UNCLASSIFIED CHEMICAL PROBLEMS OB NON-AQUEOUS FLUID-FUEL REACTORS October 15, 1952 Authors George Scatchard Herbert M. Clark Sidney Golden Alvin Boltax Reinhardt Schuhmann, Jr. NUCLEAR ENGINEERING PROJECT MASSACHUSETTS INSTITUTE OF TECHNOLOGY . Manson Benedict, Director CONTRACT NO. AT(30-1)-1359 with U.S. ATOMIC ENERGY COMMISSION NEW YORK OPERATIONS OFFICE - - o o d L ‘- e @ . . *g s 0 . & @ * & = - 9 @ . 2 -0 . . s 9 . . . ¢ & & . ¥ .8 . . * o @ . » » & - * & -0 oes - ES " s - . 4 AEC RESEARCH AND DEVELOPMENT REPC MIT-5001 ~ CHEMISTRY RT This report was prepared as a scientific account of Govern- ment—spansored work. Neither the United States, nor the Com- mission, nor any person acting on behalf of the Commission makes any warranty or representation, express of implied, with respect to the accuracy, completeness, or usefulness of the in- formation contained in this report, or thot the use of any infor- mation, apparatus, method, oF process disclosed in this report may not infringe privately owned rights. The Commission assumes no liability with respect to the use of, or from domages resulting from the use of, any information, apparatus, method, or process disclosed in this report. Clocad, by . o e e} SO0, datey - I 2 Phetostat Charge $ , [/for Acces™Rermittees - Available fr Technice! Infornjatidy Servjce ~O. Box 1001] Odlf Ritige, Ternessee 7 ¥ T~ aas0aas - aeads _2- T TABLE OF CONTENTS Chapter I General Discussion s 800008000038 08B0000E00LEORTY Chapter II Chemical Problems Associated with Fast Fused- Salt Reactors ..eseececssssscccccssscccssccasss 1. Selection of Fused-Sall Fuel Mixture ..ssecesessccccsocses 1.1 Nuclear Requirements .ceecesesccsssssssces 1.2 Chemical Requirements ..cececesscesccscas Chemical Stabilily ..ccecesvsncsncsssses Liquidus RAnge seeesecesccsscsssscssses 2. Properties of the Fused-Salt Components .seescesssccsccse 2.1 Major Components and Neptunium and Plu‘tonilm TP PR PPN OP PO PRSPPSO S 202 Fission PrOduCtS PP 0P PSP S PO EOTINSSBES O 2.3 Chlorine Balance and Radiation Decomposition .eceeececsescesesveccsanes 2.4 Deposition of Solids from Fused-Salt MiXtureB SO O P OO EPR OO REPE SR 2.5 Container Material for Fused Salts ceeees References ..;..........O..l.ll....l.l...b...l..' Chapter III Chemistry of Liquid-Metal Fused-Salt Systems .... 1. Equilibria between Solutions in Liquid Bismuth and Fused Sa-lts P22 9SSR PPSPERLATPPBEDRIEROEEDOI PSP EOSSS 1.1 Significance of Experiments .cceceesesses 1.2 Description of Experimental System ...... 1.3 Experimental Results seeseececsscsocseces 2. Correlation of Experimental ResultsS .eeesecsessscscsvsses 2.1 Model for Equilibrium between Fused Salts and LiqU.id Metals seeesensevcsnee 2.2 Qu&ntitative Res“-llts OO PO OSDEODEPOTREEOEBOIS 2.3 Distribution of Plutonium .,.ccesccesssee 2.4 CritiCism Of MOdel AN EEENEEEEENNEENE NN & N N J ~02 2725 70 o [ X7 . T ess 0 ®8 as . . . see . * - . e ¢ o . s & - * . W » - . . - - L ] . - . £+ e » . - oy S8 § - . @ ¢ & ae L * e EeBare chaes *hey L 24 w e - a0 8 . Page No. 11 11 11 11 11 12 17 17 22 29 31 31 37 40 ¢ 3. Applications of Theory to Reactor Problems s.ceceessesss 3.1 General ..eeiersecescssasccsesassscsnscs 3.2 Separation ProcesSsesS ..eceecsccccorscces Liquid-Metal Fuel .,.eevescecvcscsnsse Fused—Salt Fuel ..cecesscsnscccsssccss 3.3 Corrosion P 0088000 BRONOOLEELERREIRLS References S 8 & & O 08 DO OO EO PO OPEO TSNP ECSOPSEESTNEETS Chapter IV Separation Processing .eeesseessscsscscssscssscs l. Introduction ..eeecececscessccsonsossessnsssscssssssarae 2. Stoichiometric Requirements eessssessesssecessassassssye 2.1 Over-All Materlals Balantes ..ceceevccee 2.2 Higher IsotOpeB .eessccssscssossscscnccs 2.3 Summary of Separation Requirements and Objectives T H BB PPRNPORORPTROOEDROEOS 2.4, Preseparaltioll ceeveesescsceccccsecscecccne 3. Unit ProCesSesS cceescrcscsccssescsccnccscsnsnnscssscsnnse 3.1 Wet Processes sceesesesssccesacsscscssss 3.2 Processes for Ligquid-Metal Fuels .eceese Liquid-Gas ..scecvscccsccnsecssssssccs - Liguid-Ligquid Separations .eeeesesssss Liquid-Solid Separations seseeasseceee 3.3 Procesges for Fused-Salt Fuels sceessnes Fuged Salt-Liquid Metal Processes .... Distillation ..eeevecsscccnccesccsnasns Volatile Filuoride Processes .ssssesces Fractional Crystallization ..ceceesces Tmmiseibility in Fused-Salt Systems .. Electrolysls seseeveensccscsssssnsscnns MetathSSis .."l....l.'..'....,.;....O. 4. Tentative Flowsheets for Processing Bi-U-Pu Fuels ...... References LN N NN N NN R N NN R A RN N N I I NN NN Page No. 60 60 S EER 69 70 70 71 71 73 73 74 75 75 77 78 78 79 80 80 80 81 81 81 8l 82 82 -4- Chapter V Suggestions for Research Program ..eeceveescesse 1. 2. 3. 8. 9. Introduction .eececescssnrassscessssscsencssssccccnsssne Pure SubSTANCeS .eeeesssesssscescsssscscscsssccssssssans 2.1 Physical Properties .seessessssccccsscasns 2.2 Chemical Properties .evssesscesscccnncss MiXtUresS cueeecccecscessvaceosevsccscescessscnnsannnases 3.1 AllOYS sesvecssaseccsscccassossccansancans 3.2 Salt MixbBures cevecececcosssscscncasanae 3.3 Metal-Salt SyStemS seeescenessscscecncass Radiation Stabllity seeveessesecsscessersacsesocnansonns KineticsS seessessassanssssoscessnsscssansccssnnsannssnss Physical Properties of Fuels Geestscssrsscesneanetsesnss Unit Processes for Separation ..eeeecscsccssscssscscsscss 7.1 Liquid-Metal ProCesSSES sesssescssveses 7.2 Extraction with Fused Salts crecesncese 7.3 Gas-Metal Separations .ceesesscescsecs 7.4 SeleCtive OXidation o9 & 089 bdP oS de oy ove e 7.5 Precipitation of Intermetallic ComPOMdS 0 0O BPHIESPO PO ENPETeSESeS 7.6 Imj-SCible Liq‘lid Metals 8% %590 0dsEee 7.7 Fractional Crystallization .eeessssses 7.8 Electrolysis, Liquid-Metal Electrodes, and Fused-Salt Electrolyltes .ceeecess 7.9 Selective Chlorination ccesceescesscss Engineering Development of Complete Separation Process . conCluSion a0 PP OO ORI T OSSOSO REED OO EREDE NS References CER B P L LR EES 4B ERDPEIRDRIDIOLIREDIREROLERETDTS -~ Acknowledgments ...'.'......I..'....IDD.Q....’.........C......I. - ;.. % ¥ \’,_"*_‘ . ] Sy ¥ Y) - "1’/ ® ¢4 é & odd w w4 ae . . - ey *e “"\ . . . . e . o+ @ - s " s o \ ae ¢ ws o . . . . s e o ooh e . = . a2 » . . . o & » r} & * s . s » . . o . s a . . . . e .s .a ¢ tsae . st e Page No. g8 88 89 89 89 89 90 91 92 92 93 93 93 93 94 94 94 94 94 9% 95 95 95 96 98 e o T -5- LIST OF FIGURES Page No. FIGURE II-1. UCIB-Na01 (4) Phase Diagram ...eeeeesecccecssoe 13 II-2O UClA-“NaCl (A) Ph&Se Diagr&m C P PSS ABBIBGLETENOLETSE 14 II-3. PbCl,-NaCl (8) Phase Diagram ....covevecncenses 16 II-4. Vapor Pressure, UClA, PbCl2 (T = %) vevennnens 19 ITI-1. Rare Earth Distribution between Liquid Bismuth and Fused Sa'l-t 22 0PGRS 0S8 sedees e Pee e eR 42 I1I=-2. Equilibrium Constant as a Function of Temperature 5P S HEEDPESINEBPIELIEO SIS SEsE S 47 III-3. Comparison of Fluorides and Chlorides with Tri-valent Uranium as the Common Basis .ceeeces 48 III-4. 1Influence of Uranium Distribution Ratio on Rare Earth Distribution Ratio ..eeeeeccenscss 54 III-5. Test of Equation (16) for Distribution of Rare Earth (R) between Fused Salt and Liguid Bismuth with and without Uranium (U) Present TS C OB PSP HIEREES PR NERRORNO LSS 55 IIT-6. Ratio of Distribution Coefficients of Pu and U between Fused Salt and Met8l .esececoncsnss 58 Log [MClm] o III-7. —fwj— vs- Log [C1,] at 1000" K .....ee.n. 61 Iv-1, Flow between Reactor and Separation Plant ..... 72 Iv-2. Multistage High-Temperature Separation FlOWSheet > 058 588 806080 P BSOS HNLEDH B GOSOE B 76 IvV-3. Separation Process for U-Bi Reactor Fuel ,..... 83 IV-4. Multistage Separation Process for U-Bi - & Reacmr Fuel .....l...DC...'.‘I...’......“.... 85 IV-5. Bismuth Refining Flowsheel .ceevvcsesescsncsssssne 86 2V2% 05 rewoRe ’TIHBIJE ]:[“élo CIT-2. II-3. II-4. I1-5. I11-6. IT-7. :II];fil. III-2. ]:[I'{B. ]:[I-fi&. LIST OF TABLES Properties of Fused-Salt Components .eeeesecsss Specific Heat Dat& LR B I BN BN BN B BE B BN AR B OB R N OBK N N BE N NN BN AR RN Properties of Fission Products in Fused Chloride Fuel S S P SO0 S SS SIS LBERESIETESRPE Vapor Pressures at 500 K vevecescnsnsnsescssss Long-Lived (T>100d) Fission Products ececesess Elements with Relatively Unstable Chlorides ... Estimated Free Energy and Enthalpy Changes for Corrosion Reactions ceeeecsscssccscsssvsses Concentration of Rare Earths and Uranium in Melts Containing Liquid Bismuth and Fused Cthride a-t’ 4-60 C L N N N I N N AN N N AN N N N N N NN NN NN N Equilibria between Alkali Metal Alloys and F‘used A]—kali Chloride Melts S oS o O ePESEFeDN FEquilibration Experiments for Plutonium and Uranium *«S b be o ée bbbl eorseerivESIOOeN Equilibrium Constants for Reaction ...ceeeeesee . . @ - N w T e . . - [ . teony - sy sssldw - ¥ sesg o fe . . > s sade LEE N X ¥ *e e .0 » . L X . * » . e e P - *teOoENS Page No. ~ 18 - 21 23 26 - 28 34 35 43 P 50 - - 57 " 64 -m : -T- S CHAPTER I. GENERAL DISCUSSION Before the development of reactors using non-aqueous fluid fuels can be advanced beyond the preliminary conceptual stage, it will be necessary to have a much broader background of knowledge of the inorganic and physical chemistry of relevant substances than is now available. This report discusses the chemical problems of these reactors and suggests a program of experimental and theoretical chemical research broad enough to serve as an adequate basis for their engineering development. The three main problems of non-aqueous fluid-fuel reactors are: 1. selection of a fuel system which meets nuclear and thermal requirements; 2. control of corrosion of structural materials; 3. development of an efficient and economical separation process, Discovery of a fuel system which can be handled as a liquid and has desirable nuclear properties is of course essential to the successful development of these reactors. The work of this project and the inves- tigation of other groups considering non-agueous reactor fuels has shown the following fuel systems to be of interest: For thermal reactors: 1. Dilute solution of uranium in bismuth; , 2. Solutions of UF4 in certaig mixed fluoridgsjsuchrasxi NaF' - BeFZ, NaF - ZrFA, Li'F - Ber, or DF - NaF, For fast reactors with fuel circulated to an external heat exchanger: Solution of UClA, UCl,, and/or PuClB in certain chlorides, notably NaCl, KC1, and/or PbCL,. For fast reactors internally cooled: Concentrated solution of uranium and/or plutonium in iron, nickel, bismth, or aluminum, The degree of solubility of reactor products , such as plutonium and individual fission products, in each of these reactor fuels is ob- viously of importance second only to that of their physical and muclear properties. ' Corrosion of structural materials by these fuel systems is probably the most difficult single factor impeding development of a workable _ ;fA V\{"E‘ S s o7 * i ve . ¥ & L L - * & - * & & (R L L X X! reactor using non-aqueous fluid fuel, Chemical research should prove helpful in suggesting materials of superior corrosion resistance, and, by seeking to understand the mechanism of corrosion, should be able to suggest fuel compositions and operating conditions to minimize the rate of corrosion, Structural materials currently receiving attention for this type of reactor include: carbon steel (for components of short service 1life), stainless steel, zirconium, tantalum, molybdenum, vanadium, graphite, and beryllium (as moderator). Separation processing for non-aqueous liquid fuels can be carried out by cooling, dissolving, and wet—-processing by a modification of ex- isting processes. However, one of the major objectives in considering liquid fuels is to obtain the substantial simplifications and savings in separation costs which appear likely through adoption of high temper- ature separation procesées conducted directly on a stream of ligquid fuel without extensive physical or chemical changes in the bulk of the fuel. Even with the limited data available, a considerable number of such processes have been proposed. The processes of most immediate promise include: l. extraction between a liquid-metal phase,of which bismuth, iron, lead, nickel, alumimum, or uranium is the principal constituent, and & fused-salt phase consisting of mixed fluorides or chlorides; 2. extraction between a uranium-rich 1iquid4métal phase and molten gilver or copper; 3. fractional distillation from a liquid fuel containing volatile constituents, notably the distillation of UCl4 from a chloride fuel or UF6 from a fluoride fuel; 4. vacuum removal of noble gases and other volatile fission products; 5. selective precipitation of intermetallic compounds from a uranium—-rich liguid phase; 6, fractional crystallization of one or more constituénts from a liquid. fuel. - A chemical research program aimed at solution of the problems of fluld-fuel reactors will of necessity be concerned with the substances and mixtures 1isbél above which have potential reactor utility, but “ohe A DNOOO . & Thmare von ® . LN X ) . o - gshould not be restiricted entirely to them. Too exclusive preoccupation with these particular materials might delay discovery of other useful substances or prevent formulation of useful generalizations vwhich would not be forthcoming without data on additional materials. The most valu- able research program will have the dual objectives of (1) obtaining de- tailed information on the chemical properties of the substances listed above, and (2) making fundamental advances in high-temperature inorganic and physical chemistry. Such a research program should make extensive use of theoretical methods to correlate and interpret experimental findings. Development of an adequate theory of the propsrties of solu- tions of liquid metals and of fused salts is one of the most challenging problems facing physical chemistry today, and would be one of the most valuable aids to the development of reactors using non-aqueous fluid fuels, This report does not pretend to cover completely the chemical problems of liquid-fuel reactors, The next three chapters discuss three aspects which have arisen in our study, which are very different in their approaches, but which are conveniently grouped together in a single report. Chapter II discusses in some detail the chemical problems which have arisen in planning a fast neutron reactor with a core of fused NaCl, PbClz, 0014'mixture and a blanket of fused UGlA, except for problems of the separation process. The particular separation process planned for this reactor is discussed in Chapter V of "Engineering Analysis of Non- Aqueous Fluid-Fuel Reactors"; more general matters related to separation appear in Chapter IV of this report. The chemical problems of the thermal neutron reactor with liquid-metal fuel are somewhat simpler. They are discussed in Chapter IV of the engineering anslysis report, in the report "Metallurgical Problems of Non-Aqueocus Fluid-Fuel Reactors," and in Chapter IV of this report. Chapter III gives & detailed discussion of the experiments of Barels at Brookhaven on the distribution of various metals between liquid bis- muth and fused-salt solutions, and of other related experiments. It presents a quantitative explanation of these results in terms of a rather ‘simple, and to us plausible, model., It also discusses some of the re- - lations of these experiments to reactor technology and to separation processes. | e 27‘25 - 09 "‘“|| |m| B e A il -10- “ A, Chapter IV gives a general discussion of non-aqueous, high tempera- ture separation processes, and includes two tentatlve flowsheets for tresting liquid bismuth fuels, Finally, Chapter V gives recommendations for research related to the development of non-aqueous liquid-fuel reactors, Although it mekes no pretence to being complete, the coverage is fairly general for the more fundamentel chemicel research. Since suggestions for developmental research related to corrosion and fo ehgineering are given in the report, "Engineering analysis of Non-Aqueous Fluid-Fuel Reactors® (MIT-5002), the proposals for developmental research in this report are limited to separation processes. . 272§ 30 ¢ e Sse L -11- CHAPTER II. CHEMICAL PROBLEMS ASSOCIATED WITH FAST FUSED-SALT REACTORS H. M, Clark 1, SELECTION OF FUSED-SALT FUEL MIXTURE 1.1 NUCLEAR REQUIREMENTS For faest reactors the choice of salts containing fissionable and non—-fissionable elementg is limited to those in which the non-fissionable elements have low slowing-down power and low cross secfiions for absorp- tion and inelastic scattering of fast neutrons. As a very rough guide for setting a limit to the allowsble amounts of moderating, lov mass— number diluents, Brooks (1) suggested that the ratio of the mumber of atoms of diluent to the number of atoms of fuel (U-235) should be less than half the atomic weight of the diluent. Compounds containing hydrogen, carbon, nitrogen, oxygen, and fluorine‘are'generally unsatisfactory because of the moderating effect of these elements. Of the halides, only chlorides appear to be satis- factory., In the case of chlorides, the moderation is less than for fluorides, end the capture of fast neutrons is less than for bromides end iodides. Sulfides and phosphides may also be satisfactory muclearwise, The chlorides of sodium and lead are considered to be satisfactory as dilu- ents. Sodium chloride appears to be somewhat better than potassium chloride, | 1.2 CHEMICAL REQUIREMENTS CHEMICAL STABILITY. — Salts which are suitable muclearwise are further restricted to those which are thermodynamically stable at the operating temperature of the reactor. This temperature may be of the order of 700°-800° C. Both the trichloride and the tetrachloride of uranium and the trichloride of plutonium are stable. No consideration was given to the stability of salts other than chlorides after chlorides vwere selected for nuclear reasons. Similarly, the chemistry of plutonium chloride fuel mixtures was not investigated - s e ? * - 8 e 4§ SO0 ¢ sV - . +* ¥ - . . 