W NS e, anememie Sty SaP - e T el e - 3 4456 03L0L0OS8 3 i ORNL-4782 ¥ ! . - Pl \fl MOLTEN-SALT REACTOR PROGRAM Semiannual Progress Repott Period gnding Cerbhuahg 29, 1972 OAK RIDGE NATIONAL LABORATORY . CENTRAL RESEARCH LIBRARY DOCUMENT COLLECTION LIBRARY LOAN COPY DO NOT TRANSFER TO ANOTHER PERSON If you wish someone else to see this " document, send in name with document and the library will arrange a loan. LUCN-7969 i3 3-67, L OAK RIDGE NATIONAL LABORATORY OPERATED BY UNION CARBIDE CORPORATION e FOR THE U.S. ATOMIC ENERGY COMMISSION Printed in the United States of America. Available from National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road, Springfield, Virginia 22151 Price: Printed Copy $3.00; Microfiche $0.95 This report was prepared as an account of work sponsored by the United States Government. Neither the United States nor the United States Atomic Energy Commission, nor any of their employees, nor any of their contractors, subcontractors, or 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, ORNL-4782 UC-80 — Reactor Technology Contract No. W-7405-eng-26 MOLTEN-SALT REACTOR PROGRAM SEMIANNUAL PROGRESS REPORT For Period Ending February 29, 1972 M. W. Rosenthal, Program Director R. B. Briggs, Associate Director P. N. Haubenreich, Associate Director OCTOBER 1972 s T for the 3 quE D US. ATOMIC ENERGY COMMISSION 360108 3 ORNL-2474 ORNL-2626 ORNL-2684 ORNL-2723 ORNL-2799 ORNL-2890 ORNL-2973 ORNL-3014 ORNL-3122 ORNL-3215 ORNL-3282 ORNL-3369 ORNL-3419 ORNL-3529 ORNL-3626 ORNL-3708 ORNL-3812 ORNL-3872 ORNL-3936 ORNL4037 ORNL4119 ORNL-4191 ORNL4254 ORNL4344 ORNL-4396 ORNL4449 ORNL-4548 ORNL4622 ORNL4676 ORNL-4728 This report is one of a series of periodic reports in which we describe the progress of the program. Other reports issued in this series are listed below. Period Ending January 31,1958 Period Ending October 31, 1958 Period Ending January 31, 1959 Period Ending April 30, 1959 Period Ending July 31, 1959 Period Ending October 31, 1959 Periods Ending January 31 and April 30, 1960 Period Ending July 31, 1960 Period Ending February 28, 1961 Period Ending August 31, 1961 Period Ending February 28, 1962 Period Ending August 31, 1962 Period Ending January 31, 1963 Period Ending July 31,1963 Period Ending January 31, 1964 Period Ending July 31, 1964 Period Ending February 28, 1965 Period Ending August 31, 1965 Period Ending February 28, 1966 Period Ending August 31, 1966 Period Ending February 28, 1967 Period Ending August 31, 1967 Period Ending February 29, 1968 Period Ending August 31, 1968 Period Ending February 28, 1969 Period Ending August 31, 1969 Period Ending February 28, 1970 Period Ending August 31, 1970 Period Ending February 28, 1971 Period Ending August 31, 1971 Contents PART 1. MSBR DESIGN AND DEVELOPMENT L. DESIGN . ottt e e e e e e e 2 1.1 Molten-Salt Demonstration Reactor Design Study ........ ... ... i 2 0 I 7= 1 =3 O PR 2 1.1.2 ReaCtOT COTC . ittt ittt e e e e e e e e e e e e e e e s 2 1.1.3 Graphite Temperatures . .. ... ..ottt e 6 1.1.4 Cell COOlNG . .ottt it et it e et et et e 8 1.2 Side-Stream Processing of MSBR Primary Flow for lodine Removal ......................... 8 1.3 MSBR Industrial Design Study .. ... ...t 9 1.4 MSBE DESIgN ..ottt ittt e et e e 12 1.5 Bubble Behavior in the MSBR Primary Salt System ... ... ... . . .. . i i 13 2. REACTOR PHY SICS . .ttt ettt e e e e e e e ettt e e 18 2.1 Experimental PhySics ... . ... ..ottt 18 2.1.1 HTLTR Lattice Experiments ......... ... .. oot 18 2.2 Physics Analysis Of MSBR . ... ... . 21 2.2.1 Radiation Heatingin MSR Pumps . ... ... ... i 21 2.2.2. MSBE Control-Rod Worths .. ... e e 22 2.2.3 Molten-Salt Converter Reactors Using Plutonium . .. .......... ... i, 23 3. SYSTEMS AND COMPONENTS DEVELOPMENT ... ... . i e 28 3.1 Gaseous Fission Product Removal . .. ... ... . . e 28 3.1.1 Bubble Separator and Bubble Generator .............. ... ... 28 3.1.2 Bubble Formation and Coalescence Test .. .. ... ... ..o .. 29 3.1.3 Bubble Separator ANalyses .. .. ... ...ttt 29 3.2 Gas System Technology Facility ......... ... .. i 32 3.3 Molten-Salt Steam Generator Industrial Program . ........ ... . . . . . ... i il 33 3.4 Coolant-Salt Technology Facility . ... ... ... i 33 3.5 SAlt PUMPS . .ottt ettt e ... 34 3.5.1 Salt Pumps for MSRP Technology Facilities . . ............. ... oo, 34 3.5.2 ALPHA PUMD .. oottt ettt e et e e e e e 35 3.5.3 Molten-Salt Mixer, Laboratory Scale . ........ .. ... . i 37 4. INSTRUMENTATION AND CONTROLS . .ttt i e e e e e e 38 4.1 Transient and Control Studies of the MSBR System Using a Hybrid Computer ................. 38 iii iv 5. HEAT AND MASS TRANSFER AND PHYSICAL PROPERTIES . ....... ... ... ..., 39 5.1 Heat Transfer . ... e e e e e e e e 39 5.2 Wetting Studies . ..... .. i e e 40 5.3 Mass Transfer to Circulating Bubbles .. .. .. ... .. . . e 41 PART 2. CHEMISTRY 6. FISSION PRODUCT BEHAVIOR . . .. o e e e it eane s 45 6.1 Some Factors Affecting the Deposition Intensity of Noble-Metal Fission Products .............. 45 6.2 Effects of Selected Fission Products on Hastelloy N, Nickel, and Type 304L Stainless Steel at 650°C . . ..o\ttt i ittt e 50 6.3 Reaction of CoF3 with Tellurium . . . ... oo e e 51 7. BEHAVIOR OF HYDROGEN AND ITS ISOTOPES . .. ... . e 53 7.1 Solubility of Hydrogenin Molten Salt ... ... ... ... i i i s, 53 7.2 Initial Tritium Chemistry in the Core of a Molten-Salt Reactor .......... ... ... .. ... .. .... 53 7.3 Permeation of Hydrogen Through Metals at Low Pressures ........... ... ... ... .. .. ... 54 7.4 Influence of Films or Coatings on Hydrogen Permeation Rates ,........... ... ... ... ..... 56 7.5 Experiments on Hydrogen Evolution from Fluoroborate Coolant Salt . ............. ... ... .. 57 7.6 Apparatus for Infrared Spectral Studies of Molten Salts . .......... .. .. .. ... ... ... 59 7.7 Infrared Spectral Studies of the Chemical Behavior of BF;OH ™ and BF;0D " Ionsin Molten NaF-NaBF, ... ... ... . i 59 7.8 Thermal Stability of NaNO3, KNO;, NaNO, ,and RITEC ... ......... ... ... .. ... ... .. .. 62 8. FLUOROBORATE CHEMISTRY . ..t it et e e e e et et e e 63 8.1 Solubility of BF3 in Fluoride Melts ... ... .. e 63 8.1.1 Reactor Applications .. .... ... ...t e e i 63 8.2 Free Energies of Formation of NaFeF; and NaNiF;; Their Relationship to the Corrosion of Hastelloy N by Fluoroborates .. ........ ... ... ... 65 8.3 Preparation of Fused Sodium Fluoroborate for the Coolant Salt Technology Facility ............ 68 0. PROTACTINIUM CHEMIS T RY ... e e e e e 70 9.1 Oxide Chemistry of Protactinium in MSBR Fuel Salt . .. . ... ... ... .. ... ... . ... o .. 70 9.2 Binary Solid Solutions of PaO, and Other Actinide Dioxides and Their Exchange Equilibria with Molten-Salt Reactor Fluorides .. ... ... ... ... ... .. .. .. .. ... ... 72 9.3 Termary Solid Solutions of ThO,,Pa0,,and UQ, ... .. ... . i 74 10. DEVELOPMENT AND EVALUATION OF ANALYTICAL METHODS FOR MOLTEN-SALT REACTORS . .. ..o e e e e e e e e 77 10.1 In-Line Chemical Analysis of Molten Fluoride Salt Streams .. ...... ... ... ... .. ... ... ... 77 10.2 Theoretical Considerations of the Voltammetric In-Line Determination of Uranium(III) . .. ... e 80 10.3 Electroanalytical Studies of Titanium(IV) in Molten LiF-BeF, -Z1F, (65.4-29.6-5.0 MOIE 20) -« oottt e 81 11. 10.4 Electrochemical Studies of Bismuth(III) in Molten LiF-BeF,-ZrF, at 500°C .................. 82 10.5 Voltammetry of Chromium(III) in Molten NaBF,-NaF (92-8 Mole %) .......... ... ... ...... 82 10.6 Voltammetric and Hydrolysis Studies of Protonated Species in Molten NaBF, . ................ 83 10.7 Determination of Hydrogen in NaF-NaBF, Salts ... ...... .. .. ... . . . . L. 83 10.8 Spectral Studies of Molten Salts . .. .. ... .. 84 OTHER MOLTEN-SALT RESEARCHES . ... .. e e e 87 11.1 The Oxide Chemistry of Niobium in Molten LiF-BeF, Mixtures ............... ... ... ..... 87 11.1.1 Equilibrations of Nb, Os and BeO with Molten LiF-BeF, Mixtures . ................... 87 11.1.2 Equilibrations Involving Nickel Niobates in Molten Li,BeF, .......... ... ... .. ... ... 89 11.2 The Reaction of MoF¢ with Niobium ... ... ... . e 91 11.3 Thermodynamics of LiF-BeF, Mixtures .. ... ... ... . .. . . i 95 11.4 Electrochemical Mass Transport in Molten Beryllium Fluoride—Alkali Fluoride MiXIUTES . . . ..ottt e e e et e e et e e e e e e e e 96 11.5 Electrical Conductance in Beryllium Fluoride Rich NaF-BeF, Mixtures ...................... 98 11.6 The Disproportionation Equilibrium of UF; Solutions . ............ ... ... ... ... .. ..., 98 11.7 The Raman Spectra of Be, F,> and Higher Polymers of Beryllium Fluorides in the Crystalline and Molten State ... ... ... .. . . . . . . i 100 11.8 Raman Spectra of Molten and Crystalline Potassium Dichromate ........................... 103 11.9 Nonideality of Mixing in the Systems Li, BeF,-Lil, Na, BeF,-Nal, and C82 BeF4 B 106 PART 3. MATERIALS DEVELOPMENT INTERGRANULAR CRACKING OF STRUCTURAL MATERIALS EXPOSED TO FUEL SALT .. .. e e e e e e e e et et i 109 12.1 Examination of Hastelloy N Components fromthe MSRE .. ... . ... ... ... .. .. ... .. .. 109 12.1.1 Freeze Valve 105 . ... .. ot e e et e 109 12.1.2 Control Rod Thimble . . ... ... e e e e 111 12.1.3 Sampler Cage Rod ... ... e e 115 12.1.4 Mist Shield . . .. ... e 116 12.2 Auger Analysis of the Surface Layers on Graphite from the Core of the MSRE .. ............... 117 12.3 Auger Analysis of the Surface of a Fractured Hastelloy NSample . .......................... 122 12.4 Intergranular Corrosion of Hastelloy N . ... ... .. . . 123 12.5 Tube-Burst EXperiments . .. .. ..ot 126 12.6 Cracking of Samples Electroplated with Tellurium . ....... ... .. ... ... ... ... oLt 128 12.7 Cracking of Hastelloy N Being Creep Tested in Tellurium Vapor . ......... .. ... ... ... ..., 128 12.8 Intergranular Cracking of Materials Exposed to Sulfur and Several Fission Product Elements .. .. ... o e et e e e 134 12.9 Mechanical Properties of Hastelloy N Modified with Several Elements .. ...................... 136 12.10 Status of Intergranular Cracking Studies . ......... ... . . . . . i 141 vi 13. GRAPHITESTUDIES . .......... P 144 INtrOdUCHION . . .. e 144 13.1 Graphite Development . ... . e e 144 13.2 Procurement of Various Grades of Carbon and Graphite .................... ... .. ........ 149 13.3 Texture Determinations ... ... ...ttt ittt et it e 149 13.4 Thermal Property Testing . .. ... ...ttt e it 150 13.5 Nominal Helium Permeability Parameters for Various Grades of Graphite ..................... 151 13.6 Reduction of Helium Permeability of Graphite by Pyrolitic Carbon Sealing ................... 151 13.7 Characterization of Pyrocarbon Sealants for Graphite Using Reflected Light and Scanning Electron MiCrosCopes .. . ... ...ttt ettt 155 14, HASTELLOY N ..o e e e e et e e e e e e et et e et 161 14.1 Development of a Titanium-Modified Hastelloy N . . ... ... ... . o o i i 161 14.2 Alloys with Exceptional Strength . .. ... .. .. . e 163 14.3 Weldability of Commercial Alloys of Modified Hastelloy N . ......... ... .. ... .. ... .. ... .... 166 14.4 Electron Microscope StUudies ... ... ... ...ttt i e e 171 14.4.1 Intermetallic Precipitation in Hastelloy N .. ... ... ... .. ... .. o L. 172 14.4.2 Precipitation in New Commercial Alloys . .... ... ... .. i 173 14.4.3 Modification of X-Ray Diffraction Techniques ............. ... ... ... ............ 173 14.5 Salt Corrosion StudIes . .. ... ..ottt i e e e e e 174 14.5.1 Fuel Salt ... . . i i ittt e e e i e et e 174 14.5.2 Fertile-Fissile Salt . . ... ... ... e e 177 14.5.3 Blanket Salt .. ... ... . e 177 14.5.4 Coolant Salt ... .. ...t e e e 177 14.6 Forced-Convection Loop Corrosion Studies ... ... ...ttt i e e 179 14.6.1 Operation of Loop MSR-FCL-1A . .. ... ... it 179 14.6.2 Results from Loop MSR-FCL-1A . ... .. e et icee e 179 14.6.3 Operation of Loop MSR-FCL-2 ... ... .. . e 180 14.6.4 Results from Loop MSR-FCL-2 . ... ... e 180 14.7 Corrosion of Hastelloy Nin Steam . . ... ... . i ettt e e eee e 182 14.8 Evaluation of Duplex Tubing for Use in Steam Generators ............... ... ... ..., 189 15. SUPPORT FOR CHEMICAL PROCESSING .. ... . it e e et et e e 192 15.1 Construction of a Molybdenum Reductive-Extraction Test Stand . .......................... 192 15.2 Fabrication Development of Molybdenum Components .......... ... ... ... . ... ..., 194 15.3 Welding of Molybdenum ... ... . e e e 195 15.4 Development of Brazing Techniques for Fabricating the Molybdenum Test Loop ............... 195 15.5 Compatibility of Materials with Bismuth . ....... ... ... . ... ... ... ... . . ... 197 15.5.1 Tantalum and T-1 10 ... o i i ittt e et ettt e 197 15.5.2 Graphite . ... ... o e 198 15.5.3 Tungsten-Coated Hastelloy N . . ... ... . e e 199 15.5.4 Molybdenum ... ... ... e e 199 15.6 Molybdenum Braze Alloy Compatibility ...... ... . ... .. . .. 202 16. 17. 18. 19. vii PART 4. MOLTEN-SALT PROCESSING AND PREPARATION FLOWSHEET ANALY SIS ..o e e e e e e e e e e e e e e 16.1 Design Study and Cost Estimates of a Processing Plant for a 1000-MW(e) MSBR 16.2 Multiregion Code for MSBR Processing Plant Flowsheet Calculations .............................................................. ........................ PROCESSING CHEMISTRY . . ..ot e e e e e e e e e e et 17.1 Distribution of Lithium and Bismuth between Liquid Lithium-Bismuth Alloys and Molten LiCl .. . ... e 17.2 Solubility of Europium in Liquid Bismuth . .. ... .. . . . 17.3 Integral Heats of Lithium-Bismuth Solutions .. ........ ... .. . . . . . . . . .. 17.4 Protactinium Oxide Precipitation Studies . . .. ... ... ... . 17.5 Chemistry of Fuel Reconstitution .. ...... ... ... i e ENGINEERING DEVELOPMENT OF PROCESSING OPERATIONS .. ... ... ... ... .. .. .. 18.1 Lithium Transfer during Metal Transfer Experiment MTE-2 ...... ... ... ... ... ... ... .. .. 18.2 Operation of Metal Transfer Experiment MTE-2B . ...... ... ... ... ... ... .. .. ... ... ..... 18.3 Installation, Testing, and Charging of Materials to the Third Metal Transfer Experiment . ... ... .. . i i i e 18.4 Design of the Metal Transfer Process Facility . ......... ... ... ... . .. .. . . ... 18.5 Development of Mechanically Agitated Salt-Metal Contactors .............................. 18.6 Reductive Extraction Engineering Studies . .......... .. ... . . 18.7 Design of the Reductive-Extraction Process Facility ............ ... ... ... ... ... ... ... 18.8 Frozen-Wall Fluorinator Development .. ... .. ... . . . i 18.9 Engineering Studies of Uranium Oxide Precipitation .......... ... ... .. ... ... ... ... 18.10 Design of a Processing Materials Test Stand and the Molybdenum Reductive Extraction Equipment ... ... . e 18.11 Development of a Bismuth-Salt Interface Detector .......... .. ... . ... .. . ... . ... CONTINUOUS SALT PURIFICATION . ... e e it e Introduction The objective of the Molten-Salt Reactor Program is the development of nuclear reactors which use fluid fuels that are solutions of fissile and fertile materials in suitable carrier salts. The program is an outgrowth of the effort begun over 20 years ago in the Aircraft Nuclear Propulsion program to make a molten-salt reactor power plant for aircraft. A molten-salt reactor — the Aircraft Reactor Experiment — was operated at ORNL in 1954 as part of the ANP program. Our major goal now is.to achieve a thermal breeder reactor that will produce power at low cost while simultaneously conserving and extending the nation’s fuel resources. Fuel for this type of reactor would be 233UF, dissolved in a salt that is a mixture of LiF and BeF,, but 235U or plutonium could be used for startup. The fertile material would be ThF, dissolved in the same salt or in a separate blanket salt of similar composition. The technology being developed for the breeder is also applicable to high-performance converter reactors. A major program activity through 1969 was the operation of the Molten-Salt Reactor Experiment. This reactor was built to test the types of fuels and materials that would be used in thermal breeder and converter reactors and to provide experience with operation and maintenance. The MSRE operated at 1200°F and produced 7.3 MW of heat. The initial fuel contained 0.9 mole % UF,, 5% Z1F,4, 29% BeF,, and 65% " LiF; the uranium was about 33% ?