WASTER //,g Symposium on * REPROCESSING OF NUCLEAR FUELS sssssssssss llllllllllllllllllllllllll Institute of Metals Division The Metallurgical Society of AIME aaaaaaaaaaaaaa U. §. Atomic Energy Commission i p2td ION / Divi ical Information WSTRIBUTION Y @ RO LEGAL NOTICE This report was prepared as an account of Government sponsored work. Neither the United States, nor the Commission, nor any person acting on behalf of the Commission: A, Makes any warranty or representation, expressed or implied, with respect to the accu- racy, completeness, or usefulness of the information contained in this report, or that the use of any information, apparatus, method, or process disclosed in this report may not infringe privately owned rights; or B. Assumes any liabilities with respect to the use of, or for damages resulting from the use of any information, apparatus, method, or process disclosed in this report. As used in the above, ‘“‘person acting on behalf of the Commission® includes any em- ployee or contractor of the Commission, or employee of such contractor, to the extent that such employee or contractor of the Commission, or empioyee of such contractor prepares, disseminates, or provides access to, any information pursuant to his employment or contract with the Commission, or his employment with such contractor. USAEC Divisian of Tecknies! Infomarion Extention, Ock Ridge, Tonntues DISCLAIMER This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency Thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. DISCLAIMER Portions of this document may be illegible in electronic image products. Images are produced from the best available original document. LEGAL NO This report was propnred as an account of Government Spo States, nor the Commissiot, noT any persod acting on pehalf of A Makes any warranty of repraunuuon. enpressed oF jriplied. with Tespect 0 the BCCU- TAEY. completeness, of usefulnens ol the jnformation contained in his report or that the use of any \nformation, apparatus, method, oF proceas disclosed in this report may pot infringe g T B. Assumes any Jiabilities with respect to the use of, of {or damages resulting from the use of any information. apparatus, method, 0T process disclosed in this report As used in the above, +iperacn acting on behalf of the Commlu\on" includes any em- ployee 0T contractor of the Commisajon, or employee of such contractor, to the extent that such employee or contractor of the Commission, o7 employe® of such contractor prepares, disseminates, oY prov!dal access 1o, any information purlum. w his emp‘loyment or contract with the Commission. ar his employment with such contractor CHEMICAL SEPARATICOONF_69080] NS PR FOR PLUTONIUM AND UHAN?SIE’ISSES {T1D-4500) NUCLEAR METALLURGY, VOLUME 15 SYMPOSIUM ON REPROCESSING OF NUCLEAR FUELS Edited by P CHIOTTI Containin n A ining papers presented August 25, 26, 27, 1969 ’ El ! at | owa State University, Ames, lowa Jointly sponsored by NUCLEAR METALLURGY COMMITTEE INSTITUTE OF METALS DIVISION THE METALLURGICAL SOCIETY American | nstitute ot Mining, Metallurgical, and Petrole E um Engineers, Inc J T. Waber, Committee Chairman ond L »:MES LABORATORY of the .S. ATOMIC ENERGY COMMISSION R S HANSEN, Director ki AULION 8 (RIS QOCUMERR 8 UNW \(fi“ Available as CONF-690801 from Clearinghouse for Federal Scientific and Technical Information National Bureau of Standards, U. S. Department of Commerce Springfield, Virginia 22151 $3,00 Printed in the United States of America USAEC Division of Technical Information Extension Oak Ridge, Tennessee 37830 August 1969 iv THE METALLURGICAL SOCIETY S Paul Queneau, President Michael Tenenbaum, Past President William J, Harris, Jr., Vice President Carleton C. Long, Treasurer Jack V, Richard, Secretary INSTITUTE OF METALS DIVISION John H, Rizley, Chairman Donald J. McPherson, Past Chairman F. Lincoln Vogel, Jr., Vice Chairman William C. Leslie, Vice Chairman C. F. Stewart, Jr,, Secretary-Treasurer 1969 NUCLEAR METALLURGY COMMITTEE SYMPOSi1UM ON REPROCESSING OF NUCLEAR FUELS P. Chiotti, Chairman J. F. Watson, Co-chairman E. N. Aqua, Co-chairman CONFERENCE SESSION CHAIRMEN D. E. Ferguson, Technology and Economics of Aqueous Processing R. C. Vogel, Technology and Economics of Nonaqueous Processing S. H, Smiley, Halide Volatility Processes W. R. Grimes, Fused Salt-Liquid Metal Systems J. F. Smith, Basic Data and Thermodynamic Properties, |1 Irving Johnson, Basic Data and Thermodynamic Properties, 11 Morton Smutz, Chairman of Panel Discussion ACKNOWLEDGEMENTS The organization and the successful conclusion of this symposium were the result of the efforts of many individuals, The Symposium Committee wishes to express appreciation for the advice and guidance of the members of the Nuclear Metallurgy Committee of The Metallurg- ical Society; W, F, Sheely and J, T, Waber, Chairmen during the period of organization and completion of the symposium; Hurd Hutchins, Jr., Editor, TMS; J, V, Richard, Secretary, and C, F, Stewart, Jr.,Assist- ant Secretary, TMS. Many members of the Ames Laboratory rendered valuable assistance in planning details, scientific and technical editing of the symposium papers and in secretarial work., Special thanks are due to Morton Smutz, Deputy Director of the Laboratory; W, E. Dreeszen, Head, Information and Security Department; J. F. Smith, Chairman, Metallurgy Department; W. H. Smith for invaluable editorial assistance; and Verna J. Thompson for most of the typing and secretarial work, We also wish to acknowledge gratefully the cooperation of R. C. Dreyer and other members of the Division of Technical Information Extension, Oak Ridge, Tenn., for making the volume available in time for the symposium and the cooperation of the many authors who made this symposium possible. Unfortunately a few papers, for good and sufficient reasons, were not ready in time to be included in the symposium volume. vi PREFACE The purpose of this symposium is to review and bring into focus experience, progress and basic information on nuclear fuel reprocessing methods. The principal emphasis is on nonaqueous methods. To lend perspective and orientation, one session is devoted to the technology and economics of the more conventional and well- established aqueous processing methods. This session is followed by a series of papers describing pyrometallurgical or nonaqueous methods. Much progress has been made in the development of processes based on halide volatility and on oxidation-reduction re- actions in metal-oxide and fused salt-liquid metal systems. This is a very broad and potentially fertile area of research which ex- tends beyond the area of fuel reprocessing. Fused salt-liquid me- tal systems also are of interest in such areas as the development of fuel cells, application of wear or corrosion resistance coatings or cases on metals, methods for inorganic as well as organic synthesis, etc. The last two sessions are devoted to basic data and thermody- namic properties. Phase relations and thermodynamic data are particularly helpful in the design of new and better reprocessing methods or in the improvement of existing methods. Often the necessary data are not available and special emphasis has been Placed on the evaluation or estimation of thermodynamic data from Phase diagrams. The papers on thermodynamics have been contributed by indi- viduals from various disciplines and backgrounds. It is apparent from these papers that there is a dire need for standardization of the nomenclature employed for the thermodynamic description of solutions. Time did not permit the adoption of a common nomen- clature for the papers published in this volume. The interested reader must bear the burden of deciphering the symbolism employ- ed not only in this volume but in the published literature in general. Hopefully the various disciplines will evéntually adopt a common formalism. CONTENTS TECHNOLOGY AND ECONOMICS OF AQUEOUS PROCESSING (,f APPLICATION OF AQUEOUS REPROCESSING TO LIQUID METAL FAST BREEDER REACTOR FUEL 3 W. E, Unger, R. E. Blanco, A. R. Irvine, D. J. Crouse and C, D. Watson } AQUEOUS REPROCESSING OF THORIUM- CONTAINING NUCLEAR FUEL ELEMENTS 25 G. Kaiser, E. Merz and H. 7J. Riedel PROGRESS IN TECHNOLOGY AND ECONOMICAL ASPECTS OF AQUEOUS PROCESSING 37 P. Michel FUEL REPROCESSING IN INDIA - TECHNOLOGY AND ECONOMICS 39 H. N. Sethna and N. Srinivasan TECHNOLOGY AND ECONOMICS OF NONAQUEOUS PROCESSING MELT REFINING OF ERB-II FUEL 57 D. C. Hampson, R. M. Fryer and J. W. Rizzie REMOTE REFABRICATION OF ERB-II FUELS 77 M. J. Feldman, N. R. Grant, J. P. Bacca, V. G. Eschen, D. L. Mitchell and R. V. Strain PREPARATION AND PROCESSING OF MSRE FUEL 97 J. M. Chandler and R. B. Lindauer ENGINEERING DEVELOPMENT OF THE MSBR FUEL RECYCLE 121 M. E. Whatley, L. E. McNeese, Ww. L., Carter, L. M. Ferris and E. L. Nicholson EXPERIMENTS ON PYROCHEMICAL REPROCESSING OF URANIUM CARBIDE FUEL 123 G. E. Brand and E. W. Murbach ix )Zz TECHNOLOGICAL AND ECONOMICAL ASPECTS “% OF IRRADIATED FUEL REPROCESSING BY {FLUORIDE VOLATILITY METHODS IN FRANCE G. Manevy and Y. Rochedereux SIZING THE CHEMICAL REACTORS FOR FLUORIDE VOLATILITY PROCESSING OF FAST REACTOR FUEL G. J. Spaepen g % REPROCESSING OF THTR FUEL ELEMENTS BY - HIGH TEMPERATURE TREATMENT 53; J. Hartwig and K. H. Ulrich HALIDE VOLATILITY PROCESSES L PILOT PLANT EXPERIENCE ON VOLATILE FLUORIDE REPROCESSING OF PLUTONIUM M. A. Thompson, R. S. Marshall and R. L. Standifer LABORATORY-DEVELOPMENT OF THE FLUORIDE VOLATILITY PROCESS FOR OXIDIC NUCLEAR FUELS M. J. Steindler, L. J. Anastasia, L. E. ( Trevorrow and A. A. Chilenskas - ENGINEERING-SCALE FLUORIDE VOLATILITY STUDIES ON PLUTONIUM-BEARING FUEL MATERIALS N. M. Levitz, E, L., Carls, D. Grosvenor, G. J. Vogel and I. Knudsen | THE POTENTIAL OF THE FLUORIDE VOLATILITY PROCESS FOR FAST BREEDER REACTOR FUELS A, A, Jonke, N. M. Levitz and M. J. Steindler [ CHLORINATION-DISTILLATION PROCESSING OF IRRADIATED URANIUM DIOXIDE K. Hirano and T. Ishihara y DIRECT CHLORINATION VOLATILITY PROCESSING OF NUCLEAR FUELS-LABORATORY STUDIES A. V., Hariharan, S. P, Sood, R. Prasad, D. D. Sood, K. Rengan, P. V. Balakrishnan and M. V. Ramaniah 141 143 159 163 177 211 231 241 261 y FUSED-SALT FLUORIDE-VOLATILITY PROCESS FOR RECOVERING URANIUM FROM THORIA-BASED FUEL ELEMENTS W. Bannasch, H. Jonas and E. Podschus FUSED SALT-LIQUID METAL SYSTEMS / EBR-II SKULL RECLAMATION PROCESS gl I. O. Winsch, R. D. Pierce, G. J. Bernstein, W. E. Miller and L. Burris, Jr. 8TATUS OF THE SALT TRANSPORT PROCESS FOR FAST BREEDER REACTOR FUELS R. K. Steunenberg, R. D. Pierce and I. Johnson L{JRANIUM AND PLUTONIUM PURIFICATION BY THE SALT TRANSPORT METHOD J. B. Knighton, I. Johnson and R. K. Steunenberg L/THE REDUCTIVE EXTRACTION OF PROTACTINIUM \“\ AND URANIUM FROM MOLTEN Li.F-BeJ:TZ-';l‘hF4 MIXTURES INTO BISMUTH R. G. Ross, W. R. Grimes, C. J. Barton, C. E. Bamberger and C. F. Baes, Jr. THE REDUCTIVE EXTRACTION OF RARE EARTHS FROM MOLTEN LiF-BeFZ-ThF4 MIXTURES INTO BISMUTH J. H. Shaffer, D. M. Moulton and W. R. Grimes THE MOLTEN SALT EXTRACTION OF AMERICIUM FROM PLUTONIUM METAL J. L. Long and C. C. Perry BASIC DATA AND THERMODYNAMIC PROPERTIES, I COMPATIBILITY AND PROCESSING PROBLEMS IN THE USE OF MOLTEN URANIUM CHLORIDE-ALKALI CHLORIDE MIXTURES AS REACTOR FUELS B. R. Harder, G. Long and W. P. Stanaway 279 297 325 337 363 375 385 405 TECHNOLOGY AND ECONOMICS OF AQUEOUS PROCESSING Chairman: Don E. Ferguson Oak Ridge National Laboratory Oak Ridge, Tennessee, U.S.A. WELCOME ADDRESS Robert S. Hansen Director, Ames Laboratory, USAEC Iowa State University Ames, Iowa U.S.A. INTRODUCTORY REMARKS Stephen Lawroski Argonne National Laboratory Argonne, Illinois U.S.A. APPLICATION OF AQUEOUS REPROCESSING TO LIQUID METAL FAST BREEDER REACTOR FUEL™ W. E, Unger, R. E, Blanco, A, R, Irvine, D, J, Crouse, C, D, Watson Oak Ridge National Laboratory Oak Ridge, Tennessee U.S. A. Abstract The low-priced power that is one of the incentives for the develop- ment of the Liquid Metal Fast Breeder Reactor (IMFBR) depends in part on economic processing of spent fuel. Some form of the Purex Aqueous Solvent Extraction process is used by nearly every major fuel process- ing facility in existence, and it will be desirable to extend its application to include the processing of the IMFBR fuels, These fuels are inherently more difficult to process than present light- water reactor (IWR) fuels, owing to the fact that IMFER fuels will operate at higher specific powers and for greater burnup and will contain more plutonium than the corresponding IWR fuels, The higher specific power and burnup produce heat dissipation problems during handling and transport of the spent fuel, The high plutonium content makes short cooling a significant economic incentive, and development work will be aimed at improving our technigue for accom- modating short-lived fission products to make short-cooled process- ing feasible, # Research sponsored by the U, S, Atomic Energy Commission under contract with the Union Carbide Corporation, Introduction This paper presents a description of the problems involved in the application of aqueous chemical processing technology to the reproc- essing of spent Liquid Metal Fast Breeder Reactor (IMFBR) fuels. The present status of technology is also presented, based on the development program in progress at Oak Ridge National Laboratory and on technology available from private and AEC reprocessing instal- lations, The objectives of the ORNL development program are: first, to reduce, with time, the uncertainties that exist in applying present aqueous processing technology to future IMFBR fuels; second, to pro- vide the technology necessary to adapt IWR fuel processing plants to the processing of IMFBR fuels in the interim period before the IMFBR fuel load increases to the point that it can support a separate proc- essing plant; and to provide the technology necessary for eventual commercial IMFBR fuel reprocessing plants, The latter, long-term approach is reserved for those problems that are occasioned by the economic incentive for short-decay processing and which are of such technical complexity as to require an extensive development period. The first fuels to be reprocessed will be produced in the Fast Flux Test Reactor (FFTF) and the demonstration reactors now being proposed, The Atomic International Reference Oxide Reactor (NAA- SR-Memo-126CL), has been selected as the reference reactor for the initial reprocessing development program, since it is representative of many of the new problems in reprocessing, which are not present or are not as significant with light water reactors (IWR) fuels. These problems are derived from four factors: (1) the desire to minimize inventory costs and reprocess the fuel as soon after dis- charges as possible, which causes higher thermal power, radiation level, and concentrations of important volatile fission products; (2) a high concentration of plutonium, which requires special consid- eration for criticality control; (3) the presence of liquid sodium on, or in, the fuel rods, and () the dimensions and method of fabri- cation of the fuel assemblies, which present problems in preparing, or disassembling, the fuel prior to reprocessing. The present intention is to modify the current technology for shipping and reprocessing IWNR fuels to the extent required for IMFER fuels., The Purex solvent extraction process is used in all major fuel reprocessing facilities in existence and its important favor- able features also apply to the processing of IMFBR fuels., The unexcelled separation efficiency, the process versatility, the ease of adaptation of the process to continuous high-capacity equipment, and the vast operating experience available from major processing facilities make the aqueous solvent extraction process an obvious choice for adaptation to reprocessing of IMFER fuel. Purex Process The Purex proces, as applied to IWR oxide fuels (see Fig, 1) con- sists of fuel shearing to rupture the corrosion-resistant sheath and expose the fuel, dissolution in nitric acid, solvent extraction, and conversion of the uranium and plutonium nitrate product to oxides for refabrication into fuel elements, The spent fuel is transported from the reactor to the chemical processing plant in shielded casks, which are unloaded in a water-filled pool or canal. The fuel elements are transferred to a head-end cell, sheared into approximately 1/2 in. lengths and the product leached with nitric acid in batch dissolvers, The residual sheared, leached hulls are disposed of as solid waste. The nitric acid solution of the fuel, containing uranium, plutonium, and nearly all of the fission products, is the feed solution for the solvent extraction process. The solvent for the Purex process is an organic complexing com- pound, tributyl phosphate (TBP), in an inert hydrocarbon kerosene- like diluent such as dodecane, The solvent is brought into intimate countercurrent contact with the aqueous feed solution where the TBP extracts the uranium and plutonium into the organic phase, leaving the fission and corrosion products in the aqueous solution, The latter is stored as high-level radioactive waste. The organic solu- tion is selectivity stripped of plutonium with dilute nitric acid by reducing the plutonium valence from +4 to +3. The uranium is then stripped in a third contactor, The plutonium is further purified by additional extraction cycles or by ion exchange. The uranium and plutonium may be precipitated from their dilute nitric acid solution and converted to oxides by thermal decomposition, or they may be directly converted to oxides by the sol-gel process, in which the nitric acid is removed from aqueous solutions of pluto- nium or uranium by extraction with an amine or heavy alcohol., In the sol-gel process the uranium or plutonium, as the acid extraction proceeds, gradually forms into a colloidal suspension or sol, This can be handled like a true solution. Progressive removal of water by evaporation or by extraction with a hygroscopic solvent converts the sol to a plastic gel. Sols of plutonium and uranium can be combined and gelled together to form an oxide in which the two ele- ments are homogeneously dispersed, The gel when fired to ~ 1200°C approaches theore%ica density and is suitable for fabrication into reactor elements,‘\*’?®) The sol-gel process is designed to couple easily with the Purex process for preparing plutonium for recycle, However, uranium from IMFBR fuels has little value and will probably be stored, Makeup uranium will be supplied from diffusion plant tailings for many decades, IMFBR Fuels It is expected that oxide fuels sheathed in stainless steel will be a characteristic of the early IMFBR reactors, and, for this reason, AQUEOQOUS AQUEOUS SCRUB FEED SOLUTION 3 s (8] ! [+ < - » ORGANIC w EXTRACTANT AQUEOUS WASTE 1, Purex Flowsheet, PARTITIONING SOLUTION R §°.thERT GAS B S5 COOLANT x,;?" °-09(>_" L) 5 T - 0. Pyl ] D Qi'B 04z LAY slet ' P Oy A h i) 850 i3] - 5530, st UN#SAR\PING '5& 3R d e 47 I R A A O A T T ot g 3 : fg‘ PLASTIC BAGGING (ke v TO0YR X ] Ao S 2ol : SOl SLEEVE LRI, o g SHIPPING CASK— O .i' > - 2 ] L) e — ! ] ', r s H [ ot O O l 2! : M "s == L —= & — — — 4 —_— = e——-—— D —— == eV IIEE = = 2 o jaf{ o a o ojn{o o 5. Receiving and Handling, 14 ARGON BLOWER AND HEAT EXCHANGERS CASK-TO-CELL ACCESS PORT, CELL AIR LOCK {1} n _.? C=— T = S e o AN N P———————- /r 22y et } CLEANED FUEL STORAGE MECHANICAL REPROCESSING CELL CLEANED FUEL TRANSFER MECH and Storage Facility). Central Reprocessing Plant for Spent IMFBR Fuel (Conceptual Fuel Receiving, Unloading, Cleaning, 6. 15 91 Table 2, Effect of Parameter Variation on Fuel Shipping Temperatures (°F) Shielding-to-Fuel Basis for After Fire Accident Ratio Variable Comparison Condition Outside Skin Max, Pin (Ws/Wg) Decay time 36 subassemblies 20-day-cooled 710 1155 Lo 90-day-cooled 510 805 28 Excess 18 subassemblies, Full length (17 ft 8 in.) 555 895 56 hardware 20-day-cooled Cropped (7 £t 8 in,) 800 1291 L9 Shipment 20-day-cooled 6 subassemblies L0O0 610 106 quantity 18 subassemblies 555 895 56 36 subassemblies 710 1155 L0 Shielding 18 subassemblies, Uniform shielding 520 872 83 design 20-day-cooled Graded shielding 555 895 56 Note: Unless otherwise noted, all shielding is graded and fuel subassemblies are shipped with all hardware intact. cladding or dissolution of the cladding in molten metals such as zinc or antimony-copper alloy are interesting alternatives for some possi- ble cladding materials, but both require the physical separation and recovery of the fuel oxides, a difficult step, If it is practical to disassemble the fuel element into individual rods or small clusters, the heat dissipation problem is ameliorated and the mechanical shear can be small, relatively inexpensive, and easily maintained (see Fig, 7). Disassembly involves trimming the end fittings with an abrasive saw and slitting the shroud with a milling slitter, The wire-wrapped pins can then be handled in arrays of about 25 without exceeding the maximum temperature of 1300°F, If spring clip spacers are used in place of wire wrapping, an additional removal step to free the rods will be required, The problem of the retention of *3'I, encountered even in the processing of IWR fuels of 150 days gdecay, is an important problem with IMFBR fuels., ITodine retention factors are required that are in the range of 10%® for 150-day-cooled FFTF fuel (near the range of present practice) to 10° for future IMFBR fuel processed after 30 days decay. The chemistry of iodine is complex and much more knowl- edge is needed of the behavior of iodine in the various conditions encountered in a fuel processing plant before a treatment system can be designed with the reliability required for IMFBR fuels, As an example, the presence of organic vapors and organic iodides in the off-gas system reduces the efficiency of iodine removal steps. Early studies indicate that the organic materials can be eliminated from the off-gas by conversion to CO,, H,0, and I, in a catalytic oxidizer using Hopcalite catalyst (oxides of copper and manganese) at 150 to 500°C, A catalytic-oxidizer-charcoal-adsorber system was tested in the Transuranium Processing Plant (TRU) at Oak Ridge National Laboratory and an iodine retention factor of about 10° was demonstrated, It has been found that fission gases interstitially trapped in oxide fuels can be volatilized and collected in a relatively small volume when oxide fuels are heated in oxygen at temperatures greater than ~ 400°C., Similar volatilization is expected on oxida- tion of carbide or nitride fuels, Early studies show that nearly quantitative removal of noble gases and >90% removal of iodine and tritium is obtained on conversion of U0; to Uz0g;. The release of gases is more difficult when the fuel contains a high percentage of Pu0, (~ 20%)., However, >90% of the tritium and 20-98% of the krypton have been removed from highly irradiated 20% PuQ,-U0,. The removal of iodine from PuO,-UO, remains to be demonstrated. The efficiency of iodine removal from off-gases should be much higher if it is contained in a small volume of gas after the fuel is sheared but before it reaches the dissolver, This should greatly reduce the problems attributed to removing iodine from dissolver off-gases in current processing experience, The oxidative heat 17 51n. HYDRAULIC CYLINDER FEED MECHANISM 15/ 1n. SQ. (INSIDE DIM.) FEED MAGAZINE FIXED KNIFE {EACH OF FOUR FACES CAN BE USED) TRAVELING KNIFE 3. SQ. (EACH OF FOUR FACES CAN BE USED) SHEAR HOUSING 9% n. 5Q. 25 FUEL TUBES ] ] o J 7S (=] W @ I V] X w Tube Bundle), ORNL Semicontinuous Prototype Shear (5 x 5-in, 7. 18 treatment may have other advantages as well, Sodium contamination (failed fuel rods) would be deactivated, and the oxidation of U0, to Uz;05 involves a phase change which tends to break the oxide into a fine granular form, making the physical separation of the oxides from the stainless steel cladding a possibility and providing for a faster dissolution of the fuel in nitric acid, This improvement is probable but whether it applies also to the mixed oxide is yet to be determined, The oxidation of uranium oxide before dissolution also decreases the evolution of nitrous oxides in the dissolver off- gas., The oxidation step may also be & useful mea?s og converting the carbide and nitride advanced fuels to oxides, (7?8 Dissolution: The dissolution of solid solutions of (Pu,U)0, proceeds quite readily in nitric acid and the use of fluoride as a catalyst does not appear to be required, IMFBR fuel samples irradiated to 100,000 Mwd/metric ton have been successf?ll{ ?issolved in hot-cell tests at ORNL in 8 M HNOz in 4 to 8 hours,\®’*°/ Residues contained <0,2% plutonium and consist mainly of ruthenium and rhodium and insoluble stainless steel. Irradiated fuels prepared by the sol-gel, co-precipitation, and mechanical blending processes have been successfully dissolved, However, if a true solid solution has not been formed, the PuO, may dissolve too slowly in nitric acid, The sol-gel and co-precipitation processes inherently produce intimate mixtures of PuO, and U0,. Care must be exercised to assure homogeneous blending of cxides if a solid solution is to be produced. Elemental iodine is evolved rapidly from the hot nitric acid dissolving solution, TIodine present as the iodide oxidizes rapidly to elemental iodine, and that present as the iodate is reduced to elemental iodine by the nitrous acid formed during dissoclution of U0,. In small-scale tests at ORNL, > 99% of the iodine has been removed during dissolution by a proper arrangement of the condenser system (to avoid refluxing iodine back to the dissolver) and by adding iodide and nitrite to the system at the end of the dissolving period., It is desirable to remove iodine before extraction because of the difficulty of containing iodine in the subsequent processing steps, Extractions: Solvent extraction processes require organic extraction media, and organics are susceptible to radiation damage. However, radiation damage to the solvent, in spite of the high specific activity of the IMFBR fuels, should not be a significant problem, The estimated total single-cycle dose to the solvent in pulsed column operation is about 0,11 watt-hr/liter for the 15% TBP flowsheet 19 and about 0.25 watt-hr/liter for the 30% TBP flowsheet, These doses include an allowance of 20% of the P-y dose obtained in the extraction-scrub section to cover alpha irradiation and irradiation from ®*I (which may have accumulated in the solvent). The estimates are for columns operating in the organic-phase-continuous mode at 80% og flooding with an efficiency corresponding to an HETS of 2.5 £t.(°) "The British have demonstrated sustained solvent extraction where the single-cycle dose was about 1.l watt-hr/liter.(:l) By using high-speed conta%§gss, such as the centrifugal contactors in use at Savannah River the extent of solvent exposure to radia- tion can be held far below the limits that have already proved satisfactory in plant practice,{(®). These contactors also have high capacity and quick response and are applicable over a wide range of scale. On the bases of laboratory and hot-cell tests, we have concluded further that pulse columns, such as are currently used in many fuel processing plants, are adequate for IMFBR fuel processing. Separation of plutonium from uranium is accomplished in the Purex process by reducing the plutonium to the three-valent state to ' selectively strip it from the solvent. Ferrous sulfamate is used as the reductant in U, S. processing plants, However, IMFBR reactor fuels have a much higher plutonium content and, consequently, the use of ferrous sulfamate could contribute relatively large amounts of iron and sulfate to the waste. The sulfate also would interfere severely with plutonium extraction in the second TBP cycle. The use of ferrous nitrate, stabilized with a small concentration of a holding reductant such as hydrazine, has shown promise in small- scale cold tests,‘\'*) Tts use has the advantage of eliminating sulfate, although it still contributes iron to the waste (about as much iron as there is total weight of fission products). By appropriate con- trol of the acid concentration, the partitioning can be accomplished using as little as 25% of the stoichiometric amount of iron otherwise needed, reducing the iron contributed to the waste to about 20% of the weight of the fission products, Uranium{IV) has been studied extensively as a reductant by many investigators and is used on a production basis in France, It has the advantage of not adding metal contaminants to the waste, The U(IV) nitrate can be prepared by electrolytic reduction, by reduction with H, in the presence of platinum catalyst, or by photoactivated reduction with formaldehyde. The reaction rate for reduction with U(IV) is slower than for Fe(II), which may be of importance if short-residence contactors are used. Final purification of plutonium by extraction using secondary amines is an attractive alternative to anion exchange, the process used in present processing plants, A continuous countercurrent extraction process has obvious advantages over the batch ion exchange process, particularly where criticality considerations may be limit- ing and where the equipment must be remotely operated and maintained. 20 Shielding will probably be required, because of the a-n reactions and spontaneous fissions from the plutonium. In addition, extraction processes will couple readily to the preparation of reactor-grade oxides by the sol-gel process. Summarz The economic advantage of large-scale processing plants is So prominent that central processing facilities of capacities in the order of 5 tons/day or greater throughput are generally to be pre- ferred to a proliferation of small (< 1 ton/day) facilities. The continued expansion of the power economy and the increasing technol- ogy of power transmission makes large central power parks, in the order of 20,000 Mw, a plausible concept. An on-site processing facility to serwe this power park (~ 1 ton/day) begins to acquire favorable economics due to savings in fuel shipping end inventory costs, The essential processing problems are similar except for fuel transport. It is expected that plant design trend will be in the direction of high-capacity, small-volume equipment; this is equivalent to minimizing the plant inventory of both reactor fuel and process rea- gents, Continuous equipment (as opposed to the batch operations characterizing the industry in the past), and perhaps parallel lines to ensure operational continuity, will be easier to maintain and cheaper to operate, Minimizing the in-process inventory will serve both safety and economy considerations. Fuel casks are expensive but are most economical in large sizes (about 120 tons), Casks will be designed to ensure containment of the enclosed fuel throughout the postulated accidents that might occur during shipment, The cask seals will be designed to permit the carrier to be readily loaded and unloaded at the processing plant to minimize carrier turnaround time, Present mechanical shears are designed to accept entire subassem- blies, denuded only of their hardware. Were the fuel elements design- ed to be readily disassembled, preshipment disassembly might prove economical by decreasing the cask size and facilitating heat dissipa- tion during shipment. Smell, inexpensive, easily maintained, high- capacity shears will be especially attractive if the fuel is easily disassembled, The most noxious problem facing processing plants of the future is that of the volatile fission products, especially I, The volatile fission products, icdine, the noble gases, xenon and kryp- ton, and tritium, perhaps can be volatilized from oxide fuel at moderate temperatures (L50 to 750°C). If dilution by the cell atmosphere is minimized, the fission product gases can be very con- centrated, meking their capture and reduction to solid form efficient 21 and reliable, The advantages of continuous leachers have always been recognized, but the simpler batch dissolvers were not only adequate but even preferred for small plants, particularly where process control relied upon chemical analysis, Continuous leachers are presently under active development, in response to the obvious advantages of their small physical volume from the standpoint of criticality control, Countercurrent solvent extraction has been carried out in a variety of contactors: mixer-settlers and pulse columns and more recently in fast centrifugal contactors. The latter reduces radiation and chemical damage to the solvent, and reduces the solvent inventory (and therefore the fire hazards associated with the organics are also reduced)., Centrifugal contactors are so responsive that automatic control can be used to simplify operation. The cell ventilation and vessel off-gas systems are primary sources of routine and accidental radiocactivity release, Recycle of cell off-gas is feasible and will minimize the volume of off-gas needing routine treatment. Recycle will probably be quite economi- cal and the uge of an inert cell atmosphere (necessary for the sodium- contaminated IMFBR fuel) will become practical, The use of inert cell atmosphere throughout the plant will practically eliminate the hazard of solvent fires, In many cases the aqueous high-level radioactive wastes will con- tinue to be stored for interim periods in tanks, However, the trend is toward solidification and immobilization of these wastes as soon as possible, as dictated by safety and economics, The technology of solid ficat}on of wastes is now in the final stages of develop- ment, 1618 References 1. Chem, Technol, Div, Ann, Progr. Rept, May 31, 1968, ORNL-4272. 2. Chem, Technol, Div. Ann, Progr., Rept, May 31, 1969, ORNL-LL22, 3. E, L. Nicholson, Preliminary Investigation of Processing Fast- Reactor Fuel in an Existing Plant, oERNL;_—"TM-T'?'SET—’TEMay 19677, 4. Code of Federal Regulations, Title 10, Part 71, as published in the Federal Register, Vol, 31, No, 41, July 22, 1966, 5. A, R, Irvine, "Shipping Cask Design Considerations for Fast- Breeder-Reactor Fuel," Proceedigg; of the 16th Conference on Remote Systems Technology, Idaho Falls, Tdaho, March 11-T3, 1969, 22 11, 12, 13. L., 15, 16. B, C, Finney, R, S, Lowrie, and C, D, Watson, A Conceptual Design and Cost Estimate of an On-Site Facility for Cleaning, Disassembling, and Canning Short-Cooled LMFBR Fuel in Prepara- tion for Shipment to a Central Reprocessing Plant, ORNL TM-205C (May 19687, J. R. Flanary et al,, Nucl, Appl. 1, 219 (1965). C. Moreau and J, Phillipot, Compt. Rend, 256, 5366 (1963), W, E, Unger et al., Aqueous Processing of IMFBR Fuels Progress Report, March 15569, Wo, 1, ORNL TM-2552 (April 19697, W, E, Unger et al., Aqueous Processing of IMFER Fuels Progress Report, April 1989, No. 2, ORNL TM-2585 (May 1969). R, H, Alardice, "Reprocessing of Fast Reactor Fuels by Aqueous Methods, Part 2, Second and Third Generation Fast Reactors,® presented at Institutt for Atomenergi, Kjeller Research Establishment, Norway, August 1967. A. A, Kishbaugh, Performance of Multistage Centrifugal Contac- tor, DP-8L41 (October 15637, C. A, Blake, Jr,, Solvent Stability in Nuclear Fuel Processing: Evaluation of the Literature, Calculation of Radiation Dose, and Affects of lodine and Plutonium, ORNL-L212 (March 196B). D. E., Horner, The Use of Ferrous Nitrate as a Plutonium Reduc- tant for Partitioning Plutonium and Uranium in Purex Processes, - ebruary 7, 1969}, J. O, Blomeke et al,, "Estimated Costs of High-Level Waste Management," Proceedings of the Symposium on the Solidification and Long-Term Storage of Highly Radioactive Wastes, February 1L,-18, 1968, Richland, Washington, CONF-660208, pp. 830-L3. K. J, Schneider, Status of Technology in the United States for Solidification of Highly Radiocactive Liquid Wastes, BNWL~820 (October 1968), 23 AQUEOUS REPROCESSING OF THORIUM-CONTAINING NUCLEAR FUEL ELEMENTS G. Kaiser, E. Merz and H.J. Riedel Institut filr Cremische Technologle Kernforschungsanlage Jillich GmbH Germany Abstract The KFA-TBP 23/25-procese 1s being developed for the aqueous reprocessing of thorium-containing fuel elements. In the proposed head-end, a rapid and complete dissolu- tion of highly sintered (Th,U)O,-particles is achieved by digestion in molten potassium pyrosulphate. Fertlle materlal 1s lsolated as potassium sulphatothorate 1n a Eggcentration step and may be stored until most of the Th has decayed. 233Pa, the 27,4-day precursor of 233U, will be recovered by adsorption on powdered unfired vy- Egg. Uranium, consisting mainly of the isotopes 235U and U, 1s decontaminated by a TBP solvent-extraction pro- g8edure. A sultable flowsheet for an economical reproces- ping of the sulphate-contalning feed solutlion was deve- loped and tested in cold runs. * Work performed under a joint project sponsered by the German Federal Ministry of Science 25 Introduction Computer calculations on the long-term potential of high-temperature gas-cooled reactors in the future com~ pound network of the Federal Republic of Germany indi- cate that this reactfor type will secure a consglderable share of the generating capacity far beyond the year 2,000. The estimates for the year 1985 come up with a pa- wer output share of about 30, 000 MWg, nhia? should fur- ther increaseto about 250,000 MWe in 2010 (1), The first reactor ceneratlon ls expected to be opera- ble 1n the middle of the seventies. In order to guaran- tee a complete fuel eyele to the future reactor users, appropriate procedures for the reprocessing of spent fuel elements must be evolved in time. In the Federal Republic of Germany the development of the technology for reprocessing spent HTGR-fuels began in 1966. The results and experience gained to date 1nde- cate that in the first reprocessing period only a wet chemical process such as solvent-extraction can satis- factorily separate the feed and/or bred material from the fisslon products. This concluslen is not murprising since 1n 25 years' of experlence in reprocessing, only extraction processes have been practical on an industrial scale. KFA-TBP 23/25-Process Process~-Philosophy In planning the KFA-TBP 23/25-process we have been gulded by the following conslderations: 1. Fuel elements for high-teémperature gas-cooled reac- tors consist of pyrolytic carbon coated uranium and/ or thorium carbide or oxide partlcles dlspersed 1n a graphite matrix. On a technical scale the removal of all of the carbon, that means the structural graphite as well as the coating, appears feasible only by burs+ ning. 2, Mixed oxides should be preferable as the fuel partic- les because of cheaper production and better beha- viour under reactor conditions than the carbides. Their strong chemical resgistance against the only known dissolving agent, F -catalysed nltric acld, may require the application of a technologically ad- vanced solubilizing procedure. 26 3. The main part of the process. the solvent-extraction, shall be a TBP-procedure, Only when using the "clas- slc" extractlion medium whose technology has already been thoroughly investigated, will the setting up of a total process line be possible in the near future. i, For the present, an isolation of decontaminated tho- rilum can be renounced. The small share of the cost of the thorium on the complete fuel cycle costs makes 1t adgisible to store the used thorium until most of the 220Th has decayed. 5. The process should be capable of treating fuel ele- ments after short cooling periods. Therefore, a pro- cess step for the recovery and puriflication of the relatively long-lived 233U-precursor, 233Pa, has to be provlided. This point 1s partlcularily important with regard to the reprocessing of fuel elements from the pebble~bed reactor. The continuous loading and unloading of feed and bred elements has the highest economical efficiency only if the fuels can be re- processed immediately after belng discharged from the reactor., OQutline of the Process Based on the above considerations the flowsheet shown in figure 1 was developed. In the first step the crushed fuel elements are burned 1n a fluldized-solids furnace at temperatures be- tween 700 and 850 ©oC. The remaining thorium-uranium mixed oxide particles are then digested in a potassium pyrosulphate melt. Af- ter the reaction has been finished, the liquid melt will pneumatically be pushed out of the crucible and poured into the requlsite amount of water necessary for disso- lution. Insoluble fission product sulphates, principally ba- rium- and strontium sulphate, are separated by a cyclone separator or by fllters and the clear solutlion is then passed over a vycorglass-column to remove the protacti- nium stlll present in equilibrium. The resulting protactinium-free fuel solution of a now reduced specific activity level will be concentra- ted and adjJusted to TBP-extraction conditions by adding nitric acid and aluminium nitrate. The uranium concen- tration of the feed solution amounts to 20 g/l; it 1s 27 HEAD -END URANIUM RECOVERY AND PURIFICATION HIGR FUEL ELEMENTS 5 VOL % TBP BURNING W A FLUIDIZED Y URANIUM FISSION WASTE TREATMENT SOKDS FURNACE EXTRACTION -CYCLE PRODUCT! AND STORAGE £ Th IOy FISSION PRODUCTS 25,07 - SALT-MELT DIGESTION SPARINGLY SOLUBLE FISSION DieaOiuTION OF THE SALT-MELT 2 URANIUM PRODUCT - SULPHATES N ATER EXTRACTION- CYCLE DIL SULFURIC ACID M2C20% |procEss-saLuTION ADSORPTIVE PROTACTINUM PROTACTINIUM OXALATE SECARATION URANIUM TAIL-END HNOy AL(NO3)y POTASSIUM SULPHATC-THORATE CONCENTRATING AND { ZIRCONATE AND -CERATE ) FEED ADJUSTMENT J URANYL-NITRATE FEED SOLUTION Figure 1: KFA-TBP 23/25-Process; Schematlc Process-Flow- sheet 28 nearly free of thorium, because sparingly soluble potas- slum sulphatothorate 1s preclpitated in the concentra- tion step. Also 1ts content of zirconium and cerium should have decreased considerably compared to the inl- ttal produet solution, because like thorium both ele- ments form sparingly soluble double salts with potassiunm sulphate. The extractive separation and decontamination of the uranium with a solution of 5 vol ¥ tri-n~butylphosphate in Kerosene 1s achieved in the first extractlon cycle following a fiowsheet specially designed for the present f'eed conditilans. For the second uranium extraction cycle the applica~- tion of the published flowsheet designed for the TBP 25~ process 1s projected; the 3?93 1s intended for the ura- nium tall-end purification . Experimenta; Results Head=-End Burning No thorough description of the burning step will be given here since most of the de%a%ls are already known and partly published elsewhere (3), An essential diffe- rence of our approach compared to the other known me- thods 1s the fact that we are not using any alumina as a diluent and thermal transmltter in the fluidized bed. To avoid thermal hot spots close above the bottom flow plate, the burning gas 1is diluted by refluxing carbon dioxide. This method avolds the difficulties encountered in the dissolution of the ashes in the presence of alu- mina. Salt-Melt-Digestion The knowledge of the most effective reaction tempera- ture was of declslve importance for a rational carrying out of the pyrosulphate salt-melt dissolution. The tem- perature determines not only the kinetics of the reac- tions U0, + 21{2:3207 —> U0,80, + 2K,50, + S0, 1) ThO2 + 2K28207 —*4'Th(30u)2 + EKESOH 2) FP(oxide)+ nK28207 ———?FP(sulphate)n+ nkK, S0, 3) 29 bat also the degree of thermal decomposition of the po- tassium pyrosulphate K,5,04 > K,S0, + S0, 4) and thereby the minimum quantity of melt necessary per welght unit of mixed thorium-uranium oxide. Thorough in- vestigations showed that the most favourable temperature range lles between 700 and 750 9C. Under these conditions the pyrolysis of K 320 1s kept within tolerable limits and the reactions E) tZ 3) proceed fast enough to guaran- tee a complete dissolutiorn of 100 grams of fuel partic- &es in the 4.5-fold guantity of melting material within hours. Protactinium-Recovery The method of the sorptive protactinium pre-isolation originates from investigations of American sclentists some years ago. They succeeded in removing 97 % of the protactinium present 1in fiitrate solutions by sorptlion on powdered unfired Vycor (4). our own experiments have shown that this procedure can also be applied in the sul- phate system since the process solu%io? Setained only 3 - 4 9 of nonadsorbable protactinium (3). It should be pointed out that these data were obtalned with tracer amounts of protactinlum, however, we are qulte optimi- stlc regarding a separation of macro-quantities, since Z00DE and MOORE have already confirmed the{g tracer re- sults with mmol quantitles of protactinium ). Thorium-Iscolatlion and Feed~Adjustment With respect to the technical aspects of the process, the pyrosulphate salt-melt dissolutlion exhibits a cer- tain disadvantage 1n that very dilute solutions are ob- talned. The reason for this 1is the poor solubility of thorium sulphate, which under the conditions in question permits only thorium concentrations of about 5 g/l. Natu- rally, the content of fissile materiasl 1s comparably low too, and amounts to approx. 1 g/l with fuel particles having a thorium : uranium ratio of 5 : 1. Since the re- processing of such dllute sclutlons may hardly be done in an economical way, we declded to introduce a concen- trating step in ader to end up with solutions of 20 g/1 after the feed adjustment. The uranium losses observed in connection with the quantitative thorium precipitation are rather small; the sulphatothorate isolated in experi- ments carried out so far contalned always less than 0.1 % of the total uranium. 30 Filrst Uranlum Extraction.Cycle The chemical flowsheet developed for the first extrac- tion cycle is shown in figure 2. Extractlon~-Serub Unit The following operation conditions were chosen: feed solution with an uranium content of 20 g/1, 1 mol alumi- nium nitrate, 2.5 mol/1 of free acid and a sulphate con- centration of ¥ 0.5 mol/l. The exact sulphate concentra- tion depends on the uranium to thorium ratic of the dis- solved (Th,U)OQ-particles, 1t never exceeds however 0.5 mol/1. The aluminium nitrate concentration employed 1s a com- promise between economical aspects (minimum of solid waste, low costs of chemicals) and practical require- ments (low stage values and an uranium concentration in the ogganic phase as hlgh as possible in the feed input stage). The acld concentration of 2.5 mol/1l, which together with the acid from the scrub section results in a total acld concentration of 3 mol/l in the extraction sectiin was chosen to ensure a good ruthenlum decontamlnation 75. This fission product requires speclal attention in the first extraction cycle. A partlal pre-decontaminatlion of some other unpleasant flssion products like zirconium, niobium and cerium has already taken place durilng the concentrating step. The number of theoretical stages necessary for a bet- ter than 99.99 % recovery of uranium may be deduced from figure 3. As one can see, the necessary theoretical stages at a sulphate concentration of 0.5 mol/l1 amounts to 6. In practice about 8 stages are needed because of a 75 % efficlency of the individual stages. When the sulphate concentration is lower than 0.5 mol/l, the number of stages decreases, because the uranium distribution coef- flcients are increasing. For scrubbing of extracted fisslon product activities, 5 molar nitric acid is used. Again the high acild concen- tration serves for a good ruthenilum decontamination but it causes also a diminution of the zirconium-niobium and cerium content of the organic phase. As can be seen from figures 3 and 4 the operating line of the scrubbing sec- tion intersects the inherent uranium equillbrium line, 31 I AF 1 AX URANIUM 20 G/L TBP SVO_% ACID 25 MOL/L vO. 180 Al(NOC3 )3 ' MOL/L SULPHATE 05 MOL/L ’ VoL 100 | 4 EXTRACTION-SCRUB UNIT | l 1 AS I AW HNO3 5 MOL/L URANIUM = 0 01%, voL 27 ACID 2,97 MOL/L AlNO3)3 078 MOL/L SULPHATE & 04 MOL/L voL 127 I AU TBP 5 VOL% » URANIUM 0 G/L HND3 8,7-10-2 MOL/L voL 180 STRIPPING ~UNIT ! | 1 CX 1 Cu HNO3 107ZMOL/L URANIUM 27,5 G/L voL 72 HNO4 2,3 10 Mo ' VoL 72 1Cw 8P 5 vOL% URANIUM = 0,01% HNO 4 ~10"2 MOL/L EVAPORATOR voL 180 SECOND SOLVENT- EXTRACTION CYCLE SOLVENT RECOVERVI RECYCLE SOLVENT TO EXTRACTION-SCRUB UNIT Figure 2: KFA-TBP 23/25-Process; Chemical Flowsheet for First Uranium Solvent-Extraction Cycle 32 SCRUB-SECTION EQUILIBRIUM LINE SCRUB-SECTION OPERATING LINE, 9-0.15\ '00 EXTRACTION-SECTION [ % -— e EQUILIBRIUM LINE EXTRACTION -SEC TION OPERATING UNE.e-oJ 1 L~ ORGANIC PHASE URANIUM CONCENTRATION (g/l) ————= ) 0,01%L0SS 0 'R w0 o' M © 1of o’ o AQUEOUS PHASE URANIUM CONCENTRATION {g/l) ————= Figure 3: KFA-TBP 23/25-Process; Theoretlical McCabe-Thlele Dlagram for Extraction-Scrub-Unit 33 " 5 ] =z s 3 — L g | SCcRUB-SECTION EQUILBRIUM LINE g SCRUB - SECTION OPERATING LINE a \ - O n‘ J T R RS- £° 1 EXTRACTION-SECTION Z | EOUILBRILM LINE - =T L U 4 EXTRACTION - SECTION £ ] OPERATING LINE H r 4 2| =l 0 e ———rrrr —————rrr o' i o v’ AQUEOUS PHASE URANIUM CONCENTRATION (g/l) — Figure 4: KFA-TBP 23/25-Process; Theoretical McCabe-Thiele Diagram for Extraction-Scrub-Unit 34 i. e. a pinch~point operation is employed in this part of the extraction-scrudb unit. Stripping Unit Stripping of uranium is achieved using ©.01 molar ni= tric acid. Applying the flow condltions shown in figure 2, three theoretlcal stages, respectively four practlcal stages, are necessary to strip more than 99.99 % of the uranium out of the organic phase. Status of Development and Outlook The labosatory studies of the individual process steps are nearly completed. Therefore we have initlated the planning of a small pllot plant with a dally throughput of approx. 1 kg (ThU)O, which will be processing spent fuel elements in the H8t Cells of the KFA-JUlich by the end of 1970, The final judgement on the economy and the industrial applicability of the KFA-TBP 23/25-process awaits completlion of these studles - both experimental and pillot plant operation. 35 T. References Krdmer, H., Schulten, R. and Wagemann, X, Dle lang- fristige wlrtschaftliche Bedeutung des gasgekihlten Hochtemperaturreaktors, Nukleonik, Vol. 11, 1968, pchJ Fianary, J.R., Goode, J.H., Kibbey, A.H. Roberts, J.T. and Wymer, R.G., Chemical Development of the 25-TBP~ Process, ORNL~1993 (Revision 2}, 1956 Witte, H.0., Survey of Head~End Processes for the Re- covery of Uranium and Thorium from Graphlte-Base Reac- tor Fuels, ORNL-TM-1811, 1966 Moore, J.G. and Rainey, R.H., Separation of Protacti- nium from Thorium in Nitriec Acid Solutions by Solvent Extraction wlth Tributylphosphat or by Adsorption on pulverized unfired Vycor Glass or Sllica Gel, TID-7075, 1963 Kaiser, G and Coenegracht, 0., Aufarbeitung thorium- haltiger Kernbrennstoffe, 2zusammenfassender Berlcht fir den 1. Projektadbschnitt 1966 - 1968, pp. SO - 54 Goode, J.H. and Moore, J.G., Adsorption of Protactl- nium : Final Hot-Cell Experiments, ORNL-3950, 1967 Bruce, F.R., The Behaviour of Fisslon Products in Sol=- vent Extraction Processes, Progress in Nuclear Ener- gy, Series III, Process Chemistry, Vol. I, Pergamon Press, London 1956, pp. 130 - 146 36 PROGRESS IN TECHNOLOGY AND ECONOMICAL ASPECTS OF AQUEOUS PROCESSING P. Michel Centre D'Etudes Nucleaires de Fontenay-Aux-Roses 92-Fontenay-Aux-Roses France 37 iy 1 P A FUEL REPROCESSING IN INDIA - TECHNOLOGY AND ECONOMICS H.N. Sethna* & N. Srinivasan¥* Bhabha Atomic Research Centre Trombay, Bombay (India) Abstract The paper presents the approach to the establishment of reprocessing facilities in India in the context of the technological development and economic factors in a developing country. The impact of less expensive labour and construction costs and less developed transportation facilities on the size and distribution of reprocessing facilities is discussed briefly. The capital and operating costs for the Trombay demonstration plant and the es- timated costs for a 500 Kg per day plant under construction for BWR and CANDU type fuels are included. The plans for repro- cessing fast reactor fuels are also discussed. * Director, BARC and Member for Research and Development, Indian Atomic Energy Commission. **Head, Fuel Reprocessing Division, BARC. 39 Introduction The fuel reprocessing programme in India is developing in stages. The first demonstration plant (1) was set up in Trombay to reprocess the metallic uranium fuel irradiated in the CIRUS, a 40 MW heavy water moderated research reactor. This plant has been in operation since 1965(2). With the setting up of the power stations at Tarapur and Ranapratapsagar, a plant is under con- struction for the reprocessing of the BWR type fuel from the Tarapur power station and the Candu type fuel from the Ranapra- tapsagar power station. The third power station is under con- struction at Kalpakkam near Madras based on a Candu design. The Indian nuclear power programme also includes the setting up of a fast test breeder reactor at Kalpakkam. A reprocessing complex for the irradiated fuel from the Candu type power station as well as the fast test breeder reactor is planned for construction at Kalpakkam. The present paper attempts to outline the background to the decisions regarding the setting up of the fuel reprocessing plants in various locations in the country, the economics of the construction and operation of such facilities and the research and development programme pertaining to fuel reprocessing of thermal as well as fast reactor fuels. The fuel reprocessing plant at Trombay operates on the conven- tional Purex process with a codecontamination cycle, a partition cycle, a third uranium purification cycle and an ion exchange cycle for the final purification and concentration of plutonium. The plant has seven solvent extraction columns of the pulsed perforated plate type, two ion exchange columns for the plutonium cycle and 100 vessels including 10 evaporators for the concentration of intercycle products and high and medium active wastes. The process equip- ment involved fabrication of 100 tonnes of stainless steel. The piping inside the cells is about 100, 000 ft. The cells have heavy density concrete walls (220 Ib per cft) with thickness ranging from 5 ft at the dissolver end to 3 ft at the product end. The plant was constructed during 1961-64 at which time the cost of heavy con- crete complete with steel was Rs.17 per cft ($ 96* per cu. yard). The cost of fabrication of stainless steel equipment was Rs. 7, 500/- per tonne ($ 1550% per tonne). The over all capital costs of the Plant were as follows: Rs. Million ($ Million)* Cost of equipment, piping and services (including erection) 19.0 (4. 0) Civil Engineering Works (including plumbing, drainage and site preparation etc.) 12.0 (2.5) Electrical installation 2.5 (0.5) 40 Rs. Million ($ Million)* Ventilation and airconditioning equipment - 1.7 (0. 3) Charges towards plant and building design, supervision of fabrication, construction and installation 2.0 (0.4 37.2 (7.7 *Based Rs. 4.75 per $ prevalent when the plant was built. The plant cost can also be broken up broadly as follows: Rs. Million ($ Million) Main Process Building complete with process equipment, utilities, general services and plutonium facility 25.5 (5. 3) Waste Treatment & Storage Facilities 9.7 (2.0) Engineering costs 2.0 (0. 4) 37.2 (7.7) The operation of the plant requires 200 persons. About 1/3 of the plant is now being utilised for non-operational purposes, viz., for development work. The fixed charges are calculated at 12-1/2% on Rs. 26 million. Labour & Supervision Rs. 1.50 million Overheads Rs. 0.45 " Equipment & Stores Rs. 1.00 " Power, water & other services Rs., 0.40 " Fixed charges Rs. 3.25 " Rs. 6.60 million For the nominal capacity of 30 tonnes uranium per year the cost of reprocessing works out as Rs. 220/~ ($ 29) per kg uranium. For fuel irradiated to 1000 MWd tonne, the cost of processing with 41 respect to plutonium is Rs. 275 ($ 37) per gm upto its conversion to plutonium oxide. The operation of the plant has shown that a more economical plant with 6 solvent extraction columns (1A, 1S, 1BX, 1C, 2D & 2E), 2 ion exchange columns, 50 process vessels and 50, 000 ft of piping with half the cell volume of the Trombay Plant can be constructed for efficient operation with an O & M strength of 150 persons. Capital cost of such a plant at today's wages and prices would be about Rs. 34 million ($ 4.6 million - 1968). If such a plant is built now it would not use heavy concerete for cells nor would the thickness of the cell-walls be as high as has been used for the Trombay Plant. The cost of conventional concrete assumed is Rs. 12 per cft ($ 43 per cubic yard) for cell walls. At today's prices the cost of fabrication of stainless steel equipment would be Rs.12, 000 per tonne ($ 1600 per tonne). The cost of reprocessing in such a plant would work out as Rs. 200/kg ($ 27 per kg) of uranium and corresponding cost of Rs. 250/gm with respect to plutonium. ’ The experience in the operation of the plant has provided not only the knowhow for the design of future bigger plants but has in- dicated the fruitful areas of research and development for efficient utilisation of the resources allocated for the purpose. Power Reactor Fuel Reprocessing The first power station which is expected to go in full operation by the middle of 1969 is at Tarapur, 60 miles north of Bombay. This has two boiling water reactors of nett output 380 MWe usiag uranium oxide of initial enrichment varying from 1.8 % to 2.5 % U 5. The second power station of the Candu type is under construction at Ranapratapsagar near the city of Kota in Rajasthan. The first unit of 200 MWe will deliver power in 1970/71 and the second unit two years thereafter. The third power station which will also be of Candu type is under construction at Kalpakkam, 50 miles south of Madras. The first unit will deliver 200 MWe in 1973. The capacity of the second unit on this location is likely to be 200/300 MWe and the time schedule for this unit calls for completion in -1974/75. The power programme of the Indian Atomic Energy Commission visualises the addition of 200 to 500 MWe nuclear power every year in the form of Candu type of reactors. The approach to the reprocessing of irradiated fuel from power reactors is conditioned by non-availability of highly enriched uranium and the consequent urgent need for plutonium for the fast reactor programme to utilise the vast thorium reserves in the country, the variation in capital investment and operating costs of reprocessing plants compared with more advanced countries and the problems involved in transporting large quantities of radiocactive material over long distances in the country in the background of the develop- ment of rail and road transport. The location of power stations is based on the demand for power and the economics of nuclear power generation vis-a-vis conventional power stations. The scope for 42 the sizing and siting of reprocessing plants extends from small plants attached to each power station to a large scale central plant catering to many power stations. It is not always found feasible to have package plants in every reactor location on account of the limitations associated with the capacity of the environments to ac- cept discharge of liquid and gaseous wastes from reprocessing plants in addition to those generated from the nuclear power stations. While India has a fairly extensive railway net work the existence of two different gauges of railways and the already heavy goods and passenger traffic on the existing railway lines impose limitations for rail trans- port of irradiated fuels in some sectors. Roadways are not uniformly developed all over the country and in each instance a decision has to be taken about the feasible mode of transport. These factors coupled with the longterm nature of the decisions regarding the locations of future power stations preclude the possibility of a large size central reprocessing plant. The compromise of setting up regional repro- cessing plants near suitable nuclear power stations appears to be the solution, transporting irradiated fuel from nearby nuclear power stations wherever transportation by road or rail of irradiated fuel is safe and feasible. The first reprocessing plant for power reactor fuels, under construction at Tarapur, is based on this principle and will reprocess fuels from the boiling water reactor station which is only a kilometer away and from the Candu power station in Rajasthan 500 miles away from where a combination of transportation by road and rail is felt to be feasible. In the context of the economics of setting up of reprocessing plants in India the influence of size is not considerable above 1 tonne per day capacity, according to present day prices and skilled labour costs. Fig. 1 gives the estimated cost for construction of reprocessing plants in India for treating Candu type fuel elements irradiated to 8000 - 10000 MW days per tonne, the plant operating for 250 days. Approximately the cap'htal investment for plant follows the relation Rs. 75 x 10° (MT/day)"" =. In this connection a comparison of capital costs and reprocessing costs estimated in India with those estimated for more advanced countries say U.S.A., is revealing. The basic differences in the capital as well as the operating costs arises out of the wide differ- ences in the skilled labour cost which is lower in India by a factor 10 to 15 and the appreciable difference in construction costs. For instance, the cost of a cubic yard of concrete is reported to be $ 100 in the USA(3), against Rs. 12 per cft ($ 43 per cubic yard) in India. Similarly the cost of construction of a plant building com- plete with normal services in U.S.A. is estimated as $ 2.5 to 3 per cft(3) as against Rs. 7 per cft in India ($ 0.95 per cft). Engineering and commissioning costs for such plants form about 12 % of the total cost of construction in India whereas the figure mentioned for U.S.A. is above 20 %. Applying suitable weightage factors for reduced com- pliment of labour due to mechanised construction in the advanced countries one can estimate that a plant for 1 tonne a day that can be 43 set up in India for Rs. 75 million ($ 10 million) would cost $ 27 million in North America with no change in design. In the developed countries perforce it becomes necessary to incorporate instrumenta- tion and automation to reduce the labour costs for operation. Thus a plant of 1 tonne a day may cost $ 28 to 30 million in North America. This estimate is close to the reported cost of the N¥S pla.nt( . Fig. 1 gives the compa.riso& of costs for different capacities, based on published information ) and estimates for plants in India. A similar effect is seen in the operating costs as well. Fif- teen per cent charges on investment are reported for Canadian con- ditions(5) as against 12.5 % adopted for computation of Indian pro- cessing cost. Reprocessing cost in a 1 tonne a day plant operating for 250 days a year in India is estimated at Rs. 66 ($ 8. 8) per kg uranium as against Rs. 210 ($ 28) per kg in North America based on labour costs ratios with suitable weightage for decreased number of O & M staff and assuming all other charges like chemicals, consumables, spares, maintenance materials, power, water, etc., to be the same in both countries. fi‘ig. 2 gives the estimated op- erating costs in India and Canada'”/ and compares the operating costs with respect to the major components. It is seen that cost of re- processing one kg uranium in India follows approximately the re- lation Rs. 66 (MT/day)}* 'x' being nearer minus 0.6 below 1 MT/day and nearer minus 0. 45 above 1 MT/day. Incidence of fuel reprocess- ing costs on power costs in India decreases rapidly from 0. 45 p/Kwh to 0.17 p/Kwh when the capacity is increased from 100 kg to 500 kg per day. Thereafter the decrease is marginal; 0.12 p/Kwh for a 1 tonne a day plant and 0.06 p/Kwh for a 5 tonne a day plant (Fig. 3). From the above it will be clear that the considerations that govern the decisions regarding the capacity of reprocessing plants are quite different at least for the time being between India and the more advanced countries like U.S.A. and Canada. The unit repro- cessing cost obtained in India with a 500 Kg a day plant is obtainable only with a plant of 3 to 4 tonnes a day in developed countries. It is expected that for the next decade with the pressing demand on finance for capital investment it will not be economical nor techni- cally advantageous to set up reprocessing plants larger than 0.5 to 1 tonne a day in India. The Indian Nuclear Power Programme is based on the pro- duction of plutonium in natural uranium reactors and utilising the plutonium for conversion of thorium into uranium-233 ultimately leading to the uranium-233 - thorium cycle. This is necessitated by the limited resources of uranium and vast reserves of thorium. The two reactor systems of interest for the utilisation of plutonium are the Fast Breeder and the Molten Salt Breeder reactors. The need for separating enough plutonium for going over to these systems influences the decisions regarding the setting up of fuel reprocess- ing facilities. Normally, one would optimise the timing and the sizing of the reprocessing plants to ensure maximum load factor. 44 ~ AECL 3136 -3 ESTIMATED INDIAN COSTS <100 -] 2 - 80 2 60 400 8 % 1408 . 2 200f E 3 i 1208 | » g 100 . Q 8 80 4400 -l 3 s 1 o8& a 6% S 40 3 4 4 1 L L L L L L 02 04 06 10 2 4 6 CAPACITY - TONNES/ DAY 1. Comparison of estimated capital costs for reprocessing plants for Candu type fuel. 45 AECL INDIAN 3138 ESTIMATE TOTAL REPROCESSING cOST @ @ 400+ FIXED CHARGES ® @ LABOUR, SUPERVISION & OVERHEADS @ @ 440 200+ {20 2 ® = ® o g_wo- o 2 ® J1o ¥ 500'- @ - 5 | 1oy 40 B g .8 Q 20- ® @ g & d2 g w 10 ® 1 1 o2 04 Q6 1 2 4 ) 10 CAPACITY TONNES PER DAY 2. Comparison of reprocessing costs Candu type fuel - Purex process. 46 \ ———— |NDIAN \\ =——— CANADIAN { INFERRED) 03 40-4 -3 x E ~ x @ » o ¢ do.ad - X 8 Soef . o] 2 o 7] o w 2 3] lo-23 (7)) o w x a 3 w x « [N 0-1F & 01 1 L 1 L 1 1 | 0-2 04 0608 10 5 « 20 23 CAPACITY TONNES PER DAY 3. Reprocessing cost/Kwh (8000 MWD/Te) 30 % efficiency. 47 The initial slow pace of development of nuclear power, the setting up of our first few nuclear power stations in locations far apart over a vast country like India and the necessity for setting up regional processing plants do not, as discussed earlier, permit to avail of full load factor. The object essentially is to ensure that all plutonium produced in nuclear power stations is recovered as quickly as possible. The regional reprocessing plants are sized in a fashion that they can absorb the load from nuclear power stations in the region as and when they are set up. For instance the Tarapur Reprocessing Plant has been designed for a capacity of half a tonne uranium per day though the minimum load available is from the Tarapur Boiling water power station (25 tonnes a year) and the two units of the Ranapratapsagar power station (50 tonnes a year). This plant can very likely absorb additional load from another Candu type 400 MWe station in a location from where trans- portation of fuel should be feasible. By the present analysis, . transportation to this plant should be feasible from a power station accessible by road from a railhead in Central or Northern India. Similarly transportation of irradiated fuel should be possible from many of the southern parts of the country to the Fuel Reprocessing Plant proposed to be set up at Kalpakkam near Madras. Between these two reprocessing plants fuel from nuclear power stations of the Candu type upto a total of 2,000 MWe installed capacity can be processed. The Tarapur Fuel Reprocessing Plant The design of this plant is based on the Purex Process. A codecontamination cycle followed by partition cycle and two separate purification cycles for the uranium and plutonium are visualised. Purification cycle for plutonium will be chosen from among a TBP process and TLA process. The boiling water reactor fuel element is about 14 ft long and contains massive pieces of zircaloy and al- so0 various minor materials of construction other than zircaloy. The fuel element from the Rajasthan Power Station is only 19" long and does not contain any material other than zircaloy. The end Pieces are also not very thick. Because the plant would be treat-~ ing boiling water reactor fuel elements a decision in favor of chop- leach process was taken. In the Kalpakkam (Madras) Reprocessing Plant it is quite likely that if all fuel to be reprocessed in this facility is of the Candu type chop leach process would not be favoured and a chemical decladding process would be adopted. The choice lies among the Zirflex process, the Thermox process and sulfurjc acid dissolution. Information available on the Thermox process 0) and the few experiments carried out in Trombay on the oxidation of the zircaloy and separation of the insoluble oxide from the dissolver solution are quite encouraging for the adoption of the process. ,})n fact an extension of this process to a ""chemical chop' concept( would reduce the fine oxide to be handled and would leave a large portion of the zircaloy in compact solid form. 48 The Plant will have 11 to 14 columns depending on the plutonium purification procedure decided upon. The use of TLA appears attrac- tive on account of the fewer solvent extraction contactors involved, and hence a decrease in the cost of equipment. Experiments are bfgsng carried out in Trombay on the use of TLA for the final purification as well as on the use of uranous nitrate(?) for the partition and in the final purification cycle using TBP solvent extraction process. In any case, for partition, enough experience has been built up in Trombay on the use of ferrous sulphamate which can be used in the event of any difficulties in the use of uranous nitrate. The plutonium purification cycle at the Trombay plant uses ion exchange process. The performance with respect to decontamination as well as the capacity of the resin has been quite satisfactory. However, the final concentrations of the plutonium product have not been as good as reported. The concentrations obtained in plant scale operation are invariably lower by a factor upto 2 or 3 than those obtained in laboratory experiments under identical conditions. This experience has led to the exploration of the possibilities of other methods. Based on the results of experiments in progress, provision will be made for recovery of neptunium-237 from the third uranium cycle raffinate. The plant is proposed to be operated on high acid flowsheet for the first two cycles and at lower acidity in the third uranium cycle. For the reconversion of plutonium nitrate to plutonium oxide experiments are being conducted on continuous precipitation and continuous calcination. Experiments on the residence time indicated the possibility of designing criticality safe equipment for throughput of the order of 500 to 600 grams plutonium per hour. Fast Reactor Fuel Reprocessing It is generally contended that the economics of power gen- eration by fast breeder requires reduction in inventory of fissile material and hence can admit only a very short cooling period before the irradiated fuel is reprocessed. This contention is basically valid, but there are other areas of fuel cycle which contribute more significantly to fuel cycle costs. The proponents of short cooling time are of the view that reduction in fuel cycle cost by decreasing cooling time is far easier than reduction of cost in other areas of the fuel cycle. One has however to balance this with the problems involved in processing short cooled fuel elements and the context of such processing. The choice for fast reactor reprocessing lies between aqueous and non-aqueous methods. The handicaps adduced to aqueous pro- cesses are the degradation of the solvent and the consequent effects on reprocessing efficiency and decontamination factors, criticality problems and the problem of iodine accumulation and release. 49 Chemical solutions appear to be in sight for minimising the impact of solvent degradation on efficiency of fuel reprocessing by aqueous process. (10) This and the development of short residence time solvent extraction equipment will remove the limitations on aqueous processing due to solvent degradation. Criticality problems es- pecially with respect to the dissolver generally appear to be not insurmountable and can essentially be solved by proper designs and use of built in poisons in equipment. The iodine problem however remains to be properly understood and solved. The non aqueous processes, both the volatility method as well as the reductive salt transfer method using metal alloys, show great promise. The technological problems associated with such processes are not insurmountable in small scale units. It appears therefore that both the processes should be competative for cloge coupled systems whereas aqueous methods have potential for larger units. The problems associated with the transport of short cooled fuel may necessitate the setting up of close coupled plants in any case and the economics of aqueous versus non-aqueous methods will have to be worked out in this context. The process and technological problems associated with the fluidised bed fluoride volatility process are the behaviour of plutonium, the efficiency of conversion of plutonium to its hexa- fluoride, the decontamination factors with respect to the ruthenium, the quantitative separation of plutonium from uranium and their mutual contamination factors, the clogging of off gas filters during decladding of zircaloy clad fuel by hydrochlorination or hydro- fluorination and the heat removal during the fluorination. As mentioned earlier these problems can probably be solved if equip- ment used are not complicated in shape and design and the capacity is limited. The fluidised bed fluoride volatility process is essenti- ally suitable for uranium-235 systems and the introduction of plutonium into such a system perforce introduces some problems. While the number of pieces of equipment required for fluoride volatility system is less than what is required for an aqueous pro- cess, the materials of construction are more expensive and the problems of fabrication greater in the light of very stringent re- quirements of corrosion resistance to fluorides at high temperatures. A difference in capital investment by a ratio of 3 : 2 is expected between aqueous and volatility procesges in a plant close coupled to a 1000 MWe fast reactor station. (llf A reduction in operating costs by 50 % is also estimated. Such significant ad- vantage in capital costs may not be available in the Indian context of lower unit costs of fabrication of equipments and installation of piping. Similarly the major contribution to decrease in operat- ing costs is from reduced labour force required. The already low labour costs in a developing country dilutes the impact of this aspect of cost reduction. 50 Madras Fuel Reprocessing Complex As mentioned earlier, a reprocessing plant for treating irradiated fuel from 1000 MWe of installed power in the form of Candu reactors is planned at Kalpakkam near Madras. A Fast Test Breeder Reactor is also planned for being set up at Kalpakkam. Considerable amount of development work still remains to be done in India on the various methods of reprocessing in general and the reprocessing of fast reactor fuel in particular. Therefore the fuel reprocessing complex planned at Kalpakkam will consist of facilities for reprocessing half a tonne a day of Candu type fuel elements and research and development laboratories in the field of reprocess- ing. In this complex, facilities will exist for development work and plant scale operation for the separation of uranium 233 from thorium irradiated from the fast or thermal reactors and for developmental work and pilot plant operations of the different non-aqueous processes. All the production facilities are planned under one roof to provide for flexibility in the utilisation of per- sonnel and for economies in common services. The cells and process equipment will however be different and separate,. The thermal reactor reprocessing facility would be quite similar to the one proposed at Tarapur except that the head-end might be restricted to a chemical chop (Zirflex or Thermox) pro- cess. The other difference visualised would be the use of mixer settlers for the partition and final purification cycles, in order to reduce the height of the cells. This will provide the necessary plant scale experience in the use of mixer settlers. The demonstration facility for reprocessing fast reactor fuel will have a capacity as mentioned above of 1, 6 kg plutonium a day. The fuel is received in the form of disassembled fuel pins of U0, - Pu0l; clad in stainless steel approximately 5 mm in outer diamefer. A simple cutting procedure is preferred for cutting the fuel into small pieces. A dissolver with product at 90 g uranium and plutonium per litre is visualised. This will be diluted with nitric acid to 25 to 30 g uranium and plutonium per litre and 3 M in nitric acid. The fast fuel reprocessing line will have three cycles of decontamination without the separation of uranium and plutonium. The first cycle extraction contactor will be a 3 cm diameter pulsed perforated plate column. All subsequent contactors will be mixer settlers. The product from the first solvent extraction cycle is expected to be 30 to 35 g of uranium and plutonium per litre. After adjusting acidity this will be fed to the second cycle which will operate under the same conditions as the first cycle. The third cycle will also be a repetition of the first cycle with adjustment of acidity based on the nature of the residual fission products. Evaporation for concentration of aqueous products as well as high active wastes would be avoided. Provision will be made for recovery of neptunium-237 to minimise build up of 51 plutonium-238. The uranium and plutonium in the final decontaminated product would be separated if acquired either by TLA process of if experiments prove successful by reduction with hydrogen in presence of platinum catalyst. The uranium product will be stored to allow the decay of uranium 237 and plutonium will be precipitated as plutonium oxide for recycle. When the development work on centri- fugal contactors and stacked clone contactors prove successful the first solvent extraction column will be replaced by the equipment which proves most successful in trials. The production facilities are expected to cost Rs. 75 million ($ 10 million) of which Rs. 45 million can be assigned to the Candu type fuel reprocessing and Rs. 30 million for fast breeder fuel re- processing. The research and development laboratory to be set up separately is expected to cost about Rs. 15 million ($ 2 million). The approximate cost of reprocessing is expected to be Rs. 125 ($ 17) per kg of uranium in the Candu type fuel and Rs. 30, 000 ($ 4000) per kg plutonium in fast reactor fuel. These estimates for capital costs and reprocessing costs are perforce precise only to an order of magnitude as more details remain to be worked out. Lack of familiarity with reprocessing high burn-up high plutonium fuel has introduced an element of cautious conservatism in the cost estimates for the fast reactor fuel reprocessing. Conclusion The figures presented for comparison of costs cannot be assumed to have long term relevance with respect to absolute values. It is only recently it has been possible to attempt extra- polation of available data on the economics of reprocessing for long term planning. The results of the preliminary studies have brought out the interesting features with respect to Indian conditions. With the setting up and operation of the Tarapur Plant, the data will acquire better precision. The trend of increase in labour costs in India will have a significant impact on the construction and operating costs in the future on account of the appreciable labour component. The size of the optimum plant will thereby increase and this trend will be supported by an increase in the size of nuclear power stations and improvements in transportation by rail and road. It is foreseen that for sometime to come fast reactor fuels are best reprocessed in close -coupled plants, avoiding thereby the additional technological problems of safe transportation of high burn-up plutonium bearing fuel. Aqueous reprocessing with modifications and refinements has become the universal method for thermal reactor fuels on account of the proved performance with respect to economics as well as process efficiency. For fast reactor fuels the choice will remain open for quite some time. While aqueous methods do not have insurmountable pro- blems for extension to high burn up short cooled fuels, non aqueous 52 methods offer an attractive alternative, especially for reprocess- ing plants coupled to fast reactors upto about 1000 MWe. The Indian programme visualises development work in the various non aqueous processes as well as on short residence solvent extraction contactors for aqueous processing in connection with the reprocessing of irradiated fuel from fast reactors. References 1. H.N. Sethna and N. Srinivasan, "Fuel Reprocessing Plant at Trombay'" Proceedings of the Third United Nations Conf. on Peaceful Uses of Atomic Energy, Geneva, 1964. 2. H.N. Sethna and N. Srinivasan, "Operating experience with the fuel reprocessing plant at Trombay'' presented at the A.I.Ch. E. Symposium on Recent Advances in Re- przcessing of Irradiated Fuels, New York City, Nov. 1967. 3. W.E. Unger, et al, On site fuel processing and recycle plant- ORNL 3959, 4. T.C. Runion and W.H. Lewis, '""Construction and Operation of the West Valley Reprocessing Plant" presented at the A.I.Ch.E. Symposium on Recent Advances in Reprocessing of Irradiated Fuels, New York City, November 1967. 5. D.D. Stewart, '""The Canadian incentive for fuel reprocessing and plutonium recycle™ - AECL 3136 {Canada) 6. O. Tijalldin, "The Thermox Process'' AE-120 (Sweden) 7. T.J. Barendregt, '"Chemical decanning of fuel" Kjeller Report - KR 126. 8. N. Srinivasan, et al, Trilaurylamine as extracting agent for the final purification of plutonium - BARC 374 (India) 9. N. Srinivasan, et al, Studies on the use of uranium IV as a reductant for plutonium in purex process - BARC - 375 (India) 10. G. Lefort et al, Rapport Semestriel du Department de Chimie Dec 67 - May 68 Section 7.2 SGCR, CEA-N-1044, 11. M. Levenson, et al, Comparative cost study of the processing of oxide, carbide, and metal fast breeder reactor fuels by aqueous volatility and pyrochemical methods - ANL-7137 53 TECHNOLOGY AND ECONOMICS OF NONAQUEOUS PROCESSING Chairman: Richard C. Vogel Argonne National Laboratory Argonne, lllinois, U.S.A. 55 MELT REFINING OF EBR-II FUEL* D. C. Hampson, R, M. Fryer, and J. W. Rizzie Argonne Natlonal Laboratory, Jdaho Falls, Idaho U. S. A. Melt refining is-the neme given to the process of selective removal of fission products from highly irradiated metallic reactor fuel by & high-temperature oxidation step in which the ceramic crucible is the source of oxygen. Volatilization of fission products contributes to the overall fission-product removal., Data obtained from four years of totally remote operation with irradiated fuel has, in general, confirmed theoretical and laboratory-obtained results. The purified metal is remotely refabricated into fuel elements for reinsertion in the reactor (EBR-II). ¥Work performed under the auspices of the United States Atomic Energy Commission. 57 I. Introduction Experimental Breeder Reactor II (EBR-II) is an unmoderated, heterogenecus, sodium-cooled fast breeder reactor with a power output of 62.5 megawatts of heat (1). The energy produced in the reactor is converted to 20 megawatts of electricity through gsodium~to-water heat exchangers and a conventional steam cycle, The plant was built under the suspices of the USAEC as a demon- stration central-station fast breeder reactor. The nuclear driver fuel of EBR-II currently consists of metallic uranium enriched to 52.18% in the U235 isctope and alloyed with 5% fissium.” When approximately 1.2 to 2 atom percent of the uranium in the fuel has fissioned, the fuel 1s discharged from the reactor and processed to remove fission products, re-enrich the alloy, and restore its original metallurgical and nuclear properties. Pro- cessing is accomplished by melt refining in the direectly adjoining Fuel Cycle Facility (2,8) which is an integral part of the EBR-II cecmplex, The Fuel Cycle Facility consists of a conventional rectangular air-atmosphere hot cell and a circular argon-atmosphere cell (Fig. 1). The two cells are joined by a lock system. All process operations that expose fuel directly 1o the atmosphere take place in the argon cell. Operations involving fuel that is protected by a stainless steel jacket or cladding are accomplished in the air- atmosphere cell., Since the process does not remove all classes of fission products from the fuel, all processing operations are totally remote. Feed for the process is an irradiated fuel subassembly from the reactor. The product is a newly reconstituted and re-enriched fuel subassembly ready for insertion in the reactor. A pyrochemical process was chosen for EBR-II because of its promise of reducing the reprocessing cost associated with nuclear power. The principal characteristics of the process which promise reduced costs are 1ts simplicity, compactness, low volume of dry active wastes, and capability for handling short-cooled fuels with ¥Fissium is a term used to represent a mixture of fission product elements (atomic numbers 40 to 46) which when alloyed with uranium impart to the alloy desirable metallurgical and radiation-sta- bility properties. In this fuel, concentrations of fission- product alloying elements are: 2,5 w/o molybdenum, 2 w/o ru- thenium, 0.26 w/o rhodium, 0.19 w/o palladium, 0.1 w/o zirconium, 0.04% w/o silicon, and 0,01 w/o niobium. Silicon (0.04 w/o) is added to improve the radiation stability of the fuel. 58 66 SODIUM | DEGAS DISPOSAL| & INSPECT| MOCKUP OPERATING AREA MOLD PREPA- RATION SODIUM HANDLING OPERATIONS CONTROL Plan of Fuel Cycle Facility an sttendsrt refinetion in fuel inventories. Another advantage is the reduced criticality problem agsociated with small volumes of metallic fuels in the absence of nuclear moderators(2). No attempt will be made to evaluate the economics of the overall process, sinee this paper is concerned only with the results produced by melt refining. Melt refining was chosen as the specific pyrochemical demonstra- {ion process for the metallie fuel of EBR-II, A large number of parametric studies (3) were made to aid in predicting the disposition and. interactions of individual fission products and to define the physiecal specifications for the process, These studies were made on & amall scale with either inactive or tracer-level materials., However, some of the variables, such as buildup of isotopes of uranium with continual recycle, could not be verified in small-scale runs. Two of the primery objectives for the operation . of the melt refining process in the Fuel Cycle Facility were to eveluate the veracity of the small-scale experiments relative to full-scale (both size and activity level) runs, and to demon- strate that a pyrochemical process could be operated on a full- plant scale by totally remote means for a meaningful period of time. With regard to the chemistry of the process, the full-scale raus tornfirmed the experimental amd theoreticnl predictions. With regard to the time period, the successful operation for more than four years by totally remote means exceeded the original design specifications for the demonstration of this process. II. Head-End Processing Steps The fuel subassemblies are 92 in. long, are hexagonal in shape, and contain 91 enriched uranium-alloy fuel elements (clad pins). The fuel pins are 13.5 in. long by 144 mils in diameter, and are sealed in e stainless steel can (O mil wall) which is 18 in., long. The 6-mil annulus between the fuel pin and the cladding wall is filled with sodium which acts as 8 heat transfer medium during power operation. The external surfaces of s subassembly (Fig. 2) are covered with residual sodium after removel from the reactor. The sodium is removed from the subassembly in the interbuilding coffin (2) by an oxidation and water-dissolution procedure. After sodium removal, the subassembly is transferred to the air cell for mechani~ cal dismantling. Individual fuel elements are transferred toc the argon cell, where they are mechanically decanned and chopped into 1.5 in. sections for ease of handling (2, 8). The chopped, sodium~ coated fuel is ready for melt refining. Refabrication of new fuel elements from the melt-refined fuel is described in an accompanying paper (10). 60 19 '°|3's MAX. AFTER WELDING - =-13.500 65115 N FUEL PIN | ‘0o O LEVEL . 049 WIRE GAP Y * A © - it | ey m 180 DiA. 009 WALL —T 11: 082 - 13.500 »| 1440 T \ 1Y {200 FUEL PIN 3416 = ol SODIUM . RESTRAINER -PLUG = 8% . S — ———t— — — EBR I FUEL ELEMENT-MARK I-A STAINLESS TRI-FLUTE BLANKET; ; ge.aoo SECTION y ¥ j Z (9 FUEL ELEMENTS) fe A—s2" 2. EBR-IT Driver Fuel Subassembly III. Description of Melt Refining The irradiated, chopped fuel and a precalculated quantity of re~enrichment uranium is charged to a lime-stabilized zirconium oxide crucible.® Zirconium oxide was preferred to other ceramics because contamination of the melt by the substrate metal (Zr) is negligible, and pouring yields from ZrOp crucibles were somewhat better than those from the other ceramics considered (3). In addition, zirconia provided better cerium removel. The crucible is placed in a sealed furnace (2), and the fuel alloy is melted and liquated for 3 hrs at 1400°C under an inert argon atmosphere. During the run, the crucible is covered with a ceramic- fiber fume trap which has been shown effective in trapping cesium and iodine (CsI){6). After purification, the fuel alloy is chill- cast into a graphite ingot mold. This process is termed melt refining (2,3). Melt refining removes three classes Of fission products from the fuel, and these represent about two-thirds of the total fission yield. The chemically inert fission gases xenon and krypton are evolved on melting (3). Fission products such as bromine, iodine, and cesium, which have high vapor pressure at 1400°C, are volatil- ized and trapped by the fume trap (3,5,6). Chemically reactive fission products, such as yttrium, strontium, barium, lanthanum, and the lanthanide elements, react with the oxygen in the zirconia crucible to form oxides. These oxides remain with the crucible when the alloy is poured into the mold, Fig. 3 (3, 5). One group of fission products, atomic numbers 4O through 46, are not removed by the melt refining process. Zirconium, atomic number 4O, has exhibited fractional removals, but this is attributed 10 reaction with carbon and other impurities in the alloy rather than to the oxidative process, These fission products constitute the so-called fissium elements. For the first core lecading of EBR-II, it was felt advisable not to start with a pure uranium fuel, which has undesirable metallurgical and radiation-stability charascteristics, and would be subject to compositional changes during each resctor-processing pass. Instead, the steady-state concentrations of fissium metals in recycled fuel were calculated as a function of fissien yield, percent burnup, melt refining re- actions, dross (skull) purification and recycle (7), and pouring yields. This calculated composition was chosen as the alloy for the first core loading of EBR-II (4) (see footnote p. 1). ¥The approximate compostion of the crucible is: Cal0 , 5 w/o; HfOp, 2 w/o; Si0p, 0.7 w/o; Alp03, 0.5 w/o; Ti02, 0.3 w/o; Fe203, 0.3 w/o; balance ZrOz. 62 APPROXIMATE DISTRIBUTION OF ELEMENTS IN MELT REFINING PROCESS OFF - GAS Xe 100% Kr (C0% I trace FUME TRAP Cs 100% Cd 100% Rb 100% I >75% (Na) 100 % CRUCIBLE Y 5% Rare Earths 5% Ba 90% Sr 90% Te 10% SKULL U 5-10% Pu 5-10% Noble Metals 5-10% v 059, ‘ PRODUCT INGOT u 90-95% Pu Rare Earths 95% Ba 10% 90-95% Sr 10% Noble Metals 90-95% Te 90% (Mo, Ru, Nb, Zr,) Rh, Pd, Tc, etc 3. Fission Product Removal in Melt Refining 63 IV, Melt Refining Performance 1. Fission Product Removals Considerable preoperational research was done on the behavior of many of the fission products during melt refining ( 3 ), For direct comparison with current work, the most important work was that of Trice and Steunenberg (5). They completed a series of small-scale melt refining experiments (400-g charges) with highly irradiated EBR-II-type fuel and determined fission product distributions as well as removals. From these experiments, it was concluded that uranium, molybdenum, and ruthenium show no preference for the ingot or skull, and distribute directly according tc the pouring yield. The removels of rere earths, tellurium, iodine, cesium, and barium-strontium were generally 99% or better. The single determination on plutonium hinted at a very slight partitioin to the oxide, Plant-scale runs typically involve 10- to 12-kg charges of irradiated fuel plus makeup enriched uranium. The melt refining crucivle is 9-1/2 in, high by 6-3/8 in. in dismeter, and is covered by a Fiberfrax*(2) fume trap with dimensions of 10 in. in diameter by 5-1/4 in. high. Because of the large size of these components and the high levels of activity involved, sampling of these com- ponents was difficult and complete fission-product distributions were not obtained for plant operation. For the same reasons, and because of the heavy analytical load of routine fuel analyses (uranium, plutonium, noble metals, and trace impurities), complete fission-product analyses were not obtained for each run. Certain fisslon products were selected as representative of major classes. Cesium and ilodine were selected as representative of those elements having high volatility. Barium,lanthanum and cerium were selected as representative of the chemicelly reactive group, with cerium as a representative for rare esrths, The behavior of the gaseous fission products (xenon, krypton) was so well established that they were not checked, For plant startup, 1400°C and 3 hrs of liquation were chosen as the initial melt-refining paremeters. These runs were made primerily to confirm that plant-scale results would agree with leboratory-scale results. Accordingly, removals for these con- ditions are presented first, The removals for the volatile and the chemically reactive groups were generally 98.5% or better. *Fiberfrax is a trade name of the Carborundum Company for a ceramic fiber composed of alumina and silica. 64 Plutonium exhibited a slight preference for the skull, showing approximately 5% removal from the refined alloy. The noble metals distributed directly according to pouring yield. Data obteined from these early runs confirmed that 3 hrs and 1400°C are adequate conditions for acceptable. fission product removals, They essentially duplicate the data from the small- scale experiments ( 5). The results also confirm that technetium may be considered a noble metal. The stability of the noble mefald is more clearly presented in the next section {IV-3). The data for pluto?i?m are at low concentration levels, but they confirm previous work (3). Off-gases from the melt refining furnace pass first through the in-cell filter containing a 2-in. layer of activated carbon and a high-efficiency glass filter medium, then through a vacuum pump and an oil separator to a shielded holdup tank (500 cu ft) con- taining 1000 1b of activated carbon. When meteorological con- ditions are favorable, the gas in the holdup tank is discharged through a heated (430°F) bed of silver-nitrate-coated packing (silver nitrate tower), through a bank of high-efficiency filters for removal of submicron particles, and through a 200-ft high stack to the stmosphere, Prior to plant startup, two experiments were performed in the plant furnaces to evaluate the behavior and distribution of iodine In one of these experiments, one wire of 1131 {as palladium iodide which decomposes at 350°C) was added to 8 kg of fuel and melt re- fined at 1400°C for 3 hrs. During refining, the charge contained in the Zr02 crucible is covered by the fume trap. The fume trap collected ~ 80% of the I13l, ~20% distributed to metallic surfaces in the furnace assembly, and the crucible and skull retained approxi- mately 2%. Of the small amount that reached the in-cell filter (0.3 mCi), only 0.3% passed through it (as is typical of most radia- chemical determinations, the material balance is not perfect). Very minor amounts (0.1%) of the originsl iodine eventually appearad in the cell atmosphere. In a companion experiment, 1 Ci of elemental iodine was intro- duced into the pipe entering the off-gas delay tank at a point downstream of the in-cell filter and the exhaust pumps. Approxi- mately 25% of the charge (250 mCi) reached the tank. Three suc- cessive pumpouts (following backfilling with argon) resulted in a total of 0.14 mCi of TI13l being extracted from the tank. The percentage passed by the tank charcoal was thus 6 x 10'5% (11). Following the confirmation of the initial parameters, data were obtained to ascertain the effect of liguation time while maintaining the 1400°C liquation temperature. Table 1 presents these data. 65 During a typical melt-refining run, 2 hrs elapses between the start of the run and the time at which 1400°C is reached. The EBR-IT fissium alloy melts in the range 1030 +to 108000, so the alloy has normally been molten for 20 to 30 min before 1400°C is reached. From the data of Table 1, it is concluded that cesium and barium~lanthanum are effectively removed from the bulk alloy upon reaching 1L400°C. Todine is also effectively removed in rela- tively short liquation time at 14OOC (it has been shown that the iodine is normally deposited after volatilization as an iodide, primarily CsI (6), Table 1 Effect of Liquation Time on Fission-Product Removals et 1400°C Average Percent Removal No. of Runs on Ligquation Barium- Which Average Time (hr) Cesium Lanthanum Cerium is Based 4 > 99 > 99 99 1 3 > 99 > 99 98.5 Many 2 > 99 > 99 95.5 3 1.5 S 99 > 99 90 1 0.5 > 99 > 99 73 1 0.25 > 99 > 99 69 2 The removal of cerium, and presumably the other rare earths, on the other hand, is strongly time-dependent at 1400°C, The cerium data are plotted in Fig, 4 for illustration. This time dependency was not observed in the small-scale runs where the time was varied between 1 and 3 hrs. However, this is not unexpected, since the crucible contact-surface-to-volume ratio of the small-scale runs was a factor of three larger than for the plant-scale runs. It has been shown that diffusion through the reaction zone is probably the controlling process for cerium removal (3, 9, 10). Thus, it is reasonable that a 1-hr liquation in small-scale geometry should produce the same results as a 3-hr ligquation in the plant- scale crucible. Either of them would essentially effect quantita- tive (98-99%) removal of cerium, and additional liquation time would not produce a significant increase in removal of cerium. Scme data were obtained tco assess the effect of temperature on fission-product remocvals. Because of plant operational require- ments, however, the determinations were over relatively narrow temperature ranges. At 3 hrs, no effect of temperature was discernible within the accuracy of analytical results for Cs, Ba-La, or Ce. In order to accentuate any change that may have occurred, the tests with 2-hr 66 CERIUM REMOVAL, % b, 100 90 @ O ~J o aad | I l | I | o .5 1O 1.6 20 25 3.0 35 40 LIQUATION TIME AT 1400°C, HOURS 60 Cerium Removsal as a Function of Liquation Time 67 liquation time wereé made. Table 2 These fPesults are shown in Table 2. Effeet. . of .Ligquation Temperature On Fission Product Removals 3-Hour Liquation 2-Hour Liquation Temperature Average | No.of | Temperature| Average No,uf (og) Element. Removal(%U ZRuns | (o) Rewevad— (%)) Buns 1400 Cesium > 99 Y 1400 >99 3 Be-La 599 2 >99 3 Cerium Be5 5 95.5 3 1350 Cesium > 99 1 1350 >99 2 Ba-ILa > 99 1 > 99 2 Cerium 98.5 1 92.5 2 1300 Cesium > 99 2 1300 >99 1 Ba-La > 99 2 96.5 1 Cerium 98.5 2 7 1 Cesium shows no temperature effect over the range considered. The data for I131 are too incompléte to allow conclusions. Pre- vious work had shown complete removal at 1400°C and at least 90% removal at 1300°C (3). Barium-lanthanum was removed quantitatively during all the 3-hr runs checked, but it did exhibit a slight lowering of removal for a 2-hr liguation at 1300°C. Extension of the tempersture range would probably amplify this effect, con- sidering the chemical similarity of barium, lanthanum and cerium. The data for cerium are interesting. No apparent temperature dependence is exhibited for 3-hr liquations, but there appears a fairly strong dependence for 2-hr liquations. To correctly estimate the magnitude, the time effect must be subtracted ocut before the tem- perature dependence is established. The data are too meager, however, to allow this separation. The final conclusion about the cerium behavior is that there exists a combined time-temperature dependence that would aliow some parametric juggling in the operation of & melt-refining pro- cessing plant. In general, the data for the elements investigated confirm laboratory and theoretical estimates, and appropriate choice of parameters will provide adeguate fission-product removals. 68 2. Gross Gamma Activity Distribution Gross radiation distribution measurements were taken for one of the early melt refining runs (all readings are at 1 ft). The feed materials {corrected for decay) showed an input tc the run of 7 x 10k R/hr (12)- After being refined, the activity distribution on the various components showed the following: Reading Percentage of (R/nr) Total Refined Ingot L.5 x 103 8 Crucible (1.5 x 104) ) y + Skull (3.5 = 107) 5.0 x 10 8h Fume Trap 4,0 x 103 8 Ingot Mold 3 _— Total 6 x 10% 100 The charge for this run averaged 0.4 a/o (atom percent) burnup. At 1 a/o burnup, %ocalized radiation readings would fall in the range of 102 - 10 R/hr. The difference between charge and product can be attributed to self-shielding in the ingot and gas evolution from the refined alloy. The subsequent high radiation background within the shielded cells precluded repeating this experiment with fuel containing a higher activity level, 3. Alloy Compositional Stability and Plutonium Buildup As noted previously, the initial "equilibrium" core loading for EBR-II was calculated from knowledge of the behavior of the various fission products in the melt refining process. Compositional varilations were later followed in one batch of fuel that was re- cycled five times between the reactor and the processing-refabri- cation cycle. The fuel received 1 to 1.2 afo burnup each cyele in the reactor. None of this fuel was intermixed with other batches during this experiment and the only compesitional adjustments made were to add sufficient U-235 each cycle to compensate for that consumed in the reactor. After five cycles, the composition of the fuel had not changed {(within analytical accuracy) except for zirconium, silicon, plutonium, technicium, and U-23L plus U236. These starting con- centrations for these elements and the concentrations after five cycles are shown in Table 3. 69 Table 3 EBR-II Fuel-Alloy Composition Changes Composition After Initial Composition Five Processing Element {w/o) Cycles (w/o) Uranium-236 0 09 0.52 Plutonium None C.12 Technecium None .05 Zirconium 0.10 0.09 Silicon 0.01 0.07 The concentrations of plutonium and technieium increased because they are formed in the reactor and are not selectively removed in melt refining. Neither of them was present in the initial alloy composition., A slight increase in U-236 is also apparent. This isotope is formed in the reactor as a result of the n, & reaction on U-235., No separation of urenium isotopes results from melt refining. The increase in silicon concentration comes from small amounts of the Vycor (quartz) molds that enter the melt during the sub- sequent injection-casting operation. Continuved recycle of the metal resulted in a steady increase of silicon. No attempt was made to remove silicon, since postirradiation surveillance of the fuel showed that fuel with silicon contents in the range of 200-800 ppm ex- hibited less radiation damage than the fuel containing the original concentration of silicon (50-100 ppm). The total fission yield of zirconium is greater than that of any other noble metal in the alloy. If zirconium were not removed from the alloy by some mechanism, it would grow at the expense of the other noble-metal concentrations. The current alloy compesitions show that, if anything, zirconium has slowly decreased in concentra- tion. The analytical data for zirconium typically exhibit wide variance, and no estimate of the percentage of removal is attempted. The removal of zirconium has been correlated with the presence of carbon and cerium in the alloy ( 3)., The staviiity of the alloy is readily spparent and confirms criginal estimates of the behavior of the process. L, Silicon Addition As mentioned above, it was desirable to increase the silicon 70 concentration of the alloy to about 400 ppm. Other investigators who attempted to add silicon to uranium melts noted that quanti- tative alloying was difficult to achieve; several experiments were therefore conducted at FCF to determine the best way to add silicon to EBR-II driver fuel. The methods that were considered were; (1) direct addition of silicon metal to uranium-fissium-alloy melts, (2) direct addition of a uranium-silicon master alloy to the uranium- figsium-alloy melts, (3) direct addition of fissium silicides to uranium-fissium-alloy melts, and (L) direct addition of silicon dioxide to uranium-fissium-alloy melts. Table 4 shows the results of initially adding silicon metal to the charge (uranium and uranium - 5 w/o fissium) before alloying at 1400°C. Table L Direct Addition of Silicon Metal to Uranium And Uranium - 5 w/o Fissium Melts at 1400%C Charge(g) Stoichilometric U 31 Silicon Actual Type of Melt Concentration Concentration Uranium 11,501 253 2,15 w/o 1.95 w/o Uranium 10,351 211 2.01 w/o 1.85 w/o Uranium 7997 163 2.00 w/o 1.86 w/o U-5 w/o Fissium 10,951 * 3.8 ~ 350 ppm 300 ppm U-5 w/o . * Fissium 11,166 3.8 ~ 340 ppm 290 ppm U-5 w/o Fissium 11,213 3.8 ~ 340 ppm 290 ppm *> w/o Fissium An interesting point that resulted from the experiment is that 86.5 £ 1.5% of the silicon initially charged remained with the product for each case. Two experiments were conducted where silicon was added to a uranium - 5 w/o fissium melt through a master alloy. The master alloy was composed of U-238 and silicon (1.86 w/o Si). The desired final composition was 400 ppm Si. Results indicate that a near-perfect stoichiometric silicon pickup is possible when silicon is added by the master alloy method., This method is presently being used at the Fuel Cycle Facility for silicon addition to alloy melts. 71 Zirconium disilicide was added directly to an allcy preparation charge ag a means of adding zirconium and silicon to the fissium alloy. Results indicated that 100% of the silicon and 85% of the zirconium alloyed with the product. Vycor glass tubing (99.9% Si02) was added to a depleted-uranium charge. The oxygen introduced as Si02 reduced the melt yield; therefore, this technique was not used for adding silicon to the fuel alloy. 5. Melt-Refining Pour Yields and Throughput The yield of one melt refining operation is defined as the ratio of the weight of the chill-cast ingot to the total weight of spent fuel, enrichment adjustments (U-235 and U-238), and fissium elements charged to the furnace before the melt. The throughput at melt refining is defined as the total weight of alloy processed per week in the melt refining furnaces (there are two furnaces in the EBR-II, FCF). The throughput is directly related to the yield of each melt refining operation. The pouring yleld in melt refining is dependent on & number of conditions, such as charge size and geomeiry, furnace atmosphere, and crucible material. These conditions were not varied for the melt refining runs. The conditions that were investigated were the effect of liguation time, liguation temperature, and fission product concentration (percent burnup). For what were defined as standard operating conditions, i.e., 3 hrs of liquation in a zirconia crucible at 1400°C, the effect of burnup on pouring yield is shown in Fig. 5 The straight line is a least-squares fit of data points obtained. The effect of liquation times or temperatures on pouring yield could not be separated from the effect of fission-product removal. Pouring yilelds are directly dependent on the quantity of dross produced. Iower temperatures and the resulting lower reaction rates would require longer reaction {liquation) times to produce the same fission-product removals. Previous work with unirradiated uranium showed that the dross consisted of 10 to 20% oxidized metal and 80 to 90% occluded metal. This work also showed that higher pouring temperatures resulted in higher yields. This is probably a result of a lower viscosity of the matrix metal at higher temper- atures, which results in better "draining" of the occluded metal from the dross. The highest temperature consistent with equipment limitations was selected, 1400°C. This resulted in a 3-hr liquation time which was consistent with a single shift (8 hrs) overall operating cycle for the melt refining furnace. 72 gl POURING YIELD, w/o 5. 90 | | l | l | | | | o 4 =2 3 4 5 6 1 8 9 10 AVERAGE BURNUP OF CHARGE, a/o Pouring Yield as a Function of Fuel Burnup In addition to refining the spent fuel, one of the necessary jcbs of the melt refining furnaces is to consclidate the recycle material being returned from the refabrication process. This material comes in two bulk forms, 'heels" and "shards.," An explanation of the sources of this material is given in an accompanying paper(10). A summary of the consolidation and melt refining operation carried out in the melt refining furnace is given in Table 5. The average throughput of alloy per furnace during normal pro- cessing periods was x 25 kg per week, This included both spent and recycle fuel. The total amount of fuel refined in the Fuel Cycle Facility and the total weilght processed through melt refining is shown in Table 5 (two furnaces). Table 5 Melt Refining Throughput and Yields No. of Total Ave, Charge Total Ave, Pour Runs Charge Burnup Refined Yield {kg) (a/o) Ingots (w/o) Type (kg) Irradiated Fuel 310 2279 0.825 2090 91.7 Recycle Runs 191 2117 1974 93.2 Alloy Prepara- tion Runs 73 ToU 748 9L.2 74 Summary In the work reported on in this paper, two areas are of partic- ular interest. The first is that the melt refining process is the first demonstration of pyrochemical processing of reactor fuel on a full-plant scale. The second is that equipment can be de- signed to be operated by totally remote means on a productive schedule, and can maintain this schedule for over four years of continuous operation, The data obtained from these operations essentially confirm data that had previously been obtained on unirradiated or small- scale runs. This is noteworthy and encouraging in that designers of future pyrochemical processes may continue tc have confidence in data that were derived from laboratory or theoretical results, The melt refining furnaces were installed in the shielded cells in 1963. They were tested with unirradiated uranium melts, and in 1964 the first melts with irradiated fuel were conducted. During the 4-1/2 years that the two melt-refining furnaces were in oper- ation in the argon cell, a total of almost 600 runs of 10 to 12 kg each were made in these furnaces. The majority of the runs were made for purification of the reactor fuel. The remainder were made for blending of new (unirradiated) alloy or for consolidation of recycle material. The modular design of the furnaces permitted remote repair or replacement of any components that had failed. The longevity of the equipment and the ease with which repairs could be made re- affirmed the desirability of proper design and precperational testing of all equipment that must be operated remotely. The over- all operating availability of both furnaces for the past four years exceeded 85% of operating time. 75 10, 11. 12. References L. J. Koch, et al., "Hazards Summary Report, Experimental Breeder Reactor IT (EBR-II}," ANL-5719, May 1957, and"Addendum to Hazard Summsry Report, Experimental Breeder Reactor II (EER- II): ANI-5719 Addendum, June 1962, Argonne National Laboratory. J. C. Hesson, et al., "Description and Proposed Operation of the Fuel Cycle Facility for the Second Experimental Breeder Reactor (ERR-II)," ANL-6605, April 1963, Argonne National Iaboratory. "The Melt Refining of Irradiated Uranium: Application to EBR-II Fast Reactor Fuel," A Series of Papers in Nucl, Sci. and Eng., Vol. 6, 1959, and Vol. 9, 1961. D. C. Hampson, "Preparation of Alloy for First Core Ioading of EBR-II," ANL-6290, August 1961, Argonne National Labaratory. V. G. Trice, et al., "Small-Scale Demonstration of the Melt Refining of Highly Irradiated Uranium-Fissium Alloy," ANL-6696, August 1963, Argonne National Laboratory. N. R. Chellew, et al., "Laboratory Studies of Iodine Behavior in the EBR-II Melt Refining Process," ANL-6815, January 196k, Argonne National Laboratory. L. Burris, et al., "The EBR-II Skull Reclamation Process. Part I. General Process Description and Performance,"” ANL-6818, January 1964, Argonne National Laboratory. D, C. Hampson, et al., "Startup Experience for the EBR-II Fuel Cycle Facility. Part II. Fuel Processing," Proceedings of the 13th Conference on Remote Systems Technology, The American Nuclear Society, Hinsdale, Illinois, 1965, pp. 8-13. "A Mechanism Study of the Oxide-Drossing of Cerium Molten Uranium with Uranium Dioxide," NAA-SR-3090, 1958, North American Aviation, Inec. M. J. Feldman, et al., "Remote Fabrication of EBR-II Fuels," 1969 Nuclear Metallurgy Symposium, Reprocessing of Nuclear Fuels, Ames, Towa, Vol. 15, August 1969. Chemical Engineering Division Semiannual Report, July-December 1964, ANI-6925, Argonne National Laboratory, pp. 69-72. L. Burris, et al., "Estimation of Fission Product Spectra in Discharged Fuel from Fast Reactors," ANL-5742, July 1957, Argonne National Laboratory. 76 REMOTE REFABRICATION OF EBR-II FUELS® M. J. Feldman, N. R. Grant, J. P. Bacca, V. G. Eschen, D, L. Mitchell, and R. V. Strain Argonne National Laborsatory, Idaho Falls, Idaho U. S. A, The melt-refining process used for purifying the highly irradi- ated EBR-IT metallic fuel does not remove all of the fission products, hence refabrication of the product fuel must be done remotely. The steps inveolved in refabrication of fuel elements and subassemblies are: injection casting, pin processing (shearing and inspection), loading into a sodium-filled jacket, welding the jJacket closed and leak testing, bonding the sodium in the jacket, and assembling the completed elements into a subassembly. The subassemblies are rein- serted into the reactor. Nearly 35,000 fuel pins and elements have been fabricated by totally remote processing to tolerances of one mil or less. Process variables and yields for each step of the process have been investigated. *Work performed under the guspices of the United States Atomic Energy Commission, 77 Introduction The reprocessing of fuel from Experimental Breeder Reactor II (EBR-II) requires a system of sophisticated equipment operated re- motely in two large cells of the Fuel Cycle Facility (FCF). The fuel itself is an alloy of 52 %-enriched uranium and 5 w/o fissium. The fissium consists of fission-product elements (molybdenium, rhodium, ruthenium, palladium, zirconium, and niobium) that are not removed by pyrometallurgical processing.il) In the FCF the first series of operations includes fuel handling, decanning, and melt refining, as described in an accompanying paper. 1) The following series of operations, the refabrication steps, are described in this paper, In the argon cell, the first step is that of melting the refined ingot and injection casting the uranium alloy into Vycor molds. Next, in the pin processing operation, the fuel pins are removed from the molds, sheared to length, and inspected for length, weight, diameter, and internal porosity. Acceptable pins are loaded into stainless steel Jackets containing a precisgely measured quantity of sodium. An end closure (restrainer) is inserted in the jacket and the final closure weld completed. These fuel elements are trans- ferred to the air cell where the closure weld is leak tested. The final element operations are sodium bonding and sodium bond testing. Acceptable elements are fabricated into subassemblies, which, after final acceptance testing, are returned to the reactor. Three other operations involved in fabrication are not accomplished by remote means. The stainless steel jackets are fabricated from purchased components, and each Jjacket is lcaded with a carefully measured quantity of sodium. Other stainless steel components are purchased and fabricated into preassemblies, which when combined with g1 fuel elements, form a completed subassembly. A summary of operating experience in the FCF is presented at the end of the report. Injection Casting(2) Two of the requirements for the reactor fuel were that the dia- meter (0.1h4Y4 in.) had to be closely controlled, and the grain strue- ture had to be randomly oriented. To combine these factors with remote operation, an essentizlly new process had to be developed. Using a casting process together with controlled cocoling solved the random grein structure. The diameter problem was resolved by forcing the molten metal into tubes with a precision bore. After considerable development work, the injection-casting furnace shown in Fig. 1 was ready for use in FCF.(3,;4) With minor modifications, the furnaces have been used to produce fuel pins in the argon cell since 1964, A production summary is given in Table I. 78 6L HN0IA —=— STANDARDIZED LIFT LUGS —— | FURNACE BELL MOLD CLUSTER- (00-i20 MOLDS \J PALLET STANDH\ L// wf GAS COOLING TUBE L_ma-m SEAL IMMERSION TC ’ RR)| [‘—\ \ - CRUCIBLE pcm-:s*m—-—-——H-—%i ] v e Zr0, INSULATING BRICKS - S & | = - L NDUCTION HEATB?————L-\_HqL %_i\_\¥ _ i l »y - COUNTER BALANCED COVERS | RESISTANCE HEATERS | , | WOUCTION SPINDLE S FREEZE SEAL *h \\ PAN & BUSHING PLATE L = \ { | - VACUUM-PRESSURE BRIDGE mE ACTUATOR N J |1 INDUCTION BRIDGE [ [ FURNACE BASE a = — ] QUENCH ARGON IMLET ————-1- s T —MINERAL INSULATED - M POWER LEAD PNEUMATIC BRIDGE - L — =1 | ~ a INJECTION CASTING FURMACE e e L L L LA ) THOz COATED, GRAPHITE CRUCIBLE (ELEVATED POSITION) 1. Fuel Cycle Facility Injection Casting Furnace The ingot from melt refining is placed in a thoria (ThOp)-coated graphite crucible, which is positioned on a zirconia insulation pedestal in the furnace. A pallet stand holds approximately 100 Vycor* molds above the crucible. The cylindrical molds are sealed at the top end, have a precision bore, and also are coated with thoria. Around the crucible (which also serves as a susceptor) is the in- duction heater. The induction coil is an uncocled, 3/8-in.-dia solid molybdenum coil, which is powered by a 10,000-cycle, 30-kW motor generator. Temperature control of the melt is cobtained by using a platinum-rhodium immersion thermocouple sheathed in a sealed tantalum tube. The thermocouple passes through the center of the pallet to the crucible. There is a backup thermocouple located under- neath the graphite crucible. A gastight chamber is formed by a furnace bell in combination with a Bi-Sn eutectic-alloy freeze seal (Cerro-tru). After the furnace is charged and the seal frozen, a vacuum of 50- 100 1 is obtained. The charge is heated to 1350° C, and the automatic casting cycle is initiated. This raises the crucible into the elevated position so that the molds are immersed approximately 1-1/2 in. into the melt. Since the molds are relstively cold, a plug of metal forms in the tube. The molds are therefcre held in the melt for a short time (8 sec) before the furnace is pressurized to 1.7 atmospheres (abs.). The pressure drives the molten metal up into the evacuated molds. After 2 sec, the metal begins to freeze in the molds, at which time the crucible is lowered to its original position and a flow of cooling is recirculated over the molds. The furnace is cooled for approximately 4 hrs (to a crucible temperature of less than 200° C), the seal is melted, and the furnace is unloaded. The fuel ping in the Vycor molds are transferred to pin processing; the heel remaining in the crucible is broken and recycled to melt refining or rerun in the injection-casting furnace. One fuel pin has several small pieces cut from its center, and these samples are sent to an adjacent laboratory for chemical analysis, Pin Processing The pin processing operation consists of several individual steps.( :5) Each step is performed by an equipment module that can be individually replaced. The eguipment modules are supported on a cascade base that contains the necessary pneumatic and electrical connections., These service connections are automatically msde when the module is placed into position on the base frame. A complete set of spare equipment modules is provided so that when repairs are required on a module, the spare module can be installed and the pin processing operation can continue. *Jyeor is the trade name for high-silica glasses from Corning Glass Works., 80 1. Demolding: Vycor glass molds are removed from the castings by placing the castings, one at a time, onto a slot formed by two parallel hard-faced bars spaced slightly closer than the mold dia- meter. A single flat-edge blade powered by & pneumgtic ram pushes the mold between the parallel bhars, breaking the glass from the fuel pin. The broken glass is collected in a metal container located underneath the demclder. The cast fuel pin falls downward to a pair of rails and rolls to the front face of the democlder. 2. Sheering Station: The cast fuel pin is placed into the shear from the top, and drops into shearing position between spring-loaded pressure pads and the shear tools. One set of shear tools is sta- tionary while the second set is mounted in a movable block which is actuated by a pneumatic cylinder. Both ends of the pin are simul- taneously sheared to produce a 13.5-in. length when the movable blades slide past the stationary blades. The shear blades are fabricated of hardened tool steel (Super High Speed Coballoy), and have four cutting edges that can be rotated into position for shearing. Each cutting edge will normally shear 1000 to 1500 castings before failing. 3. Length-Measuring and Weighing Station: After shearing, the. fuel pin drops through the shear station and rolls down inclined rails below the shear to the length-measuring stetion. The length is measured by forcing the pin longitudinally ageinst a dimension- measuring transducer by means of a pneumatically operated ram. When the ram is released, the pin rolls down inclined rails to the balance, The deflection of the balance is measured through a transducer system, A movable rail section on the top of the balance raises the pin off the balance and sllows the pin to roll down to the diameter- and porosity-checking station. 4, Diameter and Porosity: From the previous station, the fuel pin falls onto a longitudinal set of rails and is fed into an air gauge by means of a push-rod mechanism operated by an electric motor. The air gauge and an eddy-current coll are mounted inside a removable brass block, and the pin is pushed through the air gauge and the eddy-current coil simultaneously, 5. Data Recording System: Signals from the in-cell pin- proces- sing equipment are fed to a data recording and processing system located outside the cell in the coperating annulus. The signals for length and weight are converted intc numerical values and displayed on a digital voltmeter. Pin-diameter values are integrated over the pin length and used by the data system to calculate the pin volume. The signal from the eddy-current porosity cecil is fed into a recorder which plots the results on a strip chart. The data system operates a key-punch machine which punches the casting-batch number, pin number (determined by pin sequence during processing), length, weight, and volume for each fuel pin on a data card. This data card then 81 becomes the permanent fabrication record for the fuel pin, and is used for accountebility and criticelity-control purpcoses in all sub- sequent fabrication steps. The operator reviews the data as exhibited on the data recording system, and then makes a decision whether o sccept or reject the pin. Rejects are transferred to a separate tray, chopped into seg- ments, 3 to 4 in. long, and recycled to melt refining. The identity of accepted pins from this station forward throughout the balance of the operations is maintained with the data card by positioning the pins in specific locations of various receptacles, Table T Summary of Injectien-Casting and Pin-Processing Production, 1964 through 1969 Total Number of Casting Runs 518 Total Weight of Alloy Charged 6074 kg Number of Castings Processed L4 ,080 Number of Acceptable Fuel Pins 36,620 Percentage of Acceptable Fuel Pins 83.1% Reason for Rejection: (a) Sheared Short 21.0% (b) Short Casting 27.2% (¢) Low Weight 13.5% (d) Diameter or External Defects 5.0% (e) 1Internal Porosity 33.0% Jacket PFabrication Long before the pins that will be loaded into the jackets are cast, the stainless steel components rust be purchased and fabricated. Each Jjacket consists of four components: tube, tip, spacer wire, and restrainer. The first three are manufactured from Type 304L stain- less steel, and the last from Type 304 stainless steel. The tubing is of welded construction with an inside diameter of 0,1560 * 0.0005 in. and a wall thickness of 0,0090 X 0,0005 in. An extensive testing program is carrie% 3ut on the tubing, including 100% pulsed eddy-current inspection 6) for wall defects. The maxi- mum acceptable defect level is 10% of the wall thickness. Acceptable tubing is cut te length and a bottom tip is heliarc-welded to each piece. The spacer wire, 0.049 * 0.0005 in. in diameter, is heliarc- welded to the tip and wrapped around the tube with a 6-in. pitch. The top end of the wire is welded to the tube with a capacitor- discharge welder. The jackets are inspected and the internal 82 volume 1s determined as described in the next section. The re- strainers, which are fabricated and inspected elsewhere, are cleaned and taken into the argon cell at the same time ag the sodium-filled jackets. Sodium Loading Sodium is employed in the 0,006-in.