4 LI e - . . @ * -* ’» o8 . s e @ she . - . * * » * & . " & - - - L. . Be aw - -12- “ | after it became spparent that a fast fused-chloride breeder does not appear to be economically attractive. As & general rule complex substances containing covalent chemical bonds are unsatisfactory because they are much more readily decomposed by ionizing radiation thsn are ionic materials and metals. LIQUIDUS RANGE. - It is reasonable to expect that the maintenance of mechenicel pumps, corrosion of the reactor shell, etc., will be less severe the lower the operating temperature. The lowest temperature to which the fused salt can be cooled is obviously slightly above the freezing point of the galt. By adding other salts to depress the freezing point, the liquid range of a salt may often be extended by several hundred degrees., Tabulations of low-melting binary eutectics have been reported by Grimes and Hill (2) and by Shaw and Boulger (3). Both the trichloride and the tetrachloride of uranium have unfavor- ably high melting points. The melting points are 835° C. and 590° C., respectively., Information on binary mixtures of uranium trichloride or tetrachloride and other salts is contained almost entirely in one report of work conducted at Brown University (4) in 1943. Data for fluoride fuel mixtures of interest for thermal reactors, such as the ARE (5), (6), (7), have been helpful in the prediction of physical properties, e.g., viscosities, of analogous chloride mixtures. The systems UClB-KCI and UClAfK01 were considered inferior to the corresponding binary systems containing NaCl because of the formation of compounds of the type 2KCl-U013 and 2KCl-~UCl4 having congruent melting points at 625° C, and 650° C., respectively. In contrast, the phase disgram for the system NaCl—UGl3 consists of & single eutectic at 33 mole per cent UCl3 with a melting point of 520° C. es shown in Figure II-1. For the system NaCl-UCl4 the compound 2Na.Cl--UCl4 having an.incongruent melting point at 4,30o C. is'formed. The phase disgram showing the eutectic at 50 moleper cent UCl4 and & temperature of 370° ¢, is represented in Figure IT-Z. The melting point of the eutectic mixture of NaCl and UCl3 wag con- sidered undesirably high for design purposes, although it is better nuclearwise because of the lower C1/U ratio. Preliminary design for the fast fused-salt converter was based on the lower-melting eutectic mixture s 12— L] Temperature Degrees C -13- 100 FIGURE II-1, 1CO0 UGlB—NaCl (4) Phase Diagram 900 800 ,/ 700 \ 600 / 500 400 300 0 20 40 60 80 Mole Percent UCI3 -14- ~‘~“’ FIGURE II1-2. 1000 . ©Ucl 4-Nacl (4) Phase Diagram 300 800 700 - 600 Temperature Degrees C 500 / 400 v 300 . 0 - 20 40 60 Mole Percent UCl, 125 34 o B : S - -15- of the NaCl—-UCl4 system. For the final design, an even lower melting mixture was sought. With respect to both nuclear and chemicel (stability and processing) requirements, PbCl2 was considered to be a satisfactory third component. The phase diagram for the NaCl~--PbCl2 gsystem is shown in Figure II-3. The data are those of Treis (8). Unfortunately, no data are availablg for either the ternary system or the Pb012~-U.,Cl4 system., Estimates of the expected melting points for the ternary system for various compositions were obtained by assuming ideal melting point lovwerings for the two fiseudo—binary systems: (UClAfNaCI, eutectic) + PbCl, and (PbClz—NaCl, eutectic) + UClA. eutectic melting points were: 320° C. for System A having the composi- tion PbCL,, 0.33 moles; NaCl, 0.33 moles; UCL, O .33 moles; and 250° C. for System B with the composition PbClz, 0.43 moles; NaCl, O.17 moles, UClA, 0.40 moles, According to data supplied by A, R. Kaufmam (3), there are three breaks in the cooling curve for System A at about 524° C.s 356° C., and 290° C. For System B there are two breaks; one at_315° C. and the other at 295o C. System B was considered to be sufficiently close to the ternary eutectic and was chosen as the fuel The two predicted ternary mixture for the fixed design of the fast, fused-salt reactor, Consideration was given to the possible application of system UClB- UClL-as a fuel mixture. This system is elso of interest in connection with the PbClz, NaCl, UClA‘mixture because of the possibility (a) that UCl3 may be formed in some way during Ope?ation of the reactor, or (b) that the addition of UGl3 masy be of use In & precipitation-type separa- tion process. Kraus (4) found a nearly linear liquidus curve up to 20 mole per cent UCl,, the same 1iquidus temperature, 565° C., for 30 per cent as for 20 per cent, and no evidence of a esutectic halt in this range. Butler (10), in the Ames wvork discussed on page 488 of Katz and Rabino- " witch (11), agrees with this, ‘Kraus also carried one 20 per cent solution to high temperatures and found a bresk in the cooling curve at 830o C. He says, therefore, "The two compounds are only slightly soluble in each other in the liquid phase." Calkins (12) found bresks at 838° c. and 545 C. for 89% U013, and says, "The UClBAUGl phase diagram might be quite similar to the UBrB--UBr4 diagram.® Calkins and Nottorf (12) found that the latter has.a eutectic at 24% UBrB, teined with samples on the UBrh side of the eutectic were mostly of a that "cooling curves ob- 3 ‘:J" 2725 15 L 3 I ‘ it « Q &‘l"i. T NERS - .' Aamm swan EerERR 3 sroe ‘ra Temperoture Degrees C -16- o | FIGURE I1-3. 1000 PbCl,-NaCl (8) Phase Diagram 900 800 700 \ 600 N 500 400 %6 . ® 40 60 ‘Mole Percent Pb 012 m -17- poorly defined solid solution type," and that "for this portion of the curve, the bresks in the cooling curve were not sufficiently sharp to permit accurate determination of the solidus line." This corresponds to the region investigated by Kraus for the UCl.B-UCl4 solution formation may account for the absence of a eutectic halt in system, and solid this work., 2. PROPERTIES QF THE FUSED-SALT COMPONENTS 2.1 MAJOR COMPONENTS AND NEPTUNIUM AND PLUTONIUM Values for the melting points, boiling points, vapor pressures, heats of formation, and free energies of formation listed in Table II-1 were teken from Brewer et al (13), and Katz and Rabinowitch (11) for the uranium chlorides, from Quill (14) for sodium and lead chloride, and from Seaborg, Katz, and Manning (15) for neptunium and plutonium chlo- rides, Values in parentheses were considered approximate in the original source; values in brackets were approximated in this work from informs- tion and methods described in "Chemistry and Metallurgy of Miscellaneous Materials: Thermodynamics" (14). Data for UCl, have been included in 3 the table, since reference to the properties of UCl_ is made in other gections of the report. ’ The vapor pressures are for the pure substances as liquids at 1000° K. For those substances which melt above 1000° K., the vepor pressures given are for the supercooled liquid at 1000° K. These were estimated by extrapolation of vapor pressure data for the liquid at higher temperatures. The variation of vapor pressure with temperature for the two major volatile components, UCl4 and PbClz, is shown in Figure II-4. As a reference temperature for stability considerations 1000° K. (727° C.) wes chosen. At this temperature plutonium trichloride is the stable chloride in the liquid state., For neptunium, which has chemical properties intermediate betwsen those of uranium and plutonium, both the trichloride and the tetrachloride are stable. Unless reducing conditions develop during the operation of the reactor, the uranium tetrachloride ghould remain unchanged. 21153 i ...... ooooooo ...... TABLE II-1. _Properties of Fused-Salt Components Normal Vapor s | sl | o, xou /e rresere, Chloride % % keal/mole 298°k | 500° K 1000° K 1000° K NaCl 800 1465 - 98.3 -91.8 | -87.4 - 76.3 | 8.2 x107° PbCL, 498 954 - 85.7 -9 | -67.8 ~52.9 | 5.3x107° ucy, 590 787 ~251.1 ~229.7 | -215.7 -184 3.9 x 1071 uc1, 835 1727 _213.5 1197.3 | -186.7 -161.8 | ~ 3 x 107° Pucl, 734 1767 -229.3 ~213.1 | -202.3 -176.3 | 1.0 x 107° NpO1, 538 (800) 237 216 T=200 ] [-170] |~ 3x107 NpCl, 800 (1527) 216 ~200 [-183 ] [-158 ] ~ 107 —8‘[- Pressure Atm, . _10- Temperature ©C 900 800 700 600 50? 400 - | B _ et 1 i ucl, e A - L—M P 2 ~ [PbCl> t TVt P MP — :0’4i Q80 090 10O 110 120 .30 140 150 YT x 103 FIGURE II-4. Vapor Pressure, UClA, PbCl (T = ) -20- s The density of the fused salt was estimated from the relation, p=0p, (1-2.5%107%) (1) vhere t = °C. and p_ = density at 0° C. The density at 0° C., obtained by assuming an ideal solution with additive molar volume, was 5.01 gm./cc. for the composition Pb:U:Na = 43140317, Densities at 25° C. for PbCl,, UCl#, and NaCl were teken to be 5.85, 4.93, and 2.165, respectively. The specific heat of the fused-salt mixture was estimated from the specific heats of the three major constituents by assuming these to be additive. The specific heats for the constituents and that for the mix- ture at 100° ¢., 300° C., and 500° C. are given in Table II-2. In addition to the problems of handling UCl4 which arise from miclear considerations, i.e., restriction of processing and storage quantities to values far below that needed for criticality, there are handling precautions necessary because of its chemical properties. The salt is very hygroscopic and reacts with water according to the following equations: UCl4 + 4H20 — UCl4 . 4H20 and Ucl 4 + H20 _— UOl'.!l2 + HC1. If oxygen is also present, the additional reaction, uoc1, + 1/2 0, UO0,C1, can take place, Even in the absence of water, the following reactions proceed at a measurable rate at about 300O C.t Ucl, + 02 — U0,C1, + Cl2 4 -2 3U02012 + 02 — UBOS + 3012. The oxychlorides, U0012 and U02 UClL. It is apparent that in the preparation of the fuel mixture and in the operation of the reactor, extreme precautions would have to be taken in order to exclude water and air. Water and air are also undesirable because of their corrosive reactions with most structural materials at elevated temperatures. Furthermore, water and other low mass-mumber sub- stances must be excluded from the reactor area because of the hazard 20 012, are not readily converted back to 2729 -91- TABLE II-2., Specific Heat Data Specific Heat, cal/gm. Substance 100° ¢. 300° c. 500° . c1 A(a) 0.0792 0.085 0.091 Pbclz(b) 0.0681 0.074 0.122¢) Nac1(®) 0.217 0.224 0.232 40 UC1 4:43 Pb012:l7 NaCl 0.079 0.085 0.109 (a) From reference (11) (b) From reference (16) (¢)yt 540° c. wr o WAL, -22- AR . resulting from their action as neutron modersators. During operation of the reactor, the isotopic composition of the uranium is altered by the formation of U-236 and U-237. Too mmch U-236 is harmful because it dilutes the fuel and is a source of undesirable Pu-238. The U-236 is meintained at an acceptable level by isotopic processing in a K-25 type separation plant, The U-237 is a short-lived (6.63d) B -7 emitter and introduces a holdup problem in the decontamina- tion of uranium to be sent to the isotope separation plant. The latter problem is considered more fully in Chapter III of the engineering analysis report. 2.2 FISSION PRODUCTS For each fission product the properties of the chemical species vhich are likely to be present in the fused-salt mixture at a temperature of 70005800°_C:¥are given in Table II-3. The selection of likely species waétbésed on thermodynemic considerations, and on descriptive information conteined in Quill (14), Sidgwick (17), Latimer (18), and Huckel (19). No attempt was made to estimate the relative proportions for the cases vhere more than one chemical species seems likely, The effect of the high flux of ionizing radiation in the core on the equilibrium distri- bution of chemical species corresponding to different oxidation states is not knbwn. Presumably the majority of the chlorides will be present chiefly in ionic form in the molten salt mixture, | The values for the melting points, boiling -points, heats of formation, end free energies of formation given in Table II-3 were obtained from Quill (14), to & limited extent from Latimer (18), and from NBS circular 500 (20). The significance of parentheses and brackets around numbers in the table is the same as for Table II-1. In addition, parentheses have been used to enclose the formulas of compounds which have been entered in the table with some reservation because of insufficient or conflicting data, As in Table II-1l, the vapor pressures are for the pure substances as liquids at 1000° XK. and were obtained for supercooled liquids in some cases. For & few of the chlorides the vapor pressures at 1000° K. were estimated from values at 500° K. These chlorides, to- gether with their vapor pressures at 500° K., are given in Table II-4. f (A * 288 2 @ - - » . . .- s e .. . s ae o & . . . . ss e . . e e 6 . e .z » . . . . s @ . o . ¢+ ® . . an e R e vvvvvv Asnewe S tanne »BEAN TABLE II-3, Properties of Fission Products in Fused Chloride Fuel Atom ¢ Weight % Thermo- Normal o Vapor of of dynamically | Melting | Boiling | Ar® AF”, kcal/mole Pressure Fission Fission | Steble Form Po%nt Po%nt 298 o o o atm, Elem. [Products |Products | (ca.1000° K.) c. C. |keal/mole| 298° K.| 500° K. | 1000° k.| 1000° K. 7n | <107 [c107° ZnC1, 283 732 ~99.6 | -88.4 | -21.1 |=-70.0 |9.507t Ga |<107% |c107% GaCl, w1 | (59 |(-88) |(-79) [(-72) |[60] [45 GaCl 78 302 ~125 ~112 -103 E 83 [1.5x0 Ge 1.5><10"3 9x10™4 | GeCl, ees (702) (- 90) (-80) [(-74) |E 58 ~1 | GeCl - 49 80 i(-0) |(-153) |[-140] |[[F119] |~6 x 10 as | 4.5a07% | a0t 2601, - 16 122 - 715 | -6L7 | -5 |[5]] |~5x 10° Se 0.22 0.15 ‘ Sez 217 723 sae sa e see LX) o~ 3 Se Cl ! -85 130 - 2000 seae sesw e “\'5x102 (S 1-4? 305 sew - 4601 - 2606 - 1501 B 14] eve Br 0.065 0.045 i Br - 8 58 e e e e *e e sse oo s : bromides KI‘ 1.03 0.73 Kr [ XN ] —153.2 ee e oo [ s e L ] Rb 2.0 1.48 RbC1 m7 | 13 ~105.1 | -98.5 | -93.8 |- 82.4 |2.3a07% Sr hod 3.3 srcl, 872 |(2027) | -197.8 | <186.0 | -178.1 | -160.4 |1.3x078 Y 2.3 1.7 xc1, 700 | (1507) ~235 219 | =08 |-186 |3.3x0™° Zr 14.9 11.6 ZrCl, 727 | (1477) 2-145; — 2-134} {-127; 6_1123 1.9X105" 2rCly 437 331 subl.|(-230 -209 -195 -160 1.2X10 M | < 3 <2 NbCL, 212 243 oo | 3.8%0! MO 12.1 lo.l MO 2597 *se s *ee ' see L ) - .o s e w 2 MoCL, 217 322 (- 79) (- 60) |[(- 47) [=17 1.2X10] MoCl, 194 268 ~90.8 | -68.8 | ~-553 |[-25 5,6X10 1 ) %) 1 aaaaa wh e aaaaa ------ w '''''' T P/ e TABLE II-3, Prcmei't_:j._g_g_ of Fission Products in Fuged Chloride Fuel (continued) PO | Atom % | Weight % Thermo— Normal o Vapor > of of dynamically | Melting | Boiling | ,s0 AF”, kcal/mole Pressure Fission Fission | Stable Form Po%nt Point 298 o o o agm. Elem. Products | Products | (ca.l000°K.) C. oc kcal/mole| 298 K. | 500 K.| 1000° K | 1000 K, Tc 3.1 2.6 Tc (2127) (4727) > * S s * 09 0@ * & _..l chlB ( 727) ® 00 L X W J o0 LN N ) L N ) ~3x10 Ru 6.2 5.3 Ru 24-27 4227 e P L N ] 80 e P ® e -2 (Ru-Clz) een se s oes T e se e ~3X10__3 RuC15 dec.627 | (1250) | - 46 - 31 -21 -1 ~ 4 X 10 1 zaa.‘l‘.m.t’.‘.l2 Rh 1085 1.62 m 1%7 3877 *e e L 3 N e L I ] L X I J RhC1, ( 677) dec.958 | - 39 - 29 - 22 - 7 1 atm.012 BhClB sa ee "55 -40 _30 - 9 see Pd 0.75 0067 Pd 1555 3167 *o P .,. *e e *e o0 _2 PAC1, 678 1027 ~ 43 - 33 - 26 - 8 ~3.0 X 10 Ag 0.015 | 0.014 AgCl 445 1564 | - 30.3 | - 26,2 |~-23.6 | -19.2 | 3.3 x 107 cd 0.033 | 0.031 cdc1, 568 97 | -93.0 |-82.7 |-76.0 | -60.8 | 4.7x107° In 0.005 0.005 InCl 225 608 - 48 - 43 - 39 [~ 30} 4e5 5 InCl, 235 485 - 9% - 80 - 72 & 551 ~1 X 107, InC1; 586 |498 subl.| -126 -111 -101 ~ 78 6.7 X 10 Sn 0.05 0.05 SnCl, 227 652 - 8l.1 |[-71.5 |-65.1 E— 49 2.2 SnCl, - 33 113 -127.4 | -110.4 |-102.0 |[= 70 ~1X10 sb 0.03 0.03 SbC1, 73 219 | -91.4 |-77.7 |-7.1 |E 59 ~ 4 X 102 Te 1.30 1.44 Te, 453 987 5 X 10‘2"2 Teclz 175 322 > 9 *ee . e .88 ~2x102 TeCl 224, 392 | - 7.4 | -58.2 |- 47.4 | 29 ~1 %10 4 1 no T OOOOO » 111111 - QQQQQQ TABLE II-3, Properties of Fission Products in Fused Chloride Fuel (contimed) Atom % Weight % Thermo—- Normal o Vapor of of dynamically |Melting | Boiling Ai° AF, kcal/mole Pressure Fission Figgion | Stable Form Po%nt Pgint 298 o o o agm. Elem.| Products| Products | (ca.l1l000K.) C. C. kcal/mole| 298~ K. | 500" K.{ 1000~ K.| 1000" K. I 0.60 0.66 I, 113 183 ‘oo ceo con ces cee Iodides xe 869 10.1 xe s e "10801 e s e * P os e se 0 see Cs 8.3 9.5 CsCl 642 1300 | -106.3 |- 99.4 |- 94.5 | —83.2 |9.3 x 1074 Ba 3.2 3.7 BaCl, 960 1827 | -205.3 | -193.3 | -185.6 | -165.7 | 4 % 10~/ La 3.2 3.7 LaC1, 852 | (xwr) | 264 oy 236 | -209 2 x 1070 Ce 8.2 10.4 (CeC1) e vee o e .. .. ~ 107 09013 802 (1727) 260 -24.3 ~233 -206 3 X 10 Pr 2.9 3.4 (Prclz) ass o9 e *ee® L I N ) *daw ® P e Nlo--‘g Prcly 6 (1707) | -258 ~241 231 ~205 4 X 10 Nd 8.9 1009 (NdClz) see co e vee ses ess ees ~~ 10-_‘? | NdC1y 760 (1687) -254 -237 -227 ~201 5 X 10 Pm 1.30 1.62 (Pm'clz) o e o o9 » eoe S o0 * o8 ~10j Pm013 737 (1667) | (=252) (-235) (=225) {(~=200) 6 X 10 Sm 1.15 1.46 sncl, 740 2027 |(-208) |(-194) |(-186) |(~170) |1.3 x 207§ | SmC17 678 dec, | (-248) (-231) (-221) |(-197) |5.6 X 10 Fu 0.09 0.12 BuCL, 727 | (2027) |(-210) |(<299) |(-292) |(~176) |1.3 x 10:2 EuCly 623 dec. | (-233) (-217) (-206) |(~183) 3.4 X 10_6 -Gd 0.01 0.01 GdCl, 609 (1577) ~245 -229 218 ~195 9.4 X 10 -gz_ -~26- TABLE II-4. Vapor Pressures at 500o K Chloride AsCl3 GeGl4 89014 SbCl SbCl TeCl TeCl W N W ™~ Vapor Pressure, atm, 7.5 1.9 x 10 7.5 1.1 x 10* 1.2 2.7 8.9 X 1077 1.0 X 1072 W -217- For some of the substances the critical temperature is below 1000° K. However, the product of the extrapolated vapor pressure given in Table II-3 and the activity is a convenient approximate measure of the vola- tility in dilute solutions, Bromine and iodine might be present in elemental form, as inter- hslogen compounds, as bromides and iodides, or as mixed halides such as UGlBBr, etc. The contribution of each fission product, as the element, to the total of fission products is given both as an atom per cent and as a weight per cent in Table II-3. The assumptions used in estimating the atom per cent values were (a) that the thermal fission yields for U-235 may be taken as a reasonable epproximation to the fast fission yields, and (b) that the reactor had operated for several months without removal of fission producte. Fission chain ylelds were obtaeined from the Plu- tonium Project (21) and from Coryell and Sugarman (22). The chain yields were assigned to the last member of each chain except for those chains containing a relatively long-lived radiocactive fission product. Such fission products with half-lives greater than 100 days ere listed in Table II-5. For the welght per cent values in Table II-3, an aversge mass mimber . was assigned to each fission element. The percentage values are of use in estimating the relative importance of various fission products on a weight basis rather than on a radiocactivity basis., For example, 82% by wieght of the fission products results from the eleven elements: Sr, Zr, Mo, Ru, Xe, Cs, Ba, La, Ce, Pr, and Nd. ‘ No single fission product seems to be a much more effective muclear poison than any other for fast reactors. Furthermore, one advantage of fast reactors is that the tolerable concentration of fission products in general is greater than for thermal feactors. When the fuel is subjected to chemical processing for fission prod- ucts, the spectrum of fission products at steady state is altered. If, for example, a uniform processing rate of 2% of the fuel mixture per day with complete recovery is assumed, the mean processing age or residence time of the fission products becomes fifty days. For a fission product production rate of 1 kg. per day, the fuel would contain 50 kg. of fission producte at steady state. Due to this short residence time, 2725 21 >3 € ‘ TABLE II-5. Long~Lived (T > 100d) Fission Products Isotope Half-Life se’? 