.7 15.4 932 + 3 71.4 14,3 14.3 934 t 3 66.7 13.3 20.0 847 + 3 482 62.5 12.5 25.0 570 £ 4 62.5 20.8 16.7 649 * 4 505 485 53.6 17.9 28.6 721 * 4 46,9 15,6 37.5 743 4 743 41,7 13,9 bty b 821 t 4 818 582 438 37.5 12.5 50.0 869 £ 5 858 575 428 34.1 11.4 54.5 >1000 £ 5 precision limits on thermal transition data are approximately * 5° bMetastable transition. 26 i - . T . a c - el Table 1.0. Invariant BEguilibriz in the DYSiEfi;KFmAlfiSMZTFQr e S e [ fommi | m T Composition \naie ) Temparatuie - s o e e 7 vpe of mguilibhrium KR A1F; Jr¥s (°c) P = 86.0 14.0 765 Futectic 2.0 36.0 590 Peritectic 6C.0 40.0 445 Peritectic 58.0 42.0 430 Eutectic 45,0 55.0 440 Futeectic 63.0 15.0 22.0 490 Butectice ~55.0 ~5.0 ~40.0 400 rutectic TKE-A1F2 invariant eguilibria are given in Table 4. 27 UNCLASSIFIED ORNL-LR-DWG 20293 VIEWING PORT GLOVE PORT WITH RUBBER GLOVE —UTILITY CONDUIT . HEATED COMPARTMENT 28 UNCLASSIFIED PHOTO 63518 Fig. 2. Visual Study Apparatus with Accessories. Filgs 3. Apparatus for Vacuum Sublimation and Distillatlion. UNCL ASSIFIED PHOTO 64598 6C TEMPERATURE (°C) UNCLASSIFIED ORNL~LR—DWG 35483 t000 800 800 1 700 600 500 400 LiF 10 20 30 40 Fig. 4. 50 60 KF {(mole %) The System LiF-KF. 70 80 90 KF og¢ 31 UNCLASSIFIED ORNL-DWGE £3--16684 AlF3 ~1400 \ \rF 645 \ L!3A|F€ r 64 \ \?\X ~540 ?"')O T e 62720 ¢ 50 T o° /o KF ——— o S 7 B s e LiF 856 350 800 7507 700 OOO / \ [“492 600 700 848 £-490 SEE INSET Fig. 5. The Bystem KF-LiF-AlFj. 32 UNCI.ASSIFIED ORNL-DWG 64—-4537 Alfy TEMPERATURE IN °C COMPOSITION IN mole Y Fig. 6. The System LiF-NaF-AlFj. 33 UNCLASSIFIED DRNML-OWG 63-41685 AIF3 1400 rky 910 Fig. 7. The System KF-ZrF,-AlF,. 10. 1. 12. 13. 14. REFERENCES Chem. Tech. Div. Ann. Progr. Rept. May 31, 1963, ORNL-3452, p. Z20. R. P. Milford, Process Development in the ORNL Fluoride Volatilily Program October 1067 to Septenber 1963, ORML 1TM-717 (Oct. 25, 1963). R. E. Blanco and €. D. Watson, "Heat-¥nd Processes for Solid Fuels,” Chap. 3, p. 23-106 in Reactor Handbook, ed. by S. M. Stoller and R. B. Richards, Vol. II, Interscience, New York, 1961. C. M. Slansky, "Preparation of Fuels for Processing,” Chap. 3, p. 75~ 124 in Chemical Processing of Reactor Fuels, ed. by J. F. Flagg, Academic Press, New York, 1961. F. S. Mertin and G. L. Miles, Chemical Processing of Nuclear Fuels, Academic Press, New York, 1958, p. 99. Fundamental Nuclear Energy Research 1962 - A Special Report, USAEC (Dec. 1962}, p. 305. F. L. Culler, personal communication. Chem. Tech. Div. Ann. Progr. Rept. June 30, 1962, CRNL-3314, p. 39. "Research and Development on Nonagueous Process - Volatility Proc- esses,” Reactor Fuel Processing, 6, 19 (1963). H. F. Sawyer and P. lowenstein, "Fuel-Element Fabrication Facilities,"” Appendix B, p. 673-700 in Nuclear Reactor Fuel Elements - Metallurgy and Fabrication, ed. by A. R. Kaufmann, Interscience, New York, 1962 {See especially Table B-2, p. 690-5). ¥F. 8. Martin and G. L. Miles, Chemical Processing of Nuclear Iuels, p. 22=3, Academic Press, New York, 1958, R. L. Boles and R. E. Thoma, Volatility Process Phase Studies -~ A survey of Molten Fluoride Solvent Mixtures Suitable for Dissolution of AlF3, ORNL TM-400 (Oct. 22, 1962). M. Hamsen, Constitution of Binary Alloys, McGraw-Hill, New York, 1958. Deutschen Chemischem Gesellschaft, "Gmelins Handbuch der anorganlchen Chemie," system No. 35, Aluminum, Part A, Issue 1, 8th ed., Verlag Chemie, Berlin, 1934, p. 346-8. N. M. Volkova and G. V. Gaidukov, Izvest. Sibirsk. Ctdel. Akad. Nauk. SSSR, 1959, 43; Chem. Abstr. 53, 21333 (1959). A. Glassner, The Thermochemical Properties of the Oxides, Fluorides, and Chlorides to 25000K, ANL-5750 (1957). 17, 18. 19. 