-thick annulus between the fuel pin and the element cladding to provide heat transfer from the fuel pin.(7) A precise amount of sodium is placed in each element 80 as to leave a maximum gas-plenum volume to accommodate irradiation swelling of the fuel and yet keep the fuel pin completely covered with sodium during reactor service. The sodium loading operation consists of two parts -- jacket prepasration and sodium-charge pre- paration. Jacket preparation consists of a visual inspection, leak testing, and meagsurement of the internal volume. The visual inspection is employed to detect damage which may have occurred during shipping, as well as to provide a double check for defective welds which may have been missed during the fabrication inspecticn. The Jackets are leak tested by pressurizing the interior of the jacket with helium and analyzing air flow passing the exterior of the jacket with a masg spectrometer. The test is capable of detecting leak rates of 10-8 sta cc/sec at a pressure difference of L atmospheres. The internal volume of each Jjacket is measured by utilizing a motorized alr gauge coupled to an electronic integrating circuit. Each jacket is placed in one of ten classifications depending on its internal diameter. Each class varies from the next by 0.0001 in. in the range of 0,1555 to 0.1565 in. Sodium-charge preparation consists of calculating the weight of sodium required, extruding sodium into wire, cutting and weighing lengths of the wire, and placing the charges in the Jjackets. The weight of sodium required for each element (0.65 to 0.85 g) is calculated by subtiracting the average volume of the fuel pins of a casting batch from the average volume of the jacket class to be used. One of the factors that affects the sodium level is the volume reduction which accompanies the transformation of the uranium alloy from the gamma phase to the alpha phase during the bonding operation. Complete transformgtion of the entire fuel pin causes a volume decrease of about 0.035 cc (about 1% of the originsl pin volume). Experience has shown that only about 0.026 cc (0.025 g) of sodium must be added to the calculated sodium charge to achieve the desired sodium level in the element, because the entire pin is not retained in the garma phase during the injection-casting oper- ation. The sodium charge is obtained by extruding sodium (obtained from the reactor's secondary sodium system) through a die at room temperature into l/8-in.-dia.wire. The wire is cut to precise lengths by using a special cutter which permits wvariation of the 83 length of the wire with a micrometer adjustment. The lengths of sodium wire are weighed to check the accuracy of the cut and are placed in the Jjackets. All the operationg requiring sodium handling are carried out in an argon-filled glovebox (impurity levels: Op - 20 to 60 ppm; HpO - 20 to 60 ppm; No - 0.5% max). Average values for pin volume and jacket volume are used for calculating the sodium charge for each element, so the final sodium levels of the individual elements vary slightly. Levels between 0.50 and 0.80 in. above the fuel pins are acceptable (Fig. 2). Less than 3% of the elements fabricated have been rejected because of high or low sodium levels, Settling and Welding The Jjackets containing the sodium charges are transferred to the argon cell in sealed polyethylene bags. The fuel pins are placed in the Jjackets and heated to allow the pins to displace the sodium and settle to the bottom of the jacket; then the combination fuel re- strainer and end plug can be inserted into the top of the jacket. The restrainer's purpose is to prevent gross fuel movement during irradiation by ratcheting mechanisms. The restrainer is sealed to the top end of the element tubing by means of a capacitor discharge weld (Fig. 3). This method was developed to provide a reliable, remotely operated welding machine for very thin-wall tubing. The weld is accomplished by placing the top of the end plug of the element 0,040 in, from a tungsten electrode, preheating the top of the element for 5 sec. with a high-frequency discharge, and discharg- ing a 0.17-farad capacitor bank charged to 150 V across the ionized path between the electrode and the end plug. The capacitor dis- charge fuses the end plug to the top of the jacket, thus sealing the element. The primary reason for rejection of welds is improper alignment of the electrode with respect to the top of the end plug, which causes a misshaped weld with lack of fusion on one side of the element. There are two other causes of unacceptable welds that occur less frequently: (a) contamination of the 1lip of the jacket with sodium causes blowouts, and (b) improperly shaped (blunted) electrode tips cause misshaped welds. The overall acceptance rate for the closure weld is almost 97% (see Table IT). Since the fuel pins and bonding sodium gare now protected, the elements are moved to the air cell for subseguent operations, Legk Testing The closure weld is tested for leaks by means of a pressure-decay leak-detecting device, The device operates on the principle of 84 EBRII MARK IA-TYPE FUEL ELEMENT TOP END " FUEL PIN (144" DIAMETER) ELEMENT OVERALL LENGTH . | g3 1812 U-5w/0 Fs FUEL PIN, TOP END g8 BONDING SODIUM —RESTRAINER CLOSURE WELD—\ o.es"—~| GAS-PLENUM VOLUME NOMINAL NUM SODIUM LEVEL (~ 665 cnv) ABOVE PIN 2. EBR-II Mark IA-Type Fuel Element ot FUEL CYCLE FACILITY EBRII FUEL-ELEMENT CLOSURE WELDING MACHINE LIFTING LUG ‘ SHIELDING-GAS ELECTRODE ASSEMBLY ELECTRODE-TO- RESTRAINER GAPPER INLET \'; I d CHILL ASSEMBLY. FUEL ELEMENT -; L 4 ELEMENT-RAISING ASSEMBLY ELECTRICAL GROUND / TERMINAL N\MACHINE BASE 3. Fuel Cycle Facility EBR-II Fuel-Element Closure Welding Machine. 86 determining a leak rate by the loss of pressure from a very small volume over a period of time., The system has many advantages over more common methods of leak testing. Use of high pressures will break through a scdium oxide film covering a small hole; it has heen demonstrated that other methods such as a mass spectrometer in a vacuum system will not detect that type of leak. Also, as discussed below, a system leak in the pressure-decay method will give a positive identification., The top weld of the element to be checked is sealed with a seal gland into a chamber of approximately 0.06 cc in volume. A metering chamber also 0.06 cc in volume is pressurized to 82 atmos- pheres from a helium gas cylinder. The metering chamber is then opened to the test chamber with the top end of the element sealed in it, producing a resultant pressure of about 41 atmospheres in the two chambers. If there is a leak in the weld, the gas is forced into the gas plenum region of the fuel element and the pressure is reduced to about 20 atmospheres. If a leak around the sealing gland exists, the pressure in the chamber approaches zero. Each weld is pressurized for 7 min, during the test. This duration in combination with the readout equipment presently in service results in a sensitivity of about 10-% cc/sec. The lower sensitivity of this leak test compared to the one carried out on the tip-to-tube weld is acceptable, since the consequences of a leak in the gas plenum are much less serious than those of a leak near the bottom of the element during reactor cperagtion, Experience has shown that the capacitor-discharge welding technique produces very few slow leaks at the closure weld. Either the weld is good or the defect is large enough to allow high leak rates and rapid depressurization of the leak-detector head. Sodium Bonding Following the completion of leak testing of the fuel-element closure weld, a sodiuT bon?ing operation is the next fabrication step for the element. 9:10) This operation is required in order to establish a high-quality heat transfer path in the 0.006-in.-thick sodium-filled annulus between the fuel pin and the jacket., The establishment of such a path will allow for the effective removal of heat from the figsioning fuel-pin material to its Jjacket and sub- sequent removal by the EBR-II primary-sodium coolant. The impact-bonding technique employed consists of heating groups of 50 elements in a special magazine within an electrical resistance furnace to a temperature of 500° ¢ (Fig. 4). When this temperature is attained, a vertically oriented mechanical impact is delivered similtaneously to the 50 fuel elements by means of a single-acting, spring-return pneumatic cylinder coupled to a hardened steel striking platen in the bonding machine. Bonding impacts are transmitted from the pneumatic cylinder's striking platen to the elements through hardened tool-steel transition pileces at a rate of approximately 87 BONDING MACHINE i .,’ L & 4. Fuel Cycle Facility Bonding Machine. 88 FURNACE 100 impacts per minute. These impacts, usually 1000 in number in any bonding sequence, cause the fuel elements to move vertically up- ward in free flight for a distance of approximately 1 to 1 l/u in. and return to their starting positions before the subsequent impact ig initiated. Forces resulting from the impacts cause the gas entrapments in the sodium thermal-bond annulus of the fuel element to move upward out of the bond region and into the gas plenum at the top end of the element, thus leaving the bond itself free of gaseous defects. Defect removal is aided by high transient pressures that apparently result in the sodium near the element lower end during the impacting process and by small, but significant, move- ment of the fuel pin with respect to its Jacket. This latter feature aids the upward travel of gas bubbles by a mechanical "wiping" action, Following completion of the impacting procedure, a carefully controlled furnace cocl-down 1s carried cut for the elements in the bonding machine, This cooling procedure ensures that the bond sodium will reach the solid state (approximately 100° C) progressively from the bottom ends of the elements toward their top ends. The controlled cooling eliminates shrinkage voids, tears, etc,, which will develop in the sodium if top-to-hottom or otherwise uncontrolled rapid cooling is allcowed to take place. Experience has indicated that when recycling the bond rejects and using up tc three bonding sequences (three operations, each of which utilizes 1000 impacts delivered while maintaining a temperature of 5000 C), a 91% acceptance rate has resulted (see Table II). The operation does cause some deformation of the tip, but this does not appear to be significant. Impacting the elements for more than three sequences may cause excessive damage. A bonding sequence for 50 driver elements requires approximately 4 hrs. Sodium Bond Tesgting Following the completion of impact bonding, the same 50-element magarine used in the bonding cperation ig positioned in the sodium bond-testing machine (Fig, 5). This machine utilizes an encircling- type coil coupled with a pulsed eddy-current system to indicate the quality of the fuel-element sodium bond, the sodium level within the element, the fuel-pin overall length, and other less important inter- pretable features.(1l) The mechanical portion of the machine incorporates a pneumatically operated turntable which allows individual peositioning of the 50 elements contained in the magazine at the prcper test location. A pneumatic cylinder raises each element from the magazine, upward through the eddy-current coil tc a position where it is grasped by an element drive mechanism, This drive mechanism raises the element upward through the eddy-current coil for the element's entire length, 89 EBR-II FUEL-ELEMENT BOND-TESTING AND SODIUM-LEVEL MEASURING MACHINE GRIPPER ASSEMBLY FUEL-ELEMENT GRIPPER LIFTING HOOKS UEL-ELEME ‘ ASSEMBLY DRIVE MOTOR FUEL-ELEMENT GRIPPER ASSEMBLY FUEL-ELEMENT BOND TEST MACHINE v LIFTING MEMBERS PNEUMATIC-OPERATED ELECTRICAL HEATER ROTARY TURNTABLE ASSEMBLY PNEUMATIC SERVICES ELECTRICAL AND ‘ CONNECT-DISCONNECT PANEL MACHINE LEVELLING LEGS 5. Fuel Cycle Facility EBR-II Fuel-Element Bond-Testing and Sodium- Level Measuring Machine. 90 then reverses its movement to downward, and returns the element to the magazine. A strip-chart recorder is synchronized to the element drive system so that a one-to-one ratio results in the test recording. This re- cording provides duplicate indications of the element's sodium bon- and sodium level; the first resulting from the upward travel of the element through the encircling eddy-current coil, and the second from its downward travel through the coil. Figure 6 presents a typical eddy-current recording for an EBR-II driver element. The pulsed eddy-current system of the machine with its integral encircling coil is able to detect discontinuities (gas voids, tears, etc.) as small as 0.015 in. in diameter in the sodium-bond annuli of EBR-II driver elements. In addition, the system is capable of in- dicating the level of the sodium column in the fuel element to an accuracy of + 0.020 in. Sodium entrapment in the gas plenum regions, and gas pockets in the sodium column above the fuel pins, are also readily discernable. Approximately 2 hrs. are required to bond test and level test a magazine of 50 elements. Although the machine has an integral electrical resistance furnace incorporated in its design for elevated- temperature sodium-bond testing, all production bond testing of EBR-II driver elements has been carried out at hot-cell temperatures, approxi- mately 90 to 95° F, at which the sodium is in the solid state. Table IT Summary of Element-Loading and Bonding Production Number of Fuel Pins Loaded into Jackets 39,750% Number of Elements Accepted by Leak Testing 38,415 Percentage of Accepted Welded Elements 96.6% Number of Elements Bonded 38,337 Number of Elements Accepted by Bond Test 34,976 Percentage of Accepted Bonded Elements 91.2% Reason for Rejection (Bond Test) (a) Voids in sodium annulus 43.8% (b) Sodium traps in gas space 4.1% (c) Bubbles between fuel pin and restrainer 3.2% (a) Sodium level 33.3% (e) Damaged elements and other rejects 15.6% *Includes supplemental fuel pins added to the system to increase production rates temporarily. 91 44 4s s ¥ " IT Fuel-Element. Sample Bond Test Trace of EBR- 6 Preassgembly and Subagsembly Fabrication A subassembly, as furnished to EBR-IT, is made of several paris (Fig. 7). FExternally, it consists of a top-end fixture, a hexagonal tube, and a lower adapter. The top-end fixture has a stem that is used for all handling, both in FCF and in the reactor; the lower adapter supports the subassembly in the reactor and contains the orifice holes that maintain the deslired coolant flow. The internal porticn of the subassembly has been changed three times during the history of the reactor. Originally, the regions above and below the fuel elements were depleted uranium blanket sections. Each section contained 18 sodium-bonded depleted uranium elements, two grids, and a tie bar. After approximately two years, the breeding ratios for the reactor had been determined, and the expensive blanket elements were replaced by solid stainless steel rods. However, even this desigh contained relatively expensive grids and required considersble assembly time, and was replaced by stain- less steel shields. These shields are still in use at the present time; each consists of two trifluted sections, offset 60° with a connecting pin., They provide approximately the same pressure drop as the blanket sections, and the offget reduces neutron streaming to the reactor components above and below the subassemblies. A1l of the components are manufactured from Type 304 stainless steel. Acceptable components are fabricated into upper and lower pre- assemblies. The upper preassembly consists of the top-end fixture, the hex tube, and one shield. The lower preassembly consists of the lower adapter, one shield, and the "T" bar grid. The latter is an assembly of 11 T-shaped bars onto which the fuel elements are placed; the bars are welded to a short hexagonal tube. The refabricated fuel elements must be assembled into a sub- assembly before they can be reinserted intc the reactor core. This operation is performed on the assembly machine located in the FCF air cell. The assembly operation consists of placing a new upper and lower preassembly in the assembly machine and performing the following operations. The 91 refabricated elements are fitied on the lower preassembly grid by using the master-slave manipulators. Each fuel element has an identity and must be placed in its proper position on the grid so that when the element is examined after irradiation, the pre- and postexamination data can be compared. The upper pre- assembly is then lowered over the 91 fuel elements and seated on the lower preassembly. Six tungsten inert-gas spot welders are then positioned at the lower end of the upper preassembly and six spot welds are made, Jjoining the upper and lower preassemblies, 33 ¥6 Top End Upper Triflute Fixture Shield Lower Triflute Shield T-Bar Grid T. EBR-II Fuel Subassembly, Core Type. Hex Tube 91 Fuel Elements Lower Adapter After the assermbly is completed, 1t must be tensile tested to ensure sound welds. The tensile tester locks the upper and lower adapters of the subassembly into fixtures and applies a tensile force of 1800 1bs. The subassembly is then moved to the straightness tester. The subassembly 1s again held in position by the upper and lower adapters and the straightness of the six flat surfaces and the lower adapter are checked with prepositioned dial indicators. The maximum permissible bow is 0.040 in. If the subassembly passes these inspections, it is returned to the reactor via a 20-ton lead- shielded cask. Summary The remote fabrication process described in this paper represents e noteworthy accomplishment from twe distinct viewpoints. The first is the production accomplishments of the process and the second is the ramifications of conducting a remote multistaged production operation, In the period September 1964 to April 1969, the facility produced 353 subassemblies, This represents more than 5 reactor-core loadings of reprocessed and refabricated fuel. In accomplishing this, the facility received LL5 subassemblies of spent fuel, refined 598k kg of fuel, and cast 607k kg of fuel., It produced L4,080 castings, of which 83.1% were of adequate quality to process. The jacketing, welding, and leak-detection operations processed 39,750 units, of which 96.6% were acceptable. The bonding and bond-testing operations processed 38,337 elements, of which 34,976 or 91.2% were accepted as reactor grade. It is noteworthy that of the over 40,000 elements that have seen reactor service, only one failed element has been detected. That one element appeared to have been defected in the final assembly operaticn (into a subassembly) and caused no serious difficulty during reactor operation. The second salient accomplishment of the Fuel Cycle Facility has been the proof in operation of the basic assumptlons of the initlal design. These, in broad perspective, were the capability to con- tinuously operate a remote facility and to provide remotely operated process equipment that could support a production type of effort, The facilities used in these operations (air cell and argon cell) were last entered by persomnnel in March 1964, Continuous multistage facility operation without personnel entry for over five years is unprecedented in remote systems technology. A relatively high availability rate for manipulators and process equipment is essential to the successful operation of any continuous remote production activity. For the 35 units which make up the mani- pulative capacity of the faeility, the availability has ranged 95 between 61 and 99%, with the average availability being 89%. The 21 units of process equipment have had operstional avallabilities of between 80% and 99% with an average availability of 92.5%. The experience garnered in the past five years in pyrometallurgical procesging and remote fabrication has provided a strong experience base for continued development of spent fuel processing that encompasses integration of reactor and plant, fast turnaround (low inventory), radioactive waste concentration, and remote fabrication, 10. 11. References Hampson, D.C., Fryer, R. M., and Rizzie, J. W., "Melt Refining of EBR-IT Fuel," to be presented at 1969 Metallurgy Symposium, Towa State University, Ames, Iowa, August 25-27, 1969, Shuck, A.B., U.S. Patent No. 2952056 (September 13, 1960). Jelinek, H.F., and Iverson, G.M., "Equipment for Remote Injection Casting of EBR-II Fuel," Nuclear Science and Engineering, Vol. 12, March, 1962, pp. L05-411. Jelinek, H.F., Carson, N.J., Jr., and Shuck, A.B,, "Fabrication of EBR-II, Core I Fuel Pins," ANL-627Lh, June, 1962. Carson, N.J.,Jr., and Brak, S.B., "Equipment for the Remote Demolding, Sizing, and Inspection of EBR-II Cast Fuel Pins," Nuclear Science and Engineering, Vol. 12, March, 1962, pp. L12-418, . Renken, C.J., "A Pulsed Electromagnetic Test System Applied to the TInspection of Thin-Walled Tubing," ANI-6728, March, 196k, Carson, N.J., Grant, N.R., Hessler, N.F., Jelinek, H.F., Olp, R.H., and Shuck, A.B., "Fabrication of EBR-II Core I Fuel Elements," ANL-6276, December, 1962, Grunwald, A.P., "Leak Testing of EBR-II Fuel Rods," Nuclear Science and Engineering, Vol. 12, March, 1962, pp. L19-k23, Sowa, E.S., and Kimont, E.L., "Development of a Process for Sodium Bonding of EBR-II Fuel and Blanket Elements," ANL-638l4, July, 1961, Cameron, T.C., and Hessler, N.F., "Assembling, Sodium Bonding, and Bond Testing of EBR-II Fuel Rods," MNuclear Science and Engineering, Vol. 12, March, 1962, pp. 424-431, Ono, K., and McGonnagle, W.J., "Pulsed Eddy-Current Instrument for Measuring Sodium Levels of EBR-II Fuel Rods," ANL-6278, July, 1961. 96 PREPARATION AND PROCESSING OF MSRE FUEL¥* J. M. Chandler and R. B. Lindauer Chemical Technoclogy Division Oak Ridge National Iaboratory U. S. A. Abstract The Molten Salt Reactor Expe imen% gMBRE) has been refueled with an enriching salt concentrate, LiF- UF), (73-27 mole %), vhich was prepared in a shielded cell in the Thorium-Uranium Recgcle Facillty at ORNL. The preparagfign process involved reducing UO ( conten 222 p g by treatment with hydrogen, coaggrtlng the U02 to UFh by hygrofluorination, and fusing the UF) with The original MSRE fuel salt, which contained 220 kilograms of 235U-238U was fluorinated to volatilize the uranium as UFg. The UFg was then absorbed on packed beds of NaF pellets. Essentially all of the uranium was recovered in six runs; less than 5 grems was lost to the caustic scrubber solution. To minimize corrosion, fluo- rination was discontinued before the uranium volatilization was complete (i.e., 130 grams of uranium wes allowed to remain in the molten salt). The uranlgm was decontaminated from fission products by a factor of almost 10 Fluorine utilization veried from greater than 70% initially to 13% (average) for the final run, and averaged 39% for all six runs. Corrosion products were removed from the barren carrier salt by reduction and filtration. Corrosion rates for surfaces exposed to fluorine during fluorination averaged 0.1 mil/hour for 47 hours. *Research sponscred by the U. S. Atomic Energy Commission under contract with the Union Carbide Corporation. 97 Intrecduction The ORNL Molten Salt Reactor Experiment (MSRE) is an 8-megawatt circulating liquid fuel reactor operating at 1200°F. The original fuel contalned gg mole % LiF, 30 mole % BeFp, 5 mole % ZrF), and 0.9 mole % The reactor first went critical on June 1, 1965, and began full-power operation in May 1966; prior to shutdown on March 26, 1968, the reactcr had operated for slightly more than one equivalent full-power year (or 72,400 megawatt hours). This paper describes the work done in 1968 at ORNL in preparation for the refuellng of the MSRE with an enriching salt concentrate, TLiF-233UF), (73-27 mole %). This concentrate was prepared in the Thorium=-Urenium Recycle Facility because the 233U used contained 222 ppm of 232y, The gamma emitting daughters of this 232U produced a radiation level of 300 Rem/hour for a 450-gram can of the starting ox ide and necessitated the use of heavy shielding and remote processing. At the reactor plant, the uranium in the original MSRE fuel salt was removed as UFg by fluorination. The barren fuel carrier salt was purified by reduction and filtration. Preparation of 7LiF-233UF)_,, Concentrate Three batches of 233U'O (1) each containing 12 kilograms of 233U were required as starting materlal for greparlng 63.4 kilograms of the fuel-enriching concentrate, This congentrate was to contain 39 kilograms of uranium (35.6 kilograms of <33U). The first run began May 9, 1968, and the third run was completed July 30, 1968. The product was packaged in 45 enrichment capsules, four 7- kilogram shipping containers, and a six-can cluster comprised of shipping containers of miscellaneous sizes. The ten salt shipping containers and the 45 enrichment capsules were delivered to the MSRE as required in the reactor enrichment schedule. Process Equipment The flowsheet shown in Figure 1 1s a simplified presentatiocn of the major eguipment components required in the process. These com- ponents are: (1) the fuel decanning station, (2) the reaction or oxide treatment vessel, (3) the salt storage and transfer vessel, (4) various containers for shipping the product, (5) the off-gas scrubbers. In addition, the process requires many other smaller items of equipment, such as: the oxide can preparation equipment, the in-cell titration assembly, furnaces for the vessels, the enrichment capsule drilling and weighing stetion, disconnect stations for electrical and instrument lines and process gas lines, work tables, and tool racks. 98 66 o 1-2 9/min HF =IO sim Hz FLOW ROT@; RESTRI%TOR FUEL AP MET ENTRY CHPEVE AQ&SOUS ADDiT'.ONY CALROD b4 GAS PENTHOUSE FILTERD CE - U036 kgs LiF~4 kgs [1-p——He PURGES ~1/2 sim CAN VoG CUTTING OCL CAN CHUCK AND DRIVE CAN OPENING AND DUMPING BOX ALT SAMPLER X GAS SAMPLE ANALYSIS SALT OFF GAS — OVERAFLOW FILTER ALARM (V\MIV\N\I\ SALT FILTER £y F2 F-3 FURNACE FURNACE FURNACE AQUEOUS KCH SCRUBBING SYSTEM : ! TI T2 T-3 REACTION SALT ENRICHING CAPSULE VESSEL STORAGE AND SALT CAN FILLING VESSEL VESSEL R vCG RHD SYSTEM 1. Simplified Chemical and Engineering Flowsheet for Preparing 233y Fuel Salt. All services and reagent sources are located in the penthouse outside the cell. The reduction and conversion processes were monitored by a thermo- couple array that was inserted into the powder in the reaction ves- sel and by measurements of hydrogen and HF utilization during the reduction step and the conversion step, respectively. Unfiltered and filtered samples of the melt were withdrawn for oxide, petrograph- ic, and metal impurity analyses. Feed Materials The 233U0; had been prepared in batches by using 7 M NHjOH (in excess) to precipitate hydrous uranium oxide from solutions that contained 10 to 40 grams of uranium per liter and were 1 M in HNO and NHhNO . The uranium in the feed solution had been purified aild isolated gn 1964 and 1965 by a solvent extraction method, followed by ion exchange. These treatments decreased the concentrations of plutonium, thorium, fission products, and corrosion products {iron, nickel, and chromium) to acceptably low levels. The hydrous oxide had been dried and packaged in nested aluminum cans. No spread of contamination or excessive radistion exposure to personnel occurred during the removal of the cans from the storage facility, during their transfer to TURF, or during their discharge from the carrier to cell G. Figure 2 shows the chemical flowsheet used for the fuel prepara- tion work. Reduction of 233Uo3 Three separate operations were required to elongate the cans of 233U'O3 to the 9-1/2-inch length required for proper operation of the can opening box. These operations involved trueing the cans and ce- menting a l-l/E-inch cap on the end of each. The elongated cans of oxide were opened, one at a time, in the decanning station, and dumped into the reaction vessel. The bed of 233y0, that had been dumped into the reaction vessel was expanded and thén heated for 2 hours at 550°C in a helium at- mosphere to remove, by pyrolysis, any traces of ammonium compounds or cther volatiles remaining from the chemical processing. The oxide bed was then cooled to L400°C before reduction with hydrogen was started. This temperature was sufficiently low to accomodate the temperature rise that would be expected from the exothermic reaction of hydrogen with UO3: 100 CHARGE U03= HEAT TREAT UOS: HYDRCGEN REDUCTION: uo, — U0, HYDROFLUORINATION: U(:)2 — UF4 EUTECTIC FORMATION: UF, + LiF —UF - LiF EUTECTIC PURIFICATION: MO + HF —MF + Hp0O MF + Hp — MO0 + HF uo 3 OVERALL REACTION + H2 + 4HF — UF, + SHZOT UF4 + LiF - UF4- LiF 27% — 73% EUTECTIC COMPOSITION ~13.2 kg U AS UO, 3 TO 5 hr DIGESTION AT 550°C; COOL TO 400°C, START 5% Hp AT 400°C AND INCREASE TC 50% Hp; TEMPERATURE RISES TO 490°C; TREAT AT 500-550°C AT 100% USAGE OF Hp; COOL TO 400°C, START 5% HF IN Hp AT 400°C; INCREASE TO 40% HF IN Hp; TEMPERATURE INCREASES TO 450°C; WHEN HF USE DECREASES BELOW 80%, INCREASE THE TEMPERATURE TO 630°C STEPWISE UNTIL HF USE BECOMES 0; COOL TO 150°C, ADD EXACT QUANTITY OF 7LiF; MELT UNDER 30% H DIGEST AT 850°C FOR 3 TO 5 hr; COOL TO 700°C, PURGE MELT 24 TO 30 hr AT 700°C WITH 20% HF IN Hp; TREAT WITH Hz FOR 75 TO 150 hr. UNFILTERED SAMPLE ANALYZED FOR OXIDE CONTENT. FILTERED SAMPLE ANALYZED FOR METALLIC IMPURITIES, PRODUCT PURITY: 2. Chemical Flowsheet for the Low-Temperature Process for Preparing the MSRE Fuel Concentrate. 101 UOg + Hy — U0, + HO + 72,000 cal/g-mole The concentration of hydrogen (in helium) was adjusted initially to 5 vol % and was graduslly increased to 50 vol % during the first k4 hours of treatment. The temperatures rose in response t¢o an increase in hydrogen concentration and then became constant. Both the location of the reaction zone and the zone movement inside the 24-inch-high bed of 233U02 were clearly defined by the tempera- ture profile. As the reaction progressed, the reaction zone rose, in the form of a band, up through ‘the powder bed. After the temperature excursions resulting from the increases in hydrogen flow had subsided (approximately 12 hours), the temperature of the furnace was increased, at the rate of 30°C/hour, until the bed temperature was 525 & 25°C. The reduction operation was contin- ued at this temperature, with 50 vol % hydrogen--50 vol % helium, at e gas flow rate of 2 liters/minute, until 50 to 100% excess over the stoichiometric amount of hydrogen had been passed through the bed. Oversll reduction time was about 56 hours. Hydrogen utilization within the bed was 100% until the reaction zone approached the top of the powder bed; then a slight decrease was observed. Hydrogen usage was determined by a material balance of the gas outflow, as measured by the in-cell wet test meter. Hydrofluorination of 233uo, Upon completion of the oxide reducg§§n step, the bed was allowed to cool to 4OO°C. The conversion of <33U0, to @ 33yF), by hydrofiue- rinstion, using HF gas diluted with hydrogen, began at U400°C end was completed at 625°C. A period of 5 to 7 days was required for the conversion. The HF gas that was supplied to the process was withdrawn from the vapor space of a heated 100-pound HF cylinder. A differential- pressure transmitter across & caplllary restrictor in the HF gas supply line was used to monitor the flow. The gas was passed through a maze of tightly packed nickel wire in a 2-inch-diameter nickel tube thet was maintained at 625°C to remove sulfur from the stream. It was then mixed with a metered amount of dry hydrogen, filtered, and introduced to the reac%ion vessel through a dip tube that extended to the bottom of the = UO bed. At the beginning of the hydroflucrination step, the composition of the gas used for hydrofluorination was 95 vol % hydrogen--5 vol % HF; the flow rate of the mixture was 2 stenderd liters/minute. Over a period of 3 to 4 hours, the HF concentration was incrementally in- creased to 40 vol % as the exothermic reaction caused the bed tempera- ture to rise from 400 to 450°C. During these initial hours, the 102 temperature within the bed was constantly monitored to determine when the temperature excursion resulting from each HF flow adjustment had ceased and when another adjustment could be made. After the HF concentration of the hydrofluorinating mixture had reached LO vol %, the reaction zone traveled, in the form of a band, up through the bed {in a manner similar to that obsggved during the hydrogen reducticn) as the 233U02 was converted to 3UFLL. The reaction U0y + 4HF —UF), + 2Hp0 + 144,000 cal/g-mole is more exothermic than the reduction reaction, but it does not have as great a tendency to cause thermal excursions. Probably, this is the result of differences between UO3, UOo, and UFu with respect to bed permeability and thermal properties. The reaction-zone tempera- tures for the three production runs were plotted as a function of time, as a control measure. The progress of the hydrofluorination reaction was also followed by obtaining a material balance of the HF in the system. The HF utilization was essentially 100% for the first five days and then decreased sharply as the reaction zone reached the top of the bed. Then the temperature of the bed was increased to 625°C, where it was held for two days to ensure completeness of the reaction. The HF utilization did not increase at the higher temperature; instead, it continued to decrease, suggesting that the reaction had been complete at the end of the fifth day of hydrcflucrination. A total of 13.5 kilograms of uranium, as 233yo , was converted to 233uF, in each of three runs; only very mincr differences in the runs were noted. Formation of the Eubtectic Salt The eutectic mixture 233UF)-TLiF (27-73 mole %) was formed by adding the stoichiometric quantity of lithium flucride powder to the uranium tetrafluoride powder and fusing the mixture. The temperature of the reaction vessel containing the stratified 233UFu and TriF powders was increased to 855°C in order to melt the lithium fluoride (melting point, 835°C). The melt was digested at this temperature for 3 hours while it was sparged with hydrogen (at a flow rate of 0.2 liter/minute) to reduce any extraneous compounds that might have been introduced duvring the LiF addition. Differences, with regard to conditions durin§ initial meltdown, were noted in the runs. In runs 1 and 3, the 2 3UFu and 233LiF had to be heated to about 850°C before melting occurred. In run 2, ini- tial melting cccurred at 650°C, nearly 200°C below the melting point of the lithium fluoride. The low-temperature initisl melting must 103 have resulted from the presence of a sizable heel of salt (melting point, 490°C) that remained in the reaction vessel from Tug %. Prob- ?bly, this heel acted as a "seed" to permit fusing of the 3 UF), and 1iF powders at the lower temperature. The 9-inch-deep pool of eutectic salt (melting point, L90°C) was treated with 20 vol % HF--80 vol % hydrogen (flow rate, 2.4 liters/ minute) for 24 hours at a temperature of T00°C to remove the last traces ¢f oxide from the salt prior to the hydrogen purification pro- cedure. At the conclusion of this treatment, an unfiltered sample of the salt was withdrawn and analyzed for oxide content. (A 1/2- inch-0D x 2-1/4-~inch-long nickel cup was immersed in the salt to withdraw a 25-gram sample.) A l-inch-long section cut from the cen- ter of the sample was analyzed petrographically and chemically for the presence of UO,. The remaining portion of the sample was sub- mitted for a compléte chemical analysis. The more rapldly obtained petrographic results were used to determine whether hydrofluorination should be resumed or whether hydrogen purification of the melt should be continued. In each of the three runs, the UQ, content was reported to be less than the lower limit of accuracy (220 ppm) for the petrographic appraisal; thus, subsequent HF treetment was unnecessary. Chemical analyses of the same samples showed oxide contents (in the product salts) of 62, 34, and 32 ppm for runs 1, 2, and 3, respectively. Purification of the Eutectic Salt The eutectic salt was purified by bubbling pure hydrogen gas (3 to 10 ppm Eéo), at a flow rate of 2 standard liters/minute, through the 10-inch-deep eutectic salt melt. The temperature of the molten salt during this reaction, MF + 1/2 H, — HF + M°, was TO0°C. The progress of the reaction was followed by titrating the effluent gas with the in-cell titration assembly. The end point of the purification was evident from the leveling off of the HF evo- lution rate at 0.025 milliequivalent per liter of hydrogen. The hydrogen flow rate was increased on several occasions during the proc- ess, and the reaction rate seemed to be almost independent of the hydrogen concentration. The chromium concentration in the melt was not affected by the hydrogen treatment. The levels to which the iron (100 ppm) and nickel (75 ppm) concentrations were decreased are believed to be near the limit of accuracy of the sampling system and the laboratory analyses. During each of the three runs, the reaction vessel was exposed for 20 days to 40 vol % HF--60 vol % hydrogen at temperatures ranging 104 from 40O to 850°C. Approximately 5 grams of nickel (from the nickel liner of the reaction vessel) was lost to each melt. This corresponds to a uniform corrosion rate of slightly less than 0.001 inch/year — a low rate for this type of process. Transfer of the Fuel Concentrate Eight operations were necessary to transfer the three 13.5- kilogram batches of eutectic salt mixture from the reaction vessel to the intermediate transfer vessel and then to the various shipping container assemblies. The transfers ranged in size from the 13.5 kilograms of uranium (4.7 liters of salt) in the production batches to the 4.3 kilograms of uranium (1.5 liters of salt) that was re- quired to fill the array of 45 enrichment capsules. The transfer operations were conducted at a salt temperature of 600°C. (This temperature had been arbitrarily selected, and the containers had been fabricated to contain the desired quantity of fuel at this temperature.) Shipping Containers The shipping containers were arranged in three arrays for the filling operations. Iater, upon completion of the filling operation and freezing of the salt, each array was disassembled into individual containers (at TURF) for shipment to the MSRE. The first array (Figure 3) consisted of 45 enrichment capsules, each of which was 3/t inch in diameter and 6 inches long and designed to contain 96 grams of uranium. The capsules were connected in series and arranged in three 15-capsule decks. The second array (Figure 4) consisted of four 2-1/2-inch-diameter by 34-inch-long cans connected in series for filling operations. Each was designed to contain 7 kilograms of uranium and had an in- strumented overflow pot. They were shipped, one at a time, to the MSRE, and charged to the drain tank. In this tank, the salt melts and runs out of the container. The third array (Figure 5) contained a group of six 2-1/2-inch- diameter variable-length cans that were similar in design and arrange- ment to those in the second array. One of these cans was used to store excess product material that was blown back from the other five cans after they had been filled to overflow. The latter cans were designed to contain 0.5 to 3 kilograms of uranium (two cans, 0.5 kilogram each; one can, 1 kilogram; cne can, 2 kilograms; and one can, 3 kilograms). 105 3. Filling Array Consisting of 45 Capsules. 106 : 4 X : ) s A . 1;5"~ as,f) iyt : ! ! . T - } , &l i ¢y X!ixxx - [ , rwxsflzg;nE’]’ ’Mg”ighffvfi”‘ ’a ! ) . Cig i * ;3;“! e g‘;;, i ’ax,figé;z 's;zgé!; i M,“ A T m:g%ingfligqfi:f, d b L PR ‘Kfii,fi”az u#fi " gs*’gi i Pz ] N} . g;, Ll Feg i 0l v ‘ xi,i‘i i i R , o sl gz i i JE: 3§ i v z';‘g i I 4. Second Filling Array: Four Cans, Each Containing 7 Kilograms of Uranium. 107 By \ , e B R i o o B £ S g gt 152;2 d B g il o i e £ U ] + i i i i u‘ i ]E FE “fis;‘ EEZ i i - £ “ B f Fig kY el - e e | ey | o g o8 M i £ 8 Zg‘u‘ o ‘ e PR ¥ i 5 e Py o K ‘z 3 T3 i i JZ I oy H f i t RN § fey i d o @ 3 t ] 3 LI ¢ s t ' ' [ pie g 5. Third Filling Array Containing Six Shipping Containers of Miscellaneous Sizes. 108 The MSRE Fuel Processing Operations The MSRE Fuel Processing Facility(e) was constructed in s smgll cell in the reactor building for two purposes: (&) to remove any sccumulated oxides in the fuel or flush selt by H.-HF sparging, and (b) to recover the original uranium charge from the fuel carrier salt in order to allow the 233U fuel concentrate to be added for the second phase of reactor operation. This facility had been operated previously, in 1965, to remove 115 ppm of oxide from the flush salt before the reactor went critical. A LiF-BeF, (66~ 34 mole %) salt that had been used to flush the reactor seven times was used for a final test of the uranium recovery process. The uranium that had accumulated (6.5 kilograms) in this salt was recovered without incident in less than 7 hours. The recovery process consisted of fluorine sparging the salt to volatilize the uranium, followed by decontamination of the gas stream with a T50°F NaF bed and absorption of the UF6 on 200°F NaF. The excess fluorine was removed by an aqueous scrubber. The corrosion product flucrides were reduced to the metals, which were filtered from the salt. Using this process, approximately 216 kilograms of uranium was recovered from the fuel salt batch in 46.5 hours. Corrosion of the fuel processing tank during fluorination of the fuel salt averaged about 0.1 mil/hour. Fluorine utilization averaged 7.7% during fluo- rination of the flush salt and 39% during fluorination of the fuel salt. Reduction and filtration produced a carrier salt with less impu- rities than the original salt. The recovered uranjum was decontami- nated from fgssion products by factors of 8.6 x 10° (gross gamma) and 1.2 x 107 (gross beta). Identifiable uranium losses were less than 0.1%. Description of the Process Fluorination. — The flowsheet used in processing the flush salt and the fuel salt is shown in Figure 6. The molten salt was forced, under pressure, through a freeze valve in the drain tank cell, through a metallic filter (backflow), and another freeze valve in the proec- essing cell to the fuel storage tank. The transfer was maede at 1000 to 1100°F, a temperature that is well above the freezing point of the salt but not hot enough to reduce the strength of the metallic filter element below a safe limit. After being cooled to 475°F, to minimize corrosion and fission product volatilizaetion during fluorination, the salt was sparged with pure fluorine or a fluorine-helium mixture at a relatively high flow 109 01T SALT SAMPLER ‘_—‘_—l SALT CHARGING T T FV AND [Fey = HIGH BAY AREA II SALT TO OR (FV, FROM DRAIN AND FLUSH TANKS STORAGE ]'A!\JK A=ar WASTE SALT N TANK 200°F NaF ABSORBERS IN CUBICLE ===’7 PROCESSING o ACTIVATED CHARCOAL CELL U :;TE‘,' PI\KIESUT!E TRAP N [ = TANK f& ReTem L MIST o nFILTER L P A C ABSOLUTE - A | E?!\EE FILTER TRAP 750°F CAUSTIC CHARCOAL TRAP oF BED NEUTRALIZER 6. MSRE Fuel Processing System. rate (approximately U0 liters/minute) to convert the UFu to UF.. When all the UF), had been converted to UF5, UF6 began to form gnd volatilize, as indicated by a temperature rise in the first absorber. When this occurred, the fluorine flow rate was reduced to 15 to 25 liters/minute to increase the absorber residence time and, in turm, to permit more efficient absorption and to increase the fluorine utilization. The gas leaving the fuel storage tank was composed of UFg, excess fluorine, helium, MoFg and some CrFg from corrosion, IF7, and other fission product fluorides. It passed through a T50°F NaF trap, where the chromium fluoride and most of the volatilized fission products, except icdine and tellurium, were retained. After being routed through the NaF trap, the gas stream, which now consisted of UFg, fluorine, helium, MoFg, IF;, TeFg, and a trace of other fission product fluorides, exited %rom the shielded cell and passed through five NaF absorbers in-a sealed cubicle in the operating area. These absorbers were heated to 200 to 250°F to increase the reaction rate end to minimize MoFg absorption. As the UFg began to load on s particular absorber and the temperature started to rise, cooling air was supplied to the absorber to limit its temperature to a maximum of 350°F. High temperatures tended to decrease the uranium loading by promoting surface absorption and reducing the penetration of the UFg to the inside of the pellets. The finasl absorber was operated below 250°F, where the partial pressure of UFg over the UFg 2NaF complex would allow only & negligible amount of uranium to reach the caustic scrubber. The caustic scrubber was charged with 1300 liters of 2 M KOH--0.33 M KI, along with 0.2 M KyB; 07, which was added as a soluble neutren poison. The reaction occurring in the scrubber was: 6F, + 12KOH + KI — 12KF + 6H,0 + 3/2 O, + KIOq The scrubber sclution was replaced with fresh solution before one- half of the KOH had been consumed, as determined by fluorine flow and calculated utilization. (Dip tube corrosion increased when the OH™ concentration was less than 1 M.) ILaboratory development of the fluorine disposal system is described in reference (3). In addition to fluorine, most of the molybdenum and iodine were removed in the scrubber. A high-surface-ares filter located downstream of the scrubber removed any particulate matter that was retained by the scrubber. During test runs, hydrated oxides of molybdenum collected at sharp bends in the line from the scrubber. A soda lime trap (a mixture of sodium and caleium hydrates) provided a detector for fluorine and a means for removing traces of fluorine from the scrubber off-gas be- fore it reached the charcoal sbsorbers. Activated impregnated charcoal traps sorbed any iodine not re- moved in the caustic scrubber. The gas exiting from the charcoal traps contained only helium and oxygen (that was produced in the il1l caustic scrubber). This gas flowed through an absolute filter and wes monitored for gamms activity and for iodine before being mixed with the remainder of the building exhaust gas, which passed through additional filters and was finally discharged from s 100-foot-tall stack. When there was no longer any evidence cf absorber heating, the flow of fluorine was discontinued, and the salt was sampled and analyzed. The uranium concentration in the salt was expected to be less than 50 ppm at this time. Reduction. — Before the fuel carrier salt could be returned to the reactor system, the NiF,, FeF,,and CrF (produced by corrosion of the Hastelloy-N fuel storage tank) had %o be removed from it (MbF6 is volatile). Because of the high concentration of nickel in Hastelloy-N, the NiF, concentration in the salt was also high. Since nickel is more noble than iron or chromium, it is reduced by hydrogen sparging at 1225°F. After the NiF, was reduced, as indicated by fil- tered salt samples, pressed zirconium metal shavings were added to the salt, and hydrogen sparging was continued to reduce the FeF2 and CrF2 to the metal form. The metals subsequently formed were then removed by a fibrous metal -filter, and the efficiency of the filtra- tion was verified by sampling of the filtered salt. Description of the Equipment Plant Layout. — Most of the processing equipment is located in s 13 x 13 x 17-foot-deep cell (Figure 7) located adjacent to the reactor drain tank cell. This cell contains the fuel storage tank (fluorin- ator), the 750°F sodium fluoride trap, the caustic scrubber, two remotely-operated air valves, and three salt freeze-valves. An adjacent cell of the same size contgins the off-gas equipment-- the mist filter, the soda-lime trap, charcoal traps, and the off-gas filter--that is located downstream of the caustic scrubber. The plant is operated from the high-bay area over the two cells Jjust described. This area contains the absorber cubicle, the in- strument cubiecle, the instrument panelboard, the sampler and sampler panelboard, hydrogen and oxygen monitors, and radiation detection instruments. The gas supply station is situated outside the building. The fluorine manifold (where two 15,000-1iter fluorine tanks mounted on trailers can be connected), the hydrogen manifold, and the pressure and fiow instrumentation associated with these two gases are also located here. The helium for purging and sparging comes from the reactor system supply. 112 Photograph of Fuel Processing Cell. T Fuel Storage Tank. — The fuel storage tank, or fluorinator, is a 50-inch-diameter, 10-foot-tall Hastelloy-N tank with sbout 30% free- board to minimize salt carryover during gas sparging. There is nc provision for cooling since the heat loss will limit the temperature to less than 1200°F after a two-week decay period. During the UF) to UFc conversion period, when the fluorine utilization is high, the heat of reacticn and the afterheat maintain the temperature of the salt at 850°F. About 12 kilowatts of electrical heat is required during the reduction operation at 1225°F. The gas inlet line has a normsl submergence of 6L inches, and the differential pressure be- tween this line and the gas space provides an indication of liquid level in addition to that indicated by weigh cells. The tank is also equipped with an ultrasegnic probe tc verify the weigh cell calibra- tion during the filling and emptying operations. NaF Trap. — The NaF trap is a 20-inch-diameter by 18-inch-high Inconel vessel that is operated at 750°F. At this temperature, volatile ruthenium, niobium, antimony, and chromium fluorides are absorbed, and uranium and molybdenum hexafluorides pass through. This trap was designed to be replaceable because of 1ts suscepti- bility to becoming plugged with volatilized chromium fluorides. Rather extensive chromium fluoride volatilization was expected during the processing of the fuel salt as the uranium concentration in the salt decreased; however, no pressure drop was detected. Caustic Scrubber. — The caustic scrubber, which is used for disposing of excess fluorine and of the HF produced during NiF, re- duction, is a 42-inch-dismeter by 84-inch-high Inconel tank with two dip lines, each with 3-3/8-inch holes. The dip lines have shutoff valves with extension handles for alternating dip lines when plug- ging occurs. This plugging, which is caused by the buildup of corrosion products in the dip line, can be eliminated in 5 or 10 minutes after use of the plugged line has been discontinued. UFg Absorbers. — Five absorbers, piped in series, are located in a sealed cubicle in the operating area situated above the processing cell. The absorbers, 14 inches in diameter and 12 inches high, are made of carbon steel. When loaded to within 1/2 inch of the top, each absorber holds about 25 kilograms of NaF. Bach absorber is mounted in an insulated can and provided with an air cooling coil and an electric heater. Salt Filter. — The salt filter(h) consists of two 4-foot-long concentric fibrous metal cylinders with & total filter area of 8.65 square feet. During the filtration of flush and fuel salts, there was no detectable pressure drop across the filter even when salts conteining sbout 10 kilograms of reduced corrosion products were filtered. It i1s not known how much of these metals remained behind in the fuel storage tank. Filtration of each of the two batches required about 2 hours. 114 Processing Results Recovery of Uranium. — The amount of urenium that was recovered has been determined from the weight increase of the absorbers. Some MoF¢g, which was also absorbed, had to be subtracted in order to ob- tain an accurate value. Although the total emount of absorbed molyb- denum is not sccurately known, random samples showed a correlation between the percentage of molybdenum on the NaF and the amount of UFg that passed through the NaF. However, such a correlation provides only an approximate value for the absorbed McFg because other factors, such as flow rate, temperature, and type of NaF, also affect the MoFg loading. Based on the correlation, a total of 1600 grams of molybdenum was absorbed. Subtracting this wvalue from the increase in absorber welght yields a uranium recovery of 216.0 kilogrems. This value is in good agreement with the amount (218.0 kilograms) which was calcu- lated to be present from the uranium charged to the reactor, the burnup, and the number of samples removed, etc. Since 0.13 kilogram of uranium remained in the salt when fluorination was discontinued, and since less than 1 gram of uranium was found in the scrubber solu- tions, 1.87 kilogreams, or 0.85% of the totel uranium, is unaccounted for. Purity of the Uranium Product. — The NaF at the inlet of the first absorber in each run was analyzed for beta and gamma emitters about 12 days after fluorination. Analyses of feed salt samfi%es showed 2.66 x 1013 gross gamme counts per minute and 3.8 x 10+3 gross beta counts per minute per gram of uranium, respectively. The average gross beta and gamma decontamination factors (DF's) for the six ab- sorbers analgzed (corrected for uranium daughters) were 1.2 x 10 and 8.6 x 109, respectively. Only the gamma radiocactivity levels in the first and second (of a totel of six) runs were appreciably above background. The only radioactive material collected in any measurable amount on the NaF absorbers with the UFg was INb. Most of this material was probably carried from the salt into the piping and equipment located between the fuel storage tank and the metallic filter up- stream of the absorbers at the end of the salt transfer operation when the pressure of the transfer gas was released. Because of this deposition of material, the calculated decontamination factor for 95 dpm per gram of U in salt . ¢ o per gram of U on absorber) is low. The calculated IF was lower at the start of run 1 when most of the niobium between the NaF trap and the absorbers was fluorinated and collected on the first absorber. The calculated DF's wvaried from 5 x 10“ at the start to 1 x 1010 at the end of fluorination. Therefore, notwithstanding the piping contamination, a considerably highe§ niobium IF was obtained in these studies than was obtained (5 x 10 ) in the CORNL Volatility Pilot Plant work using 30-day-decayed uranium. 115 Fluorine Utilization. — The efflciency of the fluorination reaction is measured by the percentage of fluorine (after the fluorine that is consumed by corrcsion is subtracted) that reacts with the uranium in the salt. Fluorine utilization is a function of melt temperature, fluorine flow rate, uranium concentretion, and dip tube submergence. Laboratory tests had shown a decrease in fluorine utilization from 8.0% to 3.8% as the temperature was reduced from 930°F to 8LO°F. Fluerine utilizations are shown in Teble I. The higher utilization (average, 39%) during the fuel salt processing was probably due both to lower gas velocities and grester dip tube submergence than were used in the laboratory tests. Utilizetion was very high before the start of UFg volatilization (averasge, 71%) and then seemed to become relatively insensitive to uranium concentration until nearly all the uranium was volatilized. Until 90 minutes before the start of UFg volatilization, the readings on the absorber inlet snd outlet mass flowmeters were the same, indicating the absence of any absorbable constituents in the gas stream. These reading were used to calculate the fluorine utili- zation. The utilization decreased when the fluorine flow rate was increased and also when MoFg and UF began to be vaporized. It was assumed that, 1f the fluorine flow rate is steady, the fluorine utili- zation does not change until volatilization of some component hegins. During the MoFg volatilization perilod (before UF6 was volatilized), the fluorine utilization was calculated by difference, using data obtained in the two previous periods and assuming that all the UFy was converted to UFg before the volatilization of UFg began. During the volatilization of UFg, the utilization was calculated from the increase in absorber weights. The inlet mass flowmeter indicated a high utilization toward the end of each run because of the accumula- tion of UFg in the fuel-storage-tank gas space. This caused e higher UFg concentretion in the exit gas stream near the end of the run. Corrosion. — Corrosion of the Hastelloy-¥ fuel storage tank durlng fluorination was one of the major difficulties encountered during the processing of the flush and fuel salts. Tests of Hastelloy-N coupons in the Volatility Pilot Plant and at Battelle Memorial Insti- tute hed shown this alloy to be superior to nickel (mainly because of the absence of intergranualr corrosion) and to be as resistant as any other material tested. The corrosion rate, calculated from the increase in chromium, iron, and nickel contents of the salt and from the amount of MoFg volatilized, averaged 0.1 mil/hour for the total fluorination period (i.e., 47 hours). FPission Product Behavior. — The principal fission preducts that form volatile fluorides are iodine, tellurium, molybdenum, ruthenium, entimony, and niobium. Jodine volatilizes as IF- and passes through the two NaF beds; it is removed by the caustic sZrubber both by reaction with KOH and by exchange with KI. At the start of proces- sing, we calculated that 404 me of I should be present in the fuel salt. Actual analysis of the caustic serubber sclutions showed that 116 ruthenium and molybdenum are formed directly by fission, and radio- active antimony has no long-lived precursors, the amounts of these isctopes in the processed salt a§§ small as compared with nicbium, which grows in from nonvolatile Zr. No ruthenium or antimony was detected on the uranium absorbers or in the scrubber solutions. Based on the lack of heat generation in the NaF' trap, we believe that the volatilized nlfibium constltuted less than 10% of the total amount of niobium (9 x 10™ curies) which would have grown in after reactor shutdown. Most of the niobium apparently was removed from the salt by deposition in the drein tank or in backflowing through the salt filter during salt transfer. At the end of this transfer operation, the transfer gas blowthrough carried a small smount of metallic nio- bium to the absorber filter. Although this filter was decontaminated prior to fluorination, some of the niobium deposited on it was un- doubtedly carried to the upstream plping. During fluorination, this niobium would be converted to volatile NbF: and would collect on the UFg absorbers (any niobium upstream of the NeF trap should be gbsorbed on the trap). Two days after the end of the fluorination operation (i.e., at the start of hydrogen reduction), a larger smount of niobium, estimsted to be 10 to 15 curies, was found on the absorber filter. Since J7Nb should have grown in at the rate of approximately 1000 curies per day, it is apparent that less than l% of the niobium in the salt was carried to the absorber cubicle. 5 only definite peak in a gamms scan mede of the filter was that of 5l\Tb at 0.77 Mev. Urenium Absorption. — Two types of NaF were used to absorb the volatilized UFg: & high-surface-sres (HSA) material (1 m 2 /grem) prepared by heating NeHF, pellets, and a low-surface-area (ISA) material (0.063 m /gram) prepared from NeF powder and water, and subsequently sintered at TO0°C. The ISA material wes originally epecified for the plant because of its higher capacity for both CrF5 (on the T50°F NaF trap) and UFg. This higher cspacity results from the slower surface reaction and the greater penetration to the inside of the ILSA pellets. The low MoFg retention shown during preopera- tional tests suggested that the reaction rate with UFg might be too slow for complete uranium absorption Yn%er the planned operation con- ditions. Subsequent laborstory tests confirmed the much slower reaction rate of UF, with the ISA material. The absorbers were, therefore, heated to increase the reaction rate, and the LSA NaF was restricted to the No. 1 end No. 2 positions of the group of five absorbers. A comparison of the two types of NaF is shown in Figure 8. Three ebsorbers loaded with each type of material were used in the first pogition, and three of each type in the second position, during six runs. In the first position, the NaF was exposed to a higher UFg concentration, which resulted in a lower loading than in the second position where the slower reection rate with the lower UFg concen- tretion permitted deeper penetraticn of the pellets. Except in one 118 Table I. Fluorine Utilization Corrected Fo Perc;ntage Run Flow Rate . . (std liters/min) Utilization Flush Salt 19.8 7.7 Fuel Salt, Overasll 18.8 39 1 - Before MbF6 Volatilization 7.1 g8 Before MoF6 Volatilization 36.0 77 During MoF Volatilization 31.5 43 During UF, Volatilization 26.2 32 2 - During UFg Volatilization 15.3 31 3 - During UF, Volatilization 16.7 29 L - During UFg Volatilization 16.2 3l 5 = During UF6 Volatilizetion 13.5 33 6 - During UF¢ Volatilization 16.9 13 288 me and 10 mc were collected during run 1 and run 2, respectively. This analysis provided a reasonable check on the icdine accountabil- ity in the fuel salt during reactor operation. However, the amount actually found was always less than tgglcalculated amount because of the loss of the iodine precursocr, Te, before decay. There was no indication that iodine deposited on the charcoal beds downstream of the scrubber. Tellurium exists primarily in the metallic state during reactor operation and is essentially removed from the fuel salt in this form by deposition and carryover to the off-gas streem; results of analy- ses showed that less than 1% remained in the salt. This residual tellurium would be converted to volatile TeFg during fluorination. Tellurium hexafluoride is not absorbed on NaF at any temperature, and is not removed in the caustic scrubber very efficlently. How- ever, since analysis indicated that the scrubber solutions contained no tellurium (i.e., the tellurium content was below the limit of detection), it is probesble that the fuel salt did contain less than 1% of the tellurium at reactor shutdown. Any tellurium passing through the scrubber would have been removed by the activated aluming in the soda lime trep. Molybdenum, ruthenium, niobium, and anbtimony also exist as metals during reector coperation, and as such, are continuously removed by deposition or by carryover to the off-gas stream. Since radioactive 117 URANIUM LOADING (kg) 8. 14.5 T 5 e NO.1 POSITION © NO.2POSITION 14.0 LSA \ NO.2 POSITION 13.5 -\ ' \ . 13.0 12.5 \ . \fi LSA NO. { POSITION 12.0 . 11.5 P \\ \ HSA NO.2 POSITION 11.0 AN ; HSA NO.1 POSITION . 10.5 . 10.0 \ 9.5 160 200 240 280 AVERAGE TEMPERATURE (°F) Comparison of LSA and HSA NaF loading. 119 320 cgee, higher loadings were obtalned at lower temperatures. In that particular case the absorber had been used in a previous run vhere g small amount of uranium was absorbed. In spite of the fact that the ILSA absorbers were operated at higher temperatures {to compensate for the slower reaction rate), the total loading was 13 to 14% higher than with the HSA material. References 1. Chandler, J. M. and S. E. Bolt, "Preparation of Enriching Salt 7LiF-233UF1,_ for Refueling the Molten Salt Reactor," ORNL-4371, March 1969, Osk Ridge National Iaboratory, Oak Ridge, Tenn. 2. ILindauer, R. B., "Processing of the MSRE Flush and Fuel Salts,"” ORNL~-TM-2578, (in prepsration), Osk Ridge National Laboratory, Oak Ridge, Tenn. 3. Cathers, G. I. et al., "MSR Program Semisnnual Progress Report for Period Ending August 31, 1968," ORNI~L3kl, October 1968, p. 325, Osk Ridge National laboratory, Osk Ridge, Tenn. L. Lindsuer, R. B. and C. K. McGlothlen, '"Design, Construction, and Testing of e Large Salt Filter," ORNI~-TM-2478, March 1969, Oak Ridge National laboratory, Oak Ridge, Tenn. 120 ENGINEERING DEVELOPMENT OF THE MSBR FUEL RECYCLE* M. E. Whatley, L. E. McNeese W. L. Carter, L. M. Ferris, E. L. Nicholson Oak Ridge National Laboratory Oak Ridge, Tennessee U. 8. A. Abstract The molten salt breeder reactor concept being developed at ORNL requires continuous chemical processing of the fuel salt, which 1is 7TLiF-BeFp-ThF, (72-16-12 mole %) containing about 0.3 mole % 233yF,. 1In order to minimize fuel inventory, the reactor and the processing plant are planned as an integral system. The main functions of the processing plant are to isolate the 233Pa (which is an intermediate in the production of 233y from 232Th) from the neutron flux and to remove the rare earth fission products, which constitute the major class of neutron poisons that are soluble in the salt. (The noble gases and noble metal fission products are not soluble in the salt.) The processing method being evaluated involves the selective chemical reduction of the various compo- nents into liquid bismuth solutions at about 600°C, utilizing multistage countercurrent extraction operations. Protactinium, which is easily separated from uranium, and from thorium and the rare earths, would be trapped in the salt phase in a storage tamk located between two extraction contactors and allowed to decay to 233y, Fluorination of the 233U from the salt entering this tank would be used as a process control method. Rare earths would be separated from thorium by a similar reductive extraction method; however, this operation will not be as simple as the protactinium isolation because the rare-earth-thorium separation factors are only 1.3 to 3.5. The proposed process employs electrolytic cells *Research sponsored by the United States Atomic Energy Commission under contract with the Union Carbide Corporation. 121 to simultanecusly generate reductant to the bismuth phase at the cathode and to return extracted materials to the salt phase at the anode. The practicability of the reductive extraction process depends on the successful development of salt-metal contactors, the electrolytic cells, and a suitable material of construction. 122 EXPERTMENTS ON PYROCHEMICAL, REPROCESSING OF URANIUM CARBIDE FUEL* Glenn E. Brand Supervisor, Analytical Chemistry and E. Wesley Murbach Supervisor, Sodium Chemistry Atomics International, A Division of North American Rockwell Corporation U. S. A, Abstract A summary of the experiments carried ocut at Atomics Internaticnal on pyrochemical reprocessing of uranium carbide fuel is described, Three processes were investigated; 1) the CARBOX process which is based on oxidation-carbothermic reductions 2) a nitride-carbide ¢ycle in which UC is converted to a nitride and then reconverted to the carbide; and 3) a fused-salt electrolysis process in which UC is anodized in a KC1-LiCl salt bath and the uranium cathode deposit is dissolved in mercury and converted to UC by reaction with propane at 350 to 600°C, Decontamination by all three processes was studied with lightly irradiated UC (nvt ~J017). The CARBOX and nitride processes remove about 75 per cent of the fission products, but plutonium losses are high. The electrolysis process removes most (98-99.9 per cent) fission products except zirconium. A 100 gram Scale oxidation exper- iment was carried out using UC irradiated to 15,000 Mwd/MTU, Batches of unirradiated UC up to 10kg were processed in scale up studies using a rotary kiln., Oxidative decladding studies were carried out in an apparatus capable of decladding 3 foot lengths of stainless gteel clad fuel. Cost studies indicate a savings of about 0,2 mil/kwh over aqueous reprocessing for a 1000-Mwe scdium cooled power reactor using a burn- up of 20,000 Mwd/MTU, *Work performed under the auspices of United States Atomic Energy Commission Contract AT(11-1)-GEN-8, 123 Introduction A program was carried cut at Atomics Intermational to investigate the potential of pyrochemical reprocessing to help achieve lower fuel cycle costs for uranium carbide fuels. It was hoped that a simpler method than the agueous process developed for the recovery of weapons-grade fissile material would beéfeasib e. With agqueous reprocessing, decontamination factors of 10~ to 10 were achieved and refabrication of low burnup fuels was carried out by direct methods. The build-up of heavy isotopes in high burnup recycled fuel will produce sufficient radiation even with com- plete decontam%n§tion to require special methods of handling during refabrication, 1 T%}s requirement suggests the consideration of 8impler processes(z’ which do not completely decontaminate the fuel, Three processes were emphasized in the present work;(q) 1) oxida- tion of the carbide to the oxide and then reduction with carboen to the carbide (CARBOX); 2) conversion to the nitride with nitrogen and reconversion to the carbide by heating in vacuum; and 3) molten salt electrolysis. Oxidation Carbothermic Reduction (CARBOX) The first step in the CARBCX process(u) (CARBothermic-reduction OXidation) is decladding of the fuel by oxidation., The resulting U0, is then oxidized to Uj0g, re-enriched by addition and reduced with carbon in a two step Process tc UC, The reactions involved can be represented by: UC +20, 202°C, U0, +C0,1 400°C > > Us0g U,0g +C 850°€, 300, +CO 3 300, +9C 1X00-1720% =ic 460t 3U0, +0 of Decladding A serieg of experiments was carried out to determine the feasi- bility o{ adapting the oxidative decladding method developed for U0, fuels., In this method the cladding is punctured at 1 ineh inter- vals along its length and exposed to air at ~400°C or steam at ~120°C, The UC reacts to form U02, which results in about a 30 per cent increase in volume, causing the cladding to split and the resultant 124 finely divided UO2 to fall out of the cladding. Typical specimens are illustrated in Figure 1, An apparatus was constructed which was capable of decladding three foot lengths., In one experiment 750 grams of UC clad in 0.723 inch OD-stainless steel, 0,010 inch wall thickness was punctured at one inch intervals and heated at #400°C in air, The rod was completely declad in 12 hours. In similar experiments using sodium bonded UC, the cladding was pierced and heated in vacuum to 600°C for three hours. The sodium was effectively removed and the exposed fuel was oxidized by steam at 120°C, Experiments using UC ifg§diated to 15,000 Mwd/MTU indicated the applicability of the process, An obvious modification is to chop the elements intc short lengths followed by exposure to oxygen or steam, Oxidation of UC A series of small scale experiments was fergied out to study the oxidation of UC with oxygen, CO, and steam.''’ It was found that the oxidation rate, and also the rate of reaction with nitrogen, was highly dependent upon the previous history of the UC. Uranium carbide which has been recently melted oxidized very slowly below about 600°C while aged (reactive) UC ignited in oxygen at about 300°C. It was shown that freshly arc melted UC was converted to the reactive form by exposing the moist air for a few days., Large slugs of hypostiochio- metric UC were more difficult to activate. The examination of both forms of UC showed no difference in photo- micrographs, x-ray diffraction patterns, and chemical analysis for uranium, carbon, and oxygen., There was some increase in hydrogen on aging but this was in the several ppm range. The only physical difference noted was a large increase in surface area (as measured by B.E.T.) on activation. In a typical sample the surface area increased from a few cm U0, +4C01 Below 525°C the react on 18 kinetically first order, above thais temperature 1t 1s zero order. A series of experlment? yas carried out to invesatigate scale up of the oxadation process, 9 These experiments were carried cut i1n a rotary kiln 16 inches in diameter by 21 inches long made of stain- less steel. It was rotated at 8 to 10 rpm. The kiln was heated electrically by movable clamshells. Reaction rates were controlled by controlling the temperature and flow of air. Several kilogram batches of are cast UC slugs were oxidized at a rate of 2 to E%kg per hour. The results of some of these experiments are shown in Table 2., As expected, unreactive material was difficult to oxadize, Some batches were converted tec the reactive form by exposure to azr at room temperature and oxidized readily. Carbothermic Reduction A series of smal sgale experiments was carried out to study the reconversion to UC. 10 Two methods were investigated. In the first 127 method the oxidized product, primgrily U 0gy was reduced with hydro- gen to UO,. The U0, was then reacted with carbon to product UC. 1In the second method the U308 was reacted directly with carbon, Table 2. 0Oxidation of Uranium Carbide in a Rotary Kiln Charge Sieve Analysis Run | Weight ?fib;‘)l Temx(’fg?t“re % (kg) =50 +200 -200 1 1,02 5.2 400 2 98 2 2.25 5.2 Loo 2 98 3 5,00 ‘5,2 520 1 98 il 4, o5* 4, 6-4,7 550 10 89 5 4,92 4. 6-4,7 550 12 86 6 4, 96% 4, 6-4,7 550 5 81 9 8,08 b, 6-4,7 550 12 75 10 10.01 4, 6-4,7 550 10 90 12 9.95 h,6-4,7 550 3 84 14 10.73 4,6-4,7 550 0 79 *The UC was only partially activated so wae treated in the kiln for further activation; results are a total ¢f three oxidation treat- ments, Experiments with UOZ utilized powdered mixtures of UO2 and graphite. In some experiments tThe mixture was pelletized using methyl methacry- late or stearic acid as a binder. The mixture was outgassed below 1200°C and then heated above 1%00°C. The reaction rates followed the gecond order equation where x is the fraction reacted in time (%) and k is the rate con- gtant., The rate constants followed the equation 10gk = 9.8 - ‘:'!'-6_,%0_0 where T is the temperature (°K). It was found that in the experiments using pelletized mixtures some carbon was contributed by the binder and a correction was applied to the amount of carbon added. Early attempts to prepare UC directly from U308 according to the equation 128 U308 +11C — 3UC +8cot resulted in a high and variasble carbon content.(lo) This was attri- buted to the fact that beth CO and 002 were produced and that the amounts depended on many variables, It was found that by carrying out the reduction in two steps the stoichiometry could be controlled. The reactions are represented by the following equations: U 850°C 0g +C 220°C, 300, +C0,, 3U0,, +9C 100°C, 3UC +6COt t The temperature was maintained below 900°C until the first reaction wag completed, The temperature was then raised to above 1300°C, Reaction rates were more rapid than observed for UQO, as the starting material, A}so the reaction rates were correlated by a 3/2 order rate equation.(ll Several scale up experiments were carried out to study the reaction of U 08 with UC. In these experiments the charge was ball milled, mixea with 1,15 wt per cent polyvinyl alcohel as a binder, and agglomerated. The mixture was then heated in an induction wvacuum furnace, These experiments were directed primarily toward the pre- paration of hypostoichiometric UC. The results are shown in Table 3. Table 3. Scale up Results on Carbothermic Reduetion of U 08 ) . Carbon gfilght Temperature (wt %) arge (eC) Theoretical Analytical (kg) 20 2000 4,66 Y, Gox 16 2000 4,68 b4, Th* 1.5 1750 4,51 4 73 0.12 1750 4. 60 4,54 0.12 1750 4,59 4,66 0.14 1900 4 48 4,81 1.2 1750 5.35 5.15 1.5 1750 4,69 4,92 *¥As-reacted. Other samples arc-melted before analysis, 129 Fission Product Behgvior Fxperiments with UC irradiated up to 15,000 Mwd/MTU showed that no fission products except rare gases and ruthenium arfiéfemoved by decladding and oxidation-hydrogen reduction to UO,. Appreciable rutheni removal can be achieved by increasing tfie oxidation tempera- 'l;u:r'e.?]'g[j1 In anither series of experiments three mixtures of irradiated UO (nvt ~20 7) and carbon were prepared which would pr?duse hypostoichio- metric, stoichiometric, and hyperstoichiometrie UC. Y These mixtures were heated at three temperatures in a vacuum induction furnacz and another set was arc melted, The results are shown in Table 4, Table 4, Behavior of Fission Products and Plutonium ‘ During Carbothermic Reduction Removal Element C:UO2 () 1560°C 1760°C 1940 Arc-Melted Pu 3,17 6 - 17 80 3,00 6 - 33 94 2.88 14 38 46 98 Ce 2,17 22 - 20 84 3,00 14 - hh 97 2.88 17 37 71 98 Sr 3.17 98+ - 99+ 99.9+ 3,00 o8+ - 99+ 99.9+ 2.88 ok 99+ 99+ 99.9+ Cs 3.17 97+ - 99+ 99+ 2.00 97+ - 99+ 99+ 2.88 97+ 98+ 99+ 99+ In another experiment U0, was mixed with oxides of cerium, samarium and neodymium and reduced with carbon. With a mixture which produced hypostiochiometric UC, essentially all rare earths were removed at 1800°C, With a hyperstoichiometric mixture 75 per cent of the samar- ium was removed below 1600°C, Heating to 1800°C removed essentially all the samarium, 20 per cent of the neodymium, and 15 per cent of the cerium, Calculations on the effect of scale up indicate that plutonium losses would not he serious on production scale carbothermic reduc- tion, but that the losses during arc-melting would be prohibitive., The most promising refabrication method would seem to be pressing and 130 gintering. Comparative Costs A study was carried out to compare fuel c¢ycle cg§ts for reprocess- ing by CARBCX with a conventional aqueous plant.(l Fuel cyele costs were computed for fuel recycled through a 360-Mwe sodium cooled power reactor., Calculations were carried out through nine cycles of 20,000 Mwd/MIU with reprocessing and re-enrichment between each cycle, Fuel reactivity was adjusted by the addition of an amount of enriched uranium so that the reactivity of the fuel at discharge was constant for each irradiation cycle, Burnup and re—ensichment requirements were made using the computer code AIMFIRE.(15 The results of these calculations are shown in Table 5. 2 \ Table 5, Cyeclic 35U Burnup Requirements for Recycled UC Fuel in the Reference SCR {(Burnup per Cycle = 20,000 Mwd/MTU) 100% Decontamination Decontamination by Process CARBOX Process Cycle Initial Burned Initial Burned %y g/ P g/mg | PPv g/g | 2P0 g/kg 1 33,94 19,04 3%,04 19,04 2 27,89 14,50 30,18 15.12 3 26,40 13,54 29,40 14,28 Y 26,04 13,33 29, 22 14,05 5 26,00 13,31 29,20 14,00 6 26,03 13,33 29,22 14,01 7 26.08 13,37 29.26 14,04 8 26,13 13,41 29,31 14,07 9 26.18 13,45 29.736 14,11 Average 27.19 14,14 29,90 14,75 The recycle of neutron absorbing fission products in fuels repro- cessed by the CARBOX process increases the consumption of fissile material but comparative fuel cycle costs indicate the burnup penalty is less expensive than the savings realized via the simpler fuel cycle., The costs are shown in Table 6. The CARBOX reprccessed fuel requires remote fabrication, but since high burnup fuels will contain isotopic impurities of the fertile and fissile materials, direct fabrication may not be feasible even with completely decontaminsted fuel, Fabrication costs and reprocessing costs were combined for the CARBCX process. The potential savings of ~0.2 mill/kwh represents a savings of $3.5 million/yr for a 1000- Mwe installation. 131 Table 6. Comparative Fuel Costs AEC Ref. On—Sltg - BTEeactor Charges Aqueous Plant aLacihy (mill/kwh) Aqueous CARROX {(mill/kwh) | (mill/kwh) Burnup 0. 806 0.806 0.856 Use 0.13%6 0.1 0.142 Loss 0.037 0.037 0.043 Conversion 0.125 0,125 0.019 Fabrication (SS clad) 0.350 0.350 0.479 Reprocessing 0.1%0 0,370 Shipping 0.113 0,041 0.011 Total 1.70 1.87 1.55 Nitride-Carbide Cycle The first step %fl the carbide-nitride cycle is the conversion of ) When UC is heated at 800 to 900°C in a nitro- UC to the nitride. gen atmosphere the reaction can be represented by huc +2N, —> 2U N, 4C 273 the carbide is pulverized during the process., The nitride is then reconverted to the carbide by heating in The reaction is reversed vacuum to 1300 to 2100°C. 2UéNj The UC can then be refabricated by arc casting or pressing and sinter- ing. Conversion to the Nitride Preliminary experiments were carried out using a thermobalance to +4C ——> 4UC +3N 1 study the rate of reaction of UC with nitrogen. The reaction rate of UC with nitrogen, as previously described for oxidation, was found to be highly dependent on the previous treatment of the carbide. Arc melted UC (even -16 +30 mesh) which has been recently prepared was unreactive to nitrogen at 800°C, After about 40 days exposure to laboratory air the UC became reactive. As with many solid-gas 132 reactions the mechanism appeared to be quite complex.(ls) A large number of experiments were carried out in which nitrogen was reacted with UC in a thermcbalance. The reaction rates were correlated by = K(1-x) 7/ where K, t, and x are rate constant, time and fraction reacted. The rate of reaction varied with time after arc melting, initial particle size, UC composition, nitrogen pressure, and temperature, The maximum rate occurs in the range of 900°C. The reaction goes virtually to completion in a few hours and the product is a finely divided powder which consists of a mixture of U2N3 and carbon., Reconversion to the Carbide The reconversion,pf, the nitride to th. carbide appears to take place in two stebs.( ) The first step w. ich converts the U’2N3 to UN T 2UéN3 —> 4N +17, occurs at a linear rate, The second step in which the UN is converted to the carbide 2UN +2C ——> 2UC +N27 follows a parabolic rate, Removal of nitrogen depends on temperature and carbon content.(lg) If the material contains less than 4.8 wt per cent carbon (hypo- stoichiometrie) sufficient nitrogen is retained to combine with the excess uranium. If the material contains 4.8 wt per cent or more carbon, the nitrogen can be reduced to a few hundred ppm at 1900°C, Arc melting usually reduced the nitrogen content somewhat and samples exposed to laboratory air lost carbon, presumably caused by the reaction with oxygen. Figssion Product Behavior Experiments were carried out on the gram scale using lightly irradiated UC (nxt‘:'lol7) to study the behavior of fission products. No fission product removal was found after the conversion to the nitride, although undoubtedly scme rare gases were released, The material was then converted to the carbide, The results are shov in Table 7. 133 Table 7. Fission Product Removal in the Reconversion Step Maximum | Time at gzgggft Removal % — Runf Temper- | Temper-{ ;4: ) Earths No.| ature ature ( Wt %) Cs Ce|Sr{| Zr-Nb | Ru| Pu Fxeluding (°C) (hr) nitrogen Ce 1] 1700 5 0.019 97 | 27197 2 5 6 51 21 1700 2 % 0.086 50 | 19|91 0 of 2 239 z | 1850 % 0.009 97 { 47198 7 15] 40 68 4| 1850 11 0.033 |91 | 25|93| & | 13| 34| 53 51 1950 2 0.010 98 | 60|99 15 30| 46 €9 6] 1950 1 0.009 |99 | 33195 17 -4 73 The product was then arc melted, and the results are shown in Table Table 8, Fission Product Removal in the Nitride- Carbide Cycle Including Arc-Melting Product Removal % Run 'I‘:Lrne Composition Rare No. Lz'ql.u;l c c 3 or-tb | R Earths min Nitrogen| Carbon ° © r B v EW Excluding Ce 1 6 0.009 5.20 | 299 | 87 |>99 36 | 271 84 95 2 2 0.0%0 4,75 [ >99 | 64 |~99 15 | 17] 65 85 3 2 0.008 5.55 1299 | 70 |>99 o | 221 66 86 4 6 0,008 5.25 [ 299 | 82 |>99 4o -178 g2 5 2 0.007 6.60 { 299 | 60 |99 b1 | 30} 53 70 6 6 0.005 6.45 | 299 | 73]-99 3 | 3] 72 84 Unless a way can be found to lower plutonium losses, arc melting would not be a satisfactory method for refabrication. Molten Salt Electrolysis A method for reprocessing uranium carbide based on molten salt electrolysis has been studied by Hansgen. This process is shown schematically in Figure 2, It consists of three steps; 1) electroly- 8is, Uranium carbide is dissolved at the anode and uranium metal is formed a8 a dendritic mass at the carhode; 2) dissolution in mercury. The cathode deposit is immersed in hot mercury. The uranium readily reacts to form a quasi-amalgam which separates from the occluded salt 134 ael 2. n ALy, 0 o PRy ELECTROLYSIS IRRADIATION IN REACTOR uc CONDENSER] ' 3 FUEL DISSOLUTION IN MERCURY FABRICATION T0 REACTOR FUE L : h Hg PRCPANE _| FORMATION OF uC ucC (POWDER) Flow Sheet for Reprocessing of UC by Molten Salt Electrolysis, phase; and 3) formation of UC. The quasi-amalgam is heated in an atmosphere of propane. The uranium reacts readily to form stoichio- metric UC and the mercury is distilled away. Although in theory any alkali or alkaline earth halide can be used in the molten salt bath, the KC1-LiCl eutectic was chosen because of its low melting point, 352°C which is below the boiling point of mercury. The occluded salt readily melts when the cathode is immersed in hot mercury and can be conveniently separated. In the experiments, the bath was prepared by adding approximately 10 per cent of either UCl; or UF) to the KC1l-LiCl eutectic and purified by bubbling chlorine through the molten mixture. The anode consisted of wafers cut from cast UC and supported by a molybdenum wire or contained in a graphite crucible immersed in the bath. The UC dissolved while the graphite remained inert. The reaction can be represented as e ——> U 4C 436" The uranium metal was deposited on a molybdenum cathode as a dentritic mass which contained appreciable quantities of occluded Balts, After a few preliminary runs, current efficiencies of from 50 to 60 per cent were obtained, The uranium deposit was immersed in mercury which was at a higher temperature than the melting point of the occluded salt. The salt melted and the freshly deposited uranium rapidly dissolved in the mercury., The amalgam was cooled and poured through a cone filter with a small hole in the bottom. The amalgam passed through the hole and the frozen salt remained behind, The salt free amalgam was heated to 350°C and contacted with a hydrocarbon gas such as propane, The reaction is 3U(Hg) +%H8 —_— ZUC +Hg +4H27 The mercury was then distilled away leaving stoichiocmetric UC as indicated by x-ray analysis, No detectable mercury remained,. In an experiment carried out to determine fission_product behavior during electrolysis, ten grams of irradiated UC (10~ nvt) was mixed with 90g of unirradiated UC to form the anode material. Six deposits were collected and each depeosit was dissolved in mercury and con- verted to carbide by distillation in a propane atmosphere., The results of this experiment are shown in Table 9. These results show that appreciable decontamination from most of the fission products can be achieved but remote refabrication would be required, The effect of the remaining fission products on nuclear reactivity should be neg- ligible. The behavior of plutonium was not determined due to diffi- culties with analysis, 136 Table 9. Observed Decontamination Factors for UC Electrolysis Hlew Deposits ment 1 2 3 5 6 Ba.f2.1 x 102 2.0 x 1o% 3.3 x 10% 5.8 x 10° | 2.0 x 102 4.2 x 102 Sr 7 x20316,7 x 105 5.4 x 105 - |30 = 103]5.2 % 107 RE [1.1 x 107 [ 5.9 x 20| 4.1 x 107} 5.6 x 107 | 8.5 x 107|4.3 x 10 Ce |85 65 5% 5 58 86 5 51 Ru - - & x 10 - 8 x107]1.1 x 10 7r | 1 1 1 3 2 3 In an attempt to eliminate the dissolution step a series of experiments was carried out to, determine the feasibility of using a molten metal as the cathode.(go) The uranium dissclved in the molten metal a8 it was electrolyzed. Two reguirements of the metal are that it have a low melting point and be compatible with the LiC1-KC1-UCl» salt bath. Lead, zinc, bismuth, and cadmium were used as cathodes.” Because of potential problems with vaporization, mer- cury was not used, The metal cathode was stirred with a tantalum rod, and the salt bath was agitated by bubbling argon. These experi- ments were carried out at 450 to 500°C, A summary of some of the results is presented in Table 10. Table 10. Electrolysis of UC Using Liquid Metal Cathodes U Content of Salt - ggn Cathode (wt %) Cathode gfflclency Initial Final I Cd 2.4 4,4 21 8 Zn b4 3.6 69 7 Zn 3.6 0.5 88 10 Zn 2.2 2.4 91 14 cd 15.3 18.0 3.1 16 Pb 6.0 4.1 86 19 Zn 3.4 12.5 2 The results show the adverse effect of over 10 wt per cent U in the sa&t bath. Apparently, due to cyclic processes involving U+2, U™ and U02++. 137 Coneclusion Three ~tential methods for reprocessing UC have been demonstrated. It appears hat arc-meltin -casting would not be satisfactory for refabrication unless plutc um losses can be controlled, The elec- trolyeis process offers sv stantial decontamination. The CARBOX process offers a potential saving of ~0.2 mill/kwh when compared to an on-site aqueous reprocessing plant. References 1. Zebroski, E., L., H, W. Alter and G, E, Collins, "Plutonium Fuel Fabrication and Reprocessing for Fast Ceramic Reactors," GEAP-3876, February 1962, 2. Burris, L., Jr. et al, "Pyrometallurgical and Pyrochemical Fuel ‘ Processing," presented at the Third United Nations Internsational Conference on the Peaceful Uses of Atomic Energy, Geneva, Switzerland, August 31 - September 9, 1964, 5. Sinizer, D. I., et al, "Terminal Status Report for the Processing Refabrication Experiment," NAA-SR-3%269, November 1959, 4, Murbach, E. W. and G. E. Brand, "Pyrochemical Reprocessing of Uranium Carbide Summary Report,” NAA-SR~-11340, August 1965. 5. Bodine, J, E., I. J. Groce, J. Guon and L. A. Hanson, "Oxidative Decladding of Uranium Dioxide Fuels,™ Nuclear Science and Engineering, Vol., 19, No. 1, May 1964, pp. 1-7. 6. Bodine, J. E., J. Guon, R. J, Sullivan and F. W. Gandolfo, "Reprocessing Studies on Irradiated Uranium Carbide Reactor Fuel," NAA-SR-7511, December 1962. 7. Murbach, E, W.,, "The Oxidation of 'Reactive' Uranium Carbide," Transactions of the Metallurgical Society of AIME, Vol., 227, 1963, pp. U88-451, 8. Murbach, E, W, and W. D, Turner, "Oxidation of Uranium Carbide by Carbon Dioxide," NAA~SR-TU82, December 1962, 9. Strausberg, S., "A Rotary Kiln for the Controlled Oxidation of UC," NAA-SR-10485, June 1965. 10, Smiley, W, G.,, "Oxidation-Reduction Reprocessing of Uranium Carbide Reactor Fuel I. Carbothermic Reduction of U02," NAA~SR-6976, March 1962, 11, Murbach, E., W, and S, Strausberg, "Preparation of Uranium Mono- carbide from Uz0g," Nuclear Metallurgy, Vol. X, Compounds of Interest in Nuclear Reactor Technology, 1964, 138 12, 13- 14, 15. 16, 17, 18. Colvy, L. J., Jr., "Ruthenium Removal from Irradiated UO, by Reaction with Oxygen to 1300°C," NAA-SR-Merio-6107, February 1961, Smiley, W, G., "Oxidation-Reduction Reprocessing of Uranium Carbide Reactor Fuel II. Behavior of Plutonium and Fission Products," NAA-SR-10738, June 1965. Mattern, K. L, and L, J. Colby, Jr., "Comparative Fuel Cycle BEvaluations, Low Decontamination Pyroprocessing, and Agueous Reprocessing Part II. UC Fuel in a Thermal Reactor,” NAA~SR-9335, February 1965, Blaine, R. A., "AIMFIRE, A Fuel Economics Code," NAA-SR-6706, October 1961, Hanson, L. A., "Reprocessing of Uranium Carbide by a Nitride - Carbide Cycle I. Kinetics of Nitride Formation," NAA-SR-8388, October 1963, Hanson, L, A., "Reprocessing of Uranium Carbide by a Nitride - Carbide Cycle JI. Kinetics of Nitride Conversion to Carbide," NAA~SR-9161, October 1964, Hanson, L. A., "Reprocessing of Uranium Carbide by a Nitride - Carbide Cycle III. Complete Cycle and Fission Product Study," NAA-SR-9278, October~1964, Hansen, W, N., "Reprocessing of Uranium Carbide by Molten Salt Electrolysis,”" NAA-SR-7660, March 1963, Iverson, M, L., and R. J. Sullivan, "Electrolysis of Uranium Carbide in Fused Salt Using Molten Metal Cathodes," NAA-SR- 10737, June 1965, 139 TECHNOLOGICAL AND ECONOMICAL ASPECTS OF IRRADIATED FUEL REPROCESSING BY FLUORIDE VOLATILITY METHODS IN FRANCE G. Manevy and Y. Rochedereux Centre D'Etudes Nucleaires de Fontenay-Aux-Roses 92-Fontenay-Aux-Roses France 141 SIZING THE CHEMICAL REACTORS FOR FLUORIDE VOLATILITY PROCESSING OF FAST REACTOR FUEL* Gustaaf J. Spaepen Head of Chemical Technolegy, SCK-CEN Mol, Belgium Abstract The choice of the type and size of the fluorinator in fluoride volatility processing depends on many factors such as heat dissipation, criticality, capacity and mass transfer. An outline is given of the present state of development studies at SCK~-CEN, Mol for the experimental determination of the main factors for design and operation of an industrial unit. The project is aimed at the reprocessing of the fuel of one 1000 MWe fast breeder reactor. .Introduction The fluoride volatility processing method is considered to be a total or partial alternative to other processing methods for fast reactor fuel. The present R and D program along this line in the SCK-CEN laboratorie?13t Mol is based on the earlier work of J. Schmets and col. « The goal of our program is the accumulation of the data, necessary to evaluate a prototype reprocessing facility based on FVP. Use of one or more volatility steps in an aqueous reproces- sing scheme is considered to be likely and hence smaller programs are set up at the side, in order to evaluate the hybrid possibilities after completion of a major FVP-step. * Agreement of Co-operation EURATOM/BELGIAN GOVERNMENT No. 016-65-1 RAP B and No. 014-65-1 RAP B 143 The all-volatility line will be composed of a number of steps including transport, reaction, absorption, condensation.. of gases and solids. These operations will have to be performed on solids of as yet unknown reactivity, impurity content, physical properties, etc. The gases to be used in the flowsheet are better known. Fluorine is the only gas to volatilize the Pu-compounds while the advantages of different chleorine- and bromine-fluorides for the UF6-volatilization have been widely discussed (2, 3, 4). Furthermore, a high heat load in the process is to be reckoned with because of radioactivity after short cooling times and exothermicity of most of the chemical reactions. Recovery of the valuable plutonium should be virtually complete. Like most of the laboratories working on FVP, we have chosern the fluidized bed as the main equipment of the flow- sheet. Development of its technology is seen in the frame of continuously feeding powdered fuel of core and axial blanket to the UF ,-reactors. Countercurrent flow of fluorine and/or C1F/C1F és provided with intermittent condensation of PuF and dilation with NZ' A schematic drawing is given in fig.1. General Reactor Conditions Most applications of continuous fluidization in non- nuclear industries have the following features in common : = beds of large diameters and large production rates -~ Dbed composition of standardized grain sizes and bulk densities - recycling of the reactant gas - c¢yclone recuperation of elutriated fines - high linear gas velocities (10 - 100 v__) - usually only one fluidized bed in the Circuit. The situation of a FVP-plant based on fluidization will be radically different : - bYeds of small cross-section and moderate out-put - bed composition of varying grain sizes and containing solids of widely different densities - recycling should be kept to a minimum because of costly purification devices in the secondary lines - 1large quantities of elutriated fines and necessity of complete recuperation hence use of "absolute" dust filters - linear gas velocities as low as possible to minimize elutriation - several beds in series. 144 ————— o e s e l'- From _.! Power | - . Core UFg reactor L Head-end _: Axial blanket 6 ““““ ) Nol*Fy i, 1“3\ solid residue(Pu) fufs Fp+PuFg N No+Fo P, Fg reaction Py F e Fa+P, Fg No P, Fg reaction N2 +R fufs Fa+FR Fs | P, Fg reaction e ] l fo R, Fs purification | T e, e ] Fa to : Solid waste 1 l—— _ 1. Scheme for Fluoride Volatility Processing (FVP) of fast reac- tor fuelo 145 Moreover, because of the high heat load and the presence of low melting compounds in the bed, accidental loss of fluidization would lead to caking of the bed and destruction of the equipment. Since linear velocities will be low, close fluidization control by sensors is essential. According to these general conditions for the use of fluid beds in FVP and because of the unknown characteristics of the high burn-up fuel of the future, several ground rules for the development program at Mel were chosen @ - determination of the fuel powder particle size distribution as a function of the fuel type and the milling characteris- tics = determination of the microkinetics of the different fuel powders by thermogravimetry - confirmation of the microkinetics in batch type fluidized beds of laboratory scale (macrokinetics) - determination of minimum fluidization velocities, fluidization stability and heat transfer coefficients for mixtures of extremely variable grain sizes and densities - elutriation behaviour of the same mixtures for linear velocities between 1.2 and 1.8 v £ - demonstration of reactors in serles with solids transport by air-1ift and with representative on-line sampling of solids and gases, leading to residence time determinations fer both - development of a fluidization sensor for linear velocities between 1.2 and 1.8 ¥ and of recuperation devices other than porous filters. Development Program The resulting development program is summarized in fig.2. Only those tasks related to the development of the main reactors for UF, and PuF, are indicated in this diagram. The knowledge thus gained on fluidized beds of unusual operating conditions, can directly be applied to the purification circuits which consist primarily of less critical absorption units, decomposers and UF6-U02 converters. Before giving some illustrative results, we first indicate briefly the progress made in our program. All the batch-type units are in full operation including : - thermal balance for unirradiated and for short-cooled, high burn-up fuel - a 100 kg UF,.-reactor automatically operated, with on-line gas resistivity cells, UFs-condensation and absorption units and scrubber - a 10 kg UF,.-reactor, remotely controlled, for low activity fuel - an on-line gamma spectrometer is installed 146 LE1 Mitling of Sintered Molten (U, Pu) 0y Coprecipitated Characterization of Alumina Fluidization Behaviour vmf, elutriation Sampling, sensor, Heatl transfer solid mixtures ! Dynamic behaviour of reactors in series: airlift sampling I 3 Microkinetics Macrokinetics Unirradiated Hfghly irrad fU,Pul)Os ! Demonstrafion in single continuous reaclor Unirradiated Irradiated } (U.Pu) 0 ¥ Evaluation 0 Proto type Development program at SCK-CEN, Mol. - two PuF -reactors, of 500 g each, with continuous solid sampling during operation - a hot cell for conditioning and sampling with instruments for fission gas release - a demonstration of continuous slab reactor with airlift transport of solids - a cold fluidization laboratory including units for macrokinetica, a low temperature microwave and cyclone equipment. Descriptions of most of the equipment and installations have been published (5, 6, 7, 8). The microkinetics of the reactions of F,, ClF and ClF3 on U0, and on simulated fast reactor core fuef are well established. The relation to the macrokinetics in fluidized bed is completed for UO, and underway for core fuel with simulated high burn-up. The same holds for the determination of the fluidization behaviour while the dynamics of slabs in series as well as the construction of continuous single reactors is scheduled for the end of 1969 and early 1970C. Results In order to illustrate the respective studies already completed in our program a selection of typical results is presented. More details can be found in the cited references or in SCK-CEN progress reports, which can be made available on request. Microkinetics (9) The microkinetics of fluorinating gases on pure UO_ -grains fit the rate eqq7§ion, according to the decreasing sphere model : (1 - F) = 1 - R't. Although physically speaking, this model is meaningless for (U Pu)O2 and F,, the same expression can be used : (1 - c)5/3 =21 - r' ¥t (table I). In fig. 3 the dependence of this R'' on F_-concentration is given for (U, 20 Pu)O, obtained by melting, hence with Pu in solid solution. The %exture of the (U,Pu)0, solid solution after volatilization of only part of the UO_ is shown in fig. 4. As could be expected, the cracked grains splinter into small fragments by the mechanical action in a fluidized bed. Consequently the macrokinetic results for this case of "molten" fuel differ very much from the thermogravimetric results, although it still seems possible to relate one another. 148 6¥%1 I R’ x 10 -] ) 200p- . _x ] ,,4””/ o * - - - - - 100t - _— ExperimentoI’/resuHs ' 2 S g R 29J?XCfi} s, Z z i e, Y /4 Y 1 1 | , _LF2 vol% 5 10 20 30 3. Influence of Fo> concentration on the rate constant R!' in the reaction (U,Pu)0p with Fo, Pu)0p at 50 wt. % reaction with Fp. 3 Micrograph (xL50) of (U Lo 150 Table I. Reaction of (U,Pu)0, (22 % PuO,) with F, (a) Influence of temperature (U,Pu)O, 87 um 3 20 vol. % F, in argon ; linear gas velocity : 10.5 cm min™ Run ¢ R' x 100 - log R' 10° ¢~ UPO 281 450 383 1,417 1.38 UPO 282 420 242 1.616 1.44 UPO 190 I Loo 136 1.863 1.49 UPO 190 II Loo 138 1.860 1.49 UPO 283 350 45 2.347 1.60 (b) Influence of grain size 400 °C ; 20 vol. % F, ; 10.5 cm pin | Run Grain size R' x 104 " UPO 191 48 207 UPO 190 I and II 87 137 (c) Influence of F, concentration (U,Pu)o2 87 um ; 400 °C ; 10.5 cm min™ | Run F2 vol. % R' x ‘1()‘[+ UPO 286 5 61 Up0 284 10 g5 UPO 190 T and II 20 137 UPO 285 30 194 UPO 189 I 50 245 UyPo 189 II 50 261 (U,Pu)O2 48 um ; 400 °C ; 10.5 cm min-’I Run F2 vol. % R X 10E4 UPO 191 20 207 UpP0 280 50 368 151 Macrokinetics (9) The macrokinetic relationship for UO2 in C1F,; and F, still follows the same rule : (1 - F)1/3 = 1 Z k't. Formatisn of intermediates UO,F., and UF4 lead however to an induction period which was not observed in microkinetics. The chosen example (fig. 5) shows almost constant fluorine efficiencies in the linear part of the curve and an abrupt decrease towards the tail. Since the curve was obtained in a batch experiment it is c¢lear that fluidization of the mixture changed too much with increasing proportion of fines. The shown curve is in fact a function (U ~ concentration, time) and sets of curves for varying fluidization parameters were used to extrapolate to continuous operation. The model was also used to estimate the operation conditions for the 100 kg demonstration unit for UFg-proeduction. Flutonium mass balance in the semi-pilot installation The results of the last step (F2 on Pu - containing residue) of a series of experimentS in the plutonium semi- pilot are summarized in table II. These experiments on macrokinetics were carried out according to a fractional factorial design (10). C1F; was used for the UFg volatili- zation and Fp for the PuFg-step. Table II lists the observations during the Pqu-formation and the over-all mass-balance of the experiments. Fluidigation behaviour Fig. 6 is illustrative of the fluidization behaviour when fines are present in the bed. Fluidization has been followed visually, by pressure drop measurements and by the microwave technique. Three regions can be observed as indicated on the graph. It is worth mentioning that the experimental results are very reproducible but differ greatly from the values calculated according to literature data (11). Powder characteristics The results of milling experiments on different types of fuel are summarized in table III. In combination with the correct choice of alumina the fuel powder is directly fluidizable. However, a better specification of the fraction (= 37 u) is of importance with regard to elutriation losses and cyclone or filter performance. Particle size measurements are therefore scheduled using a sedimentation balance technique. 152 ¢al i (1-F) V3 14 . - - . k'=70.10 4min ! 0.9 1 120°C , 10 Y% ClFy in Ny Vi=1lem s-1(STP) 08 220 g UOyin 330g Al,03 0.7 0.6- [ ] 0.5- * B L T 0 20 40 60 80 100 120 min 5. Decreasing sphere model applied to over-all reaction rate in fluidized bed. +0.9 Pl {412% (-40+60mesh ) & Theorelical 0—0(Alp 03(-40+60 mesh) *cm {U02(—37;.rm) (~37pm) values ! ‘ wilh variable bed heights ;h const bed height BAD - UNSTABLE GooD _ 20" 10+ A b A A 0 10 20 30 7 50 60 70 a0 90 100 whe Al503 100 Whe o)) 6. Fluidization behaviour with fines, Table II. Plutonium mass balance in the semi-pilot installation Experiment Time Pu- Pu- Sampling Pu- Pu mass- Remarks concentration absorption during concentration balance Last step on NaF the expe- 1in reactor -1 % init. Pu riment residue No. h ng g % init.Pu wt. % Total % Spu-26 F, 11 b4, 29 70.55 25.72 0.152 98,91 Sintered bed FP + CsF SPu-27 F2 11 hi 47 33.37 1.68 0.034 36.82 Filter failure SPu-28 F2 8 L b7 92,29 4,66 0.022 97.62 - SPu-29 F 8 Ly 29 56.00 5.74 0.063 63 .44 Filter faijilure 2 . . FP + CsF No sintering (1 wt.%) gg1 Table III., Size analysis data of the different UO2 lots Particle size Average range particle nm diameter 1 - 37 43774 +74-125 +125-149 +149-177 +177-250 +250 =TW/D) nm Weight percent w Lot 1 a sintered 8 21 25.5 6.9 9.8 13.3 15.5 77 U02 Lot 1 b idem-other 26 15 1% 5.1 5.5 13 23 50 milling parameters Lot 2 idem- 20 26 24 12.9 6.4 5.7 11.6 13.3 L6 experiments Lot 3 "molten"UO, 36 23.5 19 5 recycled - Lot 4 "molten" 36 21.5 18.5 5.5 7.5 recycled - U02-20Pu02 Conclusions The outlined program was chosen to lead directly to operating conditions for a pilot unit. It is worth mentioning that scale-up is not a matter of extrapolating the data for emall diameter beds to very large diameter beds. Instead, the concept leads to a slab reactor of small width but considerable length which is divided into compart- ments of about "laboratory"-scale. Once the system of fluidized beds in series is under conmtrol, this concept avoids the presently too complicated scaling-up of rather unusual bed composition. Parallel to the technological development work, a i} theoretical group is working on modified mathematical models for our fluid beds in order to evaluate the fresh data on mixtures for potential applications outside the scope of the programe. References 1. Schmets, J. and col., "Retraitement par fluoration de combustibles a base de bioxydes mixtes'", Colloque sur l'emploi du plutonium comme combustible nucléaire, Bruxelles, 13-17, 1967, I.A.E.A. 2. Henrion, P., Camozzo, G., Fontaine, J., Leurs, A., Schmets, J., Stynen, A., "Réactions entre quelques composés plutoniféres et le trifluorude de chlore", communication pour la Journée Nucléaire du 10 février 1968, Bruxelles. 3. BHenrion, P., Camozzo, G., Coenen, F., Schmets, J., "Réactions entre l'uranium et le trifluorude de chlore! communication pour la Journée Nucléaire du 10 février 1968, Bruxelles. 4, Breton, D.L. and col., "A Conceptual Study of a Fluoride Volatility Plant for Reprocessing Light Water Reactor Fuels", ORGDP K-1759, Dec. 24, 1968. 5. Broothaerts, J., De Coninck, A., Heremans, R., "Le circuit de ventilation et son systéme d'épuration de 1'air, dans les laboratoires chauds pour 1l'étude du retraitement par volatilisation des fluorures", XXXVII® Congrés International de Chimie Industrielle, 4-12,.11,1967, Madrid. 157 1. Heremans, R., Lambiet,C., "Manipulations et transferts d'échantillons dans les installations chaudes du programme de recherche sur le retraitement des combusti- bles par volatilisation des fluorures? 14th Meeting of the Euratom Hot Laboratory, Karlsruhe 20-23th Sept. 1967. Goossens, W., Heremans, R., "Semi-pilot installation for fluoride volatility reprocessing'", Symposium on Dry Reprocessing, European Atomic Energy Society, 28-29th October 1968, Mol. Vandersteene, J., Camozzo, G., Heremans, R., "Un laboratoire pour 1'étude de la récupération du plutonium dans un procédé de retraitement de combustibles nucléaires par voie séche", Industrie Chimique Belge - T.33 - No.7-8, 5.C.K.~C.E.N., “"Quarterly Report No. 35" R.2479, October 1 to December 31, 1968, Vandersteene, J., Parthey, H., Spaepen, G., "On Fluorination of U0, - 2 wt. % PuO, Pellets by Chlorine Trifluoride", S.C.K.-C.E.N. Internal Report 1969. Venkitakrishnan, G.R., and Bhat, G.N., "Minimum Velocity for the Gaseous Fluidization of Dissimilar Materials", Indian Chemical Engineer, July, 1965. 158 REPROCESSING OF THTR FUEL ELEMENTS BY H1GH TEMPERATURE TREATMENT AND CHLORINATION Jurgen Hartwig and Klaus H. Ulrich Fried., Krupp GMBH, Central Institute for Research and Development Germany For the reprocessing of irradiated nuclear fuel elements con- taining the fuel in form of coated particles, a process has been developed by which the coating is destroyed at high temperatures to such a degree that the fuel can be extracted from the fuel element or from a capsule containing the coated particles by chlorination. Coated particles with metal carbide intercalations of SiC, for instance, can also be handled without difficulty, The gas mixture composed of volatile fuel chlorides and, possibly, of fertile and fission product chlorides can be cooled and further treatment of the condensed chlorides can take place in an aqueous solution process, If development work is continued along the present lines, fractional distillation and revolatilization appear sufficient to permit the fractionating of halides as well. The investigated synthetic fuel elements of the THTR types con- tained the fuel and fertile material in the form of particles coated with pyrolitic graphite or with a combination of graphite and sil- icon carbide. The coated particles themselves were embedded in a graphite matrix of normal porosity (approx. 20-25%). The fuel elements were exposed to temperatures of 2400-3000 deg. C in a high- temperature furnace having an argon atmosphere. This caused the coating of the particles having a carbide core to be destroyed by decomposition of the carbon and by the formation of intercalation compounds., Polished sections show that after this high-temperature treatment, the carbide fuel is freely exposed in the porous graphite matrix so that it can be extracted via the gas phase. This unlock- ing does not have the effect of changing the original shape of the fuel elements. In the case of elements containing the fuel or fissile material in the form of coated particles with cores of 159 U-Th-mixed oxide of the composition U-Th-03, this mixed oxide is reduced upon high-temperature treatment to a mixed carbide by the carbon of the coatings, (u Th)o2 + (x + 2)C = (U Th)cx + 2C0 At a temperature of 2900 deg. C the resultant CO-equilibrium pres- sures are sufficient to cause the particlie coatings to burst, the matrix of the fuel elements disintegrating at the same time. Upon subsequent extraction of the chloride, more than 99% of the uranium was recovered. Kinetic investigations were carried out and condensation equilibria measured. The pore space of the graphite matrix and the distribution of the pores were determined. The optimum flow velocity of the chlorine was determined by gravimetric observation of the chlorination rate. The temperature dependence of the chlorination process and the extraction of the chlorides formed were investigated at temperatures between 200 and 1100 deg. C. The apparent energy of activation was found to be 4,5 kcal/mol. This, like the resultant 7372 dependence, suggest that gas diffusion through the matrix graphite is a step determining the rate of reaction. Separate condensation of uranium, thorium, and fission product chlorides was investigated at different temperatures, Process design and plant layout for the continuous operation of this process are outlined, 160 HALIDE VOLATILITY PROCESSES Chairman: S. H. Smiley Nuclear Materials and Equipment Corporation Apollo, Pennsylvania, U.S.A. 161 PILOT FLANT EXPERIENCE ON VOLATILE FLUORIDE L REFROCESSING OF PLUTONIUM M. Ao Thompson, Re Se Marshall and Rs L. Standifer The Dow Chemical Company, Rocky Flats Division, Golden, Colorado U. S. A, ABSTRACT A pilot plant fluoride volatility system for the recovery of plutonium from impure oxide and residues has been built at the Rocky Flats plant of The Dow Chemical Company. Studies have been continuing for about one and one-half years. The results obtained from this system have been encouraging and indicate that plutonium recovery and purification from impure oxide is feasible, Several significant chemical and engineering problems have been encountered and resolved, Optimization and scale up of the process will require further studies. A Prime Contractor for the U. S. Atomic Energy Commission Contract AT(29-1)-1106 163 INTRODUCTION The Dow Chemical Company operates the Rocky Flats plant, near Denver, for the Atomic Energy Commission., The plant is part of the AEC's Albuguerque Operations Office, which is charged with major responsibilities for research, development, production and storage of nuclear components, The work performed at the plant includes phases of fabrication and assembly of plutonium parts,. During the fabrication phases of the operation, plutonium scrap is generated., Because of the value and the radiocactive properties of the plutonium, it 1s necessary to recover the plutonium and recycle it through the fabrication processes rather than discard it, The present plutonium recovery operation at Rocky Flats is an aqueous process using nitric-~hydrofluoric acid solutions and ion exchange for purification., The process is well known and has been successfully used at Rocky Flats for 15 years; however, it does have significant disadvantages including a high recycle rate, major corrosion of process equipment, many process steps and hand operations, and a high generation rate of agueous and solid wastes requiring further processing. Because of the high corrosion rate of the process equipment, major replacement is required every five to 10 years. Volatile fluoride reprocessing of plutonium has been investigated at Rocky Flats for the past five years as a possible replacement for the present agueous recovery system, Potential advantages of a volatile fluoride process to Rocky Flats operations appear to be: l., Fewer process steps and less hand operations should result in lower operating costs and less exposure of personnel to radiation, 2o More complete plutonium recovery from oxides and possibly from other residues should be obtained, 3« A lower generation rate of aqueous and solid wastes requiring further processing. L4e A more continuous and remote system should lower costs and allow increased shielding to minimize radiation exposure of operating personnel. Studies were started on a gram-quantity static bed systeml’2 and were baseg largely on earlier work carried out at Argonnea’ 4,8 and Oak Ridge National Iaboratories, These studies indicated that the process is feasible to separate plutonium from impurities normally found in the Rocky Flats process streams, 164 A l-kg-scale pilot plant, utilizing a fluid bed, was therefore designed and built, and plutonium experiments were started in December, 1967. The purpose of the pilot plant is to determine the feasibility of the process on a larger scale, to test the design and operation of major process equipment on a larger scale and to obtain operating and design data to be applied to a larger (12 kg) production prototype system planned for operation in 1970. PILOT PLANT Figure 1 is a flow sheet of the pilot plant system, The fluid bed is a 2-ine,-diameter vessel, 4 ft high, with a 5-in,~diameter disengaging section, The bed is heated by a three~zone muffle furnace, Prior to entry into the bed, the fluidizing gas passes through a preheater capable of heating the gas to 550°C. Additional heaters are installed on the lower part of the bed for more complete temperature control, Dual nickel fiber filters in the disengaging section prevent particle removal from the bed, Blowback of the filters during the bed operation is possible to prevent filter plugging and excessive pressure drops across the filters, The cold traps are L=in,-dlameter vessels approximately /4 ft long. The cooling surface in one trap is longitudinal and in the other, transverse, Fach trap has 12 sq ft of heat transfer surface, Cooling of the traps is provided either by a refriger- ation unit capable of cooling the system to =40°C or by liquid nitrogen capable of -70°C, Strip heaters permit raising the temperature of the traps to 300°C for removal of PuF, (by refluorination) deposited by the radiolytic decomposition of PuFg. This decamposition is equal to about 1.5 percent of the PuF6 present per day. The traps are piped so that they can be used either in series or parallel, The hydrogen reductor is 2 in, in diameter by 18 in, longe It is designed in such a way that different techniques can be tested to reduce Pufg to PuF,, Methods that have been evaluated include a hydrogen-thérmal reduction using a hydrogen-fluorine flame for heat, hydrogen-thermal reduction using external resistance heaters, and thermal decomposition on a static bed of PuF, or Pu0,., Thermal decomposition on a fluid bed of PuF, or Pu0, was alsg tested, The most successful method to date hés been a hydrogen=thermal reduction using externmal heaters. Packed bed sodium fluoride traps are used to trap any PuFg escaping through the cold traps or the reductor, The traps are heated at 150°C during operation by external strip heaters, 165 991 % ] o HYDRGEN ” HYDROGEN REDUCTICON | FLUID BED | ] 4 Y | e L HEATER l:! D TRAP TO BOX T 1 PREHEATE ' CHARCOAL —— FLUORINE - RECYCLE COMPRESSOR HYDROGEN BURNER ARGON KOH SCRUBBER Figure 1, Fluoride Volatility Pilot Plant Flow Sheet Charcoal traps are used to dispose of any unused fluorine, These traps are externally air cooled to dissipate the heat from the exothermic carbon-fluorine reaction and are maintained at no more than 200°C. A potassium hydroxide scrubber removes HF formed during the hydrogen reductlon of PuF,, The scrubber is a 5-in,-diameter Raschig ring=-filled veasei. A canned centrifugal pump recircu- lates the solution, Excess hydrogen is burned and any remaining off=gases pass through a water scrubber and into the bullding exhaust system, The fluid bed, cold traps, and reductor are fabricated from nickel 200, All lines are 3/8-in, Monel tubing with compression fittings. All lines used to transfer PuFy are equipped with heating tape to preclude PuFg condensation and subsequent line plugging, The 120 valves in the system are all Monel bellows=- sealed globe valves with metal seats, About half of the valves are alr operated and half hand operated, The fluorine recycle compressor is a remote head, reciprocating diaphragm compressor with an intermediate chamber filled with a fluorinated oil, The process end diaphragm is Monel, A variable speed drive unit provides flow rate adjustments, The entire system, except the compressor drive, is enclosed in a glovebox in order to contain the contamination, The exhaust air from the glovebox passes through a water scrubber to remove any fluorine that may have leaked from the system into the box, About 30 separate experimental runs have been made, including four initial runs using uranium to check out the system, Each run has been designed to obtain specific data on one or more steps of the operation. Therefore, no extended production-type runs to test the potential capacity of such a system have been made, The reliability and service life of the equipment have also not been established because of the limited running time to date, In some cases, extensive modification of the system was made between runs, RESULTS Preliminary runs on the pilot plant system were made using depleted uranium in order to check ocut the equipment and make any final changes before introducing plutonium contamination,. After several runs and some equipment modification, satisfactory performance was achieved and plutonium studies were started, To date a8 total of about 20 kg of plutonium has been put through the entire system, 167 Fluid Bed Conversion of PuO, to PuFy Figures 2 and 3 show the original designs of the fluid bed. The original gas distribution system, as seen in Figure 2, was a nickel ball. The plutonium oxide, having a bulk density of Le8 gfcc, was mixed with an equal weight of -40 to +100 mesh aluminum oxide, The alumina was to act as a thermal conductivity media and as a diluent, With this system, difficulties were experienced with rapid and excessive temperature increases from 5450 to 800°C due to the exothermic Pu0, to PuF, reaction. This temperature spike caused a sintering o% the Pul, to a density of 647 g/cc and a plugging of the bed, In some cases, severe corrosion of the nickel ball resulted, By adding a vibrator to the bed, some uniformity in the depletion rate was observed, but generally the results were poor; sintering or packing occurred, the total depletion and depletion rate were poor, and there was severe fluorine attack on the nickel ball. The distributor arrangement was then changed as shown in Figure 3, A layer of 20~mesh nickel spheres, which extended into the heated zone, was used as the distribution system. Glass column tests indicated that the alumina was of 1little benefit, because the PuO, segregated from the alumina after short periods of bed operation, Therefore, alumina was eliminated in subsequent runs, Temperature excursions continued to unpredictably occur, In one case, the temperature increased from 550 to 1200°C in 30 seconds, causing the nickel distribution spheres to fuse and form a solid pluge By carefully controlling the temperature, the gas flow rate, the filter blowback cycle, and by vibrating the bed, the temperature excursions could be minimized, Depletion rates of 70 g/nr were achieved; however, a small static layer of Pu0, was observed at the nickel interface, Further studies were made on various gas distribution and bed designs using a glass colum, It was found that a cone with a 2=-degree taper provided complete agitation of all of the bed material and was much less sensitive to particle size distribu=- tion., It was further observed that the velocity of the gas at the bed inlet was sufficiently high to prevent the material in the fluid bed from dropping into the inlet line. Using this information, the fluid bed was modified as shown in Figure 4, This is the current design of the bed and 12 runs have been made with this system, The bed has functioned well using the 2= degree taper. The runs have proceeded to depletion without temperature excursions or sintering using Pu0, as feed material, A vibrator was installed on the bed and a depletion rate of 160 g/hr was achieved at a fluorine velocity of 2.3 ft/second. Increasing the velocity resulted in a decrease in depletion rate, accompanied by an inability to reduce the filter pressure drop by backblowing. An increase in depletion rate with increased 168 691 FULLY FLUIDIZED REGION STATIC 3 E E 4 p—/ PUOZ v %Z%%Z%%Z%% UNREACTED PuO, & PuFg4 20 MESH / NICKEL SPHERES /4 - Figure 2. Fluid Bed Distributor (Ball Figure 3, Fluid Bed Distributor (20- and Tapered Seat) Mesh Nickel Spheres) L Figure 4, Two-Degree Tapered Cone gas velocity above 1 ft/sec was not attained without external vibration of the bed, Table I shows the results of the fluid bed runs using the three bed designs. TABLE I SUMMARY OF FLUID BED RUNS Distributor Max, Depletion Avg Depletion Avg % Design Rate Rate Depletion Ball and tapered 56 L9 39 seat Nickel spheres 93 gl 56 2-degree tapered 160 93 98 cone Present studies are concerned with the design and operation of a continuous feeding system for the fluid bed, Purification Analytical data of impurities inthe feed and product indicates that good purification factors are achieved for common impurities found in the Rocky Flats process stream, In one run, 1000 ppm each of 13 common impurities were added to the feed material, Table II shows the purification achieved on this run in the 2~degree tapered bed, In early runs, prior to passivation of the equipment, same copper and nickel contamination was picked up from the Monel and nickel surfaces, In later runs this has not been a problem, TABLE II PLUTONTUM PURTFICAT ION Element Feed Analysis Product Analysis Purification jojo bl jo3e o1} Factor Al 1000 226 > 38 Cr 1000 96 10 Cu 1000 <3 338 Fe 1000 31 32 Ga 1000 12 83 Mo 1000 50 20 Ni 1000 10 100 171 Sn 1000 3 333 Ta 1000 <100 >10 il 1000 245 40 W 1000 <50 >20 Zn 1000 40 >100 Irapping Two types of cold traps were tested for their efficiency in collecting the PuFge. One type has transverse cooling fins and the second has longitudinal cooling fins, Cooling was by refrigeration or liquid nitrogen. Both cold trap configurations trapped over 1 kg of PuFy at an efficiency of 99.5% or better, Because liquid nitrogen cooling reduces the losses through the traps to virtually nothing, the best combination for trapping on a production scale presently appears to be a trap with longitudinal fins and liquid nitrogen for cooling, PuFz Reduction Different techniques have been investigated for the reduction of PuFg to PuF), or directly to metal, Table IIT is a sumary of the methods studied and an indication of the results. To date the most successful method has been the thermal-~hydrogen reduction to Pth. Figure 5 shows the latest design for this type of reduction. The PuFg is swept from the cold traps with nitrogen or argon, preheated to about 300°C, and reacted with hydrogen in the reductor, With extended runs using this concept, essentially complete decomposition of the PuF; was achieved. Some caking on the filters results; therefore blowback provisions are provided, Several batches of PuF, prepared by the fluorination of PuO, and subsequent reduction to the tetrafluoride have been reduced to the metal by calcium reduction. In all cases the reductions proceeded smoothly and efficiently. TABLE ITI PuF¢ REDUCTTON Method Product Results Ho~thermal Pth Good Thermal on Pth bed Pth Good with proper control H,~F, flame Pth Difficult to control Ca=thermal Pu Fine particles 172 /e | fqy- = = HEATERS Ho, N 1 = — D.—\ NN J1 N2 BLOWBACK ] Il E = PUFG N N2 _.-E_J \_ j‘\ [ OFF-GAS — — OUTLETS j.