6.5 X 10% ol 6.0 x 10°% Sr90 19.9 3y Zr93 | ~ 5 X 106y 707? 2.12 X 107y Rl1106 1.0y Pd(lm) ~5 X 106y 5123 130 d b2 ~ 2.7y 1135 >3 X 1oZy Cs 2.1 x 10% 0313 7 33y coti4 275 a P47 ~ 4y sm(151) ~ 1000 y Bl 2.0 v “ -29- however, the yield of a chain can no longer be assigned to its last mem- ber. For example, the yield of chain-95 is distributed along the chain approximately as follows: Zr95 654 Nb95 35d M°95 (s). 65% 18% 17% Similarly, the yields of other chains such as those of mass 89, 91, and 103 will be distributed along the chains, and the total yields for each fission element given in Table II-3 will be altered. For each processing rate, then, the distribution of chain yield along each chain will vary. A complete caléulation would require consideration of the increase in yield along each chain and removal of chain members by neutron capture, radioactive decay, and chemical processing. 2.3 CHLORINE BALANCE AND RADIATION DECOMPOSITION It is of interest to know whether or not chlorine is released in 4 and converting U238014 to Np and Pu. In these nmuclear processes, conservation of oxidation state cannot be the process of fissioning U235Cl agssumed, This may be illustrated by the following two fission reactions in vhich the products are represented in their most stable oxidation state: Ut4(235) + 401" + n ~—> Kr°(94) + Ba'2(140) + 201~ + CL, + 2n 2 U”’(zss) + 401 +Cl,+n —>» As+3(90) + Pr+3(14.4) + 6C1 + 2n 2 If it can be assumed that the steady-state distribution of oxidation states of all components in the fused—galt mixture is that predisted thermodynamically, i.e., that the high flux of jonizing radiation in the reactor core does not alter the steady-state distribution, the fission products could combine with more chlorine than is available from UClA. This is based on the yields in Table II-3. Where more than one oxidation state 1s stable, the lower oxidation states and chlorine are favored in the chemical steady state at higher temperatures. In the case of ruthenium, for example, the trichloride shows appreciable decomposition to the elements at temperatures around 700o C. On the other hand, for the rare—earths the trichlorides would probably be favored over the di- chlorides by reactions of the type UCl, + SmCl, === UCl, + SuCl AF° =~ 5keal. at 1000° K. 4 3 3 For the formation of Np and Pu the various over-all nuclear chemical reactions are 239 239 1 o Np“77C1, — Pu 7CL, +5 CL, s 239 1 Np 013 + 2 012 239 L———-—————+- Pu Cl3 or 239 239 1 U Cl4 ~—~—p Pu 013 + 5 012 . There would be produced, then, one atom of Cl1 for each atom of Pu produced. This neglects the additional chlorine associated with the steady-state quantity of Np in the +3 state. The presence of free chlorine is very undesirable from the stand- point of corrosion. The atom of Cl corresponding to each atom of Pu, and one atom of Cl for each atom of U vhich undergoes fission, can be taken care of by replacing the U as UClB. Additional chlorine could be reduced by the addition of elementary sodium dissolved in the sodium chloride makeup., If there were an excess of UClB, however, it would probably be necessary to add controlled amounts of chlorine in one of the proposed processing streams. The possibility of radiation decomposition of salts with the re- lease of chlorine, in the case of chlorides, has been generally recog- nized, A limited amount of research on the effect of radiation on fused fluorides containing U-235 has been conducted at QOak Ridge (5, 6, 7). Decomposition appears to be negligible, Almost no data, however, are available for radiation decomposition of fused chlorides. According to Report Y-854 (23), "A eutectic mixture of LiCl and KC1l showed no change in its cooling curve after two weeks in the (ORNL) pile." In general, substances containing ionic bonds are resistant to pernanent radiation damage., Small disturbances such as coloration may generally be removed from solid salts by thermal annealing. For molten salts,‘electronic disturbances should anneal instantaneously. *e & sse » * & a0 L o . *ey * . * . - - -31- 2.4 DEPOSITION OF SQLIDS FROM FUSED-SALT MIXTURES For a cirpulating-fuel reactor, the formation and subsequent deposition of solids in pipes, pumps, etec. can be a serious matter. For the fused-~salt fuel there are several ways in which solids may be formed other than as the result of simply cooling the salt. Since the melting point of the fuel is determined by the chemical composition, at a given pressure, the fuel composition of & reactor operating at a fixed tem- perature close to the eutectic temperature must be controlled within narrow limits to avoid precipitation of one of the major components., | Although the mole fraction of fission products (and Np and Pu) will generally be small, it is possible for the fission products to lead to solid formation in at least two ways. A few of the fission products, such as Ru, Pd, and Mo, may be present in part in elemental form with very limited solubility, and others may form relatively stable compounds such as BaHCl6 and BaZ'UCJ.8 having limited solubility. Neptunium and plutonium trichlorides should be similar to uranium trichloride and would be expected to have limited solubility in UCl4 according to Kraus (4), but to be quite soluble according to Calkins (12}, see above, Solids may also deposit as the result of thermal mass transport con- sisting of the dissolution of container material at high temperatures and deposition at lower temperatures. This is a serious aspect of the problem of corrosion of materials in contact with fused salts, although it is not limited to these high-temperature flulds., The matter is con- sidered briefly in the following section and more extensively in Chapter V of the engineering analysis report. 2.5 CONTAINER MATERTAL FOR FUSED SALTS In selecting a container material for a fused-salt reactor, two approaches may be followed. Relatively inexpensive material with fair corrosion resistance may be used with provision for frequent replacement, Alternatively, more durable but probably more expensive material may be ~used, Since the fuel is expensive and even small transfer losses cannot be tolerated, and since the very operation of replacing a fluid-fuel core or blanket is a difficult remote control operation at best, it is to be avoided. A material which is highly resistant to corrosion by fused salts is therefore preferable. ‘ - "qull||||||lr...,...,.... - » . e - . - ”* - L ] ae & * a - 2 - . e - s » . . De * :..:’ sus * s ::2?‘25 31 _32- o Techniques for handling large quantities of UCl4 at elevated tem- peratures were developed at the Y-12 magnetic separation plant (24). Unfortunately, very little information on the performance of the con- tainers is reported. Among the materials used at Y-12 in the preparation, sublimation, and enrichment of UCIA were: | 1. nickel for gas-heated furnaces used for the vapor phase chlorination of UO3 at temperatures between 350° ¢. and 450° C.s 2. type 316 stainless steel for sublimation stills; 3. Hastelloy C for charge-boats used with the stainless steel stills; 4o tantalum electrodes in the calutrons; 5. carbon collectors in the calutrons. Apparently the UCl4 in stainless steel containers, In any case, the various containers used wag contaminated with ferric chloride when processed at Y-12 were cleaned periodically and were not subjected to the continu- ous operation and higher temperature expected for the reactor core, heat exchanger, etc, Grimes and Hill (2), in a literature survey of high-temperature fuel systems, refer to experiments conducted at Westinghouse on the electrodeposition of uranium from a molten bath containing UFA’ KF, CaCl,, and NaCl at 800° to 900° C. Apparently a graphite crucible and a molybdenun cathode were used successfully. Reference is made in the Liquid Metals Handbook (25) to the use of Types 304 and 316 stainless steels up to about 900° C. for the dis- tillation of metallic potassium from a potassium chloride-sodium melt, It is stated that %attack that could be attributed to the alkali metal and/or chloride was limited to an unexplained carbonization." The fact that considerable éuccess has been experienced in finding corrosion —resistant container materials for metallic fluorides has been a source of encouragement in the design pf the fused~chlor1de reactor, - For example, Inconel appears to show some promise as a containerfigafié;ial for the fused-fluoride mixtures eing considered by the ANP Group.at Oak/f T e e s e \g Ridge (1)., At Y-12 (24) it was found that silver is the most corrosion- resistant of the practical metals for the handling of uranium hexafluoride. I1lium R has been recommended for UF4 reactors at the Mallinckrodt Chemical Works (26). | - - . CULS i3 e, R . . .: : ': : . . - . » . . » (XY . . T TRk w e L e R T il g M- T . In the absence of actual performance data for the resistance of various materials to attack by the particular mixture of sodium, lead, and uranium chlorides proposed for the fast reactor, certain predictions can be made on the basis of thermodynamic considerations. If it is agssumed that the container dissolves through the formation of chlorides of the container materials, those elements which form relatively unstable chlorides at the temperatures involved should be resistant to attack by fused chlorides. Strictly on the basis of the free energies of formation of chlorides at 1000° K., estimated from data given by Quill (14), the elements in Table II-6 should be corrosion resistant. The estimated free energy changes at 1000° X. for the reaction between the element and UCl4 to give UCl3 and the chloride listed has been included in Table II-6. The extent of corrosion by UClA some UCl3 in the circulating fuel., Obviously several of the elements in: the table are unsatisfactory for use in construction of vessel walls, should be decreased somewhat by maintaining pipes, moving parts of pumps, etc.,, because they do not have the desired . physical properties at high temperatures, or because they have poor mclear properties, or because they are too expensive, It is to be hoped that fabrication techniques can be developed for materials, e.g., molybdenur and graphite, which are not subject to these objections. Unfortunately those materials which show promise for containing molten fluorides are not necesgsarily of interest for chlorides. With respect to uranium halides, the free energies for reaction of container metals with UCl4 and UF4 differ appreciebly as shown in Table II-7. The typical reaction of uranjum with the constituente of Inconel in a uranium-containing fluoride or chloride mixture would be — 2 UX4 +M 2 UXj + sz . From the data for the constituents of Inconel in Table II-7, it can be predicted, in a qualitative way, that all three metals will react with UCl4 and UFL} Even for a valubhsf ONF° as high as +25, the extent of reaction may be sufficient tc permit thermal mass transport as pointed out by Bredig (7). The enthelpy changes for these reactions, &lso listed in Table II-7, indicete that the equilibrium constants decrease with increasing temperature for corrosion of chromium by either UCl, or UF4 and of iron by UCl hot contsiner walls, 4 42 which would lead to mass transfer from cold to -34- L T TABLE JI-6. Flements with Relatively Unstable Chlorides AF° aF°(1000° X) Forfiggégnkat C;rro:%on(a) eaction Element Chloride kcal./mole kecal./mole Carbon cee (+) cee Molybdenum M0012 ~ =16 ~ + 28 Ruthenium Ru013 - 1 + 65 Rhodium RhC1l - 5 + 17 RhCl2 - 7 + 37 Palladium PdCl2 - 8 + 36 Silver AgCl - 19.2 + 3 Tungsten UClz ~ - 9 + 35 Iridium IrCl - 2 + 20 IrCl2 - 2 + 42 Platipum PtCl + 3 + 25 PtCl, + 4 + 48 Gold AuCl + 3 + 25 (a)AFO is given for the reaction M + niCl,=> MIGln + nUCl3 Y-+ | -35- TABLE II1-7. Estimated Free Energy and Enthalpy Changes for Corrosion Reactions(a’b) AF° 1000%K, keal, A® 298%k, keal, el v, | ow, | wy |, Ni +5 + 25 + 4 + 14 Fe -9 + 15 -6 + 4 Cr - -20 + 4 -20 -1 (a)For reaction of type 2 UXA +M == 2 UXB + M'_Kz; where X = halogen, and M = container metal. (®)pata from (13) and (14) -36- ol For the application to solutions at high temper&turgirof enthalpy chenges for unit activity at room temperature, the uncertainty is so great that we should probably limit our conclusions to the hope that the temperature coefficients are small so that mass transfer may perhaps be unimportant. Then it should be possible to reduce corrosion by the addition of uranium trihalide and of small emounts of the container metal hélides to the fuel., Of the three metals considered above, nickel appears to show the least tendency to corrode, and chromium the grestest tendency. All three metals appear to react more readily with UCl4 than with UF4° The tendency of nickel to react with the chloride is as great as that of chromium to react with the fluoride at the same temperature. Even if attack of the container by a common chemical reaction is unlikely, corrosion may result simply from the solubility of the material in the molten salt. Solubility data are evailable for several metals (14, 27). One proposed mechanism for the solubility of a metal in its halides involves the formation of lower valent ions (subhalides), which may be relatively more stable at high temperatures. Thermodynamic data are not available for predicting the extent of this type of corrosion. This problem is discussed in more detail in Chapter III. It is of interest to note that the corrosion problem for U013 is much less severe than that for UClA. A typical corrosion reaction would be 2UCL, + 3 Ni = 20+ 3 Ni012 . 3 Assuming that s fuel containing UCl3 would require a higher operating temperature than one containing UClA, & value of AFC for this reaction at 1500° K. was calculated and was found to be +201 kcal, If uranium subhalides are not formed, it should be relatively easy to find corrosion- resistant container materials for UClB. Kraus (4) noted that Uc1, attacks quartz but not steel. The effect on this type of corrosion of the alloying of uranium in the container walls is discussed in the engineering analysis report. Y “ L P o A _37- REFERENCES l. Personal Communication, Brooks, H,. 2. High Temperature Fuel Systems -~ A Literature Survey; ORNL, Y-657, Grimes, W. R. and Hill, D. G., CONFIDENTIAL, July 20, 1950. 3. Survey of Nommetallie-TLiquid Coolents for Nuclear—Power Pilesg; Fairchild, NEPA-1476, Shaw, H. L. and Boulger, F. W., 31 pp., CONFIDENTIAL, May 26, 1950. 4. Phase Disgrams of Some Complex Salts of Uranium With Halides of the Alkali and Alkaline Earth Metals; Brown Univ,, M-251, Kraus, C. A,y 20 pp,, SECRET, July 1, 1943. 5."Aircraft Nuclear Propulsion Project Quarterly Progress Report, Period Ending December 10, 19513 Briant, R. C., ORNL-1170, 191 pp., SECRET, March 6, 1952. 6. Aircraft Nuclear Propulsion Project Quarterly Progress Report, Period Ending March 10, 1952; Cottrell, W, B., OfihL—1227, 214 Pp., SECRET, May 7, 1952. , . ' 7. Alrcraft Nuclear Propulsion Project Quarterly/%rogress Report, Period Ending June 10, 19523 Cottrell, W. B., ORNL-1294, SECRET, August 21, 1952. . “3. Crystallographlc and Thermal Studies of the Systems of Lead Chloride and Chlorides of Other Metals; Treis, K., Neues Jahrb. f, Mineral., Geol,, and Paleont,, Beil-Band 37, pp. 766-818 (1914). 9., Private Comminication, Kaufmenn, A. R. 10. Ames Monthly Report; MR-292, Butler, T. A., SECRET, April 28, 1944. 11, The Chemistry of Uraniumj Part I, N.N.E.S. Div, VIII, Vol. 5, Katz, J. J. and Rabinowitch, E., 609 pp., UNCLASSIFIED, (McGraw- Hill, 1951). 12. Chemical Research: General, For Period August 10 to September 10, 19443 Calkins, V.P. and Nottorf, R. W., Ames, CC-1975 (A-2889) , 15pp., SECRET, October 7, 1944. 13. The Thermodynamic Properties and Equilibria at High Temperatures . of Uranium Helides, Oxides, Nitrides, and Carbides, Brewer, L., Bromley, L. A., Gilles, P. W., and Lofgren, N. F.; UCRL, MDDC-1543, 84 pp., DECLASSIFIED, September 20, 1945. fl YR f 37 I < & . o e » . ¥ j.“ Toom -38- S E QBT S 14. "The Thermodynamic Properties of the Halides," Brewer, L., Bromley, L. A., Gilles, P, W., and Lofgren, N. L., pp. 76-192 in The Chemistry and Metellurgy of Miscellaneous Materials, Thermodynamics; N.N.E.S., Div. IV, P.P.R., Vol. 19B, Quill, L. L., Editor, 329 pp., UNCLASSIFIED, (McGraw-Hill, 1950). 15. The Treansuranium Elements; Parts I and II, N.N.E.S., Div. IV, P.P.R., Vol. 1/B, Sesborg, G. T., Katz, J. J., and Manning, W. M., Editors, 1733 pp., UNCLASSIFIED, (McGraw-Hill, 1949). 16. International Critical Tables; Vol. 5, pp. 96, 100, (McGraw-Hill, 1926) . 17. The Chemical Flements and Their Compounds; 2 Vols., Sidgwick, N. V., (0xford, 1950). ' 18, The Oxidation States of the Elements and Their Potentisls in Aqueous Solutions; 2nd Ed,, Latimer, W. M., (Prentice-Hall, 1952). 19. The Structural Chemistry of Inorganic Compounds; Hfickel, Wes ' (Elsevier, 1950). 20. Selected Valies of Chemical Thermodynamic Properties; N.B.S. Circ. 500, U.S. Dept. of Commerce, Feb. 1, 1952. 21. MNuclei Formed in Fission; The Plutonium Project, J.A.C.S. 68, R411-2442 (1946). | 22. Radiochemical Studies: The Fission Products; N.N.E.S., Div. IV, P.P.R., Vol. 9, 3 books, 2086 pp., Coryell, C. D, and Sugarman, N., Editors, UNCLASSIFIED, (McGraw-Hill, 1951). 23, Radiation Effects on Inorganic Liquids — A Preliminary Literature Search; Y-12, Y-854, Carter, E. P., 80 pp., SECRET, March 12, 1952. 24. Chemical Processing Equipment: Eleciromagnetic Separation Process; N.N.E.S., Div, I, Vol., 12, Akin, G. A., Kackermaster, H. P., Schrader, R. J., Strohecker, J. W., and Tate, R. E., 506 pp., U.S.A.5.C., T.I.S., SECRET, 1951. ' 25, Liquid-Metals Handbook; NAVEXOS P-733, Lyon, R. N., 188 pp., UNCLASSIFIED, June 1, 1950. 