°°U. The fuel circulated through a reactor vessel and an external pump and heat exchange system. Heat produced in the reactor was transferred to a coolant salt, and the coolant salt was pumped through a radiator to dissipate the heat to the atmosphere. All this equipment was constructed of Hastelloy N, a nickel-molybdenum-iron-chromium alloy. The reactor core contained an assembly of graphite moderator bars that were in direct contact with the fuel. iX Design of the MSRE started in 1960, fabrication of equipment began in 1962, and the reactor was taken critical on June 1, 1965. Operation at low power began in January 1966, and sustained power operation was begun in December. One run continued for six months, until terminated on schedule in March 1968. Completion of this six-month run brought to a close the first phase of MSRE operation, in which the objective was to demonstrate on a small scale the attractive features and technical feasibility of these systems for civilian power reactors. We concluded that this objective had been achieved and that the MSRE had shown that molten-fluoride reactors can be oper- ated at 1200°F without corrosive attack on either the metal or graphite parts of the system, the fuel is stable, reactor equipment can operate satisfactorily at these conditions, xenon can be removed rapidly from molten salts, and, when necessary, the radioactive equipment can be repaired or replaced. The second phase of MSRE operation began in August 1968, when a small facility in the MSRE building was used to remove the original uranium charge from the fuel salt by treatment with gaseous F,. In six days of fluorination, 221 kg of uranium was removed from the molten salt and loaded onto ab- sorbers filled with sodium fluoride pellets. The decon- tamination and recovery of the uranium were very good. After the fuel was processed, a charge of 2?3 U was aaded to the original carrier salt, and in October 1968 the MSRE became the world’s first reactor to operate on 233U. The nuclear characteristics of the MSRE with the 23U were close to the predictions, and the reactor was quite stable. In September 1969, small amounts of PuF; were added to the fuel to obtain some experience with plutonium in a molten-salt reactor. The MSRE was shut down permanently December 12, 1969, so that the funds supporting its operation could be used elsewhere in the research and development program. Most of the Molten-Salt Reactor Program is now devoted to the technology needed for future molten- salt reactors. The program includes conceptual design studies and work on materials, the chemistry of fuel and coolant salts, fission product behavior, processing methods, and the development of components and systems. Because of limitations on the chemical processing methods available at the time, until three years ago most of our work on breeder reactors was aimed at two-fluid systems in which graphite tubes would be used to separate uranium-bearing fuel salts from thorium-bearing fertile salts. In late 1967, however, a one-fluid breeder became feasible because of the devel- opment of processes that use liquid bismuth to isolate protactinium and remove rare earths from a salt that also contains thorium. Our studies showed that a one-fluid breeder based on these processes can have fuel utilization characteristics approaching those of our two-fluid designs. Since the graphite serves only as moderator, the one-fluid reactor is more nearly a scaleup of the MSRE. These advantages caused us to change the emphasis of our program from the two-fluid to the one-fluid breeder; most of our design and development effort is now directed to the one-fluid system. Summary PART 1. MSBR DESIGN AND DEVELOPMENT 1. Design The design study of the 300-MW(e) demonstration reactor plant was completed. The internal structure of the reactor was developed in considerable detail. Special attention was given to methods for fabrication and assembly of the different graphite pieces. Hydraulic conditions in the reactor were analyzed and tempera- tures of graphite were calculated. Cooling equipment was added to the cell atmosphere circulation systems that had been provided for heating the reactor equip- ment cells. A new analysis of some earlier experiments on the stripping of iodine from LiF-BeF, melts has shown that, if one includes the diffusion of I~ in the melt to the gas-liquid interface as a rate limiting step, there results a mathematical model which is in accord with the experimental data. This model permits making an estimate of what it would take to continually remove iodine from the MSBR fuel by side-stream processing. Design studies on the MSBE core were continued, with emphasis on the design of a slab-type graphite element. Layout studies of the core include provisions for four cruciform-shaped control rods. A computer program (BUBBLE) has been written to describe in detail the behavior of gas bubbles circulating with the salt through the MSBR primary system. This program is being incorporated into an overall program to describe the detailed behavior of noble gases in an MSBR. The MSBR Industrial Design Study Team under Ebasco Services completed the design report on their selected 1000-MW(e) reference concept during this period. 2. Reactor Physics MSBR lattice physics experiments, performed for us at Battelle Northwest Laboratories, have been com- X1 pleted and reported. In the course of preparing for a careful analysis of these experiments, we have reviewed and revised our cross-section library. The revised data for most nuclides are based on ENDF/B version II. Exceptions are carbon, lithium, and fluorine, for which other data are justified. In connection with the lattice experiments, the problem of calculating neutron reac- tion rates in a doubly heterogeneous system (coated- particle fuels in a lumped fuel rod) was studied in detail. For resonance neutron absorption, conventional methods of analysis contain conceptual errors, at least for a laminar system. For random spherical grains, the conclusion with respect to validity of conventional methods is not clear, but we found that the grain effects could be neglected for the small grains used in our experiment. Qur cross-section preparation code, XSDRN, was modified to treat the grain effects explicitly in the thermal neutron range. In connection with the preparation of specifications for MSR pumps, we calculated the amount of energy deposited in various parts of a typical MSR pump by beta and gamma rays from fission products in the salt and in the cover gas, as well as by neutrons and gammas from other parts of the primary salt loop. Reactivity worths of proposed control rods for the MSBE were calculated. Graphite salt-displacement rods and poison rods of Hastelloy N, boron in graphite, and europium in graphite were studied. In further studies of MSR operation with limited fuel processing, we calculated initial critical fuel loadings and fuel compositions as functions of time for various thorium concentrations with plutonium from light- water reactors as initial fuel loading and feed. Because of the high effective absorption cross section in 2*°Pu, it seems desirable to use a low thorium concentration, for example, 6 mole %, during the initial batch cycle, when the plutonium concentration is highest, Subse- quent cycles would use higher thorium concentrations, for example, 10 mole %, and would burn, primarily, bred 233U. 3. Systems and Components Development The performance and pressure-drop testing of the GSTF (gas system technology facility) bubble separator design have been completed on the water test loop. The final design has a 44-in. separation length, a tapered casing, and gas removal from both the swirl and recovery hubs. This design has a suitably high removal efficiency of the small-diameter bubbles associated with the 31% CaCl, test fluid and has a stable vortex under all normal operating conditions. The pressure-drop measurements of the bubble gener- ator indicated a larger than expected increase in the required gas supply pressure as the gas flow was increased to the design value. Efforts to reduce this pressure are in progress. The bubble formation and coalescence test rig was assembled and operated. The test results show that bubbles present in 66-34% LiF-BeF, immediately after agitation are larger than in the 31% CaCl, solution but are smaller than in demineralized water. A test capsule of 72-16-12% LiF-BeF,-ThF,; MSR fuel salt contained suspended material that made observation of bubbles impossible. Analyses were started with the objective of idealizing the swirl-flow bubble separator to develop expressions which would be helpful in understanding the perfor- mance of this separator. The effect on removal effi- ciency of the bubble size distribution and the turbulent diffusion of bubbles away from the vortex cavity are included in the studies. The design of most of the components for the GSTF is nearing completion, and the detailed design of the facility piping was started. It was determined that the loop is too closely coupled to permit accurate predic- tions of the pressure distributions and that variable flow restrictors would be required to properly balance the pressures. The loop will be operated initially with water to permit calibration of these flow restrictors and to obtain confirmation of the performance of the modi- fied salt pump. The preliminary system design descrip- tion for the GSTF is ready for publication. The subcontract with Foster Wheeler Corporation for a four-task conceptual design study of molten-salt steam generators was approved near the end of the period. After the completion of a surface arrangement study, Foster Wheeler will proceed with the conceptual design analysis on the unit of their selection. Task I is scheduled for completion in October 1972. The mechanical design of the coolant-salt technology facility is complete, and the instrument and contro! design is 90% complete. The fabrication and installation xii of the mechanical components are nearing completion, and the installation of the instruments and electrical control equipment is imminent. Heat transfer tests, run with water, on the concentric tube economizer for the corrosion product cold trap indicate that the tempera- ture of the salt supply to the cold trap can be lowered a satisfactory 200°F when the cold trap is operated at a maximum AT of 100°F. Detailed assembly procedures were prepared for the salt pumps to be used in the coolant-salt technology facility and the gas system test facility. The refurbish- ment of these pumps is under way. Plans were made to perform water tests with the salt pump in the gas system test facility. The ALPHA pump, equipped with Viton elastomeric seals in its lower shaft seal, has operated satisfactorily in the MSR-FCL-2 for approximately 3900 hr. Data on pump coastdown and lubricating oil flows and tempera- tures were taken for use in setting operating set points. The design of several improvements was completed. The layout and design of a small molten-salt mixer for laboratory use were completed. 4. Instrumentation and Controls The hybrid computer model of the MSBR system has been used to investigate thermal transients in the salt systems for representative perturbations. A report describing the model and some typical results is being published. 5. Heat and Mass Transfer and Physical Properties Heat transfer. Additional determinations of heat transfer coefficient for a proposed MSBR fuel salt have been made in the transitional range of Reynolds modulus (3000—4000) at low inlet fluid temperatures for which the temperature coefficient of viscosity is a large negative value [-0.2 b ft! hr' (°F)7]. In particular, for operation at Reynolds modulus 3161 without an adiabatic entrance length, the slowly varying axial gradient in wall temperature downstream from the heated length suggests the absence of eddy diffusion and hence that the flow is laminar. The heat transfer coefficient for this case is well below that predicted by the Hausen correlation for the transition region. We attribute the delayed transition to turbulent flow to the influence of heating a fluid having a negative tempera- ture coefficient of viscosity. Wetting studies. The bubble-pressure technique has been employed to determine the contact angle for the salt mixture (LiF-BeF,-ThF4-UF4; 67.5-20-12-0.5 mole %) on Hastelloy N as a function of time and tempera- ture. It was found that low-oxygen-content salt (~85 ppm) does not wet a Hastelloy N surface upon initial exposure at 1000°F, but gradually becomes strongly wetting. At 1230°F the salt returns to the partial wetting condition. The increase in wetting with time is believed to be due to moisture in the helium purge gas which results in formation of an oxide film on the surface. Mass transfer to circulating bubbles. The dependence of the mass transfer Sherwood modulus for bubbles and liquids in cocurrent turbulent flow on Reynolds modu- lus has been examined theoretically for the flow regimes distinguished by a bubble Reynolds modulus less than or greater than 2. For the former case the exponent of the Reynolds modulus is found to be 0.92, whereas for the latter case the exponent is 0.66. The experimentally observed value is 0.94. The experi- mentally determined dependence of the Sherwood modulus on bubble size is significantly greater than that predicted for either regime. PART 2. CHEMISTRY 6. Fission Product Behavior The analysis of data from an experimental array exposed in the MSRE core shows that on paired metal and graphite surfaces, tellurium and molybdenum were deposited considerably more strongly on metal than on graphite. Niobium and ruthenium were less intensely deposited, favoring metal surfaces only moderately more than graphite. Such individual characteristics will have to be considered in anticipating their behavior under MSBR system conditions. The presence of superficial grain boundary cracks on Hastelloy N samples from the MSRE fuel system has prompted a joint investigation by the Reactor Chemis- try Division and the Metals and Ceramics Division of those fission products which might have contributed to the process. Preliminary tests have exposed specimens of Hastelloy N, nickel, and type 304L stainless steel to elemental vapors of tellurium, selenium, sulfur, iodine, cadmium, arsenic, and antimony at 650°C for 1000- to 2000-hr periods. Only those specimens of Hastelloy N that were exposed to tellurium alone showed effects similar to that found in the MSRE fuel system. Nickel specimens exposed to tellurium showed much less severe attack, and all others showed little if any deleterious effect. Tellurium forms the volatile fluorides TeF, and TeF4, both of which are readily reduced to elemental Xiii tellurium by UF; (or by stronger reducing agents). Accordingly, a simple method for controlled generation of these materials would permit corrosion testing of metal specimens in molten-salt mixtures to which known and controlled additions of tellurium are being made. Experimentation has shown that the reaction of tellurium with anhydrous CoF; affords such a conve- nient generation method. Reaction of tellurium with CoF; when mixed as powders is rapid, but cannot be well controlled. When these reagents are placed in a compartmented cell, such that they can mix only upon vaporization of tellurium, TeF¢ appears as the cell is heated to 225°C; increasing pressures of TeF are ob- served as the temperature is raised to 275°C. A small quantity of TeF, appears in the TeF¢ at 300°C, and no other vapor species appear below 400°C. Apparatus for using this method to introduce tellurium into molten- salt systems is under construction. 