—/ L L 1 | 1 | 1 N2 TRANSFER —=—— Figure 5, PuFg Reductor 173 If PuF, could be reduced directly to metal, one process step could be eliminated. Small-scale studies were made on the calcium reduction of PuFg. The studies indicated that the reduction could be made at 25°C or lower; however, the resulting plutonium was in the form of fine powder rather than a single button. This would cause potential problems with further handling and fabrication of the plutonium because of the pyrophoric nature of the fine metal powder-~additional laboratory studies could probably resolve this problem, Material Balance Table IV shows the material balance for five typical runs, The high PuF, decomposition in the cold trap in the first runs reflect the largé quantity of PuFg collected and the period of time (up to two weeks) held in the trap, The results of the second through fifth run reflect the high efficiency of plutonium removal by the cold trap and the completeness of decomposition in the reductor. Regidue Processing Studies have started on the recovery of plutonium from residues by the volatility process, Initial studies on plutonium containing electrorefining salts (KC1-NaCl-MgCl,) indicate that a double fluoride salt containing plutonium is formed rather than the volatile PuFg. If this cannot be overcome, then plutonium in residue salts from molten salt processes cannot be recovered directly by fluoride wvolatility. Other studies are underway on calcium fluoride slag formed during the reduction of Pth with calcium, This slag contains some plutonium which must be recovered, Studies to date indicate that the plutonium can be.volatilized and removed from the slagj however, an unpredictable and significant temperature rise often cccurs during the fluorination., This is thought to be due to the highly exothermic reaction of fluorine with calcium or impurities contained in the slag, This reaction can be controlled by a preoxidation of the slag and by insuring that no impurities are introduced into the slag prior to fluorination, SUMMARY AND CONCLUSIONS It has been demonstrated on a 1l-kg scale pilot plant that the volatile fluoride process is feasible for the recovery and purifi- cation of plutonium from impure oxide. Based on the pilot plant studies, the process appears economically favorable when compared to the present aqueous process, Studies are continuing on 174 CYA Wi B W g g g g g g ; g Net Pu Pu Prod Pu in Pu in Pu in Pu, Unac~ Run Cum, Unac- Charge Bed Res Cold Traps NaF Trap counted For Balance counted For 830 497 235 L6 36 +16 9842 +16 1366 1005 208 130 5 +18 98.8 +34, 630 343 270 10 2 +5 993 +39 270 255 20 7 2 =14 105,2 +25 530 518 10 5 1 -l 100.8 +21 optimizing the process and on further developing it so that plutonium can be recovered from various residues, Future plans call for the construction of a 12~kg production prototype system to prove the process on a production scale, Tt is hoped to have this system operational by March, 1970. REFERENCES 1., Static Reactor Reduction of Plutonium Hexafluoride with Todine and Hydrogen, Jo D, Navratil, R, O, Wing and J. D, Moseley, RFP-993, Sept. 6, 1968, 2. Impurity Removal by Fluorination of Plutonium Dioxide Between 28 and 300°C, Je D, Navratil and R, O, Wing, RFP=1052, Septe. 5, 1968, 3e Laboratory Investigation in Support of Fluid Bed Fluoride Volatility Processes, M, J. Steindler, ANI~6753, L4e Engineering Development of Fluid Bed Fluoride Volatility Processes, C, J, Vogel, E, L. Carl, W, J, Mecham 5. Argonne National Laboratory, Chemical Engineering Division Summary Reports, 1958 to 1963. 6. Oak Ridge National Laboratory, Chemical Engineering Division Annual Progress Report for Period Ending June 30, 1962 (and each year through 1967). 176 LABORATORY-DEVELOPMENT OF THE FLUQRIDE VOLATILITY * PROCESS FOR OXIDIC NUCLEAR FUELS M. J. Steindler, L. J. Anastasia, L. E. Trevorrow, and A, A, Chilenskas Chemical Engineering Division, Argonne National Laboratory, Argonne, Illinois U. S. A. ABSTRACT Selected aspects of laboratory studies on the fluoride volatility process for spent reactor fuels are reviewed., Results obtained in studies of the fluorination of simulated oxidic fuel in a 2-in. diameter fluidized bed are outlined with emphasis on the behavior of uranium and plutonium. The described ex- periments include the use of BrF_ and fluorine as well as fluorine alone as the fluorinating agent go convert product oxides to volatile fluorides. Results are summarized on the behavior of fission product ruthenium in separations steps, on the fluorination of neptunium compounds to NpF, and the reaction of NpF, with bromine and NaF, and on process studies with irradiated fuel. % Work performed under the auspices of the U,S. Atomic Energy Commission, 177 INTRODUCTION A discussion of the fluoride volatility process must be oriented to the chemical and radiochemical properties of the fuel to be processed. This review includes selected aspects of a lab- oratory-scale program in which the processing of both LWR and LMFBR fuels was studied. Figure 1 shows the fission product content of an oxidic LWR fuel irradiated to 10,000 MWd/T and cooled 30 days. This fuel contains primarily UO,, slightly enriched in 235U, and 10 kg fission pro- ducts and "4"kg plutonium per ton of fuel. A little less than 40% of the 4 megacuries of activity per ton of fuel is emitted by those fission products having volatile fluorides. Figure 2 shows similar information for a LMFBR fuel containing axial blanket and core fuel sections as they are likely to be combined in large LMFBR reactors. The data were calculated assuming & core burnup of 100,000 MWd/T and a 30~-day cooling period. The blanket fuel contains depleted uranium oxide and 2% plutonium oxide, while the core fuel, which represents the major fraction of the total, contains ~20%Z plutonium oxide together with uranium oxide and fission products. The fraction of the total curies represented by fission products whose fluorides are volatile is not very different from that shown in Figure 1 but the total curies per ton is significantly greater in LMFBR fuel than in LWR fuel. The LWR and IMFBR oxide fuels to be processed, therefore, con- tain plutonium at concentrations of 0.5 to 20%, fission products at concentrations between 1 and 10%, and depleted uranium as the balance of the metallic fuel constituents, Although cladding is not an important concern of the present discussion, LWR fuels are generally clad in Zircaloy and LMFBR fuels are clad in a 300 series stainless steel. The processing steps of the fluoride volatility process include a head-end operation, which removes the massive hardware from the ends of a subassembly, a2 step that separates the oxide fuel from the cladding, one or more steps that convert the oxides to fluo- rides, and several process steps in which the actinide fluorides are separated from each other and purified from fission product contamination. This paper will include descriptions of laboratory studies on the conversion of oxide to fluoride carried out in a fluidized bed, a description of selected results on the chemistry of ruthenium and neptunium, and a summary of small-scale experi- ments on irradiated oxidic fuels, 178 MASS TOTAL 10,1009/T Sr, Br, Rb, Co, Ag | g4, RARE EARTHS |Zr, Sn |[ Mo, Te | Tc, Kr, Xe, Ru, Rh, Pd Ba, l A VooV ! | / | \ \ \\\\ \ | \\| | Y, In Nb, Sb ¢ CURIES 4X10 Gi/T pe NON-VOLATILE lr VOLATILE FLUORIDES. FLUORIDES 1. Fission Product Content of a Thermal Power Reactor Fuel (BWR) Irradiated to 10,000 MWd/T and Cooled 30 Days. MASS 70,000 ¢/T Ag [Cd T Rh Rb |[Sr|| Rare Earths |Zr,Sn|| Mo,Te Ic Ru | 54 | Krixe Cs Ba ; v/ \ RN N\ AN ' I / /:’ l VNl N NN ' | \ Voo N y \ \ NN N AN \ \ NN NN l / ) \ N\ v D | Y Al Nb,Sb CURIES 5.5 x 10T Ci/T NONVOLATILE VOLATILE - FLUORIDES - FLUORIDES 2, Fission Product Content of IMFBR Core and Axial Blanket Fuel; Core Burnup 100,000 MWd/T with 30 Day Cooling. 179 FLUORINATION STUDIES IN A 2-IN., DIA. REACTOR A fluoride volatility process with BrF. and fluorine as fluorinating agents was studied in a 2-in’ diameter fluidized- bed reactor with sintered alumina as the inert bed material and a feed of UO,-Pu0, pellets containing nonradiocactive oxides of elements formed in fission. The pellets, together with oxides added separately to the alumina bed, simulated the processing of spent low-enrichment UO, fuel at burnups of 10,000 and 30,000 MWd/T. The processing of simulated fast reactor fuel (100,000 MWd/T) was also studied, using a feed of U0,-20 wt 7% PuO, powder (=325 mesh) and nonradioactive fission product oxides. "The objective of the studies was the determination of operating con- ditions to minimize residual uranium and plutonium in the fluidized bed, which is discarded as waste from the process. A typical experiment consisted of oxidation of the pellets or powder at 450°C with 20 vol Z 0,, fluorination of most of the uranium with dilute (10 to 20 vol %) BrF. or fluorine at 200 to 400°C, and fluorination of the plutonium”with concentrated recycled fluorine (90 vol %) at temperatures ranging from 300 to 550°C. A typical charge to the reactor consisted of 1100 g of sintered alumina (40 to 170 mesh) as the fluidized bed and 650 g of pellets containing, in the case of the LWR fuel, 560 g uranium and 2.8 to 3.2 g plutonium and , for the fast reactor fuel, 460 g uranium and 115 g plutonium. The major components of the experimental apparatus are the 2-in, dia fluidized-bed reactor, a remote-head diaphragm pump for gas recycle, soda lime traps for disposal of excess chemical reagents and UF, produced during the BrF_ step, activated-alumina traps for the disposal of excess fluorine reagent, and sodium fluoride traps for the collection of PuF produced during the recycle-fluorination step. Detailed desériptions of these components and of the experimental procedures have been reported elsewhere,(1,2) Fluorination of Low-Enrichment UO, Fuels with BrF. and Fluorine To verify that selective fluorination of uranium from plutonium had occurred during the BrF_ step at 400°C, samples from the soda lime traps used for disposai of the UF,, unreacted BrF_., and bromine reaction product were analyzed for plutonium, “These samples contained 4 x 10~ to 9 x 1053 wt % Pu which, from ex- perience with the equipment used in sampling and grinding the soda lime in preparation for analysis, is known to be within the background level of cross-contamination. These results indicate that essentially none of the plutonium is converted to volatile PuF, during fluorination with BrF .3 an average of approximately 0.7% of the uranium charged remained in the fluidized bed after fluorination with BrFS. O0f this amount, 96% was fluorinated during 180 the subsequent recycle-~fluorination with fluorine so that only about 0.025% (0.14 g) of the original uranium remained in the final alumina bed. (3) The operating conditions used during the BrF. step (when most of the uranium is fluorinated) and during the récycle-fluorination step (when plutonium is fluorinated) appear to affect the fluorination of plutonium from the fluidized bed. For example, the data shown in Fig. 3 depict the plutonium concentrations in the fluidized bed during recycle fluorination following fluorination of uranium with BrF_ at 200, 300, and 400°C. From the relative rates of plutonium remdoval and the final plutonium concentrations in the fluidized bed, it appears that the temperature at which uranium is fluorinated with BrF_ affects the subsequent plutonium fluorination and that the uranium Should be fluorinated at temperatures below 400°C, Additional experiments in which the BrF. step was carried out at 250 and 350°C showed that plutonium fluorination was similar to the curve shown after a 300°C BrF_. step; consequently, the fluorination of uranium with BrF shoulg be carried out at temperatures from 250 to 350°C to minimize the effects of this fluorination step on the subsequent fluorination of plutonium. Hence, most experiments were completed with the BrF5 step at 300°C. The procedure used in the fluorination of plutonium also affects the final concentration of plutonium in the fluidized bed as shown in Fig. 4, These data were obtained from experiments in which uranium was previously fluorinated with BrF_ at 300°C. Temperature Sequence A (Figure 4) represents a fluorénation scheme of 1 hr at 450°C followed by 3 hr at 500°C and 8 hr at 550°C. Sequence B is similar except that in this experiment the initial period of 3 hr at 300°C was followed by a gradual increase in temperature from 300 to 550°C for 5 hr and then an additional 2 hr at 550°C. The reaction of plutonium appears to be significantly different in these two schemes, but the mechanisms responsible for this difference are not clear.t3s4) However, low residual plutonium concentrations in the bed can be achieved by performing the BrF_. step at 250 to 350°C and by starting the subsequent fluoriné step at about 300°C and increasing the temperature gradually to 550°C.(3,4) For a typical experiment with a BrF, step at 300°C and a plutonium fluorination step as shown in Figure 3, about 3% of the plutonium charged remains in the fluidized bed. The fraction of the plutonium charge remaining in the bed can be reduced if several batches of pellets could be processed with a single alumina bed without an increase in plutonium concentration. To demonstrate the reuse of an alumina bed, two sets of experiments were completed, each set using a single bed of alumina to process three batches of fuel pellets at simulated burnups of 10,000 MWd/T for one set and 30,000 MWd/T for the other.(4) These experiments 181 Pu CONCENTRATION IN FLUIDIZED BED, wt% ———300°C ——‘-—300 TO 550 °C—-"-—550°C—-{ 0.3 ® 200°C BrFs 0.2 m 300°C BrFg A 400°C BrfFs 0.l 008 005 003 002 0.01 L o [ -» 0 008} — [ ] — [ 0005 1 l ! | | I [ | | | | 0 2 4 6 8 |0 12 RECYCLE-FLUORINATION TIME, hr 3. Fluorination of Plutonium with Fluorine After Reaction of Uranium with BrF5 at Several Temperatures. 182 PLUTONIUM CONCENTRATION IN FLUIDIZED BED, wt % (el e] 0.001 URANIUM FLUORINATED WIiTH BrFy AT 300°C | =500 +k—— 550 — 450 L Temperature Sequence A, °C _| | e 300-+k— 300 10-+{550 }=— 550 - Temperature Sequence B, °C lllllll TIME, hr Effect of Temperature Sequence on with Fluorine. 183 TIME, hr the Flucrination of Plutonium were carried out by repeating part of the reaction cycle, i.e., the oxidation and BrF, steps, allowing the plutonium to accumulate, and then fluorinating the accumulated plutonium in a single fluorination step with fluorine., The progress of reaction during the oxidation and fluorination steps for the low-burnup fuel is shown in Figure 5. The fluoride content of the final bed was 3.5 wt %, which corresponded to fluorination of 2,7% of the original alumina bed., Similarly, the final bed contained 0.003 wt Z U and 0.009 wt % Pu, corresponding to ~0,01% and 1,3% of the total charge of uranfum and plutonium, respectively. Figure 6 shows changes in the fluidized-bed composition as the accumulated plutonium was fluorinated for the higher burnup fuel. The fluoride cvontent of the final bed was 10.4 wt % indicating that 8.8% of the original alumina had been fluorinated to A1lF,. The final bed also contained 0.012 wt % U and 0.009 wt Z Pu, corfesponding to <0.01% of the total uranium and 0,752 of the total plutonium charged to the reactor. Fluorination of Low-Enrichment UQ, Fuels with Only Fluorine In pilot-plant studies(5!6) with U0, pellets, a two-zone method of simultaneously oxidiziug and fluorinating a batch of fuel pellets in a fluidized-bed reactor was demonstrated under controllea temperature conditions to yield high UF, production rates, and high fluorine utilization. When this processing scheme was applied to pellets containing plutonium, approximately 80 to 100% of the uranium and 50 to 70% of the glutonium charged in the pellets was reacted in 3 hr of operation. 1,2) However, cesium, added to the bed as finely powdered CsF, appeared to influence plutonium fluorination and to increase the residual plutonium concentration in the bed. The effects of the added cesium were largely overcome (1,2) with a stepwise oxidation and fluorination procedure. In particular, fluorination of uranium at 350°C, followed by recycle~fluorination in the temperature range 350° to 550°C, was effective in reducing residual plutonium in the alumina bed to 0.009 wt %, corresponding to the retention of 3,5% of the plutonium charge. As shown in Figure 7, the relative plutonium retained by the alumina bed was reduced further by reusing a single bed to process three batches of fuel pellets with this reaction sequence. The uranium and plutonium concentration in the final alumina bed were both 0.009 wt Z, corresponding to the retention of “0.01% of the uranium and 1.2Z of the plutonium in the pellet charge. The alumina bed was not significantly altered mechanically or chemically in the sets of alumina-reuse experiments reported here, and it is concluded that a single batch of alumina can be used to process three batches of fuel pellets at simulated burnups of 10,000 and 30,000 MWA/T to effectively reduce plutonium losses for a fluidized-bed fluoride volatility process. 184 le— BATCH-I —eh—BATCH-2 —_fiy_ BATCH-3—«fe— RECYCLE - FLUORINATION —— | 20 " . r—— arer - OXIDATION; |BrFis; OXIDATION’,‘IBrFs, OXIDATION; |BrFs; 300° gc;% 02‘_1 _— 10}= N \ = /\ - I\ fl L I\ /) N /I \ \ * “ Fluoride z s 'E 5 FE / / \ a | / Y w - N 3 L 2 3 = [ z \ x 3 o'|E \\anium = — ~ S R~ o EF ~’“’\}‘ u —_ 2 L N (8] \Xl 0.0I= | - o : \ — \ = x\ = N, x X ooosl_| 1 | | 0 ¢ 0 [ 4 8 12 16 20 24 28 32 36 5. PROCESSING TIME, hr Uranium, Plutonium, and Fluoride Content of a Fluidized Bed During Alumina Reuse Experiments with Simulated 10,000 MWd/T Spent Fuel. 185 200 300°c —+— 300T0550°c——-|-—550°c—- 00— 50— CONCENTRATION N FLUIDIZED BED, wt % e Ae— - - — d“. — g—————— === ——=@="" FLUORIDE PLUTONIUM 6. Progress of Reactions During Fluorination of Plutonium Accumulated from Three Batches of Simulated 30,000 MWd/T Spent Fuel, TIME, hr 186 Pu CONCENTRATION IN ALUMINA BED, wt % S=15 e RECYCLE-FLUORINATION 20 v/0 02 [v/oF2 450°C (350°c| 350°C ‘—-—350—550°c ~—550°C—= }. O e u L 0.1 s L O BATCH | I A BATCH 2 ®m BATCH 3 Cl()l::— | | | | 0 4 8 |12 16 20 CUMULATIVE PROCESS TIME, hr 7. Plutonium Concentration in the Fluidized Alumina Bed During Alumina Reuse Experiments. 187 Fluorination of Fast Reactor Fuel with Fluorine Fast breeder fuels were simulated using a solid solutiomn U02- Pu0, powder (mean particle size ~5 um) containing 17.8 wt % plu%onium and 70.3 wt Z uranium, This powder, together with a mixture of nonradicactive oxides of the elements formed in fission, was proportioned to simulate FBR fuel with a burnup of 100,000 MWd/T. The initial scoping experiments consisted of a fractional factorial series of five experiments to determine the effects on plutonium in the final fluidized bed for (1) two levels of fuel- to~alumina ratio (0.3 and 0.6), (2) fluorination temperature (350 and 450°C) with 10 vol %Z F, for uranium fluorination, and (3) recycle-fluorination time (EO and 20 hr)} for plutonium fluorination. The 20-hr recycle-fluorination sequence consisted of 4 hr at 300°C, 3 hr each at 350, 400, 450, and 500°C, and 4 hr at 500°C while the 10-hr sequence incorporated 3 hr at 450°C, 3 hr at 500°C, and 4 hr at 550°C, This latter sequence was chosen in an effort to obtain high fluorination rates for plutonium. For the five experiments, the fraction of charged plutonium in the final bed ranged from 0.51 to 3.19% with only one ex- periment significantlyabove 1% of the plutonium charge in the final bed. The conclusions drawn from the statistical analysis are that an increase in the fuel/zlumina ratio from 0.3 to 0.6 and an increase in the temperature of fluorination with dilute fluorine from 350 to 450°C will increase the fraction of plutonium charge remaining in the fluidized bed; conversely, increasing the recycle-fluorination time from the 10-hr to the 20-hr sequence will reduce the fraction of plutonium charge remaining in the bed. The highest rates of plutonium fluorination were obtained during the initial part of recycle-fluorination sequence except in one experiment when low rates during this period apparently resulted from an initial recycle fluorination temperature of 300°C which followed fluorination with dilute fluorine at 450°C. For the remaining experiments, the production rate for PuF averaged 1.2 to 3.0 1b/(hr) (ft2) while the reactor 0perateg at 17.2 to 55.4% of equilibrium for the reaction PuF, (s) + F2 (g) 2 PuF,(g). As had been found for the fluorination of uranium, 8) the diminishing-sphere reaction model was tested and appeared to correlate the data for plutonium fluorination during the jinitial recycle~fluorination period. Rate constants of 0.7 x 10~ min~! at 300°C and 4.4 x 10-3 min~1 at 450°C (i.e., average result for 3 experiments) correspond to an apparent activation energy of 10 kcal/mole. In summary, fluorination of simulated LWR fuel using BrF followed by elemental fluorine or using dilute fluorine foliowed by concentrated fluorime converted a large fraction of uranium and plutonium to their volatile hexafluorides, which volatilized 188 from the fluidized bed, The temperature of initial fluorination was shown to affect subsequent plutonium removal from the bed. Reuse of the alumina bed to process three fuel charges reduced the fractional loss of plutonium to the alumina to 1%, Corresponding experiments with simulated FBR fuel indicated that low plutonium losses can be obtained in a single use of the alumina bed. BEHAVIOR OF FISSION PRODUCTS AND NEPTUNIUM Recent experimental work to elucidate the behavior of fission products and neptunium has been concerned mainly with behavior during two process operations: the fluorination operation, in which actintdes are separated from nonvolatile fission product fluorides and some gsubsequent operations in which actinides are separated from volatile fission product fluorides. A fuel recovery process will include a separation of neptunium from uranium and plutonium regardless of whether the neptunium is treated as a by~ product to be recovered or as an impurity to be discarded. Behavior of Neptunium in Fluorination Processes The behavior of neptunium in the fluorination operation has been investigated both in small-scale boat-reactor experimenta(g) and in the 2-in. diameter fluid-bed experiments.(3,4) Reactions of NpF, with gaseous BrF3, BrF., and F, in the boat reactor were carrieé out at 300-400°C. The neptunium product of the reaction was NpF_ with each fluorinating agent; bromine was also produced in fluorinations with BrF5 and BrF3. The fluorination of NpO, by either fluorine or BrF_ was found to proceed through tfie formation of the intermeaiate NpF (identified by X-ray diffraction analysis). The mechanism of fluorination of NpO,, therefore, parallels that of Pu0O, (which proceeds through thé intermediate PuF,), but differs from the fluorination of UOZ’ which proceeds tfirough the intermediate U02F » Apparent rate constants were obtained for both the boat reac%or and fluid-bed experiments from correlation of the data by the rate law that assumes reaction occurs at a continuously diminishing spherical interface. The apparent rate constants obtained in the boat reactor experiments were correlated by the Arrhenius equation to yield an activation energy of 20 kcal/mol for the reaction of NpF, with elemental fluorine. An independent study(lo) of the rate o? reaction of NpF4 with fluorine yielded the same activation energy. 4 The fluorination of NpO, from mixtures with simulated oxide fuel pellets was demonstrated with both BrF_ and fluorine in the 2-in. diameter fluidized-bed reactor. %he NpF, product was distributed about equally between the UF6 produced during 189 radd€fon with BrYF_ and the PuF, produced in the subsequent reaction with fludrine, The amount of neptunium added to the simulated fuel was a factor of 40 more than would be expected in 650 g of apent fuel irradiated to 10,000 MWd/T. Excess neptunium was provided so that its concentration in the bed samples would be abouve the limit of analytical detection, which was affected by the plutonium concentration. After 2 hr of fluorination with BrF (10.¥0l1 %), an average of 46% of the original neptunium remained in the fludidizrd-bed. This- result- corresponded to a reaction .rate constant for the diminishing-sphere model of 2 x 10-3 min‘l, a factor of 10 bigher than the rate constant obtained in the boat reactor tests with 33-35 vol % BrP_. The -difference-4n rate constants may be attributed to a more highly reactive NpF, being formed by reaction 6T NpO, and BrF. in the fluidized bed wince the NpF, used in the boat reactor expetiments was obtained by reaction of fipo2 with HF and oxygen at 500°C. The apparent rate constant is inversely proportional to both the particle size and the bulk density of the initial reactant, Higher specific surface areas and lower bulk densities may have been obtained in the fluidized- bed experiments. Other factors promoting higher apparent reaction rates in the fluifdized bed are better gas-solids contacting and the influence of elutriation. After the subsequent fluorinacion with fluétrine, neptunium concentration in the fluidized bed was below the limit of analytical detection, indicating that less than 7% of the original neptunium remained in the final alumina bed. It is important to note that in a fluidized-bed fluoride volatility process which utilizes both BrF. and fluorine, volatile NpF, is formed in both fluorination steps, ahd about one-half of the original neptunium will accompany each of the UF, and the PuF, products. The removal of neptunium from the fluidized bed by fiuorination is represented graphically in Figure 8 for fluorination of the fuel by BrF_ and by fluorine and is illustrated in Figure 6 for the recycle éluorination of plutonium with fluorine. Reaction of NpF, with Brz(ll) In the experiments described above, the reaction of NpF with either BrF_. or BrF, in a gas~flow enviromment with continuous removal of products from the reaction site resulted in a net production of NpF,, A reaction was observed to occur, however, in the traps in which the products of the boat reactor fluorination of NpF, with either BrF. or BrF, had been condensed. These reactions Btarteé at temperatures in the fange ~78 to +30°C, indicating that NpF, must be readily reduced by bromine. Additional experiments in which bromine was condensed onto NpF, at -78°C, then warmed to 430°C confirmed these indications. For the gas-phase reaction of NpF, with bromine at +80°C, the measured pressure change was con- Bisgant with the stoichiometry of the equation: 190 RECYCLE - FLUORINATION BrFs OXIDATION| 300 300 TO 450°C °C 300 °C 550 °C 550°C 30 A\ X 10— ‘\ p— [ | 5 — \ ] | | . | 32 — | ] b | o Lo | p— w — — @ — | —] s = | - N \ — = — \ —] = x —_— 2 < z 0= = — — Uronlum\ Plutonium ] « _ ] @ = T — z L — Wl S . - o O O.OI__L_: — [ — | — A 000! | N A | 0 4 8 12 1) PROCESS TIME, hr 8. Fluorination of Actinide Elements in a Fluidized Bed Using BrF5 and F2 as Reagents. 191 3INpF, (g) + nr (g) 2 3NpF4(s) + ZBrFB(g) The product BrF3 was idengified by its infrared absorptidn spectrum. Since UF, is not reduced by bromine, the reaction with bromine offers a means of separating UF, from NpF,. This separation process was demonstrated experimentally. Stoichiometric-excesses of bromine were condensed onto 1l:1 mixtures of UF, and NpF_ at -78°C After a reaction period at a higher temperature, the volatile materials were distilled away from the reaction.-mixture, and both the volatile and nonvolatile fractions were analyzed for neptunium and uranium, Table I shows that at 30°C the reaction did not go to completion even after several hours, >ut that after reaction at 75°C for one hour, 99.9% of the neptunium found by analysis was in the nonvolatile fractiom. (12) Reaction of NpF, with NaF VL The removal of volatile metallic fluorides from gas streams by reaction with solid NaF to form solid complex compounds has been frequently included in volatility process schemes. The reaction of solid NaF with gaseous NpF6 was studied to obtain Information on the stoichiometry of the reaction and the identity and stability of the product in order to assess the feasibility of using solid NaF to remove NpF6 from gaseous mixtures. Table T Separation of UF -NpF, Mixtures?® by Reaction with Br, s — Reaction Reaction Neptunium in Nonvolatile Fraction Temp. Time (% of Total Np Found (°C) (min) by Analysis) 30 30 98.6 30 30 50.3 30 40 35.5 30 960 87 30 2400 96 25b 60 53 75 60 99.9 75 &0 99.9 75 60 99.9 aMixtures contained ~1L0 to 200 m~ of each hexafluoride. bLiquid bromine in contact with aexafluoride mixture. 192 It was observed that, at temperatures >150°C, powdered NaF reacts readily with gaseous NpF, with the formation of a violet solid product. The equation 3NaF(s) + NpF_(g) Z 3NaF:NpF_(s) + 1/2 F,(g) was found to represent the equilibrium involved in the react%on of NpF, with NaF at 250 to 400°C, The partial pressures of flvorine and NpF, in equilibrium with the solid phase formed by reaction of NpF, with NaF were obtained by measurements of total pressure and ultraviolet absorbance of the gas phase. At a fixed temperature (350°C) over a 16-fold variation of fluerime pressure, the value of log (PNpF_) was found to depend linearly on log (fiF ) with a proportionalgty coefficient of 0.49, comparable to the valie of 1/2 expected iygm'thE'equattun. Equiitbrium constants, K ~ (PNpF )/ (PF. ) , for the reaction at 250 to 400°C, are expressed by the aqiation: log Kp(atm]'/z) - -3.147 x 10°/T(°K) + 2.784 The equilibrium constants given by the above relation were used to caluclate the extent of removal of NpF, from mixtures with UF6 in a proposed operation at the followgng conditions: (1) Gaseous UF_ -NpF, mixtures, with no initial fluorine, would be passéd through a ped of WaF pellets with the intention of maximum fixation of neptunium and minimun fixation of uranium by the solid phase. (2) The residence time would be sufficient to allow the reaction of N‘pF6 with NaF to reach equilibrium, (3) The partial pressure of UF, in the mixture would be less than the dissociation pressure of either of the complexes formed between NaF and UF6 (NaF-UF6 and 2NaF-UF6). The stoichiometry of the equation describing the equilibrium leads to the expressions: .2 61 + 61 =1/2 7 PNoF 2K PFg P prFfir and % NpF6 removed from gas = 100(1 - ) NpFGi 6’ = pressure of NpF6 remaining at equilibrium, where prFe = initial pressure ot NpF i p NpF6r 193 Kp = equilibrium constant. The initial NpF, pressures were fixed by chosen concentrations of NpF, in UF, and the dissociation pressures of 2NaF:UF, calculated from the data of Katz, (13) indicating the pressure below which the UF6 must be maintained to avoid formation of 2NaF-UF6. Figure 9 presents the results of calculations of the maximum percentage of NpF, that would be removed by one equilibrium stage for concentrations of NpF, in UF, warying from 10 to 1C00 ppm at temperatures of 250, 300, 350, and 409°C. Figure 10 presents similar information for two equilibrium stages (equilibration of gas mixture with NaF, removal of fluorine formed in the reaction of NpF, with NaF, and equilibration of the residual gas mixture with a second NaF trap). Two-stage operation would be realized, for example, if the UF, stream passed in sequence through a NaF trap, a distillation column, and then a second NaF trap. In general, the fraction of NpF, removed increases as the initial concentration of NpF, increases; the percentage of NpF, removed also increases as the temperature of the NaF bed decreases. Since the probable initial concentration of NpF, in the UF, process stream is low (100 ppm), the extent of rémoval in a single equilibrium stage is not high. Calculations were performed at 409°C since the temperature of a NaF trap must be kept above this value to pass UF, vapor at 1.75 atm, a favored process pressure, without formation of 2NaF:UF,., At these conditions, the calculations shown in Figures 9 and 10 ingicate that about 46% of the neptunium would be removed by one equilibrium stage and that about 70% of the neptunium would be removed by two equilibrium stages. Behavior of Volatile Fission Product Fluorides in the Fluorination Operation Experiments(4) determining fission product behavior during fluorination of simulated oxide fuel in the 2-in. diameter fluidized bed were limited to elements forming volatile fluorides. To aid in following ruthenium in the various processing steps, Ru was added to the alumina bed in an experiment in which the BrF5 step was carried out at 300°C. Additionally, a spark source mass spectrometric analysis was used to determine the distribution of ruthenium, molybdenum, and rhodium in experiments with the BrF step at 250 and 350°C. The results indicate that volatilization of most of the ruthenium takes place during fluorination with BrF_. at 300°C followed by only minor fluorination of ruthenium during the recycle~fluorination with fluorine at 300 to 550°C. These results are in ggreement with those obtained in boat reactor tests with 10°Ru tracer and in fluidized-bed tests with irradiated fuel. Table II gives the results of spark source mass spectrometric analyses of fluidized-bed and sodium fluoride samples for experiments with the BrF5 step at 250 and 350°C, 194 % OF INITIAL NpF, REMOVED 9. 00 ] I N | T T T T 7171 DESIGNATION | TEMP. Uf-'6 PRESSURE 30 °C {mm) A 250 18 20 8 300 89 — C 350 347 10 f— D 409 1330 —1 0 ; Lt [ Ll 10 100 1000 INITIAL CONC. NpF, IN UFs,p Calculated Maximum Percent of Initial NpF Mixture with UF6 by Reaction with NaF in 195 pm Removed From Gaseous 8ne Equilibrium Stage. 961 100 , % OF INITIAL NpF, REMOVED 10, | T 1 ¥ 1TT1 i | | | T 40 |- DESIGNATION| TEMP. [UF, PRESSURE| __| °C {mm) 30 |— A 250 18 _ B 300 89 20 — C 350 347 — ol 0 409 1330 N o AR L1 1111 10 100 Calculated Maximum Percent of Initial NpF Mixture with UF 6 INITIAL CONC. NpF IN UF,,ppm Removed from Gaseous by Reaction with NaF in @wo Equilibrium Stages. Table II Spark Source Mass Spectrometric Analyses of Alumina (4) Mo Ru Source of Sample (ppm) (ppm) I. BrF. Step at 250°C Alumina bed after oxidation 800 230 of pellets Alumina bed after fluorination 34 70 with BrF_ at 250°C Alumina bed after fluorination <5 39 with F, from 300 to 550°C Sodium f%uoride (PuF6 product 35 0.7 collector) II. BrF,_ Step at 350°C ATumina bed after oxidation 800 230 of pellets Alumina bed after fluorination 19 60 with BrF_ at 350°C Alumina beg after fluorination <5 19 with F, from 300 to 550°C Sodium f%uoride (PuF6 product 32 0.7 collector) Rh {ppm) 100 150 90 <0.3 100 300 200 <0.1 Analyses of replicate samples varied by a factor of three except for rhodium where the variations in the spark source analyses were not significant. indicate that 70 to 80% of the ruthenium and more than 95% of the molybdenum fluorinates during the BrF While the bulk of the molybdenum and ruthenium are fluorinated during the BrF the fluorine s step at 250 to 350°C, step, additional fluorination also occurs during gep. A similar distribution of molybdenum and ruthenium was observed with irradiated spent fuel pellets, also without significant variations in the spark source analyses for rhodium. Separation of Ruthenium from PuF 18 Recent experimental studies of the separation of volatile fission product fluorides from FuF, have centered on PuF_- ruthenium fluoride mixtures since past experience showed that this separation might prove difficult. ruthenium mixtures was tested in bench-scale experiments. The general procedure consisted of (a) fluorinating mixtures of 200- 400 mg of PuF 4 197 and 15-20 mg of ruthenium metal (containing 0bpy Within these limitations, the spark source data The behavior of plutonium- to permit radiochemical analysis), (b) transporting the resulting gaseous mixture by gas flow through a train of vessels, each simulating a process vessel and corresponding process operation, and (c¢) determining the final distribution of plutonium and ruthenium in the train. Early results caused a shift in the main point of attention to the operation in which ruthenium and plutonium fluorides were to be separated by preferential condensation at -10°C. At this temperature, the vapor pressure of pure PuF, is 7.89 Torr, which is greater than the partial pressure of P in the effluent gas stream of the fluorination operation. The vapor pressure of RuF_. at ~10°C is 2.6 x 10‘6Torr, suggesting that passage of the gases formed by fluorination of ruthenium- plutonium mixtures through a trap at -10°C should result in condensation of only Rqu. The results, presented in Table III, indicate that, although the amounts of ruthenium penetrating the traps at -10°C are small fractions of the total ruthenium charged to the reactor, they are orders of magnitude greater than the calculated amounts that should penetrate the trap on the basis that the solid phase in the traps is Rqu. The quantities of ruthenium penetrating the traps at -78°C are of special interest, since at this temperature, the vapor pressures of any ruthenium species are sufficiently low that only a small quantity of any species is required to form a solid phase in the trap. The comparison of observed amounts of ruthenium transpiring through the trap with theoretical amounts calculated from the vapor pressures of various ruthenium species could, therefore, identify the solid species in the trap. The agreement of theoretical and observed moles of PuF penetratins the ~78°C trap had shown that the trap is efficieng for PuF Five of the seven results in Table III show that the quantities of ruthenium transpiring at -78°C are the same order of magnitude as the quantities calculated for RuO,. If Ru0, is indeed formed, the question of the source of oxygen arises. &he formation of very small quantities of RuO, might be explained by the presence of traces of moisture in the system or by the presence of an oxide film on the ruthenium metal powder. The results of the entire set of experiments listed in Table III can be summarized as follows: (1) A small fraction of the total ruthenium charged to the system formed a compound more volatile than RuF_. in the reaction of fluorine at 500°C with ruthenium metal, Futhenium metal-alumina mixtures, or ruthenium metal-PuF, mixtures. (2) In five of seven experiments, the observed quantities of ruthenium transpiring at -78°C were the same order of magnitude as the quantities calculated for RuO,. (3) This set of experiments indicates no consistent difference between the volatility of ruthenium compounds produced by 198 661 Table III Comparison of Observed and Theoretical Transpiration of Ruthenium Compounds Moles Tranmspired Material Initial Moles at -10°C Moles Transpired at -78°C Fluorinated Ru Ru{obs) Rnggjth) Ru{obs) RuFS(th) Rqu(th) Rggé(th) RuOFL(th) Ru~Puf,, 17x107° 1.3x10™° 3.7x107°0 1.7x10°% 6.8x107'% 3.1x10™° 1.1x1077 4.9x1077 Ru-PuF, 15x107° 1.9x107° 3.2x1070 2.7x10°° 6.3x1071% 2.8x10™° o0.98x10"7 4.6x10~° Ru Only 14x107° 2.4x107% 3.7x1077 1.5x1077 6.8x1071% 3.1x107° 1.1x1077 4.9x107° Ru-Alumina 15x10° 3.8x1077 3.2x10™° 2.0x10”7 6.3x1071% 2.8x107° 0.98x10"7 4.6x10"° Ru-PuF, 17x107° 2.0x10°% 3.2x107° 4.8x1077 6.3x10°1° 2.8x107° 0.98x1077 4.6x10"° Ru Only 16x107° 9.4x1077 3.2x107° 2.9x1077 6.3x101% 2.8x107° 0.98x1077 4.6x10"° Ru-PuF 16x107° 5.8x10°7 3.2x107° 1.6x1077 6.3x10°1°% 2.8x10™° 0.98x1077 4.6x10"7 4 fluorination of ruthenium in the presence of PuF, and the volatility of ruthenium compounds produced by flucrination of ruthenium in the absence of PuFA. BENCH-SCALE STUDIES WITH IRRADIATED FUELS Eleven experiments using fluidized-bed fluoride volatility techniques to proeess irradiated metal alloy and oxide fuels were performed in a hot cell facility. The principal objectives of these experiments were: 1. to establish whether filssion products or radiation fields have a significant effect upon the recovery of the fissionable values in the fuel; 2. to determine actinide and fission element distribution from irradiated fuels and for the various process steps; 3. to determine product decontamination from the fission elements for established process steps and to test other schemes having process potential. Of the six experiments run with alloys, two were with uranium- Zircaloy irradiated to ~40% burnup of the 235y originally present and cooled 5 yr, and four experiments were with uranium-aluminum irradiated to 50% burnup and cooled either 3 or 7 months, TFive experiments were performed with uranium oxide fuel irradiated to a burnup of ~33,000 MWd/T and cooled 1 to 1 1/2 yr. Tests with Irradiated Uranium Alloy Fuels The process, which has been described in detail, (14,13,16) involves the use of a fluidized bed of alumina to provide gas contacting of the fuel., Hydrogen chloride gas separates the zirconium or aluminum matrix from the uranium by forming volatile chlorides with the zirconium or aluminum while converting the uranium to a nonvolatile chloride. A filter bed of packed alumina particles downstream of the fluid-bed reactor prevents uranium loss from the fluidized-bed. Elemental fluorine removes the uranium from the bed by converting it to gaseous UF,. Sodium fluoride beds and a cold trap operating at -78°C coilect and decontaminate the UF6' An equipment flowsheet is shown in Figure 11, The hydro- chlorination step is performed with valve V1 closed and valve V2 open. The gas mixture flows from the packed-bed filter to the condenser, through the disposable filter to the scrubber, and then to the atmosphere through the cave stack. The ZrCl (or AlCl3) generated in the reactor is removed from the gas stream 200 VENT COLD TRAP WA I SAMPLE SH SOLUTICN PARTICULATE FILTER SCRUBBER ACTIVATED ALUMINA F, DISPOSAL —=— (AS FLOW —e~w LIQUID FLOW (=) GLASS woOoL - FILTER ‘fl' H sow'non 'é. ' PACKED - BED FILTER = S o5 1 gc w - e % HF HCL N, REAGENT SUPPLY 11. Equipment Flowsheet for Studies on Irradiated Uranium Alloy Fuels. 201 by a condenser cooled by natural convection. Experience with the unirradiated fuels showed that a filter 1s required after the con- denser to remove chloride fines, which would otherwise cause plugging of the line to the scrubber. The filter is a small section of pipe, packed with glass wool, which can be removed and replaced remotely for each run, Excess HCl is scrubbed from the nitrogen gas stream by continuously recirculated H,0, NaOH, or KOH solution before discharge of the gas to the caveé exhaust duct. When the hydrochlorination step is complete, the high- temperature valves are reversed (V1 opened, V2 closed) and the fluorination step begun, During the fluorination step, UF, 1is generated in the fluidized~bed reactor, passes through traps NaF-1 and NaF-2 (maintained at 400°C to remove certain fission products), and is collected on trap NaF-3 (maintained at 100 to 150°C). A fourth NaF trap (at room temperature) serves as a backup for the NaF-3 trap. The gas flow continues to a fluorine absorption tower containing 1/4-in. diameter activated alumina spheres and then to the cave exhaust, During several experiments, trap NaF-3 was heated to 400°C and the UF, was desorbed in a stream of fluorine and then collected in cold traps cooled by dry ice. The UF, was removed from the cold traps by hydrolysis with 20 vol % nitric acid, Distribution of Fission Products Following Hydrochlorination Step The activities found to be essentially nonvolatile during the hydrochlorination step with the alloy fuels were cerium, cesium, and ruthenium, Trace amounts of cerium and cesium found in the condenser and scrubber following the runs with 5-yr-cooled uranium-Zircaloy are believed to be due to entrainment in the gas stream leaving the reactor and incomplete removal by the packed-bed filter. Cerium and cesium activities were not detected in the condenser or scrubber by gamma spectrometry following the runs with short-cooled uranium-aluminum probablg because of the masking effect of the large amounts of g5Zr and 29Nb present, The activities that were slightly volatile during hydro- chlorination were Sb, Mo, Tec, and Te, The amounts of these activities that volatilized and were collected downstream of the packed-bed filter ranged from about 0.3 to 10% of the charge activity. The activities that were predominantly volatile during hydro- chlorination were Zr, Nb, Kr, and I. Some partition of zirconium and niobium during hydrochlorination appears to occur from a con- sideration of the large amount of niobium collected during fluorination. An indication of the release of krypton during both process steps was obtained by the use of a beta monitor located in the cave stack. 202 This monitor had been calibrated previously by releasing known amounts of 85Kr to the cave atmosphere, The results of the measure- ments made during the six hot runs showed that essentially 100% of the available krypton is released to the stack during the hydrochlorination step. During the fluorination, the activity level in the stack returned to the normal background level. Two runs were made with 3-month—-cooled fuel to follow radlo- active 1311, 0f the iodine charged to the reactor, 66% was collected in the scrubber following the HC1l step o{Bthe first run and 93% in the second run. A trace amount of 11 was found in the activated alumina following thelgiuorination step. Stack-gas monitoring showed that <1% of the I was released to the stack during both the HCl step and also during the fluorination step. These results lead to the conclusion that the bulk of the iodine i1s released during hydrochlorination and can be removed efficiently by a caustic scrubber. A small amount remained in the reactor following hydrochlorination. Fission Product Distribution Following the Fluorination Step The work with the irradiated fuels showed that only trace quantities of cesium and cerium were found downstream of the packed-bed filter. The bulk of the strontium charged was re- covered with the reactor and filter beds. Ruthenium., Ruthenium (Table IV) appears to be only slightly volatile under the fluorination conditions. The amount volatilized for the first four runs ranged from 0.2 to 0.6% of that charged to the reactor. The small amount of ruthenium that did volatilize was almost entirely collected by the 400°C NaF pellet traps. Antimony. About 90% or more of the antimony remained in the reactor bed following hydrochlorination, Most of this residue was volatilized during the fluorination step and was almost completely retained by the 400°C NaF trap. A small residue (3 to 167%) remained in the reactor and filter beds following fluorination. Zirconium and Niobium. As noted, the bulk of the zirconium and niobium was removed before fluorination during the hydro- chlorination step, only 2 to 4% of the zirconium being downstream of the packed-bed filter. This amount of activity, which must pass through the packed-bed filter, raises a question as to whether the zirconium is present as a particulate solid or as a gas. If the vapor pressure of ZrF, in the gas leaving the reactor is ~47% of i1ts saturation value at 550°C, sufficient 957r would be volatilized to account for the zirconium activity collected by the 400°C NaF trap. It appears likely that the zirconium movement is due to volatilization; hence, its separation from volatile uranium hexa- 203 ¥0¢ Table IV Fission Products Volatilized and Collected Downstream of Packed-Bed Filter(l6) HC1 Step, Percent of Charge Fluorine Step, Percent of Charge Element Runs 1 and 2 Run 3 Run 4 Runs 5 and 6 Element Runs 1 and 2 Run 3 Run 4 Sb 2.8 5.7 9.7 c Sb 37.5 31 93 Ru ni1? ni1? ni1? ¢ Ru 0.45 0.23 0.6 Zr b 48.7 64 c Zr b 2,2 4.4 Nb b 44.8 61 c Nb b 4.8 30 Mo c 1.8 2.3 c Mo 43,5 c c Tc 2,7 c c c Tc 77 64.5 35 Te c 0.34 0.85 c Te c 50 57 I c c c 79 %Not detected by gamma spectrometric analysis. b Insufficient present in charge for analysis. cNot determined, fluoride depends upon efficient sorption rather than filtration. About 307% of the niobium volatilized during fluorination and was collected by the 400°C NaF trap., The reactor-bed analysis following fluorination showed that <1Z of the zirconium and niobium remained, These two results suggest that a partition between zirconium and niobium occurs during the HC1 step; i.e., a smaller amount of the niocbium forms a volatile chloride. The niobium that remains following chlorination readily forms a volatile fluoride and is almost entirely removed during fluorina- tion. Technetium, Technetium closely follows uranium in the processing steps. Only 3% of the technetium volatilized during the hydro- chlorination step. Following fluorination, the bulk of the technetium was collected with the UF, in the 100°C NaF traps. Little or no technetium was removed grom the gas stream by the 400°C NaF trap. Molybdenum. Less than 3% of the molybdenum was removed from the fluidized-bed reactor during hydrochlorination., The work with uranium~Zircaloy showed that the bulk of the molybdenum volatilized during fluorination and was distributed largely between the 100°C NaF trap and the 25°C NaF trap. Results of a run with uranium- Zircaloy showed that about 20% of the molybdenum was trapped with the uranium in the 100°C NaF trap. About 2% was collected with the uranium when an additional desorption step was employed following collection on 150°C NaF, About 12% was found in tfie reactor and filter beds, indicating that the removal of molybdenum is not complete under these processing conditions. Tellurium., Less than 1% of the tellurium volatilized during the hydrochlorination step. The bulk volatilized during fluo-— rination, passed through the 400°C NaF, 100/150°C NaF, and 25°C NaF traps, and collected on the activated-alumina trap. This trap was used principally to remove the excess fluorine from the process off-gas before its discharge to the cave stacks., Stack-gas monitoring showed that only trace amounts of tellurium were present, indicating that the alumina was highly efficient for the removal of tellurium. Only small amounts of tellurium collected with the uranium. About 0,14% of the tellurium was collected with the uranium when a 100°C NaF trap was employed; 0.0016% was retained with the uranium when a desorption step was employed. Neptunium. Complete neptunium analyses were not performed, but uranium-product analyses for the last two runs showed that ~21% of the neptunium charged was collected with the uranium, 205 Tests with Irradiated UQ, Fuels The equipment used for the experiments with irradiated uranium alloy fuels was modified by the addition of other traps and the inetalla%19? of a sintered nickel filter in the fluidized bed reactor. In each experiment, 100 g of irradiated UO, (previously declad) were added to a bed of refractory alumina in tfie fluidized-bed reactor. The UQ, pellets were converted to U OB-PuO2 fines at 450°C in the fluzdized bed by reaction with Za Vol Z oxygen in nitrogen. The sintered nickel filter prevented elutriation of uranium oxides and bed material from the reactor. The process off-gas passed through two traps containing activated charcoal to remove any volatile products prior to discharge to the cave exhaust. In the uranium volatilization step, the uranium was separated as UF, from plutonium and from most of the fission products by the uge of BrF, as a selective fluorinating agent. The process off-gas leaving the nickel filter contained UF6, Br,, BrF_, and volatile fission-product fluorides in N, dilueht. 7he off-gas was.passed through the uranium cleanup %rap where a selective sorbent, NaF at 400°C, was tested for decontamination capability. The gases continued through three additional traps in series, which removed the UF_., Br,, BrF_, and the remaining volatile fiesion-product fluorides. The first of these three traps con- tained activated alumina, which removed UF, and some fission~product fluorides and reacted with excess BrF. to give free bromine. The next trap, which contained soda lime at ~300°C, removed the bromine and some fission-product fluorides. The last trap contained activated alumina to remove moisture released by the soda lime. In the last step, plutonium was volatilized from the fluidized bed by fluorine and was collected as PuF, in the product cold trap. The off-gas from the fluidized beg reactor first passed through a precooler at 0°C where high-boiling fission-product fluorides were collected. A cold-trap backup containing NaF pellets at 350°C was used to remove trace quantities of PuF passing through the cold trap. The fluorine in the off-gas was removed by four traps in parallel containing activated alumina; parallel arrangement allowed limiting the flow of fluorine through each trap to 1 liter/min to avoid caking of trap contents. The off-gas cleanup trap following the fluorine- removal traps was filled with NaF pellets at “400°C to remove trace amounts of volatile fission products. For three experiments in which other plutonium decontamination schemes were examined, the PuF, collected in the cold trap was sublimed at 0°C and transporteg in a nitrogen stream to additiomal 206 traps where the PuF, was either adsorbed by NaF pellets at 350°C or thermally decomposed to PuF4 in beds of refractory alumina at 300°cC, The Distribution of Actinides and Fission Products The ranges of values for the volatile products collectdd during the five runs are shown in Table V, The results of the runs ghow the following: 1. The principal activities that volatilize during the oxidation step (for fuel cooled 1 year or more) are krypton and ruthenium; <27% of the total krypton and <3.3% of the ruthenium charged were volatilized., Small amounts of ruthenium and tellurium, well below the recommended limiting concentrations, were found #n the process off-gas discharged to the atmosphere. No attempt was made to trap krypton, and its discharge to the atmosphere was safely accomplished by dilution with 2000 cfm cell ventilation alr, 2. Up to ~13% of the total beta and gamma activity was volatilized with the uranium during the BrF. step. The principal gamma actlvity that accompanied the uranium was ruthenium., In addition, up to 76% of the molybdenum, 2.7% of the antimony, 0.24% of the zirconium, and 1.9% of the niobium were also vola- tilized with the uranium. Analyses to determine the amounts of tellurium, technetium, and neptunium were not completed. Table V Volatilegs Collected in Traps During Processing of Irradiated UQ, Percent of Charge Uranium Plutonium Oxidation Volatilization Volatilization Activity Step Step Step §) NA2 71-112 0.05-0.33 Pu Na? 0.025-0.95 31+63 B <0.0001"0-05 7.7_1317 007—0184 'Y <0c0001_0-5 6.9—1107 0.5-2-3 Zr 0 0-0.24 0-0.21 Nb 0-<0,0001 0.1.9 0-5.8 Ru <0.0001-3,3 44=71 3.2-14 Sb NAa <0.001-<2.7 <0,12~1,2 Mo NAg <51-76 <6-38 Te 0-0.08 NA® Na2 2NA = not analyzed, 207 More than 99.5% of the uranium and an average of <0.5% of the plutonium volatilized during the BrF_ step. Other workers (18) have shown that even less plutonium (0%) may be expected to volatilize during this step., Up to 87% of the krypton accounted for was released to the process off-gas during this step. Small amounts of ruthenium and tellurium, well below the recommended limiting concentrations, were also found in the stack gas. 3. During volatilization of plutonium with fluorine, up to v2% of the gross beta-gamma activity was transported con- currently. The principal gamma activity transported was ruthenium (3 to 14%), and up to 38% of the molybdenum and small amounts of zirconium (<0,21%), niobium (<5.8%), and antimony (<1.2%) were also volatilized. Analyses for other possible contaminanta such as tellurium, technetium, and neptunium were not completed. A small amount of krypton was found in the off-gas during the fluorine step of three of the five runs, suggesting that a small residue of uranium oxide remidined in the reactor after the BrF. step. The largest value for the krypton release was 9.6%. in addition to a small amount of ruthenium (<0.01%), a variable amount of tellurium (up to 40% of that charged) was found in the process off-gas. The results of this work can be summarized as follows: 1, The presence of fission products and a radiation field was shown to have little or no important effect upon product recovery from the fluidized bed of alumina. 2, In general, the distribution of the actinides and fission elements was found to be in accordance with expected distributions based upon the volatilities of the compounds formed. In some important instances, as exemplified by ruthenium, assignment of the compound formed is difficult since the experimental evidence suggests that more than one compound may be involved. 3. The uranium decontamination levels for the metal alloy fuels were high enough-that the preponderant gamma activity was due to the uranium isotope U. Gross decontamination from gamma activity was 510/ and that from beta activity was >100, 4., Plutonium decontamination levels from gamma activity of up to 3200 were achieved for the oxide fuel studies. The preponderant contaminant was shown to be ruthenium. 208 7. 10. References Anastasia, L. J., P. G, Alfredson, M. J. Steindler, G. W. Redding, J., G, Riha, and M, Haas, 'Laboratory Investigations in Support of Fluid-Bed Fluoride Volatility Processes, Part XVI. The Fluorination of UQ,-Pu0,-Fission Product Oxide Pellets with Fluorine in a 2-inch-Diameter Fluid-Bed Reactor," USAEC Report ANL-7372 (1967). Anastasia, L. J., P. G, Alfredson, and M. J. Steindler, "Fluidized-Bed Fluorination of U02-Pu0 Fission Product Fuel Pellets with Fluorine," Nucl, Appi.'i, 320 (1968). Anastasia, L. J., P. G. Alfredson, and M. J. Steindler, "Fluidized-Bed Fluorination of UQO,-Pu0,-Fission Pellets with BrF. and Fluorine, Part 1. %he F%uorination of Uranium, Neptuniuf, and Plutonium," submitted to Nucl. Appl. "Fuid-Bed Fluorination of UW02-PubD2-Fission Product Pellets with BrF_. and Fluorine, Part 2. Process Applications," submitted”to Nucl. Appl. Anastasia, L. J., P. G, Alfredson, and M. J. Steindler, Anastasia, L. J,, and W, J. Mecham, Ind. Eng. Chem., Process Design Develop. 4, 338-344 (1965). Anastasia, L. J., J. D. Gabor, and W. M. Mecham, "Engineering Development of Fluid-Bed Fluoride Volatility Processes. Part 3. Fluid-Bed Fluorination of Uranium Dioxide Fuel Pellets," USAEC Report ANL-6898 (1965). Anastasia, L. J., P. G. Alfredson, and M, J. Steindler, "Fluidized-Bed Fluoride Volatility Processing of UO -PuO2 Fuels with Simulated Burnups of 10,000 ard 30,000 M%d/T, Trans, Amer. Nucl. Soc, 11, 447 (1968), Anastasia, L, J., P, G, Alfredson, and M, J. Steindler, "Fluorination of Uranium Oxides in Fluidized-Bed Reactors," 1968 Tripartite Chemical Engineering Conference, Montreal, Canada (Sept. 1968). Trevorrow, L. E., T. J, Gerding, and M. J, Steindler, "Laboratory Investigations in Support of Fluid-Bed Fluoride Volatility Processes, Part XVII. Fluorination of Neptunium(IV) Fluorice and Neptunium(IV) Oxide,' USAEC Report ANL-7385 (1968). Centre d'Etudes Nucleaires de Fontenay-aux-Roses, Rapport Semestriel Du Departement De Chimie, Dec. 1967-May 1968, CEA-N-1044, p. 255. 209 11. 12, 13, 14. 15. 16. 17. 18. Trevorrow, L., E., T. J. Gerding, and M., J. Steindler, 1n "Chemical Engineering Division Semiannual Report, Jan.=-June 1967," USAEC Report ANL-7375 (1967). Trevorrow, L. E., T. J., Gerding, and M. J. Steindler, Inorg. Chem, 7, 2226 (1968). Katz, S.,, Inorg. Chem. 3, 1598 (1964). Ramaswami, D., et al., "Engineering Development of a Fluid- Bed Fluoride Volatility Process, Part 1, Bench-Scale Studies," Nuel. Appl. 1, 293 (1965). Holmes, J. T., et al., "Engineering Development of & Fluid- Bed Fluoride Volatility Process, Part 2, Pilot-Scale Studies," Nucl. Appl. 1, 301 (1965), Chilenskas, A. A., et al., "Bench-Scale Studies on Irradiated Highly Enriched Uranium-Alloy Fuels," USAEC Report ANL-6994 (1967). Chilenskas, A. A., '""Fluidized~Bed Fluoride Volatility Processing of Irradiated U0, Fuels,” Nucl Appl. 5, 11 (1968). "Chemical Engineering Division Semiannual Report, July-December 1965," USAEC Report ANL-7125, p. 68 (1966). 210 ENGINEERING-SCALE FLUORIDE VOLATILITY STUDIES * ON PLUTONIUM-BEARING FUEL MATERIALS N. M. levitz, E. L. Carls, D. Grosvenor, G. J. Vogel, I. Knudsen'? Argonne National Laboratory, Argonne, Illinois U. S. A, ABSTRACT Plutonium-bearing fuel materials were processed in an engineering- scale alpha facility as part of a program aimed at advancing the applicability of fluoride volatility methods to processing light- water-reactor fuels. A successful program of fluidized-bed fluorin- ation and thermal-decomposition studies was carried out on non- irradiated UO,-Pu0, pellet materials and PuF, powder charges in a study of key steps of flowsheets of current Interest. Overall, kilogram quantities of plutonium hexafluoride were produced, transported, and collected satisfactorily. The results of the work are expected to find application in developing recovery processes for high-plutonium materials such as fast-breeder- reactor fuels and plutonium scrap materials. *Work performed under the auspices of the U, S. Atomic Energy Commission, tPresent address: Westinghouse Electric Corp., Atomic Power Division, Cheswick, Pennsylvania. 211 INTRODUCTION Development work conducted in an engineering-scale alpha facility(l) on fluidized-bed fluoride volatility processes for the recovery of uranium and plutonium from spent uranium dioxide fuels 1s described. The program of studies was directed primarily at developing a flow- sheet for processing light water reactor (LWR) fuel typified by low-enrichment UOQ, pellets clad in Zircaloy. The results, however, have more far-reaching significance, being pertinent to reprocessing schemes (2) for high-plutonium liquid metal fast breeder reactor (LMFBR) fuels and plutonium scrap recovery processes 3) as well, Two flowsheets for LWR fuels have received major attentien to date: an all-fluorine flowsheet(4) and an interhalogen flowsheet. (3) Both have similar processing sequences: decladding with anhydrous HC1l, fluorination of uranium and plutonium to their respective hexafluorides, separation of uranium from plutonium, purification of the hexafluorides, and finally, reconversion to the oxides. Extensive use is made of flulidized beds, taking full advantage of their excellent heat-transfer and solids-mobility characteristies. Both schemes have as goals high (99%) recovery of uranium and plutonium. The major difference between them i1s in the method of fluorinating the actinides, The interhalogen flowsheet uses BrF5 as a selective fluorinating agent for the uranium, followed by fluorine for separate recovery of the plutonium, In the all- fluorine scheme, a partial separation of the uranium from the plutonium can be effected by appropriate choice of fluorination conditions (temperature and reagent concentration); generally higher temperatures and high fluorine concentrations are used to recover the plutonium. The major emphasis of work in the alpha facility was on the fluorination of plutonium-bearing fuel materials, demonstrating the feasibility of producing and transporting practical quantities of PuF,. Nonirradiated materials were used in all cases., Ex- periments were conducted on several aspects of the process: l. Two-zone oxidation-fluorination of U0,-0.5 wt % Pu0 pellets contalning simulated fission product oxides. 2. Thermal decomposition as a means of separating plutonium as PuF, from UF _-PuF_, mixtures produced in the two-zone oxidation-fluorination studies; decontamination from selected volatile fission products was also examined in these studies, 3. Fluorination of plutonium-containing materials remaining after oxidation and interhalogen (BrF_) reactions on Uo0,-0.5 wt % PuO2 pellets containing Simulated fission product oxides, 212 4, Fluorination of PuF4 powder in campaign-type experiments. Each set of experiments is discussed separately in the above order. Also, gsets 1 and 2 are treated as a unit in a discussion on material balances and the recovery of residual quantities of Plutonium from the equipment by a cleanup fluorination treatment. Experimental sets 1, 2, and 3 are described in detail in Reference 6; set 4 is described in Reference 7. The alpha facility comprises two large alpha boxes for the containment of plutonium, one containing process equipment and the other containing equipment for scrubbing and filtration of ventilation air and process exhaust gases. The process equipment in the large alpha box, shown schematically in Figure 1, includes a fluidized-bed fluorinator for fluorinating mixed-oxide fuel to UF, and PuF_, a converter reactor for converting PuF, to PuF by thermal decomposition, and a system of condensers ang chemical traps for collecting hexafluoride products. A 200-point data logger system was used to collect and process operating data. The fluidized-bed fluorinator has a 3-in., dia., 4-ft tall reaction section and is about 9 ft tall overall, with a disengaging section that expands to 15 in., and sintered-metal filters located above the bed. The fluorinator and associated process items are of nickel. The fluidized-bed thermal decomposer consists of a 2-in. dia., 2-ft tall reaction section topped by a 4-in. dia., 2-ft tall cooling and filtering section, all of Inconel, Both sintered Inconel and nickel filters were used in this unit. Both fluidized-bed reactors have inverted-cone bottoms with a single inlet at the apex for gas entrance, Solids were batch- charged through a top opening, in the case of the fluorinator, and through a side opening near the top, in the case of the thermal decomposer. Solids were generally removed from the bottom in both cases. TWO-ZONE OXIDATION-FLUORINATION STUDIES These studies simulated the primary actinide recovery step of the all-fluorine flowsheet. Batches of synthetic oxide fuel pellets were processed to a UF_ -PuF, product; the hexafluorides were collected in cold traps and then vapor-transferred to pro- duct receivers. The fluorination equipment, except for the product receivers, is shown in Figure 2. Information on rates of UF production and extent of removal of plutonium from the alumina bed and valuable operating experience were obtained in this first series of experiments on plutonium-containing materials. 213 FILTER CHAMBERS @% TO SCRUBBER Ry BAG-QUT PORT (TYPICAL) SOLIDS ADDITION\ | >,~ FILTER CHAMBERS SOLIDS . \/‘ L~ SOLIDS RECEIVER~__ . / CHARGER b <11 T | HEXAFLUORIDE ' | COLD TRAPS SUPPLY VESSELS s \J GAS SUPPLY rBAG-OUT PORT GAS PREHEATER-/ GAS SUPPLY SOLIDS RECEIVER RECEIVERS 1. Engineering-Scale Alpha Facility. 214 q1¢ PRIMARY THERMOCQUPLES T E Al,05-30" —= == —+ NICKEL - 3" BALLS SECONDARY GAS FILTER FLUORINATION REACTOR HEXAFLUQRIDE CONDENSERS { " — TO ACTIVATED Al05-FILLED — TRAPS NogF-FILLED TRAP 2. Fluorination Process Equipment. A 4 FLUORINE RECYCLE PUMP — - Each charge consisted of 8.8 kg of U0,-0.5 wt % Pu0, as 1/2-in. by 1/2-in. right-cylinder pellets containing some 19 f%ssion product oxides to correspond to 10,000 MWd/ton fuel. About 6.5 kg of alumina (Alcoa Tabular T-61, nominal 48-100 mesh) comprised the bed, which was reused for this set of three experiments. In effecting the recovery of the uranium and plutonium, the fluidized-bed fluorinator was operated first with two reaction zones and then with a single reaction zone. Initially, the pellets reside in the alumina bed in the lower portion of the reactor in the packed fluidized-bed mode. (8) The alumina was fluidized in the voids of the pellet bed, but also extended about 2 ft above the pellet zone, providing a second fluidized bed reaction zome, Oxidative pulverization of the pellets was effected in the lower zone at a temperature of 400-450°C with about 207 oxygen in nitrogen. Fluorine was admitted continuously via a side inlet to the zone above the pellets as the fluorlnating agent for the U 0 —PuO fines, which, were carried up from the pellet zone by the gas s%ream and the mixing action of the alumina. About 10% fluorine was used during the two- zone operating period with the fluorination zone at 450°C., Under these conditions, the bulk of the uranium was converted to UF_, while the plutonium was mainly converted to PuF,. Thermal conduc~- tivity cells were used to monitor the fluorine concentration in the off-gas., 9) Continuous welght readout of the cold traps provided a monitor of the UF6 production rate. After about 85% of the pellets were reacted, the operating mode was changed to single reaction zone operation by admitting the fluorine at the bottom of the reactor; oxygen flow was stopped at this point. Also, the gas flow that had been on a once-through basis was now put on recycle to conserve fluorine. The fluorine concentration was gradually increased to 80-90%, and the bed temperature was gradually raised to 550°C and maintained there for a given period (3 to 5 hr) to complete the recovery of the plutonium. Results and Discussion of Two-~Zone Oxidation~Fluorination Studies, Success of these experiments was judged primarily upon the complete- ness with which the plutonium was removed from the alumina bed. The extent of removal was determined by analysis of grab and final bed samples. Analytical data on plutonium content of grab samples for the three experiments are plotted in Figure 3 as a function of the duration of the plutonium fluorination period (designated Fluorination Cleanup Period). Residuval levels of plutonium equivalent to a maximum loss of about 1% were desired. This loss level was essentially achieved by using a single alumina bed for the three experiments, whereas the loss for a single use of the bed was perhaps as much as 5%. Percentage loss, of course, was a function of throughput, so the 216 400 450 £4004TO 450 TO 450 500 | FLUIDIZED BED TEMPERATURE, °C 500 ' 5T%°-i-——5so——J | 550 PLUTONIUM IN ALUMINA, w/o 0 002 I 2 3 | 4 5 l ! I I | 8 2 4 | | [ 6 | — ) 8 | RUNS Pu-} AND Pu-2 FLUORINATION CLEANUP PERIOD, hr‘\ 6 7 8 I I | I 10 | — RUN Pu-3 FLUCRINATION CLEANUP PERIOD, hr 3. Percent Plutonium in Alumina Grab Samples Remcved from the Fluorinator During the Fluorination Cleanup Period. 217 above results should be considered preliminary. High (near 100%) uranium removal was susgained in each experiment. Peak UF, production rates to 110 1b/(br)(ft“) were noted; average productgon rates for the three experiments were 41, 51, and 24 1b/ (hr)(ftz). Corresponding average fluorine utilization values were 55, 66, and 26%. Data indicated fluorine utilization might be optimized by automatic control of reagents, which should be implemented in future studiles. Elutriation of some plutonium (most likely as PuF,) from the bed by the fluidizing gas and subsequent deposition and fioldup of this material in the upper regions of the column were experienced. Nor- mal blowback of the filters and mechanical vibration by a pneumat- ically operated hammer was not completely effective in returning this material to the bed during the run. When the gas flow was stopped, however, at the end of a run, a portion of this material did fall back into the bed and gave higher values for plutonium content than did final grab samples. Caution must, therefore, be exercised in evaluating results of a given experiment. This holdup, of course, does not represent a loss but,rather, inventory that is recoverable by a cleanup-fluorination procedure, as was demonstrated later (see below). Subsequent experience indicated that the quantity of material retained in the upper part of the reactor reached a steady-state value and was probably a function of column design (relative surface area and geometry of these surfaces). This is a factor to be con- sidered in the design of new equipment. In the present system, holdup was on the order of 10-20 g of plutonium, In these initial experiments, even the lowest value represented a very significant fraction of a single charge; thus yield could not be used in evaluating the results., In later experiments, with 600 g of plutonium, this amount was a negligible percentage, and PuF6 production was an important parameter. These initial studies also provided some.insight into prefluorination requirements for experimental systems. It was concluded that inter- action between PuF, and nickel surfaces still occurred even after a vigorous prefluorination treatment with fluorine and accounted for a part of the PuF, holdup in the cold traps and other parts of the nickel equipment train, This holdup, on the order of grams of material, was recoverable by the cleanup-fluorination treatment, A pretreatment with CIF, was also used in two experiments, particularly to remove adsdrbed moisture that might have entered the system when the reactor was opened to the air during charging. This would not be a problem with continuously operated units, 218 THERMAL-DECOMPOSITION STUDIES Thermal decomposition provides a means to separate PuF, from UF _-PuF, mixtures, e,g., as an intermediate step in an all-fluorine flowsheet, or a means to effect some degree of purification of a PuF_ stream from volatile fluoride fission products or other im- purgties that are more stable than PuF,. On the basis of laboratory work performed by Trevorrow,(lo) thermal decomposition requires the appropriate combination of residence time for the gas and a tem- perature that favors the formation of PuF, in the absence of significant quantities of fluorine, consifiering the equilibrium PuF, 2 PuF, + F,. A fluidized-bed concept was selected for this study since the bed provides a large surface on which the reaction can occur, and isothermal conditions can be maintained for evaluation of the effect of temperature. The 10-kg batches of UF_-PuF, produced in the two~zone-reactor fluorination studies serveg as ?eed for these decomposition studies. Because of the low plutonium yield in the fluorination experiments, additional PuF, was spiked into the final batch of mixed hexa- fluoride feed go make a total of about 30 g of plutonium for the three experiments. Some volatile ruthenium and molybdenum fluoride species (speculated to be RuF_ and MoF,) were present in this material and provided preliminary information on fission product behavior in this system. A relatively fine (-100 mesh) alumina material was used as the bed so that a low fluidization velocity could be used, maximizing gas residence time. The 2-kg bed gave about a 12-in. depth (when static), and with a superficial fluidization velocity of 0.15 ft/sec, a nominal gas residence time of 10 sec was achieved., The feed was heated to about 80°C and fed as a 40% hexafluoride-60% nitrogen mixture, With the hexafluoride feed rate at about 20 g/min, a 10-kg batch of feed was processed in about 8 hr. Experiments were con- ducted at 350°C and 300°C, Results and Discussion of Thermal Decomposition Studies. The extent of separation of plutonium from the feed was determined from data obtained by analysis of feed (liquid and vapor) hexafluoride samples, bed samples, overhead grab (UF, product) gas samples, and samples from the final UF, product receivers. Although there was considerable scatter in a given set of UF, product samples, the plutonium levels were always low enough to indicate that good separation had been achieved. Separations efficiencies ranged from 99.2 to 99.99%, indicating the feasibility of this fluidized- bed separation technique, Analysis of grab samples from the bed over the course of the three experiments showed a gradual buildup of plutonium to about 219 1.46 wt %, The final bed, unexpectedly, also contained about 0,19 wt 7% uranium, The reason for uranium deposition is uncertain; it may be a result of reaction with "untreated (unfluorinated) sites' on bed particle surfaces. Less uranium codeposition occurred at the lower (300°C vs 350°C) bed temperature. The deposition of PuF, appeared to occur preferentially on the surface of the bed particles rather than in the gas phase. Thus, after a short period, the bed simulated a PuF, bed. Assuming appropriately sized PuF, could be obtained as a starting material in an actual application, a readily transportable product is made, The nature of the PuF4 coating on the alumina was not studied. Encouraging decontamination data for ruthenium and molybdenum were also obtained. Analysis of bed samples by a spark source mass spectrometric method showed ruthenium values of 0.2 ppm. The ruthenium throughput wag about 5 g £ 2,5 g, giving decontamination factors in the range 10- to 104, An accounting of molybdenum gave a decontamination factor of greater than 102, The data showed 23 g of molybdenum in the overhead UF, product and about 0.2 g in the bed (the bed analysis showed <0.0i wt %, the limit of the analytical method used). Further studies in this area are recommended. MATERIAL BALANCES To examine the disposition of plutonium in the equipment in the work to this point, an attempt was made to account for the overall quantity of plutonium introduced into the fluorinator and thermal decomposer in the course of the above-described experiments, Virtually all equipment exposed to PuF, was given a fluorination cleanup treatment, which consisted of recirculating 907% fluorine at about 300°C, Recovered PuF, was trapped on sodium fluoride, which is regarded as being 100% egficient as a sorbent for PuF,. Sodium fluoride traps were placed in such a way that the amount of plutonium recovered from a given section of the equipment train could be determined. TFor example, one trap was placed just downstream of the fluorinator; other traps were used downstream of each of the main cold traps. Lines and product receivers were treated similarly. The NaF used for sorption of PuF, was in the form of 1/8-in. by 1/8-in. right cylinders, Follcwgng PuF, recovery, the trap contents were ground individually, the resulting powder riffled to split out a representative sample, and these samples analyzed for plutonium content. The residual deposits of plutonium in the equipment were found to be distributed as follows: l, Small (vl g or less) quantities of plutonium were deposited in the lines and secondary filter between the fluorinator and the cold traps; 220 2, several-gram quantities of plutonium were deposited in product receivers and on the primary fluorinator filters (a part of the interim holdup discussed earlier); and 3. decagram quantities of plutonium were deposited in the cold traps. Mechanisms for deposition include radiation decomposition (v1-2% per day in the condensed phase), thermal decomposition, and reaction with nickel and other surfaces and with minor constituents of the system (e.g., a number of fission products were in the oxide pellets), Overall, a material balance of 887 was obtained for this initial series of experiments, assuming the input pellets were approximately 0.5 wt % Pu0, as indicated by the manufacturer. Some question re- mained as to the reliability of this input value, Nevertheless, the initial purpose of this program had been fulfilled in that operation of pilot-scale fluoride volatility systems with plutonium materials was shown to be feasible, Insight into the behavior of PuF, in these systems was gained such that more definitive ex- periments might be planned. STUDIES SIMULATING THE PLUTONIUM RECOVERY STEP OF THE INTERHALOGEN FLOWSHEET The interhalogen flowsheet proposes oxidation of oxide-pellet fuel to U,0,-Pu0, fines, followed by separate fluorination of the uranium and plutonium to hexafluorides, using first BrF_ as a selective fluorinating agent for the uranium, then fluctine to recover the plutonium. The reactions would be conducted sequentially in a single reactor. A brief program of plutonium fluorination studies was carried out to simulate the plutonium fluorination step of this flowsheet. Four experiments were made using 135-g charges of fine PuF, powder (-325 mesh) and fresh alumina beds. Two experiments involved PuF,- fission product residues in alumina, which remained after oxidation and BrF_ steps were conducted on 650-g charges of UQ,-Pu0,-fission product pellets in the laboratory fluidized-bed unit” (see”paper by M. Steindler, this volume); oxidation had been carried out for 4 hr at 450°C with about 20% oxygen (in nitrogen); uranium fluorination had been carried out at 300°C for 2 hr with about 10% BrF, (in nitrogen). About 20 g of plutonium and about 10 g of uranium remained in the bed to be processed, following fluorination, in each of these two cases. Yield data on the production of PuF, was obtained in two ways: by direct sorption on NaF and by direcg weight of PuF_., 1In the latter case,the PuF, was initially collected in the large primary traps and then vapor-transferred to a smaller cold trap that could 221 be weighed accurately. The small cold trap also had a NaF trap as a backup. Portable neutron survey meters (BF, type) were positioned at the PuF, collection points and providéd information on the accumulation of PuF, as a function of time and temperature. Starting fluorination temperatures were 200°C and 300°C, but the temperatures were programmed to increase to 550°C (with 25°C incremental changes) in response to the data provided by the neutron monitors. The characteristic re~ sponse of the neutron monitor was an increase in count rate immediately following a rise in temperature, followed by a pla- teauing of the count rate, The temperature was raised after pla- teauing was observed. Grab and final bed samples were also taken to follow the removal of plutonium from the alumina. Results of Interhalogen Flowsheet-Related Studies. Plutonium removal from alumina to residual values of 0.005 wt Z and 0.015 wt % in the bed was achieved in the PuF, and interhalogen residue experiments, respectively. Both values were satisfactory from an overall process standpoint, the former being equivalent to 99.7% removal and thelatter being equivalent to 98.7% removal. In the case of the latter, a total of some 75 g of plutonium had been processed using a single alumina bed. In general, somewhat greater retention of plutonium has been experienced when compounds of fission product elements were present, although the exact re- tention mechanism is not understood. Residual uranium levels were very low, 0.003 wt 7, equivalent to 99.97%Z removal. Yield results on PuF, were consistently good, up to 99%, in the present seriles of experiments, indicating production and transport of PuF, was feasible, The later series of campaign~type experiments (see below) conclusively demonstrated this, Poor agreement between analytical data on bed grab samples taken early in a run and the expected values on the basis of known charges of plutonium was again evidence of the problem of elutriation and served to emphasize the point that greater attention must be given to mixing characteristics of fine materials and relatively coarse (48-100 mesh) alumina. Effects of these operating variables on overall reactor efficiency needs further study. FLUORINATION OF KILOGRAM QUANTITIES OF PuF, A very significant series of campaign~type experiments were next performed, which firmly established the feasibility of producing and transporting PuF6 in engineering-scale equipment and collecting 222 it quantitatively. Information on the rates of fluorination of PuF4 from an alumina bed for scale-up purposes was also obtained. Three campaigns were conducted. Each campaign consisted of three successive experiments in which PuF, was fluorinated to PuF, with fluorine, followed by a cleanup-fluorination experiment, which in- cluded fluorination of the primary filter region and a separate cleanup of the lines and other equipment (secondary filter section and cold traps) to recover PuF, deposited as a result of alpha decom- position of PuF, (or other interaction mechanisms). A single bed consisting of agout 6500 g of 48-100 mesh prefluorinated alumina was used in each campaign. The effect of the starting temperature of the fluorination on plu- tonium behavior, i.e.,g, retention by alumina and overall recovery, was investigated. Starting temperatures were 300, 375, and 450°C in the three successive campaigns. Each experiment involved the fluorination of 200 g of plutonium (charged as -325 mesh PuF4). Thus, almost 2 kg of plutonium was involved in this program, compared with a total of about 600 g of plutonium used in some 15 experiments up to this time. The restric- tion of 200 g of plutonium per run was self-imposed because of the direct connection of the large glovebox to a mon-critically-safe aqueous scrubber. More specifically, only 200 g of plutonium, as PuF,, was to be contained in a single vessel, whereas up to 2000 g of plutonium in nonvolatile forms was allowed. For long range work, use of fixed or soluble poisons, boron Raschig rings or boron in solution should be considered in making the scrubber critically safe, The procedure for each experiment was as follows: The bed was brought to the starting temperature while fluidized with nitrogen; the gas stream was then put on total recycle, and fluorine flow was started at a rate equivalent to about 207 of the total gas flow rate. This quantity of gas was bled off at the same rate to main- tain the system pressure constant. The ratic of fluorine flow rate to nitrogen (purge and filter blowback gas) flow rate was such that a steady-state fluorine concentration of about 85%Z was attained. In each experiment the temperature was increased incrementally, 25°C every 15 min, until the final temperature of 550°C was reached, The total time for fluorination was 5 hr for each experiment in Campaigns 1 and 2. The fluorination time was reduced to 3 hr in each experiment of Campaign 3. The PuF, was collected in the two in-series cold traps used in the eariier experiments. The cold traps were operated at about -65°C, The PuF, was subsequently transferred with an inert-gas purge to NaF sorption traps. In two instances during high PuF production periods, the PuF, was trapped directly on NaF to obgain information on fluorine uti?ization and production rates and to 223 determine how close to equilibrium the system was operating., Samples from these NaF traps and from the alumina beds provided the basis for analysis of the experiments. Separate cleanup fluor- inations (2 hr of fluorination with fluorine at 300°C) were conducted on the primary fluorinator filters and remaining process equipment (cold traps, lines, secondary filter), and the amount and location of plutonium deposits in the equipment were determined. This information was useful not only for material balance purposes but also for studying the basic question of plutonium holdup as a function of plutonium throughput. If the plutonium is recoverable, this interim holdup does not represent a process loss, but merely reflects the need for an additional operating period for cleanup. Results and Discussion of PuF, Campaign-Type Experiments., Very encouraging results were obtained, as may be seen in the summary of data presented in Table 1, plutonium in the three final alumina beds, The low residual concentrations of 0.010, 0.029, and 0.022 wt %, represent a total loss of only about 0.25% of the plutonium charged. The reduction in run time from 5 hr in Campaigns 1 and 2 to 3 hr in Campaign 3 had no adverse effect on plutonium removal from alumina, Table 1 Summary of Fluorination Campaign Experiments Operating Conditions: Campalgn 1: 592 g Pu, 300-550°C, 5-hr experiments Campaign 2: 587 g Pu, 375-550°C, 5-hr experiments Campaign 3: 578 g Pu, 450-550°C, 3-hr experiments Residual Average Plutonium Fluorine Plutonium Concentration PuFg Production Utilization Material in Alumina Rate Efficiency?2 Balance Campaign (w/0) [1b/ (hr) (££2)) @ (%) 1 0.010 2.4 22 97 2 0.029 2.4 17 101 3 0.022 4,1 28 99 8Calculated as the amount of PuFg produced during the total run time compared with the amount of PuFg that could be produced at equilibrium (PuF, + ¥, z PuFg); the change in fluorine require- ment with temperature was considered in this calculationm. 224 The reduced operating time gave overall higher average PuF production rates, and fluorine utilization efficiencies were gigher in the third campaign than in the earlier campaigns. The improve- ment was also directly related to the amount of plutonium in the fluidized bed at a given time as shown in Table 2. For example, during the first half-hour of Run 2 of Campaign 3, the average quantity of plutonium in the bed was 204 g, and a fluorine efficiency of 98% was achieved. During the next half-hour, the average plutonium content was only 154 g, and the efficiency dropped to 51%. As the plutcnium content diminished during the final two hours, a further significant drop in efficiency occurred. It is likely that similar characteristics prevailed during the final period of the earlier 5~hr runs, The higher efficiencies and PuF_ production rates observed in Run 2 of Campaign 3 over those achieved in Run 1 also reflect the effect of plutonium content, the second run having started with a fresh charge of PuF, plus a heel of PuF4 from the first run equivalent to about 1/6 of a charge. Fluorine §fficiencies near 100% and production rates up toc 6.5 1b PuF6/(hr)(ft ) were obtained in the initial period of the second run. These data were obtained by collecting the PuF, directly on NaF traps and changing traps at 30-min intervals. Tfie observed high production rate is not a limit and further improvement in production rate could have been realized by starting with a high concentration of fluorine in the fluidizing gas; in the current procedure, the fluorine concentration is gradually increased from 0 to 95% in the course of gas recycle, Higher rates could also probably have been achieved by operating at a higher initial fluorination temperature, and by passing more fluorine through the bed by increasing the gas velocity and the fluorinator pressure. These results emphasize the benefits that might be realized by a continuous process, in which the required steady-state concentration of plutonium would be maintained to give a sustained high production rate for PuF,. This was the concept considered in the design con— cept study(29 for reprocessing IMFBR fuels, The material balance values, 97-101% (Table 1), lie within the range expected on the basis of a statistical sampling experiment. Sample treatment included milling (coarse grinding in a disk mill), sample splitting by riffling, and fine grinding steps, followed by chemical analysis. Gross sampling error proved to be about T4%. A complete plutonium material balance for one of the campaigns (Campaign 2) is presented in Table 3. These data show that about 99% of the PuF, charge was fluorinated to PuF, and collected in the two cold traps during the main fluorination period. The cold traps were quite efficient at the current cperating temperature of about =-65°C, Since the small loss (0.7%) sustained is about equivalent to the quantity represented by vapor-pressure consider- ations alone increased efficiency may only be achieved by operating at lower temperatures. 225 92¢ Table 2 Fluorine Efficiencies and PuF_ Production Rates U During First and Second Runs of Campaign 3 Operating Conditions Run 1 Run 2 in Period Amount of Amount of Fluorine Plutonium in Plutonium in Average PuF Conc. in Fluidized Fluorinator (g) Fluorine Fluorinator (g) Fluorine Production gate Time Fluorinator Bed Temp. Beginning End of Efficiency Beginning End of Efficiency [1b/(hr) (ft7)] Period ¢3) °c) of Period Period (%) of Period Period %) Run 1 Run 2 First Increasing Increasing 194.3 145.1 77 228.0 180.0 98 5.20 6.53 half- 6 »~ 91 450 -+ 525 hour Second 91 + 95 525 -+ 550 145.1 108.1 42 180.0 129.4 51 4.70 6.41 half- hour Final 95 550 108.3 33.7 17 129.4 35.7 20 2,25 2.97 two hours Table 3 Plutonium Material Balance--Campaign 2 Plutonium (g) % of Charge Charge 587.1 100.0 Processed to PuE6 Recovered from cold trap 1 566.4 96.5 Recovered from cold trap 2 7.6 1.3 Recovered from cold traps and lines during cleanup fluorination 8.0 1.4 Loss through cold traps 4.0 0.7% Subtotal 99,9 Recovered from Primary Sintered Metal Filters during Cleanup Fluorinatiom 4.1 0.7 Unprocessed and Lost Grab samples and reactor cleanout 0.7 0.1 Final bed (loss) 1.9 0.3 Total Accounted for 592.7 101.0 a Assumed to be recoverable by using a lower cold trap temp- erature. The problem of holdup of plutonium on the primary filters appears to diminish with increased plutonium throughput., The fraction re- tained on the filters during Campaign 2, 0.7% of the charge, was rather insignificant in terms of the quantity of material processed. In earlier studies with plutonium charges ranging from 20 to 100 g, filter holdup represented as much as 15% of the charge. Holdup, of course, is a function of total available filter area and is responsive to the filter blowback system. The recovery of material retained on filters is dependent on the ability to expose the filters to concentrated fluorine at 300°C (conditions required for cleanup). Examination of the filters after a total exposure at 300°C of about 30 hr disclosed no obvious deterioration. Much more testing will be needed, however, before conclusions about filter life can be made. The only nonrecoverable loss is probably represented by the quan- tity of plutonium retained by the alumina bed, about 2 g or 0.3% of 227 the charged plutonium. This value represents the loss sustained when the bed is used in only one campaign, Losses may be further reduced by using the same bed for several campaigns, if this technique is feasible in practice, Studies, to date, do not indicate the exact nature of the retention mechanism; this remains a subject of interest for future study. EXPERIENCE WITH NEUTRON COUNTERS AS PLUTONIUM MONITORS The use of neutron counting equipment was explcited rather widely in plutonium monitoring applications in the current program. Stan- dard neutron survey instruments with either 2.5-in. long or 10-in. long BF., prcbes were mounted at strategic locations (cold traps and NaF traps) and coupled to scalers and recording instruments to record the movement of PuF, from the reactor to the cold traps during fluor- . ination and from thé cold traps to the NaF traps during PuF, transfer operations. Operating procedures, such as tbe time-temperature pro- gram used in the fluorination experiments, were selected on the basis of the data obtained during PuF, collection in the cold traps. Similarly, the decision tc reduce the fluorination pericd from 5 to 3 hr was based on these data, Neutron response curves obtained during the first fluorination experiment of Campaign 3, which were typical of the curves cbtained in all three campaigns, are shown in Fig. 4., Figure 4a shows PuF6 sorption on NaF (Trap 1) during the first 3 hr of this experiment. The cbserved change in activity level during the first hour was in response to the collection of 86 g of plutonium. During the second and third hours, the PuF, was fed directly to the cold traps. As seen by the plateauirng in Fig. 4b, fluorination was essentially com- plete after the secend hour. The transfer of PuF, out of the cold trap and the corresponding sorption of this material ontc NaF (Trap 2) is shown in Fig. 4c. Approximately 74 g of plutonium was trans-— ferred in about 1 hr. Use cf these instruments for more quantitative measurements is limited, in large part, by equipment geometry. In addition, above about 40°C, these detecters are sensitive to temperature, Neverthe- less, their value has been proved, and further efforts to extend their use tec more quartitative applications appear to be warranted. CONCLUSTONS Fluidized-bed fluorination studies on sintered UOQ,-Pu0, pellet materials and PuF,-alumina mixtures established the %easigility of processing plutonium-bearing materials by fluoride volatility methods. Overall, some 3.5 kg of PuF, was produced, transported and recovered from the engineering-scale process equipment. Fluorination rates for both uranium and plutonium were shown to 228 be practical, and fluorination efficiencies high using recycle fluorination techniques, Neutron survey meters proved useful as plutonium monitors in the collection and transfer of PuF6. Thermal decomposition, using fluid-bed techniques, was shown to be feasible as an efficient means of separating plutonium as PuF from PuF6—UF mixtures. Some decontamination from selected fission products thag form volatile fluorides, such as ruthenium and molybdenum, was also demonstrated in this work. The method can be made continuous, which is a process advantage, The results of these studies are pertinent to the development of reprocessing methods for light water reactor and fast breeder reactor fuels as well as scrap recovery processes, o 8OO L L 2 800 T T T = c 3 ;. 3 L 1 > z 2 6001~ ofe — £ o0 - 5 £ o - — o - - > > 400 — — E 400 |— " — > z Eob 15t ] P4 A= g 200 — — g 200 — — g & 5t 15 r 1 ] w z fe) I [ 1 [ 1 z 0 & I | 1 | 1 0 | 2 3 4 ¢] | 2 3 4 RUN TIME, hr RUN TIME, hr 80 NoF TRAP 2 40 COLD TRAP | NEUTRON ACTIVITY, arbrtrary umits ) I 2 3 4 RUN TIME, br 4. Neutron Activity in NaF Traps and Cold Trap During First Fluorination of Campaign 3. a. PuF, Sorption on NaF (Trzp 1) During Fluorination; b. PuF, Collection in Cold Trap 1 During Fluorination; and c. BuF Transfer from Cold Trap \d 6 to NaF (Trap 2). 229 6. 10. References G. J. Vogel, E. L., Carls, and W. J, Mecham, "Engineering Development of Fluid-Bed Fluoride Volatillity Processes, Part 5. Description of a Pilot-Scale Facility for Uranium Dioxide-Plutonium Dioxide Processing Studies,' USAEC Report ANL-6901 (December 1964), N. M. Levitz et al, "Fluoride Volatility Plant Design Concept Study for LMFBR Fuels, (In preparation). R, L. Standifer, "A Fluid Bed Fluoride Volatility Pilot Plant for Plutonium Purification,' Paper 46e presented at the 64th National Meeting of the AIChE, New Orleans, March 17-20, 1969. "Chemical Engineering Division Research Highlights, May 1964- April 1965," USAEC Report ANL-7020, pp. 100-101. "Chemical Engineering Division Research Highlights, May 1966~ April 1967," USAEC Report ANL-7350, p. 21. N. M. Levitz et al, "Engineering Development of Fluid-Bed Fluoride Volatility Processes, Part 1l4. Processing Experience in Fluorinating Plutonium Materials and Thermal Decomposition Studies in an Engineering-Scale Alpha Facility," USAEC Report ANL-7473 (in preparationm). N, M. Levitz et al, "Engineering Development of Fluid-Bed Fluoride Volatility Processes, Part 15. Material Balance Demonstrations, Production Rates, and Fluorine Utilizations in Fluorination of Kilogram Quantities of PuF, to PuF, with Elemental Fluorine in a Fluid-Bed Reactor," UéAEC Report ANL-7468 (July 1968). J. D. Gabor and w. J. Mecham, "Engineering Development of Fluid-Bed Fluoride Volatility Processes, Part 4, Fluidized- Packed Beds: Studies of Heat Transfer, Solids-Gas Mixing, and Elutriation,'" USAEC Report ANL-6859 (March 1965). D. Ramaswami et al, "Engineering Development of Fluid-Bed Fluoride Volatility Processes, Part 11. Off-Gas Analysis," USAEC Report ANL-7339 (July 1968). L. Trevorrow, J. Fischer, and J. Riha, "Laboratory Investigations in Support of Fluid-Bed Fluoride Volatility Processes, Part III. Separation of Gaseous Mixtures of Uranium Hexafluoride and Plutonium Hexafluoride by Thermal Decomposition,' USAEC Report ANL-6762 (August 1963), 230 THE POTENTIAL OF THE FLUORIDE VOLATILITY PROCESS * FOR FAST BREEDER REACTOR FUELS A, A, Jonke, N. M. Levitz, and M. J. Steindler Chemical Engineering Division, Argonne National Laboratory, Argonne, Illinois. U. S. A. ABSTRACT At the direction of the AEC, a high priority has been given to a study to define the potential of fluoride volatility reprocessing for application to fast breeder reactor fuels. To provide the needed information, a conceptual design of a volatility reprocessing plant has been prepared, together with an extensive critique of the process. The study included the selection of a process flowsheet, preliminary equipment design, rough plant lay- out, and a discussion of the uncertainties associated with the process. The results of the study indicate that fluoride volatility processing of fast reactor fuels has significant potential advan- tages, but that the development task may be as great as for other potential methods of reprocessing. * Work performed under the auspices of the U,S. Atomic Energy Commission, 231 INTRODUCTION In order to make fast breeder reactors competitive with other forms of energy generation, it is anticipated that fuel-cycle costs of about 1/2 to 1 mill/kw hr will be required. To reduce costs to this level, considerable improvement in all parts of the fuel cycle--including reprocessing--will be required. Reprocessing technology for fast breeder fuels might evolve through the develop- ment of advanced aqueous processes or, alternatively, by the intro- duction of new nonaqueous processes, To insure that a .suitable reprocesgsing technology is available to achieve the long range fuel cycle cost objectives, it seems prudent to investigate more than one approach to the reprocessing of fast breeder fuels. In 1968, the USAEC requested that Argonne National Laboratory conduct a study to help define the potential of the fluoride volatility process for application to liquid-metal-cooled fast breeder reactor (IMFBR) fuels, The objectives of the study were to present the current technological basis for a volatility proc- essing plant in the form of process and engineering flowsheets and to define process uncertainties. The uncertainties were to be translated into key problem areas in order to provide insight into the magnitude of the development task associated with estab- lishing the volatility technology for ILMFBR fuels. Economic estimates were not included in this study. A conceptual design of a volatility reprocessing plant has been prepared, together with an extensive analysis of the process. In addition to process and engineering flowsheets, the study includes preliminary design of major equipment items, and basic layouts of the processing cells and the overall plant, The ground rules for the study were set up in advance. The conceptual reprocessing plant was to be a large central plant serving nuclear power reactors with a total capacity of 15,000 MW(e). The processing plant capacity corresponds roughly to a fuel load of one ton per day. The plant would process core, axial blanket, and radial blanket with a decontamination factor of 10° to 107/ and an overall minimum recovery of 99% of the uranium and plutonium. The reference core fuel element was based on an Atomics International preliminary design. Some of the characteristics of this fuel design are given in Table 1. The specific power of the core fuel was taken to be 200 MW(t)/metric ton and the burnup 100,000 MWd/metric ton. A fuel cooling time of 30 days before processing was chosen to help provide low out=-of-reactor plutonium inventory costs. 232 Table 1 Reference IMFBR Core Fuel Element Fuel pin diameter: 0.25 1in, Clad thickness: 0.015 in. S8 Core active height: 4,0 ft Axial blanket lenght: 1.0 ft each end No. of fuel pins per element: 217 Spacing between pins: 0.05 in. Smeared fuel density: 80% T.D. (Core) 93% T.D. (Axial Blanket) Fission-gas handling: nonvented Element shape: hexagonal, 5.4 in. across flats Shroud thickness: 0.17 in, S8 Cladding length: 144 in. (includes fission gas plenum) PROCESS FLOWSHEET The process flowsheet was selected on the basis of existing knowledge and conservative extrapolation for process steps not clearly demonstrated. The choice of process steps rested on their high potential for successful development. The flowsheet was not optimized in any sense, A simplified version of the con~- ceptual flowsheet is shown in Fig, 1, and some of the details are shown in Figs. 2, 3, and 4. The mechanical head-end scheme includes mechanical disassembly of the fuel elements, chopping of the fuel pins, and a step to separate the cladding and convert the oxide to a powder. The homogeneous powder can be sampled for input accountability and fuel burnup determinations. Details of the mechanical methed for separating fuel and cladding segments are a little indefinite at present, since only a limited amount of experimental data is available on which to base this operation. One concept involves tumbling the chopped fuel segments In a ball-mill to separate the fuel powder, but other alternatives also appear possible, The powdered fuel oxide is next fed along with powdered alumina to the first of two in-series fluid-bed fluorination units, which provide bulk separation of the uranium and plutonium as well as partial decontamination. The bulk of the uranium is removed as volatile UF, from the first reactor, and the bulk of the plutonium is removed as volatile PuF, from the second reactor, a two-stage unit. Continuous operation and staging appear feasible on the basis of the current status of fluidization technology. Fluorine gas is the only fluorinating agent used in the con- ceptual process, Fluorination at 350°C with 20% fluorine-oxygen in the first reactor converts the bulk of the uranium to UF6 and 233 1. Fluoride Volatility Process for Fast Reactor Fuels, UFg-PuFg-FP ({2 ' PARTIAL CONDENSER CONDENSER RuFg NbF, ‘ Al»O S UFs'PUFG TO IO?(g?hr SEPARATION STEP —FLUORINATOR ——— FUEL OXIDE PuFg POWDER F.P. 47kg/h = o Pere o] fPARTIAL CONDENSER CONDENSER v RuFsg 500°C NbFs PuFg TO 2 FLUORINATOR 550°C F2 l 1 1.7 kg/hr Alx03-F.P. WASTE 2. Continuous Fluorination Step. 234 GAS TO RECYCLE GAS TO RECYCLE SEPARATION STEP UFg- F.P =5 ABSORBERS RI.IF5 .NbFs. SbF5 HF TO WASTE -80°C s CONDENSER] S g CONVERTER -1 B E R DISTILLATION H o PuFe -[ COLUMNS Hoo UFg-PuFg-FP UFg 2 TO FLUORINATOR WASTE Fa - 3, Product Purification. Hp Xe) l Kr IFs ACTIVATED SODA || prvERl—e 02 {Xe | RARE GAS TRITIUM ALUMINA LIME CONVERTER [ Kr | COMPRESSION 52 (FLUORINE (10DINE (TRITIUM l (3 REMOVAL) REMOVAL) REMOQVAL) 4, Off-Gas Treatment. 235 WATER less than 5% of the plutonium to PuF_,. The UF6 production rate is approximately 100 1b/ (hr) (££2). Durgng fluorination with undiluted fluorine gas at 500°C and 550°C in the upper and lower stages of 9 the second reactor, the PuF6 production rate is about 13 1b/(hr) (£t7). The alumina stream, which cascades through the fluorination reactors and finally to waste receivers, represents the main solid waste stream, providing a vehicle for the disposal of those fission products in the feed that do not form volatile fluorides, The gas streams from each of the two fluid-bed fluorination units pass through a fission product trap (partial condenser) and a hexafluoride-collection cold trap, where separation from some fission products occurs. The UF, product stream may contain sufficient plutonium to warrant recovery of PuF, from this stream. This 1s accomplished in the present flowsheet by recombining the hexafluoride product steams and carrying out a more nearly quanti- tative separation in a fluid-bed thermal decomposition step, where PuF, is converted to solid PuF,. This step also gives further pur?fication of the plutonium grom remaining volatile fission prod- ucts, The UF, stream passes overhead and is purified by a com- bination of fractional distillation and sorption traps. The PuF, produced by thermal decomposition is subsequently refluorinated to PuF, with concentrated fluorine (n100%) at 500°C and combined wgth the desired proportion of pure UF,. The mixture is fed to a fluid-bed converter, where a dense homogeneous Pu0,-U0, particulate solid is produced by simultaneous reaction of the " hexafluorides with steam and hydrecgen. The conversion process is based on work conducted earlier on UF, alone. Excess UF6 may be converted to the oxide for use in other reactors. PLANT DESIGN The daily load of fuel to the plant is shown in Table 2. Because of the short cooling time, the heat-load from fission product decay is very large. It is assumed, therefore, that the fuel will be transported to the reprocessing plant in sodium-filled containers inside of shielded casks, or by some similar method which allows for removal of heat during transport. The removal of heat from the reaction vessels required very careful consideration in the equipment design, Heat loads are most severe in the main fluorination vessels, which have large inventories of fission products, and particularly in the uranium fluorinator, which has a large chemical heat load in addition. Maximum heat fluxes are on the order of 11,000 Btu/hr £t2, and calculations show that satisfactory heat removal can be achieved with air cooling of a finned reactor surface, The slab design for the fluorination vessel offers several advantages with regard to 236 Table 2 Daily Processing Load Plant Capacity: 15,000 MW(e) equivalent Daily Load: 6 core fuel assemblies 4 blanket fuel assemblies 875 kg uranium 83.5 kg plutonium 39 kg ;ission products 2 x 10" curles fission products Heat Load: 13.2 kW per core assembly the heat problem. It provides a greater surface for heat transfer than would a cylindrical vessel of comparable volume and in addition presents a small dimension across which the heat must be transferred. Thus, in the event of a loss of fluidization in these vessels, heat removal by conduction should avoid any serious consequences due to high temperatures at the center of the vessel, On the basis of known or estimated reaction rates and other information, the equipment sizes for the conceptual plant were calculated., The sizes of the major equipment items are shown in Table 3. Overall plant problems such as criticality, accountability, and plant safety in the event of hexafluoride release have been considered in this preliminary evaluation. The approach to criticality control adopted for this volatility-plant concept is one that avoids neutron moderation and minimizes neutron reflection to obtain a low reactivity per unit mass of plutonium in the process system, All vessels expected to contain significant quantities of plutonium are of a slab design which lends itself to safe-by-shape geometry, Preliminary criticality calculations indicate that 100 kg Table 3 Major Plant Equipment Item Size and Shape Fluorinator A Slab -~ 4" x 48" x 10’ Fluorinator B Slab - 4" x 30" x 10’ Thermal Decomposer Slab - 4" x 31" x 7' Distillation Column A Cylinder - 4" x 15°' Distillation Column B Cylinder - 3.5" x 25' Converter Slab - 4" x 14" x 7' Cold Traps Slab - 4" x 24" x various or Slab - 4" x 48" x various 237 of plutonium could be safely contained in a 4-in. thick slab reactor of nickel, 48 in. wide containing PuF, at its theoretical density (7 g/cc) and reflected top and bottom by alumina bed material. The normal operating inventory in the present flowsheet is below 50 kg of plutonium. Water is excluded from the process, both internally and externally to preclude neutron moderation. For both normal and credibly abnormal situations, an acceptable factor of safety is predicted. Accountability and burnup analyses are accomplished by sampling the fuel before it 1s fed into the first chemical process stage and by sampling the final waste streams (which establishes loss levels)., Weights and analyses of the final products (the mixed Pu0,-UQ, product of the conversion step) provide the remaining neceéssary information for accountability. Waste dispesal is accomplished by converting all wastes to gsolid form. The principal high level wastes are: (1) the alumina waste containing all of the nonvolatile fission products and (2) the ruthenium-niobium pentafluoride from the partial condensers. The alumina containing the nonvolatile fission products is discharged to waste storage cylinders, Aluminum shot or coarse powder 1s added to the waste as it is transferred to the storage cylinder to promote the transfer of heat from the center of the cylinder to the walls and thus lower the centerline temperature. These cylinders are stored under water in a storage canal to permit partial decay of fission products. The ruthenium-niobium fluorides collected in the partial condensers are removed periodically by warming the condenser and transferring the volatile fluorides in a gas stream to a bed of sodium fluoride where the fission products are sorbed, This NaF is then transferred to storage cylinders like those used for the alumina waste. A total of about 200 cylinders are required per yvear, each cylinder being 2 ft in diameter by 9 ft tall., Less than half of these waste containers require interim storage for decay of radioactivity to a level which will permit dry storage. Considerable design and layout work would be needed to estimate the size of the radiochemical processing cells and the total plant with any appreciable degree of accuracy. This is beyond the scope of the present study, which is only intended to determine process feasibility. Nevertheless, a very preliminary layout of plant equipment was made and from this, the processing cells and the building appear to be of practical sizes., Remote maintenance was selected for the most radioactive sections. Major repalr work will be done in a separate maintenance area located in a sublevel equipped with shielding windows and manipulators. 238 CONCLUSIONS The application of fluoride volatility processing to IMFBR fuels is supported by a substantial body of basic and technological information, which has been generated in reprocessing development work on other nuclear fuel materials and in various related commercial processes. Among the most pertinent areas of earlier work are: (1) commercial refining of uranium in the Allied Chemical Corp. plant at Metropolis, Illinois; (2) extensive development work on fluoride volatility processing of several types of irradiated fuels; (3) basic and pilot (kilogram) scale work on the preparation and transport of PuF_.; (4) plant-scale experience with the fluid-bed calcination of radicactive waste solutions at the Idaho Chemical Processing Plant; (5) the first planned commercial ébplication of volatility processing to irradiated fuels in the General Electric Midwest Fuel Recovery Plant. The potential of fluoride volatility for processing IMFER fuels may be measured by evaluating the feasibility of the process design and reference plant concept developed in this study. The conceptual plant has a practical size. Most of the steps of the conceptual process have reasonably sound bases in current technology. The techniques employed--continuous fluid-bed fluorination, hexafluoride cold trapping, fracticnal distillation, and pneumz2tic-conveying of solids--are basically the same as those used in the earlier work cited above. It is in the extension of their use to highly radio- active, high~plutonium fuel that uncertainty arises. Our analysis of the conceptual process has defined several key problems as follows: 1. Mechanical decladding of fuel involves difficulties from the high rate of radicactive decay heat generation; also, it may be difficult to insure that all fuel oxide has been removed from the fuel hulls by the conceptual ball milling procedure; supplementary cleanup of hulls may be required. 2. Continuous fluorimation will require the development of reliable solids feeding devices and unique equipment such as slab- shaped fluorinators and dual-stage reaction vessels., Plutonium losses in the alumina waste from the fluorination steps must be low, 3. Further development and testing is needed to insure that plutonium decontamination will meet requirements. 4, The transport and handling of solid materials in a processing plant operating with a high on-stream factor needs development and verification, 239 5. Adequate containment of process wastes gases is an essential requirement; this needs substantial study and development. 6. Further work is needed to insure that criticality safety and containment of PuF6 will meet 2ll requirements. 7. The role of sodium (introduced into the process through leaking fuel pins) and its potential effect on plutonium losses will require additional study, since it is known that sodium fluoride forms complexes with PuFe, causing plutonium to be irreversibly sorbed. Although these key problems have been defined by analysis of the reference process selected for this study, the problems are also representative of those existing in alternative flowsheets, which were considered briefly. Solutions to these problems cannot be considered simple; but none of them appears to be insoluble, nor unduly complex when compared to similar problems for other processing methods. Finding solutions to these key problems would be the first stage of a development program on fluoride volatility. We believe that the conceptual design study provided the desired insight into the magnitude of the development task required to establish the volatllity technology for LMFBR fuels. This result, however, cannot easily be expressed in quantitative terms. It is possible to state that an extensive development program would undoubtedly be required, but it cannot be said that the magnitude of the task would be either greater or smaller than that required for development of any other method of reprocessing LMFBR fuels. 240 CHLORINATION-DISTILLATION PROCESSING OF FRRADIATED URANIUM DIOXIDE Kenmei Hirano Division of Chemistry, Tokai Research Establishment, Japan Atomic Energy Research Institute, Ibaraki-ken, Japan Takehiko {shihara Office of Planning, Japan Atomic Energy Research Institute, Tokyo, Japan (formerly Division of Fuel Research and Development, Tokai Research Establishment) Abstract Using the difference in vapor pressure between the chlorides of uranium and those of fission products, a chlorination=-distillation process of irradiated uranium dioxide fuel was investigated. Uranium recovery and decontamination were poor on a single distillation. After various improvements of the process, very high decontamination was achieved by using barium chloride as a sorption-desorption medium for the vapor of uranium chlorides, The most promising process was as follows: Irradiated uranium dioxide pellets were pulverized through an oxidation-reduction cyclic process and chlorinated by the mixed gas of argon (60%) and carbon tetrachloride vapor (40%) at about 580°C. The gas containing the vapor of the chlorides of uranium and some fission products formed was passed through the barium chloride bed in the temperature range from about 500° to 100°C, and the vapor of uran- ium chlorides was sorbed on the bed. After the high vapor pressure fission product chlorides which were sorbed with uranium chlorides were preferentially desorbed by heating the bed to the temperatures of about 460° to 480°C in the gas stream of argon (90%) and carbon tetrachloride vapor (10%), the uranium chlorides were desorbed as the vapor of higher uranium chlorides by heating the bed to the tempera- tures of about 500° to 550°C in the mixed gas stream of argon (15%), chlorine (70%) and carbon tetrachloride vapor (15%). The vapor was trapped and uranium was recovered. After repeating the sorption- desorption process twice or three times, it was possible to attain decontamination factors to gamma emitters of 5.3x103 or 6,5x10%, respectively, and uranium recovery of 96% or 94%, respectively, com- pared to the single treatment value of 1,0x102 and 98%. 241 Introduction The chloride volatility process has been applied to the pr?c§ss- ing of irradiated nuclear fuel by some research groups. Gens ] applied it to the processing of irradiated uranium dioxide fuel. NaumanniZ2 applied it to the separation of uranium and plutonium in the processing of irradiated uranium dioxide-plutonium dioxide fuel, Speeckaert(3) was concerned about the application of the chloride process to irradiated nuclear fuel, and applied it to the processing of irradiated metallic and ceramic fuels. The authors applied the process to separate uranium from irradi- ated uranium dioxi?fi gfiing carbon tetrachloride vapor as the chlorination agent\7™®/ | and experimented on decontamination of uranium chlorides by a sorption-desorption process on a barium chloride bed, Preliminary Test The major components formed by chlorinating U0y with CCly vapor are UClL, UCls and UClg, and the quantity of UClg and UClg increases as the chlorination temperature rises. However, UClg and UCIg are relatively unstable, particularly UCl5. They rapidly decompose to UCly and Cly by cooling in an atmosphere which does not contain free chlorine gas. They are formed by heating UCT4 in an atmos- phere containing large quantities of free chlorine gas. This behavior was applied to decontaminate uranium chlorides. The vapor of the chlorides of uranium formed by chlorinating unir- radiated U0y powder in a gas stream of Ar-CCly vapor were passed through a bed of anhydrous CaClp, SrCl2 or BaCl2. Then, sorbed uranium chlorides were desorbed as higher chloride vapors by heating in a gas stream of Ar-Cla-CCly, Experimental results are given in Table 1. BaCly is evidently the most promising. Experimental Materials Uranium Dioxide = Uranium dioxide powder in sizes from 5 to 10un was used for unirradiated samples. The irradiated samples used were uranium dioxide pellets of 7.5 mm both in diameter and in height. They were irradiated to an integrated thermal neutron flux of about 1x1018 n/cm2 in the Japan Research Reactor-2, After letting the activity decay for a period of about 200 days or 400 days, they were chemically pulverized by oxidation (400°C, with air)- reduction (800°C, with hydrogen gas) cyclic process. The sample used in one run ranged in weight between 0.05 and 0.5 gm, 242 Table 1. Experimental Results of Preliminary Tests Temperaturel C) State of Sorption Chlori- Sorp- nation tion CaCl2 Bed SrCl2 Bed BaCl2 Bed 600 100 Almost sorbed Perfectly sorbed Perfectly sorbed 600 200 Passed a little Perfectly sorbed Perfectly sorbed 600 300 Fairly passed Passed a little Almost sorbed 600 400 Almost passed Passed a little Passed a little Desorption State of Desorption Temperature °c) CaCl2 Bed Sr012 Bed Ba012 Bed 300 Almost desorbed Almost retained Perfectly retained Loo Almost desorbed Almost retained Perfectly retained 500 —_— Fairly retained Perfectly desorbed 600 —_— Fairly retained Almost desorbed Barium Chloride - The BaCly consisted of anhydrous granules screened in four sizes of =W+5, 546, =6+7 and =-7+10 mesh with Tyler sieves. The void fraction of the BaClz fiiled in a reaction tube was about 50%. Sodium Chloride - The NaCl was coarse powder of reagent grade. Gases - Reagent grade carbon tetrachloride and market grade chlorine gas were used. Argon gas was used for dilution of the CCly vapor and for sweeping of the gas in a reaction tube assembly. Air and hydrogen were used respectively for oxidation and reduction of the UDy pellets. Trace oxygen which remained in the argon or hydrogen was removed by passing through copper gauze heated to about 400°C, and dried through a silica-gel column. Apparatus The experimental apparatus shown in Fig. | consists of a tubular furnace, which is 35 mm in internal diameter and 450 mm in length, and several reaction tubes. The furnace is of horizontal type and electrically heated by nichrome wire. The temperature in the furnace is controlled automatically. All the reaction tubes were made of transparent quartz glass and their dimensions are given in Table 2. 243 In the case of unirradiated samples, the desorption of uranium chlorides was carried out by heating the BaClz bed to the tempera- tures of L0O0C to 580°C in the mixed gas stream of 15% Ar-70% Clo- 15% cCly, vapor with the flow rate of 300 cc/min at 25°C. The desorbed chlorides were sorbed on the NaCl! bed in the tube L4-B. Then, uranium in each reaction tube was analyzed chemically. In the case of irradiated samples, the desorption of chlorides of fission products was carried out by heating to the temperatures of 450° to 550°C in the mixed gas stream of 60% Ar-40% CCly vapor or 90% Ar=10% CCly vapor with the flow rate of 95 cc/min. at 25°C, and the desorbed chlorides of fission products and uranium were sorbed . on the NaCl bed in the reaction tube 4-B, The fission products and uranium in the reaction tube 4-B were analyzed radiochemically or chemically. The desorption of the sorbed uranium chlorides was carried out in the mixed gas stream of 15% Ar-70% Ci2-15% CCly. Results Sorption of Uranium Chlorides Effects of Gas Flow Rate and Chlorination Temperature - Granular BaCly of -b6+7 mesh size was used. Chlorination-gas flow rates of 80 to 110 cc/min. at 25°C, and various chlorination temperatures were investigated, From the results obtained the optimum temperature and gas flow rate were determined to be 580°C and 95 cc/min. respectively. Effect of Size of Granular BaCly - With a fixed gas flow rate of 95 cc/min experiments were conducted with BaClp granules of -4+5 -5+6, -6+7 and =7+10 mesh. With -4+5 and -5+6 mesh granules and a chlorination temperature of 540 to 620°C the uranium chlorides sorbed . on the granules extending from the hot zone to the colder region where the temperature was about 170°C. wWith -6+7 and -7+10 mesh granules and a chlorination temperature of 540 to 580°C the uranium chlorides were all adsorbed on a much narrower region of the bed or on that part of the bed which extended from the hot zone to the region where the temperature was about 300°C, Effect of Length of BaClo Bed - The effect of the length of the BaCl, bed in the furnace on the distribution of the uranium chlorides sorbed on the bed was investigated. The lengths used were 4, 5, 6, 7 and 8 cm, The chlorination temperature, the flow rate of the chlor- ination gas at 25°C and the size of the granular BaClp were fixed to 580°C, 95 cc/min and -6+7 mesh, respectively. Experimental results are shown in Fig. 2. With a 4 cm bed, uranium chlorides condensed in the empty reaction tube 1-A where the temperature was above AOOOC, which was the temperature of the hot end of the BaCl, bed, However, this phenomenon was not observed 246 L% Bed length in furnace 4 cm 5 cm 6 cm 7 cm 8 cm 8o} - - L 70;— = - 601~ (%) 501 4o NN m-—- AN wt. % of uranium 20 N\ RESS T T . AR TR RN NN NNyt NN N AN,y S RO .. N AN RS ANANANNNN Ve L1 ;fl L1 1 b 2 02 4 5352 02 4 64 2 02 4 754372 0 2 & 86 42o02h% inside | outside inside T outside inside ' outside inside ' outside inside ' outside End of furnace N Location in reaction tube assembly 2. Effect of Bed Length on Distribution of Uranium Chlorides sorbed on BaCl2 Bed (Sample : UO2 powder 1g ; gas flow rate at 25°C : 95 c¢/min ; size of granular BaCl, : -6+7 mesh ; chlorination temperature : 580°C) with a 5 cm bed, when the temperature of the hot end was about 440°C, Thus, the highest condensation temperature of the vapor or uranium chlorides in the reaction tube with no bed was estimated to be between about 400°C and 440°C, With 4, 5, and 6 cm beds the sorbed band of uranium chlorides extended to regions of the bed where the temperature was below 170°C. With a 7 cm bed it extended to 170°C and with an 8 cm bed it was limited to regions of the bed where the temperature was 3009C or above. In the following experiments, the reaction tube 2, the chlorina- tion temperature, the flow rate of the chlorination gas at 25°C and the size of the granular BaCly were fixed to C, 580°C, 95 cc/min and -6+7 mesh, respectively. Desorption of Uranium Chlorides Effect of Bed Temperature - Experimental results are shown in Fig. 3. Good results were obtained when the bed temperature was from 480° to 560°C. The effective temperature range when the reac- tion time was one hour was slightly wider than when it was 30 minutes. 