26, Corrosion Tests Leading to the Recommendation of Illium R for * Green Salt Reactors; Mallinckrodt Chem, Works, MCW-249, Ritchie, C. F. and Teter, E. R., SECRET, 1943. SE®RET & 27. «39- "Metal-Salt Interactions at High Temperatures: The Solubilities of Some Alkaline Barth Metals in Their Halides;" Cubiceciotti, D, D. and Thurmond, C. F., J.A.C.S. 71, pp. 2149-2153, (1949). "Metal-Salt Interactions at High Temperatures: The Cerium Cerium Chloride System;" Cubicclotti, D., J.A.C.S. 71, pp. 4119-4121, (1949) . | "The Solubility of Cadmium in Mixtures of Cadmium Chloride with Other Chlorides;" Cubicciotti, D., J.A.C.S. 74, 1198, (1952), -40- SR CHAPTER III. CHEMISTRY OF LIQUID~-METAL FUSED-SALT SYSTEMS S. Golden and A. Boltax | 1. EQUILIBRIA BETWEEN SOLUTIONS IN LIGUID BISMUTH AND FUSED SALTS 1.1 SIGNIFICANCE OF EXPERIMENTS The recent work by Bareis (1) at BNL has pointed toward a poten— tially simple and powerful method for the treatment of miclear reactor fuels in order to remove and possibly to separate the products of re- action, Bareis' experiments héve been concerned with the processing of liquid uranium-bismuth alloys by means of an equilibration of such liquids with a fused-salt phase (consisting of the eutectic mixture: 59 mole 4 LiCl-41 mole % KCl). It is clear, however, that variations based upon the general principles involved in this kind of process will include the possibility of processing fused-salt fuels by equilibration with liquid metals, as well as the possibility of utilizing liquid alloys and fused salts of compositions which may differ significantly from those employed by Bareis. (For example, W. Grimes (2) of ORNL guggests the processing of fused fluorides by equilibration with liquid zine.) To assess the potentialities of the processing scheme suggested by Bareis! experiments, an investigation has been made of the thermodynamic equilibria which may be expected to occur in the systems for which ex- perimental date have been obtained. An attempt has been made to estimate the equilibrium constants of such reactions which may occur, based upon the thermodynamic functions compiled by L. Brewer (3). The data reported by Bareis have been subjected to analysis and generalization, 1.2 DESCRIPTION OF EXPERIMENTAL SYSTEM For completeness, the experimental system that has been employed by Bareis (4) will be described briefly. A KC1-LiCl eutectic mixture was prepared by heating the mixed C.P. salts slowly under vacuum until dehydrated and fused. Various salt melts were obtained by the addition of other salts to the fused eutectic mixture, The salt melt was contacted with bismuth or bismuth-~uranium . amem e eaaian e iRy - o e — e et ——— m -41- alloy to which had been added small amounts of rare-earth and other elements and which had been cleaned to remove oxides. The contacting took place in a coarse pyrex filter tube. The melt was then heated slowly under vacuum until completely fused. Pure argon gas was allowed to enter the tube above the melt, which forced the latter through the filter into a section of tube below. At this point the melt was sealed off under vacuum, The metal had the appearance of mercury, and the fused salt appeared as a water-like liquid, The sealed capsule then was placed in a constent temperature bath and allowed to equilibrate overnight at a temperature near 500O C. The following day, the capsule was allowed to cool, The salt was usually not clear, but contained suspended material which subsequent microscopic examination disclosed to be metallic bismuth. Before analysis, the sali phase was separated from the metal phase, fused again under vacuum, and filtered., This resulted in a clear salt phase. The metal ingot was washed to remove salt., Both phases were then analyzed, In later work (5) the procedure was varied slightly by first filtering the salt phase and then sealing below a cbarse pyrex filter, while the bismuth alloy was placed above the filter. The system was then evacuated, heated, and when the alloy was molten, it was filtered in contact with the salt phase by application of helium pressure. After sealing, equilibration took place at about 450° C. The capsule was cooled, and salt and metal phases separated. The salt phase was melted under vacuum and filtered., The metsal phase was washed, Both phases then were subjected to analysis. Generally the experiments with rare earths were conducted with tracer amounts, and analyses for these substances were made by counting techniques. 1.2 EXPERIMENTAL RESULTS In Figure III-1, which is taken from Bareis (1), is presented a sumnary of the distribution of the elements Ce, La, and Pr between U-Bi alloys and a fused eutectic melt of KCl-LiCl. In Table III-l, also teken from Bareis (4, 5, 6, 7) are the data pertinent to the experiments summarized in Figure III-1l. ., ‘ o 103| 2 3 4 567891 2 3 4 567891 2 3 4 567891 2 3 4 567891 2 3 4567891 : - ¥ 2 g fi’ 7 7 6 . ] 5 e z? 4 ™ Ay 3 ¥ ! 2 0 eae : E .8 @ " & "0:.: m - .. E ' e e - . b=t ': m . % A g .E ’i' % 10, 2 FIGURE III-1. E Z Rare Barth Distribution between 3 5 Liquid Bismath and Fused Salt %‘ 4 e No Uranium o 3 e Uranium Present o O n 10 10 10 1 10 CONCENTRATION OF RARE EARTHS IN BISMUTH -~ P.P.M. TABLE IIT-1. Concentration of Rare Earths end Uranium in Melts Gontainiqg Liquid Bismuth and Fused Chloride at 460° C.” Sample| Rare Rare Earth Uranium No. |Earth| ppm in Salt pp in Bi ppm in Salt ppm in Bi 166 La 5.0 0.066 5.3 900. 167 n 3.3 0.011 - 0.022 83, 770. 170 n 241, 8. 171 n The 1.08 1.4 475, 172 n 114. 0.3 - 0.32 207 " 234 2,6 - 3.5 ppb 208 n 29. 0.2 3. 17.648 209 " 53. 12 - 14 ppb 210 " 5.4 30 - 32 ppb 0.1 6.933 226 " 34. 077 229 " 54« 0,27 175 Ce 222, 6.6 0.35 320, 176 " 478, 32, 177 n 214, 5.1 180 | %. 0.96 - 1.0 6. 250. 181 " 40. 0.024 -~ 0,052 182 " 86. 0.12 - 0,17 213 Pr 83. 0.22 -~ 0.24 214 " 239. 2.2 215 " 30. 0.52 - 0.54 0.3 170. 3* . Note that BNL log data (4, 5, 6, 7) end those incorporated in BNL~125 (1) differ slightly. For consistency, since U distributions are reported only in former, these former data are used in the analysis which follows, Three significant findings are to be noted from Figure II1-1. 1. In the presence of uranium there appears to be a more or less uniform "distribution coefficient" between the’fused salt- bismuth elloy of the elements indicated above, Thus the con- centrations in the bismuth alloy of the elements indicated above are reasonably proportional to their concentrations in the fused- salt phase, when uranium is present in the system and the con- centration of these materials is not too great. A line of unit slope may be fitted to the lower curve of Figure ITI-1, which emphasizes this result, 2. Data obtained in the absence of uranium, however, are incapable of yielding a "distribution-coefficient” which does not vary gignificantly with the concentration of the materisl in the salt phase, Indeed, the slope of the upper line in Figure III-1 suggests that the concentration in the metal phase is roughly proportional to the third or fourth power of the concentration in the salt phase. 3. Inasmuch as the uranium content of the systems examined was limjted in extent, it seems reasonable to conclude from Figure ITII-1, that for increasing concentrations of the materials mentioned and for fixed uranium concentrations there will be increasing departure from a constant "distribution-coefficient,” These qualitative characterizations of the process examined by Bereis require an explanation in terms of principles that will facilitate the assessment of such a process for the separation of the products formed by a muclear reactor. Such an explanation will undoubtedly have some bearing also upon the evaluation of anticipated corrosion problems to be associated with container materiels for fused-salt mixtures. 2. CORRELATION OF EXPERIMENTAL RESULTS 2.1 MODEL FOR EQUILIBRIUM BETWEEN FUSED SALTS AND LIQUID METALS At the outset, it seems clear that there can be listed three model mechanisms which might possibly account for the experimental results vhich have been obtained, *HEBY ' - imaea L _45- 1. Physical dissolution of metals in fused ionic compounds, and vice versa. To explain Bareis! experimental results in terms of this mechanism, it would be necessary to postulate signifi- cantly non-ideal solutions, even at very small concentrations of dissolved materials. 2. Chemical reactions between metals and cations of fused ioniec compounds that determine the extent of solubility of the metals in the ionic¢ compounds, and vice versa, by virtue of hetero- geneous chemical equilibria between metal-free salts and salt- ~ free alloys. 3. A probably more realistic mechanism which involves a combination of the essential features of the two preceding mechanisms. The model which has been employed by Bareis to account for his re- sults is one in which physical dissolution of the various metals takes place in the fused-salt phase, His experiment represents, therefore, the distribution of a single chemical species between the two phases, This model stems from the investigations of Easiman, Cubicciotti, and Thurmond (8) concerning the solubility of metals in their corresponding fused halide salts. Further support for this model is to be found in the solubility of the alkali metals in fused alkali - and alkaline earth halides (9). However, the possibility of chemical reaction between metals dissolved in fused ionic compounds and the cationic constituents con-~ tained therein requires consideration of a chemical sort, if for no other reagon than completeness., In the present_report, an attempt will be made to exploit the consequences of a purely chemical mechanism (i.e., the second mechanism, above). | For simplicity, consider the simple chemical equilibrium: nM(diss. in metal) +.mNn+ (diss. in salt) = N (diss. in metal) + nM™ (diss. in salt) vhich may be written in an alternate form, | nM (diss. in metal) + mNX (diss. in salt) == mN (diss. in metal) + mMX (digs. in salt), where M, N, refer to metals, and X refers to a halide. Then for each shvasd _46- W such reaction an equilibrium constant may be written, L bed” [ , W CRNER where the brackets refer to thermodynamic activities relative to some (as yet, unspecified) standard state of reference of the individual chemical species in their respective enviromments. To obtain some idea of the magnitudes to be encountered for the equilibrium congtants, use has been made of the thermodynamic data of Brewer et al (3, 8), to estimate the standard free energy changes, AF, for the reactions, 1o Lloeg e 1 1 =M+ NCL === S MCl +3 N, where the standard states of reference are the normal state of each pure substance. The standard free energy results have been converted to equilibrium constants, 1 121 1 EII- o n (™ (we) = They are summerized in Figure III-2. Graphs of standard free energy changes for the corresponding fluoride system exist (10), and these K have been utilized to provide a comparison between certain reactions in the chloride and fluoride systems, Such a comparison is given in Figure III-3. If the activities in Eq, 2 may be approximated by the appropriate mole fractions (ideal solutions based upon pure substances as standard state of reference), it is clear, from Figure III-2, that LiCl and KC1 should serve as sufficiently good oxidants to transfer preferentially certain rare-earth elements into a fused-salt phase from a metallic phase containing bismuth, uranium, and plutonium, in addition to these rare-earth elements. This conclusion is confirmed by Bareis! (1) ex- perimental results. Also, incorporation of Be012 into the fused-salt phase will favor the transfer of uranium as well as certain rare-earth elements from a metal phase into a salt phase, This result also has been obgserved by Bareis. For a quantitative theoretical description of L. . 89012 - PuClB BeCl, - ZrClz BeCl, - 0013 dTUDES FIGURE III-2;, Equilibrium Conatant as a Function of Temperature Log K* vs, T (°K) On the basis of ons squivalent of chlorine, ' Reactions 1.3 an¥*th Identified as: NGln - MGI’l 1 1 wE ~ - NC1 MCln + =N 500 600 700 800 900 1000 1100 1200 1300 1,00 1500 Temperature (°K) . Sy v oA ! & ‘.p.“ * ; < .!0 '.6 f&! i_'l& - . * r . "o . e T e 9 ¥ ¥ AT h enoa Yasanw P amanas % Yo R asnw s ree - #ss & L1d¥0d8 1EY408 8 Log K' -1 -3 -5 -10 -2 ~15 Y h s g - - @ PP RA A uBP TANN S PR FIGURE III-3, Temperaturs (%K) *E ™ * By 2 & ¥ 2 & » & ¢ » = < * . 5 e . 9 = - » * * x . > a - * 1 1 » @ ¥ R * v * - S0 ar - Chlorides LIYDEB Comparison of Fluorides and Chlorides i with Tri-valent Uranium as the Common J Basis Log K' vs T(%k) Reaction: 1 1y o1 i 30+nuxn....3ux3+nn Identified as: Fluorides NFn O = — e —— 1500 Ay wsamdtar w —— et S _49- these results beyond the level indicated, it is clear that the ideality agsumption will require examination. However, from a qualitative, and possibly semi-quantitative viewpoint, there seems to be some measure of support for the chemical model herein described. Support for this model may be obtained: 1. from experiments dealing with the equilibria between K-Na alloys and the fused halides of these elements (11) and 2. from the experimental results of Bareis in which bismuth was employed as a metal solvent. Rinck (12) has measured the equilibrium, K(alloy) + NaCl(salt melt) == Na(alloy) + KCl(salt melt). The equilibrium constants which he reports are given in terms of mole fractions. Since independent measurements (11) on both the metal and salt phases support the assumption of such ideality, the comparison be- tween the experimental and calculated values is noteworthy, These re- sults are summarized in Table III-2. In addition to these results, there are available two experiments by Bareis (7) in which Na-Bi alloys were equilibrated in contact with eutectic LiCl-KCl melts. (The uranium present in each of the phases in these experiments was of such small molal amounts relative to the amounts of the other substances that its effect can be disregarded comparatively.) For one of these experiments (Ref. 7, Run #190), it was possible to mske a material balance which revealed that 1.86 X 102 gram-atom of 1ithium and potassium were transferred to the metal phase as against 1.97 X 1072 gram-atom of sodium which were transferred to the salt phase, agreeing to within 6 per cent. A comparison between measured equilibrium con- stants and computed values are also given in Table III-2., Because of the marked deviations from ideality, in the above sense, to be expected from alkali-metal-bismuth alloys at the temperature at which measurements were made (13), as well as the possibility of non-ideality in the salt phase at these temperatures, the comparison between measuréd and computed values is quite satisfactory, though perhaps fortuitous. Encouraged by this preliminary analysis, we now attempt to interpret Bareis! results, summarized above, To do so, the standard states of reference in any phase will be chosen to be infinitely dilute with regard -50- S TABLE IIT-Z. Equilibria between Alkali Metal Alloys and Fused Alksli Chloride Melts Tempgrature Measured Equilibrium Reaction C. Egq. Constant K + NaCl== Na + KC1 900 11.5¢2) 800 14.o(a) 460 14.4(b) 460 27.0(¢) K + LiC1==Li + KC1 460 7.3(b) 460 6.9(¢) (8) Rer. 12 (b) Ref. 7, Run #190 () Ref. 7, Rum #216 Calculated Eq. Constant (Fig.11I-2) 10.2 14.1 38.5 38.5 6.7 6‘7 It should be noted that agreement in (a) between calculated and measured values was known at the time the tables of free energy of formation values compiled by Brewer wers published, pp. 116-117. . hedee e See Ref, 8, SRRy -51- to the substances transferred to that phase. The equilibrium constants which appear in the following, therefore, represent a modification of Eq. (1) in which the activities have been replaced by mole fractions. The analysis may be simplified somewhat by noting from the equilibrium constant (Table IIT-2), o tni] [kcil kg1 =77 % [LiC1) (3) that Li K and that the amount of potassium which mey be transferred to the metal phase as a result of chemical reaction is negligible compared to the amount of lithium which may be so transferred. For simplicity, there- fore, mention of potassium is omitted, with only slight error caused by this omission. (A more complete analysis could readily take this effect into account.,) Consider, therefore, the following equilibria which may be expected to be of importance in the Bareis experiments: U(Bi-alloy) + 3Lit(salt melt) == 3Li(Bi-alloy) + U°¥(salt melt), R(Bi-alloy) + nLif(salt melt) == nLi(Bi-alloy) + R™ (salt melt), nU(Bi-alloy) + 3Rn+(salt melt) == nU3+( salt melt) + 3Rn+(Bi-alloy), vwhere R stands for a metal of interest, possibly a rare—earth. These equilibria, which are not independent of one another, have equilibrium _wal? w3 _Lg® [e™] KR [;if]n [ R ] ’ ' (5) 3/n Kyp = 3/“ [;_a] (' ) . (6) - i) | For equilibration experiments in which uranium and the rare-earth metals initially are in the bismuth alloy, & material balance requires that constants: [1i] = 3 d([UBfl +-39 [R‘“]) , | (7) where & depends solely upon the relative numbers of moles of the alloy and fused-salt phases, No. moles of salt phase (8) No. moles of metal phase ° Substitution of Eq. (7) into Eq. (5) gives, n n ([0 o3 )" (2] ) R [La¥]® [r] This equation presumably desceribes the state of affairs existing after X == equilibrium has been established between a salt phase initially con- sisting of the LiCl-KCl sutectic mixture and a metal alloy which con- tains uranium and one other element R. The quantities in brackets re- fer, of course, to equilibrium mole fractions in the appropriate phase. - Equation (9) exhibits all the qualitative characteristics observed from Bareis! experiments. First, consider the case in which uranium concentrations greatly exceed those of a rare-earth., Then, nt] K Li“j11 %fll""‘i‘ifi"‘m—n; (&) 7 [**— o, (10) so that in the presence of considerable excess uranium in the salt phase relative to R, and for ® reasonably constant, a more or less uniform distribution coefficient is predicted in accordance with Bareis' ex- perimental results. Next consider the case in which the uranium con- centration is negligible compared to the rare-earth. Then, +ln nfilntl K, |Li [7*] R[ ] [3+ [m] v ’ (|u / R - O) ’ (11) [R] nPx D so that in the absence of uranium the distribution coefficient R’”] /[R] is predicted to vary with concentration as [Rnf]-n’ which is again in accord with the experimental results, since & does not change appreci- ably from one-eXperiment to the next., For fixed uranium content, it is clear that Eq. (11) will be approached as the mole fraction [R"*] be— comes larger, again in accord with the experimental results, reoneme I S "9 284 % 488 3 S e » L ] * . L ] . . . 