7. Behavior of Hydrogen and Its Isotopes Hydrogen and helium solubilities in Li, BeF, have been determined at 600°C over the pressure range 1 to 2 atm. As expected, in both cases the solubility varied linearly with saturation pressure over the range of pressure investigated. Chemical events associated with birth of tritons through neutron capture in lithium and with thermali- zation of the energetic triton have been considered. Path length for thermalization is such that in an MSBR core only a few percent of the tritons should reach graphite before thermalization. Production of atomic fluorine during thermalization of the triton adds an insignificant fraction to the (low) concentration of F° due to thermalization of the fission fragments. Overall, it appears that previously assumed equilibria involving U3, U™, TF, and tritium should be applicable. Deuterium permeation through Hastelloy N has been measured over the pressure range from 30 torr down to less than 1072 torr. Although earlier workers have uniformly reported variance from a dependence as the 14, power of pressure at pressures below 10 to 100 torr, no such departure was found in the present work. Accordingly, the extrapolations which have been used within MSRP for calculation of tritium permeation appear correct. Initial experiments on type 304L stainless steel showed that an oxide film decreased the permeation flow and that such a film resulted in deviation from dependence on 4 power of the pressure. Presently available (and very simple) models of film behavior are inadequate to explain the observations. Measurements of hydrogen diffusing out of closed capsules containing fluoroborate from a convection loop confirmed analyses of substantial (26 to 40 ppm) concentrations of hydrogen within the salt. The mea- surements also provided an estimate of the equilibrium quotient of the reaction OH™ + 3 Cr(s) = 5Cr® + 0% + Y, H, (g) in molten fluoroborate at 535°C. Infrared absorption techniques are valuable in quanti- tative study of BF;O0H™ and BF3;0D™ ions in molten fluoroborates. These studies have now been greatly facilitated by design and fabrication of an LaFj;- windowed cell and furnace assembly. Operation of the system with NaF-NaBF, mixtures to 450°C appears quite satisfactory. Continued study, by infrared absorption, of NaF- NaBF, mixtures indicates, as before, the band at 3645 cm ™ due to the stretching mode of OH in BF;OH™ and that at 2690 cm™' due to the OD stretch in BF3;0D". These studies have, in addition, revealed that simple bubbling with D, of a clean NaF-NaBF, melt containing BF;OH™ does not lead to appreciable exchange of deuterium for hydrogen in the BF;OH™. Reaction of the admitted D, to produce D™ prior to the exchange appears necessary, though the species responsible for the oxidation of D, to D™ has not been certainly identified. Hitec, a commercial heat transfer medium consisting of NaNOQO;, NaNQ,, and KNOj;, has been considered as a coolant for an MSBR since it would certainly oxidize and retain tritium. Thermal stability of this mixture to 600°C and of the individual components to 400°C has been examined, using a mass spectrometer to identify the gaseous species evolved. 8. Fluoroborate Chemistry Measurements of the solubility of BF; in LiF-BeF, were continued, using an improved more-rapid tech- nigue. The measurements continue to show that over the composition range 50 to 85 mole % LiF, BF, solubility varies almost linearly with the thermo- dynamic activity of LiF. A study of the solubility of BF; in MSBR fuel solvent indicated that BF3 gas is potentially useful in a reactivity control system. The formation free energies of NaNiF; and NaFeF; have been estimated from heterogeneous-equilibria mea- surements. This permitted estimation of the most oxidizing conditions which can exist in the presence of Xiv molten mixtures of NaF and NaBF4without significant oxidation of nickel. Approximately 1550 1b of NaBF,-NaF (92-8 mole %) was prepared for the Coolant Salt Technology Facility. Analyses of representative salt samples show that sodium fluoroborate mixtures of sufficient purity can be prepared from commercially available materials by evaporation of water vapor during the melting process. 9. Protactinium Chemistry Studies of various equilibria involving protactin- ium(IV) and (V), MSBR fuel solutions, and solid oxides have been completed. Protactinium(IV) and (V) con- centrations have been measured as a function of temperature at oxide saturation, and the potential of the Pa*/Pa’* couple has been estimated. From these results it is evident that protactintum could be precipi- tated in a process side stream as Pa®* oxide while avoiding oxide precipitation in the main fuel circuit, where the Pa** state is maintained by control of the redox potential (U*/U%). Equilibrium measurements have been performed for the distribution of Pa** between molten MSBR solvent salt and solid solutions of Pa0,-ThO,. The equilibrium quotients obtained, together with similar data for other tetravalent actinide ions, yield a correlation with the lattice parameters of the pure actinide dioxides. The activity coefficients of ThO,, Pa0,, and UQ, in the ternary oxide solid solutions which can be precipi- tated by oxide from MSBR fuels have been estimated from a previous correlation for binary oxide solid solutions. At 600°C, 7p,q, is predicted to vary in the range 1.0 to 1.4, and the Pa/U ratio in the oxide solid solution precipitated from an MSBR fuel will be ~Y% of the corresponding ratio in the fuel. 10. Development and Evaluation of Analytical Methods for Molten-Salt Reactors Automated in-line measurement of U(III) concentra- tions in a thermal convection loop continued to show excellent reproducibility and to demonstrate the ad- ventitious oxidation of the U(IlI) by contaminants introduced during the insertion of corrosion specimens and the recovery of the U(III) by oxidation of the chromium from the structural metal and specimens. A comparison of the experimental voltammetric waves for U(II) with those computed from theoretical considera- tions has shown that, while the reduction of chromium makes a significant contribution to the early part of the wave, an excellent match between theoretical and experimental data is obtained in the region of the peak reduction current. Moreover, the presence of chromium does not introduce any significant error in the potential at which the maximum in the derivative wave occurs. The location of the derivative maximum is used to compute the U(III) concentration. Under the reducing conditions at which the loop has been operated, chromium is the only corrosion product that is present in significant concentrations. Although the voltam- metric determination of chromium is complicated by its proximity to the larger uranium wave, we have found it possible to extract a measurable chromium wave from the reoxidation curves by a stripping technique. Calibra- tions of this method are in progress. We have investigated the electroanalytical chemistry of titanium and bismuth in LiF-BeF,-ZrF,. The use of Ti(IV) as a corrosion indicator for modified Hastelloy N appears to be impractical because of the sublimation of TiF, from the melts. Conversely, TiF; was found to be stable in this solvent and exhibited a well-defined oxidation wave, conforming to a reversible oxidation of Ti(IlI) to Ti(IV), which should be useful for in-line measurement of titanium in fuel. Voltammetric and chronopotentiometric measurements of BiF; have shown that it can be reversibly deposited in a single-step three-electron reduction at a variety of electrodes. Evidence of alloy formation was noted at platinum and silver electrodes but not at graphite and iridium. Peak currents vary linearly with concentration, and diffusion coefficients of about 10 ™® cm?/sec were measured. Investigation of the voltammetric reduction of Cr(III) in coolant melts showed that, while the reduction wave occurs close to the cathodic limit of the NaF-NaBF, melts, the resolution is adequate for peak current measurements. The diffusion coefficient at 440°C was found to be about 2 X 107®cm? /sec. Additional work is needed to validate the method for in-line analysis. Continued studies of the reduction of protons from coolant melts at an evacuated palladium electrode have indicated that the previously observed instability of the protonated species in the melt is associated with the glass envelope used to contain the system. Presently, we are attempting to resolve problems associated with lower-than-predicted voltammetric currents and with poor reproducibility of the quantity of hydrogen which diffuses into the evacuated electrode during electrolysis. We are continuing to search for a method to validate the infrared pellet technique for the determination of BF;O0H " in NaBF,. While an isotopic-exchange method tends to confirm the infrared results, it is subject to excessive blanks and occasionally gives exorbitantly XV high results. We are attempting to prepare samples of negligible proton concentration and to find other methods of standard proton addition. We are currently using spectrophotometric techniques to investigate, with members of other divisions, a variety of problems associated with reactor fuel chemis- try and reprocessing systems. These include the evalua- tion of the potential of the Pa(IV)-Pa(V) couple, the solubility of CuO in fluoride melts, the measurement of U(V) in fluoride melts, and the measurement of the spectra of LiCl melts in contact with lithium alloys. New techniques developed during this work include an improved method for the introduction of protected samples to the spectrophotometric furnace and a technique for sparging a sample contained in a window- less cell. 11. Other Molten-Salt Researches The solubility of Nb,Os in molten LiF-BeF, mix- tures saturated with BeO is observed to be too high to be explained by the formation of the relatively unstable and volatile NbFs. It appears that the species formed in solution is NbOFs?* or NbOF4®". In the presence of nickel the solubility of niobium decreases by a factor of 10 or more because of the formation of nickel niobates (NiNb, O4 and NiygNb,Og). The products of the reaction of MoF¢ with metallic niobium were monitored over the temperature range 25 to 750°C. The molybdenum fluoride vapor products appeared at nearly the same temperatures at which they were observed when MoF, was reduced with molyb- denum. No mixed niobium-molybdenum fluoride species appeared, but the effusing vapors did react with tantalum to yield the previously known compound MoTaF,, and a previously undiscovered similar com- pound NbTaF,,. Oxide impurities appeared predomi- nantly as MoOF, below 450°C and predominantly as Nb,OF, from 500 to 750°C. Nonideality of molten alkali fluoride—beryllium fluo- ride mixtures is discussed in terms of (1) the application of conformal ionic solution theory to the interaction parameters and (2) the formation of fluoroberyllate ions and their noninteraction with fluoride ions. Chronopotentiometric measurements have been initi- ated to investigate the formation of resistive BeF, films on anodization of beryllium electrodes in molten alkali fluoride—beryllium fluoride melts. Electrical conductance of beryllium fluoride—sodium fluoride mixtures has been obtained for compositions between 70 and 95 mole % beryllium fluoride and temperatures between 485 and 600°C. Equilibration of pure UC, U,C;, and UC, with UF, and UF; dissolved in 2 LiF+BeF, with analysis by spectrophotometric observations has shown that UC, is the most stable of the carbides at 550°C. From these data and plausible assumptions regarding activity of the carbide phases, values for free energy of formation of the carbides can be calculated. Single crystals of the congruently melting compound Na, LiBeF, have been shown by x-ray diffraction to contain the Be,F,® ion, consisting of two BeF;*” tetrahedra sharing a corner. Comparison of Raman spectra of crystalline Na, LiBeF, with those for this compound in the molten state permits explicit identifi- cation of BeF,* and Be,F, ions in the melt. These data, along with those for molten mixtures of LiF and BeF,, where the situation is more complex, afford for the first time, direct evidence for at least the simplest of the Be-F polymers postulated by C. F. Baes in his models of LiF-BeF, mixtures. A comprehensive study of the Raman spectra of K,Cr, O, as crystalline solid, in saturated aqueous solution at 25°C, and in the molten state is under way. Since the Cr,0,* ion is isomorphic with Be,F, 77, it is hoped that this study will aid in interpretation of the spectrum of the latter ion. The phase diagrams of the systems M, BeF4-MI (M = Li, Na, Cs) have been measured for a better understand- ing of the effects of ion size, charge, and polarizability. All systems exhibit positive deviation from ideality to a varying degree. Only in dilute solutions of M, BeF, in MI is there a high degree of dissociation of the BeF4 2 ion. PART 3. MATERIALS DEVELOPMENT 12. Intergranular Cracking of Structural Materials Exposed to Fuel Salt Several components from the MSRE fuel system have been examined in more detail. These studies have shown that (1) the cracking was less in a freeze valve where the fission product concentrations were lower than in the circulating loop, (2) samples from the control rod thimble and spacer showed little or no effect of salt velocity on crack severity, (3) a sampler cage rod deformed to fracture had deep cracks in the part exposed to liquid and fewer cracks where the part was exposed to gas, and (4) samples from the outer part of the mist shield had deep cracks whether they were exposed to flowing salt or to salt mist. Intergranular corrosion has been shown to occur in very oxidizing fuel salt. However, the cracks are so Xvi shallow that it is unlikely that corrosion accounts for the cracking observed in the MSRE. Tube burst samples of Hastelloy N tested in helium and salt environments showed no effect of environment on the rupture life, but samples plated with 0.01 mg/ecm?® of tellurium failed short of the expected rupture life. Cracks 5 to 10 mils deep were present in the specimens plated with tellurium. Similar cracking has been caused in creep specimens that were stressed and exposed to tellurium vapor simultaneously. Samples exposed to tellurium vapor or electroplated with tellurium cracked when strained after being annealed at 650°C. Ni-200 and type 304L stainless steel were more resistant to cracking by tellurium. Several alloys containing small additions of sulfur and several fission products were tested; those containing tellurium or sulfur had poor mechanical properties at high temperatures. 13. Graphite Studies Fabrication studies at ORNL are continuing, but largely directed at applications other than nuclear. Because much of the work is concerned with relating physical properties to microstructure, and such rela- tions are obviously of direct concern to the radiation damage problem, the findings are reported here. Rela- tionships between strength and elasticity of graphites have been observed for the ORNL materials. Those graphites which are monolithic in structure have about 50% higher strengths for a given modulus than the more conventionally fabricated materials. A similar linear relationship between modulus and density has been found, implying that pore morphology is the key free variable dominating the mechanical properties. Measurements of thermal conductivities on lightly irradiated graphites (2 to 4 X 10?! neutrons/cm?®) have indicated much sharper reductions in conductivity than were expected at operating temperatures. The general deterioration of permeability with damage noted in both gaseous impregnated and coated samples was revealed by use of a scanning electron microscope to be due to defective surface structures. Cracking at sharp corners, soot inclusions in the pyrolytic carbons, nonuniform coatings, and numerous surface cracks have been observed. These defects were not obvious in the optical microscope, but now that we know what to look for, they can certainly be identified at relatively low magnification. Improved coatings have been produced, using propene as the source of pyrocarbon, in a new coating furmace using a mandrel support for the samples. The coatings are extremely uniform in both thickness and texture, and no microcracking has been found even at the high magnifications of the scanning microscope. Permeabili- ties in the 107 to 107'% cm?/sec range have been easily obtained. 14. Hastelloy N Samples of small commercial alloys containing from 0.5 to 2.0% Ti have been irradiated and creep tested. The alloy with 2.1% Ti has acceptable properties after irradiation at 760°C. Three small commercial alloys with nominal additions of 2% Nb and 0.5% Ti have exceptionally good strengths at 650°C. The improved strength seems due to the formation of a very fine strain-induced precipitate. Two new small commercial heats with nominal additions ot 2% Ti were received and found to have excellent weldability. Phase analysis work has shown that alloys modified with 1% Ti contain some Ni3Ti and that alloys with 2.9% Ti contain much larger amounts. The quantity of NizTi in alloys with 2.4% Ti does not seem embrittling. Several thermal-convection loops with salts containing LiF, BeF,, ThF,, UF,, and ZrF, have continued to operate and have very low corrosion rates. Several thermal-convection loops and two pumped systems containing sodium fluoroborate continue to operate. These systems show that corrosion in sodium fluoro- borate is strongly dependent on impurities, tempera- ture, and flow rate. The corrosion rate of Hastelloy N in steam continues to be <0.25 mil/year. However, the results from stressed specimens are inconclusive. The duplex nickel 280 and Incoloy 800 tubing has good properties. The sole flaw noted to date is that the nickel 280 forms intergranular cracks at low strains, but the ductility can be improved by reducing the oxide content of the nickel. 