100,1111,11[1,1!!811111[ 90+ O — 7 o 0o 8ol —5 . (%) N\ o) O treated for 30 min. o ® treated for 1 hr. = Desorption ratio 10r - 400 450 500 550 600 Temperature (°C) 3. Effect of Bed Temperature on Desorption of Uranium Chlorides in Mixed Gas Stream of 15% Ar-70% 012-15% CClk Vapor. 248 Effect of Reaction Time - The bed temperature was fixed to 500O and 540°C, The desorption of the uranium chlorides reached 99% after 90 minutes reaction for 0.5 gm sample, and after 2 hours reaction for 1 gm sample. All the uranium chlorides were desorbed after 2 hours reaction when the sample weight did not exceed 1 gm, Sorption of Chlorides of Fission Products Samples were allowed to decay for about 200 days after irradiation before they were processed, Experimental results are shown in Fig. 4, The nuclides degectable,as shown in the figure, were 95zr- Nb, ]03Ru, 106g,.10 Rh, 141ce and 1Hh4ce, Their percentages are given in Table 3. All the chlorides of zirconium, niobium and ruthenium were volatitized through the chlorination process and sorbed on the BaCl2 bed. Although most of the cerium chlorides remained in the boat without volatilization, a small fraction of them was volatilized and sorbed on the BaCly, The uranium chlorides formed were volatilized and completely sorbed on the bed. Table 3. Distribution of Main Fission Products after Chlorination-Sorption Percentage of Y-enittinggfisaion Products 1%1Ce &14hce 103hu &106Ru-106Rh 952r-95Nb Not volatilized 91 0 0] Sorbed on BaCl2 bed 9 60* 100 Passed through BaCl2 bed 0 ¢ 0.2 *About 40% of 1008 and "%ru-"%gh were already separated through chemical pulverization of irradiated UO2 pellets by oxidation-reduction cyclic process. Desorption of Uranium Chlorides in the Mixed Gas Stream of Argon and CCly Vapor Attempts were made to desorb the chlorides of fission products and not desorb the uranium chlorides in the mixed gas stream composed of argon and CCly vapor, At first, the desorption of the uranium chlorides sorbed on the BaCly bed was investigated in the mixed gas stream, The desorption temperatures tried were 4502, 4709, 480°, 490°, 5109, 530° and 550°C, The flow rate of the desorption gas at 259C and the desorption time were fixed to 95 cc/min and 30 minutes, respectively. 249 0s? Ce Ce h eI 10 103, 106, 106, |4 10 ey ‘@ g 10— s ] 5 - 5 o ol | E 10 i paased through BaCle be 1 ‘ 4 10 p— 1 1 1 1 ) 1 1 1 1 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Y-ray energy (MeV) 4, Distribution of Y-emitting Nuclides after Chlorination-Sorption. Experimental results are shown in Fig. 5. The highest temperature where the uranium chlorides did not desorb varied slightly with the concentration of CCl, vapor in the gas. The temperature was 490° and 480°C with 10 and #0& CCl, in the gas, respectively. The desorption rate of uranium chlorides was increased linearly with the bed temperature rising. Desorption of chlorides of fission products It was confirmed that the uranium chlorides sorbed on the BaCl2 bed were not desorbed in the mixed gas stream of 60% Ar-40% CCl vapor or 90% Ar-10% CCl, vapor at temperatures of 480°C or below. The desorption of the chlorides of fission products was investigated under the same experimental conditions as the desorption of uranium chlorides in the mixed gas stream of 90% Ar-10% CClh vapor. Experimental results are given in Table 4. Retention ratio and residual ratio of a nuclide A were respectively defined as follows: Retention ratio of a nuclide A Y-activity of A retained on the bed after desorptlon (1) “Y-activity of A in irradiated an pellet and Residual ratio of a nuclide A Y-activity of A retained on the bed after desorpt1on (2) “Y=activity of A on the bed before desorption As shown in Table 4, 99% of zirconium and niobium chlorides sorbed were desorbed after %0 minutes reaction at 480°C. More than 70% of ruthenium chlorides sorbed were retained on the bed. Cerium chlorides sorbed were nearly 100%-retained on the bed without desorption. 5 L L L ® 60% Ar-40% CCl, vapor g, O 90% Ar-10% CCl, vapor / 3*— - 1~ /O - —d—L—eb—eb / | | 1 + 1 (%) 1 450 W70 490 510 530 550 Temperature (°C) 5., Effect of Bed Temperature on Desorption of Uranium Chlorides in Mixed Gas Stream of Ar-CClu Vapor, Percent of uranium desorbed 251 Table 4. Behavior of Fission Products through Sorption-Desorption Process 141 144 Temp., Time 95Zr-95Nb 1OjRn &106Ru-106Rh qj?Cs Ce & Ce RtR® RsR = RtR RsR~ RtR RsR = RtR RsR o) min) ) @ & @ @ _® ® @ Not desorbed 100 100 60 100 — — 8 100 440 15 12 12 56 93 53 - 7.5 94 Lo %0 2.3 2.3 k2 70 53 - 5.3 66 4o 60 2.5 2.5 57 95 70 — 7.8 98 460 15 7.6 7.6 63 105 75 —_ 7.9 99 460 60 0.5 0.5 4o 67 59 — 6.6 83 480 60 0.2 0.2 b6 77 58 — 13 163 * Retention ratio **ReR : Residual ratio Although data for the volatilization of 137¢s through the oxidation- reduction cyclic process were not obtained, the cesium chloride formed by chlorination was almost completely volatilized and sorbed on the BaClp bed, and retained on the bed without desorption, . . . . 1t was confirmed from these results that zirconium and niobium chlorides were almost perfectly desorbed, and ruthenium, cesium and cerium chlorides were almost perfectly retained on the BaCl, bed when the bed was maintained at the temperatures of L460° to 480°C in the mixed gas stream of argon and CCly vapor. The uranium chlorides sorbed were not desorbed through these steps in the process. Processing of Irradiated Uranium Dioxide A flowsheet proposed on the basis of the experimental results is reproduced as Fig. 6. Irradiated UQp pellets are pulverized through the oxidation-reduction cyclic process and chlorinated by the mixed gas of 60% Ar-40% CCly. The vapor of the chlorides formed is sorbed on the BaCly bed., After the high vapor pressure chlorides of fission products are preferentially desorbed by heating the bed to tempera- tures of 460° to LBO®C in the mixed gas stream of 90% Ar-10% CCly vapor, the uranium chlorides are desorbed as the vapor of UCTg5 and UClg by heating the bed to temperatures of 500° to 550°C in the mixed gas stream of 15% Ar-70% Cl,-15% CCly vapor, 252 —— e e — - | L cyclic process i 1| 1 | Uo, powder}~— Oxidation-reductionl i B 1 Noble gas,IjRu)etc. ] Nuclides volatilized | 1 | Chlorination an l‘t UM | | d digfi@};gtion] t . | e e ™ - - m——_—— = - = Waste Nuclides not volatilized T (Ba:La:Ce:etc.)01x BaCl, bed UC1x Nuclides sorbed on ! (Zr:Nb:Ce:etc.)Clx Nuclides passed through bed (2riNbletc.)Cl uu Desorption of _? (Zr:Nb:etc.)01x| {1 yr----=- —— o~ = -] Formation of UCl. and/or Flow of uranium UCly and desorption T Trapping of UClg| [Nuclides desorbed and/or uc, (Zr',Nb',etc.)Clx Il L.L, i [axidation to UQé] | | [3ecovery of UOg_J = Flow of fission products Desorption on Ba.Cl2 Bed. 253 « Fission products 6. Flowsheet of Reprocessing of U0, Fuel containing Sorption- About 1x10'8 n/cm2 irradiated and about 400 days decayed UO pellets were processed according to the above mentioned scheme. Longer decay time was adopted to study the behavior of 137¢s . The effect of temperature on the preferential desorption of the sorbed chlorides of fission products was investigated and is shown in Fig. 7. The decontamination factor of Y-activity was gradually increased as the temperature was raised from 350° to 460°C, and then saturated, The zirconium and niobium chlorides which remained after the preferential desorption were treated by the mixed gas stream of Ar-Cl12-CCly vapor, desorbed and trapped with UCl5 and UClg. The cesium chloride was also desorbed and trapped with the uranium chliorides. The ruthenium and cerium chlorides were almost com- pletely retained on the BaCly. These results are shown in Fig. 8. The uranium recovéry was not affected by the desorption temperature ranging from 350 to 480°C, The main Y-emitting fission product remaining in the uranium was 137¢s, Improvement of the decontamination was attained by repeating the sorption-desorption process as given in Table 5. Corrected decontamination factor in Table 5 means the decontamination factor that was corrected by subtracting vy-activity of natural uranium from Y-activity of untreated samples and that of recovered uranium. When the sorption-desorption process was repeated twice and three times, 103 Y T T Y T T T 10°H /o_______o—-—-—o; - o) /0 = \ Decontamination factor (Y) i 1 | i 1 1 Il 340 360 380 400 420 440 460 480 Temperature (°C) 7« Effect of Preferential Desorption Temperature of Chlorides of Fission Products on Decontamination Factor of Uranium. 254 q46¢ Relative intensity - O-F - O 10 10 retained on bed Uranium recovereq Fig. 8. Distribution Y- ray energy {(MeV) of Y-emitting Nuclides after Sorption-Desorption on BaCl2 the corrected decontamination factor was increased from the value of 4.7x103 to that of 6.3x103 and infinity, respectively, The infinity means that the Y-activity of recovered uranium was less than the VY- activity of natural uranium. This phenomenon was caused by the fact that daughters of 235 and 238y were decontaminated and not yet arrived at radioactive equilibrium, Table S. Recovery and Decontamination Factor (Y) ef Uranium Recovered after Repeating Sorption-Desorption by BaCl, Bed Once, Twice or Three Times Chlorination- sorption Chlorination temp. (°C) 580 580 580 580 580 500- 500~ 500- 500- 500- Sorption temp. (°C) 100 100 100 100 100 Time (min) 10 10 10 10 10 Desorption of chlorides of fission products rirst TemPe (°C) 460 460 460 460 460 r Pime (min) 60 240 60 240 60 Temp. (°C) L60 460 460 Second myo 0 (min) 60 60 60 . Temp. (°C) 460 Third Time (min) 60 Desorption of uranium chlorides Firat Tompe (°C) 500 500 500 500 500 Time (min) 20 20 20 20 20 Temp. (°C) 500 500 500 Second m; e (min) 20 20 20 Temp. (°C) S00 Third nine (min) 20 Uranium recovery (%) 97.6 97.7 95.8 95.7 93.9 DF(Y) 1.0x10% 7.9x10% 5.7%x10° 4.2x10° 6.5x10" Corrected DF(Y) 1.1x102 8.1x102 6.3!103 4.7x103 o0 256 '03Ru, ]06Ru-106Rh, 141¢ce and ]44Ce were almost perfectly decon- taminated through the first sorption-desorption process, and 357Zr- 95Nb and !137¢s were decontaminated through the second and third sorption-desorption process. The decontamination factor of ¥- activity of uranium recovered were of the order of 102, 103 and 104 after repeating the sorption-desorption process once, twice and three times, respectively. The uranium recovered was about 98, 96 and 94% after repeating the process once, twice and three times, respectively. Discussion The equilibrium state diagraTsyf UCly~BaClg system has been determined by Kuroda and Suzuki . In this system, the double salt, Bapuclg (UC1y-2BaCip), is formed and the peritectic reaction temperature is 583°C. The composition and the temperature of the eutectic point of UCIL-BasUClg system are 58 mol % UCHy and 434°C, respectively, The vapor of uranium chlorides formed by chlorination in the present experiment was passed through a BaClz bed with a tempera- ture range from about 5009 to 100°C and cooled down. Higher uranium chlorides are relatively unstable when the temperature is lower than about 400°C and no free chlorine gas exists in the atmosphere. They are decomposed into UCly and Cl,, and the UCly formed is sorbed on the bed and BagUClg is formed. When the bed is heated in the gas stream loaded with chlorine gas, UCI4 in the BajUClg reacts with chlorine gas and UCl5 and/or UClg are formed and desorbed from the bed as vapor, The reason why uranium chlorides sorbed on the BaCl2 bed are desorbed in the mixed gas of Ar-CCly vapor at temperatures above 4909C is explained as follows: The change, AF, in standard free energy of formation of CCly vapor by the reaction C(s) + 2C1(g) = cCly(g) (3) is given by the equation AF = =26,260 - 5.16TlogT + 49.32T cal/mol of CC14(9)(A) where T means an absolute temperature, When T = 762K, 489°%, AF = 0 in Eq. (4). Therefore, the quantity of chlorine gas formed by decomposition of CCly vapor increases rapidly at temperatures above 489°C. The lowest desorption temper- ature of uranium chlorides in the mixed gas stream of Ar-CCly vapor is 490°¢C and is nearly equal to the temperature 489°C. As the con- centration of chlorine in the mixed gas increases at temperatures 257 above 490°c, lower uranium chlorides are expected to be chlorinated to higher chlorides, and then desorbed from the bed. Zirconium, niobium, ruthenium and cerium which form chlorides of higher or lower vapor pressure than that of UClL are easily separated from UCly, but cesium which forms a chloride whose vapor pressure does not differ greatly from that of UCly is rather difficult to separate. The vapor of CsCl combines with UCly on the BaCly bed and forms a double salt. This is stable in the atmosphere containing no free chlorine, and so CsCl is not desorbed in the mixed gas stream of Ar-CCly vapor., The salt is unstable in the atmosphere containing free chlorine, and CsCl is desorbed in the mixed gas stream of Ar-Cla- CCly vapor. Thus, CsCl is not desorbed with the chlorides of zircon- ium and niobium, And the double salt is decomposed through the desorption process of uranium chlorides and CsCl is desorbed with higher uranium chlorides. Conclusions On the basis of the experimental results and the discussion mentioned above, following conclusions were obtained. (1) By applying a sorption-desorption process with a BaCl, sorption bed, the decontamination of fission products from uranium was greatly improved. (2) The optimum desorption conditions of the chlorides of the fission products with high vapor pressure and uranium chlorides were at the temperatures of about 460° to 480C in the mixed gas stream of argon and CCly vapor, and at the temperatures of about 500° to 550°C in the mixed gas stream of argon, chlorine and CCly vapor, respectively, (3) The decontamination factor of Y-activity of uranium recovered after repeating the sorption-desorption process once, twice and three times were of the order of 102, 103 and 10 , respectively. 258 1. 2. 3 7 9. References Gens, T. A., 'Chloride Volatility Experimental Studies: The Reaction of U30 with Carbon Tetrachloride and Mixtures of Carbon Tetrachloride”and Chlorine', USAEC Report ORNL-TM-1258, Oak Ridge National Laboratory, August 1965. Naumann, D., "Laborstudie zur Chlorierenden Aufarbeitung Neutronenbestrahlter Urankernbrennstoffe, 1. Mitteilung", Kernenergie, Vol. 5, No. 2, February 1962, pp. 118-119. Schmets, J., Broothaerts, J., Camozzo, G, Coenen, ¥., Francesconi, A., Haegeman, M,, Harnie, R., Heremans, R., Lambiet, C., Leur, A,, Pierini, G., Speeckaert, P, and Vanderateene, J., "Retraitement des Combustibles Irradies par Volatilisation, Report EURAEC No. 998 prepared by CEN-Mol, Belgium, August 1965, Ishihara, T., Hirano, K., and Honda, T., "Processing of Uranium Dioxide Fuel by Chloride Fractional Distillation", J. Atomic Energy Soc. Japan, Vol. 4, No. 4, April 1962, pp. 231-239. Ishihara, T. and Hirano, K., "Processing of Irradiated Uranium Dioxide Fuel by Chlorination-distillation", J. Atomic Energy Soc. Japan, Vol. 5, No. 7, July 1963, pp. S49-554. Ishihara, T. and Hirano, K., "Chlorination-Distillation Processing of Irradiated Uranium Dioxide and Uranium Dicarbide', Proceedings of the Third International Conference on the Peaceful Uses of Atomic Energy, United Nations, New York, 1965, Vol. 10, pp. 530-537. Katz, J. J. and Rabinowitch, E., "The Chemistry of Uranium: The Element, Its Binary and Related Compounds", dover Publications, Inc., New York, 1961, p. 493, Kuroda, T. and Suzuki, T., "The Equilibrium State Diagram of UCl -NaCl, UC1, -KCl, UCl -CaCl and UCl, -BaCl., Systems', Denklkagaku (E&ectrochem1stry and Indus%rlal hyslcal Chemistry, Japan), Vol. 26, No. 9, September 1958, pp. 416-418, Kubaschewski, O. and Evans, E. Ll., "Metallurgical Thermochemistry", Pergamon Press Co., London, 1956, p. 331. 259 DIRECT CHLORINATICN VOLATILITY PROCLSSING ar ot CF NUCIEAR FUELS - LABCRATORY STUDIES AV, Hariharan, S.P. Sood, Rajendrs Prassd, D.D. Sood, K, Rengan, P.V. Balakrishnan and M.V. Ramanish Radiochemistry Division Bhabhe Atomic Resesarch Centre Trombay, Bombay-85 India Abstract Direct chlorinetion volatility precessing schemes that are applicable to the reprocessing of a variety of thorium~based ard uranium-based nuclesr reactar fuel systems are discussed, Laboratory investigations have centered on the application of such a non-aqueous reprocessing method for the processing of ThOZ-an and Uoz“P“OZ type reactar fuel meterlals. The oxidic fuel compositions are converted to chlorides by reactim at eleveted temperstures with chlorine gas saturated with carbon tetrachloride, Separation of fuel, fertile samd fission product chlarides is achieved by selective volatilisation and uti- lising the high volatility of higher wranium chlorides, UCls and UCI?f The use of hested refractary alumina filter beds to facilitate condensation and filtration of intermediate voletility chlarides in the separation process 14 described., An important aspect of the process flow sheet is the use of heated NaCl bed to selectively and quantitatively sbsorbd volatile uranium chlorides, Such a medium permits a variety of gas-solid reactions in the secondary purification of the uranium fraction, 261 Introduction Non-~quecus methods for reprocessing of irradiated nuclesr fuels are under development in various lsborstarlies as economic alter- natives to conventionel aqueous reprocessing methods, Of these, flwride volstility processes are in advenced stage uf develvpment for appl{qation to a variety of ursnium-based fuels used in power resctors{~3/, An important category of nuclesr fuels i1s that incorporating thorium either as metal, c%i%e‘fig,carbm In threrme} or fast breeder reactars operating on Th 32,033 fuel cycle., Re~ processing of such fuels by fluoride volatility methods is not feasible because thorium fluoride is relutively non-volatile end its sepsrvtion from other nen-volatile fission product fluorides ip difficult, Other pyrocnemical methods also have not been extended for application to thorium-based fuels, Methods besed on the volatility cheracteristics of ursnium chlorides have been explored triefly for their spplicatim to fuel reprocessing. Laboratory investigstions on chlarination pro- cessing of uranium diaxlde, \(xranisnn carbide snd Wo-Pud, fuel materials have been reported L-11 Chlarination processes for the recovery of urani\& and, thorium from graphite-bssed fuels hsve been investigsted 3) These studtes have made uwse of e voletility of urenium tetrachloride, thorium tetrachloride and to g8 limited extent volatility of urani Smcacm.oride to effzgs. the desired separation., Gens 6:11), Good\7?) and Warren Ferris have studied the volatility of plutonium tetrachloride for the recovery of plutonium in chlcaride volatility processes., These investi- gations indicate that, inspite of the fast kinetics for the con- version of plutonium trichloride to plutonium tetrachloride vapour, the recovery of plutonium is difficult because of the large quantity of chlorine required for the process, A complete fuel reprocessing scheme explolting the volatility cheracteristics of higher chlorides of uranium has not been reported. Chlorinstion and hydrochlorination are employed as a head-end step for the removal of cladding and structural materisls from many of the wanium-based fuel systems prior to the applicati.fl i§ fluoride volstility reprocessing to such fuel msterials\idsid) In "Zircex Process" for reprocessing zircsloy clsd UC, fuels, a hydrochlarination step is used {gs‘ de¢ladding prior to reprocessing of the fuel by aqueous method sl « ouch headwend treatments maeke use of the high volatility of chlorides of aluminium, zlrconium end iron, at operating temperatures of practical interest. Literature data on the stability and voletility of the chlarideb of thorium, wsanium, protactinium, plutonium and fission products indicated thet it should be possible to work out a chlor:l.nat.&a volatility process for thorium-based and ursnium-based fuels ). 262 Feanibillty of such a.process was demonstrated Ly the present laboratory scale investigations on Th0,~U0, and U0,-Pu0, type fuel meterisls, This paper describes, in brief, the results of the experimental work carried out for the study. Hvest gation and the results are reported in moare detall elsewhere Chldarinstion volatility processing methode would be amenable to. short covled resctor fuels with tonsequent—reduction in in-plent fissile material inventory, es the processing reagents have low susceptibilities to deleterious radiation damage. Criticality pro- blems would also be less in the process owing to The ab¥ence of neutron modersting envircrmants,A chlarination volatility mrocess, in addition, would have the unique adventage of utilizing the same process media-chlorine containing gas « for the head-end and re- mocessing stages, Process Chemistry Actinide elements, rare earths and slkaline earth metals form the most stable axides known, The free energy of farmation et 298%K far a these cfl.des ranges between =120 and =140 kcael per grem atiom axygen\17;83=23), {ery strong ahlorimsting sgemts tike csrbon dioxide-carbon tetrachlaride or ¢hlorins - carbon tetrachloride gas mixtures are required for the conversion of these oxides to chlerie des, Also s temperature of 500 ~ 600°C 1is necesssry for chlori- netion of ThO,, UUz and Puly at reasonsble rates, Under such condi. tions it is expected that majority of the fission products present in Jrradiated fuel would slso be converted to chlorides, As the scheme visualises the use of voletility characteristics of higher chlorides of uranium for the separation process, it is necessary that the chlarinating gas be oxidising in nsture, This is achieved by heving chlorine as one of the constituents of the chlorineting gas, Under swch conditions the highest chlarides of the various constituents, stable at chlarinating temperatures, would be expected to be present, Large differences in the volatility of the vari. ous chlorides formed can be the basis for a separstion procedure, Boiling pointe and vapor pressures at 700°C, 500°C end 399%25‘:5 some of the pertinent chlorides ere presented in Table I 5) It is seen from the data in Tsble I that major fission products may be divided into three distinct groups as regards the volatllitly of the chlorides compared to U015-U316- 1, High volatility fission product chlorides s Zr(Cl,, M0015, NbCls’ In013. 2. Intermediste volstility fission product chlarides : CsCl, RuGlB. 3. Non~volatile fission product chlorides : Rare earth chlorides B8C12, SrClzo 263 Table I, Vapor Pressure of Chlorides Vapor Pressure in mm Hg Chloride Bnnn}%fi:filmtim 700°¢ 500°¢C 300°C °c The1, 922 0.8 belx10™2 5,157 PaCl, 527 > %0 520 Leo U1, 527 > 760 510 h.b o1 277 >HO >0 > 760 PuCl, 1730 1131077 2.0x1078 7.9x1075 2rC1; 331 >0 >0 225 NGCL, 247 >0 >%0 > 760 MoCl, 268 S0 >0 > 760 Incl 498 SO >7%0 0.11 CsC1 1300 0.70 2,0x1073 6, 51078 RuCl, decomposes 0.30 3,00 30" RRCL, decomposes 2.0x10™% 3.’7::10"9, 2,5x10"17 SrC1, 2027 3,520 2.5x10° " 2.3a018 BaCl, 2100 1,5¢1070 1.6x1077 1.@x107% ceCl, 1730 107 6.ax1077 1.ox10710 6401, 1580 1.7%107° 1.5q077 g.310™" Among the sctinide chlorides of interest, PaClc is very volstile ThClI* has intcrmediatc velatility and PuCl, cah be c¢lassified as non-volatile. The date indicate that wrehium can be separated from 2 large number of fission products end from thorium in thorium- based fusls and from plutonium in uranium-besed fucls, The gseparation can be achieved either by sirple distilletion procedures or by the use of inert, high surface ares, packed bed filters at high temperatures, which cen effectively retsin chloridcs of lower volatility froam the vapor stream while allowing the highly volatile chlorides to pass through, The latter method, with refractory alumine as filter medium, wss used in the precent investigetion, By a proper selecticn of filter beds at tempersture gradient, it would be possitle to separate ThCl product chlarides, and UCl Further, since’no f tility cheracteristics very similer to ThCl chlarides, 8 from me jority of the fission from all but the most volatile sion product chleorides have volaw it would be possible H to get essentially pure ThCl,, by the use o% the se methods, in 8 264 gecandary purification, The possibility of a secondary purificetion of uwrsnium chlorides by such simple physical methods is not indicated, because a number of flssion product dhlarides and PaCl. have similar volatility characteristics. Decrease in volatility of some of these chlorides, either by reduction to lower chlorides or by formation of stable binary compound® in some chemicelly ective sorption beds, could be a possible separetion procedure, In the present investigation it was found that chlorides of uranilem seect with sodium chloride to form some stable binary compounds Uranium present in the form of UC1 UUJ-Q vapors in the chlorinsting ges stream was quanti- tatively reg- ined in a sodium chloride bed at 250-300°C, as & sharp orange rad band, At lower (100 to 200°6)} and higher (350 to 450°C) temperatures of sodium chloride bed, an increasing smount of uwranium lesked out of the bed. The valency of uranium in this orange red compound was found to be five by chemical analysis 1s reported to form compounds of the type CsUClg 6); ébably the sorption of uranium chloride proceeds through the formeti onof NaUClg. In another series of experiments where chlorination of U.On in presence of NaCl wes being tried it was observed that no vozaeue species of ursnium are formed upto $00°C inspite of the trighly chlorinating comdition., Uranium formed a green compound with sodium chloride snd remeined in the reaction zone. This compound must be UCl -2Na01 as rfggfted in the urenium tetrachloride - sodium chloride p ase diagram Appa-~ rently the activity of UCL, in this compound is very low even at temperstures as high as 600YC, It was decided to try to use these data for separation of uranium from fission mroducts having volstile chlarides, Chlorination of U02...P\102 fuels is compliceted by the reaction indicated by equation®(1) PuCl (s) + 3 c1, (g) PuC1 (g) (1) Accordi.ng to Benz(za) the equilibrium constant for this reaction is given by equation (2) log k = 6,18 - (670 - 1050°%K) (2) The value of the equilibrium constant is apmroximately 5 x 10=° at 500°C. This means that 0,15 mg plutonium will be carried as PuCl, by 1 mole of chlorine, It has been found that the actual amou%t of plutonium carried while chlorinatin% us‘anium oxide - plutonium oxide mixtures is more than this value It is difficult to recover this part of plutonium and also it contaminates the urenium fraction. To overcome this difficulty it was decided to carry out the chlorinstion experiments et much lo- wer temperature where PuClL is less stable, It is not possible 265 to chlorinate O, at satisfactary _reteswbe_lo(jg 5. however U30 can be eas chlorinated even at 350° ’ . As UO -Ptflz sa]_?d solutions having less than 20 wt pet PuD, can be eaai?y axidised to U308-Pu02 mixtures, it was decide? S.o work out & separatdon procedwe using U308—P102 mixtures}8 Experimental Procedure Materisls Simulated ThOZ-ID fuel was prepared from coprecipitated Th(IV) - U(IV% axeletes, Qxslates were ealcined to selid solution oxides at 900°C and then pelletised and sintered in hydrogen st 1650° to 17009C. The pellets comtained 7.46 wt pot 0,. The pellsts were crushed to powder and approximstely one gram sample used for each experiment. U0, was prepared by oxidising sintered U0, pellets at 450°C in air. fiuoz was obtained by calcination of plutonium(IV) axslate at 500°C in"air. Aluminium oxide, which wes used in this work as high temperature packed-bed filter material, was granular 60 mesh type-RR refractory fused Al20.. Sodium chloride which was used a8 sorption bed for wsanium chlo- rides was of snalytical grade sleved to + 52 mesh, To study fission produgt behaviour, & sypical representative mixture of fission product elements(immctive) wes prepared from pure constituents either as oxide or chlorides. Irr_al%iation of Th0..-U0, and U308 was carried out at a flux of 5 x 10 n/sec/cm toff thhows. Apparatus The apparatus consisted of a 35 mm Vycoar resction tube, 90 cm long and was heated by a four-zone electric furnace., A demountable sleeve tube febricated out of vycar and positloned suitably in down flow line of the mein reaction tube conteined the filter beds and the sorption bed. The arrangement permitted the isolation, ard control of temperature, of the individual zones, A gas handling manifold and a gas disposal set up completed the experimental equipment. A schematic drawing of chlorination appsratus is shown in Figure 1. For experiments with U,Ug-PuDp only a three zone equipment was necessary. Alsc the ehtire sssembly, except the gaa handling system, wes kept in a glove box. Procedwre The apparatus was assembled with alumine filter beds and sodiunm chlaride sorption bed in the proper heating zones and then the sample loeded into the system, The equipment was flushed with No-Cl, stream at 600°C after which the temperatures of the various zones were ad justed to decired values, Chlorination was then 266 L9¢ THERMOCOUPLE SAMPLE NN 747 NN A|203 AI203 NaCl BED-I BED-II BED-II __‘.._1'._'__0 THERMOCOUPLE 2, / Ci, Ny fl ‘ CC|4 ........ ' W/// AT FJRNACE GAS HANDLING 1, Schematic Apparstus for Chlorination of ThU,-U0, OFF - GAS _[j DISPOSAL - started by passing a mixture of C -00th over the sempls, Normal chlorination time was four hours, at the end of vhich the epparatus wes cooled while pessing a stream of nitrogen, Various zones were then analysed to determine degree of chlarination amd separation factors. In experiments incorporating simulated fission product meterials and those with irradiated solids, the fission product snelysis was carried out by radiochemical seperations and gamma spectrometry, Results and Discussion ‘l‘hOZ-UOZ Process A 012-001 0% > 10% Ba 12610° >10° >10° >10% Ru 1.3 >10° 20 - P L 2 Zr 610 >10 3.8 3.2510 Nb 2,10% >10% 50 >10° Pa 2x10% >0k 1.5 51 U 25 >103 - - Th _ - 5x103 5x10°3 Seéondary purification procedures for urapium fraction aimed at the formstion of UCL; .2NaCl from UCl;.NaCl. In one method, this was done by passing hydrogen gas ag 500-600°C, Further chloy rination of this bed with C15-CC1,-N, at 600°C resulted in the removal of essentially e11 ZrCl éndzpacl , @nd the remeining NbCls end %015. Purified urankum was leg% behird as green UGlh.ZNaCJ.. In the second method UCl§ in the NaCl bed was converted to uranium oxide by pyrohydrolyeis at’300°C end further axidation at 500°C: The oxidised materisl was-rechlorinated using-the-ehlow rinating gas at 500°C for 3 hours, Uranium oxide got converted to UCl, «2NaCl whije ZrC]'I,,’ PaCl_, MoCl, and NbCl, were volatilized eway from the reactlon %ohe, e overéll ganms decontamination factor waa 34, This is because of the 2-5% protsctinium left in the purified uranium. UDecontemination Factors after thHe secondar, purification are also listed in Table II, The conceptusl flow sheet for reprocessing ThOo-UOC, type fuel materials by direct chlorination voletility processing is given in Figure 2. Adaptation of the scheme for other thorium based fuels such as Th-U alloys, ThC,-UC, and dispersion of axide or carbide graphite sppcar practiceble., Th-U metallic fuels would react wi hydrogen chloride during decladding to form ThCl, and UClg which ere practieally non-volatile at temperstures t.ha% are commonly employed for hydroechlorination, With ssrbide type fuels, a pre- liminary axidation step to remove mstrix materials may be advantageous, U308-..Pu.02 Processing Preliminsry experiments on the chilorination of U30 -rud, adopted the optimised reaction conditions developed for U OB(QB). in these, chlorinstion, filtration amd sorption zoneg were Kepb at b,50°C, 300%C and 300°C respectively. Pure P\.flz powder did not, get chlorinated at 450°C. However in U0y sempleS having 0.08 wt pet PuC, essentially all plutonium got chlorinsted and the volatile pluton%um species distributed betieen the Al,U, filter bed and NaCl sorption bed. At & chlorination tarpera%u?e of 350°C, the amount of Pul, chlorinsted and transpired along with urenium chlorides could be brought down appreclably, AlpOz filter bed et 200°C was found to effectively decompose Pull; vapor to PuCl, solid. There was a slight retention of uranium ir the filter bed 8t this temperatwe but the decontamination factor or Pu in uranium had very much improved and it was declded to stendardise at this temperature. In experiments with U,0g = 1 wt pct Pub,, U0 was campletely chlorinated in four hou?s and passed onto Ngc?i sorption bed. Only 5% Pul, (on 10 mg basis) was converted to PuCly and reacted with chlorine to form volatile Puclh, which was efféctlvely retained in A1,0, filter bed, Only 4 to 5 mg out of 10 mg of plutonium transpgred along with wranium chloride 270 1.2 IRRADIATED FUEL TllOz- UO: CLADIN AlCly FeCly OR ZrCiy DISCARD CHLORMATOR | FrissI08 2e Flow Sheet for Reprocessing ThO,-UD, Fuel Materisls Dacls zeiy j —_—— ICH, Mty WeCly VOLATILE tCle Cl!- N‘ ccly I Ha2 ]0" ["‘a“ ] = l l UCHg-UCig , TRC e oommaron voLarie ¢ | FiTErBEDS oM | ucy,-ua,, Pacy NaCl Ucly Mec! | REDUCTION CHLORNATOR b HLORINAT ABSORBER HLORIN. FISSION PRODUCTS | TEMP GRADIENT VOLATILE FISSION 8ED PaCly, Zrd, | pyRoNYDROLYSIS Patly PRODUCTS ] 2 > [~ 3 ThClg CRUDE OISTILLATION PYRGHYDROLYSIS y NON VOLATILE FP RESIDUE ThCle ROMYOROLY SIS ThO, l prooucr —"1°" -t 2 e REFABRICATION Jo, and was collected in NaCl bed. An overall separation factor of 2,000 for plutonium in wrenium was achieved, Chlorination volatility processing of mixtures containing 1 to 17 wt pet PuO in U0 .-PuO was studied. It was found that with increasing percegtage o? Pu0y, the amount of PW, converted to PuC13 and treanspired onto Al O, bed as PuCl, decreased from L wb pect to 035 wt pet. However separa ion factors for plutonium in wenium did not improve, In experiments with irradiated UBOS’ mixed with Pul,; and inactive fissicn products materisl, the behaviour of fission products was similsar to thst in ThO, -06 experiments, All fission productes except N‘b Mo and 2r stayed in the sample boat with plutonium, NbCir, ZrCl accompenied UCL onto NaCl sorption bed, The grosg gamna deconteminat.ion ee&or after primary purification was econdary purification of wranium, as describd earlier, imp: oved . the gamma decontamination factor to A70. The decontemination factors from individusl fission products after primary end secondary pwification are 1isted in Teble III, Recovery snd purification of plutonium fraction by chlorination volatility was not tried, It is suggested that fluoride volatility mrocedures be used for the recovery and purification of plutonium residues, This step would be easier and practicel in view of the fact that Pul, residues do not contein any significant quantitles of wanium and fission products that form voletile fluorides, The conceptual flow sheet for the direct chlorination volatility processing of the low emrichment U0 -PuD type fuels is presented in Figure 3. The scheme may not be appficable to fuels with more then 20 wt pet Pu0, because they cannot be oxidised to U308-Pu02 mixtures, Table III., Decontamination Fector for Uranium in U,0q-Pu0, Processing Activity ~ Primary Separstion peciiorevion Gross Gamma 6.7 4. 710° Gross Beta - 2‘3"102 7 1, 2.1x103 o 1.1x10° 5.1x10° Ru 2.7 69 Pu Z.OX_'LO3 2.0x103 272 £L? VOLATILE WASTE AlCly FoCly ‘ OR ZrCiy Zrcia DISCARD i FILTER BED ICI,NbCIs MoCIs VOLATILE F P CCla Cly- N, HYDRO CHLORINATOR T JRRADIATED | FUEL U0, ~Pul, CCly Hy OR Hy0 CLAD IN m OXIDATION C-N Al,5S OR 2" 2 ZIRCALOY (CHOPPED) OXIDATIVE UClg=UClg | FuTER BEDS ON | UC-U Nacl UCig-Nacl | REDUCTION CHLORINATOR A ABSORBER OR CHLORINATOR m DECLADDING [U10,, PuO, VOLATILE TEMP GRADIENT VOLATILE BED PYROHYDROLYSIS - F P F.p. FF 5 2 [} L 4 S NON VOLATILE 3 ~ ss F P,Pu0,,PuCly ZIRCALOY PYROHYDROLYSIS WASTE l OISCARD FLUORINAT ION PuF L PYROMYDROLYSIS|— Pu0, |—] REFABRICATION vo, 3. Flow “heet for Reprocessing UUy-Pul, Fuel Materials Decladding of UOp-based fuels ca? ejther be done by hydrochlori. nation or by "oxidative decladding® « In the latter method slotted fuel elements are hested in air at 4L50°C. As U02 gets oxi~ dised to U,Oq, it expands, and cracks the clsdding tube along the slots, aS-PuOZ powders are released which can be sent to the chlorinat rventhough the flow sheet is shown for low errichment W -Pqu fuels, application to reactor fuels of the type U-Pu alloy and wanium carblde appears possible, Conclusion It is seen from Tables II and III that reasonsble decontamination factors can be obtained by chlorination volatility processing of irrsdiated fuels. It is expected that the deconteminstion factors will improve in scaled=up experiments, It is suggeit that fluie dised bed techniques be used for the entire process o Uslng a colum of refractory alumina particles, a number of gas-solid reactions of the type envisaged in the chlorination volatility pro- cess can be carried out with proper process control. Such tech- niques can be easily adapted in the case of UO,-Puw, fuels, as the technology for the moderste temperstures involwved is already existing, However in the case of ThO -UO type fuels, container and process materiel development woul be necessary before the scheme can be put into practice. Achww;l_edganent The suthars are grateful to Mr N, Srinivesan, Head, Fuel Reprocessing Division, Bhabha Atomic Research Centre for his interest in this work,. References 1, Steunenberg, R.K. and R,C, Vogel, "Fluoride and Other Volatility Processes", Rgactor Handbook II, Fuel Reprocessing, Interscience Inc., New York, 1961, pp. 250-307. 2, Vogel, R.,C,, Carr, WH,, Cathers, G,I., Fisbher, J,, Hatch, L.P,, Horton, R.W,, Jonke, A,A,, Milford, R.P,, Rellly, J.J. amd G. Strickland, "Fluoride Volatility Processes For the Recovery of Fissiocnable Materisl From Irradisted Reactor Fuels", A.Conf.28/1, Vol. 10, 1964, pp. 491-500. 3. Jonke, A.A,, "Reprocesaing of Nucleer Reactor Fusls by Process Based on Volatilizstion, Fractional Distillation snd Selective Adsorption", Atcmic Energy Review, Vol. 3, No.l, 1965, pp. 3-60. 274 Le 56 15. Speecksert, Ph., "A Msthod For Processing Irradisted Uranium and Uranium Compounds by Fractionel Sublimetion of Their Chlorides", Chemical Engineering Progress Symposium Series, Vol. 60, No.A7, 1964, pp. LB-55. Ishihare, T, and K, Hirano, "Chlorinstion Distillation Processing of Irrediated .Ursnium Dicxide and Uranium Dicarbide', Vol. ]-0, l96h’ pp‘ 530'537' Gens, T., "Chloride Volatility Processing of Nuclear Fuelsh, e Symposium Series, Vol., 50, No.L7, 1954, pp. 37-47. Goode, JH,, "A Laboratory Study of Separation of and Recovery of Urenium and Plutonium From Fission Products by Chloride Volatility", USAEC Report CANL-TM-828, 1964. Warren, K,S, and LM, Ferris, "Uxidation and Chlorination of D0,-PuO,", USAEC Report C(RNL-3977, 1966. Neumsn, D., "Lsborstudie zur Chlorierenden Aufarbecitung Neutronenbestrahlter Urenkernbrennstoffe, Plutoniumbtrennung" Kernergie, Vol. 6, No.3, 1963, pp. 116.21. Schmets, J., Csmozzo, G., Francesconi, A,, Godrie, P., Heremsns, R,, Plerini, G, and P. Speeckaert, "Hetraitement de Combutibles Nucleaires par Voletilisation®, A/Conf,28/1, Vol. 10, 1964, pp. 50-529. Gens, T.A,, "Thermodynamic Calculation Relating to Chloride Vols. tility Processing of Nuclear Fuels, II. The Capacity of Chlorine for Transferring PuCl, Vapor During Reaction of Uq0g~ Puuz with CClh", USAEC Report 3693, 1964. Bradlcy, M.L, and LM, Ferris, "Recovery of Uranium and Thorium From Graphite Fuels, I, Leborstory Development of Grind Leach Process®, USAEC Report (RNL-2761, 1960, pp. 29=-36. Cook, J.L, and R.,L, Hammer, “Remuval of Urenium and Thorium from Fuelled - Graphite Materials by Chlorination®, USAEC Repart CRANL-3586, 1964. Ramaswamy, D., Levitz, N.M,, Holmes, J,T, and A.A, Jonke, ¥Engineering Development of Fluidebed Fluoride Volatility Process, Part I, Bench-Scale Inveatigation of & Process For Zirconiumn-Uranium Alloy Fuel®, USAEC Report ANL-6829, 196L, pp. Li~21. Levitz, N,M,, Barghusen, J,J,, Holmes, J,T, and A,A, Jonke, "Halogenstion Studies on Nuclear-Fuel Element Materisls in a Two-Zone Fluid-Bed Reactor®, Chemica) Engipeering Progress Symposium Series Vol, 60, No.,7, 1964, pp. 8i=89. 275 16, 17. 19 . 20. 25. 5. 27. Blanco, R,E., "Preparestion of Power Reactor Fuels For Processing by Solvent Extraction", Progress in Nuclesr Energy Series III, Process Chemistry, Vol, 2, 1958, pp. 240-2L3. Socd, D,D., and A,V, Hariharan, “laboratory Investigstions on Direct Chlarination Volatility Processing of Nuclear Fuels, Part I. Process Flow Sheets, Thermochemical and Volstility Data on Chlorides®, BARC Report BARC-397, 1969. Sood, S,P., Balakrishnen, P.V., Rajendra Prasad and A.V, Hariharan, "Laborstory Investigation on Direct Chlarinstion Volatility Processing of Nuclear Fuels, Pert II, Chlorination of Uranium Oxides", BARC Report BARC=L0L, 1969. Sood, S.P,, Sood, D,D., Rengan, K. and AV, Hariharen, "Laboratory Investization on Direct Chlorinetion Volstilitvy Processing of Nuclear Fuels, Part III, Separation of Thorium and Uranium From ThOp-UOz and Processing Irradiated Fuel Materials", BARC Report BARC-,05, 1969. Ra jendra Presad, Rengan, K, and A,V, Hariharen, "Laboratory Investigations on Direct Chlorinstion Volatility Processing of Nuclear Fuels, Part IV, Separation of Urenium and Plutonium From an"PuOZ and Processing Irradiasted Fuel Materials", BARC Report BARC-IF%’ 19690 Glassner, Alvin,, “The Thermochemical Properties of the Oxides, Fluorides end Chlorides to 2500 %K", USAEC Report ANL-5750. Rand, M.H., and O, Kubaschewski, "The Thermodynemic Properties of Uranium Compounds", Interscience Publishers Inc., New York, 1963. Kubaschewski, O., "Plutonium Physico-Chemical Properties of Its Compounds and Alloys", Atomic Energy Review, Vol, 4, Sp. Issue 1, 1966. Kubaschewski, U, and E,LL. Evans, "Thermochemical Data', Metalluwrgical Thermochemistry, Pergmon Press, New York, 1958, pp. 286-309. Brewer, L., "The Fusion and Veporisation Dats of Halides", The Chemistry and Metallurgy of Miscellaneous Msterials, Thermodynamics, National Nuclesr Energy Serles IV, 198 McGraw- Hi1l Book Co, Inc., New York, 195C. Bagnel, K.,W,, Brown, D, and J,G.H., du Preeze, "Some Chlorouranate (V) end Chlorotungstate(V) compounds", Jourpsl of the Chemigsal Soclety, 1964, pp. 2603-2608. Barton, C.J., Sheil, K.J., Wilkerson, A.,B, end W.,R, Grimes, "The System NaC1-UC1,", Phase Diagramsof Nuclear Masterisls USAEC Repart OH.NL-25L§, 1954, pp. 134, 276 31. Benz, R., "Thermodynamics of PuCl from Transpiraetion Dats", Journal of Inorgenic¢ and Nuclear ehemiatrjz, Vol. 2, 1962 pp. 1191-1195. Korshunov, B.G., and N.W, Gregory, "Vapor Pressure of Zirconium Tetrachloride Above Sodium Chloride and Sodium Hexachlarozirconate (IV)", Inorgenic Chemistry, Vol, 3, 1964, pp. 451~k52. #Processing of Nuclesr Fuels of Low Enrichment. a, Separation of Fuel From Cladding", Chemical Engineering Division Research Highlights USAEC Report ANL-6875, 196L, pp. 84~87. Reilly, J.J., Regan, W,H,, Wirsing, B, and L.P, Hatch, "Reprocessing of Reactor Fuels by Voletilizstion Through the Use of Inert Fluidised Beda!, USALC Report BNI-663, 1961. 277 FUSED-SALT FLUORIDE-VOLATILITY PROCESS FOR RECOVERING URANIUM FROM THORIA-BASED FUEL ELEMENTS+) W. Bannasch, H. Jonas, E. Podschus Farbenfabriken Bayer AG, Leverkusen, Western Germany Abstract Laboratory studies have been made on the Fused-Sal+ Fluoride- Volatility Process (FSFVP) appiied to thorium-uranium oxide or carbide particles in a manner analogous to the application of FSFVP to metaliic fuel elements. Systematic investigation of hydrofiuori- nation and fluorination establlshed that the most favourable com= position of the fused salt was LiF-Naf~Zrf4 = 25-25-50 mol/o. It was also found that corrosion of the reactlon vessel (NI or Ni-rich alloys) decreased with increasing ZrFs in the melt. Volatilisation yletds ranged upto 99,8% even when ThF, was present as a precipi- tate. In view of this high yield i+ should be possible to apply this process to these oxide fuets provided further steps in the process do not give rise to major difficulties, +) Work performed under a project sponsored by the German Federai Ministry of Science. 279 introduction This paper deals with work performed in a joint development program carried out by several industrial firms In cooperation with the Kernforschungsanlage JUlich. The object of this work is to inve- stigate several possible methods of processing irradiated fuel elements containig thorium, especially those of the AVR and THTR reactor of the pebble bed type. The fuel element contalins the fuel in the form of coated particles uniformly distributed In a graphite matrix, with an inner fuel zone of the eiement being enclosed by a fuel-free graphite zone. The coated particles under investigation contain the fuel In the form of carbides (UC2-ThC,) or oxides (U0»=ThO5). Fundamentals . The concept under discussion deals with +?e 2T+emp+, to apply the known Fused=Salt Fluoride=Volatility Process 1=4) to thoria based THTR fuel elements. This process, in the following cal led FSFVP for simplicity, originated in the early fifties, when in the first instance it was developed by the ANL for the recovery of enriched uranium metal and uranium alloys on a laboratory scale. The introduction of homogeneous molten sait reactors(®) {ed to an intensifled further deve|lopment of this process culminating in a pilot plant at the ORNL{®*8) 1n which several types of fuels with different cladding and structural materiais could be processed successfully, A com- plete core loading, processed by a non-~aqueous procedure, was demon= strated for the firig t+ime In 1958/59 by the FSFVP at the Alrcraft- Reactor-Experiment. ) In the FSFVP the fuel material is dissolved by hydrofluorination in a fused fluoride bath for conversion from a solid to a liquid state, The resulting melt is fluorinated with fluorine to voltatilise the uranium as the hexafluoride, which is finally purified in an adsorption-desorption step by use of NaF or by distillation. The application of the FSFVP to ox{g? type fuels was demonstrated only in the case of short-cooled U02( , and some experiments have been performed to investigate the applicability to refr??Tory oxide fuels containing BeO, ThOy or Zr0Op at laboratory scale! . But nothing has been published about processing of thoria-based fuels from THTR-type reactors with high thorium contents by a FSFVP. Consequently the first problem of the applicability to oxide- type THTR-fuels consists in the dissolution step, i.e. the conversion of the oxides (UOp, Uz0g~ThOp) into the tetrafluorides in a salt melt. The proposed head end process, consisting of the combustion of the graphite balls, 1s used to generate the oxide mixture above. The chosen salt meit must satisfy certain specifications: the conversion must proceed at ftemperatures as low as possible, with an adequate 280 reaction rate, to diminish the corrosion probiem. The salt me!t must exhibit at these temperatures a sufficiently high solubllity for the tetrafluorides. Theserequirements must be investigated for every new combination of fuel=-molten salt, in order to find the most favourable conditions. Hydrof luorination I+ is known that the reaction of oxides with HF in fluoride melts is on{& possible In the presence of acidic fluorides, e.q. BeFp or ZrF4( ), This fact may be explained by metathesis. The oxyfluoride mixture obtained in this way is apparently able to react quickly with HF to form the tetrafluorides, in contrast to +he oxides: ThO, + Zrf, + ThOF2 + ZrOF2 ThOF, + ZrOF, + 4 HF + ThF, + ZrF, + 2 H,0 Since zircontum=bsaring systems showed favourable properties In relation to corrosion, Th? three component system LIF-NaF-ZrFg4, which Is already we | 1=known (13 » was chosen for the investigation. Diagram 1 shows this system leading to the following requirements: if 5009 Is desired as a reaction temperature, salt compositions near the equimolar point (LiF=-NaF~ZrFg=33=33-34 mol/o) should be chosen. Raising the temperature to the range of about 550 to 600°C permits sufficient freedom for the choice of +he compesition. The possibility of working with these high temperatures is |imited by the possible degree of corrosion, The dissolution step for Th0,/UO, particles with the molar ratio 5/1 and 20/1 has been Investigated as a function of the following parameters in the presence of HF: 1. ZrFg=content at constant temperature 2. temperature and without applyina HF as a function of 3. ZrFg=content 4, LiF-NaF motar ratio 1. Influence of the ZrFs-content Dlagram 2 shows the reactlion rate of ThO,/UOp-particles at 650°C at different Zrfg-contents in the melt+. Each single diagram shows the percentage of reacted material!l as a function of the reaction time for definite ZrFg=contents, which have been increased from 3 to 47 mol/o. All values have been obtained by analysis of small samples, which were taken at definite (the plotted) times. 1t can be stated that the reaction rate is increased with increasing ZrfFs-content in the met+t. Diagram 3 shows more clearly the reacted oxide fraction at certain 281 2rF, w2 PRIMARY PHASE AREAS TEMPERATURES IN *C COMPOSITIONS IN mele % ehttH INDICATES SOULID SOLUTION © WoF 2rrm SLIF ZeFyme SNeF 2ZrFine @ ENeF 2rE, TNeP $2rF, IF 2rF, NaF 42eR-3LIF 42efse LIF NaP &2rF, ® zrr, LiF F s £-482 e 1« The System LIF—NaF—ZrF4 o Oxide reacted o8 o Function of the Zirconiumtetrafiuoride -Cantent (Mol%) —_ 100 P ——— | / 50 50 / 0 10% 120% j upfl 0510 20 60 0510 2 60 0510 2 60 /) / / 50 soH 20 [ ok 0% || 230% o050 20 80 oS0 2 60 o510 20 60 0o 00 00 50 L) 50 / 5% 0% £70% 050 2 o 050 2 [ 0SD 20 Tmeimin) o 2. Reaction Rate of Oxides for 9 ZrF4-Con1-en1‘s 282 % Oxide reacted as a Function of the Zrf,-Content at various Reaction Times 100 l — 20min | so ! s / 'y é L | 10min F b 4 7 // 50 / / // Smin 30 7 10 0 3 7 H 19 2 315 37 {7 50 — Mol % ZrFi— 3. Reaction Rate of Oxides as a Function of ZrF, Content Oxide reacted as a Function of the Temperature and the Zrf;-Content 283 100 m— /*7650"6' .—‘-._“________._.... -1 ga -" .'_—l- / ="~ 580°C l’.’ ‘,n' - 70 / 1" s s . .’ e - S e e A e 530°C 50 '_r "-_”__..- { ‘i ....... ! X 2 N - amm e e Pa - - - SR G R SR W WP TR R W AP WP G T e GRS WP EE S j ! 436°C ) % ir’ .‘..-" /‘,{ v i LiF | NaF|ZrF, 10 K 4 I/ 34 |3 |32 0 5 10 2 —Time (min)— 60 4, Reaction Rate as a Function of the Temperature times (after 5, 10, and 20 min), plotted as a function of the Zrf,=- content. At 20 mol/o ZrF4 for example the reaction Is complete after 20 min, (Ratlo oxide/melt,10% g/q) 2. Influence of the temperature The dissclution step was Investigated only at temperatures lower than 650°C, corrosion being too heavy at higher temperatures. As can be seen from dlagram 4, at the lowest possible temperature of 450°C (ref. diagr. 1) the reaction rate is +too slow and the reaction in= complete. 600 to 650°C will be necessary for this step, in order to achieve a satisfactory rate of reaction. 3. Influence 6f ZrF4 without HF-application As mentioned earlier, Zrf4 is able to react with ThO2/U0, resulting in breakdown of the crystal lattice without application of HF, gene- rating an oxyfluoride mixture. Diagram 5 shows the results of this reaction type without use of HF. The Time needed for complete destruction of the particles was measured as a function of the Zrfg~content. The LiF-NaF molar ratlo was kept constant during this series. Hence at 50 molar % ZrF, there is a maximum rate. In order to study the influence of the LiF-NaF molar ratio, a subsequent investigation was carried out, in which this ratio was changed, keeping the ZrFag-content constant. 4, Influence of the LiF-NaF molar ratio The results of this series of tests are given in diagram 6, which makes it clear, that the fastest rate is achieved, when Lif and NaF are present in equimolar proportions, thereby fixing the most con- venient composition for the dissolution step: LiF - NaF - ZrF4 = 25 = 25 = 50 mol/o Simultaneously it was found that the following hydrofluorination of these oxyfluorides to the tetrafluorides takes less time than hydrofluorination of the oxide particles. 5. Solubility tests The next important question is the solubility of the tetrafluorides in the given melt, Tests concerning the solubility of ThFs in a melt of the above mentioned composition iIndicate that the simpler three component system NaF-ZrFs~ThFs, which has already been investigated, may be used as shown In diagram 7(14) | Even at temperatures as high as 650°C the solubility for ThF, Ts not very high. Approximately 5 molar % Thf, are soluble. 284 Time(min) Reaction without HF -Application as a Function of the Zrig-Content ot LiF: NaF=1:1 lsa 55 - \ / N / 15 L 35 40 45 50 L] 60 ———ue 85LIF/NaF ZrF, 65— 60 55 50 8 o 35 5. Reaction Rate without HF as a Function of the ZrF4 Content Time(min) Reaction without HF-Appiication as a Function of the LiF/NaF Molar Ratio at SOMol% 21y 50 50 L0 JL/_\‘ Reaction without|HF \ e \\ X/ \ ~ N S A Y VR —r"' 2 \\ ~ ~ S 10 Following Hydrofiuorination 0 0 20 o oi%— Y0 0 —LiF~ = 50 -NaF— 40 0 20 10 0 6. Reaction Rate without HF as a Function of the LiF/NaF Molar Ratio 285 NoF 2ThE, 4 TEMPERATURES ARE IN “C COMPOMITION IN male % 7. The System NaF-ZrF4-ThF4 Schematic of Flowsheet - Combined Fiuorination-Hydrofluornation-Equipment Stack R-AlD,-Reaclor HF -S'crubber | | PiD Product UF; Traps HF-Absorber ] E— B i B H K . fll\ (- Lo . Regulator F=Supply | hF-gmply | Orifice Meter iert-Gas 8. 286 Schematic Drawing of the used Fluorination Equipment Fiuorination = UF. = Volatilisation From Information published by the ORNL!'3? there should be no difficulty In votatilising the uranium with elementary fluorine if +he uranium Is dissolved homogeneously in the melt, Diagram 8 shows schematically the flowshest of the combined fluortnation=hydrofluorination-equipment, which was used in this investigation. As may be seen from the diagram, attempts were made to carry out both reactions In the same reactor as a "single vessel” reaction. General procedure For the investigation of the volatilisation rate of UFg, expressed as the percentage of uranium volatiiised as a function of the fluori=- nation time, the conditions have been standardised to enable comparison to be made: All serles of tests were performed with the same fluorination equipment, consisting of crucibles of nickel or nickel=rich alloys of the Hastelloy type, which had been Inserted into resl!stance heated furnaces. The amount of the starting melt was 100 or 150 g, which enabled a reaction to take place between 10 or 15 g Th02/U02 particles and fluorine=nitrogen mixtures, as desired, After ascertaining that the UFg-volatilisation itself did not cause any major difficulties, the scale has been enlarged to a 3 kg fluori- nator, which is capable of taking from 250 to 300 g of particles. Later, experiments in a hot cell will be performed on this scale also, Figure 9 shows a picture of the arrangement used on a labora- tory scale, The fluorination reactor is shown together with the absorption beds which were operated at 100 or LOOPC, respectively. To improve the heterogeneous reaction of the gaseous fluorine with the liquid melt it was agitated by means of a percolator tube (draft tube), This construction is shown in & agram 10, Results The fluorination reaction Is influenced by the following parame- ters: 1. geometry of the reactlion vessel and its fifttings 2. gas velocity 3. concentration of fluorine 4, presence of ThF, Using the arrangement shown, it was found that the following condi+ions will achieve good results: the ratio of length/diameter of the crucible or the furnace which contalns the melt should be not less than 4/1. In the case of a 150 g batch size this configuration 287 10. Photograph of the Draft Tube 288 postulates a gas velocity of 12 |/h. The application of pure undiluted fluorine is not necessary, since mixtures of fluorine with e.g. nitro- gen can achieve comparably good results; but as it can be seen from diagram 11, a dilution of the fluorine to less than 1/1 decreases the volatilisation rate markedly. Diagram 12 shows four typical volatillisation curves with different ThF4=contents. The presence of ThFg does not influence the votlatili= sation; but i+ Is remarkable that Thfq may be present additionallty as a precipitate without appreciably decreasing the volatilisation ylelds. Furthermore a common Induction pericd of about 40 to 60 min can be observed before any UFg 1s volatilised at all, This could not be detected in the absence of ThFg. Other Process Varilants In the experiments described so far, both single reactions "hydro= fluorination" and "fluorination" were carried out in a salt melt, Kal i=Chemie Hannover, another participant of the joint project, studied possibilities for the conversion of the oxides into the tetra= fluorides by reaction with HF in a fixed or fluldised bed reactor at 450°C, | f this step Is performed without a salt melt, the corrosion in the FSFVP can possibly be diminished. The corrosion at the liquid=- gas Interface is believed to be caused by the water generated after the reaction of HF with the oxides. Consequently there are 4 possible variants: 1. Hydrofluorination with subsequent fluorination In Zrf,- bearing melts, 2. Reaction to the oxyfluorides without hydrofluorination In ZrF4-bearing systems with subsequent fluorination. 3. Dry hydrofluorination leading to a UF4/U0pF,=ThF, mixture which can be fused with alkalifluorides (LiF-NaF=ThFg) with subsequent fluorination. (Proposed by Kali Chemle) 4. Fusing the UFy/UOxF,=ThF, mixture with a ZrFs=bsaring system with subsequent fluorination. Atl four possibilities have been investigated systematically, especially concerning the corrosive properties, Considering these variants, two main features can be pointed out: ZrFs=bearing systems exhibit, in fact, only a limited solubility for ThE4, but are on the other hand quite superior in relation to the corrosive attack on the construction material, Hydrofluorinations at 650°C in Zrfs-rich melts caused only a slight increase in the nickel content of the melt. ZrF4=-free systems allow appreciably greater amounts of ThF,, since ThFg Ts TTselt a component of the three component system LIF=-NaF-ThFfy; but [+ was observed that this system leads to a corrosion hardly 289 % UE voistilised as a function of the F3:Ny ratio 6 00 90 S A | 2 3:1 iy | VA 10 20 &£ 60 120 time (min) 180 i1, Volatilisation Rate as a Function of the Fluorine Concentration fll{m lqm W[ 11T Gd o S % = 1 ‘L“TZ ! L I_] ( | 7 | | L L P T L ] » | » [ ” ”0 n!lfl o [ ] w0 w ) ; % —I Kl‘w m{w :—l! ! ’ :_J.E { Q/T NS ol | ;/ % l . /k ey ] ol A% » F 12, Volatilisation of UF6 in the Presence of ThF4 290 tolerable In connection with Intergranular attack. Corroslon Studles During the investigations into ZrFa-free systems (variant 3) a severe corrosion was observed in contrast to the ZrFs-bearing systems (variants 1, 2 and 4). The nickel content in the melt increased so rapidly, that the viscosity became too high, thereby termlinating the fluorination before all uranium had besn volatilised. These obser— vations made it necessary to undertake a somewhat more systematic corrosion study. But restrictively it must be stated that all results reported here refer to the nickel content of the meit+ which was analysed by direct X-ray excitation, and no work could be performed using ultrasonic or vidigage techniques. As is shown in dlagram 13, corrosion is uneqivocally influenced by the ZrFg=-content ot the melt. This investigation is based on the hydrofluorination step, and shows additionally that nickel=rich alloys of the Hastelloy type exhibit a much higher resistivity to attack. Corrosion during fluorination does not make the FSFVP unworkable, but It Is necessary to use only melts with high ZrF,-contents In order to avoid excessive corrosion, Whereas the amount of dissolved nickel In ZrFs-free systems is Increased under the chosen conditions (3 h,Fy at 550°C) by a factor of 30 fo 40, corrosion Is decreased to such an extent, when Zrf, is added, that the factor will become only 1,5 0 2, proving thereby the superiority of the latter system. Consequently it seems favourable to combine the "dry hydrofluori- nation" with the UF.«volatllisation out of ZrFg-bearing melt systems, since the volatilisation yields in a melt are higher than those ob- tained by dry fluerination In a fixed or fluidised bed. Dry fluorination is stilt under investigation by Kali=Chemie, who have found that UFg- volatilisation yields can be increased by introducing an additional intermediary step of pyrohydrolysis. The complete process of dry fluorination would then consist of the following steps: hydrofluori- nation, fluorination, pyrohydrolysis, hydrofluorination, fluorination, Sorption Studies of UFg on NaF It is known that a separation of UFg from some other volatile flssio? gr?ggcf fluorides can be accomp!ished by adsorptin-desorption on NaF'1o= . 1509C had already pr?st +? be a feasible temperature for the separation of MoFg from UFg ~20), Before beginning investi- gations on the behaviour of MoFg, the sorption characteristics of UFg on NaF pellets were determined. Taking the proposed 150°C as an appropiate temperature for separation, in order to make further use of this sorption technique, i+ can be said that the adsorption of UFg, even at this temperature, is still quantitative. On the other hand, it is impossible to desorb any UFg out of the complex Naz[UFs] 291 % Nickel ofter 80min Hydrofluorination ot 650°C 32 30 25 20 N 15 \ 10— A ———Reinicke! \x Hastelloy N Q5 \.& m—— TS 05 10 20 Mol ZrE &7 13. Corrosion as a Function of ZrF4 Content 292 at this temperature. This leads to the conclusion that it should be possible to separate both hexafluorides. Studies in connectlion with the |lifetime of NaF-pellets from the Harshaw Chem. Co. led to the result that this product can be used at least ten times for clean, non=contaminated UF; in a cyclic process; even after the tenth cycle the uranium remained f?xed In a small, limited band. Conclusion By demonstrating that the first two main steps of the FSFVP, 1.e. hydrofluorination and fluorination, do not cause any major difficulties, it should be possible to apply this process to oxide type THTR-fuels also, provided that the next step, the further decontamination of +the volatilised UFg on NaF-beds, can be solved successfully. Furthermore, there is another fact in favour of this process: the absence of plutonium, since the fuel elements now under investigation contain thorium as breed material and highly enriched uranium as fuel. References 1. Argonne Nat, Lab. [l1l."Chemical Engineering Division Summary Report for July, August, and September 1955, ANL-5494 (Del) 2.Nov, 1955, 2, dto for July, August, September 1959, ANL-5924, Dec, 1958, 3., dto for October, November, December 1959, ANL-6101, Febr.1960. 4, dto for April, May, June 1960, ANL-6183. 5. D.O.Campbel |, G.1.Cathers "Processing of Molten Salt Power Reactor Fuels", Ind. Eng. Chem. 52, No.1 (1960), pp.41-44, 6. Oak Ridge Nat. Lab.,Ten. Chemical Technology Division, Annual Progress Report for Period ending 31. August 1960, Fused-Sait Fluoride-Volatility Process, ORNL-2993, 1960. 7. dto for Period ending 30 June 1962, ORNL-3314, 1962, 8. dto for Period ending 31, May 1963, ORNL-4352, 1963. 9, W.H.Carr, "Volatiiity Processing of the ARE Fuel", Chem. Eng. Progr. Symp. Ser. 56, No. 28 (1960), pp. 57-61, 10. G.l.Cathers, R.L.Jolly, E.C.Moncrief, "Use of FSFVP with lrradiated Urania, Decayed 15-30 Days", ORNL-Report 3280, Sept. 1962, 293 1. 12, 13, 14, 15, 16, 20. see 7; and G.l.Cathers, M,R.Bennett, R.L.Jolly, "Application of Fused-Salt Fluoride=Volatility Processing to Varfous Reactor Fueis", Chem. Eng, Progr. Symp. Ser., No.47, Vol. 60 (1964), pp.31-36, ORNL.,, Chem. Techn. Division, Annual Prog. Rep. Perlod ending Aug. 31, 1960, ORNL=2993. R.E.Thoma, H. Insley, A.A.Friedman, G.M.Hebert, “The Condensed System LiF-NaF-ZrFs, Phase Equilibria and Crystallographic Data", J. Chem. Eng. Data 10, No.3 (1965), pp.219=230; J. Phys. Chem. 62 (I [ pp. - H J. Am. Ceram. SOC. 46 (]963)’ pp.37042; R.E.Thoma, "Phase Diagrams of Nuciear Reactor Materials™, ORNL- Report 2548 (1959). see 5-11, . Houdry Process Corp., "Kinetics of Uranlum and Flsslon Product Fluoride Adsorption by Sodium Ftuoride", 60 OCR=1, Jan, 1959; Report TID 11398, L.E.McNesse, "An Experimental Study of Sorption of UFg by NaF" Report ORNL~3494, Nov. 29., 1963, ORNL-Report, ORNL-TM=522, S. Katz, "Use of High-Surface-Area NaF to Prepare MFg+ 2 NaF Complexes with U, W and Mo Hexafluorides", Inorq. Chem. 3, No.l1 (1964), pp.1598=1600. see ref. 8; and G.l|.Cathers, R.L.Jolly, E.C.Moncrief, "Laboratory- Scale Demonstration of the FSFVP", ORNL=-TM=80 (Dec.6, 1961). 294 FUSED SALT-LIQUID METAL SYSTEMS Chairman: W. R. Grimes Oak Ridge National Laboratory Oak Ridge, Tennessee, U.S.A. 295 * EBR-II SKULL RECLAMATION PROCESS I. 0. Winsch, R. D. Pierce, G. J. Bernstein, W. E. Miller and L. Burris, Jr. Chemical Engineering Division Argonne National Laboratory Argonne, Illinois 60439 U. S. A, Abstract A pyrochemical process has been developed for the recovery of enriched uranium from residual crucible skulls that result from the EBE-II melt refining process. The process involves: (1) oxi- dation of the skulls to liberate them from the crucible as a free-flowing powder; (2) addition of the powder to a halide salt and extraction of 75 to 95%Z of the nobler fission product elements from the oxide with liquid zinc at 800°C; (3) reduction of the uranium oxide by contacting the salt with a Mg-17 at. % Zn alloy at 800°C; (4) removal of 95% or more of the remaining fission products by transferring away the molten salt and the metal alloy in which the metallic uranium is insoluble; (5) dissolution of the uranium in a Zn-29 at. % Mg alloy; and (6) recovery of the uranium product by retorting and melting. The Skull Reclamation Process was developed and tested on a pilot and a prototype plant scale in the Chemical Engineering Division of Argonne National Laboratory. However, a change in the EBR-II reactor status from an experimental to a test reactor precluded the installation of Skull Reclamation Process equipment in the EBR-II Fuel Cycle Facility. * Work performed under the auspices of the United States Atomic Energy Commission. 297 Introduction The concept of on-site recovery and recycle of discharged reactor fuel has been established in Argonne's EBR-II reactor complex, Processing of spent reactor fuel and refabrication of new fuel are carried out in the EBR-II Fuel Cycle Facility.(l Fuel processing is carried out by remote pyrochemical methods. A simplified fuel cycle flowsheet is shown in Figure 1. Although the EBR-II reactor may ultimately employ plutonium alloys, an enriched uranium alloy is now used as the fuel in the core loading. The core is an assembly of 0.144 inch pins which are clad with stainless steel and thermally bonded with sodium. A process known as melt refining‘<: was developed and has been used successfully for five years to recover EBR-~II fuel in the Fuel Cycle Facility. A closed fuel cycle melt refining and refabrication of fuel remotely has been successfully demonstrated. However, because of the small scale of operation, continued operation is not justifiable economically, and so recovery of EBR-II fuel by pyrometallurgical methods in the Fuel Cycle Facility was discontinued early in 1969. In the melt-refining process the fuel pins are declad mechanically, chopped to convenient lengths, and charged along with makeup uranium to a lime-stabilized zirconia crucible. The charge is melted, heated to 1400°C, held at this temperature for about 3 hr, and then poured into a mold to form an ingot. This treatment removes about two-thirds of the fission products through volatilization of some fission elements and selective oxidation of others by interaction with the zirconia crucible. The nobler fission products such as molybdenum, ruthenium, and zirconium are not removed by melt refining. The recycled fuel is an alloy of uranium and "fissium".* To avoid an alloy of changing compo- sition, inactive noble metals were alloyed with the initial fuel in their approximate equilibrium concentrations based on an auxiliary removal of about 77 of the noble metals during each cycle for 2% burnup fuel. Experience has shown that the presence of noble metals enhances the irradiation stability of uranium. The product ingot is used to refabricate new fuel pins by injection casting. The fuel pins are inserted into stainless steel cans, welded, and assembled into new fuel subassemblies for recharging to the EBR-II reactor. In the melt refining process, inorganic, radiation-stable materials are used which permit processing of short-cooled, high-burnup metal and ceramic fuels. *Fissium is a name given to a variable mixture of fission product elements (atomic number 40 to 46) which, when alloyed with uranium, impart to the alloy desirable metallurgical properties and radiation stability. 298 EBR-II -~ BLANKET = U PRODUCT {DEPLETED\\ BLANKET U~1at % Pu SLANKED URANIUM) BLANKET RODS L Pu PRODUCT CORE (ENRICHED == U-Fs ALLOY) FUEL ENR U PINS FP VOLATILE Fp U PRODUCT )y NEW MELT SKULL SKULL FUEL U-Fs REFINING »{! RECLAMATION [—=F P PINS ALLOY PROCESS | 5-10% OF U PROCESS 5-10% OF N M NONVOLATILE F P PRODUCT 90-95% OF U INGOT 90-95% OF N M (U-Fs ALLOY} L REFABRICATION BY INJECTION |=—U MAKEUP CASTING FP = fission products NM = noble metals Fs = fissium 1. Simplified Fuel Processing Flow Sheet 299 The rapid recycle of the fuel results in reduction of fuel inventories outside of the reactor. Other advantages of the process over an aqueous process are the small volumes resulting in compact processing equipment, direct production of solid wastes, and the elimination of criticality problems associated with aqueous solutions. When the product ingot is poured in the melt refining process, about 7% of the uranium remains in the crucible as a skull which consists of a mixture of dross and unpoured metal, In addition to uranium, the skull contains about 7% of the original noble metal content and nearly all of the more electropesitive fission product metals, Ba, Sr, Y, and the rare earths. A liquid metal process, called the Skull Reclamation Process, was developed for processing the melt refining skulls. (%) This process has three objectives: (1) recovery of the uranium, (2) removal of the electropositive fission products, which have. been concentrated in the skull, and (3) removal of noble and refractory metal fission products which are not removed in the melt refining process. Since about 93% of the fuel material is recovered in the melt refining process, a recovery of about 95% of the fuel in the melt refining skull will result in overall fuel recovery of at least 99.5%. Decontamination requirements are modest since in a fast reactor, neutron poisoning is minimal and all fuel refabrication is done remotely behind heavy shielding. It is necessary to remove only amounts sufficient to avoid excessive dilution of the fuel. Fission product removals of 60 to 90% are entirely adequate for this reactor. A pyrochemical blanket process (as indicated in Figure 1) was also developed. The basic equipment and operations are similar to those employed in the Skull Reclamation Process and will not be discussed further in this report. In developing the Skull Reclamation Process flowsheet, equipment, and techniques, a total of 36 skull-reclamation runs were made in the pilot plant equipment and 18 runs were made in the prototype plant equipment. During the develcpment period, changes were made in the process to simplify the operations and reduce the overall run time from 32 hr to about 11 hr, The final Skull Reclamation Process flowsheet resulting from this work is shown in Figure 2. The primary steps in the process may be described briefly as follows: Skull Oxidation: The skull in the melt refining crucible is burned at a controlled rate in an 0,-20% Ar atmosphere at 700°C. The skull is converted into a free-flowing oxide powder which is removed from the crucible and charged to the noble metal extraction 300 10¢ MELT REFINING CRUCIBLES AND SKULLS TUNGSTEN CRUCIBLE AND SOLUTION HEEL URANIUM OXIDE A . —m - — — o —— —— — i — A ——— — — Mg-17 at. % Zn ALLOY URANIUM OXIDE DILUTE 05-ARGON OXIS[I)(.;JH-ON MIXTURE 700°C OXIDES OF URANIUM AND FISSION ~ PRODUCTS I METAL SCRAP Y 1! 47 mol % MgCl2 1 27.7 mol % NaCl = NOBLE 18.6 mol % KCI > METAL 6.7 mol % NaF Zn ———— EXTRACTION 750°C ZnClg WASTE ZINC CONTAINING Mo, Ru, Rh, Pd, Tc + FLUX REDUCTION 750°C PRECIPITATED URANIUM WASTE FLUX AND Mg-Zn SUPERNATANT CONTAINING RARE EARTHS, Ba, Sr, Zr URANIUM PRODUCT 2. Skull Reclamation Flow Sheet URANIUM PRODUCT DISSOLUTION 810°C URANIUM PRODUCT SOLUTICN ! Jat. % U 29 at. % Mg 68 at. % Zn SOLVENT EVAPORATION (650-900°C) step of the process. Noble Metal Extraction: This step 1s designed to separate the relatively noble metal fission products from uranium. These noble metals include mclybdenum, technecium, ruthenium, rhodium, palladium, silver, indium, and antimony. To effect the desired separation (at least 50% removal), the finely divided oxides are suspended in a molten halide salt. The noble metals are reduced by contacting the oxides with zinc, and the metals are extracted into the zinc, which is then removed and discarded. It is not essential that the reduced noble metal fission products dissolve in the zinc phase. Some of these elements have very low solubilities and it would be unrealistic to use sufficient zinc to dissolve them completely., Mild agitation of the zinc phase during transfer is sufficient to maintain in suspension those elements whose solubility is exceeded. Reduction: The reduction of U0, by liquid Mg-Zn alloy can be represented by the following overa%l equation: UOZ(salt) + 2Mg (Zn) + 2MgO(salt) + U+(Zn-Mg) The uranium oxide, which is dispersed in the salt phase (MgCl,-CaCl,-CaF,) as a solid, is reduced by the magnesium in the metal phase., The MgO collects in the salt phase and the uranium, which has a solubility of only 0.1 at. % in the Mg- 17 at. % Zn alloy at 700°C, precipitates from solutionm. The experimental work of Knighton et 21{6) and Martin et a1(7) showed that molten halide salts offer advantages in the reduction of uranium oxides by molten metals. The salt promotes more complete reductions and scavenges the MgO by-product of the reaction away from the product in the metal phase. The Mg-Zn and the salt containing rare earths, barium, strontium, and zirconium are pressure-siphoned from the crucible and treated as waste. Uranium Product Dissolution: The uranium precipitate is dissolved in a Zn-30 at. % Mg alloy at 810°C. The solubility of uranium in this alloy at 800°C is about 4.0 at. %Z. However, dissolutions are made with an excess of Zn-Mg alloy to provide about a 3.0 at., % uranium solution. Solvent Evaporation: The Zn-Mg-U product ingot is charged to a beryllia crucible and subjected to a low-pressure retorting operation (750°C, 1l0-micron pressure) to remove the magnesium and zinc. The condensed magnesium-zinc vapors are discarded as waste, and the uranium is consolidated into an ingot by heating the crucible to about 1200°C. This ingot is suitable for recycle to the melt refining operation. 302 Pilot Plant Equipment The equipment used in the pilot plant studies consisted essentially of two large bell-jar furnaces heated inductively and containing tungsten processing crucibles. This equipment was located inside a glovebox whose dimensions were about 13,3 ft long by 8.5 ft high by 3.3 ft wide. A dry N,-Ar atmosphere was maintained in the enclosure and in the furnaces to protect the reagents against oxidation and moisture, Figure 3 shows the glovebox, furnaces, and control panelboard before closure of the glovebox with gloveport panels. A heated transfer line is shown in position for transferring molten metal from one furnace to the other. (Two furnaces were used in the early process flowsheets, In the final flowsheet only one furnace was used and all transfers were made to a waste or product receiver.) Transfers of molten materials were made by pressurizing one of the furnaces with argon, The bell jar covers had three nozzles, one of which was used for an agitator shaft, the other two for insertion of transfer tubes and other equipment used for sampling, liquid-level measure- ment, and temperature measurement. In addition to the bell jar itself, the service nozzles in the top of the jar were water cooled to prevent damage to rubber gaskets. The temperature of the rubber gaskets was 40°C when the furnaces were at 800°C which is well below their permissible operating temperature of 150°C., During the runs, a continuous argon gas purge (5 cfh) into the three bell-jar nozzles inhibited the accumulation of condensed metals and salt around the agitator shaft, the sample port, and the transfer tube. Tungigsn crucibles were used to contain the molten salts and metals. Figure 4 shows a pressed-and-sintered tungsten crucible (12 in. 0.D. x 20 in., high) with integral mixing baffles. This is one of three such crucibles fabricated by Union Carbide Nuclear Corporation, Oak Ridge, Tennessee. The density of the crucible material is about 92 to 94% of theoretical, and the weight of each crucible is about 500 1b. The crucible is shown suspended from its lifting yoke. Below the crucible is the graphite secondary container and susceptor, an insulating sleeve and the flat strip copper induction coil. The coil was powered by a 30 kW-10,000 Hz induction unit. Surface temperature of the coil was about 300°C when the operating temperature of the crucible was 900°C. Two other tungsten crucibles of approximately the same size were fabricated, one by arc-welding rolled tungsten sections, and the other by shear forming. The welded crucible failed at the bottom weld after three process runs. The shear-formed crucible was not used in Skull Reclamation Process runs but gave excellent service in other pyrochemical process studies using halide salts, Cu-Mg-U alloys, Zn-Mg alloys, and zinc at temperatures up to 900°C. 303 Pilot Plant Glove Box Jis 304 } 4, Tungsten Crucible | 305 Internal components of the furnace that required machining were generally fabricated from Mo-30 wt %Z W alloy. These components included the mixer shaft and blades, small heat shields for the shaft bearing assembly and a large heat shield, which covered the crucible. An additional heat shield made of Hastelloy-X was located in the upper portion of the furnace bell jar. Waste and product streams were removed from the tungsten process crucilble through Mo-30 wt % W transfer tubes formed in the shape of an inverted "J". 9) The initial transfer tubes were fabricated of 5/8 in. 0.D. by 5/16 in. I.D. gun-drilled Mo-30 wt % W alloy rod, which was assembled by means of threaded elbows. Contin- uous development resulted in an improved design: two lengths of gun-drilled rod (3/4 in. 0.D., by 3/8 in. I.D.) were coupled together to form a single tube 78 in. long. In place of threaded elbows, the bends were made by hot-forming. The tube was heated to a . temperature of 800°C by means of a number of 600-watt magnesia- insulated, stainless-steel sheathed, 1/8 in, 0.D. heating cables, which were wound on the tube. The heaters were covered with insulation and then canned with a stainless steel enclosure. The enclosure protected the insulation and heaters during handling, and the annular space evacuated and filled with argon to prevent oxidation of the Mo-30 wt 7Z W tube. The material discharged from the process crucibles was collected in a graphite mold suspended from a dial scale. This arrangement permitted control over the quantity of material removed in any transfer by close observation of the scale. Prototype Plant Equipment Prototype plant equipment for the Idaho EBR-II Fuel Cycle Facility (FCF) was designed, fabricated, and tested at Argomne National Laboratory by personnel of the Chemical Engineering Division. The equipment was operated in a large argon-atmosphere enclosure equipped with gas purification and refrigeration systems. All mechanical manipulations were carried out by use of a small crane or by various hand tools operated through gloveports. These operations were performed in a manner that simulated completely remote operation in the FCF. Figure 5 shows the interior of the enclosure as well as the furnaces used in the oxide reduction and retorting steps. The primary components of the skull-oxide processing equipment are a skull-oxidation furnace, skull-oxide-reduction-and-purifi- cation furnace, transfer tube, transfer receiver equipment, and retorting furnace for recovery of the uranium product, 306 3 A “‘j:_w k g '. " ] Fi ’ — vi——-n i o ‘,' r 2 i A v / A\ | A L — o /| | ' df: f 'fi" ( 7| % 5. Interior of Inert Atmosphere Enclosure =2 Figure 6 is a drawing of the skull oxidation furnace which was developed aB Argonne and is installed in the EBR-II Fuel Cycle Facility.(l ) An earlier version of that furnace was used in the inert atmosphere enclosure to prepare skull oxides for the Skull Reclamation Process studies with prototype plant equipment. Figure 7 is a vertical section through the skull-oxide-reduction- and-purification furnace. The furnace body is made of Hastelloy C to withstand operating temperatures up to 900°C. The pressed-and- sintered tungsten crucible was fabricated without internal baffles but had a dished bottom to increase the mixing turbulence from the offset agitator. A second impeller and a baffle cage were added at a later date to enhance mixing intensity. The agitator shaft was driven by a 3 hp D.C. motor at speeds up to 1000 rpm through an open gear-and-pinion assembly. The gears were lubricated with an adherent dressing of APL grease to which molydisulfide powder had been added. The shaft seal con- tained a special high-temperature molydisulfide lubricated packing. A constant low-rate argon purge below the shaft seal prevented entrance of fumes into the packing section. The external resistance heaters (20 kW) were arranged to be opened by pneumatic cylinders to increase the rate of cooling should that prove desirable. The insulated cover as well as the transfer tube, shaft seal assembly and charging port were all sealed by fusible-metal seals. 11§ The heated transfer tube was constructed in a manner similar to that used in the pilot plant equipment.(9) The bore was larger (1/2 vs. 3/8 in. 1.D.) to increase capacity, and braided heaters were used in place of swaged heaters to reduce the number of heater circuits by permitting longer heater sections. Since the transfer tube was designed for use in an inert atmosphere, a sealed protective enclosure was not required. Construction details are shown in Figure 8. During the initial stages of equipment development a potential operating difficulty was encountered with salt and metal fumes that were released when materials were charged to or transferred out of the hot furnace. Charging difficulties were resolved by installing on the furnace a permanent hopper, which was closed by a simple 4-in, ball valve. The transfer receiver (which is mounted on a platform scale to indicate the quantity of material transferred) was placed in an enclosure from which the fumes were exhausted through a filter chamber containing high-temperature AEC filters. %* Product of Shell Development Co., California. 308 POWER LEADS CONNECTOR | ‘Y ] E ;THERMOCOUPLE CONNECTOR INSULATION R - THERMOCOUPLE . RESISTANCE HEATER \L / W STAINLESS STEEL CAN~_} /L MELT REFINING SKULL t 4 ZIRCONIA CRUCIBLE —_] e——— BELL JAR CONTAINER ! SUPPORT ik = . - PRESSURE SENSING PROBE GAS QUTLET — \ GAS INLET FREEZE SEAL f THERMOCOUPLE H : ! I | 1 I | - | : FREEZE SEAL HEATERS : | ’ I il BASE 6. Skull Oxidation Furnace 309 01¢ : r/— AGITATOR DRIVE HOTOR = M [%d GEAR DAIVE ‘—‘—‘\—\r——_ O\ T gilig THERNOCOUMLE WELL v 3 — I RESISTANCE - BEATED \ ' \ N\ TRMSFER i Jfl"fl | —— e cover Swpront . (Raaalliis Joerm . coLvan A . FUSILE METAL 7 ~ g ,’V il "//; SEAL ¢ WEATERS . NSWATED COVER —+— ’/‘ r: :4 i \ ’-, z : e HEAT SHIELDS ¢ " — HESULATOR ASSEMBLY FuRace 000y —___ | l,\; LN B M-—\ 30 % N 05 —— o % DRILLED ROD FOR 1Yy e THERBOCOULE ELL — HEATER SUPPORT T FRABE M N 7 é\musm CRUCHLE === (B=n 1D ° & B 34 DEEN) mm——-—-..._1 |\ Ao i A e Z1] 7. Oxide Reduction Furnace ] MEATER CONNECTIONS /UFTING HANDLE 316 5 STL SHEATH ELECTRIC | oLl | HEATERS THERMAL i _—“3093'?“' INSULATION | Mo - J0%W PANCAKE HEATER THREADED RING SPLASH GUARD Mo J0%W PROTECTIVE CUP MO 30%W FERRULE SEAL 8. Heated Transfer Tube 311 As shown in Figure 2, the uranium resulting from the oxide- reduction-and-purification step is removed from the furnace by dissolution in a magnesium-zinc alloy which is cast into an ingot. The final step in the Skull Reclamation Process is the recovery of the uranium from a Zn-29 at. %, Mg-3 at. % U ingot by evaporating the Zn—Mg.(lz) This operation is carried out in a compact combination still pot, condenser and collector shown in Figure 9. The assembly is located inside a bell-jar furnace to permit operation under low pressure (V10 torr). The still pot is heated inductively with about 7.5 kW power output from a 10,000 Hz generator. A thixotropically-cast beryllia crucible with interior hemispherical bottom is used to contain the metals during the retorting operation. A graphite secondary container is used to support the beryllia crucible and acts as an induction- heating susceptor. Proper performance of the equipment requires close control of power input to avoid surging or bumping. The furnace was equipped with a seismic vibration detector and a contact microphone to warn of the approach of excessive boilup. Pilot Plant Operating Experience Mechanical Performance. In developing the process some changes were made in the skull-oxide processing flowsheet shown in Figure 2. These changes were primarily in the relative quantities of reagents used. Operating conditions employed in the final pilot plant runs are shown in Figure 10. Mechanical performance of the equipment was generally satis- factory. The pressed—-and-sintered tungsten crucible and the Mo-30 wt % W agitator shaft and impeller showed excellent corrosion resistance. The improved design of the heated transfer line gave excellent service in conjunction with techniques for close control of solution-transfer operations. The transfer line was fixed in the furnace with its inlet point about 1/2 in. above the crucible bottom. All transfers were made by pressurizing the furnace with argon. When relatively complete transfers were desired, e.g., transfer of salt and magnesium-zinc after the reduction step or transfer of uranium product solution, the furnace was kept pressurized until excess argon pressure vented through the empty transfer line. In such cases approximately 95% of the available molten material was transferred. When only the metal phase was to be transferred e.g., the zinc phase following the noble-metal- extraction step, the metal transfer was initiated by pressurizing the furnace. When the desired amount of metal had been removed (as indicated by the weight of the receiver), the gas pressure was vented through a relief valve and the transfer abruptly stopped. These transfers were controlled at 90% removal of the zinc to avoid inadvertent transfer of the molten salt in which 312 GRAPHITE FELT INSULATION FIBERFRAX INSULATORS . GRAPHITE SECONDARY — ] CONTAINER COVER \ BUTT SEAL \ ,..j"\ln =L Il n 17-1/4 n STEEL HOLD -DOWN RING PROCESS CRUCIBLE GRAPHITE SECONDARY CONTAINER 9. 8n - GRAPHITE CONDENSER FIBERFRAX INSULATING SLEEVE GRAPHITE DISTILLATE L COLLECTOR 4 r / GRAPHITE FELT ZIRCONIA BRICK -7'||/ 2in I7Tn SILICON CARBIDE GRAIN INSULATION INDUCTION HEATING COIL E/// gfiflmfllfi SUSCEPTOR evens ) Retort Assembly 313 vi¢ FISSIUM OXIDE - 198 kg ZINC - 345 kg Zr0, ~ 005 kg CaF, OR Naf - 025 kg SALT - 50 kg2 Zn 70 kg FISSIUM OXIDE 5 kg Zr0; 015 kg l FLUX 20 kg? D | Ce0y - 011 kg ZINC - 475 kg Zn - 125 kg l | ZnCly - 06 kg NCBLE METAL EXTRACTION 2 hr @ 750°C 825 rpm 100 rpm DURING TRANSFER Mg - 105 kg URANIUM REDUCTION 2 hr @ 800°C 600 rpm SETTLE 05 hr Zn-Mg - 17 kg SALT - 6 8 kg Mg - 21 kg URANIUM DISSOLUTION 1 hr @ 810°C 600 rpm 16 1 kg-Zn-Mg-U 3salt used was MgCly — CaCly 10. Operating Conditions for Pilot Plant Skull Oxide Reclamation Runs WASTE Zn - 32 kg Ce02 015 kg 1 NOBLE METAL EXTRACTION (zn) WASTE Zn 66 kg 2 hr 750°C 800 rpm Mg 18 kg Zn 8 kg _CaFz or Naf®_ [ Y Zn-Mg 29 kg REDUCTION {Mg-17 at % Zn!} WASTE 2 hr BOPC o 800 rpm FLUX 23 kg Zn 27 kg Mg 4 kg PRODUCT DISSOLUTION {Zn-29 at %» Mg-3 at % U) 12 qr, 810°C 400 rpm PRODUCT 35 kg 30ne of these fluxes was used (A) 50 mol % {B) 50 mol % MgCly, 30 mol % NaCl, 20 mol % KCI, (C} 30 mol % MgCly, 50 mol % NaCl BFluoride 1on as CaF, or NaF was added either before the noble metal extraction step or before the reduction step None was used 1n run 11 CObtained from 5 wt % fissium 11. Operating Conditions MgClz, 50 mol % CaCly, for Plant Scale Skull Oxide Reclamation Runs skull oxide is suspended. The presence of moisture in the salt constituted a major problem during the early development phases of the process. The salt is highly hygroscopic and absorbs moisture from even a relatively dry atmosphere (e.g., 750 ppm water). Reaction of this moisture with molten magnesium results in evolution of hydrogen, which can cause the molten salt to foam out of the crucible. This difficulty was resolved by pretreating the salt before it was used in the furnace. Pretreatment consisted of melting the salt and contacting it with a molten Mg-Ca alloy. The Mg0 and Ca0 formed as a result of reaction with moisture in the salt was removed by hot filtration. The dry inert atmosphere in the glovebox prevented absorption of water during subsequent operations. Fission-Product Removal Ruthenium and molybdenum are soluble in the noble-metal zinc extract to about 0.33 at. % and 0.15 at. %, respectively, at 700°C, The waste zinc supernatant is transferred while the zinc is stirred at a speed of about 100 rpm to suspend the molybdenum. Table I shows fission product removals in four pilot plant runs made in accordance with the flowsheet shown in Figure 10. The removal of molybdenum varied between 19 and 33%. Ruthenium removals were good and ranged from 75 to 85%. As noted in the table, zirconium removal in the Mg-17 at. % Zn supernatant after the uranium reduction-precipitation step varied from 30 to 417 and the waste salt showed zirconium removals of 6 to 25%. Cerium removals in the Mg-Zn supernatant and waste salt were 27 to 36% and 30 to 43%, respectively. Table I * Fission-Product Material Balances in Skull Reclamation Process Runs (See Figure 10 for operating conditions) Noble Mg-17 at.Z% Metal Zn Waste Product Total Extract Waste Salt Solution Accounted for Run No. Mo Ru Ce Zr Ce 2r Mo Ru Ce Zr Mo Ru Ce zr SKR-29 33 75 36 31 37 8 47 13 7 51 80 88 75 90 SKR-30 35 85 36 32 39 25 51 13 5 50 86 98 76 107 SKR-31 19 82 29 41 43 6 11 37 7 37 30 119 91 84 SKR-32 22 79 27 30 41 13 6 13 7 35 28 92 78 78 * Values shown represent percent of fission product concentration in oxide starting material. 315 In the development of the plant-scale retorting furnace, a number of pilot-scale runs were conducted using the U-Mg-Zn products recovered from the pilot-plant reduction-furnace runs. Since the equipment very closely resembled the plant-scale retorting equip- ment (to be discussed later), details of pilot scale operations will not be reported. Briefly, the pilot runs demonstrated that beryllia crucibles were satisfactory retort containers; that the Zn-Mg alloy could be evaporated and condensed with little loss; that entrainment of uranium in the distillate was negligible (<0.01%); and that the uranium product could be consolidated into a recoverable ingot. Analyses of fission elements in the final uranium product ingot confirmed the acceptable level of decontami- nation shown in analyses of product solutions in the reduction furnace. Prototype Plant Equipment Performance Skull-Oxidation Equipment. The skull oxidation equipment (Figure 6) was tested by oxidizing separately six skulls from the melt refining of unirradiated uranium-fissium alloy and dumping the resulting skull oxide powders. The procedure involved charging the furnace with a melt-refining cruecible and skull and sealing the furnace with the molten metal seal. All these operations were carried out remotely using an electromechanical manipulator similar to the type used in the Fuel Cycle Facility. Burning was initiated by turning on the control panel which automatically regulated the burning operation. The operation of the furnace control system is fairly complex and is explained in detail in ref. 11. In essence, the system operated on a "demand cycle'". Consumption of oxygen in the furnace resulted in a drop in furnace pressure, which initiated a pump-down cycle to reduce the pressure to a slightly lower level. Oxygen was then admitted to raise the pressure and continue the burning operation. This cycle was repeated, with a slight enrichment of oxygen taking place at each cycle, until all of the metal had been converted to oxide. At that point no further drop in pressure took place and a timer control initiated a purge cycle to turn off the heater and pump out the residual oxygen. The container holding the crucible and oxide powder was then transferred to a remotely operated dumper from which the oxide was discharged into a receiver. The furnace and its control system performed in a highly satisfactory manner in these tests and in the Fuel Cycle Facility. Oxide Reduction Equipment. Eighteen plant-scale (5 kg skull oxide) runs were performed in the prototype oxide reduction furnace., The first ten runs (SRR-1 through SRR-10) were made with the furnace configuration essentially as shown in Figure 7. 316 Preliminary analytical results suggested that contacting con- ditions were not sufficiently vigorous and several changes were made in the equipment as well as the operating conditioms. An additional impeller was placed on the agitator shaft; a temporary three-bar mixing baffle cage was installed in the crucible; operating temperatures, and duration of operating steps were increased slightly. The plant-scale operating flowsheet shown in Figure 11 was used for test Runs SRR-11 through SRR-18, A specific difference to be noted between this flowsheet and the flowsheet in Figure 10 is the absence of ZnCl, in the salt of the noble metal extraction step. The process is designed to operate on a sequential basis. New batches of salt and skull oxide are charged to the crucible, which contains the U-Mg-Zn heel from the preceding product-transfer operation. If excessive magnesium is present, some reduction of the oxide from U,0q4 to uranium could take place., This uranium would be lost in the zinc extractant. In the flat-bottom crucible used In the pilot plant, the heel was large enough to require the addition of ZnCl2 to react with the excess magnesium. In the dished-bottom crucible used in the plant-scale equipment the relative size of the heel was smaller., Conversion of the magnesium to oxide was accomplished by the excess oxidizing power of the uranium oxide (U308 + U05) and the fission product oxides and no ZnCl, was needed. Because of the variety of salts and operating conditions employed in these rumns, consistently reproducible results were not obtained. However, certain general conclusions can be made regarding the performance of the equipment under these different operating conditions. Mechanical performance of the equipment was generally good. Control of furnace temperatures and transfer-line temperatures was readily achieved. Automatic thermocouple-regulated tempera- ture controllers were used in the furnace heaters, and manual rheostats were used in the transfer-line heaters. Samples of salt and metal phases were taken by means of tantalum dip tubes inserted through the charging port. This operation was carried out in a manner that could be adapted readily to remote operation with an electromechanical manipulator. All metal freeze seals including the main cover seal and the auxiliary port seals worked very successfully and would be particularly suitable for high temperature applications in a radioactive environment. The improvement in process performance that could be attributed to the use of the mixing baffle cage, the double impeller, the higher temperatures and longer duration of operating steps was relatively modest. Of the changes made, only the 317 installation of the mixing battie cage represented a significant problem with respect to costs and fabrication. Pressed-and- sintered tungsten crucibles of this size are difficult to fabri- cate even without mixing baffles, It, therefore, does not appear to be expedient to use haffles in a crucible-and-agitator combi- nation such as was employed in this furnace. All the other changes are worth retaining. In Runs SRR-8 and -9, a MgClz-SO mol % CaCl, salt was used and in Run SRR-10 the salt composition was M3012-36 mol % NaCl-20 mol 7% KCl, These changes were made to avoid the carryover of Ca™ and F~ ions in the small amount of flux that is transferred with the final U-Zn~§§ solution produced in the reduction furnace. The presence of Ca™" and F~ was thought to be deleterious to the beryllia crucible used in the subsequent retorting step. In the absence of fluoride ion, the removals of molybdenum, ruthenium, and zirconium, as . determined by product-solution analyses, appeared to be signifi- cantly enhanced (80-90% removal vs. 50% removal). However, the absence of fluoride in the salt mixtures resulted in high uranium losses (v14%) in the waste salt from the reduction step, indicating that fluoride ion is necessary to achieve a high degree of reduction of uranium oxide. Since fluoride salt was necessary to the reduction step, a simple mechanical technique was adopted to prevent the carryover of salt with the final product. A cup-shaped trap, fabricated of alumina~silica fiber, was positioned above the graphite mold into which the U-Mg-Zn product solution was trans- ferred. The metal drained through small holes in the bottom of the trap, while the salt preferentially wet the walls of the trap and was retained. The results of the final eight runs are summarized in Table I1I. In general, the three fluxes tested performed equally well with respect to uranium reduction and fission product removals. The primary factor in achieving high uranium reduction (99%) was the presence of 10 mol %Z fluoride ion, which is shown in comparing results of Runs SRR-11 and -12 with those of subsequent runs. The effect of fluoride ion in the noble metal extraction step upon removal of molybdenum and ruthenium was not clearly defined. In Runs SRR-15 and -16 with the presence of 10 mol 7 fluoride ion in the noble metal extraction step, analyses of the product solutions showed only about 35% removal of molybdenum. In runs SRR-13, -14, and =18, in the absence of fluoride ion in the salt, about 70%Z molybdenum removal was achieved. However, in SRR-17 where fluoride ion (10 mol %) was present in the noble metal extraction step, removal of molybdenum was also about 70%. The inconsistent behavior of molybdenum may possibly be due to the fact that the concentration of molybdenum normally exceeds its solubility in the zinc noble metal extract. Accordingly, the zinc transfer is made while the solution is moderately agitated (~v100 rpm). To prevent loss of salt with the zinc, the agitation 318 Table II Summary of Plant-Scale Skull-Oxide-Reduction Runs SRR-11-18 Percent Percent of Charge in a Fluoride, U Discharged Product Solution Run Flux at.% Reduced _ U Mo Ru Ce Zr 11 B None 92.2 81.1 27 10 5 23 12 B 2 96.2 96.9 28 16 4 30 13 A 10 98.6 91.6 31 16 3 34 14 A 10 99.0 95.2 34 15 2 41 15 B 10 99.3 88.5 NA 21 NA 27 16 B 10 99.0 86.5 64 16 3 55 17 C 10 99.3 90.9 29 19 3 46 18 C 10 94.5 84.4 27 12 2 26 8Flux compositions: A MgClz, 50 at. % CaCl2 30 at. % NaCl, 20 at. % KC1 Na - Not available ¢ MgClz, 50 at. % NaCl is stopped before the zinc transfer is completed. Differences in mixing speeds employed or premature termination of the stirring could result in significant differences in the amount of molybdenum transferred even though essentially equal reductions of molybdenum oxide could have occurred. Removals of ruthenium ranged from 80 to 90%. Ruthenium removals in the absence of fluoride salt in the noble metal extraction step were slightly better than in the presence of fluoride but the differences are not considered significant. Cerium removals were consistently high at about 97%. Removals of zirconium showed a considerable range between 45% and 75%. However, an average removal of 657 was considered adequate for the process. Recovery of uranium in the product solution varied from 81 to 97% based upon analysis of product solution samples. Operational errors in Runs SRR-15, ~16, and -18 led to high uranium losses in the waste streams and low uranium recoveries. The results of runs SRR-13, -14, and -17, which were made under conditions considered reproducible and consistent with flowsheet requirements, showed an average uranium recovery of 92.6%. Since pouring yields for the melt refining process averaged about 937, a recovery greater than 90% in the skull reclamation process would provide an overall uranium recovery exceeding 99%. 319 Retorting Equipment. Forty-five runs were completed in the prototype plant-scale retorting unit. 2) Pertinent distillation data from a number of these runs are shown in Table III. The furnace charge consisted of the Zn-Mg-U ingot formed in the transfer of the product solution from the reduction furnace. All runs were conducted at a nominal pressure of 10 Torr. The bulk of the volatile zinc and magnesium present in the charge was distilled at temperatures of 650 to 750°C. Vaporization of the remaining volatile material and liquation of the uranium product were achieved by increasing the temperature to 1150-1200°C for about 45 min. The resulting ingot was readily dumped from the crucible. Analyses of the ingots from the plant scale retorting furnace runs confirmed the fission product removals that were indicated by analyses of reduction furnace product samples. Satisfactory containment of the Zn-Mg distillate vapors within the graphite retorting enclosure was achieved in 43 of the 45 runs (see Fig, 9). In the two other runs, excessively high distillation rates caused turbulent vaporization of the melt, which resulted in appreciable loss of the vapors from the enclosure. The use of the seismic vibration detector and the contact micro- phone referred to earlier prevented such loss in subsequent runs. Beryllia was the only crucible material found that would retain the molten Zn-Mg-U and yet release the final uranium ingot. Thixotropically cast beryllia crucibles manufactured by the Brush Beryllium Company of Elmore, Ohio were used in the retortinag step. Four beryllia retorting crucibles were tested in these runs and each performed satisfactorily for at least 10 rums. However, after each rum about 25 g of uranium was found in the annulus between the beryllia crucible and the graphite secondary container. This loss of uranium was attributed to seepage of the Zr-Mg-U solution through the crucible wall. Efforts were made by the manufacturer to improve the performance of the cruecibles, but high resistance to seepage could be achieved only at the sacrifice of resistance to thermal shock. Since the crucibles are expensive, the very small loss of uranium was considered an acceptable penalty for achleving a high use factor. The operating experience with pilot plant and full plant scale equipment demonstrated that all the steps of the Skull Reclamation Process could be successfully performed. Although the desired levels of uranium recovery and fission product decontamination were not consistently obtained, those levels achieved were generally adequate. Improvements in overall performance could be anticipated with increased experience. Some problems were encountered in procurement of materials and process equipment which are resistant to the high temperatures of the process and corrosion by the molten salts and metals. Tungsten, molybdenum-tungsten alloy and beryllia proved to be 320 Table III Distillation Runs conducted in Plant-Scale Retorting Agparatusa Nominal pressure: 10 Torr Distillation temperature: 650-750°C, increased to 1150-1200°C for final 45 min. Weight of Weight of Average Weight of Retorted Retorting Distillation Uranium in Uranium Retorting Chargeb Rate Charge® Product Run No. (kg) (g/min) (kg) (kg) PSR-7 34.45 55 2.89 3.00 PSR-8 34.30 53 3.53 3.80 PSR-9 34.15 51 3.89 3.90 PSR-10 33.75 42 3.64 3.75 PSR-11 15.90 35 1.56 1.58 PSR-12 33.35 51 3.84 3.90 PSR-13 33.85 46 3.45 3.50 PSR-14 35.45 44 4.08 4.20 PSR~-15 30.60 41 3.49 3.45 PSR-16 34.40 52 3.37 3.20 PSR-17 15.80 35 1.36 1.40 PSR-18 35.80 47 2.51 2.55 PSR-19 34.10 50 3.00 3.05 PSR~20 35.15 47 1.97 2.10 PSR-21 36.30 46 3.25 3.35 PSR-22 13.15 33 1.09 1.15 PSR-23 12.75 35 0.94 0.95 PSR~24 15.30 36 1.27 1.30 PSR-25 36.60 46 3.00 3.00 aApparatus sized for full-scale (about 4.0 to 4.5~kg uranium basis) boperation in the EBR-II Fuel Cycle Facility in Idaho. Nominal composition: Zn-29 at. % Mg-2 to 3 at. % U. Based on uranium analysis of the product solutions prior to its transfer from the reduction furnace. dProduct is essentially uranium with 1 to 2 wt % residual fission product elements. 321 suitable materials. Although fabrication of large tungsten and beryllia crucibles was difficult and expensive, these items could be made and they gave satisfactory service, Improvement in fabrication technology would be necessary to construct crucibles significantly larger than those used in the plant scale equipment. The heated transfer lines fabricated out of gun-drilled molybdenum- tungsten alloy gave very good service in the transfer of both molten metals and molten salts. The overall mechanical performance was good and demonstrated that the equipment could be adapted to the remote operation required in a processing plant. The equip- ment and technigues employed in the Skull Reclamation Process are directly transferable to other pyrochemical processes. Some of the techniques are currently being employed at Ar%onne in the Plutonium Salt Transport Process for oxide fuels. Acknowledgments The authors wish to acknowledge the contributions of: D. E. Grosvenor, J. F. Lenc, J. H. Schraidt, J. Wolkoff in the development of the Skull Reclamation Process Equipment and the following members of the Chemical Engineering Division for construction of equipment and performance of experiments: T. F. Cannon, A. L. Chandler, P, J. Mack, K. Nishio, R. C. Paul, and K, R. Tobias. They also acknowledge the contribution of R. J. Meyer and L. E. Ross for their direction of the chemical analytic work. 322 10. 11. References Hesson, J. C., M. J. Feldman and L. Burris, Jr., "Description and Proposed Operation of the Fuel Cycle Facility for the Second Experimental Breeder Reactor (EBR-II)', ANL-6605, April 1963, Burris, L., Jr. et al, '"The Melt Refining of Irradiated Uranium: Application to EBR-II Fast Reactor Fuel", Nuclear Science and Engineering, 6, 493 (1959). Trice, V. G., Jr. and R. K. Steunenberg, 'Small Scale Demonstration of the Melt Refining of Highly Irradiated Uranium-Fissium Alloy", ANL-6696 (1963). Burris, L., Jr., I. G. Dillon and R. K. Steunenberg, "The EBR-II Skull Reclamation Process, Part I. General Process Description and Performance", ANL-6818 (1964). Johnson, T. R., R, D. Pierce, L. Burris, Jr. and R. K. Steunenberg, "The EBR-II Skull Reclamation Process, Part II. Oxidation of Melt Refining Skulls", ANL-6874 (1964). Knighton, J. B., L. Burris, Jr. and H. M., Feder, "Purification of Reactor Fuels Using Liquid Zinc", ANL-6223, January 1961. Martin, A. E., R. D. Pierce, J. C. Hesson, T. R. Johnson and A. Schneider, Argonne National Laboratory, unpublished resuilts. Winsch, I. 0., M. L. Kyle, R. D. Pierce and L. Burris, Jr., "Tungsten Crucibles in Pyrochemical Processing of Nuclear Fuels", Nuclear Applications, Vol. 3, April 1967, pp. 245-251. Grosvenor, D. E., I. 0, Winsch, W. E, Miller, G. J. Bernstein and R. D. Pierce, '"Corrosion-Resistant Heated Transfer Tubes for Molten Metals and Salts', Nuclear Applications, Vol. 5, November 1968, pp. 329-332. Miller, W. E., G. J. Bernstein, R. F. Malecha, M. A. Slawecki, R. C. Paul-and R. F., Fryer, "EBR-1II Plant Equipment for Oxidation of Melt Refining Skulls", Proc. 15th Conf, on Remote Systems Technology, 1967, pp. 43-51. Miller, W. E., G. J. Bernstein, D. C. Hampson, R. F. Malecha and M. A. Slawecki, "Fusible Metal Seals in Process Equipment", Proc. 14th Conf. on Remote Systems Technology, Pittsburgh, Oct.-Nov., 1966, Am, Nucl Soc., Hinsdale, Ill., 213-218 (1966). 323 12, Lenc, J., W. E, Miller, G. J. Bernstein, A. Chandler, 13. R. C. Paul and T. R. Johnson, '"Retorting Unit for Recovery of Uranium from Zinc-Magnesium Solutions', ANL-7503, October 1968, Steunenberg, R, K., R. D. Pierce and I. Johnson, "Status of the Salt Transport Process for Fast Breeder Reactor Fuels", This Symposium, 324 STATUS OF THE SALT TRANSPORT PROCESS * FOR FAST BREEDER REACTOQOR FUELS R. K. Steunenberg, R. D, Pierce and I. Johnson Chemical Engineering Division Argonne National Laboratory Argonne, Illinois 60439 U. S. A, Abstract The Salt Transport Process currently being developed at Argonne is a pyrochemical scheme for the recovery of fast breeder reactor fuels. The process objectives are to be able to accom- modate short-cooled fuels and to provide plutonium and uranium recoveries of 997 with fission product decontamination factors of 106 or higher. Stainless steel cladding is removed from the oxide fuel by dissolution in liquid zinc. The oxides are then reduced by a Cu-Mg-Ca alloy in the presence of a CaCljy-CaF, flux. The subsequent plutonium-uranium-fission product separations are achieved through a series of liquid metal-molten salt extraction steps. The metallic plutonium and uranium products, which are recovered by vacuum distillation of the solvent metals, are reconverted to oxide fuel by oxidation with CO, in a fluidized bed reactor. The basic chemistry of the process separations has been investigated and the major emphasis is now on the engineering aspects. Although the development effort is oriented toward a complete process for fast breeder reactor fuels, the initial steps show promise as a head-end treatment for aqueous processing. * Work performed under the auspices of the United States Atomic Energy Commission. 325 Introduction During the early development of high-temperature, nonaqueous fuel reprocessing methods, various investigators became interested in simple purification procedures in which the fuel remains in the metallic state throughout the process, Because these procedures were usually typical of those used by the metallurgical industry and were aimed specifically at metallic fuels, they came to be known as "pyrometallurgical” processes. The main objectives were to repair irradiation damage and to restore the reactivity of the fuel. Most of the proposed pyrometallurgical processes Iinvolved simple operations that could be performed in a small, on-site plant to provide rapid recycle of the fuel to the reactor. Because only modest fission product removals could be achiéved by these methods, they would require fully remote refabrication of the fuel. A major achievement in the area of pyrometallurgical fuel reprocessing was the successful use of melt refining to process the enriched uranium alloy_core fuel of the Second Experimental Breeder Reactor (EBR—II).(l'z) Melt refining consists of melting and liquating the chopped fuel pins in a lime-stabilized zirconia crucible for one to three hours at 1300~1400°C under a high purity argon atmosphere. The liquid alloy is then poured into a mold to form an ingot. During the liquation period, approximately two- thirds of the fission products are removed through volatilization and selective oxidation by the crucible. Melt refining is an integral part of a closed, on-site fuel cycle in which the fuel is discharged from EBR-II, reprocessed, refabricated, and returned to the reactor, using fully remote operations. The EBR-II fuel cycle, including melt refining, was in routine operation for over four years. Although melt refining has been successfully demonstrated and has made it possible to continue the operation of EBR-II in its present mission as a fast flux test facility, it is not a complete process in the sense that auxiliary means are required to recover the unpoured metal and oxide that remaln in the crucible as a gkull after the pouring step. The skull is, in effect, a side stream which can be processed to remove noble metals, as well as other fission products, from the uranium and thereby maintain the desired concentration of fissium™ in the fuel alloy. Approximately 7% of the uranium in the original melt refining char%e appears in the skull., An auxiliary "Skull Reclamation Process' 3) was developed to recover this material, but the process equipment was not installed in the EBR-II Fuel Cycle Facility because of other priorities. Nonetheless, the development effort on the Skull Reclamation Process contributed substantially to the technology of the pyrochemical processes currently under development. * Fissium composition (wt %): Mo, 2.78; Ru, 3.20; Rh, 0.54; Pd, 1.13; Zr, 0.93. 326 Pyrochemical processes are distinguished from the earlier pyrometallurgical processes primarily by the fact that the major fuel constituents (e.g., uranium, plutonium, thorium) are subjected to oxidation-reduction reactions, usually in liquid metal and salt solvents. Separations of flssile, fertile and fission product elements are effected by techniques such as volatilization, pre- cipitation, liquid metal-molten salt extraction and electrolysis. Although these processes tend to be more complex, they offer the possibility of much greater versatility and higher performance in terms of recoveries and fission product removal. The Salt Transport Process, which is currently under development at Argonne,_ is aimed at stainless steel-clad, uranium-plutonium oxide LMFBR fuels. The process is expected to accommodate both core and blanket material and to provide a plutonium recovery of at least 99%. Because of the relatively low value of the natural or depleted uranium used in LMFBR fuels, there may be little economic incentive for immediate decontamination and recovery of the uranium. It appears, however, that 997 uranium recovery can be achieved by the process. With a modest amount of multiple staging, fission product decontamination factors of 10% or greater should be possible. Salt Transport Process A schematic illustration of the proposed Salt Transport Process flowsheet 1s shown in Fig, 1. The process begins with fuel de- cladding and includes the subsequent steps through resynthesis of the oxide fuel material. Decladding Stainless steel cladding is removed from the fuel by selective dissolution in liquid zine, which does not react with the uranium and plutonium oxides. Fuel assemblies are immersed in the liquid zinc at about 850°C, A layer of molten CaCl,-20 at. % CaF, is maintained on top of the zinc to inhibit its vaporization and to provide a liquid heat transfer medium when the zinc-stainless steel solution i1s transferred away from the oxide. The release of some gaseous fission products (Kr, Xe, I, 3H) is anticipated during the decladding step. These are collected in the argon cover gas, which is confined for decay and further processing. Any residual zinc from the decladding step can be removed by vaporization; the stainless steel in the residual zinc is removed in the subsequent process steps, * Liquid Metal Cooled Fast Breeder Reactor 327 8Z¢ SALY TRANSPORT PROCESS FOR LMFBR FUEL {See Text for Stream Compositions) FUEL ASSEMBLIES fi— OFF-GAS FISSION PRODUCTS, SEVEN-STAGE MIXER-SETTLER Hz, Nz, €O, COz STAINLESS STERL ‘ Pu_RETORT CLADDING, Na MgClz-NaCl-KCI-MgF, Zn-thg SALT WASTE - ~ SALT, FP-3 OFF-GAS ) | — DECLADDING Ar, FP-1 t 11 T Londensate CaCIZnCaF / zn, Na Cd-Mg I_ (e METAL WASTE ' Zn-Mg-Pu In, Fe, Cr, My OFF-GAS HOLD Ar, FP-1 TANK REDUCTION SALT WASTE = CaCly-Caf CaClz, CaFy, Mg-Cu-Ca €a0, FP-2 Mg-Cu % Mg-Cu-Pu o Ma-Cu -—I—- METAL WASTE L RECYCLE Ma, Cu, FP-4, Zn i U MgCly, Pu U0y [ RECYCLE | MgCl2 | r ] In-Mg 0, Puiny U RECOVERY mop,—1— | L1 ] L Cu-Mg f Z Cu-Ma-U I I I ZnJ (AL u METAL WASTE I Hy, Ny, £0p Cu-Mg Cu, Mg, FP-4 DECLADDING- REDUCTION EXTRACTION U ACCUMULATION FLUID-BED VESSEL UNIT AND CONVERTER VACUUM MELTING KEY TO FISSION PRODUCTS FP-1 Kr, Xe, 3H, (1) FP-2 Rb, Cs, Sr, Ba, Sm, { FP-3 Y, RARE EARTHS FP-4 Zr, Nb, Mo, Tc, Ru, Rh, Pg Figure 1. Salt Transport Process for IMFBR Fuel Oxide Reduction The oxide reduction step is performed in the same vessel as the decladding step. To carry out the reduction, additional CaCl,- 20 at. % CaF, salt is charged to the vessel and the mixture o salt and oxiae fuel is contacted vigorously at 800°C with a liquid Mg-29 at. % Cu-34 at. % Ca alloy. The uranium and plutonium oxides are reduced to the metals by the calcium and the Ca0 by-product of the reaction is taken up by the salt. The alkali and alkaline earth fission products also appear in the salt, which is discarded as a waste. The remaining gaseous fission products should be released during the reduction reaction. The argon cover gas is handled in the same manner as that from the decladding step. When the reduction is completed, the metallic plutonium is in solution in the liquid metal phase and the uranium, which has a solubility of about 50 ppm, forms a bed of precipitated metal. Plutonium-Rare Earth Separation Because of the chemical similarity between plutonium and the rare earths in pyrochemical systems, the separation of yttrium and rare earth fission products is a key step in the process. The principal reason for selecting a Cu-Mg alloy as the liquid metal solvent is that it provides the highest plutonium-rare earth separation factor of any system that was investigated. The super- natant Cu-Mg-Pu alloy, including Cu-Mg wash solution, goes to a semicontinuous mixer-settler battery. Figure 1 shows a seven-stage mixer-settler, in which the first four stages are used to extract rare earths from the Cu-Mg-Pu alloy with a MgCl,-30 mol % NaCl- 20 mol % KC1-3 mol % MgF, salt phase. The MgCl, is necessary to provide the required distribution coefficients for a practical separation. The NaCl and KCl serve as diluents to lower the melting point of the salt well below the desired operating temperature of about 650°C. The small amount of fluoride is used to promote disengagement of the liquid salt and metal. Under practical conditions it appears that a plutonium-rare earth decontamination factor of approximately 100 can be achieved in each stage. Each of the four stages is operated with a captive salt phase and the Cu~Mg-Pu alloy is a transient phase. After each batch of fuel is processed, the salt from the first stage is discarded as waste, the salt in the second, third and fourth stages is moved ahead one stage, and new salt is added to the fourth stage. Plutonium Salt Transport Separation In the last three stages of the mixer-settler as it is shown in Fig. 1, the plutonium is recovered from the Cu-Mg alloy, which retains the more noble fission products (Zr, Nb, Mo, Tc, Ru, Rh, Pd....). A molten salt of the composition mentioned previously (MgClz-30 mol % NaCl-20 mol % KC1-3 mol % Mng) is used as a carrier 329 to transport the plutonium from the fifth stage through the sixth stage to the last stage. In this operation the plutonium is ex- tracted from the Cu-Mg ''donor'" alloy into the salt and then extracted from the salt into a Zn-30 at. % Mg "acceptor' alloy which has a high affinity for plutonium. In the sixth stage the salt is con- tacted with a captive Mg-20 at. Z Mg alloy to remove any residual noble metal fission products and any Cu-Mg alloy that might be entrained in the salt. The plutonium salt transport step is carried out simultaneously with the rare earth extraction step. During this operation the plutonium is removed continuously from the Cu-Mg alloy cycling through the first five stages of the mixer- settler and is transported to the Zn-Mg acceptor alloy. After the plutonium is removed, the Cu-Mg alloy is recycled to the oxide reduction step for the next batch of fuel, Since only traces of noble metal fission products and entrained Cu-Mg alloy reach the Cd-Mg alloy in the sixth stage, this latter alloy can be used for many batches of fuel before it 1s discarded as waste. The plutonium salt tramsport step 1s expected to provide a decontamination factor of 106 or greater for the noble metal fission products. Plutonium Recovery The Zn-Mg-Pu alloy from the last stage of the mixer-settler is transferred to a retorting vessel where the zinc and magnesium are removed by vacuum distillation. After retorting, the plutonium metal product is recovered as an ingot. Uranium Salt Transport Separation The metallic uranium precipitate in the decladding-reduction vessel is dissolved in a Cu-6 at. %Z Mg alloy in which the solubility of uranium is about 4 at. % (14 wt %) at 900°C. This stream is fed to a four-stage salt-metal extraction. Although it has been shown that plutonium does not coprecipitate with uranium in the oxide reduction step, a small amount (up to about 1%) may be trapped physically in the uranium bed. The first stage of the uranium treatment is used to extract any plutconium from the Cu-Mg-U alloy with a MgCl, salt phase, which is recycled to the next batch of fuel. The other tfiree stages are a uranium salt transport separation which is identical in principle to the plutonium salt transport step, but which operates under somewhat different conditioms. A Mg-17 at. % Zn acceptor alloy is used, the intermediate alloy between the Cu-Mg donor and the Zn~Mg acceptor is zinc or cadmium, the salt phase is 100Z MgClz, and the operating temperature is about 850°C. The first three contacts are made in a mixer-settler unit and the fourth stage i1s a vessel in which the uranium, which has a low solubility in the Zn-Mg alloy, is accumulated as a precipitate. A portion of the Cu-Mg alloy from the uranium salt transport separation becomes a waste containing the noble metal fission products and the remainder is recycled. 330 If the uranium is to be stored for deferred recovery, it could be alloyed with a small amount of another metal such as iron to form a liquid phase that could be transferred out of the decladding- reduction vessel and cast into ingots. Uranium Recovery The liquid salt and metal are transferred away from the pre- cipitated uranium bed in the uranium accumulation vessel. The bed is then vacuum-melted to form a product ingot. Fuel Resynthesis The metallic uranium and plutonium products are alloyed in the appropriate proportions, charged to a fluidized-bed reactor, and then hydrided and dehydrided to form a fluidizable powder. The powdered metal is first nitrided and then converted to mixed U0p~Pu0, by treatment with CO5. Status of the Process Development Most of the laboratory studies that were needed to select appropriate solvent systems for the Salt Transport Process have been completed. This work has resulted in a substantial body of data on phase diagrams, solubilities, distribution coefficients and thermodynamic properties for various metal and salt systems. This information is useful not only as a basis for the Salt Transport Process, but alsc for new pyrochemical process applications that might be envisioned. The chemical feasibility of all of the major separations has been established in laboratory-scale experiments. However, more detailed studies are needed on some items such as the mechanism of oxide reduction, the morphology and physical nature of metallic precipitates, possible radiolytic effects in the salt golvents, and the behavior of minor fission products and by-products such as curium and neptunium. Engineering investigations have been in progress for some time on general pyrochemical operations such as mixing of metals and salts, phase separations and transfers, fluidization techniques and retorting. Although the major effort is being devoted to mixer-settlers for contacting liquid salts and metals, studies have also been conducted on stirred tanks and packed columns. Small-scale tests have been performed with mock fuel assemblies consisting of 13 type 304 stainless steel tubes with a 35-mil wall and three 1/4-in. thick spacer plates. Each tube contained 12 high- fired U0, pellets and one pellet that had been ground to a fine powder. At 800°C,the zinc dissolved the tubing in less than 30 min and the spacers in 3 to 4 hr. It was found that over 95% of the 331 zinc could be transferred out of the vessel without entraining more than about 0.2% of the oxide. Other studies of the decladding procedure showed that the 310 series stainless steels, zircaloy and irradiated 304 stainless steel all dissolve more rapidly than the unirradiated 304 stainless steel. In various laboratory and bench-scale engineering tests it has been shown that both powders and high-fired forms of U02, uo —Pu02 and Pu0, can be reduced completely by the proposed method. e reduction procedure has been carried out on a scale of up to about 5 kg of U0, pellets. A series of runs were made in which the oxide reduction, plutonium- rare earth separation, plutonium salt transport and plutonium retort- ing steps were performed in sequence with about 200 g of plutonium, 800 g of uranium and 100 g of inactive fission product elements. The . results were in good agreement with the process performance pre- dicted on the basis of the laboratory data. The removal of the nobler fission products from uranium was also demonstrated in separate salt transport experiments involving about 2 to 5 kg of uranium. The recovery of uranium from Zn-Mg alloys had been investigated earlier during the development of the EBR-II Skull Reclamation Process. Only a minor amount of work has been done on the fuel resynthesis step. Uranium metal was converted to U0y by the hydriding and COZ—hydrogen treatments in a fluidized bed, but achieving close control of the stoichiometry together with a low carbon content of the product proved to be difficult. A few exploratory tests have indicated that hydriding followed by nitridation with nitrogen gas and conversion of the nitride to U0, by 002 is a more promising method. Most of the engineering effort is currently devoted to the construction of a glpvebox facility in which all the steps of the Salt Transport ProtCess can be performed sequentially on a scale of about 1 kg of plutonium, 4 kg of uranium, and 1 kg of fission products (ndn-radioactive). High temperature mixer-settlers for the multistage plqunium decontamination steps are being developed for this facility. 