8 = > & @ . st » . . LI . . a 8 - LN . . . » . . & @ . L * v @ » . se & * Ser = » . o raddan - - -53- 2.2 QUANTITATIVE RESULTS In this section an attempfig&pqgggg to describe the experimental data more quantitatively in terms of the foregoing model. Accordingly, Figure II1I-/ represents a test of Eq. (6) to estimate a value of n, The data do not permit unequivocal determination of its value, but a value between 1,5 and 2 is in reasonable accord with the data. In view of the uncertainties involved, the value n =2 is adopted here for convenience and implies that the elements lanthanum, cerium, and praseodymium, insofar as they can be grouped together as regards these experiments, exist in the salt phase as sub-halides rather than normal halides. There is some evidence for this for certain of the rare-earths (8), which are presumably similar to the elements treated here, With the assignmment of n = 2, Figure III-4 yields, KU--R = 1.1 X 10“5 . (12) Similarly, Figure III-5, which represents a test of Eq. (9), with n = 2 yields -7 Kp = 3.2 X 107" . (13) From Eq. (6), Ky = K;_ K72, and Ky = 1.0 % 1074, (14) 2.3 DISTRIBUTION OF PLUTONIUM Inasmuch as plutonium is apt to be present in many reactors now being contemplated, the manner in which this material ultimately may be separated from uranium, fission products, and solvent by a process which involves liquid alloy-fused salt mixture equilibration is of considerable interest. According to the model herein deseribed, it is necessary and sufficient to know the equilibrium extent of chemical reaction between uraniun and plutonium halide, Unfortunately, only two experiments have been reported by Bareis in which plutonium was a constituent. However, there have been reported (14) similar experiments which involve the distribution of trace amounts of uranium and plutonium betwggg‘flgsgfrénd salt phases that differ markedly from those employed by Bareis. By making the most optimistic T RED N sSABENE » e o [ EE ENXE ] aa s -54- ol w© & o0 n N uéafl RANE EARTH DISTRIBUTION RATIO [n"fl )[a] 8 4 5 & 7 8 %fla FIGORE III-4, Influence of Uranium Diatribution Ratio on Rare Earth Distribution Ratio Key: . La + Ce o Pr Fumbers refer to Brookhaven run numbers., 2 4 & 10 yRANIUM DISTRIBUTION Ratro | 0>)/[u] S B 0N S . 0B *endbesd b:fh bigiagéilg‘li . eney + . "sesee ates . Tassen 0 1 2 3 4 - 10334) 2 3 45678930 2 3 4 5678940 2 3 4 567890 2 3 4567897 2 3 45678810 ] ; . , - ks 6 ol : . ! 4 J ; . 5 i .} FIGURE ITT-5. Teat of Equation (16) for Distribution ) s of Rare Earth (R) between Fused Salt - and Liquid Bismuth with and without 3 Uranium (U) Present. o 2 Key’: « La o + Ce o . © Pr L : 10 1 Underline denotes U present, e A 8 Numbers refer to BNL LOG DATA, » - :r—r‘ 6 : 1. rete Nn::. 5 "l’:.’ n 4 » - lm 3 Yt o 2 1»’. . ’;\D e 1 i.fl.‘..‘ t lo 1 . 9 _ 8 > % - - 7 : cianes 5 R 4 3 A . - 2 10°, 5 ' o o 1 -586- ) of assumptions, namely, that the standard states of reference for both the uranium and plutonium in infinitely dilute solutions are virtually independent of the nature of the alloy or ionic compound in which they may be dissolved, it is possible to correlate the information relevant to the chemical equilibrium between uranium and plutonium distributed between ionic compounds and metals. There is some feeling (14) that the valence of plutonium in dilute solutions of ionic compounds is +2. Accordingly, we are interested in the chemical equilibrium, Pu(diss. in metal) + 2/3 U°'(diss. in salt) = pust(diss. in salt) + 2/3 U(diss. in metal), which has the equilibrium constant, e (lml)z/z' - “Pug " [ra]/ ([T Adequate data for testing the concentration dependence (or lack thereof) of Eq.(15) are at present unavailable. Those pertinent data which are available, however, are given in Table III-3. In Figure III-6, the equilibrium constants are plotted in such a way as to suggest their correlation with the temperature at which the measurements were made, Considering the diversity of the data, it is rather surprising to obtain as good a correlation as seems indlcated, | If these results are significent, and not simply fortuitous, then the ideality assumptions noted above seem reasonable with some modifi- cation., Thus, since both solid and liquid phases are represented in Table III-3, it would be desirable to estimate the effect of this difference alone, By assuming a standard state corresponding to a liguid phase in- finitely dilute in plutonium and uranium, it is necessary to estimate the standard free energy change associated with the solution of these substances in solid phases. In other words, if ideality may be assumed for the solid phases relative to a solid phase infinitely dilute in plutonium and uranium, it is necessary to modify the equilibrium constants for the latter in order to put them on a basis comparable with the constants for liquid phases. This has been done by e AR TTTTT TR e e T -57- AR - Equilibration Experiments for Plutonium and Uranium No.'s . Tenp. on Ionic 2+ 3+ K Fig. Metal| Compound | %c. | F¥ /Pu u/u Po-U 11126 U (s) UBr 900 16(a) 1 16 (1) v (1) UBr_ 1140 28(8) 1 28 (2) U (s) UBr, 1000 40(8) 1 40 (3) Mg (1) Mger_ (1) | 760 | 27(®) Mg (1) MgCl_ (1) 760 7.5 Mg (s) | Mg(CL,I), | 622 0.92(#) - @y | 12 | © Mg (s) | Mg(Cl,I), | 622 e 0.72 Bi (1)| nic1-kc1 | 460 |2.19 x 1072(®) 1 45 x 1072(®) | .37 | (6) Bi (1)| Licl-KCl 160 [1.86 x 1072(%) 0,54 x 1072(¢) | 0.1 | (7) (2) Ref. 14 (b) Re. 7, Run #232 (e) Rer. 7, Run #235 ~-58- ofony <~ B3/ Temp. (©C) ..-.;.-w.r_,‘ o8 L XX ] » 2O ES o dbobe IGURE III<6: Ratio of Distribution Coefficients of LA A N XN} L LB X B} Pu and U between Fused Salt and Metal. See Table III for Key to Points. Reaction: 3 Pu(l) + 2 uxj7=====3 Puxz + 2 U(.’ Slope gives Al = 19 kecal/g atoa Pu. te ..o : ... .‘: I.: :. E E .65 ..: CQE :'E wf* -50- estimating the change in KPuAU of Eq. (15) induced by fusion of both plutonium and uranium. As a consequence of this analysis, it appears that the hypothetical liquid equilibrium constants agsociated with the experiments which involved solid phases are somewhat smaller than those reported for the solid phases. Indeed, when such a correction is applied, considerable improvement results in the correlation. (These results are not reproduced here,) The two data which involve bismuth alloys differ significantly from the value estimated by extrapolation of the higher temperature data. This disparity emphasizes the speculative character of the correlation ‘suggested here, While possible explanations could be suggested here which would tend to improve the correlation, it seems unwise to do so. The uncertainties associated with the as yet unconfirmed concentration dependence of the equilibrium product ratio, Eq. (15), far overshadow the uncertainties to be associated with a uniform set of standard states of reference to bring all the measured equilibrium constants to a common basis of comparison, It is worth noting that, apart from the bismuth alloy data, no aignificant change in Figure III-5 results from taking the uranium and plutonium valences in the salt phase to be the same. From the slope of the curve, the enthalpy change for the reaction is approximately 19 kcal/gram atom plutonium, This value allows an estimate to be made of the entropy change for the reaction, vhich at about 900° K amounts to about 21 entropy units. This value suggests that rather marked dissimilarities exist between the uranium and the plutonium under the conditions of the experiments, and provides some support for the assignment of a valence to plutonium different from that of uranium, 2.4 CRITICISM OF MODEL In spite of a moderately successful correlation of the experimental results by the chemical quel employed here, certain deficiencies may be noted: 1. No evaluation has been made of the effect of physical dis- solution of metals in salts, and vice versa, upon the equilibria; 2. There has been an emphasis on the formation of "sub-helides" of certasin elements in the salt phése, when there is no direct evidence bearing upon this point; 3. It has not been possible to mske a quantitative correlation be- tween the velues of the equilibrium constanis determined from experiment and values estimated from the standard free energies of the reactions postulated; this deficiency may be traced to lack of informstion regarding the standard states of reference to be used for the reactants and products as well as the occurrence of non-ideality of the solutes in their respective pfiases; 4. Na quantitative distinction has been mede between cerium, praseodymium, and lanthamum, when it is expected that small differences do exist. 3. APPLICATIONS OF THEORY TO REACTOR PROBLEMS 3.1 GENERAL In spite of the limitations which have been pointed out for a chemical model of alloy-fused salt equilibria, it is of interest to ex- amine the relation of the model to the two important reactor problems, processing and corrosion. In any alloy-salt system at equilibrium there is a single oxidation-reduction potential which can be expressed con- veniently as the logarithm of the activity of the halogen, even though this activity is far too smell for direct measurement, The halogen activity is fixed by the ratio of activities of any redox pair, and fixes the ratio of activities of every redox pair. For the chemical reaction, Moy 2 - m.M + X2 = B v 2 m R ] = K [x] (16) and K" can be computed at each tempersture from the standard free energy of formation of MX . Figure ITI-7 shows log [m{m] /[m] vs. log [012] for certain chlorides at 1000° K. The standard states are the forms of ele- ments and of the halides stable at this temperature. Quantitative use -6]1~ ["rom] 8ot -30 -25 -20 -15 -10 log [c1,] =35 ~45 see s . » seesne » - * * tFeSES wdeA s - L R ose w. B e e i wm.oua & Y - B pee e w ! - "l.l.. i - . » - at e l L3 € ietentl -62~ s of such a graph would require as standard states of reference infinitely dilute solutions in the solvent metal or salt. This disgram is there- fore only illustrative, 3.2 SEPARATICON PROCESSES The significant processing questions to be recognized and described are: 1. 2. 3. 4 5. 6. What metal solvent—fused salt solvent combinastions are rela- tively inert with respect to each other? (The practical matters of solubility of uranium in the metal, or of compounds of uranium in the fused-salt phase, and of the temperature re- quired to assure stable liquid phases of proper fluidity and vapor pressure, are not considered at this point.) What substances can be added to the salt phase to effect re- moval of uranium, plutonium, and fiséion products from an alloy containing thesge substances? Conversely, what metals may be added to a metal solvent to effect removal of uranium, pluto- nium, and fission producte from fused-salt mixtures containing compounds of these substances? What substances may be added to either phase to effect a separation and concentration of the constituenis of a muclear fuel in the other phase? Is there a limit to the possible separability for the various constituents? If so, what is its value, and under what process conditions may it be obtained? Are there any equilibrium limitations imposed upon contimous processing? How is processing affected by changes in equilibration temperature? An attempt can be made to answer many of these questions in terms of the model which has been discussed. In addition, there are important practical problems that cannot be attacked by means of this model alone: 7 8. 9. What is the rate of approach to equilibrium? How complete can th%‘EBxfiiag} geparations of the phases be made? How reective is either the alloy or fused salt with regard to container materials? & P 4 6:2 . 4« wsee v o . ses . . & L ] L ] s = & . @ an » .e 8 * e . . - . . ¢ . = . . .« s « » ¢ » o . . ey sre » sEn . e a . - - as e adw -' -63- 10, How significant will radiation damage be in alloys and fused salts, snd how will processing be affected? The choice of & metal solvent-fused salt solvent pair, the members of which are relastively inert towards each other, is restricted sig- nificantly by nuclear and physicsel conéiderations. Thus, in & thermal reactor employing a liquid alloy fuel, bismuth appears to be the choice of & solvent which will effect the best compromise between limited solubility of uranium, low temperature for solidification, and small loases due to parasitic capture of neutrons, In such a case, we need know only the fused ionic compounds which are relatively inert chemi- cally toward bismuth, The fused alkali chlorides end fluorides (with the exception of lithium compounds because of nuclear considerations) would serve adequately in this application, This can be seen from Table III-4, which contains equilibrium congtants estimated for the resction (8), %Bi My == %BiXB +iy, with 1 . 3 - Bi 3] 7 [=x)" and X being either chlorine or fluorire. For fused-salt fuels, the salt melt will contain relatively large anmounts of uranium compounds, so that suitability of metsl solvenits can be inferred from Figures 111-2 and III-3. Bismuth would appear to be adequate for such purposes. For chlorides at 1000° K, the same con- clusions may be reached from differences in the ordinates in Figure III-7, The metal phase should contain elements for which the curves are well below that of uranium; the salt phase should contein elements with curves well above that of uranium, In separation from an alloy fuel, such as a liquid bismuth alloy, it will be necessary to add an oxidizing agent equivalent to the amount of material which is to be rembved.‘ A very convenient one would be BiClB, and UCL, or Cl2 might be useful in certain cases. As an illus- 3 tration of decontamination let us assume that 0.1 per cent of the Pu may A 53 . -64- TABLE IIT—-4. Egquilibrium Constants for Reaction: 1 1 - 1 1 FBL+TMG == SBLX +IM Salt (MX_) 500° K 1000° K 1500° K LiCl 1.6 x 10~%7 6.3 x 10712 3.8 x 1020 KC1 1.8 x 10°%4 1.8 x 1072 6.3 x 103 Zrcl, 1.1 x 1077 2.5 x 1078 4.5 X 1070 LaCl, 3.6 X 10727 1.1 x 107+ 9.6 x 10~ v, 3.6 x 10718 5.6 X 107 2.4 x 1078 NaF 2.5 x 10727 2.5 x 10743 7.6 X 1077 CaF, 4 x 10722 1.3 x 10716 1 x 1010 KF 1.6 X 10726 2.2 X 10712 1,9 x 10~ ZIF, 1.7 x 1077 6.1 x 1020 3.8 x 107 Lmv3 1.3 X 10“8 1.2 X 10713 3.1 x 1077 UF, 3.8 x 10~%T 2.2 x 1070 4.2 x 1077 . DA . e s & S4 o L * .« » - s » * o et e o e e e——— F _65. be extracted in the first stage. For equimolal quantities of solvents, this would correspond in Figure ITI-7 to [CL,] of 10728 atm. oOnly 3 x 1077 per cent of the U would be transferred, and only 0.01 per cent of the La would remain in the metal, To scavenge the Pu, the salt phase would be treated with another lot of bismuth which contains reducing agent equivalent to the salt to be reduced. The metal corresponding to an alkali or alkaline earth chloride in the salt phase should be satis- factory for this purpose. If the Pu is to be geparated from the U in a subsequent stage, more oxidant would be added. If three per cent of the Pu is left in the metal phase in one stage, for example, Figure III-7 indicates that the [Clé} should be incressed to 10 <7 atm. and one per cent of the U would also be extracted. The Pu can be purified by treating the salt with more bismuth-containing reducing agent as in the scavenging process described above, but this step might not compete with the wet processing which will presumably be necessary for the final purification, This equilibrium is probably not directly applicable to a fused- salt fuel. Instead, most of the U could be removed by volatilization, and the Pu (and remaining U) then deconteminated by the process described ebove for scavenging Pu. Figure III-6 indicates that at low temperatures the Pu might be re- moved without much U, However, it would certainly be necessary to re- duce a1l the U to the trivalent state. As regards the effect of procegsing temperature, it appears from the equilibrium constants which have been estimated that in general large differences may be obtained at lower temperatures that make for favorable separations, and the differences generally tend to become sig- nificantly smaller at high temperatures, However, some exceptions do exist and may be taken advantage of, as in the separation of Pu and U discussed above. In sumary, we are able to suggest the characteristics of a separations process for reactor fuels which contein uranium, plutonium, and fission products either in an alloy or in a fused salt: 05 a * ¥ . - | i atse - saw - -9 oy - - XX EY] . » “oan " . sesese - -66- ““ Liguid-Metal Fuel 1. Bismuth appears to be a chemically suitable metal solvent. 2. In order to prevent nuclear poisoning of a reactor by fused salts vhich may be entrainéd from the separation process, it has been considered desirable to restrict the choice of anion to fluoride for thermal reactors. Alkali-fluorides (excepting lithium) and alkaline- earth-fluorides appear to be feasible salts. Eutectic mixtures of 60 mole ¥ NaF -~ 40 mole % KF with a eutectic temperature of 700° C., O 53 mole % NaF - 47 mole % CaF, with a eutectic temperature of 810° CI . would seem to be suitable. For fast reactors, where chlorine is satls- factory, the choices are much wider and will not be listed here. 3. The oxidant employed may be such a substance as B1F3, or an electrochemical procedure may prove feasible (;i). 4. Temperatures should be as low as possible for decontamination. 5. Enrichment of plutonium with respect to uranium in the salt phase will probably require temperatures in excess of 800° C, At lower temperatures it seems reasonable to suppose that uranium may be oxidized in preference to plutonium. Fused—Salt Fuel l. The choice of salt melis now contemplated involve fluorides or chlorides., In either case bismuth would be adequate for the contacting metal phase only in the absence of tetravalent uranium salts., Otherwise, the bismuth will transform the tetravelent halide to the trivalent halide. 2. At low temperatures enrichment of plutonium relative to uranium may be effected in the metal phase. Since fission products such as rare- earth elements are relatively more difficult to reduce, decontamination of the plutonium will certainly be effected. Electrolytic processing, with a cathode of liquid bismuth, would merit consideration in this sort of processing scheme, In such a scheme, the bismuth cathode would pre- sumably become enriched in plutonium relative to fission products and *The experiments at Argonne National Laboratory in which pure uranium is being prepered by electrolysis from fused LiCl-KCl mixtures containing U014'have yielded a decontamination factor of between 10 and 100, vhich is roughly anticipated from the foregoing analysis. ~+or Gb ‘ I it e . 0 . . s 0 . - as @ +e * 4 0 . - s + @ " e L » se . S -67- perhaps to uranium, and adequate control of the amount transferred could be maintained easily. 3. Extraction by a liquid metal will not serve to remove fission products in a fused salt from uranium, 3.3 CORROSION The considerations given above may also be useful in studying the corrosion of contsiner vessels, in particular in studying mass transfer. In Chapter II is discussed the relation of corrosion to the free energy of formation of corrosion products, and the relation of mass transfer to the free energy and to the enthalpy change of the corrosion reactions. 