15. Support for Chemical Processing Work continued toward the construction of a molyb- denum reductive-extraction test stand. A full-size mock- up was used to establish the step-by-step fabrication sequence. Construction began on a head-pot sub- assembly prototype that involves all of the operations required to fabricate the actual component. All of the 37%-in.-OD closed-end back extrusions required for the head pots and disengaging sections were fabricated and machined. Tubing for the various connecting lines was obtained, inspected, and certified for use. A decision was made to use electron-beam welding to join the back extrusions to form the head pots and Xvii disengaging vessels. A technique was developed in which the half sections are held together during welding by molybdenum pins through a step joint, and several successful welds were made. Additional progress was made on (1) tube-tube welding of %- and 1%-in.-OD tubing using an orbiting-arc weld head and (2) vent-tube welding which requires exceptional alignment to pre- vent damage of the vent tube during electron-beam welding. Brazing experiments were continued to develop para- meters for back-brazing and sleeve-brazing operations. Portable heating equipment was developed that will enable us to field braze either inside the dry box or outside the dry box with local atmosphere protection of the heated area. Compatibility experiments of Mo, Ta, T-111, and graphite in bismuth and bismuth-lithium solutions at 600 to 700°C were continued. A T-111 alloy loop containing T-111 samples in Bi—2.5 wt % Li was operated for 3000 hr, and postoperation examinations were started. A molybdenum loop containing molyb- denum samples in Bi—2.5 wt % Li and scheduled for 3000-hr operation was started. A Hastelloy N loop that had been internally coated with chemically vapor- deposited tungsten operated for only 24 hr at a maximum temperature of 700°C before bismuth com- pletely penetrated the wall of the Hastelloy N tubing. Capsule tests on ATJ, AXF, and Graph-i-tite “A” for 500 hr at 700°C indicated that high-purity bismuth did not intrude into any of the materials. Identical capsules exposed to Bi—3 wt % Li suffered varying degrees of metal intrusion, with the most porous ATJ being completely penetrated in SO0 hr. In a slightly different test, however, an ATJ graphite crucible exposed to Bi-2.2 wt % Li at 650°C for 720 hr did not indicate any significant metal intrusion. Molybdenum braze specimens showed small weight gains when first exposed to H,-HF mixtures but larger weight losses when subsequently exposed to a molten salt at 650°C. Most of the weight loss was attributed to dissolution of iron from the braze alloy. PART 4. MOLTEN-SALT PROCESSING AND PREPARATION 16. Flowsheet Analysis A design study and a cost estimate were completed for a fluorination—reductive-extraction-—-metal transfer processing plant that continuously processes the fuel salt from a 1000-MW(e) MSBR. The design study pointed out the need for additional development work xXviil in three important areas: (1) finding materials of construction suitable for containing molten bismuth and bismuth-salt mixtures, (2) determining the chemical behavior of noble metals in an MSBR, and (3) prevent- ing entrainment of bismuth in salt leaving bismuth-salt contactors. A computer code that can be used for calculating steady-state concentrations and heat generation rates in an MSBR processing plant is being developed. The behavior of a total of 687 nuclides in 56 regions that represent the processing plant is presently treated by this code. An algorithm that has been developed for solving the set of 38,000 algebraic equations that represents the system results in solution of the equa- tions by an iterative technique with the use of an acceptably small amount of computer time. The code has been used for calculating heat generation rates and concentrations of fission products in a processing plant whose operation is based on the fluorination- reductive-extraction—metal transfer flowsheet. In the future, the code will be used for carrying out para- metric studies of this flowsheet and for making com- parative studies of flowsheets based on other processing methods such as oxide precipitation. 17. Processing Chemistry Measurements were made of the equilibrium distribu- tion of lithium and bismuth between liquid lithium- bismuth alloys and molten LiCl over the temperature range 650 to 800°C. Integral heats of formation of liquid lithium-bismuth alloys were calculated from data available in the literature. Additional data were obtained to define more ac- curately the conditions required for the precipitation of protactinium from LiF-BeF,-ThF, (72-16-12 mole %) solutions containing UF, by sparging the salt with H, O-HF-Ar gas mixtures. Equilibrium quotients for the reaction PaFs(d) + %,H,0(g) = %Pa,05(s) + SHF(g) were found to be 3.9 £ 0.5 and 21 £ 4 at 600 and 650°C respectively. In studies related to the chemistry of fuel reconstitu- tion, investigation of the reaction UF¢(g) + UF,4(d) =2 UF;(d) was continued. Gold apparatus was found to be stable both to gaseous UFg and to UF; dissolved in LiF-BeF,-ThF, (72-16-12 mole %) at 600°C. Under certain conditions, UFs disproportionated, with the rate of disproportionation being second order with respect to UFs concentration. 18. Engineering Development of Processing Operations Studies related to the development of a number of processing operations were continued during this report period. Additional information concerning the behavior of lithium during metal transfer experiment MTE-2 was obtained. The results are consistent with a recently developed correlation of data for the distribution of lithium and bismuth between LiCl and lithium-bismuth solutions. Operation of experiment MTE-2B was con- tinued in order to further study the transfer of lithium from lithium-bismuth solutions containing lithium at concentrations of 3.7 to 16 at. %. The data obtained thus far are consistent with the expected behavior of lithium in a metal transfer process system. Installation of equipment for the third engineering experiment for development of the metal transfer process (MTE-3) has been completed. To date, the equipment has been leak tested and treated with hydrogen for the removal of oxides. Bismuth for the contactor has been treated with hydrogen and filtered before being trans- ferred to the contactor. The fluoride salt was con- tacted with bismuth containing thorium before being filtered and transferred to the fluoride salt surge tank. Preparations are under way for charging the LiCl and the 5 at. % Li-Bi solution to the system; when charging of the phases is complete, we will be ready to make the first run. We are designing a facility in which we will carry out the fourth metal transfer experiment (MTE-4). The experiment will use salt flow rates that are 5 to 10% of those whict will be required for processing a 1000-MW(e) MSBR. The experiment will generate information on the rate of transfer of rare earths in equipment of a design suitable for a processing plant and will allow evaluation of potential materials of construction. In particular, we plan to evaluate graphite as a material for the salt-metal contactor. Valuable information will also be collected during the use of graphite and metal components in the same facility. Overall mass transfer coefficients were obtained with a water-mercury system in a stirred-interface contactor of the type being used for development of the metal transfer process. These experimentally determined co- efficients are quite close to those predicted by ex- trapolation of the Lewis correlation, which is based on experimental values obtained from solvent-water sys- tems at much lower Reynolds numbers than are encountered with salt and bismuth. Experiments were continued in which the rates of transfer of ®7Zr and 237U from molten salt to bismuth were measured during the countercurrent contact of XixX salt and bismuth in a packed column. Design and development work were initiated for the Reductive Extraction Process Facility (REPF), which will allow testing and development of equipment of a design suitable for use in a full-scale protactinium removal process based on fluorination—reductive extraction. A preliminary examination and a conceptual design for the system have been partially completed. The facility will allow operation of all steps of the reductive extraction process for isolation of protactinium with salt flow rates as high as about 25% of those required for processing a 1000-MW(e) MSBR. In preliminary tests of equipment for studying in- duction heating in molten salt as a potential heat source for nonradioactive tests of a frozen-wall fluorinator, arcing between the coil and the vessel partially melted the induction coil. Although the arcing was attributed to the use of argon as the inert cover gas, it prompted reexamination of autoresistance heating for the internal heat source. In autoresistance heating tests with a fluorinator mockup, the desired mode of operation was not achieved because the frozen salt layers formed with low heat fluxes were not solid and hence were not electrically insulating. Higher heat fluxes are needed in order to obtain a lower wall temperature, and a large temperature gradient is required to suppress dendrite formation in order to produce a solid, electrically insulating frozen layer. Operation of a small-scale engineering facility has been continued in order to investigate the precipitation of UO,-ThO, solid solutions from molten fluoride salt by contacting the salt with argon-water gas mixtures. The salt temperature, which was varied from 540 to 630°C, was observed to have little effect on the rate of precipitation. However, the precipitation rate has been found to be directly proportional to the rate at which water is supplied to the system, while the composition of the gas mixture has little effect on water utilization. The precipitates have been observed to settle rapidly, and the salt and precipitates have been separated by decantation accompanied by only a small amount of entrainment. Samples of the salt and oxide precipitate have shown that the two phases are not in equilibrium. A model for the precipitation process in which the solids, once formed, do not equilibrate with the salt has been found to agree quite well with experimental data. Oxide precipitation continues to appear to be an attractive alternative to fluorination for removing ura- nium from fuel salt that is free of protactinium. Design of the components for the processing materials test stand and the molybdenum reductive extraction equipment was continued, and fabrication of some of the structural parts of the test stand was started. Specific accomplishments include: design of the ex- pansion loops in the molybdenum tubing; completion of the preliminary piping drawings and the construction of a full-scale mockup of the loop; design of the molybdenum equipment support system; design of a field assembly jig and a handling system so that the loop can be field assembled in a horizontal position and transported to the point of operation, where it will be erected to the vertical position; and design of the containment vessel, the seal-welded flange, the freeze valves, and the transition joint nozzles. An eddy-current type of detector is being developed to allow detection and control of the bismuth-salt interface in salt-metal extraction columns or mechani- cally agitated salt-bismuth contactors. The probe con- sists of a ceramic form on which bifilar primary and secondary coils are wound. The ceramic form is placed inside a molybdenum tube in order to protect the coils from attack by salt or bismuth. A high-frequency alternating current, which is passed through the primary coil, induces a secondary coil current having an ampli- tude and a phase shift that are dependent on the conductivities of the materials adjacent to the coils. In tests of the probe at 550 to 700°C by the phase shift technique, the measurements were found to correlate quite satisfactorily with the bismuth level around the probe. The temperature dependence of the indicated level varies from 0.003 in./°C at the upper end of the probe to 0.009 in./°C at the lower end. It appears that the probe is a sensitive and practical indicator for determining the bismuth level or for locating the salt-bismuth interface. 19. Continuous Salt Purification System Four flooding runs and 17 iron fluoride reduction runs were made with the new column, which is packed with Y%,-in. Raschig rings having a thinner wall and 32% greater free volume than the Y,-in. Raschig ring packing used previously. The maximum flow rates that can be obtained with the new column are three times those which could be obtained with the earlier packing, although the mass transfer coefficients are only 25% of those observed previously. The use of *°Fe tracer during the studies decreased the error in the reported values for the iron concentration significantly over values obtained with colorimetric analyses. A batch experiment in which the FeF, concentration in salt was decreased by dilution and by hydrogen sparging also demonstrated the advantage of using *°Fe tracer counting rather than colorimetric analysis as the method for determining iron concentration. The packed XX column was also used to obtain salt holdup data. Salt holdup in the column was found to increase linearly from about 5% of the free column volume at a salt flow rate of 100 cm?®/min to 11% of the free volume at a flow rate of 500 cm®/min. Part 1. R. B. Briggs The design and development program has the purpose of describing the characteristics and estimating the performance of future molten-salt reactors, defining the major problems that must be solved in order to build them, and designing and developing solutions to prob- lems of the reactor plant. To this end we have published a conceptual design for a 1000-MW(e) plant and have contracted with an industrial group organized by Ebasco Services Incorporated to do a conceptual design of a 1000-MW(e) MSBR plant. This design study uses the ORNL design for background and is to incorporate the experience and the viewpoint of industry. One could not, however, propose to build a 1000-MW(e) plant in the near future, so we have done some studies of plants that could be built as the next step in the development of large MSBR’s. One such plant is the Molten-Salt Breeder Experiment (MSBE). The MSBE is intended to provide a test of the major features, the most severe operating conditions, and the fuel reprocessing of an advanced MSBR in a small reactor with a power of about 150 MW(t). An alterna- tive is the Molten-Salt Demonstration Reactor (MSDR), which would be a 150- to 300-MW(e) plant based largely on the technology demonstrated in the Molten- Salt Reactor Experiment, would incorporate a mini- mum of fuel reprocessing, and would have the purpose of demonstrating the practicality of a molten-salt reactor for use by a utility to produce electricity. The present AEC program does not include construction of an MSBE or MSDR, so these design studies have recently been discontinued to free the effort for use on problems of more immediate concern. In addition to these general studies of plant designs, the design activity includes studies of the use of various MSBR Design and Development P. N. Haubenreich fuel cycles in molten-salt reactors and assessment of the safety of molten-salt reactor plants. The fuel cycle studies have indicated that plutonium from light-water reactors has an economic advantage over highly en- riched 235U for fueling molten-salt reactors. Some studies related to safety are in progress preliminary to a comprehensive review of safety based on the ORNL reference design of a 1000-MW(e) MSBR. The design studies serve to define the needs for new or improved equipment, systems, and data for use in the design of future molten-sait reactors. The purpose of the reactor development program is to