4) Although no provision is being made for handling irradiated fuels in the facility, engineering techniques which have been, or could be developed for remote operation are being employed in many instances. A continuing effort is made to test and select materials that are suitable for use in pyrochemical process equipment. This work, which is concerned mostly with refractory metals and ceramic materials, is the subject of a separate paper in this symposium.(s) 332 Some preliminary studies have been made on a conceptual design of a Salt Transport Process with a capacity of one tonne/day of LMFBR core and blanket fuel (equivalent to about 15,000 MWe). It appears that two decladding vessels about 7 ft tall and 18 in. in diameter would be required with three fuel assemblies being declad at a time in each vessel. These vessels would most likely be made of tungsten or have a tungsten lining. The oxide reduction operation would be done in the same vessels. The mixer-settler bank for rare earth removal and plutonium salt transport would be about 20 x 24 in. and about 6 ft long. Niobium appears to be a suitable material for this unit, except in the last stage where tantalum or a tantalum lining might be used. In the plutonium retorting step, criticality considerations probably would require three units of slab geometry about 20 x 30 in. by 1-1/2 in. thick, although a single unit may be feasible. The three-stage mixer-settler for uranium decontamina- tion would be about 10 x 20 in. and 2 ft long. This equipment and the plutonium retort would both be made of a refractory metal, such as tungsten. The uranium accumulation vessel could be made of graphite and would be about 18 in. in diameter and 5 ft tall, For fuel resynthesis, two 8-in. dia and two 24-in. dia fluidized bed reactors about 6 ft tall would be used for core and blanket material, respectively. Stainless steel would be a suitable material for these units. In addition to the major process equip- ment, various other vessels would be required for making up and charging solvents, holding process solutions and disposing of waste streams. Nevertheless, it appears that a plant of this type would be quite compact. Discussion For the near term at least, LMFBR fuels will most likely be oxide, clad with stainless steel. The core fuel is expected to consist of natural or depleted U0, and about 15-20% Pu0,. The core may be operated at a specific power as high as 200 kW/kg (U + Pu) and reach a burnup of 100,000 MWd/tonne. The blanket, initially natural or depleted UOZ’ may contain up to about 4% plutonium at discharge and reach a burnup of arounfl 15,000 Mwd/ tonne. In most of the proposed LMFBR designs, the core fuel and axial blanket are included in the same assemblies and the radial blanket consists of separate assemblies. The present practice in the commercial reprocessing of light water reactor (LWR) fuels is to cool the fuel for about six months to alleviate the problems of fission product heat removal, iodine release and radiation decomposition of process solvents. However, the high plutonium content of IMFBR core fuel results in a large capital investment and a correspondingly strong incentive to minimize the out-of-reactor fuel inventory. Therefore, it is highly desirable, if possible, to decrease the cooling time to 30 days or 333 less. This short cooling time, together with the high burnup and specific power of IMFBR fuel, poses some difficult problems in fuel reprocessing. In a typical IMFBR fuel assembly containing 60 kg of fuel, the rate of heat generation from fission product decay after 30 days of cooling would be about 13 kW or 44,000 Btu/hr. It is apparent that provisions must be made to handle wvery high levels of heat and radiocactivity. For example, it would be necessary to dilute 30-day-cooled IMFBR core fuel by a factor of about 50 to reach the same heat and radiation levels as those encountered in the processing of six-month-cooled LWR fuels, A particularly troublesome problem that is anticipated with short-cooled LMFBR fuels is the 1ar§e amount of radioactive fission gases that are present (mainly 11 and 133%e). The activity of these gases in 30-day-cooled IMFBR fuel 1s expected to be about 10° times greater than that in the LWR fuels currently being processed. It appears that a pyrochemical process for LMFBR fuels could provide attractive solutions to many of these problems. The proposed zinc decladding procedure replaces the mechanical dis- assembly and chopping steps conventionally used in aqueous process- ing. A prior sodium removal step, other than simply allowing adhering sodium to drain off, should not be necessary, nor would sodium-logged fuel pins be expected to create a problem. The removal of fission product decay heat may be considerably simplified by the high operating temperatures and the presence of liquid metal and salt solvents. The gaseous fission products are collected in a small volume of inert gas and most of the iodine activity 1s ex- pected to remain in the salt waste. The process wastes are produced directly as solidified salts and metals, The process appears to be capable of handling high~fired U0y, Pu0, and either mixed or solid- solution UOZ—PuOZ. Although the Salt Transport Process is being developed on a long-range basis as a complete process for future LMFBR fuels, it might prove attractive to use the initial steps as a head-end procedure for aqueous processing of interim LMFBR fuels in existing plants. This head-end process might include decladding, oxide . reduction, a plutonium-uranium separation and partial decontamination, depending on the requirements of the aqueous plant. It is felt that this possibility deserves further consideration. 334 References Hampson, D. C., R. M. Fryer and J. W. Rizzie, '"Melt Refining of EBR-II Fuel', This Symposium. Feldman, M. J., N. R. Grant, J. P. Bacca, V. G. Eschen, D. L. Mitchell and R. V. Strain, "Remote Fabrication of EBR-1I Fuels", This Symposium. Winsch, I. 0., R. D. Pierce, G. J. Bernstein, W. E. Miller and L. Burris, Jr., "EBR-II Skull Reclamation Process', This Symposium. Pierce, R. D., W. E. Miller, J. B. Knighton and G. J. Bernstein, "Multistage Contactors for Liquid Metal-Salt Extraction', This Symposium. Kyle, M. L., R. D. Pierce and V. M. Kolba, "Containment Materials for Pyrochemical Processes', This Symposium. 335 * URANIUM AND PLUTONIUM PURIFICATION BY THE SALT TRANSPORT METHOD J. B, Knighton, I. Johnson and R. K. Steunenberg Chemical Engineering Division Argonne National Laboratory Argonne, Illinois 60439 U. S. A, Abstract Uranium and plutonium may be separated from each other and from nobler metal impurities by the salt transport method in which a solute 1s transferred selectively from one liquid alloy (donor) to another (acceptor) by cycling a molten salt between the two alloys. Data have been obtained on the solubilities of uranium and plutonium in liquid alloys of magnesium with copper, zinc and cadmium and on their distribution behavior between these alloys and salts containing MgCl,. The chemical basis for the partitioning of plutonium, uranium and impurity elements between the salt and metal phases is discussed and the factors that enter into a salt transport separation are described, It has been shown by labcra- tory and bench-scale engineering experiments that the transport of plutonium and uranium takes place as predicted from the solubility and distribution coefficient relationships. Although the salt transport procedure is being developed for the reprocessing of nuclear fuels, it has potential application to the separation and purification to other metals, as well as plutonium and uranium. * Work performed under the auspices of the United States Atomic Energy Commission, 337 Introduction The term "salt transport" has been applied to a purification technique whereby a metallic solute is transferred selectively from one liquid alloy (donor) to another liquid alloy (acceptor) by circulating a molten salt between the two alloys. The transfer takes place through oxidation of the solute by the salt at the donor alloy and its subsequent reduction by the acceptor alloy. The objective of the present studies was to investigate the chemical aspects of plutonium and uranium salt transport purification steps that could be incorporated into a pyrochemical process for fast breeder reactor fuels, Various survey papers have been published on high-temperature liquid metal-molten salt extraction methods for the processing of nuclear reactor fuels.,(1-8) certain investigations that were carried out earlier at Brookhaven National Laboratory and at Ames Laboratory are particularly relevant to salt Transport separations. The Brookhaven work involved a process that was being developed for the fuel of the proposed LMFR (Liquid Metal Fuel Reactor).(g'lz This process employed a procedure in which UCl, in a molten salt phase was reduced selectively by a Bi-Mg alloy to separate uranium from rare earth fission products. The uranium was then re-oxidized by a new salt stream and reduced again by Bi-Mg to form a Bi-Mg-U fuel alloy. At Ames Laboratory, Chiotti (13 patented a salt trans- port method for removing rare earth fission products from a Th-Mg alloy. The rare earths were selectively oxidized by MgCl, and subsequently removed from the MgCl, by reduction with a Zn-1Q0 at. % Mg alloy. This work was extended gy Chiotti and Klepfer, 14) who developed a salt transport separation employing liquid Th-Mg and Zn-Mg alloys in mutual contact with a salt, In the present studies, magnesium alloys of zinc, cadmium, and copper were Ilnvestigated as potential donor and acceptor alloys for the purification of plutonium and uranium from the fission product elements more noble than uranium by the salt transport process. These studies included measurements of the solubility of plutonium and uranium in the liquid alloys and measurements of the distribution coefficient for plutonium and uranium between liquid magnesium alloys and molten salt mixtures containing MgClz. Chemical Basis of Salt Transport Separations Partition of Solutes between Liquid Metals and Salts The salt transport separation of plutonium and uranium from each other and from the more noble impurity elements depends primarily upon differences in the distribution of uranium, plutonium, and impurity elements between liquid metal and salt solvents. In general terms, a metal, M, partitions between a 338 liquid magnesium alloy and a molten salt phase containing MgCl, by the following reaction: M(alloy) + gngCIZ(salt) z MCln(salt) + %-Mg(alloy) (1) The thermodynamic (activity) equilibrium comstant, K, can be expressed in terms of mole (or atom) fraction, x, and activity coefficient, Y, of the reactants and products as follows: n/2 n/2 xMc1 " Fug YMCl " Tvg a n/2 . .n/2 (2) xM ngCl ™M = TMgcCl K The distribution coefficient, D, for the metal, M, is defined as the ratio of the mole fraction of MCl in the salt to the atom fraction of M in the alloy: *Mc1 D-an (3) The equilibrium constant, K_, is related to the standard free energy change, AG®, for Reaction 1 by the equation - - ° o ° _ ° RT In K AG AGfMCl 2 AGfMgCl (4) where AGfMCl and AGfM gCly are the standard free energies of formation of MCln and MgClZ, respectively. Equation 4 indicates that metals whose chlorides have larger negative free energies of formation than that of MgCl, (on a per mole of chlorine basis) will tend to distribute prefiominantly to the salt phase (K > 1). When the free energy of formation of MCln has a lower negative value than that of MgCl, (Kz < 1), the metal, M, will tend to distribute to the alloy phase. By substitution and rearrangement, the distribution coefficient may be expressed in logarithmic form as AG® log D 2 3RT + (= log aM + log YM) - (— 103 aMgCl+ log YMCl ) (5) where three groups of terms are shown in the right-hand side. The first group, the single term AG®/2.3RT, depends both on the value of the free energy of formation of MCl, relative to the valug for M3012 and on the temperature. The second group of terms, -5 log log yyMs depends on the composition of the liquid magnesium %oy and on the temperature; it is indepepdent of the salt composition. The third group of terms, - 7 log aMgClz + log YMCln' 339 depends on the composition of the salt and on the temperature. In general, the distribution coefficients of various elements fall in the same order as the free energies of formation of their chlo- rides. However, large differences occur in the distribution coefficients because of solvent effects. For example, the values of the distribution coefficient at 600°C for praseodymium, cerium, and plutonium are lowered several orders of magnitude when Zn-low Mg or Al-low Mg alloys are substituted for Cu~Mg alloys. Examination of the free energies of formation of fission product metals and structural materials (Table I) shows that the salt transport process should be particularly useful for the separation of plutonium and uranium from more noble fission product metals (Zr, Nb, Mo, Tc, Ru, Rh and Pd) and from metals such as Fe, Cr and Ni, which are typical constituents of alloys used to clad nuclear reactor fuel elements. These free energies of formation suggest that zirconium would be most difficult of these elements to separate from uranium and plutonium by partitioning between salt and metal phases. Several attempts to measure the distribution coefficient of zirconium between molten salts containing MgCl, and liquid magnesium alloys resulted in values of 10~%4 or less in all cases. The actual value may be even smaller, since it 1s difficult to avoid slight contamination of the salt samples by the alloy. Thus, zirconium and all the elements below it in Table I should be separated readily from uranium and plutonium by a process in which the uranium and plutonium are extracted into a molten salt. The difference in the distribution behavior of two elements, M, and My, is expressed as a separation factor, o o = = (6) The separation factor is strongly dependent on the composition of the liquid alloy, and to a lesser extent on the temperature and on the composition of the molten salt. Salt Transport Separations A salt transport procedure for uranium is i1llustrated schematically in Fig. 1. Metallic uranium, which is initially present in the donor alloy, is oxidized and extracted into the transport salt. The nobler metals and structural metals remain in the donor alloy U(donor alloy) +~% MgClz(salt) +> UC13(salt) + %-Mg (donor alloy) (7) 340 TRANSPORT SALT (MgCl2) ________________________ | ] {77777 R ' uCls S MgCl, | 5MgCl, ucCis | N | U %Mg %Mg U DONOR ALLOY ACCEPTOR ALLOY 1. Schematic Representation of the Salt Transport Process 341 Table I Standard Free Energies of Formation of Chlorides and Thermodynamic Equilibrium Constants Ka_for the Reaction M+ 5 MgCl, = MC1 + 3 Mg at 1000°K -AGE® MC1 (kcal/g-equiv. Cl) Ka BaCl, 83.4 1.7 1051 KC1l 8l.4 1.5 105 RbCl 81.2 1.4 1010 SrCl, 81.0 1.5 104 ‘ CsCl 80.0 7.5 109 SmCl2 80.0 5.6 104 Licl 78.8 4.1 108 CaCl2 77.9 6.8 103 NaCl 75.7 8.6 106 LaCl,4 67.0 1.3 105 PrCl3 66.3 4.4 105 CeCl3 66.3 4.4 104 ThCl3 65.3 9.6 104 NdC1; 64,2 1.8 10, YC1l 6l.2 2.0 10 PuC 3 58.9 6.1 MgCl2 57.7 1.00 -3 uc1 54.0 3.8 10_4 Zrci2 49,2 1.9 10_7 MnCl2 42.3 1.9 10_10 ZnCl2 35.0 1.2 10_12 CrCl, 31.9 5.3 10_12 cacl, 30.4 1.2 10_14 FeCl2 26.6 2.5 10_37 Nb015 24,6 6.8 10—8 CuCl 22.0 1.6 10_17 NiCl 20.0 3.3 10_22 MoC1 8.0 1.9 1075 TcC.‘L3 7.0 5.7 10_12 RhCl1 5.8 4.5 10_24 PdCl2 3.8 2.8 10_3? RuCl3 1.4 1.2 10 342 When the transport salt containing the UCl, is contacted with the acceptor alloy, the reverse reaction takes place: UCla(salt) + %Mg(acceptor alloy)»U(acceptor alloy)+%MgC12(salt) (8) Therefore, the net reaction is U(donor alloy)+%Mg(acceptor alloy)-+U(acceptor alloy)+%Mg(donor alloy) (9) Magnesium chloride consumed at the donor alloy by oxidation of uranium and plutonium is regenerated at the acceptor alley by magnesium reduction of UCly or PuCl,. Thus, the salt composition remains constant throughout the salt transport operation. For each mole of uranium transferred from the donor alloy to the acceptor alloy, 1.5 moles of magnesium are transferred in the opposite direction. The increasing concentration of magnesium in the donor alloy and magnesium depletion in the acceptor alloy must be taken into account in the design of a practical process. When Reaction 9 has reached equilibrium, both alloys are in equilibrium with the transport salt, and the ratio, R, of uranium in solution in the acceptor and donor alloys is equal to the ratio of the distribution coefficients of uranium for each alloy and the salt: R = at. Z U in acceptor alloy _ D{(donor alloy) (10) at, Z U in donor alloy D(acceptor alloy) The transfer of a solute (e.g., uranium or plutonium) from the donor alloy to the acceptor alloy is achieved by circulating the transport salt between the two liquid alloys. Although a variety of methods may be used to circulate the transport salt between the two alloys, all are basically similar in that a quantity, ng (moles), of transport salt is contacted with a quantity, , of donor alloy to transfer the solute to the transport salt. A fraction, X, of this transport salt and, under some conditions (i.e., if entrainment of liquid metal occurs), a small fraction, Z, of the donor alloy 1s contacted with a quantity, n,, of the acceptor alloy to transfer the solute from the transport salt to the acceptor alloy. The transport salt is then returned to the donor alloy to complete the cycle. The fraction, F,, of the uranium or plutonium present in the transport salt—gonor system that transfers to the transport salt-acceptor alloy system is given by D 'Eg X+ 7 F = ——— (11) 343 and the fraction, FA' in the transport salt-acceptor alloy system transferred back to the transporg salt-donor alloy system is given by D S Xx+2z A nA FA = ng (12) Pym, t 1 A where D, and D, are the distribution coefficients of the solute for the transport salt-donor alloy and transport salt-acceptor alloy systems, respectively. It may be shown that after s cycles, corresponding to the circulation of a quantity, sn X, of tramsport salt, that the fraction, ¢, of the solute initially present in the transport salt-donor system transferred to the transport salt- acceptor alloy system is given by A s b= Q-7 QA-F), (13) where F = (1 - F ) (1 -F,) (if it is assumed that none of the solute is present initially in the acceptor alloy). Equation 13 indicates that the fraction transferred approaches a limiting value (given by the first factor) as the number of cycles, or the quantity of salt cycles, is increased, The maximum possible fraction, ¢ » transferred is given by max F A A“D ¢ =1l-7—=1- (14) max 1 F D A where the approximation is good to about 1%. Equation 14 indicates that if the ratio n,/n; is unity, then, for a 99% transfer, Dp should be about 100 times D,. The number of cycles, or the quantity of transport salt cycled to achieve a given percentage of the maximum possible transfer, is dependent only on the value of F. For example, to reach 997 of the maximum possible transfer when F = 0.5 requires about seven cycles; whereas, when F =0.7, 13 cycles are required. Large values of Dp favor rapid transfer. Equations 11 to 14 are derived on the assumption that all the uranium or plutonium present in the transport salt-donor alloy or transport salt-acceptor alloy systems is in solution, either in the alloys or the transport salt. However, uranium and plutonium have limited solubilities in several usable alloys. If, for example, the uranium present in the transport salt-donor alloy system is not all in solution, then the fraction of the total uranium present that is transferred to the acceptor alloy during each cycle of the transport salt would be less than would be the case 1f all the uranium present were in solution. Therefore, limited solubility of uranium in the donor alloy increases the number of cycles of 344 transport salt required to transfer a given fraction of the uranium to the acceptor alloy. The quantity of uranium transferred from the donor alloy to the acceptor alloy during each cycle will be constant (assuming equilibrium is established) until sufficient uranium has been transferred so that the amount remaining is completely in solution. If the uranium has a limited solubility in the acceptor alloy, the acceptor alloy will become saturated after a few cycles of the transport salt and the amount of uranium back-transferred to the donor alloy will reach a constant value. This constant value is lower with a saturated acceptor alloy than 1t would be if the amount of uranium in solution in the acceptor increased with each cycle of the transport salt. The overall effect of limited solubility in the acceptor alloy is to decrease the number of transport salt cycles needed to transfer a given fraction of uranium. The maximum possible fraction transferred may also be increased by limiting the uranium solubility in the acceptor alloy. Thus, in Eq. 14, the term (PA ™0)/(PD ®A) is multiplied by the solubility of uranium in the acceptor alloy expressed as a fraction of the uranium initially charged to the system. A low solubility of uranium in the acceptor alloy can be used to compensate for a large value of the distribution coefficient, Dy The fraction of the uranium initially present in the transport salt-donor alloy system that may be transferred to the acceptor alloy may be increased if the uranium present initially exceeds the sclubility. For example, in a system in which a maximum transfer of 977 is possible when the donor alloy is initially just saturated, the amount transferred may be increased to 997 by Increasing the amount of uranium present to five times the solubility limit. However, the number of transport salt cycles required is increased from about 13 for the case of 97% transfer to about 21 for 99% transfer. The same principles that govern the rate of transfer of uranium and plutonium apply to the impurities that are present in the donor alloy. Impurities for which the value of F (Eq. 11) is small compared with the value of F, for uranium or plutonium are separated by the salt transport process. Generally these impurities are those for which the value of D is several orders of magnitude smaller than the value of Dp for uranium or plutonium. An estimate of the decontamination factor, D.F., possible with the salt transport process can be obtained from the expression - % (uranium or plutonium) (15) ) (impurity) D.F. The fraction of uranium or plutonium transferred, ¢ (uranium or plutonium), will be approximately unity, i.e., essentially all will be transferred. For small values of FD’ the quantity ¢ 345 (impurity) (Eq. 13) is approximately equal to sF,\, where s 1is the number of cycles needed to transfer the desired fraction of uranium or plutonium, and Ffl 1s the fraction of impurity transferred from the donor. Substitution in Eq. 15 gilves 1 D.F. = = (16) s s(DD o X+ 2) which summarizes most of the important factors that influence the degree of decontamination obtainable with the salt transport process. The D.F. is inversely proportional to the number of transport salt cycles, s, or the amount of salt circulated, to achieve the desired uranium or plutonium transfer. It is also seen that the fraction of donor alloy entrained with the transport salt must be smaller than D! 28 X if the maximum possible decontam- ination is to be achieved. The small values of Dfi for the nobler fission products ‘(see K, values in Table I) indicate that the decontamination factor in practical systems will depend upon the extent to which entrainment of the donor alloy with the transport salt can be eliminated. Molten Salt-Liquid Alloy Systems for the Salt Transport of Uranium and Plutonium As indicated earlier, the rate of uranium or plutonium transport from a donor alloy to an acceptor alloy depends upon (1) their distribution coefficients between each alloy and the transport salt and (2) their solubilities in the two alloys. Therefore, distribution coefficients and solubilities of uranium and plutonium in various systems of process interest were investigated experi- mentally., The distribution coefficients of uranium and plutonium between molten MgCl, and liquid Cu-Mg and Zn-Mg alloys at 800°C and molten MgCl,-30 mol % NaCl-20 mol %Z KCl1l and liquid Cu-Mg, Cd~-Mg and Zn-Mg alloys at 600°C are shown in Figs. 2 and 3, respectively. 1In general, as the magnesium content of the alloy is increased from a near-zero initial value, the distribution coefficients decrease at first, pass through a minimum, and then increase. An exception to this generalization is the distribution of plutonium between the ternary salt mixture and Cu-Mg alloys at 600°C. The values of the distribution coefficients shown in Figs. 2 and 3 are nearly independent of the plutonium or uranium concentration in the alloy. A discussion of the dependence of the distribution coefficient on the ma%nesium content of the alloy has been presented elsewhere, (15) The solubilities of uranium in liquid Cu-Mg and Zn-Mg alloys and plutonium in liquid Zn-Mg alloys at 800 and 600°C, respectively, are shown as a function of the magnesium content of the liquid alloy 346 L¥E ) mol % IN SALT at. % IN METAL DISTRIBUTION COEFFICIENT (D' to! ) © 107! 1072 o3L Lt 1 ] 1] PPu 4] 10 20 30 40 S50 80 70 8C 90 1100 MAGNESIUM CONTENT IN ALLOY, ot % 2. Distribution of U and Pu between MgCl, Salt and Cu~-Mg and Zn-Mg Alloy§ at 800°C. ) mol % IN SALT at % INMETAL DISTRIBUTION COEFFICIENT (D s oel L L] 0 1 20 30 40 50 60 70 B8O 90 100 MAGNESIUM CONTENT IN ALLOY, ot % 3. Distribution of U and Pu between 50 mol % MgCl;-30 mol % NaCl- 20 mol % KCl Salt and Cu-Mg, Cd-Mg, and Zn-Mg Alloys at 600°C. in Figs. 4 and 5. The solubility of uranium in liquid Cd-Mg alloys at 600°C is also shown in Fi%. 5. Values of the solubility of plutonium in liquid Cu-Mg alloys 16) are so large that solubility is not an important factor in limiting the use of Cu-Mg alloy as a donor for plutonium., A discussion of the dependence of the solu~ bility of uranium and plutonium in liquid magnesium alloys appears elsewhere. The amount of uranium or plutonium that can be transferred in each cycle of the transport salt between the donor and acceptor alloys depends upon the amount of salt transferred and the uranium or plutonium content of the salt. At equilibrium, the uranium or plutonium content of the transport salt is the product of the uranium or plutonium content of the alloy and the distribution coefficient. mol Z M (salt) = at., Z M (metal) x D (11) To obtain a large uranium or plutonium content in the salt equilibrated with the donor alloy, both the solubility and distribution coefficients in the donor alloy-salt system should have large values, Conversely, to obtain a small uranium or plutonium content in the salt equilibrated with the acceptor alloy, both the solublility and distribution coefficients in the acceptor alloy-salt system should have small values. Mass transfer of uranium or plutonium and magnesium between the donor and acceptor alloy stops when the equilibrium uranium or plutonium content of the transport salt is the same above both alloys. The maximum uranium and plutonium content of (1) molten MgCl in equilibrium with saturated liquid Cu-Mg and Zn-Mg alloys at §00°C and (2) MgCl,-30 mol % Nacl-20 mol % KCl in equilibrium with saturated liquid Cu-Mg, Zn-Mg and Cd~Mg alloys at 600°C is shown in Figs. 6 and 7, respectively. These curves may be used to determine the compositions of donor and acceptor alloys for uranium and plutonium, For example, at 800°C, only a low magnesium (v16 at, %) content Cu-Mg alloy would be a practical donor for uranium, while Zn-Mg alloys with either low (15 at. %) or high (>60 at. %) magnesium contents would be practical acceptor alloys for uranium. The compositions of the most promising donor and acceptor alloys of those studied for uranium and plutonium are summarized in Table II. At 800°C the Cu-16 at. % Mg alloy is a donor for both uranium and plutonium, the Mg-20 at. % Zn alloy is a donor for plutonium and an acceptor for uranium, while the Zn-10 at. %Z Mg alloy 1in an acceptor for both uranium and plutonium. At 600°C none of these systems is a very good uranium donor, but both the Cu-43 at. % Mg and the Mg-20 at. % Cd alloys are plutonium donors while the Zn-15 at, % Mg alloy is an acceptor for both uranium and plutonium. It is also evident from Table II that the plutonium donor alloys are more effective donors than the uranium donor because of the relatively low solubility of uranium. 348 6%¢ 10! Pu IN Zn-Mg L] 109 10! 102 URANIUM OR PLUTONIUM CONTENT IN ALLOY, ot % o3l 111 11 11 O 10 20 30 40 50 60 70 80 90 100 MAGNESIUM CONTENT IN ALLOY, ot % 4. Solubility of U and Pu in Cu-Mg and Zn-Mg Alloys at 800°C. URANIUM OR PLUTONIUM CONTENT IN ALLOY, ot % 107'¢ 10-2 10-3 1074 1 1 | Pu IN Zn-Mg i 1 1 | o] 10 20 30 40 50 60 70 L 80 90 MAGNESIUM CONTENT IN ALLOY, at % 100 Solubility of U and Pu in Cu-Mg, Cd-Mg, and Zn-Mg Alloys at 600°C. 0S¢ = DX SOLUBILITY) e e ol 102 In-Mg URANIUM OR PLUTONIUM CONTENT IN SALT (mol % 10-3 i — I 1 I | | L1 o] i 20 30 40 50 €0 70 80 90 100 MAGNESIUM CONTENT IN ALLOY, of % 6. Maximun Concentration of U and Pu in MgCl, Salt in Equilibrium with Saturated Cu-Mg and Zn-Mg Alloys at 800°C. or Pu URANIUM OR PLUTONIUM CONTENT IN SALT (mol % = D X SOLUBILITY) 102 10-3 10°4 Y T T T O T I B [+] 10 20 W 40 530 60 TO 80 90 100 MAGNESIUM CONTENT IN ALLOY, ot % . Maximum Concentration of U and Pu in 50 mol % MgCl,-30 mol % NaCl-20 mol % KCl Salt in Equi- librium with U or Pu Saturated Cu-Mg, Cd-Mg, and Zn-Mg Al- loys at 600°C. |5°1% Table II Composition of Donor and Acceptor Alloys for Uranium and Plutonium Salt Transport Hfiglz (800°C) Maximum Content D Sol. (at.Z) in salt (molZ%) Classification® Alloy Pu U Pu U PuCl3 UCl3 Pu U Cu-16 at. % Mg 3.5 0.45 high 1.05 high 0.46 Donor® Donor Mg-20 at. Z Zn 0.71 0.30 high 0.025 high 0.0075 Donor™ Acceptor Zn-10 at. 7 Mg 0.035 0.0094 2.3 1.5 0.080 0.014 Acceptor Acceptor Maximum Content 50 mol ¥ MgCl,.-30 mol 7 NaCl-20 mol % KC1l (600°C) D Sol.{at.Z}X in salt (molZ) Classificationb’c Alloy Pu U Pu U PuCl. UClg, Pu U Mg-43 at. % Cu 0.89 0.088 high 0.0025 hith 0.00022 Donor? Acceptor Mg-20 at. % Cd 0.23 0.065 high 0.0013 high 0.000084 Domor® Acceptor Zn-30 at. Z Mg 0.0014 0.0010 0.18 0.071 0.0002570.000071 Acceptor Acceptor 3uhen high plutonium solubility exists, the distribution coefficients provide the basis for evaluating the relative donor properties of the alloys. bThe values of the distribution coefficients are increased by about a factor of two by substituting MgCl,22 mol % MgF, for MgCl,-30 mol % NaCl-20 mol % KCl and by increasing the temperature from 600 to 650°C. €A1l of the salt-alloy systems shown are acceptors for the nobler metals. The donor and acceptor alloys listed in Table II may be used in various combinations to permit (1) selective separation of plutonium from the nobler metal fission products and uranium, (2) separation of uranium from the nobler metal fission products, (3) simultaneous separation of plutonium and uranium from the nobler metal fission products and (4) separation of plutonium and uranium from each other and from the nobler fission product elements. Plutonium Purification and Recovery In the Salt Transport Process for fast breeder reactor (LMFBR) fuels, plutonium is first separated from uranium and the nobler metal fission products and then the uranium is separated from these fission products in a subsequent step. The plutonium separation employs a salt transport step with a Mg-43 at. % Cu donor alloy and a MgCl,-30 mol % NaCl-20 mol % KC1l-3 mol % MgF, transport salt™ at 600°C.” To remove any entrained Mg-Cu donor alloy from the salt and to increase the degree of decontamination of plutonium from the nobler metal fission products, the transport salt is contacted with an intermediate Cd-50 at. % Mg alloy before it reaches the Zn-30 at. % Mg acceptor alloy. The intermediate alloy is a plutonium donor and an acceptor for the metals more noble than plutonium. The following is a schematic illustration of the process, + Transport Salt _é (A) Salt (B) salt |__ (C) DONOR ALLOY INTERMEDIATE ALLOY ACCEPTOR ALLOY (Mg-43 at. % Cu) (Cd-50 at. % Mg) (Zn-30 at. % Mg) Since the solubility of uranium is only about 0.0025 at. % in the Mg-43 at. % Cu alloy at 600°C, very little uranium is extracted into the transport salt. Entrainment of the donor alloy in the transport salt is likely to be a more significant source of plutonium contamination by uranium and the nobler metal fission products (see Eq. 16). The intermediate alloy is expected to remove most of the entrained donor alloy from the salt. Since this intermediate alloy is not as effective a plutonium donor as the Mg-Cu allioy, there is a tendency for some plutonium to build up during the first few cycles of the transport salt. As the number of cycles is increased, however, this plutonium is transferred to the acceptor alloy. The fraction of the plutonium in each alloy was * The NaCl and KCl serve as diluents to lower the melting point of the MgCly, and the small amount of MgF, aids the disengagement of salt and metal phases in the contacting equipment, 352 calculated as a function of the number of cycles. After 10 cycles of the salt, about 927 of the plutonium would be in the acceptor alloy, about 8% would remain in the intermediate alloy and the amount in the donor alloy would be negligible. The plutonium in the intermediate alloy represents an inventory, which is recovered as the next batch of fuel is processed. When, after many cycles of operation, it becomes necessary to replace the intermediate alloy because of the gradual buildup of donor alloy and fission products, the residual plutonium can be recovered by additional cycles of the transport salt. When the amount of plutonium in the acceptor alloy exceeds the solubility limit (0.07 at. % at 600°C), it precipitates as a plutonium-zinc intermetallic compound (probably Pu22n17).(l7) This occurs after only a few salt tramsport cycles., If the equipment that is used to contact the transport salt with the acceptor alloy can accommodate the metallic precipitate, several batches of plutonium would be transferred to the acceptor alloy. The purified plutonium is recovered as the metal by vaporization of the zinc and magnesium at reduced pressure. If the contacting equipment cannot handle the precipitated metal, continuous means will be required for removing the acceptor alloy from the contactor, vaporizing the zinc and magnesium and returning them to the contactor. The optimum composition of the acceptor alloy depends somewhat upon the mode of operation of the contactor, since the magnesium content of the alloy decreases as plutonium is trans- ferred. However, this effect (see Eq. 8) is not large enough to interfere seriously with the operation if provisions are made for periodic additions of magnesium to the acceptor alloy. Uranium Purification and Recovery In the Salt Transport Process for IMFBR fuels,(ls) uranium 1s separated from the nobler metal fission product elements by the use of a Cu~-16 at. % Mg donor alloy, a molten MgCl, transport salt, and a Mg-17 at. % Zn acceptor alloy at a temperature of 800 to 900°C. The operation of this donor alloy may be illustrated by referring to the copper-rich region of the Cu-Mg-U system as shown in Fig. 8, This representation must be regarded as tentative as it was constructed from the Cu-U and Cu-Mg binary diagrams and limited data from the Cu-Mg-U ternary system. Point M is an eutectic in the Cu-U binary alloy. Point C is an eutectic in the Cu~-Mg binary alloy. Point N is estimated to be the ternary Cu-Mg-U eutectic. Hence, the curve M-C-N describes an eutectic valley. Point D is the composition of the liquidus of the Cu-Mg binary at 800°C. Point C is the 800°C liquidus on the eutectic valley, and C-D 1s the 800°C liquidus 1sotherm between the eutectic valley and the Cu-Mg binary. Line A-B is an operating line that is the path that the bulk composition of the donor alloy will 353 URANIUM, at.% (ucus) Yiosoec \ \ \ \ { I b 5 MAGNESIUM, at.% 8. Tentative Liquidus Surface for the Copper-rich Region of the Cu-Mg-U Phase Diagram 354 follow as the transport of uranium takes place. The line represents the change in bulk composition that results from depletion of the uranium by MgCl, oxidation and the resultant buildup of magnesium in the donor alloy. These composition changes are related by the equation U+ 3/2 MgCl, + UCl, + 3/2 Mg (18) which indicates that 1.5 moles of magnesium are introduced into the donor alloy for each mole of uranium removed. Any peint, E, lying on the operating line represents an initial bulk composition of the donor alloy. Point B is the bulk composition of the donor alloy upon completion of uranium transport. At point E the equilibrium phases present are (1) a solid phase containing uranium, which is believed to be the intermetallic compound, UCug, (2) a Cu~-Mg solid solution, and (3) a liquid phase of composition C. As uranium transport proceeds, the bulk alloy composition moves along the operating line, A-B, from point E toward point C. The compound UCu5 dissolves, replacing the uranium extracted from the liquid donor alloy by the transport salt: UCus(solid) + U(Cu~-Mg liquid) + 5Cu (Cu-Mg liquid) (19) The copper released by dissolution of UCug and Cu-Mg solid solution combines with the magnesium that is introduced by uranium reduction of MgClz to produce additional liquid of the eutectic composition (point C) at the operating temperature Cu(solid solution) + Mg(Cu-Mg liquid) -+ Cu-Mg(Cu-Mg eut.). (20) Thus, the composition of the liquid phase remains constant as the ratio of the liquid to solids increases. When point C 1s reached, all of the solids are consumed. As the transport of uranium continues, the bulk composition follows the operating line from point C to point B, and Cu-Mg solid solution is formed. The composition of the liquid phase in equilibrium with the Cu-Mg solid solution 1s represented by the 800°C 1iquidus isotherm, C-D. During this last phase of uranium transport, all of the uranium in the donor alloy is in solution. To control the magnesium buildup in the donor alloy during uranium transport, the operating line should pass through the liquidus on the eutectic valley at the operating temperature. This liquidus composition on the eutectic valley is the compositiocn the liquid phase will have as UCug and Mg-Cu solid solution dissolve, Under the above conditions the system provides (1) control of magnesium buildup in the donor and (2) operation at optimum donor compositions. The optimum donor compositions provide maximum uranium solubility and the highest uranium distribution coefficient at the 355 specified operating temperature during the transport of uranium. Multiple Separations It is possible in principle to effect multiple separations by using three or more liquid alloys in combination with the transport salt. For example, plutonium, uranium and the nobler metals might be separated from one another by a salt transport procedure in which the plutonium and uranium are collected in separate acceptor alloys. This separation can be accomplished by using (1) a Cu-low Mg donor alloy, (2) a Cd-high Mg plutonium donor-uranium acceptor alloy. At 800°C the values of the plutonium solubility and distribution coefficient are such that the plutonium content of the recycled transport salt leaving the Zn-Mg acceptor alloy is undesirably large. It is possible, however, to lower the melting point of ‘ the donor alloy by the addition of cadmium (i.e., Cu-15 at. % Cd- 1.5 at. % Mg) and to use a MgC12—22 mol % MgF, (m.p. 626°C) transport salt instead of MgCl, (m.p. 716°C) to permit operation at 700°C. The Cu-Cd-Mg alloy is a satisfactory donor for both plutonium and uranium and it retains the nobler metals. The Cd-Mg alloy is a plutonium donor, but it 1s an acceptor for uranium, primarily because of its low solubility., At 700°C, beta uranium metal is precipitated in this alloy. The plutonium collects in the Zn-Mg acceptor alloy, and if its solubility is exceeded it precipitates as a plutonlum-zinc intermetallic compound. Multiple donor and acceptor alloys can also be used to increase the number of stages of a salt transport separation. In the Salt Transport Process for LMFBR fuels the use of an additional donor alloy in the plutonium salt transport step greatly increases the decontamination of plutonium from the nobler metal fission products, Experimental Investigations of ‘ Salt Transport Separations Several laboratory-scale experiments were performed to investi- gate the feasibility of salt transport separations. The apparatus consisted of two adjacent cylindrical tantalum crucibles which were welded together with a low common wall (Fig. 9) so that the transport salt could be poured back and forth between the donor and acceptor alloys, Dams at the upper edges of the crucibles prevented overflow of the salt when the crucible was tilted to pour the salt from one side to the other. The charge to the donor crucible was 87.0 g of copper shot, 43.0 g of magnesium rod and 20.92 g of plutonium metal, The initial charge in the acceptor crucible was 178.6 g of zinc and 12.5 g of magnesium rod. The transport salt, 178.6 g of purified MgCl,-30 mol % NaCl-20 mol % 356 o] b ° N -=-TANTALUM AGITATORS PADDLE DIMENSIONS, lin. WIDE BY % in. HIGH c A S * u ) 7 € 2n. SALT o £ -l O " DOUBLE CRUCIBLE 7z SALT IN POUR POSITION | | Ign Mg 44&7} % Izn.Zn |2&19% i 9. \4 BAFFLES, 2in BY 3in HIGH Dual Crucible Used in Salt Transport Experiments 357 KCl, was charged to the acceptor vessel. The system was heated to 600°C and the salt was poured from the acceptor side to the donor side and back again a total of 20 times. The salt and metal phases were equilibrated for 30 min, at an agitation rate of 500 rpm and allowed to settle 15 min. before the salt was poured. After 20 cycles of the salt, the plutcnium concen- tration in the donor alloy was 0.181 wt %, or about 1% of the total plutonium, which indicates that about 99% had been transferred to the acceptor alloy. The plutonium content of the salt was negli- gible. Some copper (0.1 wt %) was found in the acceptor alloy, probably as a result of donor alloy entrained in the salt. These results showed that plutonium can be transported satisfactorily. The salt transport concept was also tested in a laboratory experiment %8 which irradiated cerium was used as a stand-in for plutonium.( ) The donor alloy was 60 g of Mg-44 at. 7 Cu con- taining 2.5 g of irradiated cerium and the acceptor alloy was 100 g of Zn-12.4 at. % Mg. The transport salt, 180 g of MgCl,- 30 mol Z NaCl-20 mol % KCl, was added to the donor alloy initially. The dual crucible technique was used at a temperature of 600°C. The metal and salt phases were contacted for 60 min. with agitation at 300 rpm and allowed to settle 30 min. In each cycle, 80Z of the salt was transferred. Five cycles of the transport salt resulted in 897 transfer of the cerium. The significance of this experiment is that the rate of cerium transport was in agreement with that predicted from equilibrium values of the solubilities and distribution coefficients. Five larger-scale experiments were perfeormed to investigate the behavior of uranium in a salt transport separation. 20 The feed charges, containing from 2 to 5 kg of uranium, were prepared either by reducing unirradiated oxides of uranium and nobler metal elements (Zr, Mo, Pd, Ru, and Nb) with Cu-Mg alloy or by direct addition of these metals to the Cu-Mg alloy at 925°C. 1In either case, most of the uranium at the operating temperature of 845°C was present as a finely divided precipitate. The donor alloy was Cu-12 at. % Mg, the acceptor was Zn-83 at. % Mg and MgCl, was used as the transport salt. Both alloys were contained in tungsten crucibles and the salt was cycled between the alloys by pressure-siphoning through a heated transfer line. Prior to each salt transfer the salt and metal phases were mixed for 4 to 6 min. and then permitted to settle for 5 to 10 min. After about 99% of the uranium had transferred to the acceptor alloy, the supernatant liquid alloy was transferred away from the precip- itated uranium bed, and the bed was washed with Mg-14 at. % Zn alloy. A uranium recovery of about 99% was achieved after 27 cycles of the salt. The overall removals of the nobler metals were 99.7% for zirconium and 99.9%Z or more for the others. A separation of plutonium from uranium and the nobler metal 358 fission product elements was demonstrated in a study in which the steps of the Salt Transport Process for LMFBR fuels were performed in sequence. 21 The original feed to the series of process steps was 199 g of plutonium as PuO,, 400 g of uranium as uo, and oxides of 24 representative fission product elements. In the salt transport step the plutonium was transferred at 600°C from a Mg-44 at. % Cu donor alloy to a Zn-5.2 at. % Mg acceptor alloy, using a MgCl,-30 mel % NaCl-20 mol 7 KC1l transport salt. The pressure-siphoning technique was used to cycle the salt., After 14 cycles, 99.3% of the plutonium had been transported. Various other salt transport experiments have been conducted, including ones in which the salt was cycled continuously between donor and acceptor alloys. 2) oOn the basis of mass-transfer rates, stage efficiencies and salt-metal phase disengagement results that have been obtained, further technological development of the salt transport procedure is expected. Conclusions The salt transport technique is being developed primarily as a means of separating plutonium, uranium and the nobler metal fission products in a pyrochemical process for the recovery of LMFBR fuels, To provide a basis for selecting suitable liquid metal and salt solvents, basic chemical data were obtained on the solubilities of plutonium and uranium in liquid alloys of magnesium with copper, zin¢ and cadmium and on their partitioning behavior between these alloys and molten salts containing MgCl,. It was found in both laboratory and bench-scale engineering experiments that the equilibrium solubility and distribution data can be used reliably to predict the behavior of a practical salt transport system for plutonium or uranium. Preliminary engineering studies have indicated that the procedure shows promise for technological development and it has been inccrporated into the engineering investigations for the Salt Transport Process for IMFBR fuels, Although the current development of salt transport separations is concerned mainly with fuel reprocessing, there are numerous cther applications of potential interest. For example, some work has been done on a process of this type for the recovery and purification of plutonium-238 from recycle scrap material, (23) 1t might also be applied to the processing of other high-value actinide metals (e.g., neptunium and curium) as well as other more conven*= tional metals such as the rare earths. 359 Acknowledgments The authors wish to acknowledge the technical assistance of J. W. Walsh, K. R, Tobias, R. Tiffany, R. D. Wolson and J. D. Schilb. Dr. R. P. Larsen and his associates provided valuable consultations and supporting analytical werk. 8. References Feder, H. M. and I. G. Dillon, "Pyrometallurgical Processes", in Reactor Handbook, Vol. 2, Fuel Reprocessing, 2nd ed., p. 313, S. M. Stoller, R. B. Richards (eds.), Interscience Publishers, Inc., New York (1961). Motta, E. et al, '"Pyrometallurgical Processes: Process and Equipment Development', TID-7534 (2), p. 719 (1957). Schraidt, J. H. and M. Levenson, '"Developments in Pyro- metallurgical Processing', in Progress in Nuclear Energy, Series I1II, Process Chemistry, Vol. 3, p. 329, F. R. Bruce, J. M, Fletcher, H. H. Kyman (eds.), Pergamon Press, New York (1961). Martin, F. S. and G. L. Myles, "The Principles of High-Temper- ature Fuel Processing', in Progress in Nuclear Energy, Series III, Process Chemistry, Vol., 1, p. 291, Pergamon Press, New York (1956). Burris, L., Jr. et al, "Recent Advances in Pyrometallurgical Processes", Trans. Amer. Nucl. Soc. 4 (2), 192 (1961). Lawroski, S. and L. Burris, Jr., "Processing of Reactor Fuel Materials by Pyrometallurgical Methods!, At. Energy Rev. 2 (3), 3 (October 1964). Pierce, R. D. and L. Burris, Jr., 'Pyroprocessing of Reactor Fuels, Reactor Technology, Selected Reviews", L. E. Link, Ed., TID-8540, p. 711 (1964). Burris, L., Jr., et al, "Pyrometallurgical and Pyrochemical Fuel Processing', in Proceedings of 3rd International Conference on Peaceful Uses of Atomic Energy, Vol., 10, p. 501, International Atomic Energy Agency, Geneva (1965). Dwyer, O. E., "Process for Fission Product Removal from Uranium-Bismuth Reactor Fuels by Use of Fused Salt Extraction', J. AIChE 2, 163 (1956). 360 10. 11. 12. 13. 14. 15l 16l 17. 18. 19, 20. Bareis, D. W.,, R. H. Wiswall and W. E. Winsche, "Processing of Liquid Bismuth Alloys by Fused Salts', Chem. Eng., Progr. Symp. Ser. 50, 228 (1954). Dwyer, 0. W., R. J. Teitel and R. H. Wiswall, "High Tempera- ture Processing Systems for Liquid Metal Fuels and Breeder Blankets", in Proceedings of International Conference on Feaceful Uses of Atomic Energy, Vol. 9, p. 604, IAEA, Geneva (1955). Wiswall, R. H. et al, "Recent Advances in Chemistry of Liquid Metal Fuel Reactors', in Proceedings of 2nd International Conference on Peaceful Uses of Atomic Energy, 17 428 (1958). Vol. 17, p. 428, IAEA, Geneva (1958). Chiotti, P., Regeneration of Fission-Product-Containing Magnesium-Thorium Allovs, U. S. Patent 3, 120, 435, Feb. 4, 1964. Chiotti, P. and J. S. Klepfer, 'Transfer of Solutes Eetween Liquid Alloys in Mutual Contact with Fused Salt. Application to Fuel Reprocessing', Ind. Fng. Chem. Process Design Develop. 4 (2), 232 (1965). Johnson, I., "Partition of Metals between Liquid Metal Solutions and Fused Salts", in Applications of Fundamental Thermodynamics to Metallurgical Processes, G. R, Fitterer, Editor, Gordon and Breach Science Publ., New York, 1967, pp. 153-177. Knoch, W., Argonne National Laboratory, Private Communication. Johnson, I., "The Solubility of Uranium and Plutonium in Liquid Alloys", This Symposium. Steunenberg, R. K., R. D. Pierce and I. Johnson, 'Status of the Salt Transport Process for Fast Breeder Reactor Fuels', This Symposium. Knighton, J. B. and P, J. Mack, "Salt Transport Studies", Chemical Engineering Division Semiannual Report, January- June 1967, ANL-7375 (October 1967), p. 20. Walsh, W. J. et al, "Engineering Studies of Salt Transport Separations", Chemical Engineering Division Semiannual Report, July-December 1966, ANL-7325 (April 1967), p. 26. 361 21. 22' 23. Winsch, I. 0., W. J. Walsh and T. F. Cannon, "Exploratory Plutonium Experiment", Chemical Engineering Division Semi- Annual Report, July-December 1967, ANL-7425 (May 1968), p. 30. Pierce, R. D., "Liquid Metal-Molten Salt Contactors', Reactor Development Program Progress Report, January 1969, ANL-7548 (February 1969), p. 109. Nelson, P. A. et al, "Pyrochemical Purification of Plutonium-238", Chemical Engineering Division Annual Report, 1968, ANL-7575, (April 1969). 362 THE REDUCTIVE EXTRACTION OF PROTACTINIUM AND URANTIUM FROM MOLTEN LiF-Ber—ThF4 MIXTURES INTO BISMUTH* R. G. Ross, W. R. Grimes, C. J. Barton, C. E. Bamberger and C. F. Baes, Jr. Reactor Chemistry Division, Oak Ridge Natiomal Laboratory, Oak Ridge, Tennessee U. S. A. Abstract Previous studies of the reductive extraction of protactinium and uranium from molten fluoride mixtures into liquid bismuth contain- ing thorium gave results which suggested this as a feasible method for processing MSBR fuels. Additional studies have been undertaken to confirm the equilibrium quotients for the extraction reactions 4t o . o 4+ Pa Paipy + Thigyy = Paggyy 4 Thipys Q= Dp/Ppy (D) 3 o = .0 M U 4,3 AU(F) + 3Th(Bi) AU(Bi) + 3Th(F), QTh = DU/DTh, (2) where D = / , M XM(Bi) XM(F) and to determine -whether these equilibrium quotients depend signifi- cantly on the concentration of Pa® and/or U° in the bismuth phase. Experiments have been conducted at 625°C + 5° in molybdenum- lined nickel containers using bismuth and single-region MSBR fuel carrier salt (LiF-BeF -ThF,, 72-16-12 mole %) with up to 0.3 mole % of UF; and >100 ppm 231pa., Thorium, uranium, and protactinium were added incrementally in order to duplicate, as nearly as possible, the varied concentrations that would exist in an extraction column such as might be used for the process. The experiments have yielded data which show a correlation of log D, vs log D, which would be expected from the stoichiometry of reactions (1) and (2). This seems conclusive proof that Th+4, Pat4 and are the important species, and that there are no sig- nificant interactions of U° and Pa® under the reducing conditions of these experiments. *ork sponsored by the U.S. Atomic Energy Commission under contract with Union Carbide Corporation. 363 Introduction The potential advantages of molten-salt-fueled reactors have been recognized for some time, and their characteristics have been reviewed by Briant, Weinberg, MacPherson, Grimes and others, (1-4) This type of reactor has been fully demonstrated at Oak Ridge National Laboratory with the successful operation for more than 10,000 equivalent full power hours of the 8 megawatt Molten-Salt Reactor Experiment, One of the principal advantages of molten-salt fuels is their adaptability to continuous reprocessing without the attendant high cost of refabrication of solid fuel elements. Thus, molten-salt- fueled reactors are particularly attractive as breeder reactors when it is desired to continuously recover 233y and 233pa bred into the fuel from 232Th. Previous studies by Shaffer, Barton, and Ferris (5-8) of the re- ductive extraction of protactinium and uranium from molten fluoride mixtures into liquid bismuth gave results which suggested reductive extraction as a feasible method for processing MSBR (Molten-Salt Breeder Reactor) fuels. A flowsheet proposed by Whatley and McNeese(9) for such processing includes an extraction column for reducing protactinium and uranium from the fuel salt. In this system the salt stream enters the bottom of the columm and flows counter-currently to a bismuth stream containing thorium and the reduced metals. Since protactinium is intermediate in nobility be- tween uranium and thorium, the protactinium refluxes in the middle of the column where the salt builds up the highest protactinium concentration., At this point in the column a hold-up tank is pro- vided in order to allow the 233Pa to decay to 233U in a region away from the neutron flux of the core. Thus, high protactinium concentrations in the core which would adversely affect the breed- ing ratio are avoided. While the feasibility of such a process had been indicated,(s's) i1t was necessary to confirm the equilibrium quotients for the ex- traction reactions, and to determine also whether these equilibri- um quotients depend significantly on the concentrations of Pa® and/or U° in the bismuth phase. Theory The extraction reactions are of the general type +ni o] e o] +ny L ME oM Men = hen T e 0 @ Assuming constant activity coefficients, the equilibrium quotient, G, may be represented as 364 " 1”2 g2t Ql = , (2) M n 2 [m 11°2 w17 or (Dy; )2 it . , (3 M n 2 (DMZ) 1 where XM D, = i-MiB—ll = Distribution Coefficient. (F) X denotes mole fraction, and subscripts F and Bi denote fluoride and bismuth phases, respectively. The assumption of constant ac- tivity coefficients seems reasonable since the species involved are present in the bismuth phase at very low concentrations and are in the fluoride phase as either low concentration solutes or major components whose concentrations would not be affected significantly by the extraction equilibria. When the distribution coefficients are known, Q may be calculat- ed from the relationship in Eq. (3) if the valence states of the species in the fluoride phase are known, or by using the log form, 71 log D, =-— 1log D, + C, (4) Ml n2 M2 the ratio of the valence states may be determined from the slope of a log-log plot of the distribution coefficients. Experiments were conducted in which the distribution coefficients were determined at various metal specie concentrations, Experimental The general experimental technique involved equilibrating a molten fluoride mixture (LiF-BeFp-ThF;) containing Pa and/or urani- um with bismuth containing thorium. Since Pa concentgations great- er than 100 ppm were involved in these experiments, lpa was used in order to avoid the high gamma radiation level associated with 3pa. However, the 231Pa, being an alpha emitter similar in tox- icity to 2 9Pu, required glove box containment for the experiments. The glove boxes used have been described in detail by Barton(6), The materials were purified prior to thorium addition by pro- longed (24 hours) treatment with H,-HF (10:1) followed by reduction with Hy alone. The Hy was stripped out, and equilibration was af- fected in an argon atmosphere. A schematic diagram of the equilib- ration system is shown in Figure 1. The major components of the system are a diaphragm type circulation pump, a flowmeter for 365 PURIFIED H,,A, He OR HF _ FROM GAS HEADER "7 {ALSO VACUUM) FLOW CONTRuUL © NEEDLE VALVE 0 - RING \ ) CONECTIONS PUMP o L | FLOWMETER DIP-LEG BYPASS PURIFICATION / TRAPS (PYREX BALL VALVE OR METAL) SAMPLING PORT O—RING CONNECTIONS /TO Mc DIP—LEG NaF 0-RING ) 4 NICKEL FRIT CONNECT ION\ FILTER REACTION VESSEL NaF Hg MANOMETER 1Mo LiNED) OFF GAS SALT 8i 1. Schematic Diagram of Molten Salt-Bismuth Equilibration System with Gas Recirculation. . 366 monitoring the rate, the reaction vessel, and traps containing sodium fluoride and magnesium perchlorate to remove any traces of HF and water which may be present. During the normal course of an experiment, the valves from the supply header and to the off-gas were closed. High-purity argon was recirculated thru the dip-leg. The dip-leg by—-pass was open at times such as during sampling or additions in order to avold the possibility of forcing melt up the dip-leg. A schematic detail of the reaction vessel is shown in Figure 2. In a typical experiment this vessel contained about 200 grams each of salt and bismuth, The carrier salt used in all experiments re- ported here was a 72-16-12 m/o mixture of LiF-BeFy-ThF,, which is being considered as the fuel carrier for a single-region MSBR. The bismuth was of six-nine purity, according to the manufacturer's analysis. The mixture was contained in a molybdenum lined nickel or steel vessel at a temperature of 625°C + 5°. The molybdenum dip- leg was designed to lift the bismuth and discharge it into the salt phase, since bench tests conducted in silica apparatus revealed that simple gas bubbling merely rocked the salt-bismuth interface. An assembled reaction vessel, along with a sampler of the type used is shown in Figure 3. The vessel is 1-1/2 inches in diameter and 8 to 12 inches high. The total length from the ball valve is about 20 inches. Samplers of this length containing a sintered metal frit in the bulb were fabricated from copper, molybdenum or steel. These samplers were inserted thru a teflon seal secured above the ball valve, and the experiment was sampled as desired. The material was removed from the sampler, weighed, placed in a vial, and brought out of the glove box train thru a bag-out port. The vial dropped into a plastic tube which was sealed with a thermal sealer. The alpha contaminated sample could then be safely handled outside the glove box. It was gamma scanned for 233Pa tracer or transmitted to analytical for other analyses. Procedure and Data Equilibrium distribution data were obtained in two slightly dif- ferent types of experiments. In the first type, purified carrier salt containing PaF, and UF, was brought into contact with bismuth containing enough thorium to reduce all the protactinium and urani- um and leave about 1500 ppm thorium in the bismuth. The initial salt concentrations were: UF4 - 0.2 to 0.3 mole Z and Pat% about 90 to 125 ppm. The distribution of these solutes and thorium be- tween the salt and bismuth phases was determined by analysis of filtered samples of the two phases. Figure 4 shows protactinium and thorium concentrations for an experiment of this type as a function of contact time. This experiment was extended for about 367 rrzTT N /Molten Salt 7. Liquid 84 LR AT TR R R LSRR LAY m//// AULRALLALRY ARRLALLY ;;;;;;;;;; 222l It 4 H H ‘ 4 " H H H " H H H K 4 o e Mo [ ¢ I 7 4 4 4 4 # I # 1 I ri ’ 2 1 ” 4 ’ # I 2 I [ ¢/ ] 5 z 4 # ’ 7 4 K [ kA I 2 4 4 [ 4 4 4 /4 4 % . [ [4 4 [4 I [ [ £ ¢ [ Mo ST ITITOTI OIS ITIT T TS Schematic Detail of Molten Salt-Bismuth Reaction Vessel. 2. 368 GAS OUTLET CONTROL. TEMPERATURE 3 : FIiLTER STICK GAS WNLET EXPERIMENTAL TEMPERATURE . :<:_- ‘i" e H?% Z 3. Photo of Reaction Vessel Assembly with Sampler. 80 [+] 0 pe— Pg I[N BISMUTH 1800 a 1 Q a 3 ( g o 2.9 () g‘b _E‘ = 60 1700 '5 > = 2 2 = 50 1600 - z '— o o & 40 150C + ('8 o % o \ Th IN BISMUTH— > Z 30 b—db - b 400 o o i = 2 & = 20 1300 £ 4 L L $=— Po IN SALT. § 8 10 - 1200 8 Q O L 0 20 40 60 80 100 120 140 160 CONTACT TIME (hr} 4, Protactinium and Thorium Concentrations as a Function of Contact Time Between LiF-BeFy-ThF, (72-16-12 m/o) and Bismuth at 625°C, 369 one week in order to demonstrate the stability of the system. The concentrations held essentially constant for this period. The uranium distribution is not shown on the plot; however, it stayed constant with about 987 of the uranium in the bismuth and less than 2% in the salt. The material balances in this experiment were about 100% for uranium and 95% for protactinium. The second type of experiment involved initial purification of the salt and bismuth together. The protactinium and/or uranium were added prior to hydrofluorination of the melt. Crystal bar thorium was then added in small increments after sparging with Hp and stripping with argon. By sampling the phases after each thor- ium addition, distribution data were obtained at various concen- tration levels. Figure 5 shows data obtained in an experiment of this type. The protactinium concentration is plotted as a function of thorium added. Uranium was also present in this experiment, and during the early thorium additions uranium was being reduced while the protactinium concentration held constant. Toward the end of the experiment thorium saturation of the bismuth was indicated by loss of protactinium from the bismuth on further addition of thor- ium. Co-precipitation of thorium-protactinium bismuthide probably occurred. This was indicated by the thorium analysis and by the fact that the protactinium reappeared in the bismuth phase when the temperature was raised to 805°C. In some of these '"titration'" type experiments, in which the re- ductant was added incrementally, protactinium was reduced in the absence of uranium. These data are included in Figure 6 where protactinium distribution is plotted as a function of thorium concentration in the bismuth. Plots of thorium concentration are equivalent to plots of thorium distribution, since thorium in the salt is essentially constant. The data are plotted on a linear scale in this case in order to minimize the effect of analytical scatter at very low thorium concentration. The closed circles show data obtained in the absence of uranium. The 805°C line was drawn to the single point obtained when the tem- perature was raised as mentioned previously. The 625°C data for protactinium and thorium are represented very well by a straight line; uranium concentration has no apparent effect on the distribu~- tion. The failure of the representative line to pass thru zero is attributed to trace amounts of oxidizing impurities which reacted with some of the thorium. Figure 7 shows the 625°C for protactinium and uranium distribu- tion. 1In this case a log-log plot is used, and a straight line representative of this data has a slope of 4/3. 370 80 | ! FILTERED SALT AT 630°C » FILTERED BISMUTH AT BO5°C 4%2 50 - e FILTERED SALT AT 805°C o2 FILTERED SAMPLES AT 830°C PROTACTINIUM CONCENTRATION (ppm) 40 AFTER EQUILIBRATION AT 805 A 30 / \ ® 20 A} 10 — —kz FILTERED BISMUTH AT GW ) . * i | oo : 0 04 o8 12 16 20 TOTAL THORIUM ADDED (q) 5. Protactinium Concentration as a Function of Total Thorium Add- ed to LiF-—Ber-ThF4 (72-16-12 m/o) and Bismuth Solutions. 0 T % S N N Y R o 8 - 1 e 4 URANIUM PRESENT ;//y | 16 R e R Y A — [ [ | I / [ ‘ | I | 1 S — | 7 | ! - | . | |l — LA | e o ot 10 ! i A t f f 1 | (e25°c) /> ' [ ‘ | [ 1 8 L ¢ T 1 — [ / | T l I/IL" 6 Jf //. i 4 J —;%1’:'7 | L7 R ol s [0} CO xy 0% / ¥\ o0 -80 N / fl't\' | -100 - / / L | 1 i L ] 10°6 104 10”2 1O 102 104 108 RATIO OF THE MOLE FRACTION OF METAL CHLORIDE TO ACTIVITY OF METAL 1. Chlorine Potential Diagram at 1000°K. towards reduction, even when due allowance is made for the pose sible dissolution of the metal in its own halide(12). (ii) The stebility of PuClz towards disproportionation is expected to be greater than t of UClze In initisgl stability tests, melts containing additional UClz in place of PuCl; may therefore be used, sc avoiding in the early stages of investigation the complication of working with plutonium. (iii) The valency assumed by each of the important stable fission products at any chlorine potential may be predicted; ? few examples are shown in Figure 1. Ths valency at =70 kcal.mole™', along with the calculated yields(13 s is shown in Table 1. The nett valency of the stable fission products is rather less than the +3 of the fissioned nuclide, implying that as burnup proceeds the melt will become more oxidising. Table 1. Chlorine Balance of the Major Stable Fission Products Fission Yield(13), atoms Valency in Cl atoms Product per 100 fissions NaCl:UCl3 reacted Xe, Kr 25 0 0 Rb, Cs 19 1 19 Sr, Ba 10 2 20 Rare Earths L6 3 138 Zr 22 3 66 Nb, Mo 2 0 0 Te, I 6 0 0 P4, Ru, Rh, Ag, Cd 61 0 0 Total Cl atoms reacted out of 300 available 243 8ince with properly chosen container materials the most easily oxidised material present in the system is UCl; the overall result would be an increase in the propertion of UClu in the melt. A burnup of 10% of the heavy atoms would increase the ratio of UCL, to UCl;z by 6%, which is at the predicted limit for the chlorine potential. Some method of reducing the UClh content of the melt will thus be requireds. In the gbsence of an easily oxidisable constituent such as UCl; any excess chlorine ~ however introduced = would react with the cohtainer metal. The presence of the redox system UClz/UCl, therefore provides a buffer against large sudden changes in the chlorine potential and permits close control within the range of values in which corrosion of structural materials is minimised. An initial experimental programme designed to provide data upon which an alloy corrosion testing programme would be based should therefore include: 411 (a) Development of methods of measuring chlorine potential in the chosen melt. (b) A measurement of the relative stabilities of UCl; and UCl in the melt, so defining the position of the UCl,/UClz line on the Chlorine Potential Diagram. This would involve measuring the concentration ratio of the two species as a function of chlorine potential. (¢) A similar measurement of the concentration of corrosion product chlorides for selected container metals as a function of chlorine potential. Oxide Behaviour Oxide introduced inadvertently into the fuel salt might inter= fere with normal operation either by precipitating the oxide of some component of the salt mixture or by causing enhanced corroe sion of container metals. Predictions of the behaviour of oxide in chloride melts are conveniently made by means of a modified Pourbaix D%agram gimilar to that suggested by Edeleanu and Littlewood\14), In this diegram the sbscissa is the activity of some convenient oxide = such as sodium oxide in melts containing sodium chloride « and the ordinate the chlorine potential; these may be regarded as independent varigbles. For any element the diagram is divided into areas in which one of the several possible phases is stable, e.g. metal at low chlorine potential and low oxide activity, the chloride at higher chlorine potential and the oxide at high oxide activity. The metalemetal chloride boundary is defined by equations 4 and 5 and is independent of oxide activity but dependent upon the concentration of metal chloride. The metalmmetal oxide boundary is defined by the reaction: %Clz(g) +1/z.4(8) + % Ne,0 = 1/%AOZ/2(S) + NaCl(d) 6. This boundary is dependent upon both chlorine potential and oxide activity, but is independent of the concentration of metal chloride. Its position is defined by the free energy change for reaction 6 (AFG): 20F 6 108 eya 0 = T3 RT * 218 Ppacy * 2/h'1°5(?aoz/2/aa) - log p(Cl,) Te Finally the metal chloride ~ metal oxide boundary is defined by the reaction: 412 1/z.AClz(d) + %Nazo(s) = 1/z.AOZ/2(s) + NaC1l(d) 8. This boundary is independent of the chlorine potential but depen- dent upon the activity of the metal chloride ACl_and is related to the free energy change of reaction 8 GfiFB) by? 2AF _ 8 log aNaZO = 77387 * 2/z.Log ngz/E - 2/2.10g gAClz + 2log 8yacl 9. In the same way boundaries for oxychlorides may be calculated. When an oxide or oxychloride is formed as a result of the oxida~ tion of some species initially present in reduced form (eg UO from UCl,) the oxide ~ chloride boundary is clearly dependent upon chlérine potential. In Figure 2 are plotted ths boundaries for a selection of possible salt components and container materials. In drawing the diagram the activity of the oxide has been assumed to be unity, so that the metal ~ oxide and the chloride ~ oxide boundaries define the conditions under which the oxide will appear as a separate phase. In addition it was assumed that the activities of' sodium chloride, the metal and the chloride were also unity. The manner in which the metal = chloride and the chloride «~ oxide boundaries move when the activity of the metal chloride is reduced by decade steps is indicated along the metal =~ oxide boundary. As may be expected the diagram again illustrates the stability, already seen in the chlorine potential diagram, of container materials relative to their chlorides. The following deductions on the behaviour of oxide may be made: (i) Oxide precipitation. The order in which the components of & salt mixture will precipitate when the oxXide activity is increased at any given value of the chlorine potential is readily deduced. Of all the likely constituents of a fuel melt, uranium oxide or oxychloride will be the first to precipitate. No sub= stantial increase in oxide activity is possible until the uranium has been removed from solution. The composition of the precipi=- tating phase is expected to depend upon the chlorine potential. The boundary for UOCl is shown as a broken line since its sta=- bility is uncertain. (ii) Plutonium behaviour. Even if UOCl proves to be less stable than indicated on the diagram, at all chlorine potentials greater than the lower limit of =90 keal.mole=1 imposed by the disproportionation of UCl5 uranium is expected to precipitate in preference to plutonium. “In this case also, in the initial 413 Vi kcal .mole”’ . CHLORINE POTENTIAL 2. Moo 2 -40 Mo Ni uce ' uc: Joce; Puce\” 102 H 3 Q B Fe S -60 ] PuOCe -ao - AL AN \ -100 \ - e ~26 -24 -22 -20 -18 -16 LOGIO(SODIUM OXIDE ACTIVITY) Predicted Stability of Oxides, Chlorides and Metals at 1000°K. -4 experiments uranium mey be substituted for plutonium in the melt. (iii) Oxide removal. In fluoride melts con alglng uranium, any oxide impurity is very conveniently removed (1 and indeed analysed = as water vapour by purging with a mixture of H, and HF. In chloride melts the uranium oxide and H,0/HCl boundaries are so remote that the removal of oxide with HC1l is quite impraoctical. Indeed even with high overpressures of HCl1l the melts will be very prone to hydrolysis by any water vapour present in the system. A possible alternative method of oxide removal is suggested by the diagram. The oxides of four elements -~ B, Si, Be and Al =~ are more stable relative to their chlorides than U0p. Introducing one of these chlorides into the melt should therefore remove oxide by preclpltation. All these chlorldes are volatile (boiling points 15° c, 60° Cs 520°C and 480°C respectively). The problems in introducing corrosive vapours into the melt, of maintaining an adequate concentration in the melt and of removing the excess reagent after filtration make the scheme unattractive but not impossible. (iv) Protection of uranium from precipitation. In a melt containing the chlorides of both girconium and uranium at com= parable activity the two oxides are expected to precipitate at about the same oxide activity. A similar situation spplies in the MS.R.E. fluoride fuel salt, where prior precipitation of 2r0, is ensured through ge Mass Action Law by means of a sube~ stantial excess of ZrF (7 With the high concentration of UClL required in the fast reactor chloride fuel salt such an artifice is not possible. The four halides discussed above would fulfil a similar function, but here also their volatility would present considerable problems in maintaining an adequate concentration in the melt. (v) Corrosion. If oxide is added to a melt containing no chlorides which form stable oxides, the sodium oxide activity increases to a point = still far removed from saturation in sodium oxide = at which the oxide of one of the container metals, Ni, Fe, Mo etc., becomes the stable phase. Oxide impurity in such melts is notorious in causing corrosion of otherwise resistant metal containers. The presence of uranium, however, ensures that the activity of sodium oxide is held at such a low value that the activity of corrosion product oxides is also extremely low, far removed from separation as discrete phases. Corrosion of con= tainer metals by oxide formation is therefore expected to be negligible and the major single factor in determining the extent of corrosion remains the chlorine potential of the melt. Thus the presence of a substantial amount of uranium stabilises the melt against increases in the oxide activity as well as in chlorine potential. 4Y¥5 (vi) Oxide solubility. While a uranium-conteining phase is expected to be the first phase to precipitate on adding oxide to a chloride fuel salt, the actual concentration of oxide in the melt at saturation cannot be predicteds This is a most important aspect, since oxide will inevitably be introduced into the fuel salt during operation. The solubility of oxide then determines the rate at which the melt must be reprocessed to remove oxide for a given rate of oxygen ingress in order to avoid precipitation. In the initial studies the determination of oxide solubility in the melt over a range of conditions of chlorine potentisl and temperature is therefore of prime importance. Experimental results and discussion The composition, and in particular the concentrations of uranium and plutonium, of a chloride fuel salt would be determined by the details of the reactor design under consideration. For the present studies a melt of nominal composition NaCl 70 mole%, UCl3 30 mole% was selected as typical of the compositions which might be emplgyed. The liquidus temperature at this composition is Sgi %?8)0, close to the single eutectic at 525°C and 33 mole % 3 - Preparation and analysis of salt mixtures Salt mixtures containing UClz or UCl, are hygroscopic. In the absence of a convenient method of removing oxide from the melt, the constituents of salt mixtures were initially prepared so as to minimise the oxide content. Subsequent handling was carried out in a glove box flushed with argon dried over a molecular sieve, Experimental measurements on the molten salts were made in vacuum=tight vessels, normally of nickel, under a coVer gas of either argon purified over hot uranium or hydrogen purified by diffusion through a palladium~silver menbrane. Sodium chloride wag of analytical grade and, after drying by slowly heating to 400°C under vacuum, contasined less than 10 ppm oxygens Uranium tetrachloride was prepared in 50 g. batches by reaction of finely divided, low-fired (,.00°¢C) U0, with CCl, vapour at 450 C and purified by vacuum distillation in & horizontal silica tube at 600 Con to a liner of molybdenum foil located in the cooler part of the tube. In order to minimise the carry=over of oxide contamingtion it was found essential to insert a plug of silice wool, 1 cm. deep, close to the charge of UCl,. After two distile lations and handling in the dry box the oxide concentration was less than 0.1 wt.%. This was determined either by inert gas fusion of 30 mg. samples or by weighing the residue of U0, remaining after removing the UCl; from & 1 g. sample by vacuum distillation at 600°C. Uranium trichloride was prepared by reducing the distilled UCIh.under hydrogen in batches of up to 416 200 g. in a nickel epparatus lined w1th molybdenum foil. The bulk of the reduction was effected at 550 C. The temperature was slowly raised to 650°C and maintained until the rate of evolution of HC1 had dropped to negligible proportions, when the total HCl evolved corresponded to within Z%Fof the stoichiometric amount. The oxide level, aga%n determined by inert gas fusion or by vacuum distillation at 1000°C showed no significant increase over that of the UCIA feed material. Stability of UClk in the N’a.Cl:UCl3 The stability of UCl, with respect to UCl; was measured by determining the chlorine potential in the melt as a function of the concentration ratio of the two species. The chlorine poten- tial was measured in two ways, firstly by measurement of the equilibrium partial pressure of HC1l produced when hydrogen was passed through the melt and secondly by measuring the potential of the UCl /UCl redox electrode against a reflerence electrode of known pote%tlal with respect to the standard chlorine electrode. Melt Hydrogen Equilibration Experiments. In these experiments uranium tetrachloride, dissolved in molten sodium chloride, was reduced by hydrogen to the trichloride according to the equation: c1, (a) + 2H,() = UC1z(d) + HC1(g) From a measurement of the equilibrium partial pressure (p produced when hydrogen is passed through the melt containEg% known amounts of UCl, and UCl the equilibrium quotient for the reduction reaction (QR) s derive R = (xU013/ ey, ) Prer/’ g 1o. ? e the free energy of formetion of HC1, AFf(HC1) is well known(17) the difference in the partial molar free energies of formation of UC and UCl3 in the melt, (UCl#fUClj), is readily found: Aff(UC:LLI_-UC%) = AF (HC1) + RT1nQ 1. It is this free energy function measured in the melt, rather than the corresponding function for the pure liquids, which is more approprlately used in correlating the behaviour of UClh in the melt, as in Figures 1 and 2. To measure the equilibrium gquotient Qg, hydrogen was passed at 5 e rate of ca. 30 ml.min"1 through sbout 150 g. of melt (ca. 50 ml) contained in a nickel vessel, previously pickled, dried and fired 417 in hydrogen. The HCl content of the effluent gas from the reaction vessel was measured by passing the stream through a coulometric titration cell furnished with a glass pH electrode. At the same time the volume of hydrogen leaving the coulometric cell during the titration was accurately measured by means of a climbing soap~film ges burette. The partial pressure of HCl in the hydrogen was calculated by correcting to S.T.P., gllowing for the saturated water vapour pressure in the gas burette and assuming ideal behaviour of HC1l at the low partial pressures involved (less than 10" atm.). Even with the dip~line at its maximum depth of 10 cm. the equilibrium pressure of HCl was not attained. A known partial pressure of HCl, measured in the usual way, was therefore introe~ duced into the inlet gas stream by passing the hydrogen through a bed of ferrous chloride held at a predetermined temperature. From a series of four or more measurements with inlet HCl pressures both above and below the equilibrium value the required equilibrium partial pressure was interpolated graphically. The equilibrium pressures were constant over periods of tens of hours and reprow- ducible after temperature cycling. The composition of the salt mixture initially loaded into the reaction vessel was 70 mole % NaCl with 30 mole % of mixed UCl3 and UClh, typically in the ratio 1.5:1.0. After equilibrium measurements had been made over a range of temperatures at a given composition, the UCl, content was reduced by a known amount by introducing into the melt under argon a weighed sample of uranium metal (0.5 to 1.0 g.) suspended on a molybdenum wire. A series of equilibrium measurements were thus obtained over a range of values of the concentration ratio of UCl, :UCl:, from which the equilibrium quotient was obtained from %he siope of the plotted data, Figure 3. Within experimental error the plots at each temperature extrapolate to zero pyp) at the point anticipated from the initial composition of the melt. The derived values of the equilibrium quotient Qg from two duplicate series of experiments are plotted logarithmically against reciprocal temperature in Figure 4 and summarised in Table 2. In the table are also summarised the difference in the partial molar free energies of formation of UCl, and UCl and the ratio of their activity coefficients obtained by com= bining the partial molar free energy difference with the core responding value for the pure liquids. Within experimental error the values of the function RTln.y(UClA)/}(UCIB) are independent of temperature with an average of 4.5 4+ 0.2 kecal. mole™!, Electrochemical Measurement of the UC1, /0C1. Equilibrium. The e.m.f¢ of the cell: ~ (=) Mo(Pb)/UClfiUCls in NaCl//FbCl, in NaCl/Pb(Mo)(+) 418 Ol 008 0 06 + pHCE Y 2 pH> 004 002 1 1 | 1 | { ] | O Ol 02 O3 04 Oos 06 o7 o8 CONCENTRATION RATIO UCP 4 UCE 3 SOLID POINTS FIRST SERIES; OPEN POINTS SECOND SERIES. eo0 720C 44 650C w0 580C 3+ Hydrogen Reduction of UCl; in NaCl:UCl;. PFlot of equilibrium pressure of HC1l versus UCl) :UCls ratio”(Equation 10). 419 Qr atH@ = 1 I Il 0-0010 O- 0011 O-00I12 t 1% 4. Temperature Variation of the Equilibrium Quotient Qg for the Hydrogen Reduction of UClh in NhCl:UClB. 420 has been measured at temperatures of 61k, 681, and 730°C over a range of values of the UCll‘_:UCI3 concentration ratio. Table 2. Egquilibrium Quotients for the Hydrogen Reduction of UCl, in NaCl: UClz {70 mole % NaCl) and Derived Thermodynamic Data ~ Tbmperature °c 580 650 720 10 QR’ atm? ; 2.1 4.8 10.9 RTla Qp kcal.mole™ 5.5 5.6 witoly AFf(HCI)(”) n “23.8 w23.9 «24.0 AF"(U014-UCJ. ) " =30.3 =29.5 m28.4 AFE (uc1#-uc13)( 8) " ~25.8 =249 =24.1 RT1n + (UC1 )/y(UCl ) " o5 wdpo 6 o3 The redox electrode consisted of & closed silica or alumine tube 1.2 cm. in diameter containing a small bead of lead with a molybdenum contact wire. The compartment was filled, through a hole one millimetre in diameter drilled 6 cm. from the closed end, by immersion in a bath of melt contained in a nickel vessel. When filled the electrode compartment was raised sc that the hole was about 5 mm. above the melt surface. This both defined the quantity of melt contained in the compartment and essentially prevented mixing of the melt in the compartment with that in the bulk. Adequate electrical conductivity wes achieved through the film of melt adhering to the wall of the tube. Changes in the concentration of UCl, in the compartment were produced by electrochemical oxidation %r reduction, the total charge passing between the compartment and a nickel auxiliary electrode being measured by an integrating coulometer. Ox%dagion in dilute solution has been shown to proceed in two stages electrochemical oxidation of the electrode f'ollowed by a homo~ geneous reaction of the oxidation product with UCl In order to eliminate excessive loss of molybdenum and contamlgatlon of the melt with a molybdenum dispersion, current was passed into the melt through & small pool of lead during oxidation. Conversely, during reduction the molybdenum wire was removed from the lead to avoid dissolution of the deposited uranium. The reference electrode consisted of liquid lead in contact with a molten mixture of NaCl and PbCly (4L0.0 mole % PbCls,). This was contained in a closed tube of Mullite in which sodium ions are mobile; the impedance of the electrode was less than 2.103 ohms. Electrical contact to the lead electrode was made through an insulated molybdenum wire, as in the redox electrode, so that the thermal e.m.f. was eliminated., The e.m.fs of the reference electrode against the standard chlorine electrode was 421 calculated from the free energy of formation of FoCl,(17), the composition of the melt and the activity coefficient of PbCl, in NaCl estimated by Lumsden from the phase diagramt19§. This gs in good agreement with the average value derived from electrochemical meaaurements(zoxl(O.Blp and 0.73 respectively at 700°C). The whole cell was flushed continuously with pure argon at a pressure of 20 mm. Hg above atmospheric pressure, Suitably placed holes ensured that the hydrostatic pressure was balanced through~ out the electrode compartments. The cell voltage was measured to + 0.1 mV on & Solartron Digital Voltmeter and, for convenience, the output of a Vibron Electrometer was displayed on a strip recorder., The input impedances of both the instruments (> 407 ohm) far exceeded the impedance of the cell (2.10° ohm,) In initial tests the reversibility and stability of the redox electrode were confirmed using a pair of the electrodes in a concentration cell. After an oxidation or reduction, gentle agitation of the wire contact in the compartment produced a stable e.m.fe within a few minutes; this then remained stable to + 2 mV for periods of up to several hours., Measurements were normally made some 30 min. after & concentration change. 4s is to be expected under equilibrium conditions the e.m.f. was the same whether the Mo contact was immersed in or removed from the lead pool. Attempts were made to measure the am.f. of the cell: (=) U(#0)/UC1; in NaCl//PbCl, in NaCl/Pb(Mo) (+) by depositing electrochemically a flew milligrams of uranium on the Mo wire which was either removed from the lead pool or, in separate tests, operated in a compartment free of lead. Here a stable e.n.f, persisted for only a short time, about 10 min., and the reproducibility was poor. Uranium is known to react rapidly with si%ica when both are in contact with a melt conw taining UCL,(21). In Figure 5 the observed cell e.m.f. is plotted against the logarithm of the UCl, :UClz concentration ratio. The slopes of the plots are summarised and compared with the calculated values for a onemelectron transfer in Table 3. Alsco summarised are the standard potentials extrapolated to equal concentrations of UC and UCl3 and the data used in deriving the difference in partia molar free energies of formation and the sotivity coefficient ratio. Within experimental error the difference in excess free energy, RTln y(UCl,)/¥(UClz), is ipdependent of temperature with an average of 2,6 + 0.3 kcal.mole™ . Discussion. Both methpds of measurement show that UClL is less reactive in the NaCl:UCl3 melt than in the pure liquid. The 422 400 |- 300 | b~ | E = N 200 + 100 + 1 lllillll 1 ll.llllll | 10-3 10-2 1o} CONCENTRATION RATIO Uce4luc¢3—-—- A) 6l40C B) 681°C C) 750°C POINTS FOR A AND B DISPLACED BY -50mv AND +50mv RESPECTIVELY 5. TPotential of the cell Mo/UClh‘:UCl} in NaCl//PbCl, in NaCl/Fb. 423 Table 3. Potential of the MOZQCI!'UC 3 Electrode and Derived Thermodynamic Data Temperature °C 614 681 730 observed 174 188 195 Nernst Slope, nv. calculated 176 188 199 Cell Potential at (UClb_) = (U013), Ve ~0.03 ~0,03 =0, Ol Std. Potl. of Pb/Pb012 in NaCl, v. wt.2 w124 ~1.19 Stde Potl. of UCIL'_/UCJB in NaCl, v. -1 .21 -1.18 ~-1.15 a¥F (uca, — v o m— - —_—— — — — A CRUCIBLE—— |Fm—7= |=/—/—/———== 1203 N, =R |_—— SALT PHASE COUPON Vgt ’Ei oo ILIV! ML SALT PHASE S ) | ST | e - — — - —_——— — — TuNGRSETEN == ==l —— METAL-SALT INTERFACE WIRES i - hfif COUPON 11 i1 | | i | METAL PHASE :w I|||| | | | | ! I I |;” | | . MIN i 1'}————METAL PHASE COUPCN |L'H‘r"!' ° TTV'!'TJ‘#TIT K 'l 'I!Ilh‘l SO L Figure 2, Corrosion Agitator <———— STIRRING MOTOR SAMPLING PORT ———e ARGON-VACUUM LINE DR THERMOWELL GRAPHITE SECONDARY CRUCIBLE HEATED ZONE GRAPHITE POURING MOLD CRUCIBLE AGITATOR Figure 3. Tilt-Pour Furnace 44] Solution Stability Tests Materials which demonstrate acceptably low corrosion rates in the capsule, agitator and crucible tests are further tested to determine 1f the containment material 1is inert to the dissolved uranium and plutonium that would be contained under process con-~ ditioms. The material is normally fabricated into crucibles similar to those in the tests just described. The crucible is charged with an appropriate metal-salt mixture and heated to a constant temper- ature, The crucible contents are stirred for about 1 hr and, at that time, uranium or plutonium metal i1s added. The metal-salt mixture is then equilibrated, with stirring, for times of 50 to 200 hr, during which time filtered samples of both the metal and salt phases are taken for analysis of uranium or plutonium content. If the solutions are stable these analyses remain constant but, if a solution instability is present, the uranium or plutonium analyses decrease during the time of the run. If an instability exists, samples of precipitates and of the metal and salt phases at probable precipitate locations are taken at the end of the run to assist in identifying the probable cause. Additional details of the sampling procedures and precautions are contained in an article by Winsch (4. Thermal Convection Loop Test Mass transfer effects under a thermal or concentration gradient are not adequately tested in isothermal capsule, corrosion agitater, or agitated crucible corrosion tests. In an effort to develop a test to provide this information a two-phase thermal convection loop was operated. This loop was constructed of type 304 stalnless steel and was successfully operated to demonstrate the design, thermal control, and operation of a system which would provide mass transport information. A gchematic diagram of the loop 1s shown in Fig. 4. The general loop configurationis a "figure eight''. The lower loop contains molten metal and the upper loop contains molten salt. The hori- zontal leg common to both the upper and lower loops contains both metal and salt which flow countercurrently during operation. The loop is charged with molten metal and salt from an overhead fill- tank. The location of the metal/salt interface in the horizontal leg was determined by X-ray pictures of the operating loop. This technique was not very satisfactory for determining the salt/ atmosphere interface; therefore, this interface was determined by a "dip-stick' technique. 442 i e— ) 30 By AN 77277777777, 7)) % oy B NN NN NNNN 2 77777 7 MO AN é////////// ¥, Temperature control was maintained by the use of four independent heater circuits. One circuit heated the metal loop and a second heated the salt loop to maintain the charge materials at a uniform temperature in the molten state. The other two circuits heated the metal and salt "hot' legs to elevated temper- atures thereby providing the thermal gradients necessary for thermal convection flow. When the loop was in operation about 80% of the power was introduced through the hot leg heaters. At the conclusion of the run the metal and salt were frozen in place and the loop was sectioned for metallographic and chemical exami- nation. Prototype Vessel Testing Materials which successfully complete all previous phases of testing are then fabricated into vessels which are used in laboratory and engineering investigations of the process.(s) At the completion of an experimental series these vessels are examined either destructively or non-destructively to determine their behavior under actual processing conditions. Corrosion Mechanisms Most of the corrosion that occurs in metallic specimens exposed to these metal/salt systems is the result of dissolution of the material in the liquid metal. 1In most cases the material has some solubility in the liquid metal and slowly dissolves until an equilibrium condition is reached. Transgranular attack is relatively uncommon. Intergranular attack is not uncommon, however, and is usually the result of preferential solubility of elements concentrated at the grain boundaries or of metal phase reaction with grain boundary precipitates. Uccasionally a metal phase inclusion, such as carbon, presents a desired locatior for attack. The salt produces several different correosion effects on metallic specimens. If the salt is not dry, materials such as oxychlorides, hydrogen and chlorine may be present and, at the high process temperatures, will cause serious corrosion. For this reason the salt used in these runs is purified by contact with a Mg-33.5 at.% Cd alloy at 600°C for 24 hr before use to remove any water present, This purification process is outlined in Ref. 6. The salt phase can also cause dissolution of socluble metallic specimens and, if the distribution coefficient favors transport of the material to the metal phase, this dissolution can continue beyond the apparent solubility limit. The salt phase also causes corrosion by embrittlement of some 444 materials and alloys. The nature and cause of this embrittlement is largely unknown but has the effect of seriously reducing the ductility of some materials, especially the refractory metals. This same effect may be responsible for the few cases of stress corrosion that have been observed. In addition to these effects, the presence of magneésium, a good reducing agent, in the metal phase is the cause of some corrosion, especially of ceramic materials, such as alumina, which can be reduced by magnesium. The presence of magnesium also accentuates some metallic corrosion especially in cases where corrosion resistance is normally obtained through the presence of a pro- tective oxide film which is subsequently reduced in the presence of magnesium. Results of Tests Performed The results obtained in this program are discussed in terms of general experience with different classes of materials rather than measured corrosion rates in specific systems, since these results are avallable elsewhere(7’8!9§ and are not of great general Interest. Typical results are presented to exemplify trends and indicate types of problems encountered. The reported corrosion rates are conservative and are probably higher than those that would be experienced in actual service. These rates were obtained from data taken in short times (usually <200 hr) and converted to an annual rate for presentation on a uniform basis., This resulted, therefore, in multiplying the observed corrosion by factors of 40 to 80. Small errors in experimental measurements would then result in large errors of annual corrosion rates. In taking the experimental data a conservative approach is used to prevent underestimating the annual corrosion rate., In samples of varying degrees of corrosion the '"worst case'" 1s usually taken, except for those cases in which this condition is obviously unrealistic. Other factors such as protective film formation or saturation of the metal/salt system with a soluble constituent also decrease the annual corrosion rate that would be obtained in actual service, Tests are also usually made at temperatures slightly higher (,,50°C) than those expected in process use in order to accentuate the corrosive effects of the solvent. Several techniques are used to determine the extent of corrosioen. These include: (1) metallographic examination of the specimens after exposure to the liquid metal/molten salt systems, (2) size changes of the metal coupons, (3) physical property measurements before and after exposure, (4) chemical analysis of the metal and 445 salt phases after sample exposure, and (5) electron-probe microscopy and X-ray diffraction analysis of reaction layers and precipitated solids. Ferrous Alloys Austenitic Stainless Steels Austenitic stainless steels containing appreciable concentrations of nickel or manganese are not good containment materials for the systems studied. Nickel is soluble in copper, zinc, cadmium, and magnesium, and deterioration of the steel through nickel leaching occurs. These steels, however, are normally acceptable containment materials for a purified salt in the absence of a metal phase. Ferritic Stainless Steels Those steels with a low nickel and high chromium content (such as type 405) are much more corrosion resistant to the proposed metal/salt systems than austenitic stainless steels. Neither chromium nor iron is highly soluble in the metal phases. In general the corrosion resistance increases with increased chromium content of the steel. Increased magnesium concentration in the molten metal usually makes the system less corrosive. Ferritic steels are acceptable containment materials for high magnesium content alloys and for the molten salts. Solution stability must be considered for process application utilizing these steels, however, since both UgFe and UFe; have been identified as forming in a Cu-20.8 at. % Mg-0.5 at. % U/MgCl, system exposed at 850°C to a CB-30 stainless steel crucible. The thermal convection loop studies also indicated a tendency for both chromium and iron to mass transport under a thermal gradient, and this tendency should be taken into consideration in proposed applications. The ferritic stainless steels are more difficult to fabricate than the austenitic stainless steels primarily because of poorer welding properties. Consequently, fabrication methods not requiring welding (such as casting) are recommended. Croloy 16-1 is a ferritic steel available as pipe which has been used successfully as a transfer line in copper-magnesium systems. Cast Iron, Carbon Steel, and Low Alloy Steel These materials are intermediate in corrosion resistance between the austenitic and ferritic stainless steels but, in many cases, do not possess sufficient strength for use at the elevated temper- atures of process interest. They are also susceptible to * Babcock & Wilcox Tradename. 446 uranium-iron intermetallic compound formation, and a uranium phosphide (UP or U4P,) was found when cast iron was used to con- tain a Cu-20.8 at. % Mg-0.5 at. % U/MgCl, system at 850°C. Table I presents typical corrosion rates obtained in capsule tests with various ferrous alloys. Table I Corrosion Rate of Ferrous Alloys in Pyrochemical Process Systems Conditions: Time: 200 hr Salt Phase: MgCl,-30 mol %Z NaCl-20 mol % KC1 Rotated Capsule Tésts Observed Corrosion Rate, mpy Metal Phase, at., % Temp.,°C Carbon Steel 405 S.S. 304 S.8S. 27,7 Cu 72.3 Mg 600 <5 <5 <5 77.5 Cu 22,5 Mg 600 5 <5 <5 27.7 Cu 72,3 Mg 800 39 <5 <5 77.5 Cu 22,5 Mg 800 35 65 57 67.0 Cd 3.9 Mg 29.1 Zn* 750 105 70 300 92.3 Cd 4.4 Mg 3.3 Zn® 750 <17 <8 _— * No salt present These results show that at temperatures as high as 600°C steels are only slowly corroded by Cu-Mg alloys but an increase in temperature to 800°C significantly increases the corrosion rate. Neither copper nor cadmium base systems are highly corrosive but the addition of zinc in appreciable concentrations precludes containment of these materials in ferrous alloys. In general the ferritic type 405 steel is the most corrosion resistant. Refractory Metals The refractory metals tungsten, tantalum, niobium, molybdenum, rhenium and various alloy combinations, in general, possess good corrosion resistance to the proposed process systems. This corrosion resistance is a result of their low solubility in the process systems and is not directly related to their common 447 application of high temperature service. The problems most often encountered in their application concern (1) the fabrication of refractory metal parts and (2) embrittlement of the more ductile materials, tantalum and niobium, by the molten salts, hydrogen and oxygen. The discussions in this section are related to recent data obtained in the Cu-Mg system since this and the Cd-Mg system are very similar. The Zn-Mg system is substantially more corrosive to most of these materials and their utility in Zn-Mg systems was the subject of a previous paper. 7) Table 2 presents corrosion rates of refractory metals exposed to Cu-Mg/halide salt systems. These results were obtained from weight loss and metallographic examination studies. Although they indicate little corrosion, they do not present the entire corrosion picture. Both tantalum and niobium were embrittled by this exposure but this effect is not apparent from either of the forementioned tests. This embrittlement would be of significance in service application because process vessels would lose ductility, and the probabllity of failure through mechanical or thermal shock would increase. Table 3 presents the results of tests performed to obtain quantitative data on this embrittlement. The materials were exposed as corrosion agitators, at 650°C to a MgClz—BO mol 7 NaC1l-20 mol % KC1l salt for 196-197 hr. Tantalum appears to be more corrosion resistant to the molten salt than niobium but also appears more susceptible to embrittlement. The cause and quantitative effect of the embrittlement on the two metals is currently under investigation. Table II Corrosion Resistance of Refractory Metals to Cu-Mg/Halide Salt Systems Conditions: Metal Phase: Mg-43.3 at. % Cu~0.2 at. 7Z U Salt Phase: MgCly-30 mol % NaCl-20 mol % KCl Length of Run: 150-200 hr Temperature: 750°C Corrosion Agitator Tests Sample Location Observed Corrosion Rate4 mpy Nb Ta Mo W Metal Phase <22 <1 <17 <] Metal/Salt Interface <22 <1 17 <1 Salt Phase <22 <1 <17 <1 Vapor Phase <22 <1 17 <1 * Estimated, not a direct measurement. 448 Table III Corrosion of Niobium and Tantalum by MgCl,-NaCl-KCl Systems Temperature: 650°C Salt: MgCl,-30 mol % NaCl-20 mol 7% KC1 2 Test Material Nb Ta Length of test, hr 197 196 Initial Hardness, 15T 73 60 Final Hardness, 15T Sample 1 87.5 93.5 Sample 2 87.5 94 Sample 3 88 94 Sample 4 88 93.5 Corrosion Rate, mpy* Sample 1 +2.1 +0.5 Sample 2 +1.3 +0.6 Sample 3 +1.9 +0.5 Sample 4 +2.9 +0.5 * The plus (+) sign indicates film formation. Tungsten appears virtually corrosion resistant to this system and is routinely used in laboratory and engineering process investigations; no deleterious effects have been observed after 3 or 4 years of service of some vessels., Tungsten is difficult to fabricate into many desired shapes but crucibles as large as 10 in. ID x 20 in. high have been used. Various crucible fabri- cation techniques have been tried and these results are discussed elsewhere, (10) A fabrication technique applicable to a wide variety of shapes is now under investigation. This technique involves the vapor deposition of a tungsten coating on a graphite substrate and, to date, the results have been encouraging. Molybdenum is also difficult to fabricate into intricate shapes and, since our results indicate that tungsten is more corrosion resistant, our effort has been directed primarily toward tungsten shapes. A notable exception is the Mo-30 wt Z W alloy. This alloy i1s machinable and is routinely used for agitators, exposed 449 transfer line parts, and other applications requiring machined parts. The material is only slowly attacked in process service (perhaps about 20 mpy) and has given excellent service. Graphite and Vitreous Carbon Two major problems are encountered when graphite 1s used as a containment material for pyrochemical processes: (1) coanventional grades of graphite are too porous to contain the molten salt and (2) solute uranium and plutonium react with graphite to form insoluble carbides, The porosity problem is easily solved by the uge of premium grades of graphite with a higher demsity. It has been our experience that a graphite density of about 1.72 g/cc (theoretical density, 2.25 g/cc) or higher is usually sufficient to contain the molten salt. Graphites of this density are commercially available at a modest increase in cost over conven- tional graphite, Several attempts have been made to protect the graphite surface from reaction with the solute uranium and plutonium. The major problem encountered i1s matching thermal expansion of the coating with that of the graphite. The most successful attempts to date have been with tungsten and with silicon carbide, both of whose thermal expansion coefficients (2.2 x 106 in./in./°F) are very close to graphite (vl to 4 x 10=6 in,/in./°F depending upon grain orientation of graphite). Care must be taken to obtain a graphite with an isotropic thermal expansion coefficient. A graphite crucible coated with a 0.030 in. coating of tungsten deposited by vapor deposition is now under test and results are encouraging. The coating applied by this technique is very nearly 100% of theoretical density. A small crucible (vl in. ID x 2 in. high) coated with flame-sprayed silicon carbide was also used very successfully; but, when attempts were made to produce a larger crucible (6 in. ID x 10 in, high), the coating was not dense enough to protect the graphite. Vitreous carbon was used very successfully in small scale tests and was not generally wet by the metal/salt systems in times up to 200 hr, Attempts to scale these crucibles up to larger sizes were unsuccessful because of the fragile nature of the material and because the maximum available wall thickness was 0.1 in, Crucibles (about 4 in. ID x 8 in., high) were fabricated but were found to be easily broken in service. Silicon Carbide and Alumina Silicon carbide is inert to the process solutions but is not usable because the density of fabricated parts is too low to 450 adequately contain the molten salts., A special high-density grade of silicon carbide was obtained and tested and, although its density was sufficient to contain the molten salt, three of three crucibles failed in services from what appeared to be thermal shock. Alumina has been used in many experimental investigations but is not completely immune to attack. Alumina is reduced by molten magnesium but at our magnesium concentrations of interest, the kinetics of the reduction are slow and several hundred hours exposure are readily obtainable. The reduction of alumina can also cause solution stability problems., A metallic film adhering to the wall of an alumina crucible exposed to a Cu-20.8 at. % Mg-0.5 at. % U/MgCl, melt after 144 hr at 850°C was found to be nearly pure UAl,. %ther ceramics such as boron nitride, whose B O3 content is reduced by magnesium,and zirconia and magnesia, wfiich are porous to the salt, have alsoc proved unsuitable, Conclusions Pyrochemical processes for the purification of discharged reactor fuels require suitable containment materials for a variety of molten metal and salt systems. Although the problems of meeting these requirements are difficult, materials are available for the systems of interest. Ferrous alloys are useful in process equipment in those steps where solute uranium and plutonium are not present and for applications requiring short-term exposure. Refractory metals, especially tungsten, tantalum, and niobium, are useful throughout the process. Tungsten presents fabrication problems but if the embrittlement problem with tantalum and niobium can be solved, they should be usable for nearly all process equipment. Ceramics have process utility, but their usefulness is limited to the higher density products that are resistant to molten magnesium, Acknowledgments The authors wish to express thelr appreciation to Mr. L. F. Dorsey who ably performed much of the experimental work reported in this article. Dr., R. J. Meyer, Mr. R. V. Schablaske, and their co- workers in the Analytical Laboratory are responsible for the chemical analyses reported herein and were helpful in suggesting experiments to obtain needed chemical data. 45] 10. References Johnson, I., "The Solubility of Uranium and Plutonium in Liquid Alloys', This Symposium. Knighton, J. B., I, Johnson and R, K. Steunenberg, "Uranium and Plutonium Purification by the Salt Transport Method", This Symposium. Petkus, E. J., T. R. Johnson and R. K. Steunenberg, "Uranium Monocarbide Preparation in a Liquid Metal', Nucl. Appl 4, p. 388-393, June 1968. Winsch, I. 0., K. R. Tobias, R. D. Pierce and L. Burris, Jr., "Sampling of Liquid Metals', Report ANL-7088, Argonne National Laboratory, Sept. 1965, Pierce, R. D., W. E. Miller, J. B. Knighton and G. J. Bernstein, "Multistage Contactors for Liquid Metal-Salt Extractiomn", This Symposium. Vogel, R. C., M, Levenson, J. H., Schraidt and J. Royal, "Chemical Engineering Division Semiannual Report, July- Dec. 1965", Report ANL-7125, Argonne National Laboratory, p. 35, May 1966. Nelson, P. A., M. L. Kyle, G. A. Bennett and L. Burris, Jr., "Corrosion of Refractory Metals by Zinc-Magnesium-Uranium and Halide Salt Systems", Electrochem, Tech., 3, p. 263-269, Sept.-0Oct. 1965, Kyle, M. L., P. A, Nelson and L. Burris, Jr., "Corrosion of Steels and Tantalum by Molten Cadmium-Magnesium-Zinc Systems', Electrochem. Tech., 3, p. 258-262, Sept.-Oct. 1965. Kyle, M. L., R. D, Pierce, V. M, Kolba and L. F. Dorsey, "Containment of Copper-Magnesium-Uranium/Halide Salt Systems', Report ANL-7566, Argonne National Laboratory, 1969, Winsch, I, 0., M, L, Kyle, R. D. Pierce and L. Burris, Jr., "Tungsten Crucibles in Pyrochemical Processing of Nuclear Fuels", Nucl. Appl., 3, p. 245-251, April 1967. 452 CORROSION OF A 1LOW ALLOY STEEL IN A SIMULATED LIQUID-METAL-FUEL REACTOR MIXTURE* Roger L. Suchanek and George Burnet Institute for Atomic Research and Department of Chemical Engineering, Iowa State University, Ames, Iowa 50010 U. S. A, Abstract An isothermal dynamic testing technique was used to investigate the corrosion of 2%7 chromium - 1% molybdenum alloy steel first by pure liquid lead-bismuth eutectic (55.5%Bi) and then by zirconium- inhibited eutectic to which elements had been added to simulate those fission products which would be found in a liquid-metal fueled reactor. When exposed to the eutectic alone the steel was found to be highly resistant to attack up to temperatures of about 700°C. Above 700°C the degree of attack increased with temperature until at 900°C only long fingers of ferrite remained at the metal surface in place of the original material. In tests with the simulated reactor mixture inhibited with zirconium, the temperature was raised to about 800°C before corrosion was detected. The mixture consisted of the eutectic plus Zr, U, Mg, Mo, Ce, Nd, and La as additives. Above 800°C the same increase in attack with temperature as occurred with the pure eutectic was observed. *Work performed in the Ames Laboratory of the United States Atomic Energy Commission. Contribution No. 2538. 453 Introduction Low-melting metals and alloys show promise as nuclear reactor coolants or fuel carriers because of their excellent heat transfer characteristics, low vapor pressure, frequent low neutron capture cross section and resistance to irradiation. Applications have been limited, however, because they corrode most container materials at elevated temperatures. To overcome this limitation, investigators have turned to certain additives, termed iInhibitors, which are effective in reducing corrosion through formation of impervious surface films on the con- tainer materials at the liquid metal interface. The use of such inhibitors to reduce corrosion has been investigated for many low- melting heavy metals such as mercury, lead, bismuth and the lead- bismuth eutectic. A pioneering effort was that of Nerad (1) and his assoclates at the General Electric Company when they developed the mercury boiler. It was found that as little as 1 ppm of titanium or zirconium added to the magnesium-deoxidized mercury reduced the attack of ferrous alloys to a negligible level. More recently (1954) Gurinsky, et al. (2) reported that additioms of Mg and Zr to uranium-bismuth solutions substantially reduced the corrosion of 2%% Cr - 1% Mo alloy steel. Later, in 1956, Miller and Weeks (3) discovered that Zr and Ti additions to liquid Bi inhibited the corrosion of steels by the formation of protective deposits of ZrN or TiN. Inhibition of corrosion of low alloy steels by uranium-bismuth solutions through Zr addition and formation of a surface film of ZrN and/or ZrC was reported in 1960 by Weeks and Klamut (4). Whether a nitride or carbide film formed was determined by the relative activities of the carbon and nitrogen in the steel. Romano et al. (5) summarized in 1963 the data obtained from more than 100 low alloy steel thermal convection loops which contained zirconium-inhibited bismuth and uranium~bismuth solutions. Their conclusions confirmed the corrosion resistance of the low alloy steels to zirconium~inhibited uranium~bismuth solutions. Working in this Laboratory in 1958 Clifford (6) completed a series of experiments with the lead-bismuth eutectic which led to selection of 2%7% Cr - 1% Mo alloy steel and type 430 stainless steel as container materials most suitable up to 700°C. Above this temperature none of the ferritic steels examined were satisfactory. In 1960 Stachura (7) discovered that Zr concentrations as low as 50 ppm effectively inhibited corrosion of 2%% Cr - 1% Mo steel by the eutectic. X-ray diffraction studies suggested the formation of a protective film of ZrN. 454 The purpose of this study was to corroborate the work of previous investigators who worked with the zirconium-inhibited eutectic and to determine how corrosion inhibition might be affected by those elements which would be present in the liquid metal if it were being used in a liquid metal fueled reactor. Testing Procedure The corrosion tests were conducted in two spinner units such as that shown in Figure 1. The portion of each unit which contained the liquid metal bath was placed in the open core of a cylindrical resistance furnace. The alloy steel specimens used were 4-inch lengths of schedule 40 steel pipe. A specimen to be tested was mounted at the end of a shaft which was rotated at 200 rpm while the specimen was immersed about 2 inches in the liquid metal. Purified argon at 16 psia pressure was used as an inert cover gas. Each unit was provided with an air lock through which a graphite sample crucible was lowered to withdraw samples of the liquid metal at 12-hour intervals throughout the 96-hour test period used in all runs. This period was found to be about the minimum required for buildup of a detectable level of corrosion products in the liquid metal In those cases where significant corrosion did occur. Figure 2 is a photograph of the two units as they were installed in the laboratory. In all tests the metal bath samples were analyzed for Fe and Cr content. In addition each pipe specimen was cut perpendicular to the axis and examined metallographically. In the inhibited runs powder x-ray diffraction patterns were taken of the surface scrap- ings of the steel specimens in order to determine the presence of any surface films in which Zr or the other additives might be found. The overall investigation consisted of two series of corrosion tests in which all pipe specimens were 2%% Cr - 1% Mo alloy steel. In the first series pure lead-bismuth eutectic (SS.SiWBi) was used as the corrodent; in the second the corrodent was a simulated reactor mixture inhibited with Zr. The mixture consisted of the eutectic plus 100 ppm Zr, 1000 ppm U, 300 ppm Mg, 1 ppm Mo, Nd and La, and 2 ppm Ce. The Mg served as a deoxidant to prevent Zr and the other additives from being oxidized from solution. The tempera- ture (600°C - 900°C) was held constant throughout each run. Materials Pipe Specimens The chemical composition of the 2%% Cr - 1% Mo steel used is 455 2 COMPRESSION L~ SEAL 0 | = I* VACUUM RESEARCH CO GATE VALVE STIRRER GEARS AND MOUNT ~— COOLING coILS BEARING AND COMPRESSION e g SEAL y STAINLESS STEEL ROO FILLING PORT TANTALUM THERMOCOUPLE -~ WELL ! \ N el TANTALUM ) p\usogx LINER M GRAPHITE LEVEL sAMPLING TEST SPECIMEN CRUCIBLE 1. Spinner Corrosion Test Unit. 456 Arrangement of Corrosion Test Apparatus. given in Table I. All test specimens were &4-inch lengths of schedule 40 seamless round pipe, cold drawn and annealed. Liquid Metals The Pb and Bi used to make up the eutectic for both the inhibited and uninhibited tests were of 99.99% purity. The chemical analysis of both metals is given in Table II. Additives The exact chemical analysis of the additives U, Zr, Mg, Mo, Ce, Nd and La is not known but the highest purity materials available were used. All were of at least 99% purity. . Table T. Chemical Composition of 2%% Cr - 1% Mo Test Specimens Element Weight Per Cent C D.120 Mn 0.450 S 0.010 P 0.013 Si 0.310 Cr 2.220 Mo 0.960 Table II. Analysis of Eutectic Materials . Impurity Bi (ppm) Pb_(ppm) Ag 1 ). Of these, the Handlos=-Baron eddy dif- fusion model has been used most extensively. The orig- inal theoretical treatment of this model yielded: 471 Re , Sc i Sh, = 0.00375 ——= (6) i 1 +ui/ue Sh, = k,;d/D,; (7) Sci = \)i/Di (8) Re, = du/v, (9) The subscript 1 denotes properties of the drop and U is the absoclute viscosity. Several modifications of Eq(6), required to rectify some mathematical deficiencies of the model, have been develcoped by 0lander(11) and Patel and Wwellek (120, The stagnant drop medel of internmal mass transfer ut- ilizes the solution of the molecular diffusion equation in a sphere. Transient solutions of this type do not yield an internal mass transfer coefficient directly. In- stead, the fraction of the solute initially present in the drop which has been removed is specified as a functicen of contact time, The fraction extracted is related to the internal mass transfer coefficient by: _ d k; = - g7 4n(1-£) (19) where f is the fraction extracted and t is the contact time, For the solution of the stagnant diffusion model by Newman , the fraction extracted is: (o] f =1 - é? % 15 exp [-nzfizT] {11) i n x| where T 1s a dimensionless contact time, defined by: 4Dit T = —5— (12) d Eq(ll) is inconvenient to use when the contact time is short because many terms of the sum are needed. The alter- nate expression: £ = & T o3 (13) Ll is satisfactory for nearly all applications of this model to extraction data. Because of the explicit dependence of the mass transfer coefficient on contact time in the 472 stagnant drop model, the concept of a mass transfer coef- ficient loses some of its convenlence. For extraction systems in which external and internal resistances are important, the transfer rate is controlled by the overall mass transfer coefficient: (14) where m is the equilibrium ratio of the solute concentra- tion on the external phase side of the interface to the concentration on the drop side of the interface. This relation is useful only if the quantity m is independent of concentration. Interfacial Equilibrium If there are no extraneous resistances at the phase boundary, the concentrations of the transferring solute on the two sides are fixed by thermodynamic consideratiomns. Three types of interfacial equilibria have been investi- gated in high temperature extraction systems. 1) Physical Equilibrium. When the two phases are immis=- cible liquid metals, the ccefficient m in Eq(l4) is the equilibrium distribution ccefficient of the transferring solute between the two solvent metals. If the diffusing species 1s solvent 1 transferring to an immiscible liquid 2 which is unsaturated with respect to liquid 1, the coef- ficient m is the solubility of 1 in 2. 2) Chemical Equilibrium. In fused salt-liquid metal transfer, the metal phase generally contains a solute metal which can be oxidized by an extracting agent B+n in the salt phase. Equilibrium at the interface is gov- erned by the law of mass action applied to the exchange reaction: +n +n for which BiSIn Req = T 40 (16) ivi For simplicity, the valences of the two cations S and B have been taken to be equal. The concentrations in Eq(16) have been written in terms of molar concentrations, since these are the appropriate units for describing the diffu- 473 sional transfer step. The subscript i denotes a value at the phase boundary. The activity coefficient ratios which normally appear on the right hand side of Eq(l6) have been assumed constant and incorporated into the equi- librium constant. The simple overall mass transfer coefficient concept of Eq(l4) is inconvenient to use in the present case, since Eq(16) renders the interfacial equilibrium condition non-linear. Instead, it is easier to relate the parameters of the problem (initial concentration, equilibrium con- stant and mass transfer coefficients) directly to the frac- tion of the solute extracted from the drop phase. Gen- erally, the metal phase is dispersed in the continuous salt phase. A metal drop falling through the salt will at all times be exposed to the same concentration of ex- tractant in the salt, or Bt = constant. The concentra- tion of the solute in the drop decreases from an initial value of S, to some value S while the product metal in the bulk of the drop increases from zero to B. Since this particular exchange reaction is equimolar, S, = S+B. Typ~- ical concentration profiles near the interface are depicted schematically in Fig. 1. The individual film mass transfer coefficients are de- fined in terms of the molar fluxes per unit interfacial area and the mclar concentration driving forces: _ B, +n _+n, _ . B _ Ny = ke(B By ) = k;(B;-B) (17) Z 1 Ogtn _ S . N, = kesi = k; (s si) (18) For the reaction stoichiometry considered in this ex- ample, Ny = Ng = N. The mass transfer coefficients of the two species in a given phase differ only if the diffusion coefficients differ, Since the mass transfer coefficients vary approximately as the square root or the 2/3 power of the diffusion coefficient, and since the diffusivities of similar species in the same solvent are nearly equal, it is appropriate to make the approximations: B S B S ke = ke = ke and ki = ki = ki . Eqs(16), (17), and (18) must be solved simultaneously to eliminate all interfacial concentrations and to obtain an expression for the flux N in terms of the bulk concentra- tions 8, 8,, and B N the mass transfer coefficients ke and ki, and the equilibrium constant Keq' In general, 474 Interface Metal phase Salt phase +Nn Concentration profiles near interface in fused salt- liquid metal extraction. 475 the flux depends upon the concentration 8 in a non-linear manner, A significant simplification in the analysis with ex- change occurs if the equilibrium constant of Eq(16) is very large (this situation is usually called an "irre- versible" reaction in the literature on mass transfer with chemical reaction), If Kegq * ©, either S; or BI® must be zero. If S; » 0, Eq(l8) indicates that the flux is given by N = k;8, which is the condition of transfer com- pletely controlled by resistance within the drop. On the other hand, if Bin + 0, transfer is controlled completely by diffusion in the salt phase, and by Eq(l7), N = k B*n, The actual controlling mechanism is the one which yields the smaller rate. In a drop extraction experiment, where the concentration in the metal drop S decreases with con- tact time, the rate may be controlled by external trans-— fer in the top of the column and by internmal transfer at the end of fall., The two mechanisms participate in the determination of the amount extracted in a consecutive fashion instead of by the series manner implied in the overall coefficient of Eq(l4). If Ko, is finite, Eqs(16) - (18) can be solved for S;n which is "determined by solution of: 2 +n +n to (Keq—l)y(si )< - [Keq(yB +S5) + (5 _-85)]1(5,7) + K SB = 0 (19) where vy is the ratio ke/ki. The decrease in the concentration of the solute in the metal phase as the drop falls through the salt is given by the material balance: 6 = - 'a'N (20) With N = k ($¥tn) from Eq(l8), the concentration § after a contact imé t is determined from: 6k ds _ e (S+n) = Tt (21) i 3o where SI“ is the solution of Eq{(l9)., 1In general, the in- tegral on the left of Eq(2l) cannot be performed analyt- 476 ically. The quantity usually measured in the experiment is the fraction of the solute extracted from the drop aft- er falling through a length L of salt: f =1 - S/So (22) where the contact time is t = L/u. If the internal resistance to mass transfer is neg~- ligible (y = 0), Eq(l9) can be solved. The fraction ex~- tracted 1s given by: 6t dKeqke( S - &n(1-£f) + (Keq—l)f = ) (23) 3) Isotopic Equilibrium. If the species 8 and B in the salt-metal exchange reaction are isctopes of the same met- al species, the equilibrium constant of Eq(l6) is identi- cally unity. Eq(1l9) can be solved and the flux written as: +n k (B /S ) N = = — o S (24) 1+(B /SO)Y The bracketed term in this expression is an overall mass transfer coefficient of the type given by Eq(l4). The "distribution coefficient" m is identified with the concentration ratio (B+n/SO). Isotope exchange has the capability of covering the entire range from complete in- ternal control to complete external control simply by ad- justing the ratio (BYR/s ). Slow Interfacial Chemical Reaction If the chemical exchange reaction, Eq(l5), proceeds at a rate comparable to the diffusion rates, equilibrium be~ tween the two phases at the interface cannot be attained. If the chemical reaction mechanism is the same as the overall reaction, the interface condition of Eq(l6) is re- placed by: +n N = kr(BiSi - ESiBi ) (25) Eq(l6) 1is seen to be a special case of Eq(25) which is approached as the chemical rate constant kr becomes very large. 