'Physicai‘mass transfer may result from a large variation with temperature of the solubility of the container walls in the contained fluid, This may be an importent means of transfer by liquid metals, but it is probably unimportant with fused salts. For chemical mass transfer it is npcessary that the salt of the container metal and a reducing agent both be carried by the fluid and that their equilibrium vary greatly with the temperature, Bredig (16) has suggested that the mass transfer of Cr, Fe, and Ni by fused alkali fluorides may be due to reactions such as, Fe(s) + KF(R) == FbFz(diss.) + K(diss.) , vhich are favored at high temperatures and reversed at low temperatures. It is also possible that a sub-halide of the container metal may be both oxidizing and reducing agent if it is stable at high temperatures and disproportionates at low temperatures to give the higher halide and the metal, as, 2FeX(diss.) 2= Fexz(diss.) + Fe(s) . For the thermodynamically stable halides, disproportionation increases with increasing temperature, which does not lead to mass transfer from high to low temperature, We do not know, of course, the effect of temperature upon the disproportionation of unstable sub-halides. It was shown in Chapter II that the reaction of UGl4 or‘UF4 with Ni, Fe, or Cr, such asg, 20014(3) + Fe(s) = 20C 3%aa:) + FeCl (s) ’ is almost independent of the temperature. Therefore a mixture of reasonable proportions, say at least 10 per cent of each component, should act as a buffer to prevent the chemical masgss transfer of container metal by any other reaction, For example any K which is disgolved in the salt at high temperature will react at low temperatures much more with UCl4 than with Feclz, since the former will be present in much larger quantity. If mass transfer can be avoided, corrosion can be reduced by the addition of approximately equilibrium amounts of the corrosion products. The possibility of application of these conclusions to UCl3 and UGl4 mixtures depends, of course, upon the mutuasl solubilities of these substances discussed in Chapter II. . *AEBRRS . - » L 1. 2. 3. be ., _69- REFERENCES A Continuous Fission Product Separation Procegs, BNL-125, Bareis, D. W., 23 pp., SECRET, July 1, 1951. Private Communication, Grimes, W., ORNL The Thermodynamic Properties and Equilibria at High Temperatures of Uranium Halides, Oxides, Nitrides, and Carbides, MDDC~1543, Brewer, L., Bromley, L. A., Gilles, P. ¥W., and Lofgren, N. L., 84 pp., DECLASSIFIED, September 20, 1945. Fuel Processing Interim Report, BNL Log #C-4213, Bareis, D, W., 4 pp.; SECRET, September 8, 1950. Summary Report of Fuel Processing Group Activities during 1950, BNL Log #C-4509, Bareis, D. W., 7 pp., SECRET, Jamuary 13, 1951. Fuel Processinz Interim Report, BNL Log #C-4714, Bareis, D. W., 8 pp., SECRET, April 3, 1951. Interim Report of the Fuel Processing Group, BNL Log #C-5723, Bareis, D. W., 13 pp., SECRET, Jamuary 15, 1952, "Temperature-Composition Diagrams of Metal-Metal Halide Systems,” Eastman, E. D., Cubiceciotti, D. D., and Thurmond, C. D., pp. 6-12; "The Thermodynamic¢ Properties of the Halides," Brewer, L., Bromley, L. A., Gilles, P. W., and Lofgren, N. L., pp. 104-115, 123-125; in The Chemigtry and Metallurgy of Miscellaneous Materials, Thermo- dynamicg, N.N.E.S., Div. IV, P.P.R., Vol. 19B, Quill, L. L., Editor, 329 pp., UNCLASSIFIED, (McGraw-Hill, 1950). 9. Elekitrochemie Geschmolzener Salze, Drossbach, P., pp. 101 et seq, qfifivmh(Edwards Bros., Amnn Arbor, Michigan, 1943). i e 10. Graph of -AF vs, T for Fluorides, Lee, M., ANP Project, Oak Ridge, . . UNCLASSIFIED, Unpublished. T ll.' Thermodynamics of Allo s, Wagner, C., Chapter 7, (Addison-Wesley Press, 1952) 12, Rinck, E. Ann, Chim 10, 395 (1932). 13. Der Aufbau der Zweistofflegierunzen, Hansen, M., pp. 306-308, 314-316, (Julius Springer, Berlin, 1936). 1. The Chemistry, Purification and Metallurgy of Plutonium, MUC~JCW- 223, pp. 133-144, SECRET, December 1944 . 15. Personal Communication to J Addoms from D. F. Peppard, ANL., 16, Personsl Communlcation, Bredig, M.,' RNL. -9 . & .8 .. " 488 @ I‘ ey C . . LA . *« ¢ C l P - l . o . . . * - > e a e i""' Ci _70- .. CHAPTER IV. SEPARATION PROCESSING R. Schuhmann, Jr. 1. INTRODUCTION This chapter represents a preliminary appraisal of th; problems and possibilities of separation processing for plutonium-producing re- .-: actors using non-aquecus fluid fuels such as liquid metals and fused salts, In this study some emphasis was placed on the separation prob- - lems associated with the two specific reactors selected for detailed engineering analysis by the M.I,T. Nuclear Engineering Project. These reactors required the following principal separations: 1, For the thermal reactor, the removal of plutonium and fission products from a dilute solution of these constituents and uranium in liguid bismuth, 2, For the fast-reactor core, the removal of plutonium and fission products from a concentrated solution of UCl4 in other fused chlorides. 3. For the fast-reactor blanket, the removal of plutonium and figsion' products from molten UGlA. Originally it was thought that the costs of fuel separation and reprocessing represented a large fraction of the total cost of producing plutonium in present solid-fuel reactors. Accordingly, a major incentive for the development of fluid-fuel reactors was the hope of reducing the cost of plutonium by the substantially lower separation and processing "costs which appeared possible with fluid fuels. Recent date from the AEC on the breakdown of the unit cost of producing plutonium in Hanford- type reactors and Redox separation plants, however, indicate that re- actor costs so overshadow separation and reprocessing costs that re- duction of separation costs no longer appears to be the prime incentive for considering fluid-fuel systems. This represents some shift in point of view, but the fact remains that savings in separation and reprocessing costs through use of fluid fuels might be very substantial, The separation processes now in use and proposed for solid-fuel reactors appear readily édaptable to the treatment of non—aqueous fluid anse » aTssaes TN ~71- fuels, and it is likely that one or more of these better established processes will be used for some of the more critical separations in the treatment of liquid fuels. However, the use of liquid reactor fuels mekes it possible to consider processing by continuous (or semicontinuous) removal of liquid from the reactof, passage as liquid through a simplse separation plant, and return to the reactor still in the liquid state. With this procedure the bulk of the fuel would not have to be taken through the many steps of complete physical and chemical change which now account for a substantial fraction of processing costs. 2. STOICHIOMETRIC REQUIREMENTS 2.1 QOVER-ALL MATERIALS BALANCES Figure IV-1l represents a generalized reactor and separation systen, ~with the prineipal streams of material indicated by solid arrows. For the liquid-fuel systems under consideration, the system as a whole after an initial start-up period should approach a steady state, so that the composlitions of the principal streams will not vary with time. Moreover, the fuel stream leaving the reactor will be of the same composition as the fuel in the reactor. For these conditions, it is easily shown that ¥y = axR | (1) in which y = vrate of producing this element in the reactof, in any convenient units per day. a = total steady quantity of a given element or given isotope present in the reactor, in the same units. x = fraction of reactor contents fed to separation plant per day (x can be greater than 1). | recovery of the element in the final product stream, ex- pressed as a fraction of this element in the stream from the reactor, It is assumed that the unrecovered fraction returns to the reactor. =s ) i Thus, if Pu is produced at the rate of y grams per day, we wish to limit the quantity in the reactor to a grams, and we have a separation process which will recover in finel form the fraction R of the Pu in the fuel stream, this equation gives the necessary rate of feed to the separation snsne - TaR A . - - Reactor Feed to Separation Makeup and reagents Return to Reactor sl FIGURE TV-1, Separation Process Military grade —— Plutonium —Fission Products | _,.UFg for isotopic Processing | _[Other Waste Products) Flow between Reactor and Separation Plant -ZL- e P S _73- process x, From the standpoint of maintaining a desired concentration of Pu or limiting the extent of poisoning by a given fission product, this relation shows that there is no necessity for attempting to make very high over-all recoveries in the separation plant, For example, it would no doubt be cheaper to maintain a given level of & in the reactor with x = 0.1 and R = 0,9 than with x = 0,091 and R = 0.99. RS golid-fuel systems is that the liquid fuel eliminates the severe require- mentgs on the nature of the gtream returning from the separation process to the reactor. Specifically the geparation "tailings" do not have to be decontaminated, freed of plutonium, refined, and re-formed. These less stringent separation requirements, of course, justify the hope that simple, rapid,and cheap separation processes might be used, although they fall orders of magnitude short of giving the separation factors necessary for present solid-fueled reactors. The use of liquid fuels of course in no way lessens the severe re- quirements to be met by the final products leaving the reactor system and separation plant as a whole. Thus the final plutonium product must be decontaminated to a high degree and otherwise of sufficient purity to meet military standards. The waste stream of fission products, and any other waste streams, must be substantially free of plutonium and also should not contain appreciable quantities of uranium or other valuable constituents of the fuel. 2.2 HIGHER ISOTOPES Processing requirements for reactors with high burn-up rates for the fuel, such as the fast fused-chloride reactors, are complicated by the formation of higher isotopes, U-236 and U-237. Details of this problem are considered in Chapter V of the engineering analysis report. From the standpoint of the separation plant, however, this problem re- guires the separation of uranium and its purification, decontamination, and conversion to UF6. The~UF6 is processed isotopically in a diffusion plant which returns uranium to the reactor system. 2.3 OSUMMARY QF SEPARATION REQUIREMENTS AND OBJECTIVES In the light of the foregoing discussion, the principal require- ments to be met by the complete separation process for a liquid-fuel i Ead - -74- system may be summarized as follows: 1. Re 3. 4o 5. The final product stream of Pu must be of acceptable military grade and purity. The effluent streams of fission products and other waste ma- terials must be economically low in Pu, U, and other valuable fuel constituents, The separatibn process must maintain the fuel composition in the reactor within the desired limits, in accordance with the relation y = axR which was discussed previously. Losses of Pu and other valuable materials must be low, If isotopic processing of U is required, the separation plant must yield a pure, decontaminated, and Pu-free stream of U or UF6 snitable for feed to a diffusion plant., In order to realize the full advantages of a liquid-fuel system, the following additional objectives should be considered: 6. 7. 8, 9. Simplicity of operation, manipulation, and control, especially for all parts of the process with streams high in fission pro- ducts and requiring shielding and remote control. Separation steps on the main stream of liquid fuel should be such that the bulk of the liquid fuel passes through the separation plant without chemical changes. That is, separation steps on the main stream should be in the direction of removing tconcentrates" of Pu and fission products from the liquid-fuel phage and returning the bulk of the liquid unchanged chemically to0 the reactor, Low holdup of Pu and other valuable fissionable elements in gseparation plant, | Elimination of "cooling™ period. 2.4 PRESEPARATION On the basis of present knowledge of high-temperature separation processeg carried out on 1iquid metals and fused salts, it appears un- likely that such processes alone will meet all the requirsments listed in section 2,3, and in particular may not be capable of yielding satis- factory final products in accordance with requirements 1, 2, and 5. On the other hand, the wet processes in use for solid fuels are known to be - - ~-75- capable of meeting these requirements but lead to complex and expensive procedures which do not satisfactorily meet requirements 6 to 9 of section 2,3 Accordingly, at this time the most promising way of utilizing separation processes carried out directly on liquid metals or fused salts is for presepération or rough concentration of plutonium and fission products into one or more crude products. Figure IV-2 is representativé of flowsheets based on the use of rough preseparation steps. The primary geparation steps conducted directly on liquid fuel yield relatively crude products with moderate recoveries of Pu and fission products while the bulk of the fuel passes through quickly without chemical change and is returned to the reactor. The primary Pu concentrate is retreated first in a separation step designed to yield a Pu product sufficiently decon- taminated to be fed without cooling and without heavy shielding to the finel wet refining by solvent extraction, ion exchange, or precipitation, Similarly, the primary fission-product stream is retreated for the pur- pose of scavenging any Pu or other valuable fuel constituents. The primary function of preseparation is to reduce substantially the quantity of material to be treated by the relatively elaborate and costly separation processes which are necessary to yield acceptable final products. | Another benefit which might be realized with flowsheets involving a simple preseparation process is that none of the steps in principle need be operated to attain an extremely high recovery. This possibility is realized in a flowsheet like that in Figure IV-2 in which each step re- turns unfinished intermediate products to a preceding step, or evenitually back to the reactor, The Pu-refining step would be operated to yield a final Pu product meeting military standards but does not at the same time have to yield a final Pu-free product because all the unrecovered Pu is returned to the main gystem. Thus,'the existence of a return circuit of unfinished material back to the reactor represents a further important reduction in the work to be done in the more expensive separation steps. 3. UNIT PROGESSES 3.1 NWET PROCESSES Separation processing of solid reactor fuels at present is based on wet processes, and several of these have been developed to a high degree (R X XE ¥ ] » | o N i Partial Cw%e Pu , product - Decontamination Pu M"g':%’ A of Pu _ Refining Plutonium Reactor Preseparation . Steps Scavenging ‘ : of Puand U Fission “rene from Fission Product % Products Rejects . / - A ~J NV ~" < High Te mperature Wet Process Processing of Liquid Fuel FIGURE IV-Z2. Multistage High-Temperature Separation Flowsheet s 77 to meet the severe recovery and decontamination requirements of solid- fuel systems, The wel separation processes fall into three groups: 1. Precipitation 2, Solvent extraction 3. Ion exchange These processes are well described in AEC classified publications and have been appraised with other types of separations by J. H. Arnold et al @. As already discussed in section 2 of thisg chapter, the proper role of these wet processes in liquid-fuel systems is in the steps of puri- fying or decontaminating the final products leaving the system, espe- cially the plutonium produet. On the other hand, separation schemes based entirely on wel processes would eliminate some of the msjor savings to be sought by adoption of a liquid-fuel system, 3.2 PROCESSES FOR LIQUID-METAL FUELS For reactors with liquid-metal fuels, such as the dilute bismuth solution considered in the engineering analysis report, the preseparation step necessarily requires a process or processes in which the bulk of the fuel remains as liquid metal while the desired elements are removed into another phase, Many separations of this kind are carried out on a large gcale in the refining of common metals., Also, preliminary experimental studies of liquid-metal separations on possible reactor fuels haée given promising results. Thus it is evident that liquid-metal separation processes comprise an important field for further extensive study as part of the over-all program of developing non-aqueous, liquid-fueled reactors. Also it should be kept in mind that liquid-metal separations have possible utility in processing solid fuels, A convenient basis of classifying liquid-metal unit processes is according to whether the second phase into which the elements are separated is (1) gas, (2) liquid, or (3) solid. Fach of these three classes of processes is extensively used in the large-scale refining of common base metals such as lead, copper, bismuth, tin, and iron, so that a variety of techniques are well worked out. The further development of liquid-metal separations for nuclear fuels will require mainly studies of the chemistry of specific systems and engineering adaptation of the procesgses to the special requirements of nuclear reactor systems. T TRre . aw n . a P teNe e s Ry R -78- " LIQUID-GAS. -~ In principle, separations of volatile elements and removal of dissolved gases are the simplest kinds of separation carried out on liquid metals. Since radicactive gases such as xenon and krypton and other relatively volatile elements (Ba, Sr, Rb, Cd, I, Br, ete,) will account for important proportions of the fission-product adtivity in liquid-metal fuels, it appears almost inevitable that a gas- phage separation will be one major step in the processing of a liquid- metal fuel, The common techniques of removing gases include (1) simple heating to a high temperature at atmospheric pressure, (2) flushing with an in- ert gas, and (3) vacuum degassing, In melting and refining the common metals, these separations are generally conducted in batch apparatus. Preliminary tests of the separation of gaseous fission products from melted uranium fuel elements gave promising results in the Ames Labor- atory {2). Brewver has considered this process on thermodynamic grounds (3). Experiments reported from Chalk River (4) indicate that possible volatilization and loss of plutonium must be considered, Virtually no experimental data are available on gas-metal equilibria, gas solubili- ties, vapor pressures of solute elements, rate of separation, and tech- niques of Qontinuous operation, so that at present engineering speci- fications can hardly be made of a process for removing gaseous and volatile elements from liquid-metal fuels. LIQUID-LIQUID