satisfy some of those needs. Presently the effort is concerned largely with providing solutions to the major problems of the secondary system and of removing xenon and handling the radioactive off-gases from the primary system. The design is nearing completion for the gas system tech- nology facility for use in testing the features and models of equipment for the gaseous fission product removal and off-gas systems and for making special studies of the chemistry of the fuel salt. Construction is well along on the coolant system technology facility for use in studies of equipment, processes, and chemistry of sodium fluoroborate for the secondary system of a molten-salt reactor. The steam generator is a major item of equipment for which the basic design data are few and the potential problems are many. A program involving industrial participation has been undertaken to provide the technology for designing and building reliable steam generators for molten-salt reactors. As the first step in this program a contract was negotiated with Foster Wheeler Corporation for a conceptual design of a steam generator for use with molten salt. 1. Design E. S. Bettis 1.1 MOLTEN-SALT DEMONSTRATION REACTOR DESIGN STUDY E. S. Bettis Alexander W. K. Crowley Collins J. P. Sanders H. L. Watts L.G. C.W. 1.1.1 General The design study of the 300-MW(e) demonstration plant (MSDR) using a single-fluid molten-salt reactor was completed during this period. A report on this concept was drafted and will be issued as an ORNL-CF memorandum. This summary report concludes work on the demonstration reactor plant. In completing the study, most of the effort went into examining details of the design of the reactor core. Analyses were made of graphite temperatures, hydraulic pressure drops, and flow distribution. Results of the hydraulic analyses caused us to make some changes in the manner of distributing the flow through the core and reflector. In principle, the core is the same as has been described previously, but changes have been made which make the flow less ambiguous. A significant effort was put into the design of the reflector which completely surrounds the core. Previ- ously the methods of assembling the reflector and of accommodating the differential expansion between graphite and vessel had not been defined. This has now been done so that the assembly of the reflector and core is described and structural problems have been solved. Also, hot-spot graphite temperatures for the core and reflector have been calculated. A further addition to the concept involved incorpo- rating a cooling heat exchanger into each of the three cell atmosphere circulating systems. These coolers can, by butterfly valves, be switched into the circuits in place of the cell heaters. In this way the cell atmosphere can be cooled for removing afterheat from equipment after shutdown. Use of these coolers guarantees that nothing in the cell will overheat, even after an emer- gency or accidental drain of the primary salt. Two other rather minor additions were made to the plant design. The one large tank of water to which the drain tank heat is dumped by the natural-convection NaK circuits was divided into three tanks, any two of which would provide sufficient cooling. Also, instead of boiling water from these tanks, water flows through them to remove the heat. Boiling is employed only if water circulation is interrupted. The temperature of one type of boron control rod was calculated to determine if such a rod would overheat. This calculation gave 1260°F for the hot-spot metal temperature of the rod, and this is an acceptable temperature. 1.1.2 Reactor Core The design of the core region has not changed from that reported previously! except in the top orificing, which will be explained later. Figures 1.1 and 1.2 are a plan and an elevation of the reactor which show the graphite internal structure. This core and reflector have axial salt flow only, and the flow is from bottom to top, with no downward flow as was previously used in the top reflector. In the bottom and top reflectors, graphite is fixed by bolting to two identical mounting heads which form top and bottom plena to distribute and collect the salt. The bottom and top reflector pieces are laminated out of slabs of graphite about 2 in. thick by 12 in. wide, cemented together. These laminated blocks are then machined into sectors which are bolted to flat “lands” on the dished heads. The reflector sectors form 12 concentric rings on each head, with salt flow passages between the rings. Such a mounting maintains uniform passages between the graphite sectors and rings as the vessel heats up. In order to correctly orifice the flow through the reflector, the mounting heads have I-in.-diam holes under the circular slots between concentric rings. The number of holes increases as the radius increases. The velocity of salt flow is held essentially constant and is about 5 fps. The mounting heads are welded into the reactor vessel after the reflector graphite has been attached. The radial reflector is made up of laminated blocks of two different shapes, plus some odd-shaped filler blocks used to wedge the reflector to the vessel wall. The outer reflector is made up of laminated wedges about 10 ft long by 3 ft high. These wedges are machined to fit the 1. MSR Program Semiagnnu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 3 ff. vessel wall at the eight places where they are installed around the inside perimeter. Any out-of-roundness of the vessel is accommodated by fitting each wedge to within approximately ¢ in. of the wall. When all pieces have been put into place, retaining “T" irons are fixed to the vertical metal ribs, and wedging pieces of graphite are inserted behind the “T” irons. Thus the outer reflector is made to move out with the vessel wall as it expands. 7 “ - o o o c o el Q =] &) (o] 7 // / o o] o o o o] Q Q o \ N o o o > o o o o yz (o] [e] aQ o] o] &) o] o \ < N o ol o o [} s} o o o S ! — o oEl o} o o o s} [} ol_Jo The inner layer of reflector is made up of columns of graphite approximately 1 ft wide by 2 ft thick by 21 ft long. These columns are laminated from slabs similar to those used in the core, The columns are doweled into the bottom axial reflector, but the dowels are loose enough to permit the columns to float up from the bottom reflector while still being tied to it radially. The columns are installed with “%4-in. gaps from the outer reflector pieces and between columns. Also, the bottom ORNL-DWG 72-2826A (o] o o o] o X o =} © el o “ ™> Hydrogen was used as the carrier gas in these experiments primarily to inhibit corrosion of the apparatus, with HF oxidizer added to the extent that it formed from 1 to 20 mole % of the stripping gas. The results were analyzed according to the reaction HF(g)+ 1~ =F~+ HI(g) . (1) Briefly, it was found that the logarithm of the iodide concentration remaining on the melt varied linearly with the moles of HF passed. Equilibrium between phases was assumed, which allowed evaluation of the equilibrium constant for reaction (1) directly from the slope of the depletion curve. However, it was found that the equilibrium constant determined in this fashion evidently varied with the partial pressure of HF. This indicated that some incorrect assumption was made in the analysis, or some unaccounted for phenomenon occurred in the experiment. Hence it was not possible at that point to come to a knowledgeable evaluation of the potential for stripping iodine from MSBR fuel, and the question was temporarily dropped. A new analysis of the stripping experiments, given in a forthcoming report,® has evidently resolved this difficulty. It was shown that if one postulates that diffusion of 17 in the melt to the gas/liquid interface is the rate limiting step in the desorption process, there results a mathematical model which is in accord with the experimental data in the following respects: (1) a linear decrease of In [I7] with time or moles of HF passed is corroborated; (2) the apparent equilibrium constant, based on the assumption of equilibrium between the phases, is shown to indeed vary with HF partial pressure exactly as was observed; (3) the trend of the observed dependence of apparent equilibrium constant with both melt composition and temperature is qualitatively explained. In addition, it was shown that the rate of iodine stripping in the laboratory experi- ments was consistent with calculated values based on available bubble surface area and estimated mass trans- fer coefficients. It is therefore concluded that iodine stripping from LiF-BeF, mixtures is now understood. Although further experiments under conditions which come closer to a realistic situation are thought to be highly desirable, there is at this time some possibility for 2. MSR Program Semiagnnu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 28 ff. 3. C. E. Bamberger and C. F. Baes, Jr., “Removal of Iodide from L,B Melts by HF-H, Sparging,”” MSR Program Semiannu. Progr. Rep. Aug. 31, 1965, ORNL-3872, p. 127. 4. B. F. Freasier, C. F. Baes, Jr., and H. H. Stone, “Removal of lodine from LiF-BeF, Melts,” Reactor Chem. Div. Annu. Progr. Rep. Dec. 31, 1965, ORNL-3913, p. 38. 5. C. E. Bamberger and C. F. Baes, Jr., “Removal of lodide from L,B Melts,” Reactor Chem. Div. Annu. Progr. Rep. Dec. 31, 1966, ORNL4076, p. 32. 6. R. I. Kedl (ed.), Design Bases Report for MSR '?5Xe Removal Systems (to be issued). making an estimate of what it would take to contin- ually remove iodine from the MSBR fuel. Such an evaluation will be included in ref. 6. 1.3 MSBR INDUSTRIAL DESIGN STUDY M. I. Lundin The Ebasco MSBR reference concept was completed during this period. The reactor consists of a 2-in.-thick cylindrical vessel (nominally 22 ft OD X 20 ft high) supported from the bottom. Salt enters through four inlet nozzles at the bottom at 10S0°F and exits through four outlet nozzles at the top. The vessel contains 415 tons of graphite which defines salt flow channels in the following regions: core, axial and radial blankets, inlet and outlet salt plena, and axial and radial neutron reflectors. Each region has a specific salt fraction chosen to produce the desired nuclear characteristics of that region. Based on an evaluation of the ORNL reference concept, it was decided to retain the physics characteristic of that concept for task I. This was done by preserving the salt composition, the region salt fractions, and region dimensions specified in the ORNL reference concept (case CC-120). The Ebasco concept does, however, have two minor variations in the graphite region dimensions: 1. The salt annulus between the radial blanket and the reflector was eliminated. This annulus, whose function was to provide clearance between permanent and replaceable graphite for unit core replacement, is no longer required. In the Ebasco concept, individual graphite assemblies will be replaced on a four-year schedule. Only those assemblies which cannot survive another four-year exposure interval are replaced. 2. The inlet and outlet plena have been extended into the axial blanket regions for improved flow distri- bution. Neither of these changes will make an appreciable effect on the nuclear performance of the reactor. The core and blanket moderator bundles consist of ribbed graphite plates arranged into hexagonal assem- blies 15.6 in. across flats. Fuel salt flows from the reactor into four parallel circuits, each with a salt-circulation pump in the hot leg and an intermediate heat exchanger (IHX) where the heat is transferred to the secondary salt. The IHX is a vertical sine-wave-bent tube design with a nonremovable tube bundle. Fuel salt enters the top plenum, flows downward through about 7000 tubes (%4 in. OD), and exits at the bottom. The coolant salt enters at the bottom, flows up in a mixed counter- current flow, and exits at the top. The secondary system also has four parallel circuits, each containing one IHX, steam generator, steam reheater, and circulation pump. The pump is in the cold leg to pressurize the IHX sufficiently to force any leakage to be directed into the primary system. The steam generator concept is a supercritical, once- through, helical-coil tube design. Supercritical fluid enters at the top, flows down through annular rows of unheated downcomer tubes, turns, and flows up through the heat transfer zone. The steam flows through 815 tube coils, countercurrent to the coolant salt. The steam reheater is identical to the steam generator except for its size. The steam system basically consists of a supercritical steam cycle using a tandem-compound turbine gener- ator with reheat and feedwater heaters. Except for the steam generators, reheaters, supercritical feedwater pumps, and preheater-reheater, the system utilizes conventional power plant technology and designs. The feedwater temperature is 700°F to prevent freezing of coolant salt in the steam generator. This high feedwater temperature causes the steam system to differ from a completely conventional supercritical steam system. The chemical process plant permits the reactor to operate as a breeder by removing **>3Pa and certain soluble parasitic neutron absorbers from the fuel salt. It also reconstitutes the salt and returns it to the primary system. The plant flowsheet was developed and sup- plied by ORNL. This conceptual design reflects CONOCO’s experience and judgment regarding need and location of pumps, valves, surge volumes, drain systems, safety, control system, and spatial layout. The chemical processing cell is heated to prevent salt freezing. It is of a modular design for replacement of equipment by remote techniques. The upper level contains process equipment; the lower level contains drains and storage tanks. The cooling system uses NaK and is independent of other cooling systems in the reactor building. The reactor off-gas system removes fission gases, particularly '3%Xe and tritium, from the fuel salt. A purge gas (helium) throughput of (nominally) 10 scfm, together with efficient bubble separation from a 10% salt sidestream, will keep the salt void fraction in the core to about 1% (about 0.6% volume-weighed loop average). Based on these conditions, it is speculated that a poison fraction of 0.5% (0.005 neutron absorbed in 135Xe per absorption in fissile isotopes) can be achieved with unsealed graphite. Reduction of the '?5Xe concentration in the purge gas is accomplished primarily by decay during holdup in the drain tank, and the gas is recycled directly to the bubble injector. It is anticipated that some gas cleanup via charcoal adsorption will be required. These charcoal beds consist of coils of charcoal-filled piping submerged in cylindrical water tanks. Xenon is removed from the helium by dynamic sorption. The decay heat (about 2 MW) is removed by forced circulation of water through coolers. The off-gas system will also remove krypton, tritium, and volatile hydrocarbons from the purge gas. The fuel tank—drain tank system is intended to provide a safe place to store the salt at any time under all conceivable circumstances. It also provides holdup and cooling for the purge gas during normal operation. This tank is located below the reactor cell to permit drainage by gravity. The salt (or purge gas) is cooled by 10 about 1000 bayonet tubes inside thimbles mounted into the tank head. The coolant is NaK, which circulates by natural convection through many re- dundant external cooling circuits. Fission gas decay heat (about 18 MW) is transferred to the main steam system when in operation. Otherwise, it is transferred to a closed-cycle, boiling-water, heat rejection system. The reactor cell, chemical plant, off-gas system, drain tank cell, graphite handling equipment, and emergency power generators are located in the reactor building, a rectangular class 1 structure. Seismic supports are provided for the reactor and intermediate heat ex- changers in a horizontal plane by a three-tier support structure of Inconel beams. Structural support is provided at the bottom of the reactor and heat exchangers. The reactor building provides containment against release of radioactivity. Summary of principal data for MSBR power station General Thermal capacity of reactor Gross electrical generation Net electrical output Net overall thermal efficiency Net plant heat rate Structures Reactor cell, diameter X height Reactor building Reactor Vessel ID Vessel height at center (approximate) Vessel wall thickness Vessel heat thickness Vessel design pressure (abs) Core height Number of core assemblies Radial thickness of reflector Volume fraction of salt in central core zone Volume fraction of salt in outer core zone Average overall core power density Peak power density in core Average thermal-neutron flux Peak thermal-neutron flux Maximum graphite damage flux (>50 keV) Damage flux at maximum damage region (approximate) Estimated useful life of graphite Total weight of graphite in reactor Maximum flow velocity of salt in core Total fuel salt in reactor vessel Total fuel salt volume in primary system Fissile fuel inventory in reactor primary system and fuel processing plant Thorium inventory Breeding ratio Engineering units 2328 MW(t) 1037 MW(e) 1000 MW(e) 43.4% 7856 Btu/kWhr 72 X 46 ft 290 X 160 X 200 ft high 22.2 ft 20 ft 2 in. 3in. 75 psi 13 ft 157 30 in. 0.13 0.37 22.2 kW/liter 70.4 XW/liter 2.6 X 1014 neutrons cm 2 sec™! 