477 Drop Fall Velocity In order to determine the contact time in a drop ex- traction process, the fall velocity cof the drop must be known. These velocities have rarely been measured in high temperature systems, and the fall veloecity is usuallg com- puted by assuming that the Hu-Kintner correlation (14 , which was developed for aqueous-organic systems, applies, Comparison With Experiment Drop extraction experiments provide information on the variation of the fraction extracted f for variations of the controllable parameters d (drop diameter}, t (contact time, varied by adjusting the height of the salt column) and in chemical or isotopic exchange experiments, by ail- tering the concentration ratio (B+n/SO). Generally the temperature is maintained constant at some convenient value. The experimental results are compared toc a theo- retical model such as one of those discussed above. Lack of agreement between theory and experiment can be attri- buted to one of the fellowing reasons: 1) The theoretical or empirical expressions for ki or ko (usually taken from investigations of aqueous-organic systems) do not apply to the high temperature system. 2) A slow chemical step occurs at the interface (i.e., k., of Eq(25) is not very large. 3) There is an additional resistance at the interface, due perhaps to impurities (lack of intimate phase contact due te non-wetting is very unusual in liquid-liquid sys- tems). 4) Even if equilibrium prevails at the interface, the dynamic data (Keq or m) upon which the interface condi-~ tion is based are in error. 5) The transport properties may not have been meas- ured or are poorly estimated from correlations. A review of several experiments on high temperature liquid-liquid extraction will indicate the degree with which such experiments agree with drop extraction theory. Drop Extraction Experiments The first experiments specifically designed to inves- tigate the kinetics of drop extraction with high tempera- ture reprocessing in view were those reported by Bonilla at the First Geneva Conference(1l53), To minimize the problems associated with containing reactive metals at elevated temperatures, the lead-zinc 478 system was studied at 450°C. The two partially miscible solvents were initially free of the other component. The sealed pyrex tube containing the two sclvents was brought to temperature in the position shown by the solid diagram in Fig. 2. The tube was then tilted by 105° so that the lead fell through the zinc and the zinc rose to the top. After rapid solidification, the concentrations of zinc in the lead and lead in the zinc were measured. It was hoped that the zinc would rise through the lead as a single large drop, thus permitting comparison of the extermal mass transfer coefficient in the lead phase and the intermnal coefficient in the zinc drop with drop extraction theory. However, the measured external coefficient was 30 times greater than the value computed from Eq(5), probably be- cause the zinc phase had broken up intc many small drops instead of rising as a single sphere,. Internal coefficients in a fused salt-liquid metal extraction were measured by Bonilla in the drop rise ex- periment shown in Fig. 3. The LiGCl-KC1l eutectic contained in compartment A of the pyrex apparatus was fed by inert gas pressure through tube E into the cadmium resevoir B. By forcing the salt through nozzle C, drops of 3 mm diam- eter were produced. Since cadmium is slightly soluble in the salt, but the salt is insoluble in cadmiuwm, only cadmium transfer into the salt drops occurred. The mass transfer coefficient inside the salt drops was obtained from the terminal cadmium concentrations in the collected salt. However, the data were compared to the predictions of the Higbie model (which applies to external mass trans- fer) rather than to one of the internal mass transfer models. The first salt-metal kinetic experiment which consid- ered an exchange reaction at the interface was reported by Katz, Hill, and Speirs in 1960(16) They studied the transfer of Sm from a bismuth phase to a NaCl-KC1l-MgCl, salt phase by the Sm-MgCl, exchange reaction. Metal drops of 2.2 mm diameter were introduced into the salt phase in the Vycor device shown in Fig. 4, Smooth drop entry was accomplished by moving the notch in the central rod up to the bismuth reservoir in the upper compartment, then low- ering the rod into the salt phase so that the small amount cf metal contained in the notch fell through the salt. In order to minimize the continued extraction from the puddle of metal at the bottom of the coclumn, the constriction shown in the bottom of the apparatus was provided to re- duce diffusion to and from the collected metal. The frac- tion extracted was determined by analysis of the salt aft- er 15 drops had been released in this fashion. The con- 479 \ / - A W W A Ny -— o o e e e 2. Apparatus used by Bonilla for extraction measurements in a two-immiscible liquid metal system (15). A W) * . Y - D | \rx J/ - C= O o 3. Apparatus used by Bonilla for investigation of fused salt-liquid metal extraction (15). 480 TO VACUUM OR He SUPPLY 4 /—MAGNETS Bfl» PYREX T /_:—BALL BEARING EHVMAGNET THERMOCOUPLE WELL - | “—FLANGE I M L SIDEARM Bi ALLOY—)< 5? 7 SAMPLE CUP—~:X£J51 — vvcoa/ ~SALT 4. Extraction apparatus of Katz et al (16). 481 tact time was varied by using salt columns varying from 7 to 25 cm in height. Because the equilibrium constant strongly favored accumulation of samarium in the salt phase (equivalent to a large value of m in Eq(l4)), ex- ternal resistance to mass transfer was small compared to resistance within the falling drop. The extraction rates were satisfactorily represented by the Handlos-Baron in- ternal mass transfer coefficients (Eq(6)). Both chemical and isotopic exchange in a fused salt- liquid metal system were investigated by Olander(17), Chemical exchange involved extraction of zinc from a zinc- lead alloy by reaction with lead chloride in the LiC1-KCl eutectic: Zn(m) + Pb¥2(s) = Pb(m) + zn'? (s} (26) The apparatus used is shown in Fig. 5. Metal drops of diameters ranging from 1 to 5 mm were introduced into the top of the column. As in the experiments of Katz et al(lfi), precautions were taken to minimize extraction during drop entry and after drop fall had been completed. Elimination of these end effects is essential if the extraction data are to be compared to the theoretical models described previously, all of which are based upon mass transfer only during free fall of the drop. Mass transfer during introduction of the drop into the salt was minimized (but not eliminated) by using one of the two entry techniques shown in Fig. 6. In the rod entry method, the drop was speared on the tip of a pyrex rod and lowered into a laver of solute-free salt which had been carefully poured on top of the extracting salt in the column, In the dropper entry method, drop fall was initiated by removing the plunger from the funnel stem shown in Fig. 6b. The molten metal pellet then fell first through a solute free layer and then into the extracting salt, Reduction of the bottom end effect was accomplished in the following manner. Just prior to drop release, the temperature at the bottom of the salt column was reduced below the freezing point of the salt. The falling drop encountered an advancing solidification front as it reached the end of its fall, and continued extraction was pre- vented by enveloping the drop in solid salt. Isotope exchange was studied by replacing lead chlor- ide in the salt phase by zinc chleride. The tracer zinc- 65 in the metal phase exchanged with natural zinc in the salt phase. 482 J~—POURING FUNNEL C 1= INSULATOR 1 QUARTZ TUBE 41 GLass wooL FILTER EXTRACTION COLUMN —-FURNACE —INSULATION i || —COLUMN SUPPORT | I | . ‘-———S_<—'INSULATOR ] REMOVABLE PLUG NITROGEN =™ — 5. Extraction column for fused salt-liquid metal sys=- tems used by Olander (17). 483 | -~ PELLET g/ e _—EUTECTIC SALT—\ e T _~EXTRACTING SALTWW//'/, ® ROD ENTRY DROPPER ENTRY 6. Methods of introducing metal drop into extraction column (17). ‘ 484 The fractions extracted were the same for both chem- ical and isotoplc exchange, and both varied linearly with the ratio of the concentration of extractant in the salt to the concentration of zinc in the metal. These two ob- servations suggested complete control of the extraction process by external (salt phase) mass transfer. The ex- ternal mass transfer coefficients obtained from the data were in rough accord with those predicted by the rigid drop model (Eq(l)). There were iidications that the mechanism of the inter- facial reaction 'id not follow the overall reaction of Eq(26) for all ccnditions. When pure zinc drops were con- tacted with lead chloride solutions, the recovered drop was partially encased by a black crust, which consisted primarily of tiny spheres of metal. It was hypothesized that the primary rapid reaction at the interface was: pb¥%(s) + 2zn(m) = Pb(m) + Zn;z(s) (27) The zinc subchloride begins diffusing towards the bulk of the salt but produces finely dispersed metal particles in the boundary layer by one of the following reactions: anz(s) = Zny + znT%(s) (28) cr Zn;Z(s) + pbt2(s) = Pbe + 22nY%(s) (29) This model predicts that when interfacial reaction is governed by Eq(27) rather than by Eq(26), twice as much zinc should be extracted for the same lead chloride bulk concentration (since each Pb*® reaching the metal drop moves two zinc atoms by reaction (27) but only one by re- action (26)). This feature of the subchloride model was qualitatively confirmed by the extraction data. The results for this system suggested that a number of forms of rate-enhancing interfacial turbulence, which have also been observed in aqueous-organic systems, may be im=- portant in high temperature inorganic systems as well, Drop oscillation was observed, and some experiments sug- gested accelerated transfer due to a Marangoni-type inter- facial motion. Mass transfer and drop fall velocities were measured in an immiscible l%gg}d metal extraction system by Pas=- ternak and Olander . Lanthanum-140 and barium 140 were transferred from 2-4 mm diameter drops o¢f the uranium- 485 chromium eutectic alloy freely falling through a column of molten magnesium at 1000°C. The all-graphite extrac- tion column shown in Fig. 7 was employed. At this temp- erature (which is 500°C higher than the three salt-metal experiments discussed above), not even the modest refine- ments in drop entry methods used in the salt-metal studies were feasible, The reactor-irradiated U-Cr drop was simply suspended from a tungsten wire and lowered into the molten magnesium. After melting, 1t detached from the wire and fell through the column, The progress of the drop down the column was followed by the response cof three colli- mated scintillation detectors placed at intervals along the length of the column., To minimize continued extrac- tion before the system was frozen after an experiment, a puddle of molten BaCl, was placed at the bottom of the column to receive the falling drop and remove 1t from dir- ect contact with the magnesium extractant. The drop terminal velocities measured from the responses of the three detectors were in good agreement (15%) with the Hu~Kintner relation(l , despite density differences, interfacial tension, and viscosities far beyond the range of this aqueous-organic based correlation. Mass transfer coefficients of lanthanum were measured directly by determination of La-140 activity pick-up in the magnesium ingot. By following the decay of the La-140 activity in the ingot for about three weeks after the ex~- periment and utilizing the radiochemical decay properties of the two member Ba-140-La-140 chain, the mass transfer coefficient of barium was determined as well. Since the distribution coefficient (Mg-to-U-Cr) of lanthanum was approximately 50 times that of barium, lanth=- anum extraction was expected to be controlled by transport within the drop and barium extraction by transfer in the external magnesium phase. These predicitions were borne out by the data. To within the precision of the data, lanthanum extraction followed the stagnant diffusion model of internal transfer (Eq(13)), while barium transfer agreed with the Higbie model. Conclusions Because of the rather forbidding array of problems as- sociated with working with reactive liquid metals at ele- vated temperatures, very few high temperature (>5300°C) liquid-liquid extraction experiments have been performed. Equilibrium measurements with the same systems of course encounter identical stringent restrictions on container materials and system cleanliness, but the kinetic experi- 486 Apparatus for liquid metal extraction at 1000°C (18). The furnace and radiation detectors are sketched. The magnesium charge is shown alongside the graphite extraction column. The top section of the column is a magnesium condenser. At the upper left is the stick and wire for introducing the pellet into the column; a pellet is attached to the wire. 487 ments have the additional requirement of moving one phase relative to the other in a manner which permits theoreti- cal determination of the flow patterns and hence the mass transfer coefficients. In addition, the kinetic experi- ments require reliable equilibrium data (in the form of distribution coefficients, solubilities, or two-phase equi- librium constants) in order to interpret the extraction data. Often, the lack of good thermodynamic or transport property data may be the greatest impediment to obtaining reliable mass transfer information. Where such data are available, the mass transfer studies have shown that the liquid metal-fused salt extraction kinetics are adequately described by one of the many mass transfer correlations developed for low temperature aqueous-organic systems. The extension of these empirical correlations for physical properties and temperatures far beyond the range 1in which they were developed appears justified. Although aqueous-organic correlations appear to ade- quately describe the results of the relatively crude high temperature kinetic measurements, there is no agreement as to which correlation 1is applicable for a particular set of conditions. For example, the metal phase internal coefficlients determined by Katz et al(l6) fit the Handlos- Baron model, while Pasternak and 0lander(18) found that the metal phase transfer coefficients agreed best with the stagnant diffusion model. The applicability of these two models (which differ quantitatively by a factor of five under these conditions) may depend upon the nature of the external phase; in the former, the continuous phase was a fused salt while in the latter, i1t was another liq- uld metal. As another example, the salt-metal experiments of reference 17 showed external transfer coefficients com- parable to those expected for a rigid drop, yet the ex- ternal coefficients determined by the metal-metal experi- ments of reference 18 were of the order expected from the Higbie model. These two models predict external coeffi- cient an order of magnitude apart. Retardation of mass transfer due to interfacial resist- ance has not been observed in any experiments., If slow chemical reaction is the form of interface resistance, this observation 1is not surprising. Chemical rate con- stants increase much more rapidly with temperature than mass transfer coefficients, and one would have to search diligently for a reaction between two fluid phases which was slower than diffusion in the temperature range from 500-1000°C. It is somewhat surprising, however, that inhibition of 488 extraction due to interface-seeking impurities have not been reported, particularly since these effects have been reported in aqueous—-organic systems. Since most liq- quid metals at these temperatures will react readily with impurities in the cover gas, in the solvent phases proper, or with the container material, there should have been a much more severe contamination problem than in relatively non-reactive low temperature aqueous-organic systems. The mass transfer studies of high temperature liquid- liquid systems have not yet demonstrated unequivocally that prediction methods based upon experience with aqueous- organic systems are directly applicable to pyrochemical reprocessing systems, They have shown, however, that for devices such as spray columns and mixer settlers, the lig- uid metal-fused salt and immiscible liquid metal results at least fall in the range of the common aqueocus-organic correlations. So far, no rate-limiting phenomena pecul- iar to high temperature inorganic species have been un- covered, However, problems unique to liquid metals may appear in other contacting devices such as packed col- umns where phase contact involves a third solid phase. In this case, the degree of wetting of the packing by one or both of the immiscible liquid phases may be im- portant, 489 I1, Transport Properties of Liquid Metals Interpretation of mass transfer experiments in liquid metal-fused salt systems requires a knowledge of solvent viscosities and solute diffusivities. The dimensionless Sherwood, Schmidt, and Reynolds numbers, which are used to correlate mass transfer coefficients, depend upon these transport properties. In order to apply the correlatiocmns, measurements or estimates of viscosity and diffusivity for the particular system of interest are required. In recent years, considerable effort has been devoted to measuring the viscosity of pure liquid metals and lig- uid metal alloys and also to the measurement of diffusion coefficients in liquid metal systems. Many of these ex- periments have been on systems of particular importance in nuclear reactor and pyrochemical fuel processing tech- nology. In addition to experimental measurements, methods for estimating viscosities and diffusivities have been develo~ oped., These methods involve semi-empirical correlations in which the value of an adjustible parameter is deter- mined by the existing data. Assuming that liquid metals al)l show similar behavior with respect to the adjustible parameter, the correlation can then be applied to systems for which data are not yet available. Viscosity Correlations Two methods for estimating liqui%ZB?tal viscosities are those of Grosse and Chapman . Grosse's method assumes an exponential dependence of viscosity on tempera- ture: n = a exp [Hn/RT] (30) where 1 is the viscesity in poises, H, 1s the activation energy of viscous flow in cal/g-atom, R i1s the gas constant in cal/g-atom-°K, and T is the absolute temperature in °K. Andrade's expression(21,22) is used to estimate viscosity at the melting point. n_ = 5.7 x 107" —(AT“])UZ (31) m : y2/3 where A is the atomic weight, I, the melting point, and V is the atomic volume at the melting point in cc/g-atom. Grosse finds the following empirical correlation between 490 Hn and the melting point, Tm. log,gH, = 1.348log) T -0.366 (32) From Eqs(31) and (32), the pre-exponential factor, a, in (30) can be determined. Eq(30) can then be used to esti- mate viscosity over the entire liquid range. Chapman(zo) uses the radial distribution function con- cept of the liquid state to establish a functional rela- tionship between reduced viscosity, reduced veclume, and reduced temperature. These quantities are reduced with a distance parameter (the Goldschmidt atomic diameter) and an energy parameter. Energy parameters for the liq- uid metals are empirically found to be about 5.2kT. Grosse's correlation requires the molecular weight, melt- ing point, and density at the melting point; Chapman's correlation requires the molecular weight, melting point, density over the temperature range of interest, and Gold- schmidt atomic diameter. Viscosities of the Actinides The only experimental determination of the viscosity of plutonium is that of Jones, Ofte, Rohr and Wittenberg(23) whose measurements cover the range from 645~950°C. O0fte(24) has also measured the viscosity of uranium contained in sub-stoichiometric zirconia crucibles from 1141-1248°C. The viscosity cf uranium has also been measured recently by Finucane and 0lander(23) in both tantalum and beryllia crucibles. Their measurements cover the range from the melting point to 1532°C. All of these measurements employ the oscillating crucible technique, which is an absolute mnethod requiring no calibration with a liquid of known viscosity. The data for these three measurements are shown in Figure 8 along with values predicted by the Grosse and Chapman correlations, All of the measured viscosities are higher than the predicted values, and it may be that the actinide metals behave as a class apart from other met- als. The data of Finucane and Olander for uranium show greater precision over a wider temperature range than do those of Ofte, However, the ~ 357 discrepancy between the two sets of data will have to be resolved by a third measurement. Viscosities of the Rare Earths Viscosities and densities of the molten rare earths lanthanum, cerium and praseodymium have been measured by 491 (cP) Viscosity 10 | I T | 9 | A U, exptl. Finucane and Olander _ e Pu, exptl. Jones, et al., 8 _ m U, exptl., Ofte 7 —+.— Chapman’s correlation ,calc --- Grosse’s correlation, calc. 6 a, — / // 5 / - / '/. 7 4 - | 31 _ H33° 1300 1200 900 800 700 640°C - | | o 2 | | | i | | 0.5 0.7 0.9 .0 10% (ok -3 T XBL694-2609 8. Viscosity of uranium and plutonium. 492 Wittenberg, Ofte and Rohr26,27) in the temperature range from the melting point to over 1000°C. Figure 9 compares the data with Grosse's and Chapman's correlations. Both correlations fall within the scatter of experimental data for all three metals. It should be noted, however, that the activation energy from the least squares fit is much smaller than that predicted by either of the two correla- tions. Therefore, extrapolation of the data to higher temperatures may involve considerable error. Viscosities of Plutonium Alloys Since the early 1960's, the viscosities of several plutonium alloys have been measured at the Mound Labora- tory. These alloy systems are By-Ce-Co(23), Pu-Ce(ZS), Pu—Fe(zg), Pu-U(30) ang Pu—Ga(3 . These alloys are of interest as possible reactor liquid fuels. In all cases, the addition of alloying elements to plutonium produces a viscosity increase. The effect is most pronounced in the case of Ga where a 3.3 atom percent addition causes a 50% rise above the viscosity for pure plutonium. In the Pu-U system, there is a relative wviscosity maximum, apparent in all isotherms up to 800°C, at a composition of 10.8 atomic percent uranium. This is the composition of the lowest melting alley in the system, 620°C. Vis- cosity isotherms for the Pu-Fe system show a maximum at the eutectic composition of 9.5 atomic percent iron. Vis- cosity isotherms for the Pu-Ce system are the mgst unusual cf all. Additions of Ce produce a rise in viscosity to a maximum at 5 atomic percent cerium. Viscosity then de- creases to a minimum at 14 atomic percent cerium followed by a slight rise at the eutectic composition, 16.5 atomic percent c¢erium., There is no theory to adequately explain or predict the wide variety of viscosity behavior in liq- uld metal alloys. Some systems show a viscosity minimum at the eutectic point. For such a frequently studied sys- tem as the lead-tin pair there %? disagreement in the lit- erature. Fisher and Phillips(3 observe a viscosity min- imum at the eutectic point; Kanda and Colburn cbserve linear viscosity isctherms over the complete composition range from pure tin to pure lead. Some Russian investi- gators have applied thermodynamic (heat of mixing) data to the study of alloy viscosity behavior. Burylev has shown a relationship between viscosity isotherms for Cu-Ag alloys (which have a minimum) and thermodynamic data. Eretnov and Lyubimov(33) have investigated the relation- ship between viscosity isotherms for several copper alloy systems which have maxima and heat of mixing data, while for other systems volume changes on mixing are more impeortant. 493 {cP) Viscosity 9. 1O T T l 815°C = | I I _] 8 e ® A Exptl. Wittenberg et al. ] Sr —-— Chapman's correlation,calc, n Lanthanum -~-- Grosse’s correlation, calc. | —- . .-:![:::);::uf— B gy - . | e . - 2 ® A 1000°C 950°C 900°C 850°C 10 |- i ~ 1 1 — i 8 ] 6 . T— Cerium -] o 4 Praseodymium 0.78 Viscosity of lanthanum, 494 cerium and praseodymium. Diffusivity Correlations Mutual diffusion coefficients in a number of dilute liquid meta% sglutions have been correlated by Pasternak and Olander using a modified form of absolgge rate theory. The theory was developed by 0lander (3 toc cor- relate mutual diffusion data in dilute organic systems, and uses viscosity data for both pure solvent and pure solute to estimate the difference between the free energies of activation of the viscous and diffusive processes. The correlation for liquid metals is shown in Figure 10; the dashed lines represent 25% deviation from the best line. The dimensicnless group Y and the parameter f are defined by: d 5.31|{Vil/3 Y =(Tfl m )(fi) /3 exp [0.56] (32} * * o =[] = AA where k is the Boltzmann constant, N is Avogadro's number, and AFKA and AFEB are the free energies of activation for viscosity for pure sclvent and pure solute respectively. Free energies of activation are obtained from viscosity data by the expression = E% exp [AF*/RT] (34) where h is Planck's constant. For systems where the diffusing solute is a solid at the temperature of measurement, AFgp is obtalred by a lin- ear extrapolation of AFBB values obtained in the 1liquid region according to: AF% = AH*-TAS* (35) where AH* and AS* are the enthclpy and entropy of activa- tion of viscosity obtained for pure sclute metal above its melting point. Liquid metal diffusion coefficients have often been estimated by the Stokes-Einstein equation: _ kT 6mrN (36) where r is a characteristic radius of the diffusing solute 495 96% 10. -2.0 -1.5 -1.0 0.5 O 3 Modified absolute rate theory correlation of mutual diffusion coefficients in liquid metals. 0.5 atom. The Stokes-Einstein equation is based on a hydro- dynamic model of a large solute atom moving through a con- tinuous fluid with a "no-slip" condition at the solute atom surface., For liguid metals, the size parameter r in Eq(37) 1is wusually taken as the ifionic rather than the atomic radius of the solute atom. Based on a study of viscosities in liquid metals, Eyring(3%9) concludes that the unit of flow is probably the metal ion stripped of its valence electrons, For self-diffusion, §=0 and Y=1 in Eqs(32) and (33). Therefore, if the self-diffusion coefficient for the pure solvent Dpp is known, the mutual diffusion coefficient for the solute B in solwvent A, Dpp, can be related to Dap by Eq(33): _AB exp [0.58] (37) The ratio of mutual to self-diffusion in a particular solvent according to the Stokes-Einstein equation is: - A A3 . . (38) Data for a series of solutes in liquid silver(40,41,42) is presented in Table 1., Here the Stokes-Einstein esti=- mate using Goldschmidt ionic radii of the diffusing sol- ute and solvent molecules shows better agreement with the data than the same estimate using atomic radii. Diffusivity of Uranium in Liquid Metals Figures 11 and 12 present diffusivity data for uran- ium in bismuth, zinc, cadmium and aluminum. Data for all four systems were obtained by Hesson, Hootman, and Burris {43) The aluminum data of Mitamura(4%4) are also included. The modified absolute rate theory of Eqs(32) and (33) shows a better agreement with the data than the Stokes-Einstein equation (36), although except for bismuth, both predic- tion methods are high. Dif fusivity of the Rare Earths in Molten Uranium The extraction of rare earth fission products from uranium fuel is an important part of some liquid metal 497 11. 10~ l I | T 1 B 8 - - — \ — ~. ~ 6 \'\ ~ — - . ~ — \0 ~ ~ 4 ~ — ~. ~ - ~. \\ — ‘ ‘\1\\ .\\\\ 4 A 2 - el T o A ~ v v : - 4 € Q © { L" — | - - © o8} . — v — A 0.6 — v 0.4 - A Cd exptl. Hesson et al. v Al exptl. Hesson et al. o 2 —— Exptl. Mitamura et al. ' —-— Modified absolute rate theory, calc. ---- Stokes-Einstein equation, calc. 0.1 | l 1 | H 0.8 1.0 1.2 1.4 10 T (°K™") Diffusivity of uranium in cadmium and aluminum. 498 10 LN T l T | m® Bi exptl- Hesson et al,. _ ® Zn exptl. Hesson et al,. 8~ —.— aAbsolute rate theory , calc. | ——— Stokes- Einstein eguation, calc. 7 ~ — ] ~N N \\ 6 s \ u ~ . ~N N ~ \§ N, 5T NN\ . 7 o N\ N\, Q ’ [ ] \ o AN . y AN N £ . N O 48 \. \ . \ — -n ANERN . 3 o ° ANERN = AN . ] AN » a) 3| ¢ - ® ® ° ] 2 1 I | 1 1.0 1.2 1.4 ok -1 — (°K™) T 12, Diffusivity of uranium in bismuth and zinc. 499 Table 1. Comparison of Calculated and Experimental Mutual to Self-Diffusion Coefficient Ratios for Various Solutes in Silver at 1200°C Dyp/Paa Stokes-Einstein(Eq.38) Modified using using Solute in Absolute Rate Atomic Ionic Silver Exp.?2 (Eq. 37) Radiib RadiiP Gold 0.97 0.90 1.00 0.82 Tin 1.32 1.26 0.91 1.53 Indium 1.35 1.36 0.92 1.23 Antimony 1.37 1.23 0.90 1.26 a From equations representing data (40,41,42). b Taken from C. Smithell's "Metals Reference Book", 4th Ed., Plenum (1967). 500 reprocessing schemes. It 1s therefore desirable to have available data for the diffusivity of rare earth metals in molten uranium, The only data in the literature are those of Smith(43) who measured the diffusivity of cerium in uranium over the temperature range 1170-1480°C. The experiments were carried out in fairly large tantalum crucibles (7 mm. diameter). According to the author, con- vection at the higher temperatures due to the large dia- meter increased the uncertainty of the results in the range 1400-1480°C. The value determined at the melting point of uranium, D=8,8 =x 10=2 cm /sec, seems somewhat high also. The value calculated by the Stokes-Einstein equation is 1.5 x 10-3 cm?/sec. The modified absolute rate method gives 2.5 x 107° cm“/sec. The modified absolute rate method for calculating dif- fusivities is convenient to use since it requires only viscosity data for the pure solvent and pure solute. The Stokes=Einstein equation involves an uncertainty as to the correct radius to assign the diffusing solute atom, although most workers use the ionic radius. In general, the modified absclute rate method 1s more accurate. When applied to common binary metal systems, the Fig, 10 exper- imental data are correlated to within *25%. For uranium as solute, good agreement 1s obtained only with bismuth. The 50% discrepancy in zinc, cadmium and aluminum may in- dicate that the diffusing amount 1s larger than a single uranium atom, and may be an intermetallic compound of uranium and the solvent. Measurement of Liquid Metal Transport Properties The measurement of the coefficients of viscosity and diffusivity in liquid metals 1is subject to the same exper- imental problems as the mass transfer studles discussed earlier. Because of the high temperatures involved, con- struction material and equipment complexity are severely restricted (at least within the hot zone of the furmnace). The generation of a prescribed flow field in the liquid metal is all-important in both viscosity and diffusivity measurements. In the former, the fluilid mechanices of the device must be known because the relation between the shear stress and veloclty gradients in the fluld estab- lishes the coefficlent of viscosity. In diffusion mea- surements, all fluld velocities should be zero. Viscositz The various methods which have been used to measure liquid metal viscosities have been reviewed by Thresh (46), 501 In the capillary method, the liquid is forced through a narrow tube by inert gas pressure or by the hydrostatic head of the moving fluid itself. The use of an inert gas is often incompatible with the vacuum system required to insure fluid cleanliness. Surface tension effects may in- hibit flow in the capillary. A rather large hot zone is necessary and the mechanical complexity of the open flow renders this method unsuitable for high melting reactive metals. In rotational visccometers, the liquid metal is con- tained as an annular ring formed by a central cylinder rotating at a constant speed and a stationary outer cyl- inder. The viscosity can be determined by the torque on the outer c¢ylindex. The problems of driving the inner cylinder and of measuring the tordque on the outer cylinder are difficult in high temperature vacuum systems. In ad- dition, accumulation of slag (due to impurities) at the rather small annular liquid surface between the two cyl- inders significantly affects the measurements. Oscillating crucible viscometers have been used almost exclusively for viscosity measurements above 1000°C. 1In this method, a specimen of metal is sealed in a cylindri- cal crucible suspended in the hot zone of a furmace by a torsion wire. The pendulum is given an initial twist and the damping of the torsional oscillations is directly re- lated to the viscosity. The amount of metal used is quite modest: specimens are v 2 cm diameter by v 7 cm bigh, and of an easily machinable shape. The simple geometry also permits fabrication of nearly any high temperature con- tainer material for the crucible, The crucible (or an outer refractory metal sheath) can be vacuum sealed by electron beam welding, thereby permitting complete iso- lation of the melt from the vacuum environment and attain- ment of temperatures at which the vapor pressure of the liquid would otherwise be unacceptably high for the vacuum system, The only motion required is simple torsional os- clllation, which can be initiated from ocutside the vacuum system. No measurements within the vacuum system are re=- quired; the temperature can be measured with an optical pyrometer and the period and damping constant ¢f the os- clllation are determined by the motion of a light beam re- flected from the part of the pendulum outside of the fur- nace. The liquid metal surface is relatively large * 2 cm diameter), thereby reducing the effects of surface tension and slag-wall interactions. The apparatus depicted in Fg. 13 is currently in use in our laboratory, but 1is typical c¢f the design of most high temperature vacuum oscillating crucible viscometers. 502 A\ 1 OM] 13. Oscillating crucible vacuum viscometer (25). 503 The furnace heater element (A) is surrounded by a series of tungsten radiation shields (B). The entire system is contained within a vacuum system (C). The torsion pendulum consists of three primary parts: the crucible (D) which contains the metal specimen; a rod (E) which rigidly connects the crucible with the part of the pendulum outside of the hot zone; and the external portion of the pendulum (¥} which has a polished surface to reflect a beam of light by which the motion of the pen- dulum is monitored. A hole is drilled through the pen- dulum perpendicular to its axis of rotation into which rods may be inserted and fixed such that the moment of inertia of the pendulum may be varied. The small chuck at (G) attaches the pendulum to the torsion wire. A pic- ture of the entire pendulum in shown in Fig. 14. The torsion wire (H) extends from the pendulum to a second chuck which is attached to a rotatable holder (J) which rests on a support plate (K}. The holder is attached mechanically to a rotary feedthrough (L) by which the ro- tary motion of the pendulum may be initiated from outside the vacuum system. The temperature within the furnace region is measured by an optical pyrometer (M) which is sighted through a right angle prism (N) into a hole, 1/8 inch in diameter, drilled through the bottom shield pack into the inner furnace region. Dif fusion Coefficients Most liquid metal diffusivity measurements are per- formed in "capillary" cells. In room temperature measure- ments on aqueous or organic solutions, diffusion occurs along the length of narrow bore tubing, and the concen- tration profile can be directly measured by cptical methods. With liquid metals, however, the capilillary 1is more aptly described as a tube, since diameters up to 7 mm have been used. The concentration profile cannot be measured in situ; rather the final profile is determined by freezing the system and sectioning the capillary for analysis by chemical or radiochemical means. Application of the com- mon capilliary Eeghods to liquid metals has been discussed by Niwa et al( 7, for low melting metals, a U-tube containing the pure solvent is lowered into an alloy bath. After a definite diffusion time, the U-tube is withdrawn from the bath, quenched and sectioned for analysis., Because of the very uniform temperature of the alloy bath into which the U- 504 Viscometer pendulum (25). tube is placed, temperature variations and the resulting natural convection in the diffusion region are virtually eliminated. However, the method is inconvenient for high temperature vacuum environments because of the rather sub- stantial amount of metal required for the alloy bath and because of the complications Involved in raising and low- ering the U-tube into and out of the hot zone, The bath-less capillary method utilized by Smith(45) for measurements of diffusion in uranium is suitable for reactive liquid metals at temperatures above 1000°C. Smith's apparatus is shown in Fig. 15, In this arrange- ment, both components are placed in the c¢ylindrical cru- cible which serves as the diffusion "capillary". Care must be taken to avoild convective mixing of the two com=- ponents (which may or may not be immiscible when liquid) during melting. This method 1is more susceptible to nat- ural convection mixing due to temperature gradients than is the bath method. ©Natural convection can be reduced by using specimens of large length-to-diameter ratios, although this geometry accentuates the effect of gettering of the diffusing solute by reaction with the container wall. The effects of vertical and horizontal temperature gradients on induced convective motion have been analyzed by Ver- hoeven and found to be an unlikely source of systen- atic error (theoretically at least). The mixing which occurs during melting and freezing of the sample, however, can undcubtedly contribute to high apparent diffusion coef- ficlents. 506 L0S / - fi/ 4 %) O-RING — o £73 . MOLYBDENUM H RATIATION o i SHIELDS THERMOCCUPLES A N N 4+ Y ) TANTALUM , CRUCIBLE , URANIUM & THERMOCOUPLE LEADS VACUUM PUMPS WATER COOLED BRASS FLANGE CERAMIC (McDANEL TUBE) MOLYEBDENUM SUPPORT RODS TANTALUM BUCKET GLOBAR FURNACE CERIUM 15. Diffusion apparatus of Smith (45). References ANL 7548, p. 109 (1969). Dunn, W.E., C.F. Bonilla, ¢. Ferstenberg, and B. Gross, A.L.Ch.E. Journal, 2, 184 (1956). Kassner, T.F., J. Electrochem. Soc., ll4, 689 (1967). Sideman, S. and H. Shabtai, Can. J. Chem. Eng., 42, 107 (1964). Higbie, R., Trans. Am, Inst. Chem. Engrs., 31, 365 (1935). Newman, A.B.,, Trans. Am. Inst. Chem. Engrs., 27, 203 (1931). Kronig, R. and J.C. Brink, Appl. Sce. Res., AZ, 142 (1950}). Handlos, A.E. and T. Baron, A.I.Ch.E. Journal 3, 129 (1957). Angelo, J.B., E. Ch.E. Journal, 1 N. Lightfoot and D.W., Howard, A.I. 2, 751 (1966). Rose, D.M. and R.C. Kintner, A.IL.Ch.E. Journal, 12, 330 (1966). Olander, D.R., A.I.Ch.E. Journal, 12, 1018 (1966). Patel, J.M. and R.M. Wellek, A.I.Ch.E. Journal, 13, 384 (1967). Pasternak, A.D., UCRL-16108 (1966). Hu, C. and R.C. Kintner, A.I.Ch.E. Journal, 1, 42 (1955), Bonilla, C.F., First Int'l Conf. on Peaceful Uses of At. Energy, United Nations, Geneva, 9, paper no. 122, pp. 331-40 (1955). Katz, E.M., F.B. Hill and J.L. Speirs, Trans. Met. Soc. AIME, 218, 770 (1960). Olander, D.R., Nuc. Sci. Eng., 31, 1 (1968). Pasternak, A.D. and D.R. Clander, A.I.Ch.E. Journal, 14, 235 (1968). 508 19. 20. 21. 22, 23. 24, 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. Grosse, A.V., J. Inorg. and Nucl. Chem., 25, 317 (1963). Chapman, T.W., A.IL.Ch.E. Journal, 12, 395 (1966). Andrade, E.E., Phil. Mag., 17, 698 (1934). Andrade, E.W., Phil, Mag., 17, 497 (1934). Jones, L.V., D. Oftey, W,G. Rohr, and L.J. Wittenberg, Trans. of the ASM, 55, 819 (1962). Ofte, D., J. Nucl. Mater., 22, 28 (1967). Finucane, J. and D. Olander, unpublished data. Wittenberg, L.J., D. Ofte, and W.G. Rohr, Rare Earth Research, Vol. II, ed. by K.S. Vorres, Gordon and Breach, New York, 1964, pp. 257-275. Rohr, W.G., J. Less-Common Metals, 10, 389 (1966). Ofte, D., W.G. Rohr, and L.J. Wittenberg, Plutonium 1965, Proc. of the 3rd Internat'l Conf. on Plutonium, London, 1965, ed. by A.E. Kay and M.B. Waldron, Chap- man and Hall, London (1965), pp. 405-419. Ofte, D. and L.J. Wittenberg, Trans. ASM, 57, 917 (1964). MLM-1402, Reactor ¥aels and Materials Development Plutonium Research: 1966 Annual Report, October 16, 1967, pp. 40-44. Ofte, D. and W.G. Rohr, J. Nucl. Mater., 15, 231 (1965). Fisher, H.J. and A. Phillips, J. of Metals, 1060 (L954). Kanda, F.A. and R.P. Colburn, Phys. and Chem. of Liquids, 1, 159 (1968). Burylev, B.P. translated in Russian Journal of Phys- ical Chemistry, 41, 53 (1967). Eretnov, K.I. and A.P., Lyubimov, Ukr. Ziz. Zhur., 12, 214 (1967). Gvozdeva, L.I. and A.P. Lyubimov, Ukr. Ziz. Zhur., 12, 208 (1967). 509 37. 38. 39, 40. 41. 42. 43. bb. 45. 46. 47. 48. Pasternak, A.D. and D.R. Olander, A.I.Ch.E. Journal, 13, 1052 (1967). Olander, D.R., A.L.Ch.E. Journal, 9, 207 (1963). Glasstone, S., K.J. Laidler, and H. Eyring, Theory of Rate Processes, McGraw-Hill, New York (1941). Swalin, R,A. and V.G. Leak, Acta. Met., 13, 471 (1966). Gupta, Y.P., Acta. Met., 14, 297 (1966). Leak, V.G. and R.A. Swalin, Trans, Met. Soc. AIME, 230, 426 (1964). Hesson, J.C., H.E. Hootman, and L. Burris, Jr., J. Electrochem. Tech., 3, 240 (19653). Mitamura, N. Nippon Genshiryoku Gakkaishi, 5, 467 (1963). (Data from NSA 17, 32500). Smith, T., J. Electrochem. Soc., 106, 1046 (1959). Thresh, H.R., Trans. ASM, 55, 790 (1962). Niwa, K., M. Shimaji, 8. Kado, Y. Watanabe, and T. Yokukawa, Trans. AIME, 209, 96 (1957). Verhoeven, J.E., Trans. AIME, 242, 1937 (1968). 510 * MULTISTAGE CONTACTORS FOR LIQUID METAL-SALT EXTRACTION R. D, Pierce, W. E, Miller, J. B. Knighton and G. J. Bernstein Chemical Engineering Division Argonne National Laboratory Argonne, Illinoils 60439 U. S. A. Abstract Investigations of the chemistry of plutonium, uranium, and fission products in liquid metal-molten salt systems show that good recovery of purified plutonium is possible by liquid metal- molten salt extraction. The application of a semi-continuous mixer-settler for such extractions is under development. The flowsheet involves the selective extraction of rare earths from a Mg-Cu-Pu-fission products alloy into a series of salts at 650°C, selective extraction of plutonium into another salt away from the nobler elements and uranium, and re-extraction and concentration of the purified plutonium in another alloy. The design specifi- cation 1s recovery of 99.8% of the plutonium with a decontamination factor of 106, During a run, some of the fluid phases flow continuously between stages, and the feed alloy is recirculated through the extractor several times. Other phases are only moved batchwise between runs. The resultant flow pattern makes application of conventional designs unattractive or impossible. The design selected consists of a modular assembly of mixer and settler chambers. A simple agitator- pump moves the fluids out of each mixing chamber to a settling chamber. The separated phases flow to adjacent mixing chambers through overflow spouts. Pumping and mixing performance has been investigated for various designs. A reference design has been selected, and extraction runs are ready to begin. * Work performed under the auspices of the United States Atomic Energy Commission. 511 Introduction The Salt Transport Process under development at Argonne incorporates multistage liquid metal-salt extraction for the purification of plutonium from fast reactor fuels. This paper is a report on the status of the development of equipment for the plutonium purification step. The complete Salt Transport Prgcess flowsheet is presented in another paper in this symposium.(l The feed to the extraction step is a liquid alloy of Mg-42 at. % Cu-2 at. % Pu-1 to 2 at, % fission products. Many of the fission products and most of the uranium are separated from plutonium during the decladding and reduction steps that precede the extraction step. The objective of the extraction step is to remove yttrium and the rare earth elements for disposal and to recover plutonium selectively from uranium, nobler fission elements (zirconium, niobium, ruthenium, ‘ molybdenum, technetium, and palladium), and residual cladding elements (iron, chromium, and nickel). These separations are accomplished by selective extraction of yttrium and the rare earths into a molten chloride salt, and subsequent selective extraction of plutonium into another chloride salt. The purified plutonium is concentrated by extraction back into an alloy. Neptunium and curium follow plutonium through the extraction, but the other transuranium elements are removed with the rare earths. The metal-salt extractions are the results of the equilibrium reactions, + % MgCl A - n A (alloy) 2(salt) « 2Cla(sale) T 2 M (alloy), (1) where A is an element under consideration, n is the valence of A. The extractions that can be effected are evaluated from the distribution coefficients attainable. The distribution coefficient, D, is defined as y A A X (2) ‘ A where y, is the concentration of the chloride of A in the salt (mol %), X, is the concentration of A in the metal (at. %). D The extraction factor, R, is useful in process design. It is defined as the ratio of the amount of an element in the salt to the amount in the metal: S ¥y D, § R = A _ A - = s A M xA M (3) where S is total quantity of salt (moles), M is total quantity of metal (moles), 512 The separation factor, o, for two elements is defined as the ratio of their distribution coefficients: D o _ _A = TH AB DB (4) where A and B are two different elements. The magnitude of the distribution coefficients is dependent on the equilibrium constant for the reaction (equation 1), the activity coefficients for the components in the metal and salt phases, and the magnesium and magnesium chloride content of the metal and salt phases. The application and control of these variables in designing chemical segarations is discussed in another paper in this symposium.{2) The metal and salt phases for the present separation operations were selected to optimize Plutonium-rare earth separation and to provide distribution coefficients that allow workable extraction factors at convenient salt-to-metal ratios. The selection of an extraction philosophy for the Salt Transport Process required consideration of many factors, such as composition specifications for product and wastes, plutonium inventory, criticality, waste volumes, materials of conatruction, process control, process flexibility, and phase transfer efficiency. The goal of the extraction step is recovery of 99.8%7 of the plutonium with 8 decontamination factor of 10® for fission product radio- activity. The waste streams are to be as small as compatible with decay heat removal, Batchwise operation was rejected for the multistage extraction because of the many phase transfers that would be involved and because of the difficulty in performing the clean phase separations necessary to utilize the large separation factors available, Countercurrent columns were considered for the process(3) but are not being used in the current flowsheet because they are not compatible with the flow pattern selected. A semi-continuous procedure employing a battery of mixer~settlers has been adopted. Extraction Strategy The first stages of the extraction step are designed to move rare earths from the feed solution into a waste salt phase without undue loss of plutonium. This might conventionally be performed in a series of stages as indicated in Figure 1. The scrubbing stages remove rare earths from the product, and the stripping stages prevent excessive plutonium logses. For the present metal-salt systems, a relatively small scrubbing section of 2 or 3 stages would be required but about B8 more stages would be required for stripping. The semi-continuous mode of operation which has 513 FEED ALLOY (Pu, RE, NM) WASTE SALT FRESH (RE) SALT ey | et md fethr—eret — e | pe——— o — — E—— e — —.J — FRESH ALLOY METAL (Pu, NM} STRIPPING SECTION SCRUBBING SECTION RE RARE EARTHS NM NOBLER METALS (MORE NOBLE THAN Pu & U) Figure 1. Countercurrent Extraction 514 been labeled 'metal transport' has proved very attractive. Figure 2 is the flow diagram and illustrates the following operations: (1) Alloy containing plutonium and fission products is contacted with salt in a series of stages to perform the scrubbing of rare earths. The salt is not circulated between stages but is 'captive" in each stage, (2) Plutonium is extracted from the alloy in a "donor" stage in which the plutonium extraction factor is increased. (3) The alloy is recycled through the same captive salts to perform the stripping of plutonium from the salt until the level of plutonium in the stage 1 salt is sufficiently low for disposal. (4) Metal circulation is terminated. (5) The alloy is returned for reuse in earlier process steps and fresh feed is charged to the feed tank. (6) The salt in stage 1, containing most of the rare earths, 1s discarded. (7) The salt phases in stages 2, 3, and 4 are transferred to stages 1, 2, and 3, respectively, (8) Fresh salt is added to stage 4. (9) Processing of a new batch begins with circulation of the feed alloy. Following the rare earth removal, the plutonium is selectivel extracted from the nobler elements by a 'salt transport" step.(2¥ Plutonium is extracted into a salt in the donor stage; this salt is scrubbed in the next stage with a captive alloy for further nobler metal removal and to minimize introduction of feed solvent to the product alloy; the salt is then contacted with a captive "acceptor'" alloy which extracts the plutonium; the salt, which serves only as a transport medium, is recycled to the donor stage. At the conclusion of an extraction run, the acceptor alloy con- taining the purified plutonium is removed and fresh acceptor alloy is charged. The scrub alloy has capacity for many runs, but eventually it is discarded and replaced. In normal operation, nothing accumulates in the transport salt, and it can be used indefinitely. The level of uranium and nobler metals in the Mg-Cu alloy, which is reused, is controlled by their removal in the reduction step and by discarding a small portion of the Mg-Cu alloy before recycle to the reduction step. The entire extraction step is 1llustrated schematically in Figure 3. The seven stages indicated are expected to provide the desired decontamination factor of 106, but one or two additional stages can be added if extraction experience indicates they are necessary. The composition of the process solvents are summarized in Table I, and some pertinent distribution coefficients are listed in Table II, The distribution coefficients are relatively constant over the range of solute concentrations to be encountered. From these distribution data the excellent separation factor of 1000 obtained between plutonium and cerium with the Mg-Cu solvent can be seen. The distribution coefficient for plutonium being unity allows extraction of plutonium in either direction depending on the salt-to-metal ratio (see equation 3). The scrub alloy has the desired low distribution coefficient for nobler metals (zirconium is the most difficult to separate) and a sufficiently 515 SALT SALT SALY SALT SALT FEED J J e u ALLOY ALLOY —~|PLUTONIUM[—" (Pu, RE, NM) o 2 3 o4 (Pu, NM) DONOR ' ALLOY (NM) CONTINUOUS FLOW METAL TRANSPORT (DURING RUNS) ALLOY {Pu, RE, NM) WASTE SALT FRESH (RE) SALT ' 2 3 a4 l BATCH MATERIALS TRANSFERS ‘ ALLOY (BETWEEN RUNS) (NM) RE * RARE EARTHS NM = NOBLER METALS Figure 2. Metal Transport Mg-Cu=-Pu=RE~NM-U | I WASTE i SALT saLT . : SALT TRANSPORT . SALT- SALT~- L-d I P I Py [ Pu SALT 7 * g Zn-Mg-Pu : METAL TRANSPORT .I ‘o 4 ’ ‘ - - - ¥ ' Mg-Cu-NM-U 1Mq—Cd ! ' Zn-M Mg Cu=NM-U Mg-Cd-NM-Cu " 9 g CONTINUOUS METAL FLOW DURING RUNS ——= CONTINUOUS SALT FLOW DURING RUNS -—--+ BATCH TRANSFER BETWEEN RUNS SALT = MgClg-NaCl-KCI-MgFg RE = RARE EARTHS NM = NOBLER METALS Figure 3, Plutonium Purification Step of 516 Salt Transport Process Table I Process Solvents Composition Density at 650°C (g/cc) Metal Transport - Donor Alloy Mg-44 at. % Cu 3.5 Scrub Alloy Mg-20 at. % Cd 2.8 Acceptor Alloy Zn-30 at., % Mg 5.0 Metal Transport and Salt MgCl;-30 mol % NaCl- 1.7 Transport Salts 20 mol %-KC1-3 mol % MgF2 Table II Distribution Coefficients Distribution Coefficient, D = mole fraction in salt/atom fraction in metal MgCl,-30 mol 7 NaCl-20 mol 7% KCI1- Salt Solvent 3 moi % Mng Temperature = 650°C DPu DU DCe DZr Metal Transport and Donor Stages 1 0.1 1x10° 6x1073 Scrub Stage 0.5 5x10°% 60 2x1072 Acceptor Stage 11‘:10_3 51-:10_-4 0.2 1):.‘10-9 high value for plutonium to minimize plutonium holdup in the scrub stage. The Zn-Mg acceptor alloy provides the low distribution coefficient necessary to extract plutonium out of the transport salt. The interaction of many parameters are encountered in the proposed stages. These include metal tc salt ratios, tetal volume of captive phases, hold-up times, number of metal and salt eycles, and the number of stages, A computer program was prepared and is used to calculate expected extraction performance for different combinations of parameters; however, extraction rate data are needed to permit optimization of other parameters. Selection of Contactor Configuration Mixer-settler designs used in the chemical processing industry have been reviewed. (4-9) Probably the simplest and most versatile form is one in which mixing is done 1in one vessel and settling in 517 another. A variety of flow patterns of heavy and light phases may be employed with this equipment. Mixing and settling regions are incorporated inside of a single vessel in some designs. Arrangements have been made in which the stages are stacked vertically so that only one rotating shaft is used for agitation in all of the stages. Such an extractor forms a tower in which the heavy liquid enters at the top and leaves at the bottom, and the light liquid enters at the bottom and leaves at the top. Density difference between the two immiscible liquids causes the phases to separate in each settling region after mixing and causes the light liquid to flow up the tower. Another arrangment places the stages side by side., The stages are commonly in the shape of boxes which may be compartmentalized by baffles. Each mixing stage has its own agitator shaft. Such units are commonly known as the box-type mixer-settler, Some mixer-settler designs were eliminated from consideration for a pyrochemical process because of the elevated temperature of operation and limitations in fabrication with suitable materials of construction. The unconventional flow patterns in the present process also eliminated some designs. A box-type mixer-settler was selected as the most appropriate. In the conventional box-type mixer-settler, two liquids flow continuously and countercurrently through the stages. In the present system, multiple fluids are employed. The proposed flow patterns for a box-type mixer~gsettler are shown in plan view in Figure 4. In the metal transport stages (a 'stage' consists of a mixing chamber and a settling chamber) the liquid metal flow is continuous through all stages, but the salt within each stage is captive and is recirculated within the stage. The salt in each of these stages is periodically withdrawn and moved one stage up- stream (with reference to the direction of metal flow). In the salt transport stages, the first stage is conventional in the manner in which salt and metal move in, through, and out. However, in the remaining stages the salt moves continuously through while the metal phases are captive and are recirculated within each stage, Periocdically the metal phases are removed and replaced with fresh metal. Because of the special fluid movement requirements, an extractor concept employing a pump in each stage was chosen rather than one utilizing the pumping effects of the mixers as other designers have done.(l”79 The action of liquid in the pumps is similar to that in a vertical-bowl centrifuge which is charged on its axis at the bottom and discharges at the top. The pump tube is partially submerged in the fluid, and fluid enters an inlet in the bottom located on the axis. Centrifugal force lifts fluid up the wall of the pump to discharge openings near the top. The head which is developed is roughly proportional to the square of the internal radius of the cylinder and to the square of the rotational speed. 518 M%-Cd (BATCH FEED) Mg~Cd-NM (BATCH REMOVAL) L TO EMPTY < < EXTRACTOR IF NECESSARY SALT _ {BATCH FEED) WASTE SALT (BATCH REMOVALYS ~ T Zn-Mg € {BATCH FEED) Zn-Mg-FPu O' 3 (BATCH REMOVAL)} Mg (BATCH REMOVAL} Figure 4. 519 Mg~Cu-Pu {BATCH FEED) —> METAL PHASE ——>> SALT PHASE @ AGITATOR PUMP BATCH TRANSFER PUMP Seven-Stage Mixer-Settler - Schematic Plan View The rotating tube need not be cylindrical; a conical shape would be desirable but is more difficult to fabricate. The pumps are suspended by heavy shafts which are supported and sealed above the heated zone. Mixing blades are mounted on the mixer-pumps to promote agitation. Similar pumps, but without mixing blades, are used for the periodic removal of fluids from a settling region. The discharge openings at the upper end of the tube consist of slots cut in the tube wall. The liquid collects in a chamber surrounding the top end of the tube and can be routed as required in the particular application. The flow out of the collecting chamber of an agitator-pump is directed into a settling chamber. The flow from a pump emploved to transfer fluid from the settling region of one stage to another stage is directed to the settling region of the latter stage. Equipment Design and Performance Design efforts have been concentrated on developing a universal mixer—-settler stage that is applicable to operation with either metal or salt phase captive. In selecting a design, plastic mixer- settler models were built and tested with water and organic fluids. Acetylene tetrabromide in which xylene was dissolved to alter the density and water were used as the liquid phases in the early tests. One agitator-pump configuration was also operated with molten Mg-Cu alloy in a stainless steel mixing stage. The pumping characteristics were the same for each of the fluid systems; the volumetric flow rate was independent of the density of the pumped liquid. The more recent flow tests were made using water or water-carbon tetrachloride. Design characteristics sought include high stage efficiency, simple control of flow rates, variable interstage flow patterns, and confinement of metal fumes. Figure 5 is a schematic section of the current stage design. The plastic model of this design is shown in Figure 6. This particular configuration was selected to provide a mixing chamber of about 1 liter volume and a mixer which would produce a high degree of agitation while reducing bypassing of solutions to an acceptably low level. The top baffle and sleeve prevent the entrainment of gas intc the fluids as long as the liquid level above the baffle is at least 3 centimeters, resulting in more stable mechanical operation and higher power input at any given mixer speed., In normal operation with a two phase system, both phases enter the zone above the mixing chamber and flow downward through the two zones of the mixing chamber. Here the phases are vigorously agitated and eventually enter the bottom of the pump through the inlet orifice. The phases then flow through the collecting chamber and into the settling chamber. The inverted cup surrounding 520 SPLASH CONTAINMENT Cupr | J . AGITATOR PUMP+—~’H — T ) SALT = SLEEVE TO AVOID GAS ENTRAINMENT METAL ———5 — — —— | ! | e L,E_;‘Lll.unl T SETTLING SECTION (THIS STAGE) SETTLING SECTION (PREVIOUS STAGE) ? = h ! I I|| In = = | h e " e & s —— A VERTICAL BAFFLE — HORIZONTAL BAFFLE-——4" FLOW ANNULUS—] PUMP INLET CRIFICE— Y Y ryrr»>xr vy vy rxyrvy I I X X X X AX Y IR XA X 2 ) - Figure 5, Agitator-Pump in Captive-Metal Stage 521 Figure 6. Agitator-Pump in Plastic Mixing Section 522 the pump outlet (not present in Figure 6) eliminates splashing. In the settling chamber the fluids separate, the light (salt) phase flows out through an overflow spout, and the heavy (metal) phase flows under a baffle and through an overflow spout. The flow is either back to the mixing chamber or to the mixing chamber of an adjacent stage. The settling and mixing chambers are separate vessels. This modular design was selected to increase the flexibility in mixer- settler arrangement, simplify fabrication, and ease maintenance. Fluid flows into modules through the spouts located above the liquid level as shown in Figure 5. The necessary lifting of the fluids is accomplished by the pumps. The liquid levels in a settling chamber are fixed by the vertical position of the outlet spouts (see Figure 5). The volume of each liquid phase in the settling chamber is also fixed, 1In all stages but stage 5 in Figure 4, a constant quantity of captive phase is present. Since the quantity in tramsit in the pump and collecting ring is small and relatively constant, the amount of captive phase in the mixer can be controlled by selecting an appropriate total inventory of that phase in the stage. The flow capacity of the pump and the net flow rate of the other phase establish the salt to metal ratio and total holdup in the mixing chamber, For stage 5, the metal and salt flow rates establish the salt-to-metal ratio and, with the flow~to-1lift characteristics of the pump at the operating speed, fix the mixing chamber holdup. The pumps are designed so that the variation in total liquid holdup in a mixing stage will keep the liquid surface above the top horizontal baffle and below the inlet openings. The salt-to- metal ratios are selected to obtain favorable extraction factors (equation 3). In arriving at the present design, many power inpuvt and pumping rate tests were conducted using water as a single phase. Mixing power was determined using a dynamometer which measures drive-motor torque. Pumping rates were measured for different lifts, inlet orifice sizes, and rotational speeds. Results of three pumping runs are presented in Figure 7. As shown in the figure, the pumping rate increases and then levels off as the mixer speed increases. This condition permits a fairly stable pumping rate to be maintained without precise control of mixer speed and yet provides for a reasomable range of pumping rates. During these tests the water level was maintained about 6 cm above the top baffle, Variation of the height of water resulted in a moderate corresponding change in pumping rate. Thus the pump equipped with a 0.95 cm orifice and operating at 650 rpm delivered 76, 81, and 89 ml/sec when the water level was respectively 4,8, 6.0 and 8 cm above the baffle., This change in pumping rate corresponding to changes in liquid head would tend to stabilize the flow in a 523 1 24" PUMPING RATE, cc/sec 120 1o 0.95cm ORIFICE \ A—-—-—"“'A"'_'—_4 100 — 90 — 80 fal 70 0.7Hem ORIFICE | 60 |- g———————-~0 a o ol D/ 30 0.48-¢cm ORIFICE o—""0"" Oo— —Q o |- 0 { ] I 1 | ] t I 400 450 500 550 600 650 700 750 80O 850 800 PUMP SPEED, rpm Figure 7. Effect of Orifice Size and Mixing Speed on Output of Agitator Pump (Net pumping lift 16 cm) nultistage system. The power input to the mixing chamber as related to mixer-pump speed is shown in Figure 8. This power input is nearly independent of pumping rate since the pumrping power (less than 0.5 watt) is small compared to the mixing power. Since the pumping rate and power input are both dependent upon agitator-pump speed, it would be desirable to have a mixing chamber and pump design in which power input increased more rapidly with mixing speed than did the pumping rate. Under such conditions the reduced residence time which results from increased throughput would be offset by increased mixing intensity. Figure 9 shows the relationship between power input and flow rate at various mixer speeds for the three different sizes of pump crifice presented in Figure 7, Over the range of probable operating speeds (i.e., 600-800 rpm) there is an increase in overall mixing intensity with increased pumping rate. Thus high stage efficiency is likely to be maintained as flow rates are increased. In experiments involving batch extraction of cerium from Cd-Zn-Mg alloy into MgCl,-NaCl-KCl at 600°C, about 95% equilibrium composition was reached in about 4 minutes and 100% in about 10 minutes.(ll Power input to the contactor was 1.3 watts/liter. Power input in the proposed mixing chamber design (see Figure 8) is 40 watts/liter when mixing water at 700 rpm. The power input under the same mixing speed and geometry will be nearly three times as high, or about 100 watts/liter, when mixing higher density molten salt and metal. This represents a power input which is almost 80 times as great as that used in the above batch test. Accordingly, it is estimated that 100% equilibrium conditions would be reached in a few seconds if the phase contacting were being done under batch mixing conditions. The multistage mixer-~settler will operate under continuous flow conditions with a wmean residence time of one minute or less. 1In a full backmix stage* the residence time of any infinitesimal unit of material varies from zero to infinity. When very high (n99%) stage efficlency 1s desired, the mean residence time is far less significant than is the fraction of material exposed for short times. In the salt-metal systems of interest, the rates of extraction are known to be rapid, but have not been precisely defined. Therefore, the mixing chamber was designed to minimize the quantity of material leaving the chamber with a short residence time. *A full backmix stage is one in which the composition of all the material in the stage i1s uniform at any instant so that the material leaving the stage has the same composition as the material in the stage. 525 92¢ RATIO OF MIXING POWER TO FLOW RATE, Watt/(cc/sec) 3 - 2 — 0 48-cm ORIFICE l e 0O 7t-cm ORIFICE 095cm ORIFICE 0 l | i i | 1 1 I I 400 450 500 550 600 650 700 750 800 850 900 AGITATOR - PUMP SPEED, rpm Figure 9. Power-to-Flow Ratio for Agitator Pump (Pumping 1ift = 16 cm) MIXING PCWER, watts \SLOPE =30 Figure 8. AGITATOR - PUMP SPEED, rpm Mixing Power for Operation of Agitator-Pump 527 In the present design of the mixing chawrber, a fixed horizontal baffle with an 8.9 cm dia opening is positioned at about mid-ele- vation of the chamber. A horizontal disk, 7.6 cm OD, is mounted on the mixer-pump at the same elevation. Thus the mixing chamber is divided into two zones with a 0.65 cm annvlar connection. A model has also been tested in which the annular space is further restricted by the use of a smaller opening in the fixed baffle. These configurations are intended to reduce top-to-bottom mixing and approach the performance of two backmix stages operating in series. A series of tests was conducted to determine the residence . time distribution in the mixing chamber. The tests were conducted in a non-flow system with the pump inlet orifice closed. The tests were designed to measure the approach to equilibrium composi- tion at the bottom of the mixing chamber after a small quantity of ‘ acid was injected just below the top horizontal baffle, A pair of platinum electrodes adjacent to the injection point signaled the start of injection and a pair of electrodes at the bottom of the mixing chamber (close to the normal pump orifice position) sensed the rise in solution conductivity at that point. A Midwest Model 1210 high speed recording oscillograph was used to record conduc- tivity changes which could then be converted to molarity changes in accordance with a predetermined calibration. Figure 10 is a reproduction of a typical oscillograph chart tracing for cne of these tests. This test was conducted at a mixer speed of 600 rpm. As shown in the figure, the injection time is indicated by a sharp rise in conductivity at the upper electrodes. Following a delay of about 0.2 sec there is a continuous rise to equilibrium concentration at the bottom electrodes. On the basis of this test and similar tests conducted at various mixing speeds, the rate of approach to uniform composition was determined as shown for four runs in Figure 11l. From these data it is possible to . calculate a residence time distribution curve for any net flow through the mixing chamber. If a reaction rate curve could be determined for a batch . extraction run in a mixing chamber, mass transfer coefficients could be calculated. These coefficients could then be handled in accordance with existing correlations to predict the performance of alternative designs. The extraction is much too fast to determine by conventional sampling techniques; the phases equili- brate in about 4 seconds. Nevertheless, the shape of the reaction curve can be assumed, for example: F=1- e-'Kt (5) where F is fractional approach to equilibrium t is time K is a constant which may be evaluated as K= 2,303/t 7 1s time to reach 907 completion 528 629 ELECTRICAL POTENTIAL ACROSS CELL, arbitrary units —— T+ 1531 B+ 31 11 . T ] o=y ] — N CELLAT TOPOF CHAMBER T + =1 | | 4= [ + tt+ - ++1 1 1 1 1 — - T : === bt + = =% CELL AT BOTTOM OF CHAMBER S—f———t——F =1 ESERs | ot 2”:% =3 4+ 4= = ETIME OF o - = —F 1= TIME AFTER INJECTION, sec[ | 12— [ 1 = P INJECTION 3 T e e = A T I = e = o B o o B == . g Figure 10. Conductivity Cell Readings in Mixing Chamber CONCENTRATION AT BOTTOM, PERCENT OF EQUILIBRIUM 90 - 00 ] 1000 0 80 400rp — 70 _ 60 . 50 - — 40 — = 30 +— - 20 — 10 - 0 | | I | | | ] ] ] " [ 2 3 4 5 6 7 8 a 10 TIME AFTER INJECTION, sec Figure 11. Approach to Uniform Composition in the Mixing Chamber (Pump inlet closed; nitric acid injected into water near top of chamber; see configuration in Fig. 6) 530 On this basis it is possible to predict the stage efficiencies which can be expected under a variety of mixing conditions (residence time distributions) and for a range of reaction times (see Table III). It is anticipated that experiments planned with molten salt and metal will provide sufficient data to permit the calculation of a reaction rate curve of a form such as indicated in equation 5. Evaluation of mechanical performance and determination of stage efficiency will be based upon tests to be conducted in a single stage unit using molten salt and metal at about 650°C. Components of the stage have been constructed of type 304 S5.S. The mixing chamber configuration is identical to the design used in the plastic model tests; the mixer-pump is the same unit used in the plastic model tests. Table III * Reaction Rate Versus Stage Efficiency Time to reach Efficiency for one minute mean holdup 90% of Equilibrium (percent) ~Batch Operation (seconds) Backmixed 2 Backmixed Reference Reference Stage zones in Design Design with Series Restricted Annulus 60 70 78 71 72 10 93 98.4 94,7 96.1 5 96.5 99.5 97.8 98.6 1 99.3 99.98 99.81 99.92 * =Kt Reaction rate curve is assumed to have the form, 1 - e Discussion When the single-stage metal-salt extractor has been optimized and demonstrated, a multistage mixer-settler will be constructed. The multistage unit will be used in the Plutonium Salt Transport Experiment which will involve the investigation of the entire Salt Transport Process on a scale of 1 kg of plutonium. It is expected that the multistage mixer—settler will be fabricated of niobium and tantalum, but the selection of materials is still underway.(lz) The stability of the materials with respect to plutonium is as important as their resistance to corrosion by the molten salts and metals. For example, 300 and 400 series stainless 531 steels appear to have adequate resistance to chloride salts and Mg-Cu alloys, but appear to react with dissolved plutonium. The seven-stage mixer-settler (including external drives and heaters) will be about 70 c¢m wide by 120 cm long by 150 cm high. This unit is as small as is practical because of clearances needed for agitator-pump drives and heaters. In the Plutonium Salt Transport Experiment, the extraction batch size will be about 40 kg of Mg-Cu-Pu-fission products alloy containing 1 kg of plutonium, This solutiom, which is intentionally dilute to provide a reascnable volume of feed solution, will be processed in about 4 hours. The extractor will have capacity for purifying about 30 kg of plutonium per day which is more than the expected average daily discharge of plutonium from fast reactors generating 5000 mW of electricity, The acceptor stage of an extractor will contain the largest quantity of plutonium, and the criticality limitation for this stage will determine the maximum batch size, The allowable quantity of plutonium is greatly increased by the selection of an acceptor alloy composition for which the solid plutonium phase in equilibrium with saturated solution is a Pu-Zn intermetallic compound rather than pure plutonium. It is expected that the acceptable batch size will be about 20 kilograms of plutonium. This limit might be increased if the acceptor alloy is replaced continuously or intermittently during a run. Acknowledgments The authors thank G. N. Vargo for his aid in building and operating equipment and Drs. W. J. Walsh, T. R, Johnson and R. K. Steunenberg for their interest and suggestions. 532 10. 11. 12, References Steunenberg, R. K., R. D. Pierce and I. Johnson, ''Status of the Salt Transport Process for Fast Breeder Reactor Fuels", This Symposium. Knighton, J. B., I. Johnson and R. K. Steunenberg, "Uranium and Plutonium Purification by the Salt Transport Method", This Symposium. Johnson, T. R., R. D. Pierce, F. G. Teats and E. F. Johnston, "Behavior of Countercurrent Liquid-Liquid Columns with a Liquid Metal', Dec. 1968, Preprint 24F, Sixty-first Annual Meeting of the AIChE, Los Angeles, California, Accepted for publication in AIChE Journal. Morello, V. S. and N. Poffenberger, "Commercial Extraction Equipment'", Industrial and Engineering Chemistry, 1950, No. 6, Vol. 42, pp. 1021-1035. Akell, R. B., "Extraction Equipment Available in the U. S.", Chemical Engineering Progress, 1966, No. 9, Vol. 62, pp. 50-55. Benedict, M., T. H. Pigford; Nuclear Chemical Engineering Coplan, B. V., J. K. Davidson and E. L. Zebroski, 'The Pump~-Mix Mixer-Settler a New Liquid-Liquid Extractor', Chemical Engineering Progress, 1954, No. 8, Vol. 50, pp. 403-408. Graef, E. R. and S. P. Foster, "Design of Box-Type Countercurrent Mixer-Settler Units - Factors Affecting Capacity', Chemical Engineering Progress, 1956, No. 7, Vol, 52, pp. 293-298. Holmes, J. H., and A. C. Schafer, '"Some Operating Characteristics of the Pump-Mix Mixer~Settler", Chemical Engineering Progress, 1956, No. 5, Vol. 52, pp. 201-204. Armstrong, D. R. and R. D. Pierce, '"Power Requirements for Mixing Liquid Metals', ANL~-6596, Argonne National Laboratory, Chemical Engineering Division Summary Report, Argonne, Illinois, pp. 84-87, 1962. Walsh, W. J. and T. R. Johnson, Argonne National Laboratory, Argonne, Illinois, Personal Communication., Kyle, M. L., R. D. Pierce and V. M. Kolba, "Containment Materials for Pyrochemical Processes', This Symposium. 533 * CONTRIBUTION TO THE PLUTONIUM-MAGNESIUM PHASE DIAGRAM W, Knoch**, J. B. Knighton and R. K. Steunenberg Chemical Engineering Division Argonne National Laboratory Argonne, Illinois 60439 U. S. A. Abstract The region of liquid immiscibility in the binary Pu-Mg system was determined by chemical analyses of the equilibrium liquid phases. The magnesium content of the plutonium-rich phase ranged from 0.49 at. % at 650°C to 2.22 at. % at 900°C, while the plutonium content of the magnesium-rich phase varied from 10.5 at. % at 523°C to 11.5 at. % at 900°C. The consolute temperature is estimated to be 1040 + 10°C. Thermal analysis data are given for several solidus transition temperatures. The thermodynamic properties of liquid Pu-Mg alloys were estimated from the compo- sitions of the equilibrium liquid phases assuming the alpha function {1ln yi/(l-Ni)z] to be a linear function of Nj. The estimated data are in agreement with previously reported thermo- dynamic data for dilute solutions of plutonium in liquid magnesium solutions. * Work performed under the auspices of the United States Atomic Energy Commission. *k Present address: Gesellschaft zur Wiederaufarbeitung von Kernbrennstoffen, m.b.H., 7501 Leopoldshafen, West Germany. 535 Introduction In the course of pyrochemical process development studies data were obtained on tge figlubility of plutonium in liquid ternary Zn-Mg—-Pu alloys.(l » These solubility data indicated that the liquid phase immiscibility region of the binary Pu-Mg system extended well into the ternary system, However, discrepancies were found between extrapolated data for the compositions of the two immiscible liquid phases in the binary Pu-Mg system from plutonium solubility measurements in the ternary system and the estimated immiscibility gurve for the binary system reported by Schonfeld and Ellinger.(3 The results of an experimental determination of the miscibility curve by chemical analyses of the liquid phases in the Pu-Mg system are reported below. The results of the determination, by thermal analysis, of the temperatures for several solidus transformations are also reported. Finally, the miscibility gap composition-temperature data were used to estimate the thermodynamic properties of liquid Pu-Mg alloys. Experimental The purities of the magnesium and plutonium used in the investigation were greater than 99.95 and 99.9%, respectively. The Pu-Mg alloys were contained in a tantalum crucible (5 cm I.D.,) with a tapered bottom inside a graphite liner. The crucible assembly was located inside a resistance-heated, stainless steel furnace tube which was equipped with a sampling port, a tantalum thermocouple well, a tantalum paddle stirrer, and gas inlet and cutlet tubes. The furnace tube contained a purified helium atmosphere (one atmosphere pressure) and was located in a glovebox filled with purified nitrogen. The temperature was measured to +0.5°C with calibrated chromel-alumel thermocouples. An initial charge of 64 g of plutonium and 16 g of magnesium was added to the tantalum crucible. The furnace tube was then repeatedly evacuated and flushed with helium. After the furnace had been heated to the desired temperature, the system was allowed to equilibrate for two hours at constant temperature and a stirring rate of 200 rpm. At temperatures above 875°C the helium pressure in the furnace tube was increased to 2 atm to diminish the vaporization of magnesium. During the course of the experiment, more plutonium was added to produce an approximate composition of Pu-50 at. % Mg. Finally, magnesium was added to determine the liquidus line in the magnesium~rich region of the system, The two liquid phases were sampled after the mixture had been equilibrated and allowed to settle for 15 min. The sampling device consisted of a tantalum tube 6 mm I.D. and 7 cm long, which 536 was closed at one end by a porous tantalum frit and at the other end by a threaded plug which could be attached to a stainless steel rod. The sample tube attached to the stainless steel rod was inserted into the furnace through a valve and gas seal on the sampling port. After a 5-min period the sample tube was lowered into one of the liquid metal phases and a sample was forced into the tube by pressurizing the system for 30 sec with 1 atm of additional helium pressure. The sample was then withdrawn and allowed to solidify quickly in the cool sampling port. The excess helium pressure was then released, and the sample tube, after it had cooled, was withdrawn into the nitrogen atmosphere. The frit and any material adhering to the outside of the sample tube were removed mechanically to avoid possible cross-contamination of the sample by the other liquid metal phase, The plutonium was determined with a precision of #27% by hexone extraction and alpha counting. The magnesium was determined with +1% precision by EDTA titration after the plutonium had been removed by double precipitation. Several cooling curves were determined by repeated temperature cycling of the system. The accuracy of transition temperatures obtained by this technique is estimated to be +3°C, Results and Discussion The results are given on a weight and atom percent basis in Table I and are plotted in Figure 1. The plutonium-rich phase contained very little magnesium (0.49 at. % at 650°C and 2.22 at. 7% at 900°C). The composition of of the magnesium-rich phase was almost independent of temperature between the monotectic point at 523°C and 900°C, where the plutonium concentrations were 10.5 and 11.5 at. %, respectively. The miscibility gap closes rapidly above 900°C, with an estimated consolute temperature of 1040 + 10°C. Because of the very low magnesium concentration in the plutonium-rich phase, no attempt was made to determine the monotectic point in this region. Such a determination would re- quire accurate analyses of magnesium at concentrations of less than 0. 01 wt %. The thermal halts observed in several cooling curves indicate transition temperatures of 523 + 3°C for € + 6 « e + L, and 638 + 3°C for ¢ + L, e Ly + L,. These transition temperatures differ somewhat from the tentative values given by Schonfeld and Ellinger.(l The boundaries of the 6 + L, field are only moderately well-defined. The solid solubility of plutonium in 8-magnesium is approximately 2 at. % at 550°C, The solid solubility of magnesium in e-plutonium has not been determined, but the indications are that any solid solution in this region would con- tain less than 0.5 at. % magnesium. 537 HOO 1000 TEMPERATURE, °C 2 S 10 20 30 50 80 T Il 1 H T | T i WEIGHT PERCENT MAGNESIUM | Qo L| + L2 . 638° € + L ° 2 NG Y - e+ 8 ] T T T T T T T —— i e o) 10 20 30 40 50 60 70 80 90 100 ATOM PERCENT MAGNESIUM Figure 1. Miscibility Gap in the Plutonium-Magnesium System 538 Equilibrium Compositions of Liquid Table I Plutonium-Magnesium Solutions Plutonium Rich k.3 Magnesium Rich Temp Pu Mg (°c) wt%Z at.2 wtZ at.Z wt at.% Region Lliifi 649.8 ~ - - - 48.0 8.58 649.6 99.95 99.51 0.05 0.49 - - 701.0 - - - - 51.6 9.78 700.0 99.93 99.32 0.07 0.68 - - 750.5 - - - - 52.1 9.96 749.3 99.88 99.32 0.12 0.68 - - 800.1 - - - - 51.4 9.71 800.0 99.78 97,88 0.22 2,12 - - 851.0 - - - - 53.3 10.4 891.3 - - - - 54.5 10.9 900.7 - - - - 54.2 10.8 899.0 99.77 97.78 0.23 2,22 - - 949.9 - - - - 55.4 11.2 950.9 99.75 97.60 0.25 2.40 - - 975.1 - - - - 63.6 15.1 975.4 99.2 92,65 0.8 7.36 - - 1025.0 96.9 76.1 3.1 23.9 82.2 32.0 Region etLo 600.0 - - - - 21.5 9.75 Region 0+L 609.9 - - - - 21.4 2.69 616.1 - - - - 19.4 2.39 630.2 - - - - 15.2 1.79 632.0 - - - - 11.5 1.30 * Values obtained by difference. The for magnesium and the magnesium-rich phase for plutonium. 539 52.0 91.42 48.4 48.5 78.6 80.6 84.8 88.5 90. 97. 97. 98. 98. .22 70 plutonium-rich phase was analyzed Thermodynamic Properties of Liquid Plutonium-Magnesium Alloys The thermodynamic properties of liquid Pu-Mg alloys were estimated from the compositions of the equilibrium liquid phases using the method repogrted by Wriedt. 4) 1n this method the "sub-regular" model, ¢®) in which the alpha function is assumed to be a linear function of composition is used, i.e., if In v 0 = @ (1-N) where y; is the activity coefficient of the ith component at an atom fraction of Ny, then @, = A+ B (1 - Ni) (2) where A and B are temperature dependent empirical coefficients. From the necessary condition that the activities of each component be equal in the two equilibrium liquid phases, two independent relations involving A, B, and the compositions of the two liquid phases may be derived which may be used to compute values of A and B. The details are given by Wriedt. (%) This simple two-parameter equation was used in prefer?nge to the four-parameter Lumsden equation(6) which Sundquist 7 reported to give a better represen- tation of the thermodynamic data for systems with miseibility gaps in view of the fact that the radius ratio parameter used in the Lumsden equation is approximately unity (Yp, + Mg = .993) for the Pu-Mg system. For computational purposes, values of the composition for the liquids in equilibrium were read from a smooth curve drawn through the experimental data points at temperatures from 650°C to the consolute point. These values of the composition and the computed values for A and B are given in Table II1. The activity coefficients for plutonium and magnesium computed from these values of A and B are given in Table III. The values marked with an asterisk in the table are for a hypothetical supercooled liquid which cannot be obtained in the laboratory. It is of interest to note that the values of the activity coefficient for plutonium in the infinitely dilute solution (Ny, = 1) increase with temperature up to about 950°C and thén dectease. This behavior is in agreement with the positive temperature dependence of the activity coefficient for plutonium in dilute magnesium solutions computed from molten salt-liquid alloy distribution data.(8) The values of the activity coefficient reported in the latter study are 11.5 and 12.2 at 700 and 800°C, respectively, which are in good agreement with the results of the present computation. Computed values for the free energy of mixing for 700 to 1000°C as functions of composition are shown in Figure 2. Straight lines 540 20001 AGXS 975 1500 040 cal /mole 10001 500 ZSGwmx L 3 s Figure 2. Computed Values of 0G4, and 1G*® for the Plutonium-— X Magnesium System 541 Table II1 Values of the Parameters A and B as Function of Temperature 4 4°] 1 2 % Mg Yoo N A B 0.995 0.005 0.0975 0.9025 8.2899 -6.0260 0.9925 0.0075 0.0980C 0.9020 7.4625 -5.1179 0.9895 0.0105 0.0990 0.9010 6.7977 -4.3961 0.9864 0.0136 0.1007 0.8993 6.3046 -3.8755 0.9827 0.0173 0.1026 0.8974 5.8618 "~3.4034 0.9788 0.0222 0.1055 0.8945 5.3928 -2.9169 0.969 0.031 0.1125 0.8875 4.8180 -2.2950 0.9272 0.0728 0.150 0.850 3.6025 -1.1968 0.870 0.130 0.215 0.785 3.0918 -0.9603 0.759 0.241 0.350 0.650 2.6641 -0.7551 0,55 0. 45 0,55 0.45 2.390 -0.551 T able III Calculated Pu-Mg System Activity Coefficients YMg Nyg 700 800 900 950 975 1000 1025 1040 °C 0 135 78.0 51.1 39.2 20.2 13.6 9.8 8.73 * * * * 0.1 35.0 25.2 19.10 16.2° 10.3° 7.67 6.00 5.58 * * * * * 0.2 11.97 9.99° 8.54 7.81 5.87° 4.70° 3.92 3.78 * 0.3 5.20° 4.82% 4.48° 4.31% 3.66" 3.22% 2.74 2.m1 * * * * * * * 0.4 2.79 2.76° 2.710 2.69° 2.48° 2.23° 2.04° 2.05 * * * * * * * 0.5 1.80 1.84° 1.86° 1.88° 1.83" 1.70° 1.61° 1.63 * * * * * %* * 0.6 1.34 1.39° 1.42% .44 1.44% 1.397 1.34% 1.36 * 0.7 1.13 1.16° 1.19° 1.20% 1.22° 1.19% 1.17 1.18 * * * * * 0.8 1.033" 1.05° 1.066 1.08" 1.09° 1.08 1.07 1.08 * 0.9 1.0027° 1.009 1.013 1.02 1.02 1.02 1.02 1.018 yYPu Nyg 700 800 900 950 975 1000 1025 1040 °C 0.1 1.072 1.061 1.052 1.047 1.035 1.030 1.026 1.024 * * * 0.2 1.294" 1.248° 1.212° 1.1900" 1.144* 1.123 1.106 1.096 * * * [ 3 +* * 0.3 1.705° 1.588" 1.496 1.450° 1.339 1.287 1.245 1.225 * * * %* * * 0.4 2.379% 2.214% 1.966° 1.866 1.648" 1.542" 1.459" 1.425 * * * * * * * 0.5 3.408 2.981° 2.674 2.503 2.119" 1.921 1.771 1.719 * * * * 0.6 4.860° 4.194° 3.711° 3.451" 2.824% 2.473" 2.217" 2.147 * * * * * * 0.7 6.693 5.822" 5.165 4.823 3.875 3.272" 2.848 2.768 * * * * * * 0.8 8.630" 7.795 7.084° 6.740° 5.435" 4.423" 3.738 3.675 F3 * * * * 0.9 9.62 9.831° 9.409" 9.291" 7.733" 6.312 4.990 5.008 * 1 10.41° 11.41 11.89 12.46 11.08 8.424 6.747 6.990 2. NPu log YMg = _-—2,3026 [A+B (0.5 + Nflg)} _ Mg . log Ypy = 33026 [¢A + B Nfig)] N = Atom fraction * = immiscible region 543 10000} AH mix }1o { | Y ~ . 1 © 7 @ 7/ stlnfi\\; @ 8 / AN S E 4 / \ 4 E 3 / \ D O ) // \\ 4 :3 “ / \ x \ G / \ 2 <18000: / 1+ 5 . / \ X / \ q I \\ - / \ / \ / \ { \\ / \ Z(XI)" I XS \ T 2 ! AG,, \ / 10001 4 11 ! / + 35 + — {g 100 fi%Mg———- & 8 300 AGmix Figure 3. Computed Values of AH v BSpixs AG*S and AGmix for the Plutonium-Magnesium System at 1273° 544 Eix tangent to the AG 4, curves in the regions of the two minima define the miscibility gap for each temperature., The rapid closing of the miscibility gap between 950 and 1000°C is related to a rapid decrease in the magnitude of the excess free energy between 950 and 1000°C, which is found to be related to unusually large values of the excess enthalpy and entropy of mixing as seen in Figure 3 for a temperature of 1273°K (1000°C). However, it should be noted that these estimates of the thermodynamic functions are based on the experimental data for the miscibility gap; the large change in their values with temperature above 900°C is a result of the rapid closing of the miscibility gap. Unfortunately, this region of the phase diagram is not well established experimentally and, therefore, the indicated temperature dependence of the thermodynamic data should be accepted with reservations. Acknowledgments The authors are indebted to Mr, J. D. Schilb, who assembled the apparatus and performed preliminary studies of the plutonium- magnesium system and to Mr. E. T. Kucera who performed the chemical analyses. The authors are also grateful to Dr. Irving Johnson for advice concerning the thermodynamics of this system. 545 4, 8. References Knighton, J. B., I. Johnson and R. K, Steunenberg, "Uranium and Plutonium Purification by the Salt Transport Method", This Symposium, Knoch, W., "Solubility of Plutonijum in Liquid Cu-Mg Alloys', ANL-7225 (1966) p. 30-31. Schonfeld, F. W. et al, "Plutonium Constitutional Diagrams", Progress in Nuclear Energy, Metallurgy and Fuels, H. M. Finniston and J, P. Howe, Eds.,, Pergamon Press, New York, 1959, Series V, Vol. 2, p. 587. Also note R. P. Elliot, "Constitution of Binary Alloys, First Supplement', Mc@Graw-Hill Book Co., New York, 1965, p. 596-597. Wriedt, H. A., "Calculation of Activities in Binary Systems Having Miscibility Gaps', Trans. Met. Soc. AIME, 1961, VOl. 221’ p- 377_3835 Hardy, H. K., "A Sub-Regular Solution Model and Its Application to Some Binary Alloy Systems', Acta Met., 1953, Vol. 1, p. 202. Lumsden, J., Thermodynamics of Alloys, The Institute of Metals, London, 1952, p. 335. Sundquist, B. E., '"The Calculation of Thermodynamic Properties of Miscibility-Gap Systems', Trans. Met. Soc. AIME, 1966, Vol. 236, p. 1111-1122, Johnson, I., J. B. Knighton and R. K. Steunenberg, ''Thermodynamics of Dilute Solutions of Plutonium in Liquid Magnesium', Trans. Met. Soc. AIME, 1966, Vol. 236, p. 1242-1246, 546 * THE SOLUBILITY OF URANIUM AND PLUTONIUM IN LIQUID ALLOYS Irving Johnson Chemical Engineering Division Argonne National Laboratory Argonne, Illinois 60439 U.S. A . Abstract A method is presented for the estimation of the solubility of uranium or plutonium in liquid ternary alloys. This method requires information on the phase relations and thermodynamics of the three binary systems. The method also may be used to determine the approximate stolchiometry of the solid phase in equilibrium with the liquid alloy. The method is illustrated by analyses of the experimental data for the solubility of uranium in liquid zinc-magnesium-uranium and plutonium in liquid zinc- magnesium-plutonium alloys. * Work performed under the auspices of the United States Atomic Energy Commission. 547 Introduction Pyrochemical methods for the recovery and purification of uranium and plutonium have process steps whose effectiveness is determined by the wvalues of the solubility of these actinides metals in liquid alloys. For example, the pyrochemical method (1) for the separation of plutonium from uranium depends on large differences in their solubility in certain liquid magnesium-rich alloys. Also, the effectiveness of the salt transport step for the recovery and purification of uranium and plutonium 2 depends in part on their solubilities in the donor and acceptor alloys. The importance of the solubilities of uranium and plutonium in liquid alloys to the development of pyrochemical processes led to the thermodynamic analysis which is the subject of this paper. Solubility The solubility of uranium or plutonium in a liquid alloy is defined as the uranium or plutonium content of the liquid alloy, expressed in either weight percent (wt pct), atomic percent (at pct), or atom fraction, (at fct), when in equilibrium with a solid (or another liquid) uranium~ or plutonium-rich phase. This definition of solubility, which differs from that commonly used by the chemist, was adopted as a practical expedient during the early pyrochemical studies when it was not feasible or necessary to completely characterize the equilibrium solid (or liquid) phase. We also believe that it is incorrect to state solubilities in liquid metallic solutions in terms of the formula for the solid phase in equilibrium with the solution, since the formula of any molecular entities (species) which may exist (if any do) in the metallic solutions would not be expected to be related in any simple way to the formula of the solid phase. However, it is not to be assumed that information concerning the composition and structure of the equilibrium solid (or liquid) phase is not needed for the complete specification of solubility equilibria, but only that this kind of information is not needed to derive a numerical value for the solubility which is extremely useful for practical process development. Indeed, the thermodynamic analysis of solubility equilibria, which will be presented, requires knowledge of the composition and thermodynamic properties of the solid (or liquid) phase in equilibrium with the liquid metallic solutionm. Binary Systems Binary systems of uranium or plutonium with low-melting solvent metals, such as zinc, cadmium, magnesium, will be considered. For such systems we may represent solubility equilibria by the equation: MXm(s) 7 M(soln) + mX(soln) (1) 548 where MX is the formula for the solid phase in equilibrium with the binary liquid alloy of M(U or Pu) and X (Zn, Cd, Mg). The atom fraction of M in the saturated liquid phase, fifi, is given by the equation: AGf° s _ m _ s _ s In xy = —RT In Yy -0 1n ay (2) in which AGf° is the standard free energy of formation of MX (ger g—-atom o%mM), Y, 1s the activity coefficient of M, and n a¥ the activity of X 'in the saturated solution. The reference ates for M and X used for AGfMX » Y, and a_, must be the same. Two special cases will be con51dered gn detail: (1) when w=0, i.e., pure solid M is the equilibrium solid phase and, (2) when m is equal to a positive constant over an extended temperature range ("line" compound). Equilibrium phase: Pure solid U or Pu. For this case, when pure solid uranium or plutonium is the equilibrium phase, Eq. (2) becomes s s xy = vy (3) If the reference state for vy, is pure super-cooled liquid M, then Eq. (3) may be written: In 3y RT s (s) (4) where AG_ is the free energy of fusion and the partial excess free energy of M in the liquid solution, both at temperature T. If the free energies are written in terms of the corresponding enthalpies and entropies, then Eg( §4) becomes : xs (s) (AH + HM (AS + S ) s In Xy = RT . (5) Since, generally, the quantity (AH + H§s(s)) is positive, the solubility is found to increase wi ? }ncrease in temperature. However, in the Cd-U system Hfi is negative and has a larger absoclute value than AH_, thereby leading to a retrograde sclubility of uranium in liquid cadmium over part of the temperature range. (This is the only example of a retrograde solubility which has been reported for a liquid metallic solution.) If the enthalpy and entropy terms are independent of temperature and composition, then a plot of the logarithm of the solubility vs the reciprocal of the absolute temperature may be fit by a straight line, If the functional dependence of y,, on at fct and temperature is known, then the solubility may be computed by solving Eq.S(B) for ; alternatively, if the dependence of the solubility, , on temperature is known, then the dependence of y,, on at fct and temperature may be determined provided the form of the fupctional relation is known. For example, if the Darken equatiomn is applicable to the binary system, then 549 s W= ge e all - =R’ (6) where Y is the activity coeff1c1ent of M in the infinitely dilute solution and o is a constant; both Y and a are temperature dependent. The reference state for vy, in Eq. (6) is pure solid M. We were able to estimate the solubility of uranium in liquid cadmium over the temperature range where pure solid uranium is the equilibrium phase by the use of this technique;(3) computed values of the solubility and its retrograde temperature dependence were in excellent agreement with observed values. Equilibrium phase: Intermetallic Compounds of U or Pu. For the case when an intermetallic compound of uranium or plutonium is the equilibrium phase, Eq. (2) may be modifigd in several ways. If the solubility is extremely small, then a_, would be expected to be nearly unity and Eq. (2 educes to the f AHffiXm £ % 7 ASfyux ET In ’%Si = RT - R (N where AHfMXm and A?fMXm_ar? }he enthalpy and entropy of formation of MX,,,» whereas HMM are the partial molar excess enthalpy and entropy of M in the saturated liquid solution. The forms of Eq. (7) and Eq. (5) are similar; a plot of the logarithm of the solubility vs the reciprocal of the absolute temperature is usually fit by a straight line, although curvature has been found in some cases. When curvature is observed it is not advisable to attempt to interpret empirical constants determined from the plot in terms of variations of the enthalpies or entropiles with temperature or composition, in view of the approximations made in deriving Eq. (7). An exact form of Eq. (2) is obtained by the substitutions AGffix _ a;s(s) _ mais(s) - (1 - s)m _ m M M RT (8) hence o =xs(8) xs (s) AHE - s s.m HM HX In xy (1 -x)" = RT - ASE® st(s) §xs(s) m R (9) If x> is of the order of 0.0l or less ({.e., 1 at pet) the detection of differences between Eq. (9) and Eq. (7) will depend upon the accuracy of the solubility data. It is also to be noted 550 that since the excess quantities are, in general, composition dependent, caution should be used in the interpretation of the slope and intercept of linear plots of logarithm of solubility vs 1/T in terms of differences in enthalpies and entropies. Nevertheless, plots of the logarithm of the solubility vs 1/T are extremely useful for the detection of errors in experimental data and for interpo- lation. Abrupt changes in the slopes of such plots may usually be interpreted as evidence for changes in the composition of the equilibrium solid phase. Thus, transformation temperatures, such as peritectic temperatures, may be estimated from the temperature at which the line segments intersect. This technique is partic- ularly useful in those systems in which the transformations may be sluggish and therefore difficult to determine accurately by the use of thermal analysis or alloy-annealing methods. If the free energy of formation of the intermetallic phase and the dependence of the partial excess free energies on at fct are known as functions of - temperature, the solubility mey be computed. An example 1s given in Table I where results are given for the computation of the solubility of plutonium in liquid Cd-Pu solutions; these results were comp?t?d from thermodynamic data derived from galvanic cell studies 6) of the two intermetallics, PuCd,, and Pqu6. In this system the partial excess free energy of piutonium was found to be fit by an equation of the form: Gad = (1= %, 0% (A + Bxy) (10) where A and B are linear functions of the absolute temperature. The partial excess free energy of cadmium was derived from Eq. (10) using the Gibbs~Duhem equation. The free energy of formation of both PuCd,, and PuCd, are also linear functions of temperature. These relations were substituted into Eq. (8) and the resulting transcendental equation solved for xp,, at the temperatures shown in Table I. Good agreement between observed and calculated values of the solubility were obtained. These results are presented only as an example of the application of Eq. (8) to the computa- tion of the solubility of an intermetallic compound; the good agreement between computed and observed values of the solubility was expected since the galvanic cells were operated with one of the electrodes a saturated liquid alloy. However, the concen- tration and temperature dependence of the excess free energy was determined from measurements made on cells with unsaturated liquid alloy electrodes. Ternary Systems Ternary systems of uranium or plutonium with mixtures of two low-melting solvent metals such as magnesium-zinc or magnesium- cadmium will be considered. Only the solvent-rich region of the ternary systems will be treated, although the methods may, in 551 Table I Comparison of Observed and Calculated Solubility of Plutonium in Liquid Cadmium Temperature Solubility (at fct) Solid Phase (°C) Obs? Calc 335 0.00156 0.00152 Pqu11 351 0.00222 0.00223 Pqu11 388 0.00549 0.00500 Pqull 408 0.00773 0.00680 Pqu6 443 0.0109 0.0101 Pqu6 504 0.0187 0.0181 Pqu6 554 0.0284 0.0275 PuCd, 603 0.0414 0.0402 PuCd, 632 0.0544 0.0501 Pqu6 aReference 6. qu. (8). principle, be extended to the whole ternary system. In a ternary system, solubility equilibria may be represented by the following general equation: MXmYn(s) 7 M{soln) + mX(soln) + nY¥(soln) (11) where MXmX is the formula for the solid phase in equilibrium with the ternary liquid alloy of M (Pu or U), X and Y (Zn, Mg, Cd). The atom fraction of M in the saturated liquid phase, Xy is given by the equation.AGfo MX Y m n s ____ mn _ s _ s _ s 1n Xy = RT In Yy ~ In ay - m 1n ay (12) in Whlch AGEy is the standard free energy of formation of (per gMX%om of M , YM the activity coefficient of M in the liqu?d solution, and ay and ay the respective activities of the solvent metals X and Y in the liquid alloy. The a priori calcu- lation of values of the solubility using Eq. (12) for the general case, without the introduction of some simplifying assumptions, requires extensive thermodynamic data for the csystem and is not of great practical utility. Fortunately, in the case of the solubility of plutonium and uranium in mixtures of certain low-melting metals, a number of methods for the estimation of the required thermodynamic 552 data have been found. To illustrate these methods we shall discuss in detail the solubility of uranium and plutonium in liquid zinc-magnesium solvent mixtures. Uranium Solubility in Liquid Zn-Mg Alloys. Experimental values of the solubility of uranium in liquid zinc-magnesium alloys are shown in Fig. 1 as a function of the at fct of magnesium in the liquid for temperatures of 600, 700 and 800°C. The experimental data are from the work of A. E,. Martin(7), J. B. Knighton(a) and coworkers, It 1s seen that the solubility of uranium decreases initially, passes through a minimum value, increases to a sharp maximum and then decreases smoothly with increase in the at fct of magnesium in the liquid solution. The sharp maximum in the solubility is seen to move toward smaller wvalues of the at fct of magnesium as the temperature 1s increased. The region to the left of the sharp break corresponds to a part of the ternary system in which the equilibrium uranium-rich solid phase 1s a uranium-zinc intermetallic compound, whereas in the region to the right of the break the equilibrium uranium-rich solid phase is pure metallic uranium. The sharp break corresponds to the peritectic reaction Uzn_(s) < U{s) + nZn (sat'd soln). (13) Since the liquid phase is in equilibrium with pure metallic uranium, the activity of uranium is equal to unity both at the maximum and along the solubility line to the right., As the temperature is increased, the stability of the uranium-zinc intermetallic compound is decreased and, therefore, the equilibrium represented by Eq. (13) occurs at greater zinc concentrations, i.e., lower at fct of magnesium, The equilibrium corresponding to the solubility line to the left of the break point is UZnn(s) pa U(soln, xs) + nZn(soln, xZn) (14) s where = 1 and (x! = at fct of Mg at the break :gint) Tthgranium songility, xUTMES given by: 1 “C¥bzny 1n y$ - o In a° 15) n X, T - lny;-nlna, ( where AGffi is the standard free energy of formation of UZn, and, Y$ and a7 ~aPe the activity coefficient of uranium and the activity o¥ zinc, respectively, in the equilibrium liquid zinc-magnesium- uranium solution. The solubility line to the right of the break point corresponds to the equilibrium U(s) b U(soln, x;)] (16) hence s s In Xy = =1ln YU 7 since afi = 553 URANIUM SOLUBILITY, ATOM FRACTION o ™o S () O L] 7 i L ° r *8 (] |O-2 — [ \ » A » A‘\ L] 0 A ] o | i'- . - L A L) y 9, 103 | N - A 4 2 adp A n o L) A P\ . n] 1077 - O 800 Ao a T00*C & e00°C » Q J A 107 ) PO R S 1 1 1 1 ] 0 [+N] 0.2 O3 0.4 0% 06 (o 2rg 0.8 09 [£4] MAGNESIUM, ATOM FRACTION L. Solubility of Uranium in Liquid Zn-Mg-U Alloys at 600, 700 and 800°C. Data from Ref. 8. 554 We have found that the solubility of vuranium in liquid zinc- magnesium-uranium solutions may be estimated by assuming that (1) the activity coefficient of uranium in the solution is the geometric mean of the values for zinc-uranium and magnesium-uranium solutions and that (2) the activity of zinc in the sclution is equal to the value for a binary zinc-magnesium solution of the same zinc concentration. The first assumption may be written s— 1n Yy = xann YU,Zn + ng In YU,Mg (18) where vy 7n @0d Yy,Mg are the activity coefficients of uranium in binary zinc-uranium solutions, respectively. The solubility of uranium in liquid zinc-magnesium solutions was estimated using Eq. (15), (17) and (18). The standard free-energy of formation of the uranium-zinc intermetallic compound, AGffi n.? was compu%sg from the results of high temperature galvanic ¢Bll studies. The results from these same galvanic cell studies were used to compute the activity coefficient of uranium, yg Zn» 1o liquid binary zinc-uranlum sclutions. The activity coefficient of uranium in liquid magnesium—-uranium sclutions, Yy .Me> Was computed from the data for the solubility of uranium in li&u%d magnesium. 10) The zinc activities were computed from the equation reported by Chiotti and Stevens(11) for the ZIn-Mg system. Unfortunately, the formula for the intermetallic phase has not been unambiguously established. Studies of the zinc-uranium system at Argonne National Laboratory(lz) have established the existence of two or more intermetallic compounds whose formulae probably range from UZny, to Uing 5. These compounds appear to differ only slightly in stability and, therefore, it has not been possible to determine exactly the formula of the equilibrium solid phase (in the binary zinc-uranium sgstem) as a function of temperature. The galvanic cell studies ) of the temperature dependence of the free energy of formation of the intermetallic compound in equilibrium with the liquid phase, which ranged from just above the melting point of zinc (420°C) to about 700°C, did not indicate significant changes. Sharp changes in the temperature dependence would be expected if there had been different equilibrium intermetallic phases present which differed significantly in stability. To compute the solubility we have tried values of n between 8,5 and 12; the best agreement between observed and calculated values of the solubility was obtained with wvalues of n between 8.5 and 9.5. It should be noted that it is possible, and highly probable, that the formula of the equilibrium solid intermetallic phase will depend on the zinc activity in the liquid solution. We would expect this variation of the formula to take place in discrete steps, with the value of n decreasing as the activity of zinc in the solution is decreased. Consequently, the value of n which is found to correspond to the best fit to the solubility data may be an average of the values for the several intermetallics involved. 555 The results of computations of the solubility of uranium at 600°C in liquid zinc-magnesium-uranium solutions are shown in Fig. 2. It is seen that the best agreement between computed and observed values of the solubility is obtained for a value of n between 8.5 and 8.75. The greatest differences between computed and observed values occur in the regions of the minimum and the sharp maximum in the solubility. The difference in the region of the minimum is probably mostly caused by the use of a value of n too small for the region of high zinc activity. To correct for this difference would require the computation of the initial part of the solubility using one value of n and the latter part using a second value of n. The obvious scatter of the experimental data (probably caused in part by a lack of equilibrium between the solid phase and the liquid phase when the latter was sampled for analysis in the solubility measurements) makes a meaningful computation of this type impossible. The over- estimate of the solubility in the region of the sharp maximum is believed due to the failure of Eq. (18) to accurately predict the activity coefficient of uranium in the liquid solution. It is, in fact, surprising that this simple equation works as well as it does. A somewhat better estimate of the solubilit{3§an be made by using the approximation %}z§t suggested by Darken ( and later by Alcock and Richardson for the activityxgoefficient: AG 5 = _ _Zn-Mg Inoy, = %5010 Yy,zn T %™ Yu,ug T TET (19) where AG;E_ is the excess free energy of mixing of the binary zinc—magnes%fim solvent system. This latter term was computed from the equation given by Chiotti and Stevens. 1 It is seen (dashed lines of Fig. 2) that for a value of n = 9.2 a somewhat better overall agreement between computed and observed solubility values is obtained when the D=A-R approximation Eq. (19) is used for 1n Y3, rather than the geometric mean approximation, Eq. (18). However, neither of these approximations take into account the possible dependence of the activity coefficient of uranium on the uranium concentration. If some thermodynamic data for the ternary system were available, then the relation reported by Darken{(l3 could be used to estimate Y in the liquid zinc-magnesium-uranium solutions. Better estimates for the activity of zinc in the ternary solutions would then also be available. Lacking sufficient data on the ternary system to apglg the thermodynamically exact quadratic formalism of Darken, 15) we have studied the utility of a less exact treatment, which is similar to the "interaction parameter" formalism of Wagner. (16) Thus, in place of either Eq. (18) or (19) is written ) o) 1n Yy = xann YU,Zn + ngln YU,Mg + ax (20) where YE,Zn and Yg,Mg are the activity coefficients for uranium 556 ot © OBSERVED ~—— GECMETRIC MEAN APPROX FOR y, /A0 ~—— D-A-R APPROX FOR y, S - URANIUM SOLUBILITY, ATOM FRACTION — it T Toad /| 1 1 1 1 1 | 1 c ol o2 o3 0.4 05 os o7 o8 09 10 MAGNESIUM, ATOM FRACTION 2, Comparison of Observed and Computed Values of the Solubility of Uranium in Liquid Zn-Mg-U Alloys at 600°C. Experimental Data from Ref, 8. o0& © OBSERVED d — COMPUTED ; Uz"ll 0"~ = ] I o < _© ™ g et 3 ol (e] o® 1 1 1 1 1 i 1 1 L ¢ 4] o2 a3 4 o8 as or -1} -1 o MAGNESIUM, ATOM FRACTION 3. Comparison of Observed and Computed Values of the Solubility of Uranium in Liquid Zn-Mg-U Alloys at 800°C. Line Computed using Eq. 20. Experimental Data from Ref. 8. 557 in liquid zinc and magnesium, respectively, at the limit of zero uranium concentration and a is a constant ("self interaction parameter'). All three parameters are functions of temperature. The constant a may be determined from one value of the activity coefficient of uranium in a ternary solution (for example, one value of the solubility in the ternary system to the right of the break), if values for the activity coefficients in the two binary systems are known. We have tested the utility of Eq. (20) by the computation of the solubility of uranium in zinc-magnesium-uranium solutions at 800°C. The comparison of the observed and computed values of the solubility shown in Fig. 3 indicates that Eq. (20) is a good approximation to use for this system. Scolubility of Plutonium in Liguid Zn-Mg Alloys. The solubility of plutonium in liquid zinc-magnesium alloys has been measured by Knighton and coworkers. (17) Qualitatively, the solubility in the zinc-rich region 1s similar to that observed in the region to the left of the maximum in the U-Zn-Mg system, i.e., the solubility decreases initially, passes through a minimum value and then increases with increasing magnesium concentration. In this region of the phase diagram various plutonium-zinc intermetallic phases are found to be in equilibrium with the liquid Zn-Mg-Pu alloys. Several studies of the plutonium-zinc system have been re- ported(18'21). Cramer and Wood(21) report that there are at least five distinct phases in the most zinc-rich region of the system. One of these phases, which melts congruently at 928°C, is PuyZnyy (Th,2n;4 rhombohedral structure). On the plutonium-rich side of Pu,Zn;, is a phase Puiny z; the crystal structure of which has recentiy been reported.(2 ) Cramer and Wood reported three phases on the zinc-rich side of PuyZnjy; However, they did not establish the exact stoichiometry of these phases but indicated that two of the phases may be polymorphs of a single phase. They reported transformations at 700, 770 and 811°C, and considered the trans- formation at 770°C to be a polymorphic transformation. It would appear that the Pu-Zn system is similar to the U-Zn system in the zinc-rich region but that the stability of the various Pu-Zn compounds is enough different that transformations readily occur between them and may be detected using thermal analysis methods. We attempted to estimate the solubility of plutonium in liquid Zn-Mg solutions using Eq. (15) and (18) with the appropriate values of the parameters for the Pu-Zn and Pu-Mg systems. No single value of n could be found which would allow the solubility at 600 and 700°C to be computed over the entire magnesium concen- tration range of the solubility measurements (0-0.42 at fct at 600°C; 0-0.28 at fct at 700°C). To establish the apparent formula for the equilibrium solid phase for each range of magnesium concentration we have used the 558 following method. The value of the activity coefficient of plutonium for each composition was computed from the solubility data using Eq. (15) rearranged in the form s AGfIo’uZ s 1in Ypy = TRT - 1n Xpy = 1n a, s (21) where AGfp was computed from the thermodynamic data for the Pu- Zn system%%%?n and ay, from the data of Chiotti and Stevens 1 for the Zn-Mg system. The activity coefficient of plutonium in the liquid alloys may also be computed from the data for the distribution of plutonium between a molten salt mixture (50 mol pct MgCl,-30 mol pct NaCl-20 mol pct KCl) and liquid Zn-Mg-Pu solutions. The chemical reaction for distribution equilibria is: PuCl,(salt) + 3/2 Mg(alloy) < Pu(alloy) + 3/2 MgCl, (salt) (22) hence, = - 1] 1n Ypu In D + 3/2 1n aMg In Ka (23) where yp,, is the activity coefficient of plutonium in the liquid alloy, D, the distribution coefficient (mol fct PuClg in salt/ at fct Pu in alloy) and K] = K, YPuclq M [K, is the thermo- dynamic (activity) equilibrium constant f8F feaction (22), Ypyc1 the activity coefficient of PuClj, and ay c1 the activity of 3 MgCl, in the molten salt]. The value of Ea aas established from the value for yp, in the binary Pu-Zn system, as has been discussed in detail elsewhere.(23) The values of n and AGffiuZn which gave the best agreement between values of Ypy computed usifig Eq. (21) and (23) were determined graphically, as illustrated in Fig. 4. Values of 1In Ypy Were comguted from the distribution data (the data previously reported( 3) were combined with more recent data obtained by Knighton(ZA)) and are shown by the open circles and the straight line in Fig. 4. Values of 1ln Yp, Were then computed using Eq. (21) and the solubility data with values of n between 8.5 and 12, The value of AGffiuznn was computed from the thermodynamic data for the Pu-Zn system.(20) Values of 1ln Yp, computed using values of n =11 and 8.5 are shown on Fig. 4. It 1s seen that the values of 1n yp, computed using n = 11 (shown as diamonds) are in good agreement with the straight line up to about = 0.2 but systematically deviate from the straight linexggove XMg = 0.2, Above %y, = 0.2, values of 1n YPu (shown as triangles) computed using n = 8.5 are seen to lie below the straight line by a nearly constant quantity. If the value of AGfp,y, 1s increased by about 1300 cal/mole, the values of ln yp, computeg using Eq. (21) and n = 8.5 are seen (as squares) to be in good agreement with the straight line. We conclude that, at 600°C, the formula for the solid phase in equilibrium with liquid Zn-Mg-Pu alloys in the composition range 0 < <0.2 is PuZnjj and in the composition range 0.2 F values for both solutes except for one-position changes for tantalum in terbium and holmium. Tungsten is less soluble than tantalum in each solvent. Values of the solubility parameters of these solvents, 04 g1, Were calculated from the enthalpies of solution by the equation ) + | Al = ] s calc i ecalc [(AHi - AHf) /Vi (14) The values of 6s are compared with (EVS/‘\_/'S)i/2 Values( 12) in Table V. calc Enthalpies of Elements Which Form Compounds with Plutonium. Plutonium reacts with carbon, thulium, and rhenium to form compounds in the solid phase over the temperature and composition range of interest.{3) If the reference state in these systems is the supercooled liquid solute and the solute is treated as dissociated in the liquid solution, the reaction may be writfen as: XA (4, supercooled) * Y Pu(4) = PuyAX (s) , (15) and PuyAX(S) + Pu(4) = Pu(4, sat'd with A), (16) Equation (5) is valid if the entropy changes that occur in going from supercooled liquid to compound to the hypothetical solid A in equilibrium 571 Table IV The Partial Molar Enthalpies of Solution of Tungsten and Tantalum in Liquid Rare-Earth Metals Afi* , kcal/mole Solvent Element W Ta Yh 51 38 Eu 47 37 La 42 36 Sm 38 35 Pr 37 34 Nd 36 34 Ce 35 32 Gd 33 28 Y 31 25 Tb 30 26 Dy 29 25 Er 28 22 Ho 25 23 Tm 25 22 Lu 24 21 Sc 18 17 Table V Published and Calculated 6s Values 65 Va.lu.Qs Element Published® Cale, with W Data Calc, with Ta data Yb 46 b3 57 Eu 38 56 58 La 68 62 59 Sm 50 64 60 Pr 65 64 60 Nd 61 65 60 Ce 72 66 62 Gd 69 68 66 Y 72 70 70 Tb 70 72 69 Dy 61 73 70 Er 63 74 73 Ho 62 77 72 Tm 57 77 73 Lu 76 79 74 Sc 78 87 80 Ref (12) 572 with solution are negligible. The enthalpies of solution calculated under this assumption were independent of temperature; the average deviation in the value as a function of temperature was less than 0.1 kcal/mole. The average enthalpies are listed in Table VI. Average correction factors for these enthalpies, as calculated by Eq (3), are also given in this Table. The geometric mean law is not valid so that no attempt has been made to calculate solubility parameters for these systems. Table VI The Partial Molar Enthalpies of Solution for Carbon, Thulium, and Rhenium in Liquid Plutonium Element Afi*, keal /mole (1- Ni)z, aveg, C 10.54+ 0.1 0.94 Tm 9.2+ 0.2 ¢.90 Re 10.5% 0.1 0.95 Enthalpies of Elements Which Form Solid Solutions with Plutonium. Both zirconium and titanium form solid solutions with plutonium from 700 to 1000°C. The equilibrium reaction for zirconium is €-Pu (s, sat'd with Zr) + Pu(4) = Pu(k, sat'd with Zr); (17) for titanium, the reaction below 770°C is €-Pu (s, sat'd with Ti) ¥+ Pu(2) = Pu(4, sat'd with Ti); (18) and for titanium above 770°C B~Ti (s, sat'd with Pu) + Pu(4 )= Pu(4, sat'd with Ti). (19) Since the excess enthalpy terms are absorbed in Afii*, the principle thermodynamic difference in these solutions from simple eutectic systems can be attributed to the configurational entropy in the solid solution. This term is - R In Nis, where Nis is the mole fraction of solute i in a saturated solid solution at a given temperature. The derivation of the basic solubility equation under these conditions results in the expression AH™ =TAS,-R|InN>-InN, | , (20) i f i i 573 where AI—{i** is equal to Afiixs/ [1- Ni:\2 + AHf . Values for N.° for zirconium and titanium were estimated from the ph_ggg diagrams. {4,5,13) Ppartial molar enthalpies of liquid solution AH; were calculated by Eq (20). The enthalpy of the plutonium-zirconium system was 4.2 % 0,1 kcal/mole from 700 to 950°C. The enthalpy of liquid solution for the titanium-plutonium system was 4.1+ 0,1 kecal/mole from 700 to 75000 and 5.3 £ 0.1 kecal/mole from 800 to 1000°C. The average correction factors for these enthalpies, as calculated by Eq (3), range from 0.5 to 0.8. However, estimates of solid phase solubilities are also approximate, and the AH; values are limited to the significance of Eq (20). The shift in the titanium enthalpy was attributed to the change in the solid phase at 77 0°c. Solubility parameters were not calculated because of the formation of solid solutions in these systems. Conclusions In this paper, enthalpy values have been calculated from solubility data by utilizing regular solution theory. Since solubility is an exponential function of the difference between the entropy and enthalpy divided by the absolute temperature, small errors in the solubility data may result in significant deviations in the latter. The approximate enthalpies can be used for qualitative estimates of solubility; refined estimates are possible with the correction factors. However, it remains necessary to measure the solubilities under the conditions of interest to obtain guantitative data, Acknowledgment The author thanks J.A, Leary for his advice and encouragement during this study. References 1. D.F. Bowersox and J.A. Leary, J. Nucl. Mater., 21, 219 (1967). 2, D. F. Bowersox and J.A. Leary, J. Nucl. Mater., 27, 181 (1968). 3. F.W. Schonfeld, "Plutonium Phase Diagrams Studied at Los Alamos, " in The Metal Plutonium, Eds. A.S. Coffinberry and W.N. Miner, University of Chicago Press, Chicago, 1961, pp 240-254, 574 10. 11, 12, 13. D.M. Poole, M.G. Bale, P.G. Mardon, J.A.C. Marples, and J.L. Nichols, "Binary Alloys, " in Plutonium 1960, Eds. E. Grison, W.B.H, Lord, and R.D. Fowler, Cleaver-Hume Press, London, 1961, pp 277-278, J.A.C. Marples, J. Less-Common Metals, 2, 331 (1960). D.H. Dennison, M.J. Tschetter, and K. A. Gschneidner, Jr., J. Less-Common Metals, 10, 108 (1966). D.H. Dennison, M.J. Tschetter, and K. A. Gschneidner, Jr., J. Less-Common Metals, 11, 423 (1966), P. Chiotti, M. F. Simons, and J.A. Kately, '""Calculations of Thermodynamic Properties from Binary Phase Diagrams, " this Symposium. E.A. Guggenheim, Proc. Roy. Soc. (London) A148: 304 (1935). J. H. Hildebrand and R. L. Scott, The Solubility of Nonelectrolytes, 3rd Ed., (Reinhold Publishing Co., New York, 1950) pp 320-345, R. Hultgren, R.L. Orr, P,D. Anderson, and K. K. Kelley, Selected Values of Thermodynamic Properties of Metals and Alloys, (J. Wiley and Sons, New York, 1963). E.T. Teatum, K.A. Gschneidner, Jr., and J. T. Waber, '""Compilation of Calculated Data Useful in Predicting Metallurgical Behavior of the Elements in Binary Alloy Systems, " Los Alamos Scientific Laboratory Report No. LA-4003 (1968). J. M. Taylor, "A Study of Phase Equilibria with Plutonium-Zirconium Alloys, " Battelle Northwest Laboratories Report No. BNWL-402 {1968). 575 BASIC DATA AND THERMODYNAMIC PROPERTIES, I1 Chairman: Irving Johnson Argonne National Laboratory Argonne, llinois, U.S.A. 577 THEORETICAL CONCEPTS FOR HIGH TEMPERATURE TERNARY MOLTEN SYSTEMS AND THEIR IMPLICATIONS IN PYROMETALLURGICAL PROCESSES Milton Blander and Kjell Hagemarka Science Center, North American Rockwell Corporation Thousand Oaks, California 91360 U.S. A Abstract Fundamental theories of ternary systems will be reviewed with a discussion of the simi;aritigs and differences between terg re- _ ciprocal systems (A ,B /X ,Y ) additive ternary systems (A ,B ,C /X ) and metallic systems. In dilute solutions of two components one may define association constants between pairs of ions in terms of funda- mental energy parameters. In concentrated solutions, deviations from ideal solution behavior are related to non-random mixing. Theoretical criteria for the predicticn of phase diagram behavior, solubilities, extraction behavior, liquid-liquid misecibility, and other thermodynamic properties important in pyrometallurgical processing will be presented for simple systems. aPresent address: 3M Company, St. Paul, Minnesota. 579 Introduction Pyrometallurgical processing is generally carried out in multi- component systems and an understanding of theoreticasl concepts is necessary to define and optimize processing methods. In this review we shall first discuss theories of ternary systems, which are the simplest members of the class of multicomponent systems. Our dis- cussion applies equally to molten salt systems and to metallie systems for which there have been parallel developments of funda- mental concepts. The extension of the concepts discussed to higher order multicomponent systems is straightforward. In the last section we will present some examples of chemical behavior in systems which illustrate the theoretical concepts. Ternary molten salt systems consist oi fgur_iogs: a system con- sisting of two cations and two anions (A ,B /X ,Y ) is a ternary reciprocal system and one consisting of three citigns_agd cne anion (AT,BT,CT/X") or three anions and one cation (A /X ,Y Z ) is an additive ternary system. The metallic systems which are analogous to these can be understood if one forma$1y+co§sigers electrons (e ) as anions. Then interstitial alloys (A ,B /e ,X )+ar$ eguigalent to reciprocal systems and substitutional alloys (A ,B ,C /e ) are equiv- alent to additive systems. The theories which have been developed for molten salts and for alloy systems should exhibit many similarities and aside from specific differences in the models, many of the derived relationships are equivalent. We shall discuss thermodynamic theories for phenomena as phase diagram behavior, solubilities, extraction, and liquid-liquid miscibility which are important in pyrometallurgical processing. Definitions To establish a framework for discussion several quentities need to be defined. The chemical behavior of a compconent in solution is, of course, determined by its chemical potential, p, which has the undesirable mathematical property that it approaches -« as the con- centration of the component approaches zero. This property is cir- cumvented by defining another guantity, the activity, a, which usually lies in the range of C to 1] u=u° + RT 2n a where u° is the chemical potential of the pure component. To make the definition of activity more meaningful one must intrecduce relationships which cannot be derived purely thermodynamically but require the introduction of statistical ¢oncepts. The simplest relationship for molten saelts depends g¢n the definition of an ideal solution as first discussed by Temkin.(:) The activity of an ideal solution of a salt Aan is defined as n 8iq = NIXNX 580 where NA is the caticen fraction of A and N, is the anion fraction of X X _ A N o X A In X In cations anions where n are the number of moles of the ions indicated. For interstitial alloys there are analogous relaticns. For additive ternary systems and substitutional alloys either NA or NX is unity. A1} real solutions deviate from ideal solutions. This deviatien is characterized by the activity coefficient y such that vy = (a/aid) = (a/NiNfi) If the ideal solution is defined in a way which is consistent with the limiting laws, then vy will always be a positive finite number. In sddition to the definition of y there are severaml other quantities which are often utilized to characterize deviations from ideal soluticn behavior: The molar excess free energy GE = RTZNifinYi all components The partial molar excess free energy of a component i &® = BT an Y, 1 1 The molar enthalpy of mixing AHm =H- ENiHi and the partial molar enthalpy of sclution E B(EE/T) BAHHA i - /Ty anij All these quantities are interrelated and often parallel each cther. Solutions Dilute in Two Components The thermodynamic solubility products (K,,) end enthalpies of solution (AHgo1y,,) of an insoluble salt can be predicted for reciprocal gystems and interstitial alloy systems by the us?2?€3§(fiimple cycle first proposed by Floeod, Fgrland, and Grjotheim, As an 581 example, the process of dissolution of a solute such as silver chloride in an alkali nitrate AgCl (solid or liguid) = AgCf (= dilution in KNOj3) can be broken up into three steps AgCe{solid or liquid) + MNO3{liquid)—™ MC&(solid or liquid) + AgNOs{(liquid) (1) MCe(solid or 1liquid) — MC2(= dilution in MNO;) (2) AgNO3(liquid) — AgNO3(e dilution in MNOj3) (3) Consequently -RT 2n K, = 8G) + AG?T + AG3T (4) and BH, .= AH) + AHp + AHj (5) where the superscript ST signifies standard states. If data are available for the reaction (1), then one may calculate X and AH from a knowledge of steps (2) and (3) which involve only bggagi systems. The complexity of the problem is thus reduced from that of understanding a large number of ternary systems to understanding a much smaller number of binary systems for whichthereare generally more available data or where reasonable estimates may often be made. The use of this cycle for calculating K. and AH has been tested for the dis- solution of silveP halides®$fi'molten alkali nitrates. The results in Table I show good agreement between measured and calculated values. A similar cycle can be applied to metallic systems, e.g. to the calceulation of the solubility product of the oxides of one metal in solution in another metal. This cycle is most useful where step (1) ?afies the largest contributions to the totals in egquations (L) and (5), Only for special cases in which activity coefficient cor- rections are negligible can one deduce solubilities directly from solubility products. Generally, theions of insoluble salts tend to associate in solution and form clusters and ion pairs. This leads to significant negative deviations from ideal solution behavior which must be taken into account in order to calculate truesolubilities. Figure 1 illustrates this for a case in which the corrections are relatively small. If a solution of Ag and €2 ion in molten KNOj were ideal, then the solubility of AgCR would be equal to the activity of AgNO3 and of KC& (the activities 582 10 09 140 144 148 {1582 {56 160 'OOO/rO(oK) 1. Solubilities and Activities in Molten KNOCj (Ref. 3) B+ Yy~ gt Yy~ g* Y~ gt Y~ ‘4 At @ gt + Y- g+ @ B* g+ Y~ gt Y~ B+ Y~ gt Y~ A\ J/ Y AL /! N M v- B* ' Bt Y~ BY v~ Y- At @ BY ¢ Y~ gt @ B*t gt ‘o B+ Y~ gt e gt Y~ 2. Mecdel for the Definition gf AAj, the Specific Bond Free Energy for the Ion Pair A -X . 583 ¥86 Measured Values of -RT &n K Table I. , AH » and Their Comparison with Calculated Values (3)(k) SP soln. ~RT &n KSP’ kcal./mole AH, kcal./mole Solute Solvent Measured Calcd, Measured Calcd. Agl NaNO3 29.5 31.5 29.5 29.8 AgI KNO 3 26.7 28.2 27.0 AgBr NaNC3-KNOg3 21.7 22.1 22.4 22.9 Agl NaNC3-KNO4 28.4 29.9 27.9 28.4 AgCe KNO 5 17.0 17.8 19.2 19.6 A are based on standard states such that vy = 1 at infinite dilution). The solubility of AgCLf is higher than these activities as the result of the formation of associated species in solution as AgT + 027 = AgCe (6) AgCL + C&~ = AgCy (1) AgCa + Ag' = agycet (8) which make the activity coefficients of AgNO3 and KCL in solution less than unity. The fact that the plotted activities of AgN03 are higher than those of KC&, even though the concentration of Ag eqguals that of C& , indicates that equilibria as (7) which form species containing more chloride ions than silver ions have larger association constants than equilibria as (8). In many other cases the activity coefficients are so low that sclubilities are orders of magnitude higher than for an ideal solution. This is true not only for reciprocal molten sslt systems but also for ternary interstitial metallic systems. Con- sequently, it is important to understand association equilibrias in dilute solutions and the influence of these equilibria on the con- centration dependence of activity coefficients. Let us consider a solvent BY dilute in A+ and X ions. For metallic systems, Y 1is an electron. The association of A and X to form an A -X pair may be visualized as in the two dimensional representation of Figure 2. The interchange of the circled X and Y ions gives rise to a specific bond free energy change AA;. From the statistical mechanical analysis of a quasi-lattice model it can be shown that K;, the asso?%%f%?TTsonstant for the equilibrium A +X % AX in Figure 2 is given by K; = Z(exp(-AL;/RT)-1) (9} where Z is a coordinstion number which ranges from sbout L-6 for molten salts. In reciprocal systems and interstitial systems, AA;, is largely related toc the exchange of nearest neighbors. In simple systems where long range interactions do not predominate and where A and X are simple spherical ions one might expect that 44 is independent of temperature and is truly an energy AE;. In order to test this, one mast make measurements over a wide range of temperatures. The calculation of K; from experimental data on the activity coef- ficients of the components AY or BX can be simply made for the case in which all components and species at low concentrations obey Henry's law. Then it can be shown that + (K1K2 - 1 n Yay =-K1N > K$INZ + (2K K12-K{N Ny + ... (10) X AX and 585 + (KiKpp - 3 KON2 + (2K)Kp-K2)N, N + ... (11) in YBX = -K]_N ANx A and by comparison with a(gicLaurin serles expansion one can derive a method for evaluating K from limiting slopes 9 in vy ¢ 4n vy -K; = 1lim ——Efi__gg = 1lim -—?fir—igg (12) A X NA*O NA+O Nx+0 NX+O For metallic sclutions K; is the ?e%ative of the atomic intefacgion coefficients introduced by Wagner'9) and Chipman and+Elliott 10/, Figure 3 exhibits some of the data for the system Ag + CL” in NalNO 3 which have been utilized in conJuncEion with equation (12) to evaluate K; for the formation of Ag - €%~ ion pairs over a range of temperatures. Y?luT? og 4A; derived from such measurements at several temperatures (11)(12)(13 are given in Table II and are independent of temperature within experimental errors for all reasonable values of the coordination number. Consequently, equation (9) can be used to predict the tem- perature coefficients of K; where A and X~ are simple spherical ions. Similar data hayg been obtained for metellic systems as shown in Figures 4 and 5 ) and Table II1I{14) yhere the interaction coefficients ki = -K;. An extensive check of the temperature dependence of AA, foé metallic systems has not been made but AA; appears to be temperature independent in some of the measurements which have been made (see Figures % and 5 from reference 6). More complete tests would help to confirm this. The quasi lattice model also leads to expressions for the higher association constants(5)(T) which are somewhat more complex., Similar equations o}? f?r edditive molten salt systems and for substitutional alloys TI{15) with a modification in the definition of AA; which is the specific free energy change for the exchange of next nearest neighbors (considering electrons as anions). This modifi- cation is illustrated in the two dimensional model in Figure 6 which illustrates the associmtion of next nearest neighbor cations B++c+i_rB+_C+ An important conclusion may be deduced from equation (9). If AA, is zero, then K; is zero. If AA; is negative, then the two ions involved associate and K; can be a large number if AA; is very negative. On the other hand, if AA; is positive, it means that the "associating" ions repel each other which leads to negative values of K;. This pecularity is paralleled by the behavior of second virial coefficients of some gases (which are the negative of association constants) and 586 0.4 T=331°C 364°C — N, =0.30x10"3 f/ Ag =4U. X / 0.3 a02°¢ 385°C b -] S / 423°C Zo02 /- Va x S 2 43s°c/‘ 500°¢C 04 — 3. 4 (x 1073) Temperature Dependence of - log v (Ref. 13). Neo AgN03 in NaNOj R\ N N X log O | o I =) Log D for Sulphur in Alloys of Co £€er and Nickel at 1300°C 0, 1Lkoo°C A and 1500°C O. (Ref. (in alloy)/ys(ln Cu) 0 I -0 ~‘~\<‘\~ ~_ -0-2 ~ X - Q —0'3 N [~ — (o3} a L _p.a A o— —;quf’/ O o 0 -0-5 0] Q-2 0-4 0:6 0-8 1-0 NFe 5. Log D for Sulphur in Alloys of Cop ger and Iron at 1300°C 0, 1L00°C 4 and 1500°C O. (Ref. D=y (in alloy)/ys(ln Cu) 588 AV X oA NG A < 8 x + x & < N N SN + ¥ At x o N S x B x 4+ x At ¥ At ox ot At xm at 6. Two Dimensional Model for the Association of Next Nearest Neighbor, 589 Table II. Values of AA; Obtained From the Comparjscn,of Theory with Experimental Data%ll (12)?333 Ag® + CL” 1in KNOj 623 6.12 5.85 5.62 552 643 6.17 5.89 5.66 498 658 6.21 5.93 5.69 L60 675 6.17 5.87 5.64 396 696 6.18 5.88 5.63 348 709 6.17 5.86 5.62 315 + - Ag + CZ 1in NaNOj 604 5.10 4,83 4,62 277 637 5.12 L. 84 4,62 226 658 5.17 4,88 4.65 205 675 5.10 h,81 4,57 176 696 5.13 4,83 b,59 160 711 5.12 4,81 4,56 146 773 5,14 4.82 4,55 110 + - Ag + C2 in (Na-K)NOj 506 5.6 5.4 5.2 1050 551 5.57 5.33 5.13 644 658 5.67 5.38 5.15 302 752 5.72 5.L0 5.13 180 801 5.62 5.28 5.00 133 aKl in mole fraction units. 590 Teble IIT. Atomic Interaction Coefficients in Liquid Iron, l600°C(lh) Values of ki,j = - K Added Component i component J H C N 0 si S C +427 +11.1 +12°% - 6.4 +10 +6.0 0 + 3.5 - 13 Al + 6.7 - 0.3 -1300 e 6.5 Si 3.2 11.2 + 5.4 - 27 3.4 7.6 P i aaees v + 8.9 11 5.7 S . s e - 12 ‘s -3.8 v . - 8.0 -20.5 - 57 Wb - 1.5 -23 Ta e e -25.2 Cr . - 5.1 - 9.6 - 8.8 ..... =h.7 Mo - 3.5 - k.5 1.4 W - 2.3 - 1.1 6.4 Mn - 1.4 - b.5 07 0? -5.7 Co . 2.9 2.7 1.7 Ni 2.9 2.6 1.4 1.2 0 Cu 4.2 2.4 - 2.5 ... -3.2 2o v 3.6 - e - 3.0 Sn con 0 3.4 0 591 it arises because of the definition of an ideal solution in which K; is zero for the random number of pairs (N NX in reciprocal systems or NBN in additive systems). If these pairs of ions repel each othe? Ehen K; < 0. An examinetion of equation (9) for positive values of AA) shows that the most negative value of K, possible is -Z. Thus, the positive values of K| have no fundamental limit but the negative values do. In dilute solution, this means that it is easier to lower than to raise activity coefficients of solutes sig- nificantly by the addition of a third constituent. This is con- sistent with much of the data on metallic systems where values of K; (the negative of the Wagner coeffi?ifi?t k,,) range from very large numbers to small negative numbers. 1 Tfig only wvalue of K| less than -12 in Table III taken from reference (14) is questionable. A precise estimate of AA; is not possible at present but relative values for similar compounds may be often guessed from the crude assumption of additivity. If pair bonds are additive then for reciprocal systems o) ~ |AG AA] = EE?) (13) where AG® is the standard free energy for the reaction AY(2) + BX(%)— AX(2) + BY(2) (1k) Of course, real systems dc not exhibit this additivity so that equation {13} may be used only for crude guesses. Analogous relations held for interstitial alloys. DNonadditivity is manifested in yet another way. If one lecoks at the eguilibrium (7) for the association of a second chloride ion with a silver icn, if there is no contri- bution to the bond free energy which is related to the non-ideality of the AgNO3-MNO3 and MCR-MNO3 binary systems, then additivity of pair bond interactions means that the specific bond free energy {03 this process, AA; = AAy. Trom a generalized quasi-lattice model 5 4(2-1) (B1Bp=2B1+1) (15) Kle = ) where B, = exp (-AAi/RT). When Afdp= AA, ko = {221 (g1 < (2L (16) (6)(16)(17)(15)(T) Simple guasi-lattice theory is based on the assumption of the additivity of the energy of pair bond intersctions (AA = AA] = AAp = ..., = MA, = ...). However,many systems depart from this ( assumption and require the generalized treatment in dilute solutions. 5) 592 Solutions Dilute in One Component The theory for calculations of solutions of one component in dilute solutj olvent has been discussed in several papers. (6)( lé??l7? 19?%7?TK5? With minor differences in the definition of the parameters, one obtains the same relations for reciprocal and sdditive ternary mixtures. All of the calculations are based upcn the additivity of pair bond interactions, and, to the degree that there are deviations from this, real systems might be expected to deviate from theoretical celculations. Even for those cases where such deviations are considerable, theory pro- vides an understanding of the parameters and phenomena which govern solution behavior and ensbles one to make semiguantitative pre- dictions. The theoretical equations can be expressed in several different forms, each of which provides & different insight into the problem. Hagemark has derived expressions for ternag¥ s¥stem5 at any com- position from the quasi-chemical theory 1 for the additive ternary system AX-BX-CX (or A-B-C). In the limit of N, -~ O an expression from his general equations for the activity coefficient of the component CX (or C) can be derived | | ] o [a] e =< Q E' A S in Yo = (17) n YC(A) -y, -2 n (NA+KNB) where 7 YC(A)YB RN e This is exactly equivalent to = moS? gfimplex expression derived earlier by Alcock and Rlchardson and for cases in which Y. =Yg =1 it is equivelent to an expression derived by Wagner. I% equation (17) we have redefined Hegemark's parameters and sub- stituted the nomenclature of Alcock and Richardson. Here vy, is the activity coefficient of CX (or C), v, of AX {or A), y, Of BX (or B) in & binary mixture of AX + BX (orf A + B), v.(A) iS the act1v1ty coefficient of CX (or C) at infinite dilution in AX (or A) and YQ (B) is for CX (or C) at infinite dilution in BX {(or B). The T quantIties y.(A) and y.(B) are independent of N, and N vhereas Y and Yg &re functions of NA and NB. 593 For the model system when N, =+ C c NA-a Z Y, = | —— (18) A NA NB-a 2 Yo = (19) B NB where a4 = / ENANB 2/2 (20) 1/1eh, Nk v (4))27% o) and Yg(&) = v,(B) Physical insight into the consequences of equation (17) can be gained by rearranging the equation to the more complex form E - G (AB) in vy, =N, n YC(A) + Ny &n YC(B) -t cb(NA) (21) The first two terms of (21) are linear and the third term is E G (AB) _ mr = Ny 4n oy, + Ny dn vy (22) where GE(AB) is the molar excess free energy of the binary mixture of AX + BX which is zero when N, or NB = 0 and generally has a maximum or minimum which is at fi = N = 0.5 in the model system but may deviate from this in reafi sys%ems. The function ¢(NA) is given by the expression ¢(N,) = -ZN, N, l:f&—A n (1+1\1A Gf'{—K-)} %q; in (1-4-1\1B (K—la] (23) where in Alcock and Richardson's notation 594 (2k) and K is a function of N,. The function ¢ is always negative, is generally smell, and is Zero when N, or NB = 0. Consequently, when the binary system AX-BX exhibits very large deviations from ideal solution behavior, -G (AB) has a large maximum or minimum which is to be added to the linear terms. Such behavior may have & prefound influence on the chemical behavior of components in pyrometallurgical processes. We will illustrate this behavior in a later section. The term ¢ tends to lower the maxima or minima to an extent which depends solely on K. Equation (17) may be rewritten for reciprocal systems and is equivalent to limiting forms which can be ?er%ved from the gquasi-lattice theory for reciprocal systems. 17 For the component AX in dilute solution in the binary system AY - BY one obtains the expression In Y,y = &n YAX(AY) - Z in [NA+KNB] (25) where v,.(AY) is the activity coefficient of AX at infinite dilution in AY afié where the analogue of equation (24) is o] AG 2 YaxAY)Ygy T K =W—e (26) Yrx Yay where AG° is the free energy change for reaction (14). Eguetions (25) and (26) do not take into account the interaction between _ next nearest neighvtor A and B cations in the vicinity of an X anion. In the limit of N, = 1 equations (25) and (262 are exactly equivalent to the Flood, ?¢rland and Grjotheim cycle. 2) Equations (25) and (17) can be seen to be equivalent if Ypx in (25) is equivalent to YaYe in (17). Concentrated Solutions The thermodynamic properties of concentrated ternary solutions have been treated %S Shree papers utilizing perturbation theory for recig— rocal systems 1’ and the quasi~chemical approach for reciprocal(l7 and additive ternary systems,(l5) These theories provide a funda- mental Justification for many of the equilibrium chemical properties and provide a means for the prediction of chemical behavier, phase diagrams, and of liquid-liquid miscibility. Although these theories apply only to simple systems, they enable us to understand the basis for the behavior of more complex systems. 595 Conformal ionie solution theory(22) is a statistical mechanical second order perturbation theory which haes been applied to reciprocal systems.(2l) For the activity coefficients of the com- ponent AX, it leads to the relatioq RT &n v,y = -NpN 2G° + N_N.(N -Ny JA BY BY 'Y BXY +NY(N N_+N . N_) Ny, Ny + NB(N N +N.N_)x Aaxy BV tNANy A apx N (NN Aoy + NN (NN AN N -N N DA (27) where AG® is the standard free energy chenge for reaction (1k4), and the A parameters are defined by the excess free energies of the binary system indicated. TFor example E AGT (AX-BX) = N Nphyny (28) The parameter A can be approximated by the equaticn A = - (aG°)2/2ZRT (29) by & comparison of its theoretical form with the quasi-chemical theory. For the simple quasi-binary system AX-BY, equation (28) reduces to 2(1-N = 2 -— O RT 40 v,y NBY[ AGT+) A By asyt gy’ (A axy* aBx~Bxy~*any’ + (1-N_ ) (3N, -1)A] (30) BY BY Since the theory is only second order, it is not exact for cases in which AG™ > ZRT or where regular solution theory expressed by equations as {28) are grossly incorrect. Despite the deviations of real systems from the assumptions in the theory, this theory can be useful in helping to understand and predict thermodynsmic behavior when the deviations from the assumptions are not gross. For example, the liquidus temperatures of the Li, §||F, Ci system calculated from conformal ionic solution theory are given in Figure T and can be seen to correspond well with the measured liquidus temperatures shown in Figure 8. A close examination of the relation between the calculated liquidus temperatures and the parameters in the theory provides considerable insight into under- standing phase diagrams of such systems. For example, the magnitude of the bowing cut of the isotherms for LiF from the LiF corner is related to the magnitude of AGY and to the fact that LiF is a member of the stable pair of salts (LiF + KC&). The large phase field of LiF and the small one for KF 1s related to the fact that 596 LIF AA=16 3 kcal./mole Z=6 1 . . . 6 7 B 9 10 LiCt N K KClI T. Liquidus Temperatures Calculated for the Li, K||F, C2 System. 848 606 498 T72 8. Liquidus Temperatures Measured in the Li, K||F, Ce System. 597 LiF 1s a member of the stable pair of salts and exhibits positive deviations from i1deal solution behavior whereas KF 1s a member of the unstable pair (KF + Li1C%) and exhibits negative deviations from 1deal behavior. The extension of the KCR phase field in the di- rection of the LiF-KF binary 1s related to the large negative value of ALiKF' Similar rationalizations may be made for other systems. If we examine equation (27) more carefully, we may relate each of the terms to less general prior concepts. T?e first term has been discussed by Floed, Férland, and Grjotheim 2) and has been related to nearest neighbor cation-anion interactions. The next four terms were first proposed by Férland 23a)yho related them to next nearest neighbor cation-cation or anion-anion ilnteractions. The last term 1s a correction for non-random mixing analogous to g similar term in the quasi-lattice theory. 17 The significance of non-random leln% can be seen in Figure 9 2 in which RT Rn Y 18 plotted versus N for the LiF-KCL quasi-binary system. Elne calculated withouf the inclusion of the kst term of equatlon (27) deviates from the plotted points which were calculated from liquidus temperatures. The inclusicn of the non-random mixing term leads to an "S" shaped curve which corresponds to the measurements. The non-random mixing alters the concentration dependence of the activity coefficients significantly. Thus, a calculation of the consolute temperatures, T , (below which the solution at some com- position separates 1nto tWo separate liguid phases) without thais last term leads to values of T which are much too high and leads to the 1ncorrect preaiction of an extensive miscibility gap in the L1, KIIF C& system. The inclusion of the last term provides a much lower and better estimate of TC o o 860, tamxamy™ o ey (31) ¢ T 5.5R 11R where the numerical factors 5.5 and 11 in the denominator replace the values 4 and 8 for random mixing. The calculation of T from equation (31) appears to be con51stent % g? g? llquld—ilquld miscibility gaps i1n reciprocal systems (2 2 There are many reciprocal salt systems which exhibit miscibility gaps which can prove to be useful i1n extraction processes. The quasi-chemical approximations for reciprocal systems applies only to systems in which only nearest neighbor interactions are significant and where the pair bond interactions are additive. The activaity coefficients for the component AX are given by 1-Y Ln Yax = Z 4n 17 (32) Y 598 6000 Calculated without non- / random mixing term ’ 5000 . 4000 N £ = 3000k 2000 Calculated with the inclusion of the non-random mixing term 1000 | | | | l 0 I 2 3 4 .5 6 (1=Nyp)? 9. Calculated and Measured Activity Coefficients in the LiF- KC& Quasi-Binary System. (O - Values Obtained from Liquidus Temperatures). 599 where Y is given by E - Ny'NAX l-g 1- NA_NY+NA¥ exp (+4A/RT) (33) where AA is the specific bond free energy for formation of an A-X palr bond for the additive case. When the next nearest neighbor interactions are not large relative to nearest neighbor interactions, an spproximate correction to equation (32) can be made simply by adding the Férland correction terms which are the 2nd, 3rd, 4th, and Sth terms on the right hand side of equation (27).{23a) This correction should not be adequate for cases in which the next nearest neighbor interactions asre not small relative to the nearest neighbor interaction terms. In effect, the fundamental energy parameter AA in equation (33) should be concentration dependent. From an examination of the model (as in Figure 2) one cen deduce one possible ad hoc form for AA when |AA| is not very large Y, YL Y Y X 7oA = 4G° + RT | 4n —#5 + 4n ?? + 4n %,, + 2n -?%TT (3k) Yay YBX YBX YAY where the ' designates the binary AX-AY, '' BX-BY, ''' AX-BX, and "' AY-BY. For cases in which equaticns as (28) are valid equation (34) becomes 7Ah = AGC + (N_-N.){x v i) gy =gy ) + (NN ) (A 53 (A pnx P apy! (35) Such an ad hoc concentration dependent energy term needs a closer examination in order to use it in a manner which is consistent with limiting forms which have better fundamental Jjustifications. When non-random mixing is very ypronounced, one would have to substitute for N B NX and NY in equation (35) to take into account the fact that %he average ionlic environments are not characterized by these ionic fractions. For example, N, would be substituted by Y, NX by (l—I) and N, and N, by analogous but mcre complex expressions, Such ad hoc correc%io s are consistent with observed deviations from simple theory 2T) which have not been explained previously. The quasi-lattice model has been applied to additive ternary systems (A-B-C or AX-BX-CX).(15) The theory takes into account interactions between A, B, and C. In the model for the system A-B-C,energies are defined for the exchange of nearest neighbor 600 pairs (in AX-BX-CX these are next nearest nelghbor pairs) Exchange Energy change 2A-B % A-A + B-B Beyp (36) 2A-C = A-A + C-C be, (37) 2B-C = B-B + C-C begq (38) Each of these energies is related to thermodynamic parameters in the binary systems and in real systems they can be evaluated from measurements in the three binaries A-B, A-C, and B-C. 1In addition to these three parameters a coordination number, Z, enters the theory. The partial molar excess Ifree energy of component C, EC’ for example, is given by the expression E_Z o G, == RT &n _%Q (39) c 2 NC where o represents the fraction of the total number of nearest neighbor palirs wbfich are C-C pairs. For random mixing this is equal to N2 and G, = 0. If the C atoms attragt each other o > N2 tHe numbér of pairs exceeds random G, > O and the com- ponent € exhibits positive deviations from idéal solution behavior. If C atoms repel each other on the average o < N2 and C exhibits negative deviations from ideal behavior. To évaluate e one must solve the equations N.=a + o + o {(40) AEAC T kT 2 l-e QAC - [(wy+8) - gy = oyplage + (NA—aAB)(NC - “BC) =0, (41) 601 e ’T ) l-e aBC - [(Ng#NG) ~opq = applag, + (NB-uAB)(NC—aAC) = 0, (L2) i AEAB l-e kT aABz - LN g} - oy = apalayg + (NA-GAC)(NB-aBC) = 0, (L3) where a,, is the fraction of pairs which are i-j. Except for a few simple iimiting forms, these non-linear equations must be solved numerically. The influence of each pair interaction on thermodynamic behavior can be investigated by the use of these eguations. Even though real systems do not often conform tc the simple model used, the equations are an important guide in a complete understanding of multicomponent systems. As an illustration of this, in Figure 10, we exhibit caleculations of phase diagrams from the theory for the simple case in which the three salts have a melting point of 1000°C and an entropy of fusion of 6 e.u. The energy parameters Ae,, are related to deviations from ideality in the binary systems. %gr example the quasi-lattice theory leads to an expression for the total excess free energy of a 50-50 mixture of 1 and |} cg_s(ij) =-Zrran (3 (exp(aey,/2 k1) + 1)1 (4E) and for the partial molar excess free energy of a component i at infinite dilution in J =ZX Ae Ef(J) = -—‘??——Li (45) where XX is Avogadros number, Consequently, calculations in the ternary system can be made using information obtained from the binary systems only. Figure 10a exhibits liquidus temperatures for an ideal system. In Figures 10b-f a value of the cocordination number Z = 10 has been used. Figure 10b exhibits the case in which 602 8ll three binaries exhibit negative deviations from idesl behsavior (Ae/XT = 0.25). The liquidus isotherms for component C bow out away from the C corner of the triangle. Figure 10c exhibits the case 1n which all three binaries exhibit positive deviations from 1deal behavior and Ae/kT = -0.25 for all three. The 1sotherms for component C bow towards the C corner of the triangle. The magni- tudes of the howing in or out depend on the magnitudes of Ae. The value of Ae C and Ac o influence the magnitude as is 1llustrated by comparing F%gures 10%, 10d, and 10e which were calculated for AEA /KT = Ae C/kT = +0.25, 0 and -0.25 respectively with Ae,_/kT = +O.55. The Eow1ng out of the ligquidus isotherms for C means that the quasi-binary C + AO B exhibits positive deviations from 1deal solution behav1or'?e9&%1ve to the two binaries AC and BC. This phenomenon can be understood simply in the lamit of N, = 0 given 1in equation (21) where for the cases such as in 10a, b, &, 4, and e where ¢(N. ) 1s small, &n vy, 1s close to the linear sum ofi the wvalues of &n v, In the AC and BC Binaries minus GC(AB), where G.(AB) for 10b, 104, ang 10e 1s negative and n v, has a maximum at X, =7"1/2. 1In Figure 10f we exhibit a more realistic case in which Ae = 0, bte, /KT = 0.25, Ae_. /KT = -0 25 so that AB 1s 1deal, AC exhibits negative and BC positive deviations from ideal behavior. This case parallels the behavior of real systems such ags LiF-NaF-KF, NaF-KF-RbF, and LiF-NaF-RbF. By ccmparison of measured phase dia- grams with such thecoretical calculations we can learn which parameters and 1nteractions lead to particular features in the phase diagrams. Thus, the shape of liquidus curves can tell us much about ionic interactions and conversely, & knowledge of binary systems can be an a1d 1n predicting phase behavior and other thermodynamic properties in ternary systems. It should be remembered that liquidus curves are isoactivity curves, Thus, as an example, for cases in which AEA = Ag e if cne varies the ratio of B to A at constant values of N ghe ac%1v1t1es and activity coefficients go through a maximum 1f EE > 0 and the AB binary exhibits negative deviations from ideal befigv1or, and they go through a minimum if Ae < 0 and the AB binary exhibits positive deviations from i1deal behavior. These and the analogous maxima and minima 1n reciprocal systems are important chemical properties of ternary systems and exert a profound influence on their chemistiry. For the case 1n which Ae = Ae one may derive a simple relation for the quasi-binary mixture of component C end an equimolar mix- ture of AB 2(1-8.) E;g:RTznyC:%RTzni]—l- C (46) C 1+ /l-hNéTl—NC)(l-E) where 603 S09 10e 10£f \(T7// +025 10, Calculated Phase Diagrams for Additive Ternary Systems. Isctherms are Plotted Every 20°. AS = 6 T, = 1000°C, Z = 10 The values of the energy parameters for each diagram are AC BC 1 AB kT kT kT a 0 0 0 b 0.25 0.25 0.25 c -0.25 -0.25 -0.25 d 0 0 0.25 e -0.25 -0.25 0.25 £ 0.25 -0.25 0 GE(AB) §E(A) E = L (exp(ae,./2kT)+1) (exp(-4e, ./kT) = exg|- 2L 2.t o \EXPVOEsp P AC El= 7 T RT 7 “RT (L7) Equation (46) is analogous to equations for reciprocal systems and if we expand (46) up to second order terms in ( -E) and sgbstltute from (47) keeping terms up tc second order in G {A) and Gy 5(AB) we obtain N.(2-3N,) 2 ég = RT n v, = (1-N,)? [@g(A)—Gg.s(AB)) - ng— G}E(A) Go 5(AB)) + ... (L8) Equation (48) has the same form as the analogous equation (30) for reciprocal systems and it leads to the prediction of an "S" shaped curve for a plot of RT &n v, vs. (1=N.)¢ which is similar to that in reciprocal systems. The quantity G (AB) which is the total excess free energy of mixing of a 50-50 mlxfigre of A and B, is defined in equat] on (44) and corresponds to the quantity AG® in equation (30) and G.(A) is defined in equation (45) and corresponds to X, .. The conso ute temperature calculated from equation (46) for this case in which AEAC = AEBC is a5(a)-6E _(aB) C 0.5 Tc = 2 -ZR fn (1 - <) (L9) which is apalogous to equation (31). Gg 5(AB) is defined in equation ‘ (L4) and GC(A) in equation (45). ) Thus, theory leads to the prediction of miscikility gaps which are completely enclosed within the ternary composition triangle. Applications In this section we will illustrate the application of some of the principles discussed in previous sections. If we consider the solu- bility product of AL,03 in a solvent as NaF calculated by the cycle illustrated in equations (1)-(3), then for the reaction analogous to that in equation (1) 606 A%,03 + 6NaF — 3Na,0 + 2ARF3 at 1200°K the free energy change is over 600 kcal. Consequently, one would predict an extremely small solubility product for AL,03 in NaF. Even with fairly negative values of the standard free energies of solution corresponding to equations (2) and (3) and even with feas- ible deviations from Henry's law, the predicted solubility for this case is stil]l extremely low. However, polyvalent ions as At and 0 “in & uni-univalent salt should associate strongly in a manner analogous to the equilibria (6), (7), and (8) so that +3 = + ALY+ 0 = A0 (50) AT + 07 = Ag0j (51) AR0Y + apt3 = pg,0te (52) 28207 = AL,0, 7, ete. (53) with a tendency to produce species with low charge density (e.g. (50), 51), and (53)). From an electrostatic point of view when the A% 3 concentration is not high species of low charge density are energetically more favorable than igecies of high charge density which contain more than one Af icn. In s%%v?r halide systems in molten alkali nitrates, it is well xnown(3) (4 that with the addition of a common ion (either silver or halide ions) to an alkali nitrate the solubility of the silver halide first decreases to a minimum and then increases with further additions. This phenomenon results largely from the association equilibria involving spe¢ies of stoichiometry different from the silver halide (e.g. Ag,X or AgXE) which have a net charge. Analogously, for Af,03, the addition of ARFj3 or Naj;0 to the solvent NaF, if large enough, should sclubilize Af,03. For ALF3 additions in this reciprocal system the important species involved in solu~ bilization should have a ratio of A2/0 which, (on the average) is greater than 2/3 (e.g. species as A%0 , A%,0, , etc., are present). The temperature dependence of equilibria involving these species is predictable from equations as (9). The temperature dependence is probably very large so that theory may help in defining ad- vantageous conditions for solubilization. At high concentrations of A%F3 (> 15 mole %) in NaF the formation of more highly charged species containing more than one A%L*3 ion as A2,0%" becomes less unfavorable because the high average charge density of caticns (due to the high concentration of A2*3 ions) will stabilize species of high charge density. In a similar manner, association equilibris of constituents in dilute solution in molten irocn alloys are important in deoxidation 607 and desulfurization reacticns and the temperature dependence describted by equations as (9) and (15) aid in predicting the temperature dependence of the equilibrium quotients for these reactions if the parameters AAi are temperature independent. One potential application which has not been utilized in a practical manner is the utiliza?ig? of immiscible salt pairs for extraction equilibria., Kennedy 29) has published en academic il- lustration of this. He has measured the distribution of T2 ion between two immiscible salts KNO3-AgBr. He defined a measured distribution coefficient KO where K, = Np,+(KNO3) /N, 4(AgBr) (54) and an association constant K; for the association equilibrium in KNOj3 % + Br~ = T2Br (55) N T Ky = fi_&@%_:_ (56) T¢+ Br and a "true" distribution coefficient, K, is defined as Noepy (K¥03) K = NTQ+(AgBr) (57) + such that in the hypothetical case in which all T# ions in KINOj existed as T&Br then K = K. If the concentration range is such that TL ions exist as T+ or TeBr species in solution in KNOj3 then K K =K+ o] KINB; (58) Thus from measured values of K, at various concentrations of free bromide ions one can evaluate K; for the equilibrium (55). 1In the presence of Ag+ ions N_- could be calculated from the total bromide concentration and a knowledge of the asscciation of bromide with silver. Figure 11 exhibits a plot of K, versus 1/N . ot five temperatures ranging from 450-550°C. The distribution coefficients are strong functions of the complexing anion concentration. Values of K; for the equilibrium (55) were evaluated from the ratio of the intercept to the slope of the curves in Figure 11. Values of K 608 l.B— 5500 161 510° () .4+ 490" 270° 1.2+ O 450° o} ] 1 1 1 L 1 0 {00 200 300 400 500 600 I Ngr 11. Distribution of Thallium Bromide Between KNO3 and AgBr. 609 in mole fraction units ranged from 26 at 450°C to 19 at 550°C. The temperature coefficients of K; (and of the slopes)are consistent with the predictions from equation (9) for a constant value of AA; and the theory can aid in predicting the temperature dependence of K ., The ratio K/K; in equation (58) can be calculated from a cycle similar to the FFG cycle (equilibria (1), (2), and (3)) TeCa(in AgCL) + KNO3(liquid)-— TANO3(liguid) + KC2 (59) TRNO3(1iquid)—o TANO3(~ dilution in KNOj3) (60) KC%— KC2 (= dilution in KNO3) (61) One can calculate AG?T if one knows the activity coefficients of TeCr in AgCl as well gs the standard free energies formation of all the components. The ratio (K/K;) is then given by AG?S + Ang + AG?? = -RT 4n(K/K;) (62) Immiscible sal Byases for extraction equilibrias have been utilized by Moore 2 for equilibria of potentially practical interest in the additive ternary system LiC2-KC2-A%C23. He meagured distribution ccefficients between the two immiscible phases LiC% and KCf-AfCL3 where the immiscibility is related to the very negative value of the excess free energy of mixing of the KCi-ARCLj system which corresponds to the system A-B in Hagemark's theory. Table III gives values of the distribution coefficients of 9 salts at 625°C where - mole fraction in KALCL (63) KD mole fraction in LiC4R Table III. Distribution Ccefficients for Metal Chlorides in the System: LiCL~KAeCL, at 625°C. (29 U02C22 0.84 UCL3g 0.0k PuCR3 0.0h FeClj 9.1 RuCf 3 0.24 SnCle 0-9 SrcCh, 0.014 NaCg 0.56 CsC2 18.1 610 From an examination of Hagemark's equations we may raticnalize the relative values of . Qualitatively, values of > 1 mean that the salt reacts more strongly with ALC23; or KCZ than these two react with each other. Values of %8 < 1 generally mean a relatively weak interaction of the salt with KC& or ALCRl3. This information can be guessed from a knowledge of the binary systems. The inter- actions with LiCR phase, which are not large for these cases needs to be taken into account too. Other additive ternary and reciprocal systems with similar miscibility gaps should exhibit separations which are potentially useful. Another sapplication of these principles by Morrey and Moore(BO) was in the extraction of UCL; by molten AL in the binary system KC2-A%CR3. From an examination of equation (21) where the binary system KC2&-ALCL3 corresponds to the A-B binary we see that the very negative excess free energy of mixing of KC&-ALCZ; will con- tribute to a maximum in the activity coefficients for a component C as UCL3. UCR3 interacts weakly with KC2 and 1f it also inter= acts weakly with or has a positive free energy of solution in AfCL3 then there should be a maximum in the activity coefficients of UCZ3 in this system as a function of the composition of the KCe-A2CL3 solvent. The maximum will tend to be at compositions where G has a minimum, e.g. at about 50-50 mole % for KCR-ALCLj3. Such a maximum will lead to & maximum in the extraction cceffi- clents as was observed by Morrey and Moore. Similarly, maxima in vy, should prcduce minima in solubility which, for example, have been observed for PuFj3 in the LiF-BeF, and NaF-BeF, binaries at composit%oni which probably are close to minima in G for the binaries.'3! The activities of "Fe0" in the steelmaking system Fe0-Ca0-S8i0, is consistent with the expectations from Hagemark's calculstions (which apply only to simpler systems). Taylor and Chipman(32) have measured the activities of Fe0 in this system and these are plotted for the guasi-binary system Fe0-Ca,;S510, in Figure 12. If we calpulate Ln y, o from equation (48) by the ad hoc replace- ment of (G,-G (AB)) by Ye free energy of formation of 3 Ca,Si0, (which is approximately equal to the excess free energy of mixing per two equivalents) and of the mole fractions by equi- valent fractions we obtain the sclid "S" shaped curve in Figure 12 which correspoends well with the measurements. Even without a Justification of this ad hoc calculation it is clear that the "S" shaped character of the curve is related to clustering of Fe 2 with 0”2 ions. This association leads to a concentration de- pendence cof the activities o; FeQ and a phase behavior similar to that found for LiF in the Li , K ||F ,C¢ system. 611 <19 log YFe0 T=1600°C . - 12. Plot of log Yp Néa = equlvalegg fraction o? CasS10, (N 1 Ca )2 versus (Né )2 1n the Fe0-Ca,;S10, Quasi-Binary Conclusions Fundamental concepts can provide considerable insight into the chemistry of multicomponent systems. Although theories are some- times approximate cor apply only to simple systems, they neverthe- less provide a basis for guessing the important parameters and variables necessary for choosing and optimizing pyrometellurgical processes., 613 10. 11. 1z, 13. References Temkin, M., '"Mixtures of Fused Salts as Ionic Solutions," Acta Physiocochim. URSS, 20, 411 (1945) Flocod, H., T. Férland, and K. Grjotheim, "The Relation Between Concentration and Activity in Molten Mixtures of Salts,’ 2. Anorg. Allgem. Chem., 276, 289 (195L). Blander, M., J. Braunstein, and M. D. Silverman, "Heats of Solution of Sol:ds in Molten Reciprocal Salt Systems,” J. Am. Chem. Soc., 85, 895 (196L4). Blander, M. and E. B. Luchsinger, '"Soluticns of Solids in Molten Reciprocal Salt Systems," J. Am. Chem. Soc., 86, 319 (1964). Blander, M., "Quasi-Lattice Model of Heciprocal Salt Systems. A Generalized Calculation,” J. Chem. FPhys., 34, b32 (1961). Alcock, C. B. and F. D. Richardson, 'Dilute Solutions in Molten Metals and Alloys," Acta Met., 6, 385 (1958), Tupis, C. H. P, and J. F. Elliott, "Generalized Interaction Co- efficients. Part II. Free Energy Terms and the Quasi-Chemical Theory," Acta Met., 14, 1019 (1966). Braunstein, J., M. Blander, and R. M. Lindgren, "The Evaluation of Thermodynamic Association Constants in Solutions with an Application to Molten Salt Solutions," J. Am. Chem. Soc., 84, 1529 (1962). Wagner, C., Thermodynamics of Alloys, Addison-Wesley Publishing Co., Inc., Reading Mass. - London (1952), Chipman, J. and J. F. Elliott, "The Thermodynamics of Liquid Metallic Solutions" in Thermodynamics and Physical Metallurgy American Society for Metals, Cleveland, Chioc (1950) p. 102. Blander, M., F. F. Blankenship, and R. F. Newton, "The Thermo- dynamics of Dilute Solutions of AgNO3 and KCL in Molten KNOj3 from Electromotive Force Measurements. I. Experimental," J. Phys. Chem., 63, 1259 (1959). Hill, D. G., J. Braunstein, and M. Blander, " Electromotive Force Measurements in the System AgNO3-NaC&-NaNO3; and Their Comparison with the Quasi-Lattice Theory," J. Phys. Chem., 64, 1038 (1960). Hill, D. G. and M. Blander, "Electromotive Force Measurements in the System AgNOj3 and NaC% 1in Equimolar NalNO3;-KNOj3 Mixtures and Their Comparison with the Quasi-Lattice Theory," J. Phys. Chem., 65, 1866 (1961). 614 1k, 15. 16. 17, 18. 19. 20. 21, 22, 23. 23a 24, 25. Lewis, G. N, and M. Randall, Thermodynamics, revised by K. S. Pitzer and L. Brewer, McGraw-Hill, N.Y. - Toronto - London (1961). Hagemark, K., "Thermodynamics of Ternary Systems. The Quasi- Chemical Approximation," J. Phys. Chem., T2, 2316 (1968). Blander, M., "The Thermodynamics of Dilute Solutions of AgNO3 and KCZ in Molten KNOj3 from Electromotive Force Measure- ments. II. A Quasi-Lattice Mode," J. Phys. Chem., 63, 1262 (1959). Blander, M. and J. Braunstein, "Quasi-Lattice Model of Molten Reciprocal Salt Systems," Ann. N.Y. Acad. Sciences, 79, Art. 11, 838 (1960). Richardson, F. D., "The Solutions of the Metallurgist - Retrospect and Prospect"” in Physical Chemistry of Process Metallurgy, Part I, G. R. St. Pierre - Editor (1959) pp. 1-26. Blander, M., "Thermodynamic Properties of Molten Salt Solutions" in Molten Salt Chemistry, M. Blander - Editor, Interscience, N.Y. - London (196L) pp. 127-237. Guggenheim, E. A., Mixtures, Oxford University Press, London (1952). Blander, M. and S. J. Yosim, "Conformal Ionic Mixtures," J. Chem. Phys., 39, 2610 {1963). Reiss, H., J. L. Katz, and 0. J. Kleppa, "Theory of the Heats of Mixing of Certain Fused Salts,” J. Chem. Phys., 36, 1uk (1962). Blander, M., and L. E. Topol, "The Topology of Phase Diagrams of Reciprocal Molten Salt Systems," Inorg. Chem., 5, 1641 (1966). Fgriand, T., "Properties of Some Mixtures of Fused Salts," Norg. Tek. Vitenskapsakad., Ser. 2, No. 4 (1957). Blander, M. and L. E. Topol, "Complex Ion Formation and Non- Random Mixing in Molten Reciprocal Salt Solutions," Electro- chim. Acta, 10, 1161 (1965). Belyaev, I. N., "The Formation of Two Liquid Phases in Inor- ganic Systems," Usp. Khim., 29, 899 (1960); Russ. Chem. Rev., 29, 428 (1960). 615 26. 27. 28. 29. 30. 31, 32, Ricei, J. E., "Phase Diagrams of Fused Salts" in Molten Salt Chemistry, M. Blander, editor, Interscience, N.Y. - London (1964) p. 239. Meschel, S. V., J. M. Toguri, and O. J. Kleppa, 'Ternary Excess Enthalpies in Simple Reciprocal Fused-Salt Mixtures. III. Effect of Ionic Size," J. Chem. Phys., 45, 3075 (1966). Kennedy, J. H., "Fused Salt Distribution Studies. The Dis- tribution of T&Br Between KNOj3 and AgBr,”" J. Chem. Eng. Data, 9, 95 (196L). Moore, R. H., "Distribution Coefficients for Certain Actinide and Fission Product Chlorides in the Immiscible Salt System: LiC2-KARCLy ," J. Chem. Eng. Data, 9, 502 (196h}. Morrey, J. R. and R. H. Moore, "Thermodynamic Evidence for Complex Formation by Actinide Elements in Fused KC2-AfCL3 Solvents," J. Phys. Chem., 67, 748 (1963). Reactor Chemistry Division Annual Progress Report, ORNL 2931 (1960) Off. Tech. Serv., Dept. of Commerce, Washington, D.C. Taylor, C. R. and J. Chipman, "Equilibria of Liquid Iron and Simple Basic and Acid Slags in a Rotating Induction Furnace," Trans. AIME, 154, 228 (1943). 616 THE CHEMISTRY AND THERMODYNAMICS OF MOLTEN SALT REACTOR FUELS* C. F. Baes, Jr. Reactor Chemistry Divisicn Oak Ridge National Laboratory Oak Ridge, Tennessee 37830 U. S. A, Abstract The chemical development which has been carried out, largely at Oak Ridge National Laboratory, in support of the Molten Salt Breeder Reactor (MSBR) concept has produced a fairly quantitative description of the chemistry and thermodynamics of actinide, fission-product and structural metal fluorides in molten LiF-BeF, solutions. This information is summarized here (1) in terms of important hetero- geneous chemical equilibria, many of which have been measured direct- ly; (2) in terms of free energies of formation of solutes in molten 0.67 LiF-0.33 BeF, solutions; and {(3) in terms of standard electrode potentials, some of which have been measured directly. This informa- tion may be applied to a variety of important aspects of molten salt reactor chemistry: The extent of corrosion reactions between the container material -~- a nickel base alloy (Hastelloy N) containing Fe, Cr, and Mo -- and the fuel can be specified from hydrogen reduction equilibria. Salt purification procedures can be specified for the removal of structural metal cations, oxides, and sulfides from a knowledge of the hydrofluorination equilibria which are employed. The stability of Th, Pa, and U fluorides toward chemical reduction and toward precipitation a® oxide similarly have been established by measurements of heterogeneous equilibria which include oxide solubility and oxide exchange reactions. The behavior of a large part of the yield of fission products can also be predicted on the basis of these studies and available thermochemical data. Finally, the chemical studies which are reviewed here and the thermo- dynamic information which they have produced suggest the importance of the redox potential, determined by the extent of reduction of UF, to UF3 in the fuel, in controlling many features of MSBR fuel chemistry. *Research sponsored by the U.S. Atomic Energy Commission under contract with Union Carbide Corporation. 617 Introduction The Molten Salt Reactor Experiment (MSRE) has been operating successfully at ORNL since mid 1965. It is a reactor which is fueled by a molten mixture of .65 LiF-.29 BeF, -.05 ZrF, containing <0.01 mole fraction of fissile UF,, first as uranium-235 and more recently as uranium-233. It is moderated by graphite and contained in a nickel-base alloy (Hastelloy N) developed at ORNL for this reactor concept. The principal objective of the MSRE has been to demonstrate the compatibility of this combination of materials in a molten salt breeder reactor (MSBR) as a means to achieve breeding of 233U from 232Th with thermal neutrons(1-3). A molten salt breeder is, perhaps more than any other, a chemist's reactor; the fuel is dissolved and circulated in a relatively unique coolant, the 233pa which is bred must be removed scon after it is formed to avoid the loss of a significant fraction by neutron capture, and finally the fission products produced during operation can be removed con- tinuously to improve the neutron economy. Extensive chemical development therefore preceded the design and construction of the MSRE to establish the means and the extent of materials purification required, the chemical stability and cor- rosiveness of the fuel, the chemistry of uranium therein, and the radiation stability of the fuel. Continuing studies during the operation of the MSRE have given increased attention to the chemistry of the fission products and the means for their removal, and to a method for removing 233Pa on a short cycle. Attention is also being given to methods of on-stream control of fuel chemistry. These extensive studies have been reviewed recently by Grimes(4’5§. In a previous review,(6) I have summarized the equilibrium measurements, and some emf measurements, which had been carried out in LiF-BeF, melts, mostly near the composition 0.67 LiF-0.33 BeF,, as part of this chemical development program. These were combined with available thermodynamic data to provide a list of formation- free energies and electrode potentials in 0.67 LiF-0.33 BeF,. Observed effects of melt composition on solute activity coefficients were also reviewed. Since then, enough new information has appeared to permit a considerable extension and some revision of this com- pilation. This is presented here in much the same form as previously. The purpose of the present review, as previously, is to extend the usefulness of the chemical studies which have been carried out in support of the MSBR program by the systematic application of the methods of equilibrium thermodynamics. This seems especially appropriate and should be especially rewarding here since in molten fluoride systems reaction rates are expected to be relatively high with equilibrium conditions thereby often closely approached. 618 Reactions in 0,67 LiF-0,33 BeF, Table I summarijizes the reactions for which values of the equilib- rium constants are known from direct measurements, or from a com-— bination of other measured equilibria and/or thermodynamic data. In these reactions, and elsewhere, the following notation is used to denote the various states: (c), crystalline solid; (g), gas; (d), dissolved at low concentration in 0.67 LiF-0.33 BeF,; (ss), solid solution. In the majority of reactions, F is the only anion involved and for simplicity the dissolved reactants and products are written in the molecular form, though this is not intended to imply the actual species present in sclution. When anions other than F appear in the reactions, lonic species are written. In some cases this is done to avoid the choice of a molecular «component in solution and in others it is done to indicate that complete dissceiation of amn anion (e.g., 027) and a cation (e.g., Zr"T) is assumed. The concentration scale is the mole fraction; e.g., r, = Pwr,/ wp, * OLiv t Pger,’ Gas pressures are expressed in atmospheres, and, at the low pressures and high temperatures involved, gases are assumed ideal. The stand- ard states for the reactants and products generally can be seen from the form of the equilibrium constant. Here, and elsewhere, the standard state for most solutes is the hypothetical one meole fraction ideal solution in 0.67 LiF-0.33 BeF,. Exceptions are LiF, BeF;, Be2+, i7", and F~. For these major components the solvent composi- tion is taken as the standard state; i.e., aLiF’ aBeFQ’ aBe2+’ aLi+ and a_- all are unity in 0.67 LiF-0.33 BeF,. The values of K in Table I strictly apply only to solutions in the solvent 0.67 LiF- 0.33 BeF, which are sufficiently dilute that Henry's law may be assumed. The effect of changing salt composition and increasing solu- te concentrations will be considered presently. The expression given for the numerical value of the equilibrium constant log K = a + b(103/T) while adequate for evaluating K as a function of T for most purposes, generally should not be taken as a reliable measure of the heat of reaction (AH = -2.3 R(b)) or the entropy of reaction (AS = 2.3 R(a)) since, over the narrow temperature range where a and b usually were determined, there can be large compensating errors in these constants. From the equilibrium data in Table I and from the literature sources indicated, free energies of formation have been compiled for 619 029 Table I. * Reactions in 0.67 LiF-0.33 BeF Log K = a + b (10°/T) (700-1000°K) 2 Hydrofluorination of Oxides 101 HF(g) = HF(d) 102 2HF(g) + 02 (d) = 2F (d) + H ,0(®) 103 HF(g) + OH (d) = F (d) + H,0(g) 104 OH(d) + F (d) = 0°(d) + HF(g) 105 20H7(d) = 0°7¢d) + H.0(g) 106 Be0(c) = Be' ' (d) + 05—(d) 107 2HF(g) + BeO(c) = BeF,(d) + H,0(g) 108 zr0,(c) = 2e(a) + 2 202 () 109 4HF(g) + ZrOZ(c) = Zth(d) + 2H20(g) 1106 Zr02(c) + ZBer(d) = ZrFa(d) + 2Be0(c) 111 4HF(g) + ThO,(c) = ThF,(d) + 2H,0(g) 112 4HF(g) + UO,(c) == UF,{(d) + 2H,0(g) Hydrofluorination of Sulfide and Iodide 113 2HF{g) + §27 (@) = 2F (d) + H,5(g) 114 HF(g) + I (d) == F (d) + HI(g) Reduceable Metals 115 Hz(g) + Cer(d) = Cr(c) + 2HF(g) 116 2HF(g) + %‘Hz(g) + %Cr 04(c) == CrF,(d) + JH,0{g) 117 H,(g) + FeF,(d) =* Fe(c) + 2HF(g) 118 Fel“z(c) - I‘er(d) 119 Fe0(c) <= Fe2* (d) + 0°7(d) 120 FeO(c) + BeF3(d) == FeFz(d)+ Be0(c) K** . a b UT—-OK K Source Ky /(B - 5.17 1.31 0.02 12 (1’H O)I(P ) (X52-) - 4.01 8.42 0.08 107-106 O)I(P P Eop™) 1.04 2.09 0.06 14 (P )(xog_)/(xofl—) 5.05 -6.34 0.1 103-102 2 - — (Pnzo) (on-)/(on_) 6.09 4.25 0.14 2(103)-102 X02- - 0.39 -2.63 0.08 15 )/(P ) - 4.40 5.80 0.02 14 (x 4+) (xoz ) - 3.06 -5.94 0.l6 15 (e 0) (xzfl, 2/ (Py ) -11.08 19.90 0.2 2(102)+108 X, ¥, - 2.29 -0.69 0.2 108-2(106) (e l{20) (x )/(P ) -10.48 12.27 0.2 112+136 r 1y 0) (xUF ?/(PHF) -10.48 9.90 0.2 109-135 2 ° K 1 (Pst)/(PHF) (X52-) Log K(873°K) 18 (B F (R (X 2) 0.39 2,08 0.1 19 . -9.06 0.06 20 /) /(P )(XCIFZ) 5.12 -9 20 P32y (X _j‘zQLE“’_Z/z - 2,66 -1.17 0.1 2(202)-3(210)-115 (B y )1 . -5.31 0. 20 ) /(sz)(xFeFZ) 5.20 5 02 20 FeF 2.45 -3.05 0.01 20 (Kp, 2+) (X,2-) -0.73 -3.91 0.1 202-212-102-117 Xop - 0.3 -1.28 0.07 119-106 €2 129 ' W Table I (Cont'd.) *k Reduceable Metals K a b % og K Sourcet 121 By(g) + NLF,(d) = NL(C) + ZHF(g) B/ By ) By ) 8.3 -3.60 0.04 20 122 MiF,(c) == NiF,(d) XouF 0.30 -~2.07 0.01 20 123 N10(e) = M2t (@) + 077 () (xmzi) (x52-) -2.76 -~4.20 0.1 202-214-102-121 124 N10(c) + BeF,(d) = NiF,(d) + BeO(c) XuiF - 2.37 -~1.48 0.14 123-106 125 —gflz(g) + M0F3(d)= Mo(c) + 3FHF(g) (1’111,)?‘/(PHZ):”Z(KMoF ) Log K(769°K) 1.0 21, 1214122 126 2H,(g) + Mol (c) = Molc) + 2H,0(g) (B0 1 (Pr1p) ? 2.56 -3.61 0.7 2(202)-217 127 %Hz(g) + NbFg(g) = Nb(c) + SHF(g) (PHF)SI(PHZ)”Z(PMFS) 12.99 -19.83 0.7 5(201)~-216 Reactions of Silica 128 510,(c) + 4F(d) = 2 02'(d) + SiF,(g) (P51p4)(x02_)2 3.06 -10.93 0.4 2(202-102)-4(201)+ 129 SiF,(g) == S1F,(d) Kgyp) / (Psyp,) Log K(793°K) = - 3.0 20° i_(e)i_g 130 2540,(c) + 2BeF(d) = Bey$i0,(c) + SiF,(g) PsiF, 6.28 -7.93 0.02 28 11 2es10, () = 3510,(c) + Be' (&) + 027(d) Xo2- - 161 -Lso 0.2 (128130 Actinides 132 moz(c)-—:m‘“’(d) + 2 02'(d) ry44) ("02“)2 ~- 2,46 -4.57 0.2 133+136 133 10,(e) = U@ + 2 027 (a) Kya) (K2-)° - 2,46 -6.95 0.2 108-135 134 Uoz(fi) + ZBer(d) = UF,(d) + 2Re0{c) XL'F4 ~1.69 -1.70 0.24 133-2(106) 135 2r0z(e) + UF,(d) < ZxF,(d} + U0z(c) (erFh)/(xUFls) -~ 0.60 1.01 0.08 29 (Rygy oq) K, 136 ThO,(ss) + UF,(d) = ThF,(d) + U0,(ss) n 0.00 2.38 0.04 11 2 4 4 2 (Xtno, ’SS) Kyp,) 137 S + UF, (d) = UF,(d) + HF(g) @, )& /Py ) %ge,) 407 -9.33 0.02 10 272 4 3 HF “'UFy VT Hy UF4 1 138 UF4(c) == UF4(d) Xypq 1.99 -4.03 0.04 33 139 PuF4(c) = PuF,(d) xPuF3 1.32 -3.15 0.02 g 140 1, 0 (c) + 2BeF,(d) = PuF,(d) + Be0(c) X - 2.83 -1.11 1.4 =(202-107)-3(201)+ 227273 27772 3 2 PuF, 311-1/2(232) 229 Tab le I {Cont'd.) x* P, is in atmospheres; X X [ K** a b a K Sourcei Lanthanides Log 141 LaF3(c)== Lan(d) XLaF3 1.58 -3.38 .02 8 142 CeF3(c)== CeFa(d) XCeF 1.64 -3.38 .02 8 . 3 143 NdF3(c)-— NdF3(d) XNdF3 0.95 -2.59 .02 36 144 SmF._(c) == SmF. (d) X 0.81 -2.37 .02 36 3 3 SmF3 - * The Notations (c), (d), (g) and (ss) indicate, respectively, the solid, dissolved, gaseous, and solid solution states. i i is mole fraction and, for 0.67 LiF-0.33 Ber, is equal to (moles of i/kg salt/30.03). Underscored numbers are reference citations; other numbers refer to entries in this Table {(101-144), in Table II (201-233), in Table ITI (301-320), or in Table IV (401-441). pure components (AGf) in Table II and for dissolved components (Aaf) in 0.67 LiF - 0.33 BeF, solutions in Table III. An alternative, and very useful means of summarizing the avail- able information has proven to be electrode potentials for half cell reactions. These, listed in Table IV, are referred to the half- cell reaction (429) HF(g) + e - F (d) + 1/2 Hy(g), E° =0 Thus the potential for each half-cell reaction in Table IV corre- sponds to AG (= - nFE°) for the complete reaction obtained by com- bining the given half-cell reaction with half-cell reaction 429 above. (In the text the various reactions, and AGY, AGE, and E° values all will be referred to by numbers which identify the entries in the various tables.) As usual, if two half-cell reactions are combined to give a complete reaction, their potentials are simply combined algebraically to give E° (and hence AG°) for the complete reactions. If, instead, two half-cell reactions are combined to give a third half-cell reaction, the volt-equivalents (nE°) must be combined algebraically to obtain the volt-equivalents of the third half-cell reaction. Activity Coefficients In several of the investigations cited in Table I, the effect of melt composition on the equilibrium under study.was determined. The observed effects are presented in Fig. 1 as activity coef- ficient curves, For solutes the activity coefficients are unity at the reference composition; for the solvent components at this com- position they are vy F = 1.5 and YBeF, - 3. The various curves were derived as follows: 2 LiF,BeF;. The activity coefficients of these components were recently re-determined by Hitch and Baes(7), using the cell Be|LiF, BeF,|H,, HF,Pt for which the cell reaction was taken to be Be(c) + 2HF(g) - BeF,(d) + Hy(g) (The potentials observed at 0.33 BeF, in this cell gave directly the E° values for the Bett/Be couple (407, Table 1V), CeF3,PuF3. These curves are based on the observed variation of the solubility of the tri-fluorides with melt composition reported by Ward et al{8) and by Barton(9), NiF,. This curve is an estimate based on the observations that the solubility of NiF, varies less with composition in NaF-ZrF, melts than does the solubility of CeFj. UF,,ThF,. The variation of Yyp, ¥as estimated from the variation ——— 4 623 ¥29 Table II. Formation Free Energies of Halides and Chalcogenides a6t = a+ b (1/10%) kcal/mole (700-1000°K) Comp'd. a b %% Source Comp'd. a b I Source” 201 HF(g) -65.19 - 1.01 0.4 26 218 MoF (g) -370.88 69.59 0.7 40 202 HZO(g) -59.07 13.03 0.1 26 219 RuFS(g) -213,41 43.0 2 ZE 203 H,S(g) -21.45 11.61 0.2 26 220 LaF4(c) -416.06 57.16 2 35 204 HI(g) - 1.56 - 1,79 0.1 26 221 CeFy(c) -417.83 57.59 2 24,35 205 LiF(c) -146.50 23,11 0.7 26 222 NdF,(c) -394.02 56.58 2 35 206 BelO(c) -146.,02 24.94 0.24 -407-107 + 202 223 SmFS(C) -398.42 56.58 2.5 35 207 CF4(g) -223.30 36.54 0.5 29_ 224 TaFS(g) -455.1 44 .0 2 24 208 SiF4(g) -386.26 34.71 0.3 26 225 WF6(g] -418.50 68.04 5 26 209 Si0,(c) -216.55 42.10 0.5 26 : 226 Tho,(c) -292.40 44.55 1 24 210 Cr04(c) -270.79 6l.64 0.5 21 | 227 UF (<) -356.48 49,36 2 309-138 211 Fer(C) -168.62 32.98 0.9 313-118 } 228 UFd(c) ~452.0 67.4 2.5 gi 212 Fe0(c) -62.71 15.15 0.25 21 : 229 UOZ{C) -258.0 40.0 0.8 41 213 Nin(c) -156.33 37.65 0.9 314-122 ' 230 UFG(g) -509.92 65,04 0.6 41 214 NiO(c) -56.26 20.35 0.2 21. 231 PUFS(C) -370.0 57.5 4 34 215 ZrOz(c) -260.44 44.44 0.4 26 232 Pu203(c] -427.84 61.82 5 éi 216 NbFS(g) -416.70 54.40 2.5 gé 233 Pu02(c) -252.67 45.90 2 34 217 MOOZ(C) -134.65 37.75 3 21 * Underscored numbers are reference citations; other numbers refer to entries Table I (101-144), in Table III (301-320), or in Table IV (401-441). in this table (201-233}, in Table II1. Formaticn Free Lnergies for Selutes in 0.67 LiF-0.33 BeF — 3 2 AGE = @ + b (1/10%) keal/mole (700-1000°K) Solute* a b OAEE Source™* 01wt P —161.79 16.58 0.8 205, 7 02 La”t 4+ 3F -400.58 49.93 2 141 + 220 303 ot + I 402,35 50.01 2 142 + 221 046 NTT 4 I -382.19 54,22 2 143 + 222 305 Smot 4+ 30 -387.59 52.87 2.5 144 + 223 306 Belt + 2F —243.86 30.01 0.8 - 407 + 2(201) 307zt 4 4F —452.96 65.05 2 109+215+4(201) - 4t - 2(202) 08 ™+ 4F —491.19 62.40 2 226+136+310-229 309 LT+ 3F 338,06 40.26 2 310 + 137 - 201 310 ¥t 4 arT —445.92 57.85 2 134 M 3%362;206) 31 put + 3 -355.59 51.45 4 139 + 231 312 cett 4 2r S171.82 21.41 0.9 -1I5 + 2(201) 313 Felt + 2F -154.60 21.78 0.9 _117 + 2¢201) 34 W2+ 2F —146.87 36.27 0.9 -121 + 2(201) 315 Moot + 3F AGT(769°K)=-194.3 1.5 -125 + 3(201) 316 H + F - 71.20 22,65 00 101 + 201 317 Be™ 4 0% L1301 26.70 0.4 106 + 206 318 et + 527 Agf(s73%) <79 ~407 + 203 - 113 319 Be?t + 206 —212.53 67.58 0.6 —407 + 2(202-103) 320 BelT 4+ 217 - 97.56 24.89 0.8 -407 + 2(204-114) * The standard state of the ion 1s the hypothetical mole fraction solution in 0.67 LiF-0.33 BeF,, with the exception of Li+, Bel* and F~, for which the standard state is the solvent itself. *% Underscored numbers are reference citations; other numbers refer to entries in this Table (301-320), in Table I (101-144), in Table II (201-233), or in Table IV (401-441). 625 329 Table IV. Electrode Potentials vs E° = a + b(r/10%) HF/Haz, F~ in 0.67 LiF-0.33 BeF, (700-1000°K) Half Cell Reactilon a b EUiOOOOK Uy Source# 401 L1t + e- — Li(e) -3.322 Q.763 -2.559 Q.04 201-301 402 Ce3* + 3e” — Ce(c) -2.989 0.767 -2.222 0.034 3(201)-303 403 1a?t + 3e” — Ia(c) -2.963 (0.766 -2.198 0.034 3(201)-302 w04 am®t 4+ 3¢ — sm(c) -2.932 0.808 -2.124 0.04 3(201)-305 405 Na?* + 3e” — Nd(e) -2.697 0.828 -1.870 0.034 3(201)-304 406 Th*t + 4e~ — Th(c) -2.498 0.720 -1.778 0.03 4(201)-308 407 Be?* + 2e” — Be(c) -2.460 0.694 -1.765 0.002 7 408 Pult + 3e” — Pule) -2.313 0.788 -1.525 0.06 3(201)-311 409 Ut 4+ 3e” — U(e) -2.059 0.626 -1.433 0.03 3(201)-309 410 Ut 4+ 4de” — U(c) -2.007 0.671 -1.336 0.015 4(201)-310 411 Zr*t + 4e” — Zr(e) -2.084 0.749 -1.335 0.012 4(201)-307 412 U%t v em — U -1.851 0.807 -1.,045 0.004 137 413 SiF,(g)+4e™— 51(c) + 4F° -1.360 0.420 -0.940 0.02 4(201)-208 414 TaFs(g)+5e™—~ Talc) + 5F -1.120 0.425 -0.695 0.024 5(261)-224 415 3Cr,03(c) + 3Bet" + 3e—Cr{c)+2Be0) -1.251 0.599 -0.652 0.007 3(201)+2 (206-306) -3 (210) 416 Hy0(g) + e” — 3H,(g) + OH- -0.414 -0.206 -0.620 0.012 104 417 HI{g) + &= — IH.{(g)} + I~ -0.413 -0.077 -0.49%0 0.016 -114 418 H,0(g) + 2¢” — H,(g) + 02 -0.835 0.398 -0.437 0.009 -102 419 Cr*t + 2e- — Cr(e) -0.898 ©.508 -0.390 0.007 115 420 Ho8{g) + 2e~ — Ho(g) + S°- EC (873°K)<-0.346 -113 421 $I,(g) + e — I° -0.345 0.001 -0.344 0.016 204-114 422 WoFs(g) + 5e — Nble) + 5F° -0.787 0.516 -0.272 0.03 127 423 OH™ + e~ — 1Hp{g) +0%- -1.257 1.002 -0.255 0.015 104 LZg Table IV (continued) Half Cell Reaction a fel B000% “x Source™ 424 MoOp(c) + 2Be?t+ 4e™—Mo(c) + 2Be0(c) -0.754 0.563 -0.191 0.03 2(202-107)-217 425 3S5,(g) + 2e~ — 87" E°(873°K)<-0. 101 203-113 426 I(g) + e” — I 0.451 -0.544 -0.094 0.016 421, 26 427 4HF(g) + C(c) + 4e~ — CH,(g) + 4F 0.228 -0.278 -0.050 0.002 26 428 Fe®t + 2e — Felc) -0.527 0.516 -0.011 0.002 117 429 HF(g) + e~ — 3Ho(g) + F~ 0 0 0 430 Mo?t + 3e” = Mo(e) FP(769°K) 0.053 125 431 Nio(e) + Be?t + 2e™—Ni(e) + Be0(c) -0.514 0.595 0.081 0.001 202-214-107 432 WiF,(c) + 2e~ — Ni(c) + 2F- -0.563 0.860 0.298 0.004 121 + 122 433 WFg(g) + 66~ — W(c) + 6F -0.19¢8 0.536 0.338 0.04 6(201)-225 434 oY+ e” — JH () -0.261 1.026 0.765 0.004 101 435 Ni?t o+ 260 — Ni(e) -0.357 0.830 0.473 0.004 121 436 30,(g) + 2¢” — 02~ 0.445 0.116 0.561 0.009 202-102 437 MoFg(g) + 6e” — Mo(c) + 6F" 0.147 0.547 0.693 0.02 6(201)-218 438 CF (g) + 4e~ — C(c) + 4F- 0.406 0.440 0.846 0,02 4(201)-207 439 UFg(g) + 2e~ — U*T + 6F” 1.439 -0.200 1.240 0.05 -410-230 440 RuFs(g) + 5e~ — Rulc) + 5F° 0.976 0.417 1.393 0.024 5(201)-219 441 3P, (g) + &= — FT 2.827 0.044 2.871 0.017 201 *Underscored numbers are reference citations; other numbers refer to entries in this table (401-441), Table I (101-144), Table II (201-233), or Table III (301-320). 40 Z/I’ = 65 30 / 0337 Be 2 20 i L] o 3 | %, ~ I 15 | k A 5 5 I Th, o The — & I ”ncg:nfis m g | © NiFp W 2.90 > L.O = > = N 280 Q I f *fi\\\\\\\\ 06 V \K.GTTU; 1.47 Q. 40.30 0.35 0.40 045 " "MOLE FRACTION BOF2 1. Activity Coefficient Variation With Composition in LiF-BeF, Melts. 628 of K reported by Long and Blankenship(lo) for the reduction of UF, to UF3 at various melt compositions, with the assumption that YUFg varies in the same way as vy . The resulting curve for y is PuF UF assumed to represent YTh .y as well since, in a recent study by Bamberger et al(1l) of the exchange of UYt and Th“* between a molten fluoride solution and an oxide solid solution (136), K showed no dependence on the melt composition, thus suggesting YTthlYUFu = 1. HF. This curve is based on the HF solubility measurements (101) of Field and Shaffer(12), The following examples illustrate the use of these activity coef- ficients in calculating equilibrium quotients at compositions other than the reference composition (122) NiF,(c) 2 NiF,(d) Xyir, = K122/Vyyp, (107) 2HF(g) + BeO(c) 7 BeF,(d) + H20(g) P ?%29)2 - K107/XBeF2YBeF2 HF The activity coefficients curves in Fig. 1 have been used in a few instances to make small corrections in measurements reported at salt compositions other than 0.67 LiF-0.33 BeF,. It is evident from the figure that activity coefficients generally do not show large variations and hence these corrections were small. It has been noted previously(6) that the Yyp Curves (Fig. 1) fall between the curves for 0.67 YLiF and 0,33 YBe?z' Thelr location within the envelope defined by these two curves, except for Yyp* depends upon the ratio of charge to radius (z/r) for the cation M2+. As yet, insufficilent data are available to decide whether some analogous correlation exists in MSBR fuel solvents where ThF, as well as LiF and BeF; is present as a major component. Studies such as the recent solubility measurements of Barton et al(l3), however, should soon improve our knowledge of medium effects in such three-component solvents. Hydrofluorination and Oxide Chemistry Oxide is a commonly occcurring impurity which enters LiF-BeF; melts by reactions of water vapor to form OH and 02~ (the reverse of 102 and 103) or by the dissolution of structural metal oxides. Since BeO, ThO;, U0, and certain other oxides are sparingly soluble in LiF~BeF, melts, the chemistry of oxide and its removal has been 629 of considerable interest. Oxide is removed from these melts by treatment with an HF-Hj; mixture which reverses the hydrolysis reaction and evolves water vapor. The equilibria involved (102-109) have been studied b{ Mathews and Baes(14) ip LiF-BeF, melts and by Hitch and Baes (15) in LiF-BeF,-ZrF, melts. The results show that the removal of oxide by hydrofluorination can be made quantitative 16), Indeed, this treatment was used to determine oxide, as evolved water, in measure- ments of the solubility of BeO and Zr02(15). More recently it has been used in a method developed by Apple g£'§£(17) for the remote determination of oxide (typically 60 ppm) in MSRE fuel samples. The formation of OH during hydrofluorination impedes the removal of oxide and, moreover, if this hydroxide is incompletely removed during the treatment, it will subsequently decompose slowly to produce 02~ and HF (104) or H,0 (105). Thus hydroxide is an oxidizing impurity (423). Since HF is not very soluble in LiF-BeF, melts (101), the presence of OH™ in small amounts after hydrofluorina- tion is easily detected by sparging with a dry inert gas to deter- mine if more HF is evolved than would be expected from dissolved HF alone. Hydrofluorination of Sulfide and Iodide Hydrofluorination as a purification treatment also removes sulfide which can arise from the reduction of any sulfate impurities present in the raw materials. Its removal is important because of the sensitivity of nickel and nickel-base alloys, used for con- tainment, to sulfide embrittlement. Only the lower limit of the equilibrium constant for the formation of H,S (113) could be deter- mined by Stone and Baes in their measurements because of reaction of Hy;S with the container after it was evolved from the melt. Preliminary measurements have been made by Freasier.gg_gl(lg) of the conversion of I~ to HI by HF (114). The results show that iodide is the stable state of iodine under mildly reducing conditionms, and that it is removed by hydrofluorination. This information is useful in predicting the behavior of fission product iodine in a reactor fuel. 1In addition, it suggests a possible method of removing 6.7 hr 1351 from the fuel before large fractions decay to 135Xe, the most significant neutron poison of all the fission products. Structural and Noble Metals The hydrogen which accompanies and follows hydrofluorination in the usual purification treatment reduces a number of metal fluoride impurities. The reduction equilibria in the case of the fluorides CrF, (115), FeF, (117), and NiF, (121) have been studied carefully by Blood(zo), who found them to be increasingly easy to reduce in 630 that order. As a result of its nobility in MSR fluorides, nickel is widely used for their containment. The alloy used for the MSRE is a nickel-base alloy containing 7% Cr, 5% Fe and 167% Mo (Hastelloy N) which is very nearly inert to the fuel salt under normal reactor conditions. This is not because of favorable corrosion kinetics or because of a protective film of corrosion products, but simply because, in the absence of oxidizing impurities, there are no thermodynamically favored chemical reactions (Fig. 2); i.e., the container metal is essentially at equilibrium with the fuel salt. Blood also measured the solubilities of FeF, (118) and NiF, (122), which are only sparingly soluble in LiF-BeF,; melts. In addition, it may be calculated (119, 123) from the previously cited reactions and available thermochemical data (21} that FeO and NiO, when present as saturating solids, give still lower concentrations of Fet* and Nt Thus at equilibrium with both Fe0® and Be0, 0,67 LiF-0.33 BeF, should contain ~1.5 m/o FeF, at 600°C (120); at equilibrium with NiO and Be0 it should contain 0,007 m/o NiF, (124). Analogous calculations suggest that Cr;03 is in a sense also an Yinsoluble" oxide since its hydrofluorination and reduction to CrF, should not occur readily (116). Thus, at equilibrium with 0.1 atm HF, 1 atm H,, and Cr,03(c), the melt should contain an amount of CrF, at 600°C given by 3/2 -7 Xepp, ¥ % 10 /(PHZO) Preliminary emf measurements of Meyer gg_gl(zz) for the cell Mo |MoF4(d), LiF,BeF;||LiF,BeF,|N1iF, (c),Ni for which the cell reaction MoF3(d) + 3/2 Ni(c) =~ Mo(e) + 3/2 NiF,(e) was assumed, suggest that MoF; lies between FeF, and NiF; in ease of reduction. MoO, should be another "“imsoluble' or resistant oxide, since its reduction to the metal by hydrogen (126) is, in effect, more difficult than the reduction of dissolved MoFj. In the absence of sufficient knowledge about the identity and stability of lower fluorides of niobium, tantalum, ruthenium, or tungsten, estimates of the minimum reactivity of these metals were obtained from available thermochemical data for NbF5(23), TaF5(24), RuFg 25) and WFg 26), Half cell reactions involving these metals, and chromium, iron, and nickel are listed in Table 1IV. The fol- lowing electrochemical series and E° values result at 1000°K: 631 2£9 MFy MOLE FRACTION -2| 0.003 Mole fraction total U T T T T —— —Equil. with Cr, Fe,Mo,Ni / ~ — Equil. with Hastelloy N / - 600°C / - - UF, \ / CrE 2. Variation of Equilibrium Concentration of Structural Metal Fluorides and the Distribution of Iodine as a Function of the UF,/UF3 Ratio in an MSR Fuel. CONCENTRATION RATIO - + TaFs5/Ta,F~ =0.695v Mo3" /Mo ~.05v at 769°K cr2t/cr ~0.390v WFg/W,F~ 0.338v NbFs/Nb,F~ -0.272v NiFP/ND 0.473v Fett/Fe -0,011v RuFs/Ru,F- 1.393v The relative nobility of nickel and the surprising reactivity of tantalum, because of the stability of TaFg, are noteworthy. Reactions of $iC; Molten fluorides are widely considered difficult to contain with- out corrosion of, or contamination by, the container. This unsavory reputation probably results from the tendency of fluorides to attack glass and silica, thus preventing the study of fluorides with the ease and convenience afforded by these materials. As we have seen, this reputation is undeserved in the case of LiF-BeF, melts, which are readily contained in metals under mildly reducing conditions. Furthermore, it may be predicted from the previously described studies of the hydroflucrination of oxide and from AGE for S10, and SiF, that the reaction with 5i0, to produce SiF, (128) is quite limited. This has been confirmed by Bamberger et al(27,28) who found that if the SiF, is confined within the system, or if it is introduced with the cover gas, then vitreous Si0, is essentially non—contaminating and quite durable, save for crystallization, at temperatures to 700°C. If the SiF, produced by reaction 128 is swept away, its pressure falls and the oxide ion concentration increases in the melt until are fixed (131) by the precipitation of a new solid phase, Be,SiOy. Actindides At a few mole percent of ThF, in 0.67 LiF-0.33 BeF, (the exact amount being uncertain because of unknown activity coefficient effects) ThO, becomes the least soluble oxide., U0, is considerably less soluble and becomes the least soluble oxide at <0.1 m/fo (134). Zr0; replaces U0, as the least soluble oxide phase when the XZrFu/ XUFI+ ratio exceeds a value in the range 1.5-6, depending on tem- (29) perature (135) and ZrF, concentration . Hence ZrF, was added to the MSRE as an oxide getter to prevent the precipitation of UO, in the reactor. Romberger et al, found that Zr0O, does not dissolve significant amounts of UDZ(§U). Since ThO, and U0, form a continuous serles of solid solutions, Bamberger et_gk(ll) were led to study the exchange equilibrium of Th** and U"T between (U-Th)0, solid solutions and an LiF~BeF, melt containing UF, and ThF, (136). ©UO, is considerably less soluble than ThO, in the fluoride phase and so U** strongly distributes to the oxide solid-solution phase. The results of this study have 633 permitted an improved estimate of Aaf for ThF,(d), as indicated in Table III (308). Measurements of the reduction and extraction of protactinium into molten bismuth by Ross et al(31) provides strong evidence that protactinium is present as PaF, in LiF-BeF, melts under reducing conditions. While as yet there are insufficient data to estimate quantitatively its stability, PaF, is more easily reduced into bismuth than is ThF, and less easily reduced than UF3, hence we may estimate Aéf(Pan(d)) nv ~412 kcal/mole (1000°K) Early measurements by Shaffer et qi_(32) showed that protactinium is precipitated from molten fluorides by oxides. The oxide of pro- tactinium which is precipitated has not been determined as yet. Long and Blankenship(lo) studied the partial reduction of UF, to UF3 by hydrogen, both in the solids and in molten fluorides (137). In an MSR fuel, as we shall see, the U“Y/U3t redox couple can be used as a redox buffer and indicator to establish and control the redox potential., From the measurements of Jennings et a1(3 ) the solubility of UF3 is quite limited in 0.67 LiF-0.33 Ber, being only 0.2 m/o at 600°C (138). Barton(9) has measured the solubility of the sparingly soluble PuF3 (140), and it appears that this is the stable form of plutonium under mildly reducing conditions. From these results, and available thermochemical estimates for PuFj(c) and Pu,03 34) , we may estimate, though with considerable uncertainty, the solubility of Pu;03(140). Lanthanides As indicated by the estimates of AGf for LaF3(c), CeF3(c), NdFz(c), and SmF3(c) (Table II, 220-~223) the trivalent lanthanides are among the most stable of fluorides. These values are based on recent measurements of AHf from fluorine bomb calorimetry and other estimated AHE values, all by Rudzitis and VanDeventer(35) and on ASf values from the reported ASf values for CeF3(24) and the lanthanide oxides(35) and the assumption that the difference [ASf(LnF3) - % ASf(anoa)] is a constant for all the lanthanides. Since the limited solubilities of these fluorides have been measured by Ward et al 8) and Doss et al( 6) the AGE values can also be calculated (302-305, Table III). The resulting E° values place the Ln? */Ln couples second only to Li*/Li in reducing power (402-405, Table 1IV). 634 MSRE Fuel Chemistry The chemistry of a single fluid MSBR fuel salt, which should approximate the composition ©.72 LiF-0,16 BeF,-0.12 ThF¥,, probably will not differ greatly from the chemistry of the melts described above, provided that large departures from equilibrium conditions are not produced by the radiation fluxes of an operating reactor, Effect of Redox Potential A large fraction of the chemical reactions which have been described are redox reactions, as is attested by the list of half- cell reactions in Table IV. The degree of oxidation ¢f each of these couples when present in a fuel salt at equilibrium is deter- mined by the overall state of oxidation of the system, or, more briefly, by the redox potential. The most convenient measure of the redox potential is the ratio XUF /XUF in the fuel, since this couple can act as a buffer; i.e., b 3 with a significant fraction of both UF, and UF3 present, the amount of each can considerably exceed the amount of any other oxidizable or reduceable substance in the system. Thus any condition which might otherwise alter considerably the amounts of these other oxi- dizable or reduceable substances will instead merely change slightly the ratio of UF, to UFj3, and so only slightly change the redox potential. In Fig. 2, wherein the concentrations of the structural metal ions Cr2+, Fe2+, Mo3+, and Ni?t are plotted as a function of the UF,/UFy ratio, we see that when the amount of reduced uranium exceeds 17 of the total uranium, this buffer effect appears. The oxidized metals all become rapidly insignificant in concentration compared to the increasing UF; concentration. Not only is corrosion of the container alloy thus arrested completely, but any oxidant that enters the system will oxidize the UF; rather than the container. Two curves are shown for each metal-ion reaction, indicating the equilibrium of the metal ion with the pure metal (dashed lines) or with Hastelloy N (solid lines). The latter has been assumed to be an ideal metal solution. The assumption that such an alloy phase exists at equilibrium witn the fuel is no doubt a better approxi- mation of the situation in a reactor than is the assumption of the separate pure metal phases, In either case, the conclusion to be drawn is essentially the same, namely, that with more than 1% reduction of the UF, to UF3 in the fuel, there should be no signifi- cant corrosicen. Four years of operating experience with the MSRE supports this conclusion. It is probable that in MSBR fuels, with a few tenths of a mole % total uranium present, the UF,/UF3 ratio will be maintained in the range 10-100, the upper limit being set by the oxidation of 635 container-chromium and the lower limit being set by the approach to UF3 saturation. Included in Fig. 2 is a curve representing the distribution, as a unitless concentration ratio, of (fission-product) iodine between the fuel salt and a gas phase. It is seen that the iodine should be present as I in the fuel salt under normal reducing conditioms. Indeed relatively strong oxidizing conditions (XUFH/XUF3 >108) are necessary to produce significant amounts of iodine in the gas phase. Figure 3 indicates the conditions under which the noble-metal fission products should appear in their upper, volatile oxidation states. While these elements may be at a high state of oxidation during the instant after thelr birth, at equilibrium in a system with a normal UF,/UF; ratio, the only such volatile fission product we have any reason to expect is NbFg, and this should not be detected in the off-gas system until the UF,/UF3 ratio exceeds 10“. The volatile fluorides MoFg and PuFs should not appear until the system is so oxidizing that C¥, in one case, or UFg in the other, also appears. Included in Fig. 3 are curves for TaFg and WFg which are not fission products. Operating experience with the MSRE has shown that niobium activity, while not found in the fuel under normally reducing con- ditions, does appear in solution under more oxidizing conditions. While the UF,/UF3 ratio at which this appearance of niobium in solution occurs is not yet accurately known, its behavior is con- sistent with the calculated curve for NbFg in Fig. 3. Alternatively, a soluble lower valence state of nlobium may be involved 37), 1n any case, it seems possible that fission product niobium will serve as a useful redox indicator. Anomalous Behavicr of Noble Metal Fission Products In the MSRE, the noble-metal fission products just discussed -- Nb, Mo, and Ru -~ have been found to behave in a curious and intriguing fashion(4), Large fractions of the expected yield of these activities have been found to concentrate in the region of the salt-gas interface in the pump bowl. When this was first observed, there was considerable temptation to speculate that these metals concentrated there because they were oxidized to their upper- valent, volatile fluorides. But as can be seen from Fig. 3 and Table IV, except for NbFg, they are themselves strong oxidizing agents which should be reduced quickly by the UF3 in the fuel or by the container metal. Rather, at the present writing, it appears far more likely that the mechanism of this concentration effect involves finely divided metallic particles of these fission products which are trapped at the salt-gas interface because they are not wetted by the salt. 636 LE9 m.) t PRESSURE (a Ty =T 1 .~ 1 - - CF, ) MoFg -/ (A5 // 1:(§\ _lt) i 7 10'° e e N / RuF5 . — |o"'l5 600°C n ‘O-rZO 5 1 L 1 1 i 10 x 10'° 10'® X, ur, / "UF, 3. Variation of Partial Pressure of Volatile Fluorides as a Function of UF,/UF3; Ratio in an MSR Fuel. CONCENTRATION RATIO Since it is necessary to sparge an MSBR fuel salt vigorously with an inert gas to remove 135Xe, it will be important to determine if and how efficiently these noble metal fission products can be swept away at the same time. Hydrogen Couples Figure 4 indicates the behavior of a number of redox couples which involve hydrogen. While it is unlikely that a reactor would be operated under a hydrogen atmosphere, no doubt there has been appreciable amounts of hydrogen in the MSRE from the decomposition of traces of lubricant o0il which entered the high temperature region of the circulating pump. More important, in an MSBR it is possible that an HF-H, gas treatment will be desirable at some point to control the UF,/UFj3 ratio, or to remove oxide or iodide. With the conditions as specified in Fig. 4 (1 atm hydrogen at 600°C) we see that all the couples shown are reduced under a wide range of the UF,/UF3 ratio, with the reduced species being in the fuel salt. Then, if hydrofluorination is to be used as a means of removing oxide or iodide, it must be done under conditions which are more oxidizing than normal. Hence such a treatment should be followed by re-reduction of the fuel salt., If hydrogen is allowed to reach the graphite moderator, some hydrocarbons may be formed. Chemical Consequences of Fission The expected behavior of the important fission products in a molten salt reactor is indicated in Table V. The noble gases Xe Table V. Chemical Consequences of Fission in an MSBR Fission Product Assumed Eq. Ox. State Yield* vz (Z) (Y) Br + I -1 0.015 - 0.015 Kr + Xe 0 0.606% 0 Rb + Cs +1 0.004 0.004 Sr + Ba + 2 0.072 0.144 Lanthanides + Y + 3 0.538 1.644 Zr + 4 0.318 1.272 Nb 0 0.014 0 Mo 0 0.201 0 Te 0 0.059 0 Ru 0 0.126 0 1.953 3.049 *From Ref. 39. **With rapid stripping from the system. 638 6€9 CONCENTRATION RATIO HF )/ HF () NI @] T s \\ ®) XL 3' A S. 4. Partial Pressure and Distribution of Protonated Species as a Function of UF,/UF3 Ratio in an MSR Fuel. PRESSURE (atm) and Kr are known to be quite insoluble in molten fluorides(38). The Groups VII, I, II, III and IV fission products should be dissolved in the fuel salt in their normal valences. The remainder —- of which Nb, Mo, Tc and Ru are the Important contributors -- are expected to be reduced to the metallic state under the reducing conditions which would normally be maintained in a reactor. Also indicated in Table V are values of the product of the fission yield(39) and the valence of each fission product. This serves to show that with reducing conditions maintained and rapid removal of Xe and Kr, the sum of the charges on all the fission products is less than the +4 per mole of uranium burned, being nearly +3. Hence as burnup of the uranium fuel (mostly UF,) proceeds, UF; or other reducing agent must be added to maintain reducing conditions. Otherwise, the deficiency of cation equivalents will be made up by the oxidation of UF; present and then by corrosion of the container; i.e., the fission process is oxidizing. Conclusion It is fair to say, I think, that as a result of the supporting chemical studies which have accompanied the development of the Molten Salt Breeder Reactor concept, a sound understanding of the chemistry of such molten fluoride fueled reactors has been achieved. The most significant conclusion to be drawn is that such a reactor system -- with partially reduced UF, in an LiF-BeF,-ThF, solvent, contained in Hastelloy N, and moderated by graphite -- is a chemically stable system which is essentially at equilibrium and free of corrosion. Further, the equilibrium chemical behavior of the important fission products is reasonably predictable. The current successful operation of the MSRE supports these conclusions. 640 10, 11. References Rosenthal, M. W. and P. R. Kasten, '"Molten Salt Reactors - Their Performance, Objectives and Status," Presented at the American Nuclear Society Meeting, Washington, D.C., Nov. 11-15, 1968. To be published in Nuclear Applications. Bettis, E, 5. and R. C. Robertson et al., "MSBR Design Features and Performance," Presented at the American Nuclear Society Meeting, Washington, D.C., Nov. 11-15, 1968. To be published in Nuclear Applications. Perry, A. M., '"MSBR Reactor Physics," Presented at the American Nuclear Society Meeting, Washington, D.C., Nov. 11-15, 1968. To be published in Nuclear Applications. Grimes, W. R., "Molten Salt Reactor Chemistry," to be published in Nuclear Applications. Grimes, W. R., '"Chemical Research and Development for Molten Salt Breeder Reactors,'" USAEC Rept. ORNL-TM-1853, Oak Ridge National Laboratory, June 1967. Baes, C. F., Jr., "The Chemistry and Thermodynamics of Molten Salt Reactor Fluoride Solutions," Thermodynamics, Vol. 1, International Atomic Energy Agency, Vienna, 1966, pp. 409-433. Hitch, B. F. and C. F. Baes, Jr., "An Electromotive Force Study of Molten Lithium Fluoride-Beryllium Fluoride Solutions," Inorganic Chemistry, Vol. 8, 1969, pp. 201-207. Ward, W. T. et al., "Solubility Relations Among Rare-Earth Fluorides in Selected Molten Fluoride Solvents,' USAEC Rept. ORNL-2749, Oak Ridge National Laboratory, Oct. 1959. See also, Ward, W, T. et al., Chemical and Eng. Data, Vol. 5, No. 2, April 1960, pp. 137-142. Barton, C. J., "Solubility of Plutonium Trifluoride in Fused- Alkali Fluoride-Beryllium Fluoride Mixtures,'" J. Phys. Chem., Vol, 64, 1960, pp. 306-309 Long, G. and F., F., Blankenship, Reactor Chemistry Division Ann, Progr. Rept. for Period Ending Jan. 31, 1965, USAEC Rept. ORNL- 3789, pp. 65-72, Oak Ridge National Laboratory, April 1965. Bamberger, C. E,, C. F. Baes, Jr., and A. L. Johnson, Reactor Chem. Div. Ann. Progr. Rept. for Period Ending Dec. 31, 1968, USAEC Rept. ORNL-4400, Oak Ridge National Laboratory, in press. 641 12. 13. 14, 15. 16. 17. 18. 19. 20. 21. 22. Field, F. E. and J. H. Shaffer, "The Solubilities of Hydrogen Fluoride and Deuterium Fluoride in Molten Fluorides," J. Phys. Chem., Vol, 71, 1967, pp. 3218-3222, Barton, C. J., L. 0. Gilpatrick, J. A. Fredricksen, "Solubility of Cerium Trifluoride in Molten Mixtures of LiF-BeF,, and ThFy,' Presented at 157th Meeting, American Chem. Soc., Meeting April 13-18, 1969, Minneapolis, Minn., submitted for publication in Inorganic Chem. T Mathews A, L., and C. F, Baes, Jr., "Oxide Chemistry and Thermo- dynamics of Molten LiF-BeF, Solutions,'" Inorganic Chem., Vol. 7, 1968, pp. 373-382, Hitch, B, F, and C. F, Baes, Jr., Reactor Chemistry Div. Ann. Progr. Rept. for Period Ending Dec. 31, 1966, USAEC Rept. ORNL~- 4076, pp. 19, 20, Oak Ridge National Laboratory, Mar., 1967. Hitch, B. F. and C. F, Baes, Jr., Molten Salt Reactor Prog. Semiann. Progr. Rept. for Period Ending Feb. 28, 1966, pp. 133- 136, Oak Ridge National Laboratory, June 1966. Apple, R. F. et al., "Determination of Oxide in Highly Radio- active Fused Fluoride Salts - Hydrofluorination Method," pre- sented at the Winter Meeting, American Chem. Soc., Phoenix, Ariz., Jan. 17-21, 1966, To be published. Stone, H. H., and C. F, Baes, Jr., Reactor Chem. Div, Ann. Progr, Rept. for Period Ending Jan. 31, 1965, USAEC Rept. ORNL-3789, pp. 72-76, Oak Ridge National Laboratory, April 1965, Freasier, B. F., C. F. Baes, Jr., and BH. H. Stone, Reactor Chem. Div. Ann., Progr. Rept. for Period Ending Dec. 31, 1965, USAEC Rept. ORNL-3913, pp. 38-40, Oak Ridge National Laboratory, Mar. 1966, Blood, "The Solubility and Stability of Structural Metal Difluorides in Molten Fluoride Mixtures,'" USAEC Rept. ORNL- CF-61-5~4, Oak Ridge National Laboratory, Sept. 1961. Elliott, J. F. and M. Gleiser, "Thermochemistry for Steel Making," American Iron and Steel Institute, Addison-Wesley Publ. Co., Reading, Mass., 1960. Meyer, N, J., C. F. Baes, Jr., and K. A, Romberger, Reactor Chem. Div. Ann., Progr. Rept. for Period Ending Dec. 31, 1967, USAEC Rept. ORNL-4229, pp. 32, 33, Oak Ridge National Laboratory, Mar, 1968. 642 23. 24, 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. Greenberg, E., C. A, Natke, and W. N. Hubbard, "Fluorine Bomb Calorimetry. X. The Enthalpies of Formation of Niobium and Tantalum Pentafluorides,'" J. Phys. Chem. Vol. 69, 1965, pp. 2089~2093. Kubaschewski, 0., E. L. Evans, and C. B. Alcock, Metallurgical Thermochemistry,' Pergamon Press, 1967. Porte, H. A., E., Greenberg and W. N. Hubbard, "Fluorine Bomb Calorimetry. XII. The Enthalpy of Formation of Ruthenium Penta- fluoride," J. Phys. Chem., Vol. 69, 1965, pp. 2308-2310. ""JANAF Thermochemical Tables,'" Clearing House for Federal Scien- tific and Technical Information, U,S. Dept. of Commerce, Aug. 1965. Bamberger, C. E., C. F. Baes, Jr., and J. P. Young, "Containment of Molten Fluorides in Silica: Part I. Effect of Temperature on the Spectrum of U*' in Molten BeF,-LiF Mixtures," J. Inorg. and Nucl. Chem.,, Vol. 30, 1968, p. 1979. Bamberger, C. E. and C. F. Baes, Jr., Reactor Chem. Div. Ann. Progr. Rept. for Period Ending Dec. 31, 1968, USAEC Rept. ORNL- 4400, Oak Ridge National Laboratory, in press. Eorgan, J. H. et al., Reactor Chem. Div. Ann. Progr. Rept. for Period Ending Jan. 31, 1964, USAEC Rept. ORNL-3591, pp. 45-46, Oak Ridge National Laboratory, May, 1964. Romberger, K. A., C., F. Baes, Jr., and H. H. Stone, "Phase Equilibrium Studies in the U0,-Zr0, System,' J. Inorg. and Nucl. Chem., Vol. 29, 1967, pp. 1619-1630. Ross, R. G. et al., "The Reductive Extraction of Protactinium and Uranium From Molten LiF-BeF,-ThF, Mixtures Into Bismuth,' This volume. Shaffer, J. H. et al., "The Recovery of Protactinium and Uranium for Molten Fluoride Systems by Precipitation as Oxides,' Nucl. Sci. and Eng., Vol. 18, 1964, pp. 177-181. Jennings, W., F. A, Doss, and J. H. Shaffer, Reactor Chem. Div. Ann. Progr. Rept. for Period Ending Jan. 31, 1964, USAEC Rept. ORNL-3591, pp. 50-52, Oak Ridge National Laboratory, May 1964. Oetting, F., L., '"The Chemical Thermodynamic Properties of Plu- tonium Compounds,' Chem. Rev., Vol. 67, 1967, pp. 261-297. Rudzitis, E. and E. H. Van Deventer, "Chemical Engineering Div. Ann. Rept., July-Dec. 1967," AEC Rept. ANL-7425, pp. 122-123. Holley, C. E., Jr., E. J. Huber, Jr., and F. B. Baker, "The 643 36. 37. 38. 39. 40, 41, Enthalpies, Entropies, and Gibbs Energies of Formation of the Rare Earth Oxides," Progress in the Science and Technology of Rare Earths, L. Eyring, Ed., Vol. 3, 1968, pp. 343-433, Doss, F. A., F. F, Blankenship, and J. H. Shaffer, Reactor Chem. Div, Ann. Progr. Rept. for Period Ending Dec. 31, 1967, USAEC Rept. ORNL-4229, pp. 39,40, Oak Ridge National Laboratory, Mar. 1968. Senderoff S. and G. W, Mellors, "The Electrodeposition of Coherent Deposits of Refractory Metals: IV. The Electrode Reaction in the Deposition of Niobium," J. Elect. Soc., Vol, 1, 1966, pp. 66-71. Watson, G. M. et al., "Solubility of Noble Gases in Molten Fluorides: In Lithium-Beryllium Fluoride," J. Chem. and Eng. Data, Vol. 7, 1962, pp. 285-287. Blomeke, J. 0. and M. F. Todd, "Uranium-235 Fission Product Production as a Function of Thermal Neutron Flux, Irradiation Time, and Decay Time," USAEC Rept. ORNL-2127, Oak Ridge National Laboratory, Nov. 1958. Settle, J. L. H. M. Feder, and W. N. Hubbard, "Fluorine Bomb Calorimetry. II. The Heat of Formation of Molybdenum Hexa- fluoride,”" J. Phys. Chem., Vol. 65, 1961, pp. 1337-1340. Rand, M. H. and 0. Kubaschewski, '"The Thermochemical Properties of Uranium Compounds,' Interscience Publ., New York, 1963. 644 CALCULATIONS ON THE SEPARATION PROPERTIES CF THORIUM-URANIUM FUELS BY CHLORII'E VOLATILIZATIONst E.Fischer, M,Laser, and E.Merz Kernforschungsanlage Jiilich GmbH, 517 Jiilich/Germany Institut fiir Chemische Technologie Abstract The composition of the gas phase resulting from a high temperature chlorination of uranium-=thorium fuels was computed for different process conditions. From these values the theoretical uranium and protactinium losses caused by a codeposition of thorium and uranium or protactinium chlorides, respectively, were calcu- lated. The results have shown, that tolerable uranium losses can be achieved, if the partial pressure of uranium is very low or the deposited thorium chloride is purified by a sublimation procedure, *Work performed under a joint project sponsered'by the German Federal Ministry of Science. 645 Introduction The High Temperature Gas Cooled Thorium Reactors of AVR and THTR type operate with graphite balls con- taining thorium uranium oxide or carbide particles, coated with pyrocarbon and silicon carbide. The uranium content of the fuel elements is relatively low. A typi- cal sphere which weighs nearly 200 g contains only 10 to 20 g of particles in which the uranium thorium ratio is approximately 1:25, corresponding to a uranium con- tent of less than o.5% related to the full sphere. The protactinium content varies with the conditions of reactor operation. One calculation has given a wvalue of 31 mg Pa per sphere. For the reprocessing of these fuel elements we are developing a combined chlorination solvent extraction process, the so~called CHLOREX process (1). The balls are broken and ground to a particle size of less than 300 um with a small portion of fines below 50 um. The material is then fed into a fluidized bed and heated to 1000 or 1100°C. In the presence of graphite the heavy metal oxides are converted to the volatile chlo~ rides by chlorine. Also most of the fission products are volatilized under these conditions. The chlorides are deposited in a condenser, dissolved in 2 M hydro- chloric acid and purified by solvent extraction with long chain aliphatic amines or, after conversion of the chlorides to the nitrates, with tributyl phosphate. The resulting solution may be used for the refabrica- tion of new particles by the sol gel technique. At the present state of development we do not in- sist on a separation of the chlorides by fractional condensation, because the isclation and purification of protactinium and uranium is easily carried out by aqueous process steps. However, in the future we will try to achieve separation by condensation of thoriur chloride at 500 to 600°C and uranium and protactin a chlorides at room temperature, In this case the thordum fraction may be easily converted to a storable form and discarded to the waste, But it is essential that thorium chloride be deposited nearly uranium free. The first inspection of the vapour pressure curves (fig.1) shows that the volatilities of thorium chlori e, uranium pentachloride and hexachloride are very diffe- rent, so that one should expect a good separation. Some laboratory experiments have given encouraging results, but in other runs the uranium content of the thorium chloride was too high. Therefore we have calculated 646 L¥9 log p [atm] | 3] L o 1. CeClj | 1 ] i S N W N | ] 100 200 300 400 500 600 700 800 1000 1500 2000 Temperature (°C] Vapour Pressure Curves of some Chlorides. the gas phase equilibria between the most important components carbon monoxide, chlorine, carbon oxychlo- ride, thorium chloride, uranium tetrachloride, uranium pentachloride, and uranium hexachloride and the cor- responding activities of uranium in the condensed thorium chloride. From these values we have calculated the theoretical uranium losses under different process conditions. The results of these calculations may deviate from experimental data to a greater or lesser degree, because we must estimate some of the thermodynamic data, However, the variation of some param- eters shows the influence of the different operating conditions, Finally the results have suggested requirements for the construce tion of the condenser, Fundamentals The chlorination of thorium oxide in the presence of carbon at high temperature is given by the equation Tho + 2 C + 2 Cl1l gy ThC1 + 2 CO 1 2(s) (s) 2(g) 4 (g) (g) () The chlorination of carbides can be described by ThCz(s) + 2 012(g) — ThClh(g) + 2 c(s) (2) Under the same conditions uranium forms the three chlorides UCl),, UClg, and UClg; protactinium forms PaClgsy the fission products form their respective chlorides; and silicon carbide forms sili- con tetrachloride., The main gas phase components are connected by the following equilibria: CO + Cl,e=3 COCl, (3) UCLy () *+ 1/2 Clg==2 UCl (&) PR UCL,(g) + Clp &= UCl (5) These equilibria are all determined by the chlorine potential. Moreover, the vapourization equilibria of ThClh, UClu, UCl5 and UCl6 must be considered. The thermodynamic data used for the calculation of the equilibria are given in table 1. (2). 648 Table 1. Thermodynamic Data, _ -2 _AH298 48298 ¢ =a + bT + T kcal 1_cal L1053 L1022 mole deg-mole¢ a .10 ¢ 10 Co 26.40 7.3 6.79 0.98 =-0.11 Cl2 0.0 53.3 8.82 0.06 -0.68 COCl2 53.3 69.1 (14.51* uc 206. 1. .0 * ll}(g) 6.6 91.5 | 26 UCl5( ) 237.7 96.0% 20,0 * 0,01%* UCl6( ) 255.6% 101.3%] 30.0 * 0.005%* -1 log p = A.T + Belog T + C-T + D [atnfl A B C D ThClll-(S) »12,900 11.42 ThClh(l) - 7,980 6.69 UClh(s) -11,3%0 -3.02 20.33 UCl5(s) - 5,500 7.82 UC16(5) - 4,000 7.32 *Estimated Values The equilibrium constants of reactions 3 to 5 were calculated according to equation aH as T T 8 k,576.T ¥,57h H H R Ac -dT with a T = A 298 + 1298 p T and ASp = A8298 ¥ 1298 Acp/T-dT 649 (6) The temperature dependence of the equilibrium con- stants are approximated by the eguations: CO + Cl,&= COCl, log K1 = 5820/T - 6.64 - 0.000242-T log K, = 7300/T - 5.90 c C uc UCly(g) + Cla ™= Ullg(y) log K., = 10820/T - 9.80 3 The partial pressures were calculated by a computer.(3) The following data were put in: ‘ (i) the temperature functions of the equilibrium constants K1, K2, and K3; (ii) the temperature functions of the vapour pressures of UCl,, UCl, UClg, and ThCl; (iii) The mole numbers of CO, Cl ThCl,, and U 2° total’ If the gas phase, containing the reaction products and an excess of chlorine, is cooled from the high re- action temperature, thorium chloride, and at lower tem- perature uranium chlorides will condense. However the separation of thorium and uranium chlorides cannot be quantitative, if one or more uranium species are soluble in the thorium chloride. In this case the thermodynamic activity of the uranium chloride in the thorium chlo- ride is given by the equation P ay = g (7) %y | Py= partial pressure of the respective uranium species in the system o . P ;=vapour pressure of the respective pure uranium species. That means, that uranium is deposited together with thorium chloride even if the partial pressure in the gas phase is much lower than the vapour pressure of the pure uranium compound. 650 Since UCl, and ThCl, have the same crystal structure and similar "lattice parameters (4), UCl, is soluble in ThCl, . The solid solution should nearly approximate the 'ideal case and that means that the activity co- efficient should be approximately unity and RAOULT's law can be assumed to be valid in the whole range of concentration. On this basis equation 7 changes to PUCla XUClu = o (8) ucl, XUClh= mole fraction of UClu in ThClh This enables one to calculate the amount of uranium tetrachloride in the thorium chloride, There is nothing known as to the solubility of ura- nium pentachloride and hexachloride in thorium chloride. The hexachloride, however, can be ruled out because the activity in the thorium chloride would be so low that it would not make an appreciable contribution to the uranium content in the thorium chloride matrix., However, the solubility of uranium pentachloride in thorium chlo- ride cannot be excluded. Indeed, the pentachloride could not be found in deposited uranium tetrachloride (5), however it is known that some solubility usually occurs even if the crystal structure of the soclute is very different from that of the matrix. The activity co- efficient, however, can be very different from unity in this case, so that one cannot calculate the mole fraction of the uranium pentachloride in thorium chlo- ride until exact data are available. Protactinium in the presence of chlorine is stable as pentachloride, The vapour pressure data were recently published by WEIGEL (6). The behaviour of this compound with respect to the solubility in thorium chloride is certainly similar to uranium pentachloride. Results of the Calculations The composition of the gas phase resulting from the fluidized bed chlorination of thorium uranium oxide is partly given by the reaction equation and by the composition of the fuel. The chlorine partial pressure is variable and depends on kinetic effects, resulting in a more or less exhaustion of the chlorine, which is 651 fed into the fluidized bed. This exhaustion may be de- fined by 4 exhaustion = 2 _X moles heavy metal chlorides‘100 2 x moles heavy metal chlorides + x moles Cl 2 Figures 2 and 3 show the partial pressures of the main components ThClh, UClh, UC15, UCl6, co, C12' and 00012 as a function of the temperature for 50 % and 1 % ex- haustion. The U:Th ratio is 5:95 and is nearly identical with the ratio in an AVR fuel element. The main difference between the two diagrams is the lower heavy metal chlo- ride concentration in the latter case. The broken line gives the vapour pressure of solid uranium tetrachloride. The difference between the calculated partial pressure pUClh and the vapour pressure pOUCl in the range of ThC1l, condensation represents the thermodynamic activity of UCl1, in ThCl,. One can see, that a low uranium chlo- ride con%ent results in a low uranium tetrachloride ac- tivity in the solid phase. From the different slope of the two curves the ac- tivity of uranium tetrachloride is a function of the tem- perature resulting in fractions of thorium chloride with different uranium content. In most cases the thurium chloride, which condenses later is richer in uranium than the first fraction. Thereforswe have calculated the activity and the mole fraction of uranium tetrachlo- ride in the condensed thorium chloride for temperature intervals of 10 degrees. After the summation of these results one can calculate the fraction of the total uranium, which is codeposited with thorium chloride. Some results of the calculations are given in fig.4, in which the uranium loss ((UCl, codeposited with ThClu)/ UClh(total))is shown as a function of the mole fraction of uranium in the gas phase for the chlorination of oxides and carbides. One can see that the uranium loss increases in a remarkable way with increasing mole fraction of uranium, For a given U:Th ratio the losses are somewhat lower in the case of chlorination of carbides. A higher total pressure, which drives reactions 4 and 5 more to the side of the penta and hexachloride, results in very high uranium losses, because the partial pressure of uranium tetrachloride increases by a factor of 3 to 5 when the total pressure increases by a factor of 10, 652 Partial Pressure [ atm ) 100 =< 7 cocr; / ThCl, 107! | | / co MP. 102 |- A UClg ' ‘ / 1073 |~ / / p°UCM/ UCl, 1074 / [ 10-5 ] J | | 0 200 400 600 800 1000 Temperature [°C ] 2. Composition of the Gas Phase Resulting from the Chlorination of (U 50 % chlorine Th }o._. Exhaustion. 0.05°70.95'"2 653 100 _ C . _1 102 00t 7--—.— co S ThCl, pOUCI, ~ 107 |- € / o " / 3 g 107 |- [ [P NS o - / 5 / a UcCt, 10°5 10 - / ! [ 1 | 0 200 400 600 800 1000 Temperature [°C) . Composition of the Gas Phase Resulting from the Chlorination of (U Th )0 . 1 4 Chlorine Exhaustion. 0.05770.957 72 654 G99 ( UCI[, N ThCll. )/UTotQ[ 010 0,05 Oxides (5:95) 10 atm Carbides (5:95) latm Oxides(5:95) 1atm Oxides (40:60)1atm i 1 I 0,01 0,02 0,03 Mole Fraction U in the Gas Phase ] 4, Uranium Losses as a Function of the Mole Fraction of Uranium in the Gas Phase. For a given mole fraction of uranium in the gas phase the uranium losses decrease with decreasing thorium mole fraction, a higher U:Th ratio. From the technological standpoint the that means,the losses are lower for fuels with dependence of the uranium losses from the chlorine exhaustion is of greater interest, The diagram fig.5 shows, that for a given chlorine exhaustion the chlorination of oxides is more advantageous than chlorination of carbides, the bheavy metal chlorides are diluted by This dilution is more effective than the chlorine potential according to equation The thermodynamic activity of uranium according to the equation Pyci a = ——a > Plyer 5 is higher than that of the tetrachloride up to 2.5 in the interesting temperature because we cannot say anything about the efficient, it is impossible to calculate tion of the pentachloride in the thorium since carbon monoxide. decrease of the 3. pentachloride (9) by a factor range. But activity co- the molar frac- chloride matrix. The total uranium losses caused by the solubility of both uranium tetrachloride and pentachloride are there- fore certainly somewhat higher than those given in fi- gures 4 and 5. Also for protactinium pentachloride we cannot assume an ideal solution. Nevertheless we have calculated the protactinium losses using an activity coefficient of unity. For a thoritum oxide fuel with a Pa:Th ratio of 1:440 (corresponding to a Pa:Th ratio in an AVR fuel element) we got a loss of 0.3 % at 50 % chlorine ex- haustion. Under the same conditions the uranium losses caused by the solution of uranium tetrachloride in tho- rium chloride amounts to 1,92 %, The actual protactinium losses decrease with increasing activity Technological Consequences coefficient. The chlorination reactor must be operated at 1000 to 11009C because most of the fission product chlorides must be also volatilized to reduce the uranium retarda- tion by the fission products in the fluidized bed. Under these conditions 90 % or more of the heavy metal oxides are chlorinated and volatilized as chlorides during the first 20 or 30 minutes. The gas velocity 656 should be as LS9 Oxide(5:95) 10 atm 010 _ arbide(5:95) o 1atm © 2 ’"\‘; Oxide(5:95) S 1atm £ 005 |- £ il () - Oxide{40:60) 1atm | —— ] | 0 20 40 60 80 Chlorine Exhaustion [ %] 5. Uranium Losses as a Function of the Chlorine Ex- haustion. low as possible to reduce the blowing out of the fines. Mostly we operate near the minimum fluidization velocity. Therefore the chlorine exhaustion can amount to 50 % and the uranium losses are too high, There are two ways to decrease the percentage of uranium codeposited with thorium chloride. The simplest method is to dilute the reaction gas by injection of chlorine behind the chlorinator. This is partially veri- fied, for instance, in condensers in which the reaction gas is cooled by injection of cold chlorine. The dis- advantage of this method is, that the quantitative de- position of uranium hexachloride becomes very difficult at low uranium partial pressures. Another way of condensation is the deposition of thorium chloride in a column filled with an inert ma- terial and having a small temperature gradient. When the main reaction is over and the column is flushed with nearly pure chlorine, the thorium chloride is transported along the column, The thorium chloride sublimes several times and the uranium content is re- duced strongly. First experiments have shown, that ura- nium losses are lower than 0.5 %. Now we are developing these two types of condensers in laboratory scale. References 1. Fischer,E., G.Kaiser, M.Laser, E,Merz, H.J.Riedel, and H.Witte, "Development of Combined Reprocessing Procedures for Thorium-containing Fuel Elements", JAEO Panel on Reprocessing of Highly Irradiated Fuels, Vienna, May 1969. 2. Kubaschewski,O0., E.L1.Evans, C.,B.Alcock, "Metallur- gical Thermochemistry", Pergamon Press 1967. 3. Kirchner,H., M.Laser, and W.Schidlich, "Fortran- Programm zur Berechnung der Gasgleichgewichte bei der Chlorierung", KFA-Report, in preparation., 4, Brown,D.,"Halides of Lanthanides and Actinides", John Wiley and Sons Ltd., 1968. 5. Kanellakopulos,B., private communication. 6. Weigel,F., V.Crespi, and M.Krumpel, "Der Dampfdruck des Protactinium(V)-chlorids”, jrd International Protactinium Conference, Elmau, Germany, 1969. 658 CALCULATION OF THERMODYNAMIC PROPERTIES FROM BINARY PHASE DIAGRAMS™ P. Chiotti, M, F, Simmons* and J, A, Kateley Institute for Atomic Research and Department of Metallurgy lowa State Univerfffig, Ames, lowa 50010 .S A, ABSTRACT Basic thermodynamic relations for the description of binary phase boundaries are developed. Their application to the calcula- tion of thermodynamic properties from phase boundary data and in checking the consistency of phase boundaries with known thermo- dynamic data are outlined, Thermodynamic properties cannot be calculated from phase boundary data alone, Data for the pure com- ponents and assumed empirical parametric relations can be employed to calculate useful information in favorable cases. The use of empirical relations in the correlation of known properties for binary systems is demonstrated and their application to the calcu- lation of thermodynamic properties for the liquid phase in simple eutectic systems and for the solid phase in miscibility gap systems was investigated, +Work was performed in the Ames Laboratory of the U.S, Atomic Energy Commission, Contribution No. 2552, *Present address: U.S. Army, Fort Lee, Virginia. 659 INTRODUCTION The molar properties for a pure stable substance in the absence of surface or field effects may be represented by some function Y = Y(P,T}), where Y is any molar property and P and T are the pres- sure and temperature, respectively., A unique P-T surface describes each of the states of aggregation, The intersections of these sur- faces project on the P-T plane as the unary phase diagram'. For binary systems the molar properties are functions of three variables, Y = Y(P,T,X). The complete phase diagram can be represented in three dimensional P,T,X space. Most binary phase diagrams of metal- lurgical interest are constant pressure, | atmosphere, sections of the general three dimensional diagrams, The intersection of the surfaces, Y = Y(C,T,X), where C is one atmosphere, project on the T,X plane as the usual binary phase boundaries®. C(onsequently, the phase boundaries are thus related to the state properties, but these properties cannot be evaluated from the phase diagram alone, addi- tional information is needed. |In principle, all of the thermodynamic properties may be calculated from the binary phase boundaries if we know the correct form of the equation relating the free energy or activity coefficient to the variables P,T,X or to T and X if P is fixed. Unfortunately no general equation of state of this type has yet been developed. Various parametric equations have been devel- oped or prgposed by Hildebrand and Scott3, Lumsden™, Krupkowski5, Guggenheim®, and other authors, Partial summaries have been given by Wagner/ and Perry®., The use of the Van Laar, Margules, Scatchard-Hamer equations as well as relations of a more general type to represent the activity coefficients of components in binary and ternary systems has been discussed in some detail by Wohl9, Some basic considerations and developments in the theory of alloy phases have been more recently reviewed by Kleppalo, Orriani and Alcock!!, and Turkdogan and Darken!2, The thermodynamic properties of the metallic elements at one atmosphere pressure are quite well known, This is also true for the pure components of many salt systems. These data permit calcu- lation of the thermodynamic properties for solutions from phase boundary data in some favorable cases. However, the development of a relation for the excess free energy as a function of temperature and composition usually requires the evaluation of four or more empirical parameters and calculations are tedious, The advent of the modern computer has alleviated much of the computational drudgery and there has been a growing interest in such calcula- tions!3-17, Phase diagrams and thermodynamic data are particularly helpful in the development of new or more efficient reprocessing methods for nuclear fuels as well as in other engineering applications and in the development of the theory of alloy phases and heterogeneous equilibria in general. The purpose of this paper is to derive and summarize general thermodynamic relations which describe the phase 660 boundaries of constant pressure binary systems and which can be employed to check the consistency of the phase boundaries with thermodynamic data or conversely to calculate thermodynamic proper- ties from phase boundary data. The application of empirical rela- tions to the calculation or estimation of thermodynamic properties from simple eutectic and miscibility-gap binary-phase boundaries will be evaluated. BASIC RELATIONS FOR T-X DIAGRAMS For a heterogeneous system at equilibrium the chemical potential, Wi or fii and the fugacity, f;, for a particular component i must be constant throughout the system; the fugacity must be equal in each phase present. This fact is the basis for the derivations which are to follow, Subscripts will be used to indicate the component and superscripts to indicate the phase in question, The fugacity or escaping tendency and the chemical potential were defined by Lewis and Randal1183by the relation G; =RT In f; + B (n where B is a function of temperature only., Since there is no way of fixing an exact value to B we are restricted in our measurements to relative values or (G; - 6;°) = AG; = RT In £;/F;° (2) where Gio and fio are the chemical potential and fugacity, respec- tively, of component i in some reference or standard state, and AG; is the relative partial molar free energy. The standard state, uniess otherwise indicated, will be taken to be the pure component at one atmosphere pressure at the same temperature and in the state of aggregation of the solution in question, The quantity AG; then represents the isothermal change in free energy for the process of adding one mole of pure component i to a large reservoir of solu- tion which essentially remains fixed in composition., This process will be indicated, in the case of a liquid solution, and component A for example, by the relation A(4) - A(g,s0ln), AG = AGp = RT In ap (3) Equation 2 also defines the activity a; and for condensed phases a; = fi/fi0=XiYi (L") where X is the mole fraction and Y is the activity coefficient. With two phases, | and Il in equilibrium we can write from Eq. & t ] o,l _ IR A o’|| X6 =GR (5) 661 where the superscripts ' and '' refer to phase | and Il respectively. Taking logarithms of both sides, assuming constant pressure, and differentiating with respect to 1/T gives 1 1 O ] 11 (N} 0 il d In X, d In Y, d In f,’ d In X. dInl, d In f.? 6 i, i, i - - i (6) d 1/T d /T d /T d i/T d 1/7 d /T The left side of this relation is restricted to compositions and temperatures defined by the boundary between phase | and the two- phase field of | and |l while the right side is restricted to the boundary between phase Il and the same | plus |l phase field. Along a phase boundary ¥ is a function of both temperature and composi- tion. Consequently for a two component system In ¥ 3 InYy d 1n Y=(a ) dU/T)+(——~w—) dX (7a) 3(17T) P,X X Jp. 1 which, for any particular component, leads to dlnY _ £§_+ . (a In ¥ ) d In X (7b) P,T a(1/m) It can also be shown that [} 1 L Y | d ln(f?’ /£50) -aH?’ = (8) d(1/T) R In these expressions M. is the relative partial molar enthalpy and AH?? ' is the standard enthalpy change for the transformation of pure component i from the state of aggregation of phase |l to that of phase I, Substituting (7) and (8) in (6) and rearranging terms gives i ' ! " d In Xi /o In Yi AHi d In X, —_— |+Xi —_— -}--—=_._.._—I d(1/T) 2 X; R d(1/T) T,P ] _n 1yt Jf3n v A AH?’ T+ X —— t— " — (9a) P} Xi R R T,P Let gi. represent the terms in brackets, and since d{(1/T) = (-I/Téde and d In X = (1/X)dX we can write 662 RT dxi 1 - RT dxi ' 1 OII__.I — == g;; - AHi == — 9 - fl\Hi + AHi’ (9b) X. dT X, dT The relative excess partial molar free energy is defined as A8 = RT In ¥; and for a binary solution, the Gibbs-Duhem relation yields 1 1 and an analogous relation holds for phase |1, The slopes of the phase boundaries for any two coexisting phases in a binary T-X diagram must satisfy Eq. (9b). This equation has been derived by a somewhat different procedure by Williamson!Sb, The slopes of the phase boundaries can also be expressed in terms of the relative excess partial molar entropies, A§¥S The enthalpy terms in Eq. (9b) can be combined to yield a single term Afii which is the isothermal change in enthalpy for the equi- librium transfer of one mole of component i from phase 11 to phase . The free energy change for this transfer is zero and it follows "™ and since 85; = £5%5 + 2519 £q. (9b) may be that AHi"" = TAS written as 1 RT dX; " ¥ . _‘l - AS’;S’ +R In — + AS?’ . Xi dT ; dT Xi The pair of Eqs. (9b) can be combined to eliminate one of the slopes to yield ' 1 1 R ¥ I I | XZ i X] 0 dX] _ X] AHl + X2 AHZ XZ X] dT RT .t where AH.” has the significance discussed above. This equation has been derived by Kirkwood and Oppenheimlgc,but has no particular advantage over the Eqs. (9b) or (11) and will not be considered further. Eq. (12) is one form of the Gibbs-Konovalow relation which has been applied to the description of binary phase bound- aries by Franzen and Gerstein! The chemical potentials for a component in two equilibrium phases are equal, and therefore ] — MG, = 8B, - AG?’ 663 which also can be written as =X5 , -XS,“ _ e O’II_,I - AG; = RT In(X;/X.) - &G, (13) This equation and Eq. {5) are general and hold regardless of the number of components or phases present whereas Eg. (3) is restricted to a univariant (one degree of freedom) system and if more than two components are present Eq. (7a) must be modified, The composition terms and phase boundary slopes in these equations can be obtained directly from the phase diagram, It should also be noted that Egs. (9b) and (11) may be obtained by appropriate differentiation of Eq. (13). Only constant pressure diagrams will be considered in the following presentation. The entropy, free energy, and enthalpy terms are functions of both temperature and composition and can not be evaluated without additional information, Another relation which is helpful in the analysis of phase dia- grams results from the fact that the sum of the changes occurring in any state property along a closed path must be zero or $dy = 0. (14) EUTECTIC SYSTEMS WITH NEGLIGIBLE OR LIMITED SOLID SOLUBILITY Considerable information can be calculated from the phase bound- aries of a simple eutectic system if the thermodynamic properties of the pure components are known, For each two-phase region there are two equations of the type (9b) and two of the type (13) which describe the phase boundaries, For the A-rich liquidus, £, which is in equilibrium with an A-rich solid solution, o, Eqs. (13) give - £ - y f XS, 4 L =xs,o _ o _ us MGy MGy RT In X3/x," - 4G, (15a) MRS 4 ARKS® o T In XY - A (15b) B B B’ "B B A similar pair of equations hold for the B-rich liquidus and B-rich solidus boundaries and another pair for the two boundaries below the eutectic temperature. In (15a), A6, ¥ is negligibly small in accord with the condition that the solid pPhase @ is nearly pure A, , With this simplification Eq. (15a) may be used to calculate Afifis’ for T,X points defined by the A-rich liquidus, The free energy of fusion of any component i is calculated according to the relation AG?’FUS = AH? - TAS? (16) where the enthalpy and entropy terms are functions of temperature 664 and may be expressed as AH? = AH. + s¢, dT (17) d InT. (18) — AHi and Asi are the standard enthalpy and entropy of fusion, respec- tively, at the melting temperature T and ACp is the difference in the heat capacity of the liquid and solid states of pure component i. If the temperature span is small or if ACp is small the integrals in (17) and (18) may be relatively insignificant and consequently are sometimes neglected. In the latter case Eq. (15a) reduces to xs , 4 - L. o 3 % AG, = -RT ln(XA/XA) - MM, + TAS (19) A A Equation (15b) cannot be simplified and unless both components, A and B, have the same crystal structure it is not possible to calcu- late the last term on the right side. This term represents the free energy of fusion of pure B in the alpha form, If A and B have the same crystal structure then this term is evaluated simply as the free energy of fusion of B, otherwise it is necessary to know the free energy of transition of B to the o form, Below the eutectic temperature the two solid phases are only very sparingly soluble in one another, If the crystal structures are the same, the last terms in the analogous pair of equations are zero and the relation for the B component in the alpha solid solution becomes ATES Y = RT 1n(x¥/xB) = AR® - TaSXSH® (20) BB B B and a plot of log(Xg/XE) against (1/T) will yield a straight line from which AHg and ASES*¥ may be determined from the slope and intercept, respectively., The curve represents the solubility as a function of temperature of a sparingly soluble solute B in the A- rich solid solution and Afig and A§§5'“ are the limiting values of the relative partial molar enthalpy and excess entropy respectively, If the two solids do not have the same crystal structure the plot will still yield a straight line but the slope and intercept yield (AHg + Afig’afl“) and (A§§S’“ + &SS’BfiQ) respectively. This fact is sometimes neglected in the interpretation of such data, The same arguments apply for data on sparingly soluble solutes in liquid 665 metals., Methods of calculating or estimating the free energy of allotropic transformations for pure metals have been investigated by Kaufman!9 and Roy and Kaufman20, The Eq. (9b) also can yield useful information., For points along the A-rich liquidus it takes the form 2 y A RT dX A & =4 fus —~ — 9, = M+ M , (21) xA dT AA A A L and as the melting temperature is approached g:A and AHA approach unity and zero respectively. Consequently it reduces to the well known melting point lowering equation RT? ax? ax ¥ anfus A _ fus B = A - — - AHA —_— = - 5 - (22) X dT dT 2 RT A XAfll Similar equations may be written for the B-rich liquidus and at the eutectic temperature g&A = géB and Eq. (21) and its B-component analog may be combined to give dT =L fus, _ & /dT =4 fus XB (:;I) (AHB + AHB ) = XA (——) (AHA + AHA ). (23) B ’ A Equation (15a) may be written as fus axs , 4 L, 0 ] fus - = + A T(ASA + 85, R In xA/xA) AHA Hy (24) which can be combined with (21) to eliminate the two enthalpy terms and at the eutectic temperature further combined with its B-com- ponent analog to give 4 [dT fus =XS , 4 4, By _ A[dT_ XB(?) (ASB + 85,77 - R In xB/xB) = X, 7 B /g A (25) fus =xs , 4 4, (ASA + ASA =R In XA/XA) It should be noted that Eqs. (23) and (25) apply only for the three phase equilibrium at the eutectic temperature and that dT/dXp does not equal -dT/dXp, these two quantities are the slopes of the B-rich 666 and A-rich liquidus curves at the eutectic temperatuEe, resp%ctively. A1l the quantities in these two equations, except AFg and AHp in Eq. (23) and Agfis’t and Agfis’z in Eq. (25) may be determined from the phase diagram or calculated from the known properties of the pure components, |If any one of the four unknown quantities is determined the other three may be calculated from these equations and the relation Afi?s = Afii - TA§?S. Consequently these relations are helpful in checking the consistency of the phase diagram with the thermodynamic data or vice versa, Further evaluation of thermo- dynamic properties from the phase diagram requires additional in- formation. SYSTEMS WITH COMPLETE MISCIBILITY |f the components A and B have the same crystal structure, it is possible to have complete solid miscibility at high temperatures and the formation of a solid miscibility gap at low temperature. The liquidus=solidus curves may show a maximum or a minimum or may show neither as in the silver-gold system. Liquidus=Solidus Curves, Liquidus-solidus curves which meet only at the melting temperatures of the pure components yield very little direct thermodynamic information. The sliopes of the solidus and liquidus are related to the heats of fusion of the pure components by Eqs. (9b). The simple melting point lowering Eq. (22) does not apply since the solidus slope cannot be neglected, Similarly the Eqs. (13) cannot be simplified. These equations can be employed to calculate T,X points for the liquidus and solidus curves if thermodynamic data for the liquid and solid solutions are available. If both the liquid and solid solutions obey Raoult's law the excess free energy terms are zero and T,X points are simply related to the free energy of fusion of the pure components, If a minimum exists, the two curves are tangent at the minimum and dXfi/dT = dXi/dT = o, and X& = XX, Eq. (9b) becomes igdeter- i minate and it may be shown that this requires that g§- = g7, or Y B ii 208%S 4 208757 X’ X T Eq. (13) shows that the difference in the excess free energies be- comes equal to the respective free energies of fusion and may be calculated from data for the pure components, Other useful relationships may be illustrated by consideration of the following elementary isothermal reactions or processes, 667 A(s) = A(L) AGA A(L) ~A(L,s0ln) 4G (27) A{s,soln) = A(s) - 5, A(s,soln) = A(2,s0ln) AG = 0 The sum of the free energy terms yields Eq. (13), Since Xfi = XR the ideal partial molar free energies cancel and the difference in the excess partial molar free energies may be determined from the free energy of fusion which is the result indicated by (13}, It should be noted that although the free energy of transfer of a mole of A from the solid to the liquid phase is zero this is not true =54 =5=] . for 855 © of AHA ~. The sum of the entropy terms gives = ASfuS + Agz -s _ , fus =xs , 4 =XS,S A A AT ASA = AS + A4S - A . (28) A A A As in the case for the free energy the ideal entropy terms cancel. Stmitarly the melting of the solid solution at the minimum may be analyzed in terms of the following isothermal cycle, XAA(s) + XBB(S) - AXABXB(S) t } ¥ (29) X A(L) + X B(L) - AXABXB(x) The sum of the free energy terms, starting with the top reaction and proceeding clockwise, gives, respectively, y) fus fus _ o= XghGy T = X AG T =0 (30) At the minimum temperature the free energy of fusion is zero and (30) may be written as L _ ,.X5,5 xs, & _ fus fus b= 86%%0% - a6k X\ 86,1 + x 86015, (31) AGS - AG m Therefore a knowledge of the free energies of fusion of the pure components permits a calculation of the difference in the integral free energies of mixing. The corresponding enthalpy and entropy sums are 7 s _ fus fus fus BH- - BH = X, BH, ™" + X BH " - MM (A, B, ) (32) 668 5 £ fus fus fus AS_ - AS = XAASA + XBASB - AS (AXABXB). (33) The ideal entropies of mixing cancel and the left side of the last equation reduces to the difference in the excess entropies of mix- ing, |If the entropy of fusion for the solid solution is estimated by the method outlined by Kubaschewski and Evans2!, the sum of the terms on the right side is approximately zero for a disordered sotid solution, The above relations hold equally well for a maximum in the liquidus-solidus curves, Equation (31) has been employed by Wagner22 to calculate the dif- ference in the excess molar free energy at the liquidus minimum for a number of systems, Miscibility Gap Maximum, At the maximum in a miscibility gap the first and second derivatives of the partial molar free energy and the second and third derivatives of the molar free energy of mixing with respect to composition are all zero as described by Darken and Gurry23. Consequently = -2 and| —— = = (34) aX, X; ax. 2 X, T,P T,P must also be satisfied. As was observed for a liquidus-solidus maximum or minimum dX/dT is infinite and Eq. (9b) becomes indeter- minate. However, at the miscibility gap maximum the sum of the enthalpy terms in {9b) becomes zero and Eqs. {34) apply. For tem- peratures below the critical temperature Eqs. (9b) and (13) must be satisfied, SYSTEMS WITH COMPOUNDS The maximum in a liquidus curve may be associated with a con- gruent melting line compound., The relation of the thermodynamic properties of the liquid phase to those for the solid or compound phase may be formulated in terms of an isothermal cycle such as (29) above. The free energy change for the top reaction in this cycle may now be considered as the standard free energy of formation per gram-atom of the compound instead of per mole of compound. The sum of the free energy terms is identical with (30) with the term AG replaced by the more conventional term AGO(AXABXB) used in relation to the standard free energy of formation of compounds. The enthalpy terms and entropy terms give sums analogous to (32) and (33). For temperatures below the melting point of the compound the terms AG%; AHfi and ASé represent the integral quantities per gram-atom of supercooled liquid with the composition of the compound, 669 Another isothermal cycle which can prove useful involves the equilibrium between the solid compound and the saturated liquid of liquidus composition and temperature., For convenience, consider a mole of compound with the stoichiometry AB, This isothermal, iso- baric cycle may then be expressed as A(2,s0ln) + B(£,soln} = AB(s) ¢ 4 ¥ (35) A(2) + B(4) < AB(4) The liquid solution AB(4) is a hypothetical supercooled liquid with the same composition as the compound or solid phase. The first relation in the cycle implies that the solid compound AB(s) dissoci- ates and dissolves as A and B atoms in the liquidus solution, The sum of the free energy terms is A'Gfus ng - 206, + AG; + AEA =0 (36) and the sum of the enthalpy and entropy terms are respectively, fus =4 _ By + BHy LS - 2AH + AH + My =0 (37) = ks fus ? =4 =& ASAB + ASAB - 288 + AsB + ASA =0 (38) The term Afiig represents the change in enthalpy for the transfer of one mole of AB from the liquid phase to the solid AB phase, AHy represents the molal enthalpy of mixing for the supercooled lqutd of compound composition (XA 0.5 in this case) and the term Afim is the corresponding molar free energy of mixing. Equation (36) may be written as (88, - 85,) + (8B, - ) = -a6y3° (39) or P . x bk (AGXS _ Ast. ) + (AGXS’£ Afi;s’") + RT In _%_% - _AG;;s. XA B If the liquid solution is ideal the excess free energy terms are zero and the relation x:xé’ Ar{gs N f“'5(1' .T) In &8 _ 88 = (40a) Xy Xg RT RT defines the symmetrically spaced ideal-liquidus-points on either side of the compound, Here T¥ is the normal melting temperature 670 o and XA,XB the composition of the compound. At some temperature T below the melting temperature the right side of (40a) may be calcu- lated if the entropy of fusion is known or can be estimated. Let exp -AS:;S(T*-T)/RT = C, then (40a) gives 2 **= Xg = Xg * c(xAxB) 0 (40b) which may be solved for the two values of XB for a particular tem- perature below T, Wagner2h has developed procedures for calculating or estimating the standard free energy of formation of line compounds from phase diagram data and data for the pure components. His relations are more complicated than the above equations and the assumption of ideal or regular solution behavior for the liquid is necessary in order to calculate the free energy of formation. The above rela- tions and procedures have been employed to calculate some thermo- dynamic quantities for alloy systems for which experimental meas- urements were incomplete2>, EMPIRICAL RELATIONS Further calculations of the free energy, enthalpy and entropy for binary solutions requires the use of some parametric relation for the temperature-composition dependence of these quantities, If such a relation is assumed for the relative partial molar excess free energy for one of the components, the relation for the other component as well as relations for the enthalpy and entropy of the solution in general are fixed, A number of calculations have been carried out with the assumption that MRS = (2 + X1 - X,) (412) and a= o +BT, b=¢q' + B'T, and @, &', B, B' and n are constants, Integration of the Gibbs-Duhem equation and basic thermodynamic relations lead to =X5 _ n-1 __n_ 2 85,7 = [a + bX,” (X, - =5 )] Xy (41b) and n A6%% = x. x_ [a + -bx—A ] (41c) m A"B n+1 The assumption that a and b are linear functions of the temperature is valid if Alp is negligible and is a reasonable assumption if the temperature span to be considered is not large, The enthalpy quan- tities corresponding to the three equations are obtained by divid- ing through by the temperature and differentiating with respect to 1/T, The equations have exactly the same form with o and «' replacing a and b respectively. The entropy relations are obtained 671 by differentiating with respect to T. These equations also have the same form with B and B' replacing a and b respectively. The free energy is a function of temperature and composition while the enthalpy and entropy are a function of composition only. The above relations have been used to correlate the measured thermodynamic properties for several binary systemsZ2,2 Equation (4la) reduces to the regular solution approximation when either b or n is zero and to the so-called sub-regular approximation when n is one. When n is noninteger there is no simple relation be- tween the respective parameters when the components are reversed, that is, when the retation (4la) is assumed to hold for the B component, The following equations which are analogous to those proposed by Guggenheim® do not have this limitation and in calculations based on phase diagram data it is immaterial which component is considered to be A or B. These equations when limited to six parameters take the form EXS — 2 - 2 8Gx {a + bX, + (&) & 4000~ \ _~HERSH AND KLEPPA P — e 708 T \\ ? 2000 n=1.00 ! 1 1 I | 1 ! ! ! xAgCI _ 8000} [ &) x > 9 = 2 6000 n=5.00 e * n=1.00 n=075 e ——————— < 4000 = —— l | | | HER?H AND KLEPPR hl_ Q. 0.2 0.3 0.4 05 Q6 oN4 08 09 Xkel —m Fig. 1. Molar Enthalpy of Mixing as Calculated from Phase Diagram y Data and as Determined Experimentally by Hersh and Kleppa? 678 Table VI, Data for LiCl-KC! System, Afixs(LiCI) calculated with Equation (43) and the parameters of Table V. Temp, LiCl=rich xs _ .2 n oK Liquidus B icr = Xeerla * DX ey XKcl n=0,75 n=1,00 n=5,00 Liquidus data 630 0.405 -673 -674 -674 -675 660 0,371 -566 -564 -564 -564 700 0.32] -425 ~424 "N -426 740 0.267 -295 -295 -295 -298 780 0.208 -179 -180 -180 -182 820 0,142 -84 -84 -84 -81 KC]-rich =X ) el n Liquidus Beyey = XLicala * DX Koy = ) X KC1 630 0.412 -1174 -1176 -1182 -1209 670 0.445 -1021 -1021 -1023 -1022 710 0.483 -862 961 -861 -847 750 0.526 ~703 703 -700 686 790 0.576 -544 =544 -541 =542 820 0.618 -1428 =430 -L426 -L4s Miscibility Gap Systems, Some variation in the above procedure was employed in calculating the parameters for several miscibility gap systems. In the following development the solution on the A- rich side of the gap is referred to as the | or prime phase and the solution on the B-rich side as the Il or double prime phase. Only systems in which phases | and 1l have the same structure are con=- sidered, Therefore the free energy of transition term in Eq. (13) is zero and it may be written as =xs,"' AT - AE?S’“ = RT In x:/x; . (45) The substitution of Eqs. (4la) and (41b) in this Equation gives [N ] aAB + bBB = RT In XA/XA (46) and TR aAA + bBA = RT In xB/xB (47) where = 2y, 2414 = (y2yieyMy L 2y iy Ag = D) - D B = (D - (XD and 679 I +1 D1 - Iy (- =T _ n+1 = X A A A (X B 2, 2,001 Ay = () - (x), 8 ] The X values represent mole fractions taken from the A-rich and B-rich phase boundaries at a common temperature or temperature horizontal on the phase diagram, Phase boundary data taken from Hansen and Anderko35 and El1iot36 are given in Table VIl for the systems investigated. The gold- nickel data are for the solid miscibility gap boundaries while the other four sets of data are for liquid miscibility boundaries. The temperature spans for the Cd-Ga and Cu-Pb systems are quite narrow and only three data points were employed for each system in the calculations described below, Table VI, Miscibility Gap Data System System A B A B o I 11 o I 1 K XB xB K XB xB Au - Ni Cd - Ga 623 0.085 0.987 555 0.227 0.725 723 0.130 0.980 560 0.265 0.715 823 0.185 0.970 565 0.350 0.630 923 0.290 0.950 1023 0.480 0.900 Pb - Zn Cu - Pb 692 0,060 0.997 1227 0,147 0.670 773 0,090 0.991 1238 0.173 0.619 873 0.160 0.981 1258 0.236 0.499 973 0,300 0.949 1071 0.720 0.720 Bi - Zn 692.5 0,370 0.994 723 0.421 0.988 773 0.520 0.975 823 0.625 0.944 850 0.690 0.920 At any given temperature both Egs. (46) and (L47) must be satis- fied., Consequently, it was possible to write six relations for the systems with only three data points. A larger number was pos= sible for the other systems. The value of n was arbitrarily varied from 0 to 6.0 in increments of 0,2 and @, B, &' and B' calculated, Except for the Bi-Zn system it was again observed that the n value which gave the lowest standard deviation for the input data (RT Tn Xj/X; for Egs. (46) and (47)) did not yield relations which best descrlbed the enthalpy data. Fairly good results were 680 obtained by assuming a and b to be temperature independent and determining the n value which gave a minimum overall deviation for the complete set of data., This n and mean a and b values were taken to represent the best fit for the input data at the mean temperature. Then employing this value of n, mean values of a and b were computed from the data below the overall mean temperature and from the data above this mean temperature, For the Cd-Ga and Cu-Pb systems the middle temperature was employed in each set of data. This yielded three values of each parameter, a and b, and three mean temperatures for the three groupings of the phase diagram data. The temperature dependence of a was then estimated graphically. A straight line passing through the overall mean a,T point and with a slope which was the mean of the slopes of the two straight lines connecting this central point with the other two a,T points was taken to be the most appropriate representation of the temperature dependence of a or of the corresponding values of « and 8. The parameters o' and B' were determined by the same procedure., The results of these calculations are given in Table VII|Il, The set of parameters for the Bi~-Zn system was determined by the first procedure described above. The thermodynamic properties calculated with these parameters for this system and similar data for the other systems are compared with literature data in Figs, 2-4, The agreement is seen to be reasonably good, The calculated critical temperatures are 1300, 1140, 860, 580, and 12009 for the Au-Ni, Pb-Zn, Bi-Zn, Cd-Ga and Cu-Pb systems respectively, which compare with the literature values of 1085, 1070, 878, 568 and 1203°%K, respectively. The agreement is rather poor for the first two systems. Table VIIIl, Parameters Determined from Miscibility Boundary Data =XS . y2 n AG Xg(a + bX,) A Components Parameters A - B n a 0} Au Ni 1.2 6130 - 0.582T -365 - 6,386T Pb in 0.2 oh24 + 0,9777 -3908 - 3.7327 Bi Zn 0.2 12896 - 5,256T ~14200 + 5,768T cd Ga L,7 2571 5339 Cu Pb 3.8 5525 30424 - 17,277 681 0 | CAlLCULLTEbl o 00— EXPERIMENTAL™ 1750~ P 7 ~200|\— 44 1500 // V- / \ 300Hy- P — _ eso— / \ _ ® / 2 \ o \ E -400}-|- 7 — & o000 \ = \ / 3 \T © -500, /' — g 750 - E / T 3 -600\, )/ — < sool ‘\—- ‘ \_J/ ~— CALCULATED | 700/— - s "~ — -EXPERIMENTAL™ I N I S 0 02 04 06 08 10 O 02 04 06 08 I0 Au XNi Ni Au XNi Ni 0 l CAILC [msoT L r_j —— CALCUL 100 - experMENTALY | '290TT - 4 7N\ -200}|- — 1000}— / \ o / ° / 2 873K / 2 / \ E _300- — E goo— \ S /’ 3 / ‘; -400— ) — geoo— / ~ O / T / 4 -500}— \ // — < 4001— —\- — CALCULATED ) Z _ |/ _ 800~ = 2001/ . - — EXPERIMENTAL | 1 1 ] | . 0 02 04 06 08 10 O 02 04 06 08 10 Bi X0 Zn Bi Xgn Zn Fig. 2. Molar Free Energy and Enthalpy of Mixing for the Gold=Nickel and Bismuth-Zinc Systems, 682 ] 50|~ CALCULATED -- - EXPERIMENTAL -100( \— — [+ - 700K E —|50’—\ - ° \ J 8 -200—\ I £ \ , 3 -250—\ /] \ / -3oor_ N // — I 0O 02 04 06 08 10 Cd XGa Gao ‘ T -100}— — \ 1263K -2001 — | é -300—“ im > \ / S -400— N I— E . / 3 -500— S L—— CALCULATED 600 EexPERIMENTAL T | 0O 02 04 06 08 IO Cu be Pb Fig. 3. AH,, Cal/mole m Cal/mole AH T T T ] 600— /) RN — /4 500— — 400} — 300— N — \ 200 \ ——CALCULATED 100| - - -EXPERIMENTAL —\ l 0 02 04 06 08 10 R (500}— — (250 — —] 1000+— — TSOF— — 500 — — CALCULATED 250(/— -\ [ L1 ] 0O 02 04 06 08 IO Cu be Pb Gallium and Copper-Lead Systems. 683 Molar Free Energy and Enthalpy of Mixing for the Cadmium- ¥89 Fig. L, T 7 T//TI\ / -503}— '/ — / o) | /4 ] r ! 926k E _is0H- ! _ —_ / 3 A . -200(- — o (] < 250"~ — — CALCULATED -300—_ __EXPERIMENTAL ] N I 0 02 04 06 08 1O Pb xZn Zn Molar Free Energy and Enthalpy of Mixing for the Lead-Zinc System, AHm Cal/mole Y, Q o | F N— —— CALCULATED —\ --- EXPERIMENTAL \ | 0O 02 04 06 08 1O Pb xZn Zn SUMMARY AND CONCLUSIONS The thermodynamic properties of equilibrium phases are retated to the phase diagram boundaries but cannot be evaluated from phase boundary data alone. The basic relations {11), (13) and (14) are helpful in correlating experimental thermodynamic data with binary, constant pressure, phase boundary data. Simplifications are possible for various boundary conditions which are encountered in different types of phase diagrams or phase boundaries. Considerable thermo=- dynamic information may be calculated for the liquid phase in simple eutectic systems from the liquidus curve data and the properties of the pure components. Calculations for other binary systems are re- viewed, If a parametric equation is known to describe the temperature- composition dependence of the relative partial molar excess free energy of one of the components, then in principle it is possible to calculate the parameters from phase boundary data and to com- pletely describe the thermodynamic properties of the phases involved, It was shown that the relation =X5 2 2 6B, = XB(a + bX, + ch), A in which a, b, and ¢ are linear functions of temperature, can ade- quately describe the known excess free energy, enthalpy and entropy for a large number of binary solutions, There are six parameters to be determined. The evaluation of these parameters from phase diagram data was not pursued in detail in this investigation., How- ever, considerable work was done in evaluating the parameters in the relation BEX® = Xa(a + bX}) which also has been found to adequately describe a number of binary solutions. Here a and b are linear functions of temperature and the exponent n is assumed to be a constant. With this relation it is necessary to determine only five parameters, The application of this relation to the calculation of the thermodynamic properties for the liquid phase of several eutectic salt systems and eutectic metal systems was investigated, Parameters could be found which closely reproduced the input AGA> and AGXS values calculated from the liquidus T,X data. However, no reliable criterion was found which adequately reproduced the known enthalpy or entropy data. The free energy was not particulariy sensitive to the choice of n and could be closely reproduced with a wide range of parameters. This is not particularly surprising in view of the relation AE?S = oA, - TA§?S from which it is seen that Afi?s can be satisfied by infinite sets of Afii and Ag?s values. At constant composition Afii and Ag?s are 685 both insensitive to small temperature changes. The same problems were encountered in evaluating parameters from liquid or solid miscibility gap data. However, results were suffi- ciently encouraging to warrant further investigations. The calculated enthalpy or entropy values are sensitive to com- putational procedures. The computations carried out in the present work were based on the use of Eq, (13). At least one of the param- eters can be eliminated by combination of Eqs. (13) and (11). Equation (11) also involves phase boundary slopes which are more directly related to the enthalpy or entropy values. These possibil- ities and the Egs. (42) are being investigated, REFERENCES 1. J. E. Ricci, The Phase Rule and Heterogeneous Equilibrium, D. Van Nostrand Co., Inc., New York, (1951), p. 30. 2, G. Tammann, Text Book of Metallography, Translation by R, S, Dean and L, G, Swenson, Chemical Catalog Co., New York, 1925, pp. 166=178, 3. J. H., Hildebrand and R, L, Scott, The Solubility of Nonelectro- lytes, Reinhold Publishing Corp., (1950). 4, J. Lumsden, Thermodynamics of Alloys, The Institute of Metals, London, (1952), 5. M, A, Krupkowski, Bull, Intern, Suppl, Polska Akad. Umiejet=- nosci, Krakow, Ser. A. Vol. 1-4, pp. 15-45 and 219-235, (1950- 51). 6. E. A, Guggenheim, Thermodynamics, 5th ed., North Holland Pub- lishing Co., Amsterdam, 1967, p. 196. 7. C. Wagner, Thermodynamics of Alloys, pp. 47-53, Addison-Wesley Press, Inc., Cambridge, Mass., (1952). 8. R, H, Perry, C, H, Chilton and S, D. Kirkpatrick, editors, Chemical Engineers' Handbook, 4th ed,, Section 13, pp. 8-13, McGraw-=Hi 11 Book Co., Inc., New York, (1963), 9. Kurt Wohl, Trans. Amer, Inst, Chem, Engrs., 42 215 (1946), 10, 0, J. Kleppa, in Liquid Metals and Solidification, pp. 56-86, American Society for Metals, Cleveland, Ohio, (1958). 11, R, A, Oriani and C, B, Alcock, Trans. Met. Soc. AIME 224, T104 (1962). 686 15, 17. 18a. 18b, 18¢c. 18d. 19. 20, 21, 22. 23, 24, 25. 26. E. T. Turkdogan and L, S, Darken, Trans. Met. Soc. AIME 242, 1997 (1968). — (a) M. F, Simmons, Correlation of Thermodynamic Data with Binary Eutectic-Type Phase Diagrams, M,S, thesis, lowa State University Library, lowa State University, Ames, lowa. 1966, (b} USAEC Report 15-1500, National Bureau of Standards, U.S. Department of Commerce, Springfield, Virginia, (1966), p. M-23. B. E, Sundquist, Trans. Met. Soc. AIME, 236, 111 (1966). R. Hiskes and W, A, Tiller, Mater. Sci. Eng., 2, 320 (1967/68), P. S. Rudman, Thermodynamic Analysis and Synthesis of Phase Diagrams, to be published in Advances in Materials Research, Vol, 1V, John Wiley and Sons, New York, N.Y,, 1969, M. Blander and L. E. Topol, Inorganic Chem., 5, 1641 (1966), G, N, Lewis and M, Randall, Thermodynamics and the Free Energy of Chemical Substances, lst edition, McGraw-Hill Book Co., New York, N,Y., 1923, p. 205, A. T. Williamson, Trans, Faraday Soc. 4D 421 (1944). J. G, Kirkwood and |, Oppenheim, Chemical Thermodynamics, McGraw-Hill Book Company, New York, N,Y., 1961, H, F, Franzen and B, C, Gerstein, A.l,Ch.E. Journal 12 364 (1966). L. Kaufman, Acta Met. 7, 575 (1959). P. Roy and L, Kaufman, Acta Met. 13, 1277 (1965). 0., Kubaschewski and E. LL. Evans, Metallurgical Thermochem- istry, 3rd ed,, Pergamon Press, New York, N.Y., 1958, pp. 187-192, C. Wagner, Acta Met, 2, 242 (1954), L. S. Darken and R, W. Gurry, Physical Chemistry of Metals, McGraw-Hill Co., Inc., New York, 1953, p. 330, C. Wagner, Acta Met, 6, 309 (1958). P, Chiotti and R, J, Hecht, Trans. Met. Soc. AIME, 239, 536 (1967). P. Chiotti and J. T. Mason, Trans, Met. Soc. AIME 687 27. 28, 29- 30, 31. 32, 33. 34, 35. 36. R. Hultgren, R, L, Orr, P, D, Anderson and K, K, Kelley, Selected Values of Thermodynamic Properties of Metals and Alloys, John Wiley and Sons, inc., New York, N.Y., 1963, J. W, Mellor, Higher Mathematics, Dover Publications, Inc., 1955, L4th ed., p. 557. L. S. Darken, Trans. Met. Soc. AIME 239, 80 (1967). M. Blander, Thermodynamic Properties of Molten-Salt Solutions, USAEC Report ORNL-3293, 1962. A. Glassner, Thermochemical Properties of Oxides, Fluorides and Chlorides to 2500°K, USAEC Report ANL-5750, 1957, E. Chu and J, J. Egan, New York Academy of Sciences Annals 79, No, 11, 908 (1960). C. E. Wicks and F, E, Block, Thermodynamic Properties of 65 Elements: Their Oxides, Carbides and Nitrides, U.S. Bureau of Mines Bulletin 605, 1963, L. S. Hersch and 0. J. Kleppa, J, Chem. Phys, 42, 1309 (1965). M. Hansen and K, Anderko, Constitution of Binary Alloys, 2nd ed., McGraw-Hill Book Co., New York, N.Y,, 1958, R. P, Elliot, Constitution of Binary Alloys, First Supplement, McGraw-Hill Book Co., New York, N.Y., 1965, 688 TIERMODYIIAMIC ANALYSIS AND SYNTHESIS OF PHASE DIAGRAMS Peter S, Rudmen Department of Physics Technion - Israel Institute of Technology Haifa, Israel Computer calculations for synthesig, the deriving of phase diacrams from solution thermodynamic date are described, Computer caleulations for analysis, the deriving of solution thermodynamics from phase diagrams, are described. Consideration is limited to substanticlly disordered, binary systems, Examples ol calculationc for the Au-Ii, Al-Sn and Ag-Au systems are presented, 689 Introduction Let us consider the wideiy used free energy-common tangent construction for determining phase limits in a binary system, In Fig.1 are schematically presented such curves at some temperature T, fer equilibria among the phases &, s and ¥ in the bina}y system A-B, Also presented is a phase diagram which is consistent with the free energy curves at T., Clearly, at each temp- erature vhere we know the ffee energies of each phase we can construct the phase limits, We shall call this proceedure synthesis, The free energies F,, F, and F, are defined for all compositions while the phase diagram defines only a limited number of discrete compositions at each temp- erature, Thus at each temperature the thermodynamic information content of & phase diagram is much less than we need to know, Hovever, possibly this infor- mation content can be increased sufficiently by con- sidering many temperatures to allow the derivation of the thermodynemic functions, We shall call this proceedure analysis, The synthesis and analysis compubations are tecdious and not practical without an electronic computer. Eerein we shall formulate the thermodynamic equations, discuss their solution, and present some examples of computer syntheses and analyses for the real systems Au-Ni, Al-Sn and Ag-Au, A detailed report of this work, including computer programs, is being published elsewhere (1), Thermodynamic Formulation To perform both the synthesis and analysis operations it is necessary to analytically express the free energy as a function of composition and temperature, We shall generate all of the necessary thermodynamic functions from a relative integral molar free energy of the form F' = (1-x)aF, + xaFy + F> - 1870, (1) where AFE[AFB) = free energy of transformation of pure A(B) from reference structure to phase structure, SID = ideal integral molar configurational entropy, Xs _ . F = excess integral molar free energy. 690 Composition ('L o) Abiaugz 8ai4 3injpiadway A by Free Energy~-Common Tangent Construction 1« Phase Diagram Graphical Synthesis 691 EQe (1) is clearly best applicable to systems that are substantially disordered and present consideration has becen so limited., The crucial step in employing Eq. (1) is the choice of the analytical form for F°°, We shall limit conslderation to expansions of the form M FXS = x(1=x) 2%% afi =, (2) where " F = m th expansion consteant, Thus (M +1) expansion constants are employed to define the compositlon dependence of FA at each temperature, Eqe.(2) is not a very original choice, It is essent- 1ally the expansion introduced by Margules in 1895 and it is by far the most widely employed form today. The usefulness of Eqe(2) derives from the fact that it generally well represents the data with a small number of constants, It is particularly important in analysis to minimize the number of expansion ccnstants. There is no proof that Eq.(2) possesses the minimum number of con- stants nor that it is the best of all peossible forms, Indeed many other expansion forms have becen employed (2-) for thermodynamic data extrapolation, but whether or not they possess any advantage for analysis remains to be tested., The tempcrature dependence of FXS enters thrcough the expansion ccnstants ai(T). Perhaps the most logical F proceedure is to similarly expand the a 's in a power series in T, and this is in fact Jjust what Hiskes and Tiller (5) have done, However, in order to more readily relate to thermodynamic data tabulations we have based our itemperature expansion on the relation P = B - 1) );c dT-Tf (c®S/myar, (3 where TO = reference temperature, H§(S°XS)= excess enthalpy (entropy) at To ’ c®S = excess specific heat. xS The expansion of C in T determines to a large extent 692 the generallity of the applicgtion. In principle by employing a sufficiently genersl expansion for C (T) we could include systems with specific heat anomalies (Curie points, order-disorder transformctionS,e..). However, at this stage of progress we are having suffic- lent problems with simple systems and we have not tried to apply the calculotions to systems where the specific heat is a function of temperature, It thus follows from Eq,(3) that the most general form for the temperature dependence of the expansion constants that we have used 1s s ) (& aIFn'_ = a;'i nTBISn + T!a:]' 2 (L]-) where Tt = (T-T.) - T4 n(T/T ), ° 8 s gi ’ gg = temperature independent expansion EM XS XS . constants for s S amd C in expansions of the form of Ege(2). When only the first.4Mgt1) terms ira&g. (k) are non- zero, we say that we have carried out an expansion to Order(MT). We similarly define the complete composition- femperature expansion as of Order(Mx,MT), and which will require (MX+1)(MT+1) expansion constants, In Table 1 the present notation is related to various nomenclatures that have fregquently been used to describe solution thermodynamics behavior, Table 1. Solution Thermodynamics Nomenclature Conventional Present Notation Nos, Expan. Const., (Mx+’l ) (MTH ) Ideal Order(~-1,0) 0 Regular- Quasi-Chemn, Order(0,0) 1 Sub=-regular Order(1,0) 2 Regulear Order(Mx,O) Mx+1 (Most general Order(M,,2) 3(Mx+1) expansion of present scheme) 693 The difficulty of the calculations by pre-computer methods 1s perhaps best exemplified by the fact that it has long been appreciated that an Order(1,1) expansion generally provides the minimum realistic description for a binary alloy system, but neither syntheses nor analyses have been reported for non-reguler solutions, thet is, for MT70 ° To complete the desoription of the temperature depen- dence of the free energy function of Eq.{(1), we anzlog- ously emplcy _ Al t AF, = AH,(1-T/T,) + T}4C, » (5) where AHK = enthalpy of transformation of pure A = ( H, (phase structure) = H,(ref. stre) )m_mb » A A T—TA ACA = change in specific heat of pure A on trans., » Tz = equilibrium transformation temp. for pure A » t t ¥ = - - Ty = (2-T,) Tfn(T/TA) . Analogous expressions hold for the transforrmotion of B, SInthesis The miscibility cap is the simplest system fo treat since it involves the thermodynamics of a singlc phase only, Consider the miscibility gap of Fig.2a: we wish to calculate the gap limits X, and Xy at arbitrary T, The usual graphical common tanent method is i1ilustrated in Fig.2b ., Another method of solution is by simultan- eously solving the pair of partisl molar free energy equations: F’I\g(f}:,‘,T) - FI:(XE,T) =0 ] (63-) i F%(x1,T) - F%(XE,T) 0 (6by At each T we thus have two equations in thz two unknowms X, and Xy 0 A graphical solution is illustrated ir Fig.zc. In order to numerically solve Eqs,(6} we must first explicitly express the partial molar free energles in terms of the expansions given previously., Thic is readilly accomplished by employing the well knovm 694 1000~ Temperature, °K cal. /mole cal./mole 2e Miscibility Gap Graphical Synthesis by Pree Energy-Common Tangent ard Partial Molar Free Energy Constructions 695 Fis = 75 L r% x , (72) F‘gs = PP o (1-x)9F%% ox (7b) yielding (8a) ¥ M 2 F r m A x Z__ (m+‘|)(am = A )x s =0 M xS (1-x)2‘2:; (m+1)a§ = . (8b) m=0 ! I /nalogously the ideal partial molar free energles are readlly generated ylelding D F.0 = Rfn(1-x) , (9a) FéD = RTAn x . (9b By definition we have M _ XS ID {104a) Fy=Fp t T P o S L sIP (10) Our proceedure 1s to take the experimental thermo- dynamic data, least squares f£it it to expansions of the form given previously and generate the (Mx+1)(MT+1) expansion constants, and then to solve the simultaneous pair, Eqs,(6), for X and xo for various assumed btemp- eratures, The simultaneous pair is transcendental and cannot be solved by elementary mears but they are readily solved by a computer by numericzal proceedures, Tn Pig.3 is shown the misclbility gap in the Au-Ni system, The experimental gap(6) and the gap synthsized from thermodynamic data(7) can be compared. It is quite clear that there is an inconsistency but is not our purpose here to attempt to evaluate the source of error. Let us now consider the general case of synthesis of equilibria between phases of different structures: phase 1 and phase 2 , Agaln the boundary compositions X, and x_ arc determined by the simmltaneocus solution ot a paig of partial molar frce energy equations which we now write as 696 L69 1200 o o o - o X X © ® [ ] o X ] L J 1000 o X ° L J o . x L ] ° - o X e )-'( - ) e Experimental Gap L g o Xe > 4 = . O Order (I,1) Synthesis from Experimental . @ 800~ °© Xe Thermodynamics « (=% ® « E o oy 2 o X X Order (I,l) Synthesis from Thermodynamics of ® [ | ox o Order (i,1) Analysis of Averaged Phase Oiagram Qd . < ox X 600 X ] ] | i | . 1 | Au 2 4 6 8 NI X, Atomic Fraction 3. The Au-Ni Misecibility Gap Ff1(x1’T) - Ffiz(xz’T) = 0 » (113) F§1(X1,T) - F%z(XZ,T) =0 , (11by If we adopt the structure of phase 1 as the reference structure, then thre partial molar frce energies take the form: For phase 1 }I - P o= P rL, (122) Fp, = Fio + FLD | (12Db) For phase 2 F= 7S D sar, (130 ] S, . . . . Fya (Fle) is given by Eq.(8a) with the expansion const- ants anpropriate for phase 1 (phase 2) and Ffi? (Fgg) is given by Eg,(8b) with the expansion constants appropriate for phase 1 (phase 2). FiD and FéD are given by Eqs.(9), and 4F , (AFB) are given by Eqs.(5). In Fig.li is shown the liquid/solid equilibrium for the Ag-Au system with both the experimentally determined boundaries and the boundaries that we have computer synthesized from thermodynaemic data(7). The liquidus? agree fairly well but the solidus! differ badly. Again we shall not attent to find the origin of the dlscrepency. There is a particular case of equilibria with a structure change that we must consider seperately: when one of the terminal solubilities approaches zero., Let us take phase 1 as the zero solubility phase. Then there is effectively only one unknown, X9 the (A+2)/2 boundary, As x1fl’0, F£Q+ -00, so0 that from Eq.(12b) Fg1 becomes indetcrminant., Thus we eliminate Eq.(12b) from consicderation and Eq,(12a) alone suffices to determine the one unknown Xx,. This reduces to solving Fap = 0 » (14) with Fflz given by EQ.(133)0 698 669 °K Temperature , 1340 L ° f— as [ — . ° XX — oo — ™ e XX f— 00 ° . X X _1 1300 o0 —_ ] e XX - 00 — ° ° X X — 00 — X X — LoR o) X; ¢ Experimental Phase Boundaries — 1260}— oo O Order {(1,l) Synthesis from Experimental — XX Thermodynamics X Exact Synthesis, ldeal Solutions h—(D — | Assumed 220 | | | | 1 | Ag .2 4 6 8 Au X, Atomic Fraction lte The Lgd/(Lad + fce) and (ILgd + fec)/fec Phase Boundaries of the Ag~Au System Mg.5 presents an example of a system of this type,the liquid/solid equilibria of the Al-Sn system. Here we find very good agreement between the experimentally determined liquicdus(6)} and our computer synthesis from thermodynamic data(7)., While this agreement does not conatitute & proof th& beoth dabtas are correct, it certainly raises confidence in their reliability, Anglysis Again as in synthesis let us first treat the simplest . case, the miscibility gap. The miscibility gap limits xgl) and xél) are kXnown at each temperature T(l). Again EqS. (6) provide the solution but now the unknowns are the ' (Mx+1)(MT+1) expansion constants, the a_ '3, Since Eqs.(6) provide two equations at each temperature, solutions to Order(Mx,MT) can be obtained by applying Eqs.(6) to at least Nmin=(Mx+1)(MT+1Y2 different temp=- eratures, In order to include data from an arbitrary number of temperatures szNmin s we perform a least squares fit, At each temperature T(i) we let the LHS!'s of Egs.(6) be v#l) and vél) respectively, We then form N . . = T leiihE e wih?] (15) =1 and (Mx+1)(M’I‘+1) equations are generated by the operations . 087282 = 0, (16 ) ¢ i H° E° 8° s° with afi = 8, through.aMx s 25 throughaMX and C o & through aMx. The (Mx+1)(MT+1) equations so generated are linear in the unknowns, the Q's, and can trus be readily solved by well developed natrix methods, However, as the number of unknowns increeses we have found that the solutions tend to become erratic., A detalled study of the stablility of the analysis calculation 1is greatly to be desired but yet to be done, Nevertheless 700 10L Temperature, °K 1000 p L 3 -— O @ _I O ] «© «0 800— 0 — s — e Experimental Phase Boundary O —_ Qe Qe © Order (I,1} Synthesis from Experimental Thermodynamics o 600— o — L ® I I | I [ I I I l Al .2 .4 .6 .8 Sn X, Atomic Fraction 5. The (Al + Lgd)/Lgd Phase Boundary of the Al-Sn System in a rough and intultive way we can sum-up our experience by defining a reliability factor R: Ea f_\_f T J > n Li/zooo N, A% (17) where Ni = nos, phase diagram composition inputs, No = nos, of analysis output constants = (Mx+1)(MT+1) for miscibility gap analysis , Ax = experimental error in composition input data, If R» 1, the analysis is relidble, while if R & 1 . then the analysis is unrcliable. In Table 2 are presented the enthalpy and entropy ‘ data for the Au-Ni feec solution, both from an Order(1,1) analysis of the experimentally determined gap(6é) and from directly determined thermodynamic data(7). With N;=3k, 4x¥,005 and No=u we get from Eq.(17) that R¥.85, that is, the analysis is on the threshold of reliability. Table 2, The Relative Integral Molar Enthalpy and Entropy at 1150 K of fec Au-Ni Alloys HE, cal/mole Sg, cal/mole=-deg | e gl | e s 2 970 1121 1.38 1.70 ) ru 1590 1879 1.98 2.143 . 6 1780 2075 2.0l 2.16 . L & 1215 1515 138 1.76 Our previous consideration of systhesis of this system showed that the experimental thermodynemics is not consistent with the experimental phase diagram, so it 1s not surprising that the analysis cderived thermo- dynamics do not agree better with experimental thermo=- dynamics, While the discrepency is not large, still to Judge which is the most relieble is bcyond our present competence, The formal extension of analysis to equilibria between phases of different structure is a straight- 702 forward extenslon of the miscibility gap analysis, Although we have developed such computer programs, in ractice the results have been unreliable, This can at east be partially understood by consideration of the reliability faator R, There will be at least twice as ety constants to bs determined per input point, There will in general be (Mx+1)(MT+1) constants for each phase and possibly a few unknowns for the transfo_mations phase 1~> phase 2 of pure A and pure B {see E£g.(5) ) Thus at the mement the general applicaticn of analysis 1s not possible., However, since we co not reclly under- stand the origin of the instebilivies in che calculation, we have really no feelin; for whether they are insur- mountable or whether better algorithms and other expan- sions can lead to reliable solutions, It is clear that the phase diagram literature is a large cnd largely untaped source of valuable thermodynamic data. Thus the rewards for developing successful analysis arc ;5 eat indeed and the pursult of further investigations 1into the mathematics of analysis appear to be called for, References 1. Rudman, P, S.,"Thermodynamic Analysis and Synthesis of Phase Diagrams", Advances in Materials Research, Vol, IV, John Wiley (1969) New York, 24 Hildebranc, J. H, and Scott, R. L.,The Solubility of Nonelectrolytes, Reinhold (1950) Iliew York. 3¢ Lumscen, J., Thermodynomics of Alloys,The Institute ol lletals (1952) London, i. Chiotti, P, and Hecht, R. J., TS-AIME 239,536(1967). 5, Hiskes, R, and Tiller, 'I, A,., llater, Sci, Ing, 25320(1967/68)., , Constitution of Binary 6. Hansen, M, and Anderko . )} New York, L] AMloys, MeGraw-Hill (195 K 8 7o Hultgren, R., Crr, N. L., Anderson, ?. D. and Kelley, X. Koy Thermodyncmic Properties of Iletals and Alloys, John Jiley (1963) llew York, 703 PREDICTION OF TERNARY THERMODYNAMIC PROPERTIES AND PHASE DIAGRAMS FROM BINARY DATA* N.J. Olson Institute for Atomic Research and Department of Metallurgy, Iowa State University, Ames, lowa, 50010 U.S. A. G. W. Toop Technical Research Centre, Cominco Ltd., Trail, British Columbia, Canada With the use of binary data, excess molar free energy values and phase diagrams have been calculated for a number of ternary systems. The equations used are considered to be rigorous for regular ternary systems and empirical for nonregular systems. The calculations are compared with experimental results where possible., These results are found to give reasonable estimates of AFXS for the Pb-Sn-Cd, Sb-Cd-Pb, Bi-Cd-Sn, and Cd-Pb-Bi systems at 773°K, the Ca0-Si0,-FeO system at 1873°K, and the Fe-Mn-Ni system at 1232°K. Ifi'xase diagrams have been calcu- lated for the Pb-Sn-Zn system at 926°K and the Ag-Pd-Cu system at 1000°K. *Work performed in part at the Ames Laboratory of the U. S. Atomic Energy Commission. Contribution No. 2549. 705 Introduction Darken's work(l) provided the pertinent beginning for the math- ematical treatment of ternary thermodynamics. He showed how the Gibbs-Duhem equation may be used to calculate the excess molar free energy of a ternary system from experimental excess partial f%)lar free energy of one comp?cfient in the ternary. Later Wagner*©/, Schuhmannl( » and Gokcen derived other useful ternary relationships dealing, essentially, with mathematically consistent solutions of the same problem. The above approaches minimize the amount of experimental data needed to define the thermodynamic properties of a given ternary, and they may be extended to higher order, multicom- ponent systems. However, it would be useful to find methods for predicting the thermodynamic properties of a multicomponent system from a limited amount of data as a supplement to experi- mental measurements and analytical studies. This has been re- cognized with the considerable work that has been done to find models to predict or extend thermodynamic properties and phase diagrams, throu%’x gsrious interpolative schemes, in binary and ternary systems'” 7Y/, Equations derived from Darken's wo }1((1) ay be used to pre- dict te lar f es 18-19) i {él&ry excess molar free energies and phase dia- grams from binary data. This work is reviewed and addition- al calculations on the Pb-Sn-Cd, Cd-Pb-Bi and Ca0-8i0,-FeO systems are presented. Although the equations are rigorous only for regular solutions, they give results in fair agreement with ex- perimental measurements for several nonregular systems. Calculation of Ternary Excess Free Energy Using Binary Data and the Regular Solution Model The excess molar free energy of a ternary solution (AF*%) in terms of the activity coefficient ¥, of component 2 may be written as o In vy, dN T -N,)?2 2 L“_AO'—‘L‘—':OHC-———;Z Z [ ~~ 2z (W) (1 - Ny 2(3) (1) fa— 706 Yy = ternary activity coefficient of component 2 Ni = ternary mole fraction of component i yi(j) activity coefficient of component i in the i - j binary Ni(j) mole fraction of component i in the i - j binary When the binary data are known, Eq. (1) lends itself directly to interpolative approaches because an estimate of the first integral in the equation is all that is needed to calculate AF¥X®., However, the physical significance of any such estimate must be given serious consideratio%. There are perhaps many ways of solving this kind of problem( 1), but a good beginning would be to make any estimate of In y, meet the following conditions: [ln Y, = In 72(1)} N3 = 0 [1“ Yo = In 72(3)] N, = 0 (2) [AFXS = (AFT:)Nl/Ns] N, = 0 That is, any analytical substitution of In ¥, into Eq. (1) must force Eq. (1) to collocate with the known, ginary’, boundary data. It has been found(18) that the following substitution for ln v, into Eq. (1) would fulfill these conditions, AFXS (- Nz’z[ P\1T-3] Nl/N3 . (3) XS XS AP L [N1 aF, ) s N3 AF.2-3] RT N, RT N, TRT N, X5 F1-N)% A RT N /N, (4) 707 The terms in Eqs. {3) and (4) are defined in Fig. 1l(a). If the 2-1, 2-3, and 1-3 binaries are regular solutions for which In 'yi(jg = sds ozi ] (lh(lfg()j)) and ai—j is a constant, then Eq. (4) simplifie follows ’ %5 AF RT - MiNo%.p * NpNy%p 5 + NiNjey o0 (5) Therefore, it is apparent that the introduction of Eq. (3) into Darken's Eq. (1) gives the ternary excess molar free energy in terms of the & function in a form characteristic of a regular solu- tion(5» 15) and it is for this case that Eq. (3) is considered to be rigorous. At N2 = 0, Eq. (3) reduces to the equation presented b Darken“), Alcock and Richardson - , and Oriani and Alcockh5) to determine the activity coefficient of component 2 in dilute solu- tion with components 1 and 3. Examination of Eqs. (3) and (4) indicates that they are path dependent for nonregular solutions, i.e., the calculated value of the ternary excess molar free energy will depend on the choice of component 2. Hence, it would be desirable to obtain similar ex- pressions which are path independent. This may be done readily by chosing the path geometry shown in Fig. 1(b) and making sub- stitutions into Eq. (5) as follows, XS |:AF2-1 NN “2-1] RT N, T-X, NN, (6) The expression for the ternary excess molar free energy accord- ing to this geometry is XS X5 xS A - q-np? | A + 1 -N)2 A2 o ’ RT N, /N RT N,/N 1/ Ny 2/ N3 xS + -N2)2 [AFI—3J RT ’ Ny /N, (7) and, the resultant ternary expression for In Yo is (19), XS . 2-1 In Y, = |:(1 - N3) In Y1) + NS (1 - N3) BT :lN /N, XS 2-3 +[(1 - Nl) In y2(3) + Nl(l - Nl) BT ] /N ) AFXS NZ 3 - (1 - N,) 1-3 RT JN,/N, (8) RT lN,/Ng, Fig. 1(a) - Definition and location of terms used on the ternary composition diagram for Eqs. (3) and (4). Fig. 1(b) - Definition and location of terms used on the ternary composition diagram for Egs. (7) and (8). 709 The terms of Egs. (7) and (8) are shown in Fig. 1(b). In summary, Eqs. (3), (4), (7) and (8) meet the collocation cri- teria defined in Eq. (2). The introduction of either Eq. (3) or Eq. (8) into Eq. (1) makes the first integral on the right-hand side of Eq. (1), when integrated from N, = 1 to Ny =0, have the same numerical value as it would have in the actual system, even if the system is nonregular. This means that the terminal slope, at N, = 1, of the AF*® surface along a path with constant Ny /N3 is the same as for the actual system. However, the integral would be exact, in general, when integrated from Ny, = l to any NZ only for regular systems. The possible graphical difference between the calculated and actual integrals is shown schematically in Fig. 2. Since the calculated AF*S surface using either Eq. (3) or (8)in Eq. (1) reduces to the known binary values, it is expected that the most accurate calculated data would be near the sides and apexes of the ternary triangle and the largest errors would occur in the central regions. It would now be instructive to calculate the ternary excess mo- lar free energy of several systems where experimental results are known. Using Eq. (7), results were calculated for five sy- stems and they are shown in Figs. 3-7. Liquid standard states of the pure components were used in each case. The experimental binary and ternary data for the Pb-Sn-Cd and Cd-Pb-Bi systems at 773°K, Figs. 3 and 4, were taken from Elliott and Chipman(?’z), and experimental data for the Ca0-5i0,-Fe0 system at 1873°K, Fig. 5, has been summarized by Elliott 33)2. For the Sb-Cd-Pb system at 773°K, the experimental binary and ternary data has also been determined by Elliott and Chipman(32). The calculated results for the Bi-Cd-Sn system at 773°K are shown in Fig. Z Z?d compared with the experimental ternary results by Melgren 3 Experi- mental binary data for the Bi-Sn system were taken from Melgren, and Elliott and Chipman's Bi-Cd and Cd-Sn data were used for the calculations. Eq. (7) was used for the calculations because of its path inde- pendent nature. However, Eq. (4) will produce similar results if the calculated excess free energy curve is taken as the mean from three integration paths corresponding to the selection of each component as component 2. This may be verified by matching the appropriate excess freje energy contours in Figs. 3-5 with Figs. 3-8 in Toop's work(18), Elaboration on the advantages and dis- advantages of using Egs. (4) or (7) for the purpose of calculations is given in the discussion. Calculation of Ternary Phase Diagrams The regular solution model has been used to calculate common tangent points to ternary free energy of mixing surfaces in order to determine phase boundaries in ternary systems involving 710 In Yaq) U-N7“fl2 n 7p / (l‘Nz)z J POSSIBLE ERROR ) x} = (1-N2) AREA 2 Ni/N - N)AREA 1 3 | - N3 AREA 3 Fig. 2 - Possible graphical difference between the actual ternary integral in Eq. (1) (solid line, area 2) and that calculated from Eq. (3) or (8) (dashed line) when integrated from N2 =1to N2 = 0. Neg — Fig. 3 - Calculated AF*°(cal/mole) using Eq. (7) (dashed line) compared with measured values{unbroken lines) for the Pb-Sn-Cd system at 7T73°K. 711 Pb 300 4 400 /200 300 8 200 9 s 0 Cd Lo e | 2 3 Ng, ——» Fig. 4 - Calculated AF™® (cal/mole) using Eq. (7) (dashed lines) compared with measured values (unbroken lines) for the Cd-Pb-B1 system at 773°K. 5102 CaQ Fig. 5 - Calculated AF~°/4.575 T using Eq. (7) (dashed lines) compared with measured values (unbroken Iines) for the CaO—S102—FeO system at 1873°K. 712 Fig. 6 - Calculated AF™® (cal/mole) using Eq. (7) (dashed line) compared with measured values (unbroken lines) for the Sb-Cd-Pb system at 773°K. Fig. 7 - Calculated AFXS {cal/mole} using Eq. (7) (dashed lines) compared with measured values (unbroken lines) for the Bi-Cd-Sn system at 773°K. 713 miscibility gaps(5' 10, 24), Meijering(s’ 10) used Eq. {5) to calcu- late ternary excess free energy of mixing values after he had de- termined appropriate @j.j constants. An alternate expression which gives AF*® for regular solutions as a function of binary values of AFX% along composition paths with constant Ny/N,, N,/N3 and Nl/l\'}3 has been given in Eq. (7). This expressu)n for AFXS is more useful for the empirical calculation of ternary excess free energy values for nonregular systems because actual binary AF*S data may be used in the expression rather than attempting to fmd suitable constants for Eq. (5). Further results of this feature of Eq. (7) are illustrated in Table I where calculated excess free energy values for the Ni-Mn~Fe system_at 1232°K are compared with experimental data of Smith, et al. . Table I. Calculated AF*® values using Eq. (7), in parentheses, compared with measured values for the Ni-Mn-Fe system at ‘ 1232°K. All of the excess free energy values are negative. NMn Npe/ Ny o/1 /9 3/1 11 /3 9/t 1/0 0 0 979 2891 3660 2894 1760 0 (0) (979) (2891) (3660) (2894) (1760) (0) 0.10 727 1506 3038 3636 2908 1866 272 (727) (1450) (2920) {(3450) (2790) (1700) (272) 0.20 1420 2022 3180 3513 2788 1809 379 (1420) (1950) (2980) (3283) (2610) (1600) (379) 0. 30 2025 2448 3217 3349 2593 1679 406 (2025) (2370) (2990) (3020) (2310} (1380) (406) 0.40 2441 2666 3092 3072 2353 1515 404 . (2441) (2590) (2820) (2714) (2030) (1140) (404) 0.50 2551 2620 2850 2744 2081 1324 379 ‘ (2551) (2560) (2560) (2340) (1700) (920) (379) 0.60 2402 2394 2495 2353 1763 1107 338 (2402) (2380) (2210) (1926) (1420) (730) (338) 0.70 2056 2010 2034 1880 1395 867 .- (2056) (1980) (1780) (1430) (1000) (500) (287) 0.80 1513 1466 1440 1307 - 610 220 (1513) (1410) (1180) (931) (660) (380) (220) 0.90 791 744 708 617 453 291 108 (791) (710) (580) (450) (310) (180) (108) 714 Although regular solution equations have been shown, in this paper, to give calculated thermodynamic quantities which agree quite well with experiment for single-phase nonregular ternary systems, care should be exercised i1n the use of the equations to predict thermodynamic properties of multiphase ternary systems in which strong compound formation 1s suspected. This precaution 15 consistent with the simple regular solution model which for ne- gative values of @, . will indicate a tendency toward compound for- mation, but even very large negative values of @ _. will not give multiphase binary or ternary systems involving a Jdlstlnct stable compound. Hence, calculated ternary free energy data using Eq. (7) mMght be expected to vary between being rigorous and poor, 1in the following order, for ternary systems which are. a) regular solutions, b) nonregular single-phase liquids in which random mixing 1s nearly realized, c¢) nonregular single-phase solids, d) nonregular multiphase systems with binary compounds but no ter- nary compounds, f) nonregulat multiphase systems with highly stable binary and ternary compounds. The calculated data will be expected to be least accurate for the last two cases. To empirically predict phase boundaries in ternary solutions with the use of Eq. (7), the problem involves calculating and dis- playing the free energy of the mixing (AF) surface and then deter- mining common tangent points or tie lines on this surface. To achieve this, a method for graphically contouring the free energy of mixing surface was considered. In Fig. 8, calculated AF/RT contours are shown for a hypothetical regular ternary system in which all the ¢ _; values are constant and equal to 3. The contours were drawn with'a mechanical plotting device and the phase diagram consisting of three single -phase regions, three two-phase regions and one three-phase region, ts clearly defined. Because of the symmetry of this system, the tie lines across the two phase re-~ grons are parallel to the triangle sides. For more complex non- regular ternary systems in which the directions of the tie lines were unknown, the contouring method was found to be less promising. The method adopted 1n this paper 1s more general and involves two-dimensional plots of ternary activity curves. The principle used 15 that tie lines indicating two-phase equilibria join conjugate phases @ and B for example, for which aj{a) = aj(8), az(@) = a(8) and a3(a) = a3(B). Tie lines may be determined by plotting the ternary activities of two components along an i1soactivity line for the third component, and the unique points where the above equalities hold may be found graphically. The ternary activities of all three com- ponents may be found with the use of Eq. (8). For the present work, this method was used to calculate the phase diagrams for the Pb- Sn-Zn system at 926°K and the Ag-Pd-Cu system at 1000°K. The Pb-Sn-Zn System at 926°K 1(6) Lumsden's mode has been shown to define the the rmodynamic 715 proFf t1es of the Pb-Zn system with a liquid phase miscibility gap at 926°K. The miscibility gap should extend into the Pb-Sn-Zn ternar f)iom the Pb- Zn binary as illustrated by exper- imental evidence at 793°K. Using Eqs. (7) and (8) it is possible to calculate the ternary liguid miscibility gap with Lumsden's eq- uations for the Pb-Zn system and appropriate binary data for the liguid phases in the Pb-Sn and Sn-Zn systems at 926°K. The AF:® a(ta6 for the Pb-Sn system at 773°K was taken from Hultgren, et al(3 Since this system is considered to be regular, the necessary free energy data at 926°K could be obtained readily by assuming that the excess free energy is independent of tempera- ture. Free energy data for the Sn-Zn system is available(36) at 700°K The integral molar free energy of mixing, AF, for this system was calculated at 926°K b)g assuming that the AC,, values for the various alloy compos1t10ns were independent of temperature between 700°K and 926°K. Using the appropriate binary data, the isoactivity lines for zinc, shown in Fig. 9, were calculated using Eq. (8). By com- puting apy and ag along an isoactivity line for Zn, there must be one unique pair of composition points where ap (L ) = app(Ls) and agn{Ljy) = agp(Lp) which defines the two-phase %oundary as shown in Fig. 10. After using the same procedure along all of the iso- activity lines of Zn, the entire miscibility gap was calculated and it is defined by the dashed line in Fig. 9. The final calculated phase diagram for the Pb-Sn-Zn system at 926°K is shown in Fig. 11. The shape of the calculated miscibility gap at 926°K compares reasonably well with the shape of the miscibility gap at 793°K as shown in Fig. 11. In drawing the latter curve, the Pb-rich phase boundary was modifie% slightly to be consistent with the cur- rent Pb-Zn phase diagram The Ag-Pd-Cu System at 1000°K The Ag-Pd-Cu system should contain a region of solid phase immiscibility at 1000°K due to the miscibility gap in the Ag-Cu system. Since no experimental phase equilibria exists for this system, to the authors' knowledge, the ternary miscibility gap at 1000°K was predicted using Eqs. (7) and (8). Both the Ag-Pd and Pd-Cu systerms are single-phase (fCéI) at this temperature and the rmodynamic data are available(3 Thermodynamic data for the terminal regions of the Ag-Cu system are available at 1052°K so that extrapolations had to be made over the two-phase region. For the purpose of extrapolation, it is useful to use the parameter @, defined as, AFTS. N, @ .+ N, @, . g . = ==l = 1) j 14 (9) i-] RT N.le (36) 716 Fig. & - Fig. 9 - Contours of the AF/RT surface and the resulting phase diagram for a hypothetical regular ternary system with binary a values equal to 3. "‘_NZn_""" Calculated a,,_ values in the Pb-Sn-Zn system at 926°K with predmte%{ miscibility gap (dashed line and triangles). 717 o T S oacTivITY MISCIBILITY GAP —— — . —— W — ——— — — — —— — A .2 .3 .4 5 .6 7 .8 .9 1.0 Ny, /(1= Ng,) — Fig. 10 - Graphical determination of a tie line indicating two- phase equilibria along an iscactivity line for Zn in the Pb-Sn-Zn system at 926°K. 718 where the &'s need not be constant. In the Ag-Cu system at 1052°K, it was found that § is constant in both of the terminal single -phase regions and it was assumed that &Ffggeg = AF{§52°K in these regions. In the two-phase region of this binary, it was assumed that @ varies linearly between the constant value 2. 86 on the Ag-rich side, and 3.48 on the Cu-rich side. This gave values of @ over the entire system and permitted calculation of a AF versus composition curve for the Ag-Cu system at 1000°K. The AF curve is shown in Fig. 12 and it a(ggsees well with the phase diagram adopted by Hultgren, et al . Isocactivity curves for Cu were calculated using Eq. (8) and the ternary miscibility gap was established in the same manner used for the Pb-Sn-Zn system. The phase diagram, given in Figs. 13 and 14, shows a predicted miscibility gap which extends over a substantial region of the ternary. Discussion The equations presented in this paper are considered to be rigorously applicable to regular ternary solutions. Although the method is empirical for nonregular ternary systems, the expres- sions give a fair approximation of measured ternary excess free energy for the systems shown in Figs. 3-7 and Table I. There are several reasons which help one to decide whether to use Eqs. (3) and (4), or Eqs. (7) and (8). First, one should keep in mind that Eqs. (3) and (4) are path dependent, but this factor has not b E.'él) shown to be a major contribution to errors in the calculations!18). The use of Eq. (4) along a single path differs from the results calculated with Eq. (7) for the systems in Figs. 3-7 and Table I by less than 5%. When, for example, binary data is only available for dilute solutions of N.,, and one is only interest- ed in the dilute region of the ternary, Eq. (7) can not readily be used since it requires data for the 1-2 and 2-3 binaries for higher N, ,., compositions than the ternary N, compositions. However, E%M; can easily be used because it never requires binary data at No(i) compositions greater than N>. Graphical aids are also pos- sible with Eqs. (3) and (4)(18) which result in requiring about 1/3 the number of calculations (along a single path) associated with Eqs. (7) and (8). However, in order to receive the full benefit of the interpclative method of this paper, Fig. 2, all three components should be used for component 2 and the calculated result taken as the mean of the three independent sets of calculations. This would be time consuming for the methods used to predict phase diagrams, so Eqs. (7) and (8) were used for the phase diagram determination. Previous diS(:ussion(lS’ 19) has been given on why regular so- lution equations might be applied, with some empirical success, to nonregular systems containing more than two components. 719 Fig. 11 - Calculated phase diagram, with tie lines, for the Pb- Sn-Zn system at 926°K. The dashed line indicates the experimental miscibility gap at 793°K. 300 200t | OO} AF O -100F r ~200 Fig. 12 - Calculated AF values {cal/mole) across the Ag-Cu system at 1000°K. 720 NCQ —— Fig. 13 - Calculated a values in the Ag-Pd-Cu system at 1000°K with predicted miscibility gap (dashed line and triangles). Fig. 14 - Calculated phase diagram, with tie lines, for the Ag-Pd-Cu system at 1000°K. 721 Further support of this view, as well as a clearer definition of the limitations involved, may be found in the basic assumptions of the simple regular solution model. The assumption of random mixing and constant coordination rumber allows calculation of the number of nearest neighbor bonds. The assumption of constant atom-atom interaction then permits an energy summation to be made to give the integral molar heat of mixing, or the excess free energy, for the regular solution. The regular solution equations which result, for example Eqs. (4) and (7) show that at compositions near any bi- nary system i-j, the excess free energy due to the i-j interaction is predominant. For compositions in which the concentrations of 1 and j become small, ihe i-j contribution rapidly diminishes due to the factor (N + N. ) Now, any rnultlcornponent system is bounded by a series of bi- nary systems, all of which contribute to the multicomponent field with the addition of 3rd, 4th or nth components. Considering any binary i-j composition, even though the excess entropy may be nonzero, the actual AF¥S value is known and may be used as a starting peoint. As other components are added and are present in dilute or moderate concentrations, they are likely to be incorporated into the system in a random manner while not altering a.pprema.bly the i-j coordination or interaction. Hence, the predominant i-j contribution to the total excess free energy might be expected to follow the form of Eqs. (4) and (7} at least at compositions in the vicinity of the binary edges of the multicomponent system. This would most likely be so if the components are substitutional. This kind of agreement between measured data and the regular solution equations has been presented in this paper. The accuracy with which the excess free energy curves extend into the interior of the multicomponent field will clearly be better for systems which do not exhibit abrupt changes in excess free energy as might be en- countered with the occurrence of a ternary compound. The e%uatlons of this paper have been extended into quaternary systems and the approach may readily be extended to higher order systems although the graphical relationships become very complex. The methods presented in this paper are considered to give a fair approximation of ternary phase diagrams exhibiting miscibility gaps as shown in Figs. 11 and 14. It is apparent that this method may be applied to more complex nonregular systems with more than one miscibility gap. If two or more miscibility gaps intersect and the directions of the tie lines at the points of intersection are not parallel, the tie lines at the points of intersection will form the sides of a three phase triangle. This general feature is shown ideally in Fig. 8. In the calculation of thermodynamic properties of multicomponent systems with the equations presented, the binary excess free energy 722 data that should be used are those whichcorrespond tothe most stable single phase solid or liquid solution in each binary at the temperature considered. Acknowledgements This work was supported mainly from funds made available to the University of Washington through NASA Grant NsG-484-Multi- disciplinary Research on the Nature and Properties of Ceramic Materials while N. J. Olson was a graduate student and G. W. Toop was Assistant Professor of Metallurgical Engineering. References 1. L.S. Darken: J. Am. Chem. Soc., 72, 2909 {1950). 2. C. Wagner: Thermodynamics of Alloys, Addison Wesley Press, Cambridge, Mass., p. 19 (1952) 3. R. Schuhmann: Acta Met., 3, 219 (1955). 4. N.A. Gokcen: J. Phys. Chem., 64, 401 {1960). 5. J.L. Meijering: Philips Res. Rept., 5, 333 (1950); 6, 183 (1951) 6. J. Lumsden: Thermodynamics of Alloys, The Institute of Metals, London, p. 335 (1952). 7. H.K. Hardy: Acta Met., 1, 202 (1953). 8. C. Wagner: Acta Met., 2, 242 (1954). 9. J.L. Meijering and H.K. Hardy: Acta Met., 4, 249 (1956). 10. J.L. Meijering: Acta Met., 5, 257 (1957). 11. C. Wagner: Acta Met., 6, 309 (1958). 12. F.D. Richardson: The Physical Chemistry of Steelmaking, John Wiley and Sons, Inc., New York, p. 72 (1958). 13. L.J. van der Toorn and T.J. Tiedema: Acta Met., _Ei, 711, (1960). 14. H.A. Wriedt: TMS-AIME, 221, 377 (1961). 15. R.A. Oriani and C.B. Alcock: TMS-AIME, 224, 1104 (1962). 16. K. Okajima and R.D. Pehlke: TMS-AIME, 230, 1731 (1964). 723 17. 18. 19. 20. 21. 22. 23. 24, 25. 26. 217. 28, 29- 30. 31. 32. 33. 34, 35, 36. 37. O. Kubaschewski and T.G. Chart: J. Inst. Metals, 93, 329 (1964 -65). G.W. Toop: TMS-AIME, 233, 850 (1965). N.J. Olson and G.W. Toop: TMS-AIME, 236, 590 (1966). .J. Olson and G. W. Toop: TMS-AIME,245, 906 (1969). .E. Sundquist: TMS-AIME, 236, 1111, (1966). S . Darken: TMS-AIME, 239, 80 (1967). .S. Darken: TMS-AIME, 239, 90 (1967). .T.J. Hurle and E.R. Pike: J. Matls. Sci, 1, 399 (1966). .H.P. Lupis and J.F. Elliott: Acta Met., 14, 529 (1966). .H.P. Lupis: Acta Met, 16, 1365 (1968). .B. Alcock and F.D. Richardson: Acta Met., 6, 385 (1958). OOOOU[“_F‘UJZ .B. Alcock and F.D. Richardson: Acta Met., 8, 882 (1960). C.B. Alcock: Phys. Chem. Metallic Sol. Intermetallic Compd., Symp., Teddington, Middlesex, Eng., 1958, Vol.1I, 2E, p.2. F.D. Richardson: Phys. Chem. Metallic Sol. Intermetallic Compd., Symp., Teddington, Middlesex, Eng., 1958, Vol. 1I, 6A, p. 2. Carl-Erik Froberg: Introduction to Numerical Analysis, Addison-Wesley Press, Reading, Mass. p. 172 (1965). J.F. Elliott and J. Chipman: J. Am. Chem. Soc., 73, 2682 (1951). J.F. Elliott: J. Metals, 7, 485 (1955). S. Mellgren: J. Am. Chem. Soc., 74, 5037 (1952). J.H. Smith, H. W. Paxton and C. L. McCabe: J. Phys. Chem., 68, 1345 (1964). R. Hultgren, R.L. Orr, P.D. Anderson and K. K. Kelley: Selected Values of Thermodynamic Properties of Metals and Alloys, John Wiley and Sons, New York, (1963). A.S.M. Metals Handbook, p. 1268, (1948). 724 CALCULATION OF THERMODYNAMIC QUANTITIES FROM PHASE DIAGRAMS INVOLVING INTERSTITIAL-TYPE PHASES E. Rudy Oregon Graduate Center Portland, Oregon U. S. A. Abstract The thermodynamic fundamentals relating phase relationships in binary, ternary, and quaternary systems to the thermodynamic properties of the phases partaking in the equilibria are reviewed and discussed. Cases considered in detail include partition equi- libria in ternary and quaternary systems and the role of the three- phase field in the determination of the relative stabilities of exist- ing as well as hypothetical phases. Following a brief discussion of the methods employed in es- tablishing pertinent phase equilibrium data, the application of the thermodynamic equations is demonstrated on a number of recently investigated ternary systems. 725 ThCoS, 17’ THERMODYNAMICS OF FORMATION OF Tthel7, Th2C017, Tthl ThN1i ThCu AND ThNi, FROM ELECTROMOTIVE FORCE MEASUREMENTS#* 5’ 4’ 2 N. J. MagnaniT, W. H. Skelton, and J. F. Smith Institute for Atomic Research and Department of Metallurgy Iowa State University, Ames, Iowa 50010 U. S. A. Abstract Solid electrolyte electromotive force cells have been used to determine the Gibbs free energies, entropies, and enthalpies of formation of several binary phases of thorium with iron, cobalt, nickel, or copper. CaF, was employed as the electrolyte, and it is an ionic conductor over the temperature range of the measurements, 600-850°C. The data indicate a decreasing stability of intermediate phases with defined stoichiometry, ThxMy, as M changes from nickel to cobalt to iron. The data further indicate that the Gibbs free energies of formation per g-atom of thorium are nearly constant for the thorium-poor phases within a given system. This latter obser- vation is compatible with crystallographic information which shows that the near-meighbor atomic coordination around the thorium atoms in the thorium~poor phases consists exclusively of transition-metal atoms with the coordination geometries being closely comparable from phase to phase. The crystallographic data alsc show bond distances which are compatible with the order of stability with the greatest relative contraction occurring for thorium-nickel bonds, intermediate for thorium-cobalt bonds, and least for thorium-iron bonds. * Work was performed in the Ames Laboratory of the U. S. Atomic Energy Commission. +Present address: Sandia Corporation, Albequerque, New Mexico 127 Introduction Johnson(l) has surveyed the literature through late 1963 and has reported that thermodynamic data were then available only for the thorium-aluminum, thorium~bismuth, thorium-magnesium, thorium- mercury, and thorium-zinc systems. Additional data have subsequently become available for the thorium-mercury system(z)as well as data for the thorium-lead system‘®/. Limited information has also been found for the thorium-gilicon system(4s5). Since thorium is a fertile material, thermodynamic data on its metallic systems are relevant to the design of pyrometallurgical processing metheds for reactor fuels and blanket materials. The present investigation was therefore undertaken to add systematically to the needed informatiom. The series, thorium-iron, thorium-cobalt,. thorium-nickel, and thorium- copper, wherein the atomic number of the second component increases by one from one alloy system to the next, was chosen for measure- ment because the temperature-composition diagrams had already been reliably determined and because the fifth member of the series, thorium-zinc, had already been thermodynamically investigated. The intermediate phases which cccur in these binary systems show definite crystallographic relationships, and the pattern of occurrence of the various stoichiometries implies a systematic variation in bonding interactions. Procedure & Results The thorium-iron and thorium-cobalt phase diagrams, as proposed by Thomson(6), are shown in Figs. 1 and 2. The thorium-nickel phase diagram in Fig. 3 and the thorium-copper phase diagram in Fig. 4 are reproduced from the compilations of Hansen and Anderko(7) and of %lgiott(s)r?sFectivel{. The terminal % lubilities of thorium in iron(6), cobalt(6}, nickel (9} ang copper(®’ are in all cases and at all temperatures ~2 at. 7 or less. A Raoultian approximation indicates that within the precision of the present measurements such limited solubility is of negligible significance. Among the inter- mediate phases, only ThCos exhibits a range of solid solubility and then only between 83.5 and 85.0 at. % cobalt(6) at 1100°C; this is also of negligible significance. Solid-state electrochemical cells of the type, . Th, ThF4|CaF2]ThF4, M, ThM , (a) Th, ThF4|CaF2|ThF ThM_, Thi_, (b) 4! were employed for the accumulation of data. Here M is iron, cobalt, nickel, or copper, and M and ThM; or ThM, and ThM, are neighboring phases in the temperature-composition diagram and can exist in chemical equilibrium. All phases were solid throughout the experi- ments. Half cell reactions for these galvanic cells can be written as 728 1600 [ lr_Tthew | | | T | | //] ) - 462 Th Fe5 / 14122 T~ / 1400 F==== N " A — 3 N o - B AN Th Fesz L / —- 1212 \\ r c // © 1200 == = - o |2OO N * / bt ~ wl \\ // x 1000 \ /. 9e2 - Rk T 875° N_—— S 800 . o = ] = 800} - 400 | | { | 1 ] | 0 10 20 30 40 S0 60 70 B8O 90 10C ATOMIC PERCENT THORIUM Fig. 1. The thorium-iron phase diagram 1600 T - ¢Th z(iowTh c ' l ! In ! /I/ M 1462 Os Q o - ) / / 1400 [P8Z R aX1437° 2 < p— / N T 1300° i ees | N / \ ' o / S 1200 \ 1188% ~ ° 125 \\ _I_'37° ol N 1HOQ® — e e //' ~a // N | AN // 1037 W 1000 | N\ 975° B ) | '<_I | r 800 | . Wl o f = | W 600+ | - | | 400 ol ] | | | ] 0 10 20 30 40 50 60 70 80 90 100 ATOMIC PERCENT THORIUM Fig. 2. The thorium-cobalt phase diagram 72 9 Fig. 4. The thorium~copper phase diagram 730 — ! Tho Ny ThNig _ \ yl ° = 1400 = 350° N |f 3087 ] z } — o 1200° o 1200 i S % \/ > ) (6age | 1050 x 1000 S - & = w = e00[~ — 400 i ] | | ] ] ] 0 e} 20 30 40 50 60 70 80 20 100 ATOMIC PERCENT THORIUM Fig. 3. The thorium-~nickel phase diagram 1600 T 1 T I T | 7 I / 1400 | 5 Y 3 A = o (& o~ / = £ L / —_ = - - / ¢ 1200 f } | / | / W I005° 700/ 5 S o ot A ST A = 26 I 1 o ~970° W - o = w F — 400 ] | | | | | ] | o) 0 20 30 40 50 60 70 80 20 100 ATOMIC PERCENT THORIUM Th + 4F - ThF, + 4e (1) 4 and _ _ he + ThF, + xM > ThM_ + 4F (2a) or e + ThF, + (z/y—z)ThMy > (y/y-z) ThM +4F (2b) s0 that the cell reactions should be Th + xM = ThM (3a) or X Th + (z/y-2) ThMy# (y/y-z) ThM_. (3b) If the cells are operated feversibly at fixed temperature, EMF values provide direct measures{10) of the Gibbs free energies of the cell reactions through the relation /_\.GT = -4FE where % is the Faraday constant and E is the open circuit EMF. If pure solid phases are chosen as standard states and if the limited solubilities are neglected, the data are directly convertible to standard free energies of formation. Oriani(ll) has discussed electrochemical techniques for thermo- dynamic measurements and has concluded that in reversible cells the electrolyte must exhibit ionic conductivity only, the electro- positive metal must have a single, defined valence state with respect to the anions of the electrolyte, and only one reaction should occur at each electrode interface. 1In the present experiments the first criterion is met since the values of both Ure{l2) and Patterson(13) confirm that solid CaF, is an ionic conductor throughout the temp- erature range of measurement. With respect to the latter criteria, the compilation by Hamer, Malmberg, and Rubin{(14)of an electro- motive force series for sclid and liquid fluorides indicates that, in the cells in question, CaFj should be unaffected by the cell reactions, the transition metals should remain in the reduced metal- lic state, and thorium should exist both as a metal and as a fluoride with the changes in its state corresponding to the reaction within the cell. Though there are problems with solid electrolytes,(15a16) the suitability of the present electrochemical cells is corroborated by the results of Aronson and coworkers(17,18,19) who have used similar cells for the determination of the thermodynamics of form- ation of thorium carbides, borides, and sulfides. Furthermore, Oriani's review suggests experimental tests for reversibility, since, in a cell operating reversibly, the EMF should be time independent at constant temperature, should have the same value irrespective of whether the temperature has been approached from above or below, and should recover to the same value after current is passed through the cell in either direction. During the accumulation of the present data, any measurement which failed these reversibility criteria was rejected. 731 Alloys were prepared from material whose impurity analyses are shown together with that for ThF, in Table 1. The iron analysis is the manufacturer's analysis for high-purity electrolytic iron while the iron which was used in this investigation was a super-purity grade and should have impurity concentrations lower than those in- dicated in the Table. The ThF,; was also further purified from that analyzed for the Table. This purification was achieved by heating the ThF,; with ammonium bifluoride at 150-180°C for 12 hours and then slowly heating to 350°C while passing dry air over the fluorides. This should have reduced any residual oxides or oxyfluorides. The CaF, which was used for the electrolyte was from Electronic Space Products Inc. and was quoted as being 99.95% pure. Alloys were prepared by arc melting the elements in an argon atmosphere that had been purified by melting zirconium several times. The arc melted alloys were successively inverted and re- melted several times to facilitate homogeneity. Alloys that passed through peritectic transformations on cooling were sealed in tanta- lum crucibles under a vacuum and equilibrated at 25C° below the transformation temperature for 10-14 days. Alloys that passed through eutectic transformations were not similarly heat treated. Debye-Scherrer powder photographs of all alloys were taken to verify the presence of only the two equilibrium phases. The thorium- nickel alloys that passed through a peritectic transformation were also checked with an electron beam microprobe. In no case were non~equilibrium phases found. The alloys were very reactive and were therefore stored in a dry box containing an argon atmosphere. This reactivity tends to in- crease with increasing thorium content and the present measurements were therefore limited to the thorium-poor regions of the alloy systems. For x-~ray measurements fine powders were prepared and sealed into x-ray capillary tubes withir the dry box. All steps for the preparation of electrode pellets were performed in the dry box except the actual pressing operation. Alloys that were brittle were crushed with a diamond mertar and those that were not brittle were filed with a tungsten carbide file. Thorium was filed with an ordinary file and the filings were then passed through a group of small magnets to remove any iron introduced by the file. All of the powders were passed through a 60 mesh screen to remove large part- icles and 20 wt % ThF, powder was then added. The mixture was pressed in a one-half inch tungsten carbide die at 30,000 psi. The resultant electrode pellets were from 2 to 5 mm thick. Electrolyte pellets were prepared by pressing one-half to two grams of CaFy at 7,500 psi in a one-half inch tungsten carbide die, them hydrostati- cally pressing the pellets at 50,000 psi, and finally sintering them in an induction furnace under vacuum at 1000°C for 15 minutes. A schematic diagram of the experimental apparatus is shown in Fig. 5. In this apparatus two electrochemical cells were operated 732 Table 1. Impurity analyses of alloying elements and ThF4 in ppm Impuri ty The T, cu® N9 Co® Fef H <1 - - - 6 - B - <0.5 - - - - C 50 - - 50 60 <100 N 20 - 20 - - 0 80 - - 80 100 - Na <10 - - - - - Mg <20 55-120 - - - - Al 25 <25 - - - - Si <20 50-100 <0.1 - 10 - P - - - - <30 - S . - <1 76 10 <60 Ca <20 150-500 - - - - Ti <20 - - - - - Cr 20 - - - - - Mn <20 <20 <0.5 - 30 - Fe <20 - <0.7 70 80 - Ni <20 - <] - 600 - Cu - - - 10 30 - Zn - - - - 30 - Se - - <1 - - - Y <100 - - - - - Ag - - <0.2 - - - cd - 0.2 - - - - In - 35-45 - - - - Sn - - <1 - - - Sb - - <1 - - - Te - - <2 - - - Au - - ———WATER COOLING COILS \o RING 1 SPACER BRASS NUT THERMOCOUPLE TUBE QUARTZ PUSH ROD QUARTZ TUBE ELECTRODE LEADS QUARTZ TUBE ' Ta ELECTRODE CONTACT | ELECTRODE N |11 CaF, PELLET PELLETS =] = Ta DISK CaFp PELLET Ta PLATE QUARTZ SPACER Fig. 5. Electromotive force apparatus 734 simultaneously. Tantalum was used to provide electrical contact between the external circuit and the cell electrodes. Spectro- scopic analyses showed no evidence of interaction between the tantalum and cell electrodes. Cell temperatures were controlled to *1°C. Temperatures were measured with a chromel-~alumel thermo-~ couple, and calibration showed that standard tables could be used to convert thermocouple EMF's to temperatures without significant error (<0.1%Z). EMF measurements were made with a Leeds and North- rup K-3 potentiometer. C(Cells were operated under purified helium at a positive pressure of two inches of mercury above atmospheric. Absence of spurious EMF's in the circuitry was verified by making a dummy run with no electrode pellets between the tantalum contacts. EMF measurements on various alloy compositions were made over the temperature range, 850-1120°K. The experimental data are plotted in Fig. 6 where the linear representations are least-squares fits to the experimental points. In all cases data were taken from two independent cells, and in most cases two different alloy compositions were employed. TFrom these data, standard free energies, enthalpies, and entropies of phase formation were derived, and values for these thermodynamic functions at 973°K, which is near the mean of the temperature range, are listed in Table 2. The quoted uncertainties are based upon the root mean square deviation of experimental points from linear relationships. Table 2. Thermodynamic functions for the formation of selected binary alloys of thorium with iron, cobalt, nickel and copper. Phase 467594 ~88%g73 ~AR% oo (kcal/g-atom) (e.u./g-atom) (kcal/g-atom) ThZNil7 5.50 £ 0.02 0.44 £ 0,15 5.93 + 0.15 Th2CO17 3.17 £ 0.04 0.79 + 0.21 3.94 £ 0.20 Tthel7 1.82 * 0.04 1.21 + 0.23 2.99 £ 0.22 ThNi5 8.70 £ 0.03 1.69 £ 0,18 10.35 = 0.18 ThCO5 5.08 = 0.05 2.12 + 0.29 7.14 £ 0.32 ThCu4 4,38 + 0.01 -1.49 + 0.12 2.93 + 0.12 ThNi2 9.93 + 0.09 0.77 * 0.70 10.68 = 0.68 735 | 1 | [ I v Cu~5% Th 240 - ACu=7%Th _§——atg—Y um e 220 70 = o Ni-29% 0 0 0 °© o 0 150 o © ‘ %80 = 4 Ni-nwTh 2 o o 560 [~ O Ni-15%Th 4~o A—g g 540 i 580 o g — T —O0—0——0o—0__ — 3 560 = o Ni-3% Th w \‘-——_x R = 340 — 4 Co-N%Th \‘\ — v -13% 320 Co~13%Th . Q‘\\‘ B v 300 —J 340 —oco-3%Th D 4 320 |- & Co-5%Th T T e,—a0 4 - N x 200 o — A Fe-5%Th —'-O—-—o_.ob (@] 180 = 4 Fe-7%Th 6% Ro-90_ ] 160 [~ | | | 1 1 | 850 900 950 1000 1050 1100 1150 TEMPERATURE (°K) Fig. 6. Experimental data from various alloys with indicated compositions being in atomic per cent 736 Discussion Crystallographic data indicate that ThjNijs crystallizes(zo) in a double~layered hexagonal structure while ThCoj7 and Th2Feqpy crystallize(zo’21 in a closely related triple-layered structure whose prototype is Th2Znj7. The ?hases, ThFes5, ThCo5, and ThNig, are isostructural and crystallize 20) in the hexagonal structure whose prototype is CaCus and which is crystallographically also closely related to the ThPNijy and Th2Zny; type structures. In the ThoM37 and ThMs5 phases with M being iron, cobalt, or nickel, indi- vidual thorium atoms are coordinated to either eighteen, nineteen, or twenty neighboring atoms of the M species with the average thorium coordination number being eighteen and one-half. For both the ThyM;7 phases and for the ThM5 phases, the sequence of magnitudes for the free energies of phase formation is compatible with the crystallographic data in the sense that the interatomic Th-M bond distances in the intermediate phases show contractions from the average of the Th-Th and M-M distances in the parent metals which are greatest for the thorium-nickel phases, intermediate for the thorium—-ccbalt phases, and least for the thorium-iron phases. This statement is valid for the ThMg sequence because even though the free energy of formation of ThFeg has not yet been measured, the free energy of formation of ThFey; cannot be more negative than -2.9 kcal/g-atom; otherwise, ThyFe]7 would not be a stable phase and would spontaneously decompose to ThFeg and iron with a reduction in total free energy. It is interesting to note that conversion of the values in Table 2 from kcal/g-atom of compound to kcal/mole of thorium yields values of -52.2%0.2 for ThoNiyy, -52.220.2 for ThNis5, -30.1#0.4 for ThoCoj7y, ~-30.510.3 for ThCog and ~17.3%#0.4 for ThyFey7. From a quasi-chemical view the formation of an intermediate phase results from the energy reduction associated with the bonding of one species to a second unlike species. On the basis of the closely comparable thorium environments in the crystallographic structures, it is therefore not surprising to find closely comparable free energies of formation per mole of thorium for the two nickel phases and also for the two cobalt phases. It would seem reasonable that the same situation should hold true for ThyFei7 and ThFes, and thus a free energy of formation for ThFeg of -17.3%0.4 kcal/mole of thorium or -2.88£0,07 kcal/g-atom of compound can be considered as a good estimate. An entropy of formation for ThFegmay also be estimated from the entropies of formation of the ThpM;7; and ThMg phases in the manner shown in Fig. 7. There it may be noted that the shift in AS§y3 from ThNis to ThCos is essentially parallel to the shifts in AS§y3 from ThyNi;, to ThpCoyy to ThpFej;. A similar shift from ThCos to ThFes can be used as a basis for estimating a value of -2.5 e.u./g-atom for the entropy of formation of ThFeg. Combination of the free energy and entropy estimates for ThFes; indicates that the enthalpy of formation should be near to -5.3 kcal/g-atom. The reliability of these entropy and enthalpy values is expected to be less good than the free energy value by about an order of magnitude. 737 30 T — T Fe Co Ni 25 (~ . JhFes — i ThC05 ~ \\ _ ThNisg 5 -'6 LS - — o \_ Thz FE|7 2 ~ ~ 10 - \\ — " o ~4Jha2Corr 7] g ~ ! ~ 0S5 |- ~ - ThoNi7 0 1 1 | 26 27 28 ATOMIC NUMBER Fig. 7. Trends in the entropies of formation of the Th2M17 and ThM5 sequences 738 bel Fig. 8. TT R O Th—Ni \ \\ A Th—Co € N -1 o \ ] N\ § ————— 1"\\ \l-\ 1 o ~ AN » N o \ N \\ — n N — ~ < o~ - ~ \ _— e & N ~ ™~ ~ \ ~ - N \ Y | | | | o 0.4 ’ 0.6 0.8 1.0 MOLE FRACTION THORIUM Free energies of alloy formation as functions of composition for the thorium-iron, thorium- cobalt, and thorium-nickel systems with experimental data being connected by solid lines and estimated data being connected by dashed lines Finally, estimates of the free energies of formation of the remaining thorium-nickel, thorium-cobalt, and thorium-iron phases can be made by a simple extension of the quasi-chemical approach. In this extension the free energies of formation of the Th2Mj; and ThMs5 phases may be taken as the base, and the free energy of forma- tion per mole of thorium for any ThxMy phase can then be taken as lower than that for the ThoMy7 and ThMs5 phases in direct proportion to the difference in the number of M atoms coordinated about the thorium atoms. For instance in ThNij, the nickel coordination around thorium is twelvefold(zo), and thence AG§73 should be (12/18.5)(-52.2) kcal/mole of thorium which converts to -11.3 kcal/g-atom and is within 14% of the experimental value in Table 2. Similar testing of the procedure on the thorium-zinc system with ThpZnjy as the base yielded estimated values for ThZng,, ThZnp, and ThoZn which averaged ~20% deviation from the experimental values of Chiotti and Gi11(22), Comparable agreement has also been ob- tained between estimated values and preliminary experimental values for ThCo3, ThNi, and ThyNig. Plots of integral free energies of alloy formation against mole fraction thorium are shown in Fig. 8 for the thorium-nickel, thorium-cobalt, and thorium~iron systems. The dashed portions of these plots have been estimated in the manner outlined and, on the basis of the indicated tests of the estimating procedure, should have a reliability of about *207%. No similar estimates were made for the thorium-copper system since the crystal structure of ThCu, is as yet unknown. In this system stoichiometries of ThjCujy and ThCug do not occur as stable phases, and, in view of the value of the free energy of formation of ThCu4,, it is evident that the free energy of formation for ThCujy must be less negative than -2.3 kcal/g-atom and for ThCus; less nega-~ tive than -3.6 kcal/g-atom. References 1. Johnson, I., "Thermodynamics of Plutonium, Thorium, and Uranium Metallic Systems', Proceedings of the International Symposium on Compounds of Interest in Nuclear Reactor Technology, The Metal- lurgical Society of AIME, Vol. 10, 1964, pp. 171-192. 2. Jangg, G. and F. Steppan, "Dampfdruckmessungen an binaren Amalgamen', Zeitschrift fur Metallkunde, Vol. 56, 1965, pp. 172-178. 3. Gans, W., 0. Knacke, F. Mfiller, and H. Witte, "Dampfdruckmessungen am System Blei-Thorium'", Zeitschrift fur Metallkunde, Vol. 57, 1966, pp. 46-49, 4. CGrieveson, P. and C. B. Alcock, "The Thermodynamics of Metal Silicides and Silicon Carbide', Special Ceramics, ed. by P. Popper, Heywood & Co., London, 1960, pp. 183-208. 740 12. 13. 14. 15. 16, 17. Robins, D. A. and I. Jenkins, '""The Heats of Formation of Some Transition Metal Silicides", Acta Metallurgica, Vol. 3, 1955, pp. 598-604. Thomson, J. R., "Alloys of Thorium with Certain Transition Metals. IV. The Systems Thorium-Iron and Thorium-Cobalt', Journal of the Less-Common Metals, Vol. 10, 1966, pp. 432-438. Hansen, M. and K. Anderko, Constitution of Binary Alloys, McGraw-Hill, New York, 1958, p. 1048. Elliott, R. P., Constitution of Binary Alloys, First Supplement, McGraw-Hill, New York, 1965, p. 386. Hessenbruch, W. and L. Horn, '"Der Einfluss kleiner Beimengungen von Thorium auf die Lebensdauer von Heizleiterlegierungen', Zeitschrift fur Metallkunde, Vol. 36, 1944, pp. 145-146. . Darken, L. S. and R. W. Gurry, Physical Chemistry of Metals, McGraw-Hill, New York, 1953, p. 431. Oriani, R. A., "Electrochemical Techniques in the Thermodynamics of Metallic Systems", Journal of the Electrochemical Society, Vol. 103, 1956, pp. 194-201. Ure, R. W., Jr., "Ionic Conductivity of Calcium Floride Crystals'", Journal of Chemical Physics, Vol. 26, 1957, pp. 1363-1373. Patterson, J. W., Iowa State University, 1968, private communication. Hamer, W. J., M., S. Malmberg, and B. Rubin, "Theoretical Elec- tromotive Forces for Cells Containing a Single Solid or Molten Fluoride, Bromide, or Todide", Journal of the Electrochemical Society, Vol. 112, 1965, pp. 750-755. Schmalzreid, H., "The EMF Method in Studying Thermodynamic and Kinetic Properties of Compounds at Elevated Temperatures", Proceedings of the Symposium on Thermodynamics held in Vienna, 22~27 July 1965, Vol. 1, International Atomic Energy Agency, Kiukkola, K. and C. Wagner, ''Measurements on Galvanic Cells Involving Solid Electrolytes', Journal of the Electrochemical Society, Vol. 104, 1957, pp. 379-386. Aronson, S., '"Thermodynamic Properties of Thorium Carbides from Measurements on Solid EMF Cells", Proceedings of the International Symposium on Compounds of Interest in Nuclear Reactor Technology, The Metallurgical Society of AIME, Vol. 10, 1964, pp. 247-257. 741 18. 19. 20. 21. 22. Aronson, S. and A. Auskern, '"The Free Energies of Formation of Thorium Borides from Measurements on Solid EMF Cells", Proceedings of the Symposium on Thermodynamics held in Vienna, 22-27 July 1965, International Atomic Energy Agency, Vienna, 1966, pp. 165- 170. Aronson, S., "Free Energies of Formation of Thorium Sulfides from Solid-State EMF Measurements", Journal of Inorganic and Nuclear Chemistry, Vol. 29, 1967, pp. 1611-1617. Florio, J. V., N. C. Baenziger, and R. E. Rundle, "Compounds of Thorium with Transition Metals. II. Systems with Iron, Cobalt, and Nickel", Acta Crystallographica, Vol. 9, 1956, pp. 367-372. Johnson, Q., G. S. Smith, and D. H. Wood, "Intermetallic 2-17 Stoichiometry: The Crystal Structure of ThpFey; and ThjCoq;", Acta Crystallographica, in press, (UCRL-71021 Preprint 1968). Chiotti, P. and K. J. Gill, '"Phase Diagram and Thermodynamic Properties of the Thorium-Zinc System'', Transactions of The Metallurgical Society of AIME, Vol. 221, 1961, pp. 573-580. 742 £he ACTIVITIES ADSORPTION AG-PD~-CU AGBR ALLOY ALLOY ALLOY ALLOYS ALLOYS ALLOYS ALLOYS ALLOYS ALUMINA AMERICTUM ANALYSTS ANOMOLOUS AQUEQUS AQUEGUS AQUEOUS AQUECQUS BACLZ BEHAVIOR BEHAVIOR BERYLLIA BT1-TH BISMUTH BISMUTH BOUNDRIES BOUNDRIES BOUNDRIES BRFS BURNUP CADIUM CALCULATED CALCULATED CALCULATION CALCULATION CALCULATION CALCULATEONS CAQO-S102-FED CARBIDE CARBIDE CARBIOE CARBIDE CARBON CARBON CARBOX CD-PB-BI CHALCOGENIDES CHEMISTRY CHEMISTRY CHEMISTRY CHLDRE X CHLORIDE KEYWORD INDEX SOLUBTILITIES AND ACTIVITIES IN MOLTEN KNO3 ADSORPTION OF UF& ON NAF AG-PD-CU SYSTEM DISTRIBUTLON OF THALLIUM-BROMIOE BETREEN KKO3 AND AGBR EBR-II FUEL ALLOY STABILITY AND PLUTONIUN CONTENT EXTRACTION OF PROTACTINIUM AND URANIUM FROM LIF-BEF2-THF4 INTO B8I-TH ALLOY CORRISTON OF LOW ALLOY STEEL BY LIQUID LEAD-BISMUTH ALLOY DISTRIBUTION OF U AND PU BETWEEN MGCL2 SALTS AND CU-MG AND ZIN-MG ALLCYS THERMODYNAMIC PROPERTIES OF PLUTONTUN-MAGNESIUM ALLOYS SOLUBILITY OF URANTUM AND PLUTCNIUM IN LIQUID ALLOYS SOLUBILITY OF URANTUM IN ZIN-MG ALLOYS SOLUBILITY OF PLUTONIUM IN ZN-MG ALLOYS SILICON CARBIDE AND ALUMINA AS SALT-METAL CONTAINERS MOLTEN SALT EXTRACTIDN OF AMERICIUM FROM PLUTONIUM THERMOOYNAMIC ANALYSTS AND SYNTHESIS OF PHASE DIAGRAMS ANOMOLOUS BEHAVIOR OF NOBLE METAL FISSTON PRODUCTS IN MSRE FUEL AQUEOUS REPPOCESSING LMFBR FUEL AQUEOUS REPROCESSING OF THORIUM-URANIUM FUELS TECHNOLOGY AND ECONOMICS OF AQUEOUS PROCESSINGy FRANCE TECHNOLOGY AND ECONOMICS OF AQUEOUS PROCESSINGs INDEA SORPTION AND DESORPTION OF CHLORIDES ON BACL2 BEHAVIOR OF NEPTUNIUM DURING FLUORINATION ANOMOLOUS BEHAVIOR OF NOBLE METAL FISSION PRGOUCTS IN MSRE FUEL BERYLLIA CONTAINERS FOR ZN-MG-U EXTRACTION OF PROTACTINIUM AND URANIUM FROM LIF-BEF2-THF4 INTD BI-TH ALLOY RATE OF TRANSFER BETWEEN FLUORIDE SALT AND BISMUTH REDUCTIVE EXTRACTION OF RARE-EARTHS FROM LIF-BEF2-THF4 INTO BISMUTH CALCULATION OF LIQUIDUS BOUNDRIES FOR TERNARY SALT SYSTEMS THERMODYNAMIC DESCRIPTION OF BINARY PHASE BOUNORIES PREDICTION OF TERNARY BOUNDRIES AND THERMODYNAMIC PROPERTIESFROM BINARY DATA FLUORINATION OF U0Z WITH 8RFS AND F2 LMFBR FUEL BURNUP CORROSTON OF STAINLESS-STEEL BY ITNC,CADIUM AND MAGNES IUM CALCULATED AND OBSERVED PROPERTIES FOR EUTECTIC SYSTEMS CALCULATED AND OBSERVED PROPERTIES FOR MISCIBILITY GAP SYSTEMS CALCULAYION OF LIQUIDUS BOUNDRIES FOR TERNARY SALT SYSTEMS CALCULATION OF THERMODYNAMIL PROPERTIES FROM BINARY PHASE DIAGRAMS CALCULATION OF THERMODYNAMIC QUANTITIES FROM PHASE DIAGRAMS INVOLVING INTERSTITIAL TYPE PHASES CALCULATIONS OF SEPARATION PROPERTIES FOR THORIUM—URANIUM FUELS BY CHLORIDE VOLATILIZATION CAQ-STD2~-FED SYSTEM REPROCESSING OF URANIUM CARBIDE FUEL, CARBOX PROCESS MOLTEN SALY ELECTRDLYSIS OF URANTUM CARBIDE FUSED-SALT FLUORIDE VOLATILITY PROCESS FOR THORIUM-URANTUM DXIDE OR CARBIDE FUELS SILTICON CARBIDE AND ALUMINA AS SALT-METAL CONTAINERS REACTION OF U022 WiTH CARBON TO FORM UC CRAPHITE AND VITREQOUS CARBON AS SALT-METAL CONTAINERS REPROCESSING OF URANIUM CARBIDE FUtsi, CARBOX PROCESS CD-PB-BI SYSTEM FREE ENERGY OF FORMATION OF CHLORIDES AND CHALCOGENIDES CHEMISTRY AND THERMODYNAMICS OF MOLTEM SALT FUELS HYDROFLUDRINATION AND OXIDE CHEMISTRY IN LIF-BEF2 MSRE FUEL CHEMISTRY CHLOREX PROCESS CHLORIDE VOLATILITY PROCESSING OF THD2-U02 AND U02-PU0O2 FUELS L CHLORIDE CHLORIDE CHLORIDE CHLOR IDE CHLORTIDE-ALKALT CHLORIDES CHLORIDES CHLORIDES CHLORIDES CHLORIDES CHLORINATION CHLOR INATION-D! CHLORINE CHROMTI UM CLADDING COEFFICIENTS COEFFICIENTS COMPATABILITY CONTAINER CONTAINERS CONTAINERS CONTAINERS CONTAINERS CORRISION CORRISION CORRISION CORROS [ON CORROSION CORROSTON CORROS ION CORROSION CORRDSION CRAPHITE CU-MG CU-MG CU-MG-U CYCLE DATA DATA DECLADDING CECOMPOSITIDN DESIGN DESORPTION DEVELOPMENT DIAGRAN DIAGRAM DIAGRAMS DIAGRAMS DIAGRAMS DIFFUSION DISSOLUTION DISSOLUTION DISSOLUTION DISTRIBUTION KEYWORD INDEX COMPATABILITY AND PROCESSING PROBLEMS OF MOLTEN URANIUM CHLORIDE-ALKALI CHLORIDE FUELS MOLTEN CHLORIDE FUEL CONTAINER CORRISION UOCL2 AND UQCL IN CHLORIDE SALT CALCULATIONS OF SEPARATION PROPERTIES FOR THORJUM-URANIUM FUELS BY CHLORIDE VOLATILIZATION COMPATABILITY AND PROCESSING PROBLEMS OF MOLTEN URANTUM CHLORIDE-ALKALI CHLORIDF FUELS SORPTION AND DESCRPTION OF CHLDRIDES DN BACL?2 VAPOR PRESSURE OF CHLOREDES THERMODYNAMIC PROPERTIES OF CHLORIDES FREE ENERGY OF FORMATION OF CHLORIDES AND CHALCOGENIDES TYHERMOOYNAMIC DATA FOR CHLORIDES HIGH TEMPERATURE TREATMENT AND CHLORINATEION OF COATED PARTICLE THTR FUELS CHLORINATION-DISTILLATION OF [RRADIATED URANIUM DIOXIDE CHLORINE POTENTTAL AND CORRISION SOLUBILITY OF TUNGSTEN,TANTALUM,NIOBIUM,VANADIUM, MOLYBDENUM AND CHROMIUM IN LIQUID PLUTONIUM DISSLUTION OF CLADDING IN MOLTEN METALS INTERACTION COEFFICIENTS IN LIQUID IRON DISTRIBUTION COEFFICIENTS IN LICL-XALCLS SYSTEM COMPATABILITY AND PROCESS ING PROB{EMS OF MOLTEN URANIUM CHLORIDE-ALKAL! CHLDRIDE FUELS MOLTEN CHLORIDE FUEL CONTAINER CORRISION BERYLLIA CONTAINERS FOR ZIN-MG-U CONTAINERS FOR PYRDCHEMICAL PROCESSES CRAPHITE AND VITREQUS CARBON AS SAET-METAL CONTAINERS SILICON CARBIDE AND ALUMINA AS SALT-METAL CONTAINERS MOLTEN CHLORIDE FUEL CONTAINER CORRISION CHLORINE POTENTIAL AND CORRISION CORRISION OF LOW ALLOY STEEL BY LIQUID LEAD-BISMUTH ALLOY CORROSION OF HASTELLOY-N DURING FLUORINATION CORROSION OF HASTELLOY-N DURING FLUORINATION CORROSION IN ZRF4 FREE SALT TYPES OF CORROSION TESTS FOR SALT-METAL SYSTEMS CORROSTON OF STAINLESS-STEEL BY ZINC,CADIUM AND MAGNESIUM CORRDSION OF NIOBIUM AND TANTALUM BY MGCLZ2-NACL-XCL CRAPHITE AND VITREOUS CARBON AS SALT-METAL CONTAINERS DISTRIBUTION OF U AND PU BETWEEN MGCL2 SALTS AND CU-MG AND IN-MG ALLOYS SOLUBILITY OF URANIUM AND PLUTONIUM IN CU-MG AND ZN-MG CU-MG—-U PHASE DIAGRAM FISSION PRODUCT REMOVAL IN NITRIDE-CARBIDE CYCLE THERMODYNAMIC DATA FOR CHLORIDES PREDICTION OF TERNARY BOUNORIES AND THERMODYNAMIC PROPERTIESFROM BINARY DATA STAINLESS—-STEEL OECLADDING IN ZINC THERMAL DECOMPDSITION IN UF6-PUF5 MIXTURES VOLATILITY PROCESS PLANT DESIGN AND PROCESSING LOAD SORPTION AND DESORPTION OF CHLORIDES ON BACL2 ENGINEERING DEVELOPMENT OF PROCESSES FOR MSBR FUEL CU-MG~U PHASE DIAGRAM PLUTONIUM-MAGNESTUM PHASE DIAGRAM CALCULATION OF THERMODYNAMIC PROPERTIES FROM BINARY PHASE DIAGRAMS THERMODYNAMIC ANALYSIS AND SYNTHESTS OF PHASE DIAGRAMS CALCULATION OF THERMODYNAMIC QUANTITIES FROM PHASE DIAGRAMS INVOLVING INTERSTITIAL TYPE PHASES MEASUREMENT OF VISCOSITY AND DIFFUSION IN LIQUID METALS DISSOLUTION OF CLAPDING IN MOLTEN METALS DISSOLUTION OF PLUTONIUM—URANIUM OXIDE PYROSULPHATE DISSOLUTION OF OXIDE FUEL DESTRIBUTION DF FISSION PRODUCTS AFTER HYDROCHLORIMATION OF URANEUM FUEL SvL DISTRIBUT ION DISTRIBUTION DISTRIBUTION DISTRIBUTION EBR-1I EBR-TI ECONOMICS ECONDMICS ECONOMICS EFFECT EFFECT EFFECT EFFICIENCIES EFFICIENCY ELECTROCHEMICAL ELECTROCHEMICAL ELECTRODE ELECTROLYSIS ENERGIES ENERGY ENGINEERING ENGINEERING EQUILIBRTA EQUILIBRIA EQUILIBRIUM ERR-T1 ERR-[I EUTECTIC EUTECTIC EVOLUTION EVOLUTION EXPERIENCE EXPERTIENCE EXTRACTION EXTRACTION EXTRACTION EXTRACTION EXTRACTION FABRICATIDN FISSION FISSION FISSION FISSION FISSION FISSION FISSTON FISSION FLUORIDE FLUORIDE FLUDRIDE FLUORIDE FLUORIDE FLUQRIDE FLUDRIDE DISTRIBUTION OF FISSION PRODUCTS AFTER FLUORINATION DISTRIBUTION OF U AND PU BETWEEN MGCL2 SALTS AND CU-MG AND ZN-MG ALLOYS DISTRIBUTION OF THALLIUM-BROMIDE BETWEEN KNO3 ANC AGBR OISTRIBUTION COEFFICIENTS IN LICL-KALCL4 SYSTEM EBR-IT FUEL ALLOY STABILITY AND PLUYONIUN CONTENT REMOTE FABRICATION OF €£BR-I1 FUEL TECHNOLOGY AND ECQONOMICS OF AQUEQUS PROCESSING, FRANCE TECHNOLOGY AND ECONOMICS OF AQUEDUS PROCESSING, INDIA TECHNOLOGY AND ECONOMICS OF FLUCRIDE VOLATILITY PROCESS IN FRANCE EFFECT OF FREE-FLUCRIDE ON FLUDRIDE-SALY AND LIQUID METAL EQUILISBREA EFFECT OF OXIDE IN MOLTEN SALTS EFFECT OF UF4—UF3 RATIO IN NMSRE FUEL FLUORINE EFFICIENCIES AND PUF6 PRODUCTION RATES EXTRACTION RATE AND SYAGE EFFICIENCY ELECTROCHEMICAL MEASUREMENT OF UCL4/UCL3 EQUILIBRIUM SOLYID STATE ELECTROCHEMICAL CELLS ELECTRODE POTENTIALS IN LIF-BEF2 MOLTEN SALT ELECTROLYSIS OF URANIUM CARBIDE FREE ENERGIES OF SOLUTES IN LIF-BEF2 FREE ENERGY OF FORMATION OF CHLORIDES AND CHALCOGENIDES ENGINEERING DEVELOPMENT OF PROCESSES FOR MSBR FUEL ENGINEERING SCALE FLUDRIDE VOLATILITY STUDIES ON PLUTONTUM BEARING FUEL EQUILTBRIA [N SALT-METAL SYSTEMS EFFECT OF FREE-FLUCRIDE ON FLUORIDE-SALT AND LIQUID METAL EQUILIBRIA ELECTROCHEMICAL MEASUREMENT OF UCL4/UCL3 EQUILIBRIUN ERR-I1 FUEL REPROCESSING ERR-IT SKULL RECLAMATION PROCESS PURIFACATION OF UF4-LIF EUTECTIC CALCULATED AND OBSERVED PROPERTIES FOR EUTECTIC SYSTEMS [ODINE EVOLUYION AND RETENTION TODINE EVOLUTION AND COLLECTION PILOT PLANT EXPERIENCE ON PURIFICATION OF PLUTONIUM BY FLUORIDE VOLATILITY PILOT PLANT EXPERTIENCE WITH SKULL-OXIDE PROCESS EXTRACTION OF PROTACTINIUM AND URANIUM FROM LIF-BEF2-THF4 INYO BI-TH ALLOY REDUCTIVE EXTRACTION OF RARE-EARTHS FROM LTF-BEF2-THF4 INTO BISMUTH MOLTEN SALT EXTRACTION OF AMERICIUM FROM PLUTONIUM MOLTISTAGE LIQUID METAL-SALY EXTRACTION EXTRACYION RATE AND STAGE EFFICIENCY REMOTE FABRICATION OF E8SR-II FUEL FISSTON GAS VOLATILIZATION ANO COLLECTION VOLATILTZATICN OF FISSION PRODUCTS FISSION PRODUCT REMOVAL IN NITRIDE-CARBIDE CYCLE VOLATILIZATION OF FISSION PRODUCT FLUCRIDES DISTRIBUTION OF FISSTON PRODUCTS AFTER HYDROCHLORINATION OF URANIUM FUEL DISTRIBUTION OF FISSION PRODUCTS AFTER FLUORINATION FISSION GAS RELEASE IN PYRQCHEMICAL PROCESSING ANCMOLOUS BEHAVIOR OF NOBLE METAL FISSION PRODUCTS IN MSRE FUEL TECHNOLOGY AND ECONOMICS OF FLUORIOE VOLATILITY PROCESS IN FRANCE CHEMICAL REACTORS FOR FLUQORIDE VOLATILITY PROCESS PILOT PLANT EXPERIENCE ON PURIFICATION OF PLUTONIUM BY FLUORIDE VOLATILITY FLUDRIDE VOLATELEIYY PROCESS FOR OXIDE FUELS ENGINEERING SCALE FLUQRIDE VOLATILETY STUDIES ON PLUTONIUM BEARING FUEL POTENTTAL OF FLUORIDE VOLAYILITY PROCESS FUSED-SALT FLUORIDE VOLATILITY PROCESS FOR THORIUM-URANIUM OXIDE DR CARBIDE FUELS 9%L FLUORIDE FLUORTIDE-SALY FLUDRIDES FLUORINATION FLUDRINATION FLUOR INATION FLUORINATION FLUORINATION FLUORINE FLUORINE FORMATION FORMAT ION FRANCE FRANCE FREE FREE FREE FREE~-FLUORIDE FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUEL FUELS FUELS FUELS FUELS FUELS FUELS FUELS FUELS FUELS FUSED FUSED-SALT F2 GAS GAS HASTELLOY-N HASTELLOY=-N HYDROCHLORINAY L HYDROCHLORINAT I KEYWORD INDEX RATE OF TRANSFER BETWEEN FLUORIDE SALT AND BISMUTH EFFECT OF FREE-FLUORIDE ON FLUORIDE-SALT AND LIQUID METAL EQUILIBRIA VOLATILIZATION OF FISSION PRODUCY FLUORIDES CORROSTON OF HASTELLOY-N DURING FLUORINATION CORROSTION OF HASTELLOY-N DURING FLUORINATION FLUORINATION OF UQ2 WITH 8RFS5 AND F2 BEHAVIOR OF NEPTUNIUM DURING FLUORINATION DISTRIBUTION OF FISSION PRODUCTS AFTER FLUDRINATION REACTIONS OF (U.PUID2 WITH FLUORINE FLUOGRINE EFFICIENCIES AND PUFS& PRODUCTION RATES FREE ENERGY OF FORMATION OF CHLORIDES AND CHALCOGENIDES THERMDDYNAMICS OF FORMATION OF TH2FE17,TH2CO17,TH2NI17,THCOS, THNIS ,THCU4 AND THKI2 TECHNOLOGY AND ECONOMICS OF AQUEOUS PROCESSING, FRANCE TECHNOLOGY AND ECONOMICS OF FLUCRIDE VOLATILITY PROCESS IN FRANCE CORROSTON IN ZRF& FREE SALT FREE ENERGY OF FORMATION 0Of CHLORIDES AND CHALCOGENIDES. FREE ENERGIES OF SOLUTES IN LIF-BEF2 EFFECT OF FREE-FLUORIDE ON FLUORIDE-SALT AND LIQUID METAL EQUILIBRIA AQUEOUS REPPOCESSING LMFBR FUEL LNFBR FUEL BURNUP FUEL SHIPPING AND RECEIVING PYROSULPHATE DISSOLUTION OF OXIDE FUEL ERR-I1 FUEL REPROCESSING ERR-IT FUEL ALLOY STABILITY AND PLUTONIUN CONTENT REMOTE FABRICATION OF EBR-~II FUEL PREPARATION AND PROCESSING OF MSRE FUEL FEED MATERIAL FOR MOLTEN SALY FUEL ENGINEERING DEVELDPMENT OF PROCESSES FOR MSBR FUEL REPROCESSING OF URANIUM CARBIDE FUEL, CARBDX PROCESS HYDRCCHLORINATION OF URANYTUM FUEL ANO VOLATILIZATION OF URANIUM FUEL DISTRIBUTION GF FISSION PROOUCTS AFTER HYDROCHLORINATION OF URANIUM FUEL ENGINEERING SCALE FLUORIDE VOLATILITY STUDIES ON PLUTONIUM BEARING FUFEL MOLTEN CHLORIDE FUEL CONTAINER CORRISION MSRE FUEL CHEMISTRY ANOMCLOUS BEHAVIOR OF NOBLE METAL FISSION PRODUCTS IN MSRE FUEL EFFECT OF UF4-UF3 RATIO IN MSRE FUEL PUREX PROCESS FOR LWR OXIDE FUELS AQUEDUS REPROCESSING OF THORIUM-URANTUM FUELS HIGH TEMPERATURE TREATMENT AND CHLORINATION OF COATED PARTICLE THTR FUELS FLUORIDE VOLATILITY PROCESS FOR OXIDE FUELS CHLORIDE VOLAYILITY PROCESSING OF THO2-UO2 AND UD2-PUD2 FUELS FUSED-SALY FLUORIDE VOLATILITY PROCESS FOR THORIUM-URANIUM OXIDE OR CARSIDE FUELS COMPATABILITY AND PROCESSING PROBLEMS OF MOLTEN URANIUM CHLORIDE-ALKALI CHLORIDE FUELS CHEMISTRY AND THERMODYNAMICS OF MOLYEM SALY FUELS CALCULATIONS OF SEPARATION PROPERTIES FOR THORIUM-URANIUM FUELS BY CHLORIDE VOLATILIZATIDN MASS TRANSFER AND TRANSPORT IN FUSED SALT-LIQUID METAL FUSED-SALT FLUORIDE VOLATILITY PROCESS FOR THORIUM-URANIUM OXIDE OR CARBIDE FUELS FLUORINATION OF UD2 WITH BRFS AND F2 FISSION GAS VOLATILIZATION AND COLLECTION FISSION GAS RELEASE IN PYROCHEMICAL PROCESSING CORROSTON OF HASTELLOY-N DURING FLUORINATION CORRDSION OF MASTELLOY-N DURING FLUODRINATIDN HYDROCHLORINATION OF URANIUM FUEL AND VOLATILTIZATION OF URANIUM FUEL DISTRIBUTION OF FISSION PRODUCTS AFTER HYDROCHLORINATION OF URANIUM FUEL L¥L HYDROFLUORINATIE HYDROFLUORINATI HYDROGEN INDIA INTERACTION INTERPHASE TNTERSTITIAL TODINE TODINE IRDN KNO3 KNO3 LEAD-B ISMUTH LICL-KALCLS LIF-BEF2 LIF-BEF2 LIF-BEF?2 LIF-BEF2 LIF-BEFZ-THF4 LIF-BEFZ-THF 4 LiQuro LIQUID LTQUID LIQuUID LIQuUID LIQUID L1guUID LIQUID LTQUIO LiQuIbus LMF8R LMFBR LWR MADRAS MAGNESTUM MASS MASS MATERIAL MEASUREMENT MEASURE MENT METAL MET AL METAL METAL~-SALT METALS METALS METALS METALS METALS MG-CU-CA MG-IN MGCL2 MGCL2-NACL~KCL MISCIBILITY HYDROFLUORINATION OF U02 HYDROFLUORINATION AND DXIDE CHEMISTRY IN LIF-BEF2 REDUCTION OF UCL& WITH HYDROGEN TECHNOLOGY AND ECONOMICS OF AQUEOUS PROCESSING, INDIA INTERACTION COEFFICIENTS IN LIQUID IRON INTERPHASE MASS TRANSFER CALCULATION OF YHERMODYNAMIC QUANTITIES FROM PHASE DIAGRAMS INVOLVING INTERSTITIAL TYYPE PHASES T0ODINE EVOLUTION AND RETENTION TO0DINE EVOLUTION AND COLLECTION INTERACTION COEFFICIENTS EN LIQUID TRON SOLUBILITIES ANC ACTIVITIES IN MOLTEN KNO3 DISTRIBUTIDN DF THALLIUM-BROMIDE BETWEEN KNO3 AND AGBR CORRISION OF LOW ALLOY STEEL BY LIQUID LEAD-BISMUTH ALLOY DISTRIBUTION COEFFICIENTS IN LICL-KALCLS& SYSTEM REACTIONS IN LIF-BEFZ SALY FREE ENERGIES OF SOLUTES IN LIF-BEF2 ELECTRODE POTENTIALS IN LIF=-BEF2 HYDROFLUORINATION AND OXIDE CHEMISTRY N LIF-BEF2 EXTRACTION OF PROTACTINIUM AND URANEUM FROM LIF-BEF2-THF4 INTO BI-TH AlLOY REDUCTIVE EXTRACTION OF RARE-EARTHS FROM LIF-BEF2-THF4 INTO BISMUTH PARTITION OF SOLUTES BETWEEN LIQUID METALS AND SALTS EFFECT OF FREE~-FLUORIDE ON FLUORIDE~SALT AND LIQUID METAL EQUILIBRIA CORRISION OF LOW ALLOY STEEL BY LIQUID LEAD-BISMUTH ALLOY VISCOSITY AND TRANSPDRY PROPERTIES OF LIQUID METALS MEASUREMENT DF VISCOSITY AND DIFFUSION IN LIQUID METALS MOLTISTAGE LTQUID METAL-SALT EXTRACTION SOLUBILITY OF URANIUM AND PLUTONIUM IN L{IQUID ALLOYS SOLUBILITY OF TUNGSTEN,TANTALUM,NIOBIUM,VANADIUM,MOLYBDENUM AND CHROMIUM IN LIQUID PLUTONIUN INTERACTION COEFFICIENTS [N LIQUID IRON CALCULATION OF LIQUIDUS BOUNDRIES FOR TERNARY SALT SYSTEMS AQUEQUS REPPDOCESSING LMFBR FUEL LMFBR FUEL BURNUP PUREX PROCESS FOR LWR OXIDE FUELS MADRAS REPROCESSING COMPLEX CORROSION OF STAINLESS-STEEL BY ZINC,CADIUM AND MAGNESIUM MASS TRANSFER AND TRANSPORT IN FUSED SALT-LIQUID METAL INTERPHASE MASS YRANSFER FEED MATERIAL FOR MOLTEN SALT FUEL ELECTROCHEMICAL MEASUREMENT OF UCL4/UCL3 EQUILIBRIUM MEASUREMENT OF VISCOSITY AND DIFFUSTON IN LIQUID METALS EFFECT OF FREE-FLUORIDE ON FLUORIDE-SALT AND LIQUID METAL EQUILIBRIA MASS TRANSFER AND TRANSPORT IN FUSED SALT-LIQUID METAL ANOMOLTUS REHAVIOR OF NOBLE METAL FISSION PRODUCTS IN MSRE FUEL MOLTISTAGE L IQUID METAL-SALT EXTRACTION DISSOLUTION OF CLADDING IN MOLTEN MEYTALS PARTITION OF SOLUTES BETWEEN LEQUID METALS AND SALTS VISCOSETY AND TRANSPORT PROPERTIES OF LIQUID METALS MEASUREMENY OF VISCOSITY AND DIFFUSION IN LTIQUID METALS SOLUBILITY OF TUNGSTEN AND TANTALUM IN RARE-EAATH METALS OXIDE REDUCTION WITH MG-CU-CA REDUCTION OF UO2 WITH MG-IN DISTRIBUTION OF U AND PU BETWEEN MGCL2 SALTS AND CU-MG AND ZN-MG ALLOYS CORROSTION OF NIOSIUM AND TANTALUM BY MGCL2~NACL-KCL CALCULATED AND OBSERVED PROPERTIES FOR MISCIBILITY GAP SYSTEMS B¥L MOLTISTAGE MOLYBDENUM MSBR MSRE MSRE MSRE MSRE NACL-UCL3~-UCLS NAF NAF NEPTUNIUM NIOBIUM NIOBIUM NITRIDE-CARBIDE NOBLE NPF6 GXTDATION OX1IDE OXIDE DX IDE OXIDE OXIDE OXIDE OXIDE OX1IDE OXIDE PARTICLE PARTITION PB~SN-CD PB~SN-IN PHASE PHASE PHASE PHASE PHASE PHASE PILOT eILor PLANT PLANT PLANT PLANT PLUTONTUM PLUTONIUN PLUTONTUM PLUTONTUM PLUTONIUM PLUTONIUM PLUTONIUM PLUTONLUM PLUTONIUN PLUTONTUM—MAGNE PLUTON [UM-URANT PLUTONIUN KEYWORD INDEX MOLTISTAGE LIQUID METAL-SALY EXTRACTION SOLUBILITY OF TUNGSTEN, TANTALUM,NIOBIUM,VANADIUM,MOLYBDENUM AND CHROMIUM TN LIQUID PLUTONIUM ENGINEERING DEVELOPMENT Of PROCESSES FOR MSBR FUEL PREPARATION AND PROCESSING OF MSRE FUEL MSRE FUEL CHEMISTRY ANOMOLOUS BEHAVIOR OF NOBLE METAL FISSEON PRODUCTS IN MSRE FUEL EFFECT OF UF4-UF3 RATIO IN MSRE FUEL SOLUBILITY OF OXIDE IN NACL-UCL3-UCL4 ADSORPTION OF UF6 ON NAF REACTION OF NPFS6 WITH NAF BEHAVIDOR OF NEPTUNIUM DURING FLUORINAT ION CORROSION OF NIOBTUM AND TANTALUM BY MGCL2-NACL-KCL SOLUBILITY OF TUNGSTEN,TANTALUM,NIOBIUM,VANADIUM,MOLYBDENUM AND CHROMIUM IN LIQUID PLUTONIUM FISSION PRODUCT REMOVAL IN NITRIDE-CARBIDE CYCLE ANOMOLOUS BEHAVIOR OF NOBLE METAL FISSION PRODUCTS TN MSRE FUEL REACTION OF NPF6 WITH NAF OXIDATION OF UC PUREX PROCESS FOR LWR OXIDE FUELS DISSOLUTION OF PLUTONIUM-URANIUM OX IDE PYROSULPHATE DISSOLUTION OF OXIDE FUEL FLUORIDE VOLATILIYY PROCESS FOR OXIDE FUELS FUSED-SALT FLUORIDE VOLATILITY PROCESS FOR THORIUM-URANIUM OXIDE OR CARBIDE FUELS OXIDE REDUCTION WITH MG-CU-CA EFFECT OF OXIDE IN MOLTEN SALTS SOLUBILITY OF OXIDE IN NACL-UCL3-UCLS HYDROFLUORINATION AND OXIDE CHEMISTRY IN LIF-BEF2 HIGH TEMPERATURE TREATMENYT AND CHLORINATION OF COATED PARTICLE THTR FUELS PARTITION OF SOLUTES BETWEEN LIQUID METALS AND SALTS PB—SN-CD SYSTEM PB=-SN-ZN SYSTEM CU-MG-U PHASE DIAGRAM PLUTONFUM-MAGNESTUM PHASE DIAGRAM CALCULATION OF THERMODYNAMIL PROPERTIES FROM BINARY PHASE DIAGRAMS THERMOOYNAMIC DESCRIPTION OF BINARY PHASE BOUNDRIES THERMODYNAMIC ANALYSIS AND SYNTHESIS OF PHASE DIAGRAMS CALCULATION OF THERMODYNAMIC QUANTITIES FROM PHASE DIAGRAMS INVOLVING INTERSTITIAL TYPE PHASES PILOT PLANT EXPERIENCE ON PURIFICATION OF PLUTONIUM BY FLUORIDE VOLATILITY PILOT PLANTY EXPERIENCE WITH SKULL—OKIDE PROCESS TARAPYR REPROCESSING PLANT PILOYT PLANT EXPEREENCE ON PURIFICATION OF PLUTONIUM BY FLUORIDE VOLATILITY VOLATILITY PROCESS PLANT DESIGN AND PROCESSING LOAD PILOT PLANTY EXPERIENCE WITH SKULL-OXIDE PROCESS PILOT PLANT EXPERIENCE ON PURIFICATION OF PLUTONIUM BY FLUORIDE VOLATILITY ENGINEERING SCALE FLUORIDE VOLATILITY STUDIES ON PLUTONTUM BEARING FUEL PLUTONTUM RARE-EARTH SEPARATION URANIUM AND PLUTONIUM PURIFICATION BY SALT-TRANSPORT SOLUBILITY OF URANIUM AND PLUTONIUM IN CU-MG AND ZN-MG MOLTEN SALT EXTRACYION OF AMERICIUM FROM PLUTONIUM SOLUBILITY OF URANIUM AND PLUTONIUM IN LIQUID ALLOYS SOLUBILETY OF PLUTONIUM IN IN-MG ALLOYS SOLUBILITY OF TUNGSTEN, TANTALUM,NIOBIUM,VANADIUM, HOLYBDENUM AND CHROMIUM IN LIQUID PLUTONIUM PLUTONTIUM-MAGNES [IUM PHASE DIAGRAM DISSOLUTION OF PLUTONIUM-URANIUM OXIDE EBR-11 FUEL ALLOY STABILITY AND PLUTONIUN CONTENT 6%L PLUTONTUN-MAGNE POYTENTIAL POTENTIAL POTENTIALS PREDICTION PREPARATION PROCESS PROCESS PROCESS PROCESS PROCESS PROCESS PROCESS PROCESS PROCESS PROCESS PROCESS PROCESS PROCESSES PROCESSES PROCESSING PROCESSING PROCESSING PROCESS ING PROCESSING PROCESSING PROCESS ING PRODUCT PROPUCT PRODUCT PRODUCTION PRODUCTS PRODUCTS PRODUCTS PRODUCTS PROPERTIES PROPERTIES PROPERTIES PROPERTIES PROPERTIES PROPERTIES PROPERYIES PROPERTIES PROPERTIESFROM PROTACT INIUM PROTACTINIUM Py PU PUF & PUF6 PUREX PURIFACATION PURIFICATION PUREFICATION THERMODYNAMIC PROPERTIES OF PLUTONIUN-MAGNESIUM ALLOYS POTENTIAL OF FLUORIDE VOLATILITY PROCESS CHLORINE POYENTIAL AND CORRISION ELECTRODE POTENTTIALS IN LIF-BEF2 PREDICTION OF TERNARY BOUNDRIES AND THERMODYNAMIL PROPERTIESFROM BINARY DATA PREPARATION AND PROCESSING OF MSRE FUEL PUREX PROCESS FOR LWR OXIDE FUELS REPROCESSING OF URANIUM CARBIDE FUEL, CARBOX PROCESS TECHNOLOGY AND ECONOMICS OF FLUORIDE VOLATILITY PROCESS I[N FRANCE CHEMICAL REACYORS FOR FLUORIDE VOLATILITY PROCESS FLUORIDE VOLATILITY PROCESS FOR OXIDE FUELS POTENTTAL OF FLUORIDE VOLATILITY PROCESS YOLATILITY PROCESS PLANY DESIGN AND PROCESSING LOAD FUSED-SALT FLUORIDE VOLATILITY PROCESS FOR THORIUM-URANIUM OXIDE OR CARBIDE FUELS ERR-IT SKULL RECLAMATION PROCESS PILOY PLANT EXPERIENCE WITH SKULL-OXIDE PROCESS SALT TRANSPORT PROCESS CHLOREX PROCESS ENGINEER ING DEVELOPMENT OF PROCESSES FOR MSBR FUEL CONTAINERS FOR PYROCHEMICAL PROCESSES TECHNOLOGY AND ECONOMICS OF AQUEOUS PROCESSING, FRANCE TECHNOLOGY AND ECONOMICS OF AQUEOUS PROCESSING. TNDIA PREPARATION AND PROCESSING OF MSRE FUEL VOLATILITY PROCESS PLANT DESIGN AND PROCESSING LOAD CHLORIDE YOLATILITY PROCESSING OF THD2-UD2 AND UD2-PUDZ2 FUELS FISSION GAS RELEASE IN PYROCHEMICAL PROCESSING COMPATABILITY AND PROCESSING PROBLEMS OF MOLTEN URANYUM CHLORIDE-ALKAL! CHLORIDE FUELS FISSION PRODUCT REMOVAL IN NITRIDE-CARBIDE CYCLE VOLATILIZATION OF FISSION PRODUCT FLUORIDES SOLUBILITY PRODUCT IN SALT SOLUTIONS FLUDRINE EFFICIENCIES AND PUF6 PRODUCTION RATES VOLATILTIZATION OF FISSION PRODUCTS DISTRIBUTION OF FISSTON PRODUCTS AFTER HYDROCHLORINATION OF URANIUM FUEL DISTRIBUTION OF FISSION PRODUCTS AFTER FLUORINATION ANOMOLOUS BEHAVIOR OF NOBLE METAL FISSION PRODUCYS IN MSRE FUFL THERMODYNAMIC PROPERTIES OF CHLORIDES VISCOSITY AND TRANSPORY PROPERTIES OF LIQUID METALS THERMDDYNAMIC PROPERTIES OF PLUTONTUN-MAGNESTIUM ALLOYS CALCULATIONS OF SEPARATION PROPERTIES FOR THORIUM-URANIUM FUELS BY CHLORIDE VOLATILIZATION CALCULAYION OF THERMODYNAMIC PROPERTIES FROM BINARY PHASE DIAGRAMS EMPIRICAL RELATIONS AND THERMODYNAMIC PROPERTIES OF BINARY SYSTEMS CALCULATED AND OBSERVED PROPERTIES FOR EUTECTIC SYSTEMS CALCULATED AND OBSERVED PROPERTIES FOR MISCIBILITY GAP SYSTEMS PREDICTION OF TERNARY BOUNDRTES AND THERMODYNAMIC PROPERTIESFROM BINARY DATA PROTACTINIUM RECOVERY EXTRACTION OF PROTACTINIUM AND URANIUM FROM LIF-BEF2-THF4 INTO B8I-TH ALLOY REACTIONS DOF {U,PUYO2 WITH FLUDRINE DISTRIBUTION OF U AND PU BETWEEN MGCL2 SALTS AND CU-MG AND ZN-MG ALLOYS SEPARATION OF RUTHENIUM FROM PUF6 FLUORINE EFFICTIENCIES AND PUF6 PRODUCTION RATES PUREX PROCESS FOR LWR OXTDE FUELS PURIFACATION OF UF4-LIF EUTECTIC PILOT PLANT EXPERIENCE ON PURIFICATION OF PLUTONIUM BY FLUORIDE VOLATILITY URANIUM AND PLUTONIUM PURIFICATIAON 8Y SALT-TRANSPORT 0sL KEYWORD INDEX PYROCHEMICAL FISSION GAS RELEASE IN PYROCHEMICAL PROCESSING PYROCHEMICAL CONTAINERS FOR PYROCHEMICAL PROCESSES PYROSULPHATE PYROSULPHATE DISSOLUTION OF OXIDE FUEL QUANTITIES CALCULATION OF THERMODYNAMIC QUANTITIES FROM PHASE DIAGRAMS INVOLVING INTERSTIVIAL TYPE PHASES RARE-EARTH PLUTONIUM RARE-EARTH SEPARATION RARE-EARTH SOLUBILITY OF TUNGSTEN AND TANTALUM IN RARE-EARTH METALS RARE-EARTHS REDUCTIVE EXTRACTION OF RARE~EARTHS FROM L IF-BEF2-THF4 INTO BSISMUTH RATE RATE OF TRANSFER BETWEEN FLUDRIOE SALYT AND BISMUTH RATE EXTRACTION RATE AND STAGE EFFICIENCY RATES FLUORINE EFFICTIENCIES AND PUF6 PRODUCTION RATES RATIO EFFECT OF UF&~UF3 RATIO IN MSRE FUEL REACTION REACTION OF UD2 WITH CARBON TO FORM UC REACTION REACTION OF NPF6 WITH NAF REACTIONS REACTIONS OF {U,PUID2 WITH FLUORINE REACTIONS REACTIONS IN LIF-BEF2 SALT REACTORS CHEMICAL REACTORS FOR FLUORIDE VOLATILITY PROCESS RECEIVING FUEL SHIPPING AND RECEIVING RECLAMATION ERR=II SKULL RECLAMATION PROCESS RECOVERY PROTACTINIUM RECOVERY REDUCTION REDUCTION OF UO02 WITH MG-IN REDUCT ION SKULL-OXIDE REDUCTION REDUCT ION OXIDE REDUCTION WITH MG-CU-CA REDUCTION REDUCTION OF UCL4 WITH HYDROGEN REOUCTIVE REDUCTIVE EXTRACTION OF RARE-EARTHS FROM LIF-BEF2-THF4 INTO BISMUTH RELATIONS EMPIRICAL RELATIONS AND THERMODYNAMIC PROPERTIES OF BINARY SYSTEMS RELEASE FISSION GAS RELEASE IN PYROCHEMICAL PROCESSING REMOTE REMOTE FABRICATION OF EBR-1I FUEL REPPOCESSING AQUEOUS REPPOCESSING LMFBR FUEL REPROCESSING AQUEDUS REPROCESSING OF THORIUM-URANIUM FUELS REPROCESSING TARAPUR REPROCESS ING PLANT REPROCESSING MADRAS REPROCESSING COMPLEX REPROCESSING ERR-IT FUEL REPROCESSING REPRDCE SSING REPROCESSING OF URANIUM CARBIDE FUEL, CARBDX PROCESS RUTHENTUM SEPARATION OF RUTHENIUM FROM PUF6 SALY FEED MATERIAL FOR MOLTEN SALT FUEL SALT MOLYEN SALT ELECTROLYSTS OF URANIUM CARBIODE SALT CORRDSION IN ZRF4 FREE SALY SALY SALT TRANSPORT PRGCESS SALT RATE OF TRANSFER BETWEEN FLUORIDE SALT AND BISMUTH SALT MOLTEN SALT EXTRACYION OF AMERICIUM FROM PLUTONIUM SALT UOCL2 AND UOCL IN CHLORIDE SALT SALT THEORETICAL CONCEPTS FOR TERNARY MOLTEN SALT SYSTEMS SALY SOLUBILITY PRODUCT IN SALT SOLUTIONS SALT CALCULATION OF LIQUIDUS BOUNORIES FOR TERNARY SALT SYSTEMS SALT CHEMISTRY AND THERMODYNAMICS OF MOLTEM SALT FUELS SALT REACTIONS IN LIF-BEF2 SALTY SALT-L IQUID MASS TRANSFER AND TRANSPORT IN FUSED SALT-LIQUID METAL SALT-METAL EQUILIBRIA IN SALT-METAL SYSTEMS SALT-METAL TYPES OF CORROSION TESTS FOR SALT—-METAL SYSTEMS SALT-METAL CRAPHITE AND VITREDUS CARBON AS SALT-METAL CONTAIMERS SALT-METAL SILICON CARBIDE AND ALUMINA AS SALT-METAL CONTAINERS SALT-TRANSPDRT URANIUM AND PLUTONIUM PURIFICATION BY SALT-TRANSPORT SALTS PARTITION OF SOLUTES BETWEEN LIQUID METALS AND SALTS SALTS DISTRIBUTION OF U AND PU BETWEEN MGCL2 SALTS AND CU-MG AND IN-MG ALLODYS 164 SALTS SALTS SEPARATION SEPARATION SEPARATION SHIPPING SILICON SKULL SKULL-DXIDE SKULL-OXIDE SOLUBILITIES SOLUBILITY SOLUBILITY SOLUBILITY SOLUBILITY SOLUBILITY SOLUBILITY SOLUBILITY SOLUBILITY SOLUBILITY SOLUTES SOLUTES SOLUT EONS SORPTION STAINLESS~-STEEL STAINLESS-STEEL STEEL SYNTHESIS SYSTEM SYSTEM SYSTEM SYSYTEM SYSTEM SYSTEM SYSTEMS SYSTEMS SYSTEMS SYSTEMS SYSTEMS SYSTEMS SYSTEMS SYSTEMS TANTALUM TANTALUM TANTALUM TARAPUR TECHNOLODGY TECHNOLODGY TECHNOLOGY TERNARY TERNARY TERNARY TERNARY TESTS HEAT TRANSFER IN MOLTEN SALTS EFFECT OF OXIDE IN MOLTEN SALTS SEPARATION OF RUTHENTUM FROM PUFé PLUTONIUM RARE-EARTH SEPARATION CALCULATIONS OF SEPARATION PROPERTIES FOR THORIUM-URANIUM FUELS BY CHLORIDE VOLATILIZATION FUEL SHIPPING AND RECEIVING SILICON CARBIDE AND ALUMINA AS SALT-METAL CONTAINERS ERR=~IT SKULL RECLAMATION PROCESS PILOT PLANT EXPERIENCE WITH SKULL-OXIDE PROCESS SKULL-DOXIDE REDUCTION SOLUBILITIES AND ACTIVITEIES IN MOLTEN KNO3 SOLUBILITY SOLUBILITY SOLUBILITY SOLUBTILITY SOLUBILETY SOLUBILITY SOLUBILITY SOLUBILITY SOLUBILITY OF URANIUM AND PLUTONIUM IN CU-MG AND ZN-MG OF OXIDE IN NACL-UCL3-UCL4 OF URANTUM AND PLUTONIUM IN LIQUID ALLOYS IN TERNARY SYSTEMS OF URANIUM IN ZN-MG ALLOYS OF PLUTONIUM IN IN-MG ALLOYS OF TUNGSTEN, TANTALUM,NIOBIUM,VANADI UM, MOLYBDENUM AND CHROMIUM IN LIQUID PLUTONIUM OF TUNGSTEN AND TANTALUM IN RARE-EARTH METALS PRODUCT IN SALT SDLUTIONS PARTITION OF SOLUTES BETWEEN LIQUID METALS AND SALTS FREE ENERGIES OF SOLUTES IN LIF-BEF2 SOLUBTLITY PRODUCT IN SALT SOLUTIONS SORPTION AND DESORPTION OF CHLORIDES ON BACLZ2 STAINLESS-STEEL DECLADDING IN ZINC CORROSION OF STAINLESS-STEEL BY ZINC.CADIUM AND MAGNESTUM CORRISION OF LOW ALLOY STEEL B8Y LIQUID LEAD-BISMUTH ALLODY THERMODYNAMIC ANALYSIS AND SYNTHESIS OF PHASE DIAGRAMS DISTRIBUTION COEFFICIENTS IN LICL-KALCL& SYSTEM CD-PB-BI SYSYEM CAO-SI02-FEQ SYSTEM PB-SN-CD SYSTEM AG-PD-CU SYSTEM PB=-SN=-IN SYSTEM EQUILIBRIA TN SALT-METAL SYSTEMS TYPES OF CORROSION TESTS FOR SALT-METAL SYSTEMS SOLUBTLITY IN TERNARY SYSTEMS THEORETICAL CONCEPTS FOR TERNARY MOLYVEN SALT SYSTEMS CALCULATION OF LIQUIDUS BOUNDRIES FOR TERNARY SALT SYSTEMS EMPIRICAL RELATIONS AND THERMODYNAMIC PROPERTIES OF BINARY SYSTEMS CALCULATED CALCULATED AND OBSERVED PROPERTIES FOR EUTECTIL SYSTEMS AND OBRSERVED PROPERTIES FOR MISCIBILITY GAP SYSTEMS CORROSION OF NIGBIUM AND TANTALUM BY MGCL2-NACL-KCL SOLUBILITY SOLUBTILITY OF TUNGSTEN.TANTALUM,NIOBIUM,VANADIUM,MOLYBDENUM AND CHROMIUM IN LIQUID PLUTONIUM OF TUNGSTEN AND TANTALUM IN RARE-EARTH METALS TARAPUR REPROCESSING PLANT TECHNOL DGY TECHNOLOGY TECHNOL OGY SOLUBILITY AND ECONOMICS OF AQUEOUS PROCESSING, FRANCE AND ECONOMICS OF AQUEDUS PROCESSING, INDIA AND ECONOMICS CF FLUORIDE VOLATILEITY PROCESS IN FRANCE IN TERNARY SYSTEMS THEDRETICAL CONCEPTS FOR TERNARY MOLTEN SALT SYSTEMS CALCULATION OF LTQUIDUS BOUNDRIES FOR TERNARY SALT SYSTEMS PREQICTION OF TERNARY BOUNDRIES AND THERMODYNAMIC PROPERTIESFROM BINARY DATA TYPES OF CORROSION TESTS FOR SALT-METAL SYSTEMS est THALLIUM-BROMID THCOS THEORETICAL THERMAL THERMODYNAMIC THERMOOYNAMIC THERMOD YNAMIC THERMODYNAMIC THERMODYNAMIC THERMODYNAMIC THERMODYNAMIC THERMODYNAMIC THERMODYNAMIC THERMODYNAMICS THERMODYNAMICS THORTUM=-URANIUM THORIUM—URANIUM THORT UM-URANT UM THO2-y02 THTR TH2COL17 TRANSFER TRANSFER TRANSFER TRANSFER TRANSPORT TRANS PORT TRANSPORY TUNGSTEN TUNGSTEN ug2-pPuo2 URANIUM URANIUM URANTUM URANTUM URANTUM URANTUM URANIUN KEYWORD INDEX DISTRIBUTION OF THALLIUM-BROMIDE BETWEEN KNO3 AND AGBR THERMODYNAMICS OF FORMAT ION OF TH2FELT,TH2C017,FH2N117,THCOS,THNIS, THCU4 AND THN12 THEORETICAL CONCEPTS FOR TERNARY MOLTEN SALT SYSTEMS THERMAL DECOMPOSITION IN UF6-PUF6 MIXTURES THERMODYNAMIC PROPERTIES OF CHLORIDES THERMODYNAMIC PROPERTIES OF PLUTONIUN-MAGNESIUM ALLOYS THERMODYNAMIC DATA FOR CHLORIDES CALCULATION OF THERMODYNAMIC PROPERTIES FROM BINARY PHASE DIAGRANMS THERMODYNAMIC DESCRIPTION OF BINARY PHASE BOUNDRIES EMPTRICAL RELATTONS AND THERMODYNAMIC PROPERTIES OF BINARY SYSTEMS THERMODYNAMIC ANALYSIS AND SYNTHESIS OF PHASE DIAGRAMS PREDICTION DF TERNARY BOUNDRIES AND THERMODYNAMIC PROPERTIESFROM BINARY DATA CALCULATION OF THERMODYNAMIC QUANTITIES FROM PHASE DIAGRAMS [NVOLVING INTERSTITIAL TYPE PHASES CHEMISTRY AND THERMODYNAMICS OF MOLTEM SALT FUELS THERMODYNAMICS OF FORMATION OF THZ2FEL17+TH2CO17+TH2NI17,THCOS, THNIS,THCU4 AND THNI?2 AQUEOUS REPROCESSING OF THORIUM-URANIUM FUELS FUSED-SALT FLUORIDE VOLATILITY PROCESS FDOR THORIUM-URANIUM OXIDE OR CARBIDE FUELS CALCULATIONS OF SEPARATION PROPERTIES FOR THORIUM-URANIUM FUELS BY CHLORIDE VOLATILIZATION CHLORIDE VOLATILITY PROCESSING OF THO2-UOZ AND UO2-PUO2 FUELS HIGH TEMPERATURE TREATMENT AND CHLORINATION OF COATYED PARTICLE THTR FUELS THERMODYNAMICS OF FORMATION OF THZFE17,THZCO17,TH2NIL7,THCOS, THNIS, THCUS AND THNIZ RATE OF TRANSFER BETWEEN FLUDRIDE SALT AND BISMUTH HEAT TRANSFER IN MOLTEN SALTS MASS TRANSFER AND TRANSPORT IN FUSED SALT-LIQUID METAL INTERPHASE MASS TRANSFER SALT TRANSPORT PROCESS MASS TRANSFER AND TRANSPORT IN FUSED SALT-LIQUID METAL VISCOSITY AND TRANSPORY PROPERTIES OF LIQUID METALS SOLUBILITY OF TUNGSTEN,TANTALUM,NIOBIUM,VANADIUM, MOLYBOENUM AND CHROMIUM IN LTQUID PLUTONTIUM SOLUBILITY OF TUNGSTEN AND YANTALUM IN RARE-EARTH METALS REACTIONS OF (U,PUI02 WITH FLUORINE DISTRIBUTION OF U AND PU OXIDATION OF UC BETWEEN MGCL2 SALTS AND CU-MG REACTION OF U002 WITH CARBON TO FORM UC ELECTROCHEMICAL MEASUREMENY OF UCL4/UCL3 EQUILIBRIUM REDUCTION OF UCL4 WITH HYDROGEN ELECTROCHEMICAL MEASUREMENT OF UCLA/UCL3 EQUILIBRIUM PURIFACATION OF UF4-LIF EUTECTIC EFFECT OF UF4-UF3 RATIO IN MSRE FUEL ADSORPTION OF UF6 ON NAF THERMAL DECOMPOSITION IN UF6-PUF6 MIXTURES UOCL2 AND UOCL IN CHLORIDE SALT HYDROFLUORINATION OF vo2 REACTION OF 102 WITH CARBON TO FORM UC FLUORINATION OF UD2 WITH BRF5 AND F2 REDUCTION OF U022 WITH MG-IN CHLORIDE VOLATILITY PROCESSING OF THOZ-UQZ AND UD2-PUD2 FUELS REPROCESSING OF URANIUM CARBIDE FUEL, CARBOX PROCESS MOLTEN SALT ELECTROLYSIS OF URANIUM CARBIDE HYDROCHLORINATION OF URANIUM FUEL AND YOLATILEZATION OF URANIUM FUEL DISTRIBUTION OF FISSION PRODUCTS AFTER HYDROCHLORINATION OF URANIUM FUEL CHLORINATION-DISTILLATION OF IRRADIATED URANIUM DIOXIDE URANIUM AND PLUTONIUM PURIFICATION BY SALT-TRANSPORT SOLUBILITY OF URANTUM AND PLUTONIUM IN CU-MG AND IN-MG AND ZN—MG ALLOYS £9L 363 405 547 5513 566 264 490 501 450 141 143 163 177 211 231 236 261 279 1718 111 194 200 645 327 446 347 349 553 558 320 291 URANTUM URANTUM URANTUM URANTIUM VANADIUM VAPDR VISCOSITY VISCOSITY VITREOUS VOLATILITY VOLATILITY VOLATILITY VOLATILITY VOLATILITYY VOLATILITY YOLATILITY VOLATILITY VOLATILITY VOLATILIZATION VOLATILIZAFION VOLATILIZATION VOLATILIZATION VOLATIL TZATION ZINC TINC IN-NG IN-MG IN-MG IN-MG IN-MG-U IRF4& A4 — EXTRACTION OF PROTACTINIUM AND URANTUM FROM LIF