25. 24, 29. 20. 31. 32‘ [ [8 1§ MBRP Quar. Progr. Rept. Jan. 31, 1959, ORNL~2684, p. 114, B. J. Sturm, "Preparation of Inorgenic Fluorides,” p. 186-7 in Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1960, ORNL-2931. B. J. Sturm, Stannous Fluoride as a Component of Molten-Salt Reactor Fuels (unpublished work). MSRP Progr. Rept. March 1 to Aug. 31, 1961, ORNL-3215, p. 122. R. E. Thoma et al., "Molten Fluoride Mixtures as Possible Fuel Reproc- essing Solvents,” p. 257-9 in Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1963, ORNL-3417. T. B. Douglas and J. L. Dever, J. Am. Chem, Soc. 76, 4826-9 (1954). landolt-Bornstein, "Elgenschaften der Materie in IThren Aggregatzastan- den," Part 4 in Kalorische Auslandsgrossen, Springer, Berlin, 1961, p. 199, A, B. Trenwith, "Lithium Fluoride,” p. 174-7 in Mellor's Comprehensive Treatise on Inorganlc and Theoretical Chemistry, Vol. II, Suppl. II, Longmans, Green, London, 1961. A. G. Bergman, A. K. Nesterova, and N. A. Bychkova, "A Visual~Polythermal Method for the Investigation of Silicate Systems,"” Doklady Akad. Nauk. SSSR, Khimiya, 101, 483-6 (1955); Chem. Abstr. 49, 15418z (1955). M. M. Vetyukov and G. 1. Sipriya, "Viscosity of Melts of the Systems LiF-A1F3 and Na3AlFe-1isAlFg," Zhur, Prikl, Khim. 36, 1905-9 (1963). R. B. Ellis, "Surface Tension of Fused Zinc Chloride," Southeast Re- glonal Am. Chem. Socilety Meeting, Nov. 14-16, 1963, Program and Abstracts, p. 87. H. J. Emeleus, "Nonvolatility Inorganic Fluorides,” pp. 10 and 40 in Fluorine Chemistry, Vol. I, ed. by J. H. Simons, Academic Press, New York, 1950Q. L. Brever, "The Fusion and Vaporization Data of the Halides," Paper 7, p. 193-275 in The Chemistry and Metallurgy of Miscellaneous Materials -~ Thermodynamics, ed. by L. L. Quill, McGraw-Hill, New York, 1950; Metallurgical Lab. Report CC-3455 (1946). R. K. Steunenberg and R. C. Vogel, "Fluoride and Other Halide Vcla- tility Processes,” Chap. 6, p. 250-312 in Reactor Handbook, Vol. II, Fuel Reprocessing, ed. by S. M. Stoller and R. B. Richards, Interscience, New York, 1961. A. F. Wells, Structural Inorganic Chemistry, 3rd ed., Oxford, England, L. H. Ahrens, Geochim Cosmochim Acta, 2, 155-169 (1952). 33. 32. 4 i'%‘j G. W. Gray, Molecular Structure and Properties of Liquid Crystals, Academle Press; New York, 1962. G. H. Brown and W. G. Shaw, "The Mesomorphic State,” Chem. Reviews, 57, 1049-1157 (1957). T . J. Barton et al., "The System NaF-KF-AlF3,"” p. 32 in Phase Diagrams of Nuclear Reactor Materials, ed. by R. E. Thoma, ORNL-2548 (Nov. 2, 1959, 3\ INTERNAL DISTRIBUTION ORNL~3594 UC~4 — Chemistry TID~4500 (31st ed.) 1. Biology Library 78-87. E. H. Guinn 2-%4. Central Research Library 82. C. B. Guthrie 5. Reactor Division Library 89. C. A. Hortoun 6~7. ORNL — Y-12 Technical Library 90. R. W. Horton Document Reference Section 21. C. E. Larson 8-62. Laboratory Records Department 92. H. F. McDuffie 63. Laboratory Records, ORNI R.C. 93. H. G. MacPherson 64. C. J. Barton 2% . Gleb Mamantov 65. J. BE. Bigelow 95. R. P. Milford 66. C. M. Blood 96. M. J. Skinner 67. G. B. Boyd 97. G. P. Smith 68. M. A. Bredig 98-107. B. J. Sturm 69. J. C. Bresee 108. J. A. Swartout 70. W. H. Carr 109-118. R. E. Thoma 7L. W. L. Carter 119. A. M. Weinberg 72. G. I. Cathers 120. J. P. Young 73. 3. W. Clark 121. L. Brewer (consultant) 74. W. BE. Clark 122. F. Daniels (consultant) 75. F. L. Culler 123. R. W. Dayton (consultant) 76. H. A. Friedman 124, E. A. Mason {(consultant) 7. W. R. Grimes EXTERNAL DISTRIBUTION 125, Research and Development Division, AEC, ORO 126-717. Given distribution as shown in TID-4500 (31lst ed.) under Chemistry category (75 copies — OTS)