8.3 X 10'4 neutrons cm ™2 sec™} 3.5 X 1014 neutrons cm ™2 sec™? 3.3 X 10! neutrons cm 2 sec™! 4 years 669,000 1b 8.5 fps 1190 2292 4440 200,000 Iv 1.06 Primary heat exchangers (for each of four units) Thermal capacity, each Tube-side conditions (fuel salt) Tube OD Inlet-outlet conditions Mass flow rate Total heat transfer surface Shell-side conditions (coolant salt) Inlet-outlet temperatures Mass flow rate Overall heat transfer coefficient (approximate) Primary pumps (for each of four units) Pump capacity, nominal Rated head Speed Specific speed Impeller input power Design temperature Secondary pumps (for each of four units) Pump capacity, nominal Rated head Speed, principal Specific speed Impeller input power Design temperature Fuel salt drain tank (one unit) Qutside diameter Overall height Storage capacity Design pressure Number of coolant U-tubes Thimble OD Number of separate coolant circuits Coolant fluid Maximum heat load Maximum transient heat load Fuel salt storage tank (one unit) Storage capacity Heat removal capacity Coolant fluid Coolant salt storage tanks (four units) Total volume of coolant salt in systems Storage capacity of each tank Heat removal capacity, first tank in series Steam generators (for each of four units) Thermal capacity Tube-side conditions Inlet pressure Inlet-outlet temperatures Mass flow rate Total heat transfer surface Shell-side conditions {coolant salt) Inlet-outlet temperatures Mass flow rate Steam reheaters (for each of four units) Thermal capacity Tube-side conditions (steam at 580 psi) 11 Engineering units 583 MW(t) 3/8 in. 1300-1050°F 23.45 X 10° 1b/hr 13,000 fi2 850—1150°F 17.6 X 100 b/hr 850 Btu hr~! ft=2 (°F)~! 16,000 gpm 150 ft 890 rpm 2625 rpm (gpm)?-3/(f)0-73 2350 hp 1300°F 20,000 gpm 300 ft 1190 rpm 2330 rpm (gpm)2-5/(ft)0-75 3100 hp 1300°F 14 ft 21 ft 2500 ft3 55 psi 1000 3in. 8 NakK 18 MW(t) 53 MW(1) 2500 ft3 1 MW(t) Boiling water 8400 ft3 2100 ft3 400 kW 490 MW(t) 3950 psi 709-1006°F 2.6 X 10° Ib/hr 3929 ft? 1150—850°F 15.5 X 10° Ib/hr 86.4 MW(t) Inlet-outlet temperatures Mass flow rate Shell-side conditions (coolant salt) Inlet pressure Inlet-outlet temperatures Mass flow rate Turbine-generator plant (see “General” above Number of turbine-generator units Turbine throttle conditions Turbine throttle mass flow rate Reheat steam to IP turbine Condensing pressure (abs) Boiler feed pump work (steam-turbine-driven), each of two units Booster feed pump work (motor-driven), each of two units 1.4 MSBE DESIGN H. A.McLain D.W. Wilson Layout studies of the MSBE core reported previ- ously” have continued, but using a core constructed of slab-type graphite elements rather than those of the prismatic design. The advantage of the slab-type ele- ment is that it is easier to fabricate; particularly there is no central hole requiring pyrolytic graphite coating. Consideration was given as to how a slab element could best be fitted into the overall cylindrical core layout. The individual slabs are incorporated into a moderator element measuring 7.27 X 6.93 in., with each moderator element, except those at the edge of the core, containing five graphite siabs having cross- section dimensions of about 7.27 X 1.25 in. (slab width does not include the ribs). The slab ribs are arranged to form 0.17-in.-thick salt flow channels giving the desired 15% salt fraction in the core. Lifting and suspension support for each moderator element is supplied by its center slab. This support is transferred to the other slabs by use of two cross ribs attached to the center slab by graphite bolts. A floating head allows orificing of the channel flows and provides space for differential ex- pansion between the center slab and the other slabs of the element. An isometric view of the moderator 7. MSR Program Semiannu. Progr. Rep. Aug 31, 1971, ORNL4728, pp. 11-15. 12 Engineering units 650—1000°F 1.5 X 10° Ib/hr 228 psi 1150—850°F 2.73 X 10° 1b/nhr 1 3500 psia, 1000°F 7.18 X 10° Ib/hr 540 psia, 1000°F 1.5 in. Hg 19,700 hp 7777 hp element containing these features is shown in Fig. 1.6. Some of the moderator elements used at the edge of the core contain only three graphite slabs to approximate the cylindrical core geometry, as shown in Fig. 1.7. The basic moderator elements are modified at five points in the core to form circular salt channels about 2%, in. in diameter. Four of these are required for the control rods, and the fifth, which is at the center of the reactor, is for the insertion of sample specimens. The locations of these channels are shown in Fig. 1.7. The four control rods would be fabricated of Hastel- loy N in a cruciform shape, with each designed to drop into one of the 2'-in.-diam salt channels. A graphite extension would be placed on the end of the control rod to protect the core graphite from damage, and to provide orificing of the salt flow through the control channel. Scratching of the core graphite by the edges of the cruciform is prevented by graphite buttons or strips fastened to them. A typical layout of such a control rod is shown in Fig. 1.8, It was decided that the control-rod drive mechanisms should be of the positive coupled design, and three types were considered for preliminary study. They were the magnetic jack, the roller nut, and the rack and pinion. All three filled the preliminary requirements of simplicity of operation, operational experience, and speed of response. The three-coil magnetic jack finally was selected for use as a future reference because of its simplicity, reliability, standardization of design, and extensive use in the present water power reactors. ORNL-DWG 72-7579 Fig. 1.6. MSBE fuel element. 1.5 BUBBLE BEHAVIOR IN THE MSBR PRIMARY SALT SYSTEM H. A.McLain L. W. Gilley® T. C. Tucker® A digital computer program, BUBBLE, has been written describing in detail the behavior of the gas bubbles circulating with the salt through the MSBR 13 primary salt system. This program is now being incor- porated into a larger digital computer program describ- ing the detailed behavior of the noble gas in the MSBR. The intent of this effort is to either confirm the results of the simplified calculations of Kedl® describing the noble-gas behavior in the MSBR or to make improve- ments where required. In the BUBBLE program it is assumed that the number of bubbles per unit salt volume and the sum of the gas dissolved in the salt and present in the bubbles per unit salt volume are constant throughout the salt loop (with the exception of the small volume of salt between the gas separator and the bubble generator). It is assumed also that the bubble size distribution at any location within the salt loop is described by the distribution function given by Kress:'? Ol3 1/2 fidy=4 (7> d* exp (~od?), in which d = bubble diameter, 4Ny 23 a= 6D ’ Ny = number of bubbles per unit salt volume, $ = void fraction. The rate of transfer of the dissolved gas to the bubbles can be described by the relation dCdt = —ka(C — Hp), where C = concentration of gas dissolved in salt, k = mass transfer coefficient, a = surface area of bubbles per unit salt volume, H = Henry’s law constant, p = pressure, t = time. For the bubble size distribution function given above, the surface area of the bubbles per unit volume of salt is 8. Mathematics Division. 9. MSR Program Semiannu. Progr. Rep. Aug. 31, 1968, ORNL-4344, pp. 72-74. 10. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 57-59. ORNL-DWG 72-7580 14 426 in. Iy TS _%Mwwnononowonowowowh. PSS o R _.w.w%&&.»mv e A A YA AY <= R R S R IR %K S 0o 2 holi|le%e A <] POODBD__”Q““A . ” TS P PSS AT X e SE S oo S5 o 2 55 e X < :“92 b 3292 e Pl VAN < X «. BEXXRK = = pedstelete’e eetelel jsl e NV ST b RSN TRCI AR oth teso el 008 51 N BRSSP RA L Griettee T T A =T =1 . ;fiv V=L = ~ R SR A ' e AR s R SR T L] X < 50 XX K2 9 | TN .ob»o»owowowo.o.o.%k_w"“ FORIR IR KSR RS g i p——— W - v‘. y —7" s | = 2 e 8 AN APNP NP DRSS ““O.QQMO %%0@ CARSAR e ,' S ¥ ¢ Fig. 1.7. MSBE core plan view. 15 ORNL-DWG 72-7581 ! 8in ) N A \\\\\\\\\\\\\\\\\\\\‘ \ N SECTION A-A 70in. A A Y ) SECTION B-B B Y 12in. 1 Fig. 1.8. MSBE control rod. Assuming perfect gas behavior and neglecting the surface tension effects, the volume fraction of the bubbles is qs:(N“C)R]-}/ps where N = total sum of gas per unit volume of salt, R = gas constant, T = temperature. The above relations are combined to give 2/3 %f—= ~4216N /3 k@}) (N — C)*3 (C - Hp). This relation is solved by numerical methods to give the dissolved gas concentrations throughout the primary salt loop. Once the dissolved gas concentrations are known, all of the other parameters relating to the bubbles can then be determined by using the appro- priate relations. In the parts of the primary salt loop defined as straight channels, fuel channels, piping, etc., the term to the left of the equal sign in the above relation can be rewritten as dc _ dc dt dx ’ where u = salt velocity, x = distance. The value of the mass transfer coefficient to the bubbles, k, is assumed to be that predicted by the relations reported by Kress'' correlating the results of his bubble mass transfer experiments. Similarly, in the portions of the primary salt loop defined as plenum regions, pump volute, heat exchanger plenums, etc., the term to the left of the equal sign in the above material balance can be rewritten as dc _ ¢ dt dv’ where V = volumetric salt flow rate, v = volume. In these regions the value of the mass transfer coeffi- cient to the bubbles, k, is assumed to be that predicted by the Calderbank and Moo-Young relation:* 2 Ply 1/4 2/3 k=013 [CL28E ()T p Sc where P/v = power dissipated per unit fluid volume, g, = Newton’s law conversion factor, Sc = Schmidt modulus, u = salt viscosity, p = salt density. No attempt has been made so far to account for any mass transfer between the bubbles and the salt within the gas separator and the bubble separator. The gas separator is treated simply as a unit that removes a specified fraction of the bubbles but does not affect the bubble size distribution. An equal amount of bubbles is then replaced in the salt by the bubble generator. Therefore, in the piping between the gas separator and the bubble generator, the total gas per unit salt volume is reduced by the amount EN - 0), where § = gas separator efficiency. In this same piping the area for mass transfer to the bubbles is reduced by the factor £. Some initial cases calculated by this program indicate that the transport rate of a gas to the bubbles (product of the mass transfer coefficient and the bubble surface area) is not a significant function of the volume-to- surface mean diameter for a given void fraction of bubbles in the reactor. This is not too surprising, since the bubble mass transfer coefficient correlation of Kress' 3 states that in flow regimes where the buoyancy forces are insignificant compared with the turbulent forces, the mass transfer coefficient is proportional to 11. MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 41—-43. 12. P. H. Calderbank and M. B. Moo-Young, “The Con- tinuous Phase Heat and Mass Transfer Properties of Disper- sions,” Chem. Eng. Sci. 16, 39-54 (1961). 13. C. E. Bamberger and C. F. Baes, Jr., “Removal of Iodine from L,B Melts by HF-H, Sparging,” MSR Program Semiannu. Progr. Rep. Aug. 31, 1965, ORNL-3872, p. 127. the volume-to-surface mean diameter. Since the Kress correlation for this flow regime applies to most parts of the MSBR primary salt system and since by definition the bubble surface area is inversely proportional to the volume-to-surface mean diameter, the gas transport rate is essentially constant for a given bubble void fraction. With the number of bubbles per unit salt volume held constant, increasing the void fraction increases the 17 mean diameter of the bubbles, the surface area, and the mass transfer coefficient. This results in large increases in the gas transport rate to the bubbles. Also, significant variations in the transport rates occur throughout the primary salt loop as a result of differences in tempera- ture that affect the mass transfer coefficients and differences in temperature and pressure that affect the mean bubble diameter. 2. Reactor Physics A. M. Perry 2.1 EXPERIMENTAL PHYSICS 2.1.1 HTLTR Lattice Experiments G.L.Ragan Q. L. Smith The experimental program of the Battelle Northwest Laboratories which measured the neutron physics parameters of a simulated MSBR lattice at temperatures to 1000°C has been reported.! We are ready to proceed with corresponding calculations, whose validity will be tested by comparison with the experimental results. During this report period we have updated our cross- section library and made and tested code modifications, some of which were described in our last progress report.? Basic cross-section library. In preparation for the HTLTR analysis, we have compiled a new cross-section library in our standard 123-energy-group format. This did not represent any major revision in our cross-section data, but is part of our continuing effort to keep MSR production cross sections up-to-date. With the notable exceptions of lithium, carbon, and fluorine, the data for the new library were obtained from the ENDF/B version II files. The cross sections for ®Li and "Li were prepared from ENDF/B version III data. For carbon, the thermal scattering data were prepared by a code using the incoherent crystalline scattering model. The capture cross section of carbon was normalized to 4.0 mb at 2200 m/sec (in contrast to the ENDF normalization of 3.4 mb) to account for impurities in nominal MSR-grade graphite. The capture cross section in our carbon has the correct 1/v dependence in the epithermal and fast range, in contrast 1. E. P. Lippincott, Measurement of Physics Parameters for an MSBR Lattice in the HTLTR, BNWL-1633 (1972). 2. MSR Program Semiannu. Progr. Rep. Aug 31, 1971, ORNL-4728, pp. 16—18. 