SEPARATIONS. - Many possibilifies exist for separa- tions by treating the liquid metal with a second immiscible liquid phase. The extracting phase may be (1) another liquid metal, (2) a fused- halide mixture, or (3) slags, such as oxides, silicates, alkalis, ete. Slag-metal systems form the basis of a majority of the separations made in smelting and refining ordinary bagse metals, but at this time the halide-metal systems show the most immediate promise for processing re- actor fuels. Bareis () has shown that fused—chloride mixtures will re— move rare-earth fission products from dilute solutions of uranium in bismuth. These data and their theoretical interpretations were considered in detail in Chapter III. It might be pointed out also that fused chloride-liquid metal systems are the basis of a number of commercial metal-refining processes, collectively termed gselective chlorination processes because the behavior of a given metal in this process depends L _79- primarily on its chemical affinity for chlorine as measured, for ex- ample, by the free energy of formation of the chloride. Brewer (3) has suggested that separations of rare-earths, plutonium, and uranium might be made by gelective oxidation of their liquid alloys. This process merits comsideration for reactor fuels based on liquid bis- muth because bismuth itself is not readily oxidized and because the selective oxidation procéss is now successfully used in commercial bis- muth refining. The separated oxide might be removed either as a solid or as an oxide slag. If ah oxide slag is produced, it probably will be desireble to add auxiliary oxide fluxes to lower the slag melting point and to improve the selectivity of the separation. For example, caustic 'soda and sodas ash are used for these purposes in selective-oxidation procegses of lead refining. Commercial separations based on systems with two immigeible liquid metals are rare, but this combination is of some interest because work at the Ames Laboratory (6) has indicated that liquid silver and perhaps liquid copper will extract plutonium from liquid uranium. It would not be surprising if other useful systems of two immiseible liquid metals were discovered, but so far none has been proposed involving liquid bis- muth as one of the phases, LIQUID-SOLID SEPARATIONS. - Two general methods of separating im- purities from liguld metals as solids mey be digtinguished: 1. fractional crystallization or liquation, in vhich the original liquid metal is cooled slowly to cause precipitation of an in- soluble phase, and | 2. precipitation by adding a chemical which reacts with the im- purity to form a stable solid compound. In either case, the solid phase is separated after precipitation by ékimming, decantation, or filtration, depending on engineering re- ‘quirements, , As an example of fractional crystallization, uranium can be re- moved from solution in liquid bismuth by cooling to near the freezing point of pure bismuth. The uranium,separafies as solid UBi2 and near 285° C. only 0.038 % U is left dissolved in the liquid bismuth, A number of processes have been proposed for selectively pre- cipitating various constituents of liquid reactor fuels. These may be 78 QQQQQ -80- .« classified on the basis of the kind of compound precipitated as follows: (1) oxides (3), (2) carbides (2, 3), (3) nitrides (2, 3), and (4) inter- metallic compounds (6). All four groups show promise on physico- chemical grounds, and preliminary tests ét Ames have given favorable indications, 3.3 PROCESSES FOR FUSED-SALT FUELS As with liquid-metal fuels, it appears that the separation pro- cessing of fused salts will require the use of established wet processes for the steps of purifying the final products which leave the reactor system. In this section are described briefly several possible methods of directly processing fused salts. These methods at present are to be considered primarily for preseparation and other primary separations steps not requiring high separation factors. FUSED SALT-LIQUID METAL PROCESSES. — The experimental studies of Bareis (5) and the theoretical considerations of Chapter III have dealt primarily with the use of fused salts to treat ligquid-metal fuels. When liquid bismuth containing U, Pu, and rare—earths is treated with fused halides, the rare-sarths tend to enter the fused-salt phase ahead of the U and Pu so that a direct separation of figsion products can be made while leaving the valuable fissionable elements in the liquid- metal fuel. On the other hand, if a fused-salt fuel were treated with liquid bismuth, the rare-earth metals would not be extracted into the bismuth until all the U and Pu had been removed from the salt. Thus, the order in which the metals tend to enter the liguid metal under re- ducing conditiouns meansg that application of the process to fused-salt fuels would require a roundabout and complicated flowsheet. DISTILLATION, - Many of the halides of the metallic elements pre- sent in reactor fuels are sufficiently volatile that distillation or selective volatilization appears to be an attractive possibility for primary separations of the constituents of fused-haslide reactor fuels. This unit process is considered in some detail in the engineering analysis of the fused-chloride reactor. The proposed flowsheet for this reactor, with a fuel consisting primarily of NaCi, PbClz, UClA, and PuClB, involves a preseparation step in wvwhich UCl4 and a portion of the fission products are distilled from the salt mixture while the ~ . L <0 . T — ey - . amttiibay 81 nonvolatile PuCl3 remains in the distillastion residue. No data are available to predict the behavior of the noble gases xenon and krypton in the distillation of mixed halidgs, but it appears likely that separation of these fission products will be relatively straightforward. VOLATILE FLUORIDE PROCESSES. - Processes based on UF, volatilization and involving ClF3 or BrF3 as fluorinating agents are under investigation at several AEC installationg (_, 8,9, 1 __)‘«’k Apparently it is possible to decontaminate uranium by a volatile fluoride process (7), so such a step may be worthy of particular consideration for reactor systems in which UF6 must be prepared for isotopic processing in a gaseous-diffusion plant. FRACTIONAL CRYSTALLIZATION. ~ Constitution diagrams for various binary salt mixtures indicate various rough separations which might be made by cooling the melt and removing the solid phase. More complete data on the congtitution of binary, ternary, and multicomponent salt systems are needed, however, before any significant appraisal can be made of the chances of making useful separations by fractional crystal- lization, For example, Calkins (11) believed that the U0134U01A system might have a simple eutectic with extensive solid solution formation on the UCl, rich side, like the UBr 4 3 believed that UCl3 and UCl4 are only slightly miscible as liguids (See Chapter II, 1.2). If Calkins is correct, crystallization of UCLl with Pu013 and trichlorides of fission products might give a useful rough separation. --UBr4 system. Kraus (12), however, 3 IMMISCIBILITY IN FUSED-SALT SYSTEMS., — If Kraus is correct about the U’ClB--UCl4 system, however, there is the very interesting possibility of a two-liquid separation process in which plutonium and/or rare-earth trichlorides are concentrated in one phase while UCl4 and other con- stituents are recovered in the other liguid phase, Further work on the constitution of fused-salt mixtures may disclose other two-liquid systems which might form the basis of useful separations. ELECGTROLYSIS. — Electrolysis is a potential method of separation for fused-salt fuels, but data are not available for a significant ~ W * - | 51 -82- “' anslysis of the method at this time., Electrolysis in fused salis has been used for the purification of urenium at Argonne (13). METATHESIS. - Selective removal of plutonium snd/or rare—earth fission products from fused-haslide fuels by treatment with solid com- pounds such as oxides and carbides appears thermodynamically feasible. This type of process involves a metathesis, with the metallic element from the fused-halide combining with the oxygen or carbon of the added reagent to form a solid insoluble product. Brewer (3) has suggested several specific processes of this kind. 4. TENTATIVE FLOWSHEETS FOR PROCESSING Bi-U-Pu FUELS Insufficient data are available to justify engineering specifica- tions of any complete separations process for liquid-metal fuels. How- ever, enough information is available on specific unit procesSes of treating liquid-bismuth solutions to justify some speculation as to possible combinations of individual unit processes into complete flow- sheets, Accordingly, this section presents a few tentative flowsheets ‘which may be of some use as a guide in planning further work. These flowsheets involve primarily the following unit processes: Fused salt-liquid metal separations Vacuum removal of noble-gas fission products Selective oxidation Fractional erystallization Also it is assumed that one of the established wet processes will be used fdr the final steps of decontaminating and purifying the plutonium. Figure IV=3 gives a relatively simple type of flowsheet which is gimilar to the flowsheet described in Chapter IV of the engineering analysis report. The fused salt-liquid metal step is used to make a rough concentrate of Pu and rare-carths for feed to wet processing. In order to obtain a reasonable recovery of Pu in the fused-salt phase, the fluorine activity will have to be maintained relatively high (for example, by adding BiFB) and a substantial fraction of the uranium probably will be drivem into the salt phase. As has been shown in Chapter III, high temperatures are expected to favor the selective concentration of Pu relative to U in the salt phase, SRV, "se » . anop " . .e Military grade " Plutonium = Uranium Reject . - Noble Gases Bi,U,Pu;tFP Vacuum D ISES ) ; _»Volatile Fission Degassing Products to Waste Reactor ~—— NaF, CaF,+BiF3 ¥ | . | Fused - Salt Pu and Fission Wet Processing Bi+ U Preseparation | Product Concentrate (Redox) (Some Pu) (Some 1) ------ !!!!!! L] FIGURE IV-3, Separation Process for U-Bi Reactor Fuel Fission Products -88- -84- 4‘%& Figure IV-4 gives a flowsheet involving three separate steps of fused-salt treatment of the liquid-metal fuel, In each of these steps the activity of fluorine is regulated (increased by adding BiFB; de- creased by adding Na or Ca) to control the direction and extent of separation, In the fused-salt decontamination step, the objective is to separate & maximum of fission products even at the expense of driving a portion of the Pu and U into the fused salt. In the scavenging step, reducing conditions are provided to drive virtually all the Pu and U back into the liquid metal. Under these conditions, part of the fission products will be returned to the fuel in the scavernging step, but it is hoped that the fused-selt product of this step will be sufficiently low in Pu and U to be discarded as the principal fission-product reject. The fused--selt product of the Purseparation step should contain only fluorides with dissociation pressures close to those of PuFB. As com- pared to the flowsheet in Figure IV-3, this flowsheet should permit use of & simpler and smaller wet plant with a less serious shielding problem. Some doubt exists as to whether an adequate separation of Pu from U can be made in the fused ssli~liquid metal system. An alternative flowsheet to meet this difficulty is obtained by substituting a selec- tive—oxidation step for the fused-salt Pu-separation step in Figure IV-4. This alternative is based on the expectation that Pu would be oxidized shead of U (3), te give a plutonium oxide concentrate for feed to the wet plant, | It would not be surprising if a process such as that shown in Figure IV-4 would allow some fission products to build up in the liquid- metal fuel, The partial flowsheet in Figure IV-5 is one way of separa- tifig and discarding elements which fall in this category, but in view of the probable relatively high cost of the alkasli metals is best suited for application to a fraction of the total stream of liquid metal through the separation plant. In principle, this procedure should bleed from the system all elements which will not tend to come out in the final salt solution of fission products, in the gaseous waste product, or into the Pu-refining circuit. The sbove—discussed flowsheets by no means exhasust sll the possi- bilities which might be devised from present information, but further speculation without more quantitative data on the unit processes hardly eppears warranted at this time, " > L J e - e . L4 - e @ - . - e - * & @0 LR 2 . - EE R RN N1 000000 ------- ------ Pu Refining (Redox) Gaseous NaF,CaFz Fission Products + Bi F3 | | t vacuum Decontamination Degassing By Fused Salts 7 s S Reactor Na or Ca eb/%/ NoF CaFz + BiFs3 l Q\)c"/ S l P4 // |~ / S : Pu Concentrate cavenging Fused Salt [Tn NaF -CaFa | of Pu and U Pu " N_OF Ca fz from Fused Concentration |Refining Reject Salts. (U,Pu) Fission Praducts in Fused Salt to Waste N Liquid- Mefal Circuit A L™ R . Wet Processing FIGURE IV-4. Multistage Separation Process for U-Bi Reactor Fuel Military Grade Plutonium 1 oo o l ...... ...... oooooo ...... Cla g Liquid Bi Portion of Liquid Bi-U-Pu Chlorination After Degassing and or! nq ' Partial Decontamination with Fused Salts Crude Pu and U Chlorides | | | | | | , _ A cEL L D S S— Return to a Fused Salt-Liquid Metal Step in main system | | with K and Li to — Refined Bi Maintain Salt Balance Free of Pu and U | Conventional Bi Refining or Discard and buy New B Fission | Product Bleed FIGURE IV-5. Bismuth Refining Flowsheet —98- 3. bo 7. 9. 10. 11, 12. 13, A | - 87- REFERENCES Survay of Processes for Separation of the Products of Nuclear Reactoré, Reactor Science and Technology, TID-71, Arnold, J. H., Benedict, M,, Elsey, H. M., Finneran, J. A., Golden, S., and Grosselfinger, F. B,, pp. 96-112, SECRET, April 1951. Preliminary Survey of a Thermal Method for Removing Figsion Products, Developed by the Amég_Chemical Group, CN-437, Spedding, F. H., Johns, I. B., Newton, A, S., Voigt, A, F,, and Sullivan, W. H., 4 pp., SECRET, 1943. Hiéh Temperature Decontamination and Sseparation Processges, Univ. of Calif., UCRL-314, Brewer, L., 11 pp., SECRET, May 6, 1949. Simplified Methods for the Processing of Reactor Fuels, KAPL-670, Ahmann, D. H. and Tevebaugh, A. D., 25 pp., SECRET, Dec. 6, 1951, A Continuous Fission Product Separation Process, BNL-125, Bareils, D. W., 23 pp., SECRET, July 1, 1951. Chemical Research - Chemistry of Plutonium, CN-1058, Butler, T. A., Voigt, A. F., Wolter, F. J., Ayres, J. A., Hein, R. E., Tevebaugh, A. D., Johng, I. B., 12 pp., SECRET, Oct. 1943. Fluorination of Massive Metallic Uranium with Ligquid Chlorine Trifluoride, K-831, McMillan, T. S., Xirslis, S. S., Barber, E. J., 27 pp., SECRET, Nov. 12, 1951, Pilot Plant Studies of the Recovery and Decontemination of Irradiated Uranium by the Chlorine Trifluoride Process, K~Slé, Gustison, R. A., 23 pp., SECRET, June 13, 1952, ' The Argonne Fluoride Volatilitx Process ~ Status Report, ANL-4709, Hyman, H, H. and Katz, J. J., 210 pp., SECRET, Feb. 10, 1952, Recovery of Plutonium and Fission Products from Reactor Pot Residues of the Chlorine Trifluoride Process, K-817, Benton, S. T. and Gustison, R. A., 16 pp., SECRET, Sept. 12, 1951. Chemical Research: General, For Period August 10 to September 10, 1944, Calkins, V. P., Ames, CC-1975 (4-2889), 15 pp., SECRET, October 7, 1944. _ Phase Diagram of Some Complex Salts of Uranium with Halides of the Alkali and Alkaline Earth Metals, Brown Univ., M-251, Kraus, C. A., 20 pp., SECRET, July 1, 1943. Private Communication to J. N. Addoms from D. F. Peppard, ANL. flwlfl@qppuuyfi..gfi.;:{ SS? _g8- oG CHAPTER V, SUGGESTIONS FOR RESEARCH PROGRAM 1. INTRODUCTION The following suggestions for a research program are based upon the needs which we have found in connection with the study ofliquid- fuel reactors and related separation processes. No attempt is made to relate these suggestions to current research projects because of our incomplete knowledge of these projects. We do know, however, that some items vhich we suggest are now being studied. The greatest need is for theoretical studies. It is generally recognized that our understanding of liquids is far behind our knowledge of either gases or solids. Of all liquids and liquid mixtures, the theory of fused selts is probably the most limited, particularly for salt systems contasining complex ions., The theory of liquid alloys, particularly those containing intermetellic compounds, is almost as limited. Many of the systems of interest in reactor technology are rich in complex ions or intermetallic compounds. | It is inherent in theoretical research, however, that it cannot be programmed, except for detailed developments. Each advance in theory opens up new fields for development, This is well illustrated, in a modest way, by Chapter III of this report. We do all know that insight vhich may lead to & new theory depends greatly upon the breadth of knowledge, Therefore, cur appreciation of the importance of theory should prevent the limitation of research to systems of immediate prac- tical interest, For example, it is unlikely that bromides or iodides, or lithium or cesium salts will be used in or with reactor fuels. Yet they are so important in the series of the halides and of the alkalis that their behavior in mixtures with the salts of uranium and other metals should not be ignored. Primary emphasis must be placed, of course, on materials of practical interest. The elements of greatest importance in this work are U, Pu, and Np. They are listed above in the order of decreasing importance but also in the order of decreasing knowledge. The substances of present principal interest are the metals, their A RN e e T aeti ¥ e oty _8o- chlorides and fluorides. However, more knowledge of oxides and carbides might lead to better separation processes. 2. PURE SUBSTANCES 2.1 PHYSICAL PROPFRTIES Before fluld-fuel reector processes can be designed in detail it will be important to have reliable information on physical properties of fluorides and chlorides of U and Pu, particularly melting points, vapor pressures, densities, viscosities, specific heats and latent heats, and similer information on solvent metals and salts, where such data are not aveilsable, 2.2 CHEMICAL PROPERTIES Relisble data on the heats of formetion and free energies of formetion of the fluorides and chlorides of U, Pu, and Np, of fission- product elements, and of iron, nickel, chromium, zirconium, molybdenum, tantalum, and other structural materials are of key importance in de- vising fused salt-liquid metal separation processes, and in understand- ing corrosion by fused salts, Brewer'!s valuable compilations of the thermodynamic properties of compounds of U (1), Pu (2), and other elements (3) are noteworthy contributions, but many of the entries are necescsarily based on inaccurate experiments and many only on anslogies, These should be replaced by results of precise experiments, In partic- ular, some of the properties for Pu derivatives are assumed to be the gsanme as for the corresponding U derivatives, and they are, therefore, of little value for studying the separation of these elements. HNo values for Np are available. Any sub-halides of U, Np, Pu, of the fission products, or of the structural metals, even those which are too unstable to exist in the pure form, ere extremely important for reactors. OQur knowledge of them is meager indeed, 3. MIXTURES 3.1 ALLOYS Phese diesgrams for many binary alloys containing U asre available and ere summarized in recent articles in the Journal of Metallurgy and cove - "0 d . . . . e : & » . ol - *ae " - —awea i sroen aeEee 59 -90- -‘i Ceramics (4, 5). There is need for correspondingly complete information on binary alloys of Pu, to help select fuel systems for breeder reactors and to suggest possible separation processes for liquid-metal fuels. Information is also needed on ternary alloys of U, Pu, and the promising solvent metals, bismuth, iron, nickel, and alumimum, Phase diegrams for alloys of each of these solvent metals, U, and individual fission pro- ducts would be helpful in working out separation procedures, and phase diegrams for each of these solvents metels, U, and individual structural elements would aid in the study of corrosion. In all of these studies, the liquidus curves will give valuable in- formation, but we should also know the compositions of the solid phases, particularly those of intermetallic compounds. Measurements of the activities of the components, where obtainable from vapor pressure or electromotive force measurements, will be particularly valuable in understanding these systems, The recent book of Wagner (6) gives an excellent picture of methods of studying alloys and of the information vhich is importent. Barly work at Ames, recently confirmed at Chalk River, has shown that Pu may be selectively extracted from molten U by molten Ag or Cu. Complete liquid-liquid equilibrium diagrams for the systems U-Pu-Ag and U-Pu-~Cu including the solubilities of U and the effects of added com- ponents should be studied., The distributions of individual fission pro- ducts between U and Ag and Cu also need to be known, The presence of gases in reactors may be so disturbing that it is important to know, as functions of the temperature and pressure, the solubilities of the gaseous fission products, particularly of the noble gases, xenon and krypton, in liquid urenium, liquid bismuth, and in other potentiasl fuele. The vapor pressure of zinc in liquid zine~bismith—uranium alloys, and of mercury in liquid mercury-bismuth-uranium alloys may be of prac- tical interest since these alloys have been suggested (7) as fuels to be used with evaporative cooling. 3.< SALT MIXTURES The needed studies on salt mixtures are very similar to those described above for alloys, but with complex salts instead of inter- metallic compounds. The important methods of study are essentially . 00 LA RN X3 - - & - sahasas Sy | _91- the same: 1liquid-solid and liquid-liquid equilibria or phase diagrams; vapor pressure and electromotive force measurements to determine activ- ities. More of the salts have vapor pressures large enough to be measured for activity determinations and to be of practical importance in re- actors and especially in separation processes. There is an additional complication that an element may be present in a salt mixture in more than one state of oxidation. The oxidation-reduction equilibria of pairs such as U-Pu or U- fission product need to be investigated., Sub-halides which are not known as pure substances may be stabilized in salt mixtures, and may have considerable practical importance, especially in corrosion. This is one of the more pressing problems for research. The system UClB--'UCl4 has been described as giving two immiscible liquids, and also as forming a simple eutectic with solid solutions on the UClA-rich side (Chapter II)., Knowledge of the solubility of UCl3 in UClA,iS very important to reactor design. Moreover, if PuGl3 and the rare—earth chlorides behave approximately like UClB, this system may be used for a convenient separation process. The nature of the process would depend upon the state of the trichloride-rich phase. More detailed and accurate knowledge of this system is particularly important.‘ A few fused-chloride mixtures cqntaining U613 and UGlADhave been studied (8), and several fused-fluoride mixtures containing UFA'have been studied at Qak Ridge in connection with the ARE project. Work on the chlorides particularly needs extension. 3.3 METAL-SALT SYSTEMS Equilibria between liquid alloys and fused salis, as discussed in Chapter I1I, afford the bases for separation processes, and also means of studying the behavior of fused-salt systems which mey be particularly important in understanding corrosion. Studies of the equilibrium between a single salt and the metal corresponding to its cation, such as fihoée of Eastman, Cubiccilotti, and Thurmond () should be extended, and an attempt should be made to de- termine the nature of the species present in the salt phase. The metals studied should include U, Pu, fission products, bismuth, alkalis, al- kaline earths and structural metals. The metal phase may be solid, et G P 2 . - e 2R, men - - - » e - & - - * * . -~ A . a0 " @ TeAER e THGRD - * » -a - » L] ragses XK K J _92- SRR liquid; OT' vapor. Measurements on systems containing two metels and their salts should ineclude in addition the determination of the distribution of each element between the two phases, Finally, studies of more complex sys- tems, such as those of Bareis (see Chapter III) but including higher concentrations, should be extended to give the distribution ratios of importent pairs as a function of the temperature and the compositions of both metal and salt phases., A radiocactive material or any material easily measurable in small quantities may be used as indicator for the oxidation potential in the system, The distribution ratios of greatest practical importance in reactor technology will probably be Pu-U and individual fission products - U, with the metsl largely liquid bismuth and the salt largely fused mixtures of alkali and/or alkaline earth chlorides or fluorides, but theoretical studies may well involve a much greater range, Studies of these equilibria offer one of the most power- ful tools for studying fused salts. 4. RADIATION STABILITY Studies should be made upon the effect of reactor radiation on chemical equilibria in fused-salt systems. In fused chlorides and fluorides the activity of chlorine or fluorine is important in relation to corrosion in reactors. Sub-halides may be importent in this relation also, In this project it was considered unwise to consider any fuels containing complex lons, such as nitrate or sulfate, because of the greater risk of decomposition by radistion. If these salts are inter- esting in other ways for a thermal reactor, the effect of radiation on them should be studied. 5. EKINETICS The rate of approach to equilibrium in metal-saslt systems may be as important as the position of equilibrium so that it may be desirable to study the rates of various processes, including diffusion, phase separation, distribution between phases, and chemical reactions, both homogeneous and heterogeneous., The high temperatures and relative simplicity of the reactor systems tend, however, to reduce the importance of rate studies. Ao d“IIIIII'IIII. ' o . - *r & Sas ¢ o9 . . T eSS - *e - - *res - - - ¥ . . l.# b o X . 'y afls » JRE 5. 6. PHYSICAL, PROPERTIES OF FUELS Physical properties of cértain pure substances must be known as & preliminary to reactor design. The study of the properties of the mix- tures used as fuels is of even more direct importance. Such properties are density, viscosity, heat capacity, thermal conductivity, and vapor pressure, The viscosity, thermal conductivity, and vapor pressure may vary greatly with compound formation, and they may be decisive in the selection of reactor fuel. 7. UNIT PROCESSES FOR SEPARATION The objectives in unit process research are the development of effective and economical separation steps and the quantitative under- standing of the basic process variables and how they are related to each other. The logical starting points for developing'separations processes are favorable thermodynsmic date for the primary reactions, but the mak- ing of a usable process requires meassurement and integretion of data on energy requirements, materisls flow and materials handling, reaction and mass transfer rates, apparatus design, materials of construction, etc., Unfortunstely, the special requirements to be met by separation processes for reactor fuels in general will not be met very well by many of the techniques and operating procedures which are standard in the metallurgical industry. 7.1 LIQUID-METAL PROCESSES The discussion below deals with processes in which a separation is made from a ligquid-metel phase into a second phase which may be solid, liquid, or gas, depending on the process. 7.2 EXTRACTION WITH FUSED SALTS. - On the basis of present knowledge, the treatment of liquid-metsl fuels with fused salts promises to be a very useful process for separation of rare-earth fission-products, U, and Pu from liquid metals. The research program on this process should be considerably expanded beyond the present program at Brookhaven and the theoretical study now being started by Norfih American Aviation. . If it is found that single-step extractions do not give the necessary separation factors, techniques of counter-current multi-plate extraction should be developed similar to those used in conventional solvent extraction. Cos il 7.3 GAS-METAL SEPARATIONS, — Some of the fission products may be permanent gases and others may be sufficiently volatile to be separated from the liquid fuel by boiling, vacuum distillation, or flushing with inert gas. At present virtually no date on solubilities or relative volatilities are available which could serve as the basis for designing a contimious process of removing gases from liquid metals., 7., SELECTIVE OXIDATION. — Many of the procedures of fire refining of common metals are based on selective oxidation. Brewer suggests that Pu might be separated from liquid U in this way (10), and a similar separation should be feasible if the Pu and U are initially present in liquid Bi. 'A serious problem to be considered in connection with this and other processes in which a solid phase is precipitated from liquid metal is to find a suitable method of performing the solid-liquid geparation, especislly if the quantities of solids are small. It might be possible to add an inert flux so that the oxides could be more easily separated in a ligquid phase. 7.5 PRECIPITATION OF INTERMETALLIC COMPOUNDS. — Work at Ames some years ago indicated that Pu could be precipitated from liquid U in the form of intermetallic compounds with tin and possibly other metals., This type of process is of particular importance in present-day lead refining. 7.6 IMMISCIBLE LIQUID METALS., — Systems with liquid silver or liquid copper in contact with liquid uranium or iron-uranium alloys are said to be under investigation at Ames. 7.7 FRACTIONAL CRYSTALLIZATION. - Constitution disgrams and equili- brium data show vhat separations are possible by cooling and crystal- lization of liguid alloys, but considerable further research on tech- niques of phase separation and contimuous operation is needed to ‘develop a complete separation process. 7.8 ELECTROLYSIS, LIQUID-METAL ELECTRODES, AND FUSED-SALT ELECTROLYTES. The work on fused-galt extraction of liquid Bi alloys suggests the possibility of electrolysis with two liquid Bi alloy electrodes and a fused-salt electrolyte or with various combinatioms of oneg}aguid Bi electrode and one solid electrode. A_A‘!’," I veiw “vesew - L I ] THILe Sanany . & ET L suremyg . - cvne - r1eanee i _95- 7.9 SELECTIVE CHLORINATION. ~ Another process similar in principle to the fused-salt process would involve passing chlorine gas through the ligquid metal to form a s a lt phase of chlorides of metals derived from the liquid fuel. This might be a good way of separating U or U and Pu from a decontaminated Bi-U-Pu alloy. 8. ENGINEERING DEVELOPMENT OF COMPLETE SEPARATION PROCESSES On the basis of the limited quantitative information now available on the individual unit processes, an extensive program of testihg com—~ plete processes would be premature., However, the available data do justify a limited program, and this program should be expanded rapidly as the unit processes are developed. In particular, modest programs could be planned now for the engineering development of complete pro- cegses for the following feed materials: 1. Irradiated Bi-Pu-U liquid alloys. 2. Liquid fuels based on Fe~U systenm, 3. Melted Hanford slugs. 4e Liguid fuels based on Al-U systen, 5. Irradiated UCl 4 in fused alkali chlorides. 6. Irradiated UF4 in fused alkali fluorides. 9. CONCLUSION It should be emphasized again that at the present state of develop- ment of non-aqueous fluid-fuel reactors,research on the design of com- plete reactors may well be subordinated to studies of unit processes. Still more important is the basic study of the properties of the ma- terials and systems related more or less closely to those used in or with such reactors. The most important research, however, will probably " be the broader and more fundamental theoretical studies suggested by the reactor problens, (s .1 - . sehae [ *a0en e - - e e 96- oPa 3. he 5. 9. REFERENCES The Thermodynamic Properties and Equilibria at High Temperatures of Uranium Halides, Oxides, Nitrides, and Carbides, MDDC-1543, Brewer, L., Bromley, L. A., Gilles, P. W., and Lofgren, N. L., 84 pp., DECLASSIFIED, Sept. 20, 1945. "The Thermodynamic Properties of the Halides," Brewer, L., Bromley, L. A., Gilles, P. W., and Lofgren, N. L., pp. 76-192 in The Chemigtry and Metallurgy of Miscellanecus Materials, Thermo- dynamicgs; N.N.E.S., Div., IV, P.P.R., Vol. 19B, Quill, L. L., Editor, 329 pp., UNCLASSIFIED, (McGraw-Hill, 1950). The Thermodynamic Properties and Equilibria at High Temperstures of the Compounds of Plutonium, BC-88, Brewer, L., Bromley, L., Gilles, P. W., and Lofgren, N. L., Oct, 10, 1947. The Binary Alloys of Uranium, Journal of Metallurgy and Ceramics, TID-65, Buzzard, R. W. and Cleaves, H. E., pp. 25-53, SECRET, July 1948, The Solubility of Uranium and Thorium in Liquid Metals and Alloys, Journal of Metallurgy and Ceramics, TID-65, Hayes, E. E., and Gordon, P., pp. 130-141, SECRET, July 1948. Thermodynamics of Alloys, Wagner, C., (Addison-Wesley Press, 1952). Personal Communication - Kaufmann, A. R. Phase Diagrams of Some Complex Salts of Uranium with Halides of the Alkali and Alkaline Farth Metals, Brown Univ,.,, M-251, Kraus, C. A., 20 pp., SECRET, July 1, 1943. n"Temperature-Composition Diagrams of Metal-Metal Halide Systems," Eastman, E. D., Cubicciotti, D. D., and Thurmond, C. D., pp. 6-12 in The Chemistry and Metallurgy of Miscellaneous Materials, Thermodynaemics, N.N.E.S., Div. IV, P.P.R., Vol. 19B, Quill, L. L., Editor, 329 pp., UNCLASSIFIED, (McGraw-Hill, 1950). "Metal-Salt Interactions at High Temperatures: The Solubilities of Some Alkaline Earth Metals in Their Halides," Cubicclotti, D. D. and Thurmond, C. F., J.A.C.S. 71, pp. 2149-2153, (1949). "Metal-Salt Interactions at High Temperatures: The Cerium Cerium Chloride System," Cubicciotti, D., J.A.C.S. 71, pp. 4119-4121, (1949). c{ -98- AR The short-term nature of this project made it imperative that we draw on exigting information at other installations wherever possible. We are very grateful for the active co-operation in this which we en— countered in every case, The assistance of the New York Operations Office of the AEC in many phases of our work proved invalusble, The Division of Reactor Develop- ment and the Qperations Analysis Staff of the AEC in Washington con- tributed much-needed advice and date. We alsoc wish to express our ap— preciation of the generous amounts of time and information extended to members of our staff gt the following AEC installations and Contractorst offices: Argonne National Laboratory Dow Chemical Co. Battelle Memorisl Institute Iowa State College Brookhaven Nationel Laboretory Knolls Atomic Power Laboratory Detroit Edison Co, North American Aviation Co, University of California Advance reactor information obtained from other groups saved a great deal of time here and served to expedite comparison with project reactors., We are indebted to the Washington Office of the AEC for data on the Jumbo and Aqueous Homogeneous Reactors and to the KAPL staff for information about their Pin-Type Fast Reactor. The members of the project!s Advisory Committee were Harvey Brooks, John Chipman, Charles D. Coryell, Edwin R. Gillilend, and Harold S, Mickley. This committee, in addition to its assistance in shaping the general course of the project, made many valusble, specific contributions to the wvork. Among the project's consultants, Henry W. Newson provided many useful suggestions, particularly in the field of reactor controls. Warren M, Rohsenov spearheaded the mechanical design activities. David D, Jacobus helped crystallize structural designs of the reactors. George Scatchard and Waelter Schumb contributed to and coordinated the work in chemistry, Herbert H. Uhlig provided valuable interpretation of corrosion problems. » reedbhe 10. -Q7- "The Solubility of Cadmium in Mixtures of Cadmium Chloride with Other Chlorides," Cubicciotti, D., J.A.C.S. 74, 1198, (1952). High Temperature becontamination and Separation Processes, Univ. of Calif., UCRL-314, Brewer, L., 11 pp., SECRET, May 6, 1949. toen Ve are grateful to Karl Cohen of the Walter Kidde Nuclear Labora- tories for allowing two physicists from his staff to join the project for the summer. Lee Haworth of the Brookhaven National Laboratory also helped at a critical time vwhen he made it possible for Dr. Jacobus to spend a week with the project, As Executive Officer, Williem E. Ritchie facilitated the work of the project in many ways and helped it to work effectively.