18 to the ENDF representation, which is arbitrarily set to zero at 4 keV. The cross sections for fluorine are not available on the ENDF/B library, so for some years we have used data of our own evaluation. In preparing our new cross sections we revised the ! ° F(n,a) cross section downward by about 50% to take into account recent data. QOur updated 123-group library contains cross sections for all important MSR nuclides at the following temperatures: 293.6, 573, 900, and 1273°K. Carbon is available also at 1000°K. Resonance-groups treatment. Like many other reactor lattices, the MSBR lattice in the HTLTR was doubly heterogeneous, being composed of moderator material surrounding fuel rods that were in turn heterogeneous dispersions of grainy materials. During this report period we performed some calculations for resonance- energy neutrons to test the method we had developed for dealing with the double heterogeneity. As described below, we found that for the simple laminar geometry in the test problem the method is inaccurate and it is better simply to homogenize the fine heterogeneity in the treatment of resonance-energy neutrons. (The treat- ment of thermal neutrons is a different matter, dis- cussed later.) It is customary to define a self-shielding factor, at each energy, by the equation #(rod) M(lattice cell) - ¢(grain) _ ¢(grain) . Mlattice cell) ¢(rod) o_ o (1) The first flux ratio is the fine heterogeneity factor; the second, the gross heterogeneity factor. Existing treat- ments assume separability of the two factors, that is, that they can be approximated by evaluating the fine heterogeneity factor for an infinite medium of grainy rod material, then using cross sections disadvantaged by this factor in a smeared-fine-structure representation of the rod in evaluating the gross heterogeneity factor. Several different prescriptions®™® have been proposed for the fine heterogeneity factor. The gross factor is usually obtained by the Nordheim integral treatment (NIT), using codes such as GAM-II®* and GAROL.” We have developed a method that is similar in principle to the others in that we assume separability, but we obtain both factors by NIT calculations. The GAM-II part of the XSDRN® code was suitably revised to perform a double NIT treatment, making two successive passes (treating fine and coarse hetero- geneity) through the fine energy mesh of each reso- nance. Also, the GAM-II equations were modified to yield, at each energy, a moderator-region flux that agrees with that obtainable by the GAROL code, which solves explicitly for the neutron collision densities and thus avoids some of the approximations in GAM-II. We have used a set of simple laminar problems to test important features of our method by comparison of results from double NIT with those from a reliable reference calculation using the discrete ordinates code ANISN.? (Laminar geometry permits use of ANISN and also the exact calculations of escape probabilities and Dancoff factors.) In the reference case (case 2 in Table 2.1), absorber grains were simulated by ThO, slabs 0.0040 cm thick (comparable to the 0.0059-cm mean diameter of the particles in the HTLTR experiments). These were alternated with carbon slabs 0.022 c¢m thick to build up a laminated region 1.04 cm thick simulating an absorber rod. Finally, these laminated absorber regions were alternated with 6.6-cm-thick graphite moderator slabs to simulate the main lattice. Two other 19 cases were also considered: one with coarser structure in the absorber region (case 1 in Table 2.1) and one in which the absorber region was homogenized (case 3). The atom densities and volume fractions in the refer- ence case were preserved in the other two cases. The specific results that were compared are (1) the absorber-to-matrix flux ratio (I') in the laminated absorber region at the peak energy (21.7 eV) of the lowest major thorium resonance and (2) the thorium absorption cross section (0,) averaged over the lattice cell and the energy range (17.98 to 25.42 eV) covering the resonance. Both the ANISN and the double-NIT calculations used the 403 energy groups set up by 3. R. K. Lane, L. W. Nordheim, and J. B. Sampson, “Resonance Absorption in Materials with Grain Structure,” Nucl. Sci. Eng. 14, 300-396 (1962). 4. M. W. Dyos and G. C. Pomraning, ‘“Effective Thermal- Neutron Cross Sections for Materials with Grain Structure,” Nucl. Sci. Eng. 25,8-11 (1966). 5. P. Walti, “Evaluation of Grain Shielding Factors for Coated Fuel Particles,” Nucl. Sci. Eng. 45,321-30(1971). 6. G. D. Joanou and J. S. Dudek, GAM-II, a B3 Code for the Calculation of Fast Neutron Spectra and Associated Multigroup Constants, GA-4265 (September 1963). 7. C. A. Stevens and C. V. Smith, GAROL: A Computer Program for Evaluating Resonance Absorption Including Reso- nance Overlap, GA-6637 (August 1965). 8. N. M. Greene and C. W. Craven, Jr., XSDRN: A Discrete Ordinates Spectral Averaging Code, ORNL-TM-2500 (July 1969). 9. W. W. Engle, Jr., A User’s Manual for ANISN, a One- Dimensional Discrete Ordinates Transport Code with Aniso- tropic Scattering, K-1693 (March 1967). Table 2.1. Comparison of resuits? from double NIT and from ANISN for the lowest major thorium resonance Case 1: coarser absorber slabs (0.0200 ¢cm Description of problemb Case 2: nominal absorber slabs (0.0040 cm Case 3: infinitesimal absorber slabs (fine structure Code Conditions thick) thick) homogenized) r o, r a, r o, 1. ANISN Laminated absorber region 0.672 c 0.919 ¢ 1.000 ¢ alone 2. ANISN Lattice cell 0.827 12.87 0.973 12.98 1.000 12.99 3. Double Lattice cell, approximate 0.588 11.97 0.824 12.58 1.000 12.86 NIT escape probabilities 4. Double Lattice cell, exact escape 0.669 12.35 0.909 12.76 1.000 12.88 NIT probabilities S. Double Related spherical-grain 0.749 12.26 0.944 12.54 1.000 12.60 NIT cases (see text) ar is for 21.7 eV, the peak of the lowest thorium resonance. For both ANISN and NIT, Efa is averaged over the 403 energy groups (17.98 to 25.42 eV) set up by GAM-II for that resonance (Au = 0.00086). Other resonances are ignored. bThe problem is more fully described in the text. CThis calculation did not treat the full cell, so g, is not defined. GAM-1I for that energy range. The ANISN calculation was done at S-32 quadrature (required to get I' to an estimated accuracy of *0.003) and scattering order P-1 (insensitive). Four major conclusions can be drawn from the results summarized in the first four lines of Table 2.1 — line 5 will be discussed below. 1. The separability assumption leads to large errors in I'. Compare the true I' values of line 2 with those of line 1, obtained by considering the laminated absorber region alone. 2. For conditions existing in the laminated absorber, small errors in escape probability result in much larger errors in I'. The line 4 T" values agree well with line 1, but a significant error is contributed by the use of approximate escape probabilities (line 3) even though the approximation used was the excellent one, due to Nordheim, that is used in GAM-II and GAROL. A similar sensitivity to errors in the Dancoff factor was noted. 3. The grain effect (depression of g, relative to case 3) is overestimated severalfold by double NIT — taking the ANISN result in line 2 to be correct. 4. Hence, the best procedure found in this study is to homogenize the fine structure and treat the gross heterogeneity by an NIT calculation — which is the procedure that gave the NIT results of the last column. [t must be emphasized that the above conclusions are based on a laminar model. A comparable study, based on a more realistic model, would be of interest, but it appears to be very difficult. The extent to which these conclusions are valid for other geometries is an open question, but an indication of the magnitude of the effects in the HTLTR analysis, relative to those in the above study, is afforded by the following series of calculations. We set up three cases comparable to the slab cases of Table 2.1, but with HTLTR-type geometry: spherical absorber grains in cylindrical absorber rods arranged in a square-lattice array. The absorber grain diameters were chosen to have the same mean chord lengths as did the corresponding absorber slabs of Table 2.1; the respective grain diameters were 0.0600 c¢m, 0.0120 ¢m, and infinitesimal (homogenized fine structure). The absorber rods had the same mean chord length as did the 1.04-cm-thick laminated absorber stabs; hence, their diameter was 2.08 c¢m. All atom densities and volume fractions were kept the same as in the slab problems; the resulting square-lattice rod spacing was 5.00 cm. The ANISN code cannot treat this geometry, so results comparable to lines 1 and 2 of Table 2.1 cannot be given. Nor are results comparable to line 4 given, 20 since exact escape probabilities are not available. The results given in line 5 are calculated by the double-NIT code, using the same Nordheim-type approximation for escape probabilities as was used for line 3. Thus lines 3 and 5 may be compared to evaluate geometrical differences in resonance self-shielding effects. Let us define the grain effect as the fractional decrease in g, relative to that for the corresponding homogenized-fine-structure case. Then for case 1, the grain effect for line 3 (grains represented as slabs) is found to be 6.9%, while that for line 5 (spherical grains) is only 2.7%. For case 2, the grain effect is 2.2% for slabs and 0.5% for spheres. The grain effects for slabs are found to be severalfold too large, taking line 2 (ANISN) to be correct. Errors are due both to the separability assumption and to the escape probability approximation, and both errors have the same sign. The same two errors are present in the grain effects estimated from line 5, so that they, too, may be assumed to be overestimates. Hence, the true grain effect for 0.0120-cm-diam spheres (line 5, case 2) may be taken as less than 0.5%. The mean ThO, grain diameter in the HTLTR mixture was 0.0059 c¢cm, about half that of line 5 of case 2. Since the grain effect in line 5 appears to vary linearly with grain diameter, we estimate that the true grain effect for the HTLTR grains is less than 0.25%. If the double-NIT treatment over- estimates the grain effect by more than a factor of 2, as it did in the slab study, neglecting the adjustment is better than making it. Hence we plan to neglect this grain effect, simply homogenizing the fine structure and doing a normal NIT calculation of the resonance-groups cross sections. The resulting cross sections should be high by less (perhaps much less) than the 0.25% adjustment that a double-NIT treatment would give. Since less than 26% of the neutron absorptions occur in the resonance energy range, neglecting the grain effect should underestimate the multiplication factor k by less (perhaps much less) than 0.00065 (i.e., 0.26 X 0.0025). Thermal-groups treatment. Although grain effects were found to be negligible in the resonance energy range, that is not the case at thermal energies. The spatial variation of flux for the thermal groups is reasonably small within the graphite and ThOQ, powders that comprise most of the fuel mixture, but thermal fluxes are significantly depressed in the kernels of the coated particles that supply the fissile material. These particles have a kernel about 0.0300 ¢cm in diameter and consist mainly of ThO, and ?33UQ, in a 3:1 ratio. The kernels are surrounded by a 0.0100-cm-thick carbon coating. The average thermal flux in the kernels is about 2% below that in the rest of the fuel mixture, so that significant reactivity effects result. We are preparing to determine the thermal-range self-shielding for the fuel nuclides from an XSDRN problem in which the main lattice (cylindrical fuel rods in a square array) is represented as an equivalent spherical problem. A typical kernel, its carbon coating, and the appropriate quantity (smeared) of ThO, and graphite powders that should be associated with that kernel constitute the innermost three regions. The fourth region, consisting of homogenized fuel, extends to a diameter giving a mean chord length equal to that of the actual cylindrical fuel rod. Region 5 contains the main lattice moderator and has the correct volume, relative to the fuel. The XSDRN code has been modified to give multigroup cross sections, averaged over the innermost three regions. These cross sections take into account the spatial self-shielding effects of the grainy fuel, in the presence of the correct amounts of other materials, realistically disposed. They will be used whenever fuel is explicitly specified in further problems. In particular, they will be used in the fuel rod of a cylindrical XSDRN problem, in which the rod is correctly specified and the associated square-lattice cell is cylindricized in the usual manner. The resulting cell-averaged cross sections will be used in any further problems where the main lattice is to be represented without specifying its detailed spatial struc- ture. 2.2 PHYSICS ANALYSIS OF MSBR 2.2.1 Radiation Heating in MSR Pumps J. R. Engel Some preliminary calculations were performed to estimate the amount of radiation heating in the tank of an MSBR pump. Since this problem is common to all molten-salt systems, the reference-design MSBR'® was chosen to represent the entire class because the heating there will be greater than in lower-performance systems. In anticipation of the need to provide cooling for metal surfaces, a pump tank design was proposed with cooling shrouds that direct a flow of fuel salt over those surfaces. Since such provisions would affect the heat production in the structural members, estimates were made both with and without the shrouds in place. 10. R. C. Robertson (ed.), Conceptual Design Study of a Single-Fluid Molten-Salt Breeder Reactor, ORNL-4541, pp. 58-61 (June 1971). 21 The radiation that heats an MSR pump tank comes from several sources. One of the more conspicuous is the source associated with the noble gases and their daughters in the gas space above the salt pool. For molten-salt pumps of the type being considered, it is estimated that a salt flow of 50 gpm will enter the tank as leakage around the drive shaft and other seals, bringing with it noble gases at the same concentrations as those circulating in the loop. These gases were assumed to escape into the gas space and to eventually be swept out by the purge flow of gas that enters around the pump shaft. In the case where no cooling of the tank was assumed, the seal leakage was the only noble-gas source considered. Pump tank cooling could, no doubt, be accomplished in a variety of ways, but one convenient salt source is the return flow from the drain tank. This stream would amount to about 110 gpm per pump in an MSBR, and at least part of the salt would have previously been stripped of gaseous fission prod- ucts in the drain tank. To provide a basis for calcu- lations, this stream was assumed to flow into the pump bowl with complete release of the gaseous fission products to the gas space. The second major source of radiation in the pump tank is the salt pool in the surge volume, which was assumed to be 10 ft*. (The salt inside the volute is a more intense source, but it is also shielded from the pump tank by the volute itself and by the salt in the pool.) The intensity of this source was estimated from the rate of throughput and the ages of the various streams entering the pump for the two situations described above. A third, but much smaller, source of radiation heating is the gamma and neutron *‘shine” on the pump tank from other components — primarily the reactor vessel — in the primary loop. Previously reported! ! heating rates were used to estimate the contribution from this effect. The heating effectiveness of the sources inside the pump tank is strongly influenced by the presence or absence of the cooling-salt layer and the shroud required to direct the flow over the tank surfaces. With no cooling, we assumed that all the radiation from the gas space was directly incident on the surrounding surfaces, so that beta, as well as gamma, heat was deposited in the tank walls. In addition, we assumed that all daughters formed by noble-gas decay would deposit on the tank walls and contribute to the heat source. With salt cooling, the tank wall was presumed to be completely shielded from the beta radiation emitted 11. MSR Program Semiannu. Progr. Rep. Aug 31, 1969, ORNL-4449, pp. 63-67. by the noble gases and their daughters. However, an additional gamma source from the cooling salt itself had to be considered. (A l-cm-thick layer of salt was assumed.) The shielding provided by the shroud and cooling salt was neglected in estimating the heating due to gamma radiation. Table 2.2 provides a summary of the results that were obtained. Values are given both for the energy flux (beta and/or gamma as appropriate) to the tank wall and for heat generation in the tank. An exception to this is the contribution from reactor vessel shine, because the earlier results were reported only in terms of heat generation. In estimating the energy flux from the gas space, surface deposits, and surface cooling, we assumed that all the radiation emitted would be incident on some surface (e.g., the volute support cylinder would radiate beta and gamma energy toward the tank and vice versa) with a total area of 100 ft2, The salt pool was treated as a spherical source radiating toward surfaces 1.4 m away. Of the incident flux, all of the beta energy and all gamma energy with £ < 0.2 MeV were assumed to be absorbed; only 0.3 of the higher-energy gamma flux was allowed to deposit in the metal. Although these estimates of pump-tank heating are clearly rough approximations, they indicate that surface heat fluxes from the pump tank will probably be in the range of 3000 to 6000 Btu hr™! ft—2, depending on the internal configuration of the pump. Heat production in the pump impeller and scroll case may be somewhat higher, but these items are cooled by the rapidly circulating main salt stream. 2.2.2 MSBE Control-Rod Worths J. R. Engel Some preliminary survey calculations were made to estimate the reactivity worth of several potential control-rod materials and configurations in the core of the 150-MW(t) molten-salt breeder experiment.!? The materials considered included pure graphite and Hastel- loy N, for which no new compatibility questions would be raised, and the more conventional neutron poisons, boron and europium, which we presumed could be dispersed in graphite to give a usable material. Boron- impregnated graphite has been produced, so such rods, while short-lived in a high neutron flux, would pre- sumably be inexpensive and might be acceptable if used only for shutdown purposes. Europium-graphite rods 12. MSR Program Semiannu. Progr. Rep. Aug 31, 1971, ORNL-4728, p. 11. 22 Table 2.2. Estimated radiation heating in MSBR pump tank Uncooled tank Cooled tank Radiation source incident Heating Incident Heating energy rate energy rate (W/em?) (W/em?) (W/cm?) (W/cm?) Noble gases 0.65 0.49 0.59 0.18 Noble-gas daughters 0.73 0.39 1.19 0.36 Salt pool? 0.36 0.11 0.39 0.12 Cooling salt? 3.77 1.17 Reactor vessel shine ~0.1 ~0.1 Total 1.09 1.93 4Gamma source of 3.25 W/cm? in uncooled tank and 3.56 W/cm? in cooled tank. bGamma source of 3.77 W/cm3. would have a substantially longer life but would require development and would be more expensive. For each poison evaluation, criticality calculations were made in one-dimensional cylindrical geometry using the 123-group neutron-transport program XSDRN.? The reference reactor for each case contained a 2%-in.-diam cylinder of fuel salt on the core center line. The change in k¢¢ produced when part of this salt was displaced by the control rod was taken as the reactivity worth of the rod. A solid 2-in.-diam rod was used for each of the materials, and an additional calculation was made for a 2%- by %-in. cruciform Hastelloy N rod. Appropriate neutron transport proper- ties were computed for the central cell with each rod material in place. Table 2.3 lists the reactivity worths that were obtained for a single control rod at the core axis. The range of values given for the boron rod illustrates the kind of worth variation that could be attained with different loadings. It appears that, except for the pure graphite rod, acceptably large reactivity effects are attainable with any of the materials. Graphite rods are considered, in any case, only as low-worth regulating rods to provide operational reactivity control, but not shutdown or long-term shimming. Since it is quite likely that multiple control rods would be required in a reactor to provide for redun- dancy in safety action, some estimates were made of the total reactivity worth of four rods in a square array around the core axis. Figure 2.1 illustrates the decrease in poisoning that is realized as a rod is displaced radially from the core axis. The rapid decrease in worth at positions beyond ~30 cm suggests that greater radial displacement would be undesirable. (The boundary of the graphite-moderated region of the reactor lies at 57 cm.) In addition, the increasingly steep flux gradients Table 2.3. Worths of various control rods placed on vertical axis of MSBE core . Worth? Material % sk/k) Pure graphite, 2-in.-diam +0.06 Hastelloy N 2-in.-diam —2.5 2Y,-in. cruciform ~1.5 Natural boron in graphite, 2-in.-diam 10 at. % —-5.0 3at. % -3.9 1at. % -2.8 0.2at. % -1.3 Europium in graphite, 2-in.-diam 10 at. % —6.1 @Relative to channel filled with fuel salt. ORNL-DWG 72-7586 %1.0' i i | ‘[ } T s o T T N I A e HNAEEEEEEEEE T T T 16 20 24 28 32 RADIAL POSITION (cm)} Fig. 2.1. Relative worth of one control rod in MSBE as a function of radial position. that cause the decrease add considerably to the uncer- tainty of the worth values. Since mutual shadowing effects depend, to some extent, on the worths of the control rods, no attempt was made in Fig. 2.1 to include worth reductions due to shadowing. Thus the actual worth of clusters of rods near the core axis would be somewhat lower than indicated by this figure. For the Hastelloy N rod considered here, mutual shadowing effects appear to be less than 10% of the nominal worth for radial positions of 20 cm or greater. 2.2.3 Molten-Salt Converter Reactors Using Plutonium H. F. Bauman Molten-salt reactors that will breed if processed continuously can be operated as advanced converters if the processing is done only at six- to eight-year intervals.! 3 In the last semiannual report we described some calculations for a reactor having a core designed 23 for breeding but operated as a converter with 235U or plutonium feed.'* We also discussed the effects of changes in neutron energy spectrum during operation on the effective cross sections of plutonium. During this report period we modified our calculational procedures to minimize errors due to these effects and made an exploratory calculation of an MSR with initial salt composition appropriate for startup with plutonium. ROD code modifications. We have extended the capability of the ROD code!® for the calculation of the lifetime performance of molten-salt reactors with batch processing, to cover cases, such as startup with plu- tonium fuel, in which the lifetime-averaged reaction- rate coefficients are not suitable for part of the lifetime, in particular, the startup period. For such cases, we have programmed ROD to recalculate the reaction-rate coefficients at specified intervals during the lifetime. Since the core composition changes most rapidly at the beginning of life, we have arranged for shorter intervals to be used for the first cycle and for the beginning of each cycle. The use of from 20 to 40 intervals has given good results for plutonium startup cases. The interval time, as well as the fuel withdrawal option, has been included in the “specified variable” options. Another capability, important in plutonium startup, is to be able to change the salt composition during the reactor lifetime. We have programmed ROD so that, if desired, different salt compositions may be specified, by cycle, for the first four cycles. In addition, separate broad-group cross-section sets may be used, by cycle, in the first three cycles. Some additional new capabilities in ROD are the calculation of a lifetime-averaged fast (damage) flux and a lifetime fissile material balance. The power normaliza- tion has been improved by calculating and using average values for v (neutrons per fission) and the energy release per fission based on the fuel composition, rather than the fixed values used previously. The running time for single cases can now be shortened by about a third by the use of a restart tape, from which the atom densities, fission densities, and fluxes from a previous case can be recalled. For the study of initial conditions, such as the plutonium initial loading cases reported following, the ERC search and/or equilibrium calculations may now be bypassed. 13. A. M. Perry and H. F. Bauman, “Reactor Physics and Fuel-Cycle Analyses,” Nucl. Appl. Technol 8, 208 (1970). 14. MSR Program Semiannu. Progr. Rep. Aug 31, 1971, ORNL-4728, pp. 21--25. 15. H. F. Bauman et al,, ROD: A Nuclear and Fuel-Cycle Analysis Code for Circulating Fuel Reactors, ORNL-TM-3359 (September 1971). 24 Initial fissile plutonium loading. The initial fissile 0.01 salt fraction. The core diameter was adjusted in loading for an MSR varies with thorium concentration the calculations to give the same peak flux of neutrons in the fuel salt over the attainable range. We studied this with £ > 50 keV (damage flux) in each case, namely, effect by calculating the initial loadings of fissile that equivalent to a graphite life of 30 years in a plutonium in a fixed-moderator'® MSR with five 2250-MW(t) plant operating at 0.8 load factor. The different fuel salt compositions ranging from 0 to 14 fissile material was assumed to be typical first-cycle mole % ThF,. For this study we used a simple reactor plutonium from light-water reactors (23°Pu/2%%Puy/ model consisting of a spherical core with 0.12 salt ~ 241py/242Py:60/24/12/4 at. %). Because of the rec- volume fraction surrounded by a 78.4-cm reflector with ognized sensitivity of neutron resonance cross sections to plutonium concentration, we first estimated the 16. The term “fixed-moderator’” is used to indicate that the initial plutonium Conce'ntrations for each Ca_se and used core is designed so that the limiting fluence for fast-neutron ~ ~ODRN to prepare nine-group cross sections appro- damage to graphite will not be exceeded in a 30-year nominal priate for those concentrations. ROD was then used to reactor life (24 equivalent full-power years). calculate the critical loadings of plutonium. Further Table 2.4. Initial fissile plutonium loading in a 1000-MW(e) fixed-moderator molten-salt reactor ROD calculations in spherical geometry Salt fractions: Core 0.12 Reflector 0.01 Reflector thickness 78.4 cm Salt volume outside of core 18.4 m3 Case identification LOS5 LO6 L04 LO7 LO08 Carrier salt, mole % 67/33/00 65/32/03 64/30/06 67/23/10 69/17/14 LiF/BeF,/ThF, Core C/Th ratio o 698 365 226 167 Fissile loading, kg 119 339 596 1026 1425 Core radius,? cm 555 550 542 522 468 Core volume, m3 716 697 667 596 429 Core power density, W/cm?> Peak 7.83 7.81 7.81 7.83 7.83 Average 3.13 3.20 3.34 3.70 5.00 Ratio 2.51 2.44 2.34 2.12 1.57 Fast flux fraction at 0.15 0.33 0.45 0.59 0.69 center of core (£ > 1.86 eV) Fraction of fissions in 0.005 0.007 0.011 0.019 0.047 reflector Initial conversion ratio 0.08 0.58 0.69 0.75 0.75 Core C/Pu ratio (x10%), calculated 239py 30 10 5.7 3.0 1.7 240py 76 26 14 7.6 4.2 C/Pu ratio (X10%) used in preparing cross sections 239py 18 6.1 3.2 1.8 0.9 240py 50 17 8.6 4.7 2.5 Indicated change in resonance-group cross sections? 239py 1.04 1.04 1.12 240py 1.11 1.13 1.26 Core radius adjusted to give peak damage flux (pp = 4.5 X 10'3 neutrons cm™2 sec™, £ > 50 keV) equivalent to 30-year graphite life at 0.8 plant factor. bThe factor by which the cross section used would be expected to change if it were reweighted at the calculated C/Pu ratio. information on the calculations and key results are given in Table 2.4. As expected, the calculated plutonium loadings in- creased sharply with increasing thorium concentration (decreasing moderator ratio). Concurrently, as the neutron energy spectrum became harder (as evidenced by the fast-flux fraction at the center of the core) the power distribution flattened, and the core could be made smaller at constant peak damage flux. This resulted in more neutrons entering the reflector and higher fission rates in the 1% salt volume there. A large increase in reflector fissions is the first symptom of insufficient moderation in this type of reactor; the large increase in fissile loading seen between the 10 and 14 mole % ThF, cases suggests that further decreases in C/Th ratio would result in large increases in fissile loading. As shown by the C/Pu ratios in Table 2.4 the calculated critical concentrations of Pu were somewhat lower in every case than the concentrations used in preparing the nine-group cross sections. Because of time limitations, we did not perform an iteration of pre- paring new cross sections and recalculating critical loadings. From some of our earlier cross-section prepa- rations we were able to estimate the change in the effective resonance-group plutonium cross sections that could be expected from the observed difference in the C/Pu ratios. The expected change in the 2*°Pu reso- nance-group cross section for case LO8, for example, is about 25%, as shown in Table 2.4. The error in the critical loading from this cause is no doubt significant (on the order of a few percent) in this case. An iteration of cross-section preparation, while it would reduce the error, would not be expected to change the major conclusion of this study, which is that, in the startup of a fixed-moderator MSR with recycle plutonium fuel, the carbon-to-thorium ratio should be not less than 200, and preferably should be between 300 and 400. Lifetime performance with plutonium and uranium feed. In the last semiannual report we presented data showing that a fixed-moderator MSR designed as a breeder with continuous fuel processing can be oper- ated as a converter with batch processing using enriched uranium fuel. Qur studies showed further that, if LWR plutonium fuel is used under the same conditions, a considerable hardening of the neutron spectrum can be expected. In order to study this and other effects of plutonium fuel in MSRs, we have shifted our investiga- tion to a more general core design, with a uniform salt fraction over the core, rather than the zoned core of the 25 breeder, which was designed to flatten the fast flux distribution for a particular 233U equilibrium fuel composition. We selected the same reactor model as described in the preceding section on plutonium load- ing. However, in this study the core diameter was adjusted to obtain the lifetime-averaged (rather than startup) peak damage flux required for a 30-year core life at 0.8 plant factor. In the simple batch process assumed for this study, the uranium is removed from the fuel salt at the end of a cycle, by the fluoride volatility process. The re- maining salt, containing fission products and any plutonium present, is discarded. (At present, there is no economical process for recovering plutonium.) To avoid discarding large quantities of plutonium, we propose, even in plutonium feed cases, to switch to a uranium feed near the end of a cycle, permitting most of the plutonium to burn out. We have completed an exploratory plutonium feed case using the revised calculational methods described in the first section. We used the results of a number of earlier calculations to select conditions which we believe are near optimum. The reactor lifetime consists of four 6-efpy (effective full-power year) cycles with a switch to enriched uranium feed for the last two years of each cycle. Based on the critical loading study, we selected a salt composition containing 6 mole % thorium for the first cycle and 10 mole % for the remaining three. The reaction rate coefficients were recalculated at intervals ranging from 1.5 to 6 months in the first cycle and from 3 to 12 months in the remaining cycles. The broad-group cross-section sets used were changed after the first cycle to account for the change in carbon-to-thorium ratio and average plutonium concentrations at this point. The fuel nuclide inventories, the feed rate, and the conversion ratio are plotted over the reactor lifetime in Fig. 2.2. The conversion ratio increases as 2>3U builds in and averages just under 1.0 for the last two cycles. The change in critical loading is indicated by a separate bar on the feed-rate graph at the start of each cycle. The first bar represents the initial plutonium critical loading; the next the additional plutonium required for criticality as the thorium concentration is raised to 10%. The final two bars represent the excess mixed uranium (mainly 233U) over that required for criti- cality at the start of the final two cycles. This uranium is fed back to the system first, as required, before normal feed is resumed. Only a few kilograms of plutonium feed were required for the last two cycles, CONVERSION RATIO INVENTORY (kg) FISSILE FEED RATE (kq/mo) 0.8 .6 2800 2400 2000 1600 1200 800 400 120 80 40 -40 26 ORNL-DWG 72-7587 OPERATING TIME (equivalent full-power years) Fig. 2.2, Conversion ratio, fuel nuclide inventories, and fissile feed rate for plutonium feed case A37.20. SN T B N | T — — \ CONVERSION RATIO } B I A R I i T T T ] I T T I T I f T I I "’80 000 kg Th ~13O ;000 kg Th ~130,000 kg Th ~130,000 kg Th 6mole %o ThF4 (10 mole % ThF4)__.{ I_—._(10 mole % ThF, —-—1 (10 mole % ThF,) I ! | - T — fiF T i e | e T | o —— T —= 2 233 233, s 233P0+233U/V 233 I | 2 / / / k= — INVENTORIES 233 233 .7, /-] Pa + %3 234, ‘fl 234 239p // 235, 235, 235, | -— _‘"-—_.J ==l 1 - M— Pu 235y Pu 238y BRED 233y 235 |- _BRED 233y 233y N N | H_I REACTOR FEED RATE T L\_\fl ] 0 2 4 6 6 8 e} 12 12 14 16 18 18 20 22 24 Table 2.5. Lifetime-averaged performance of typical MSCR designs, comparing plutonium and enriched uranium feed Lifetime: four 6-efpy cycles Thorium concentration, mole %: First cycle 6 Other cycles 10 Case identification A37.20 A41.1 Feed Primary Pu(1)4 23s5yb Secondary® 235y None Core diameter, cm 1060 1020 Initial fissile loading, kg 574 1453 Lifetime fissile material balance, kg Purchases Plutonium 3272 0 235y 1165 4569 Discard: plutonium 71 20 Recovery at end of life 233y 2145 1978 235y 318 386 Net fissile requirement 1902 2185 Conversion ratio, lifetime averaged? 0.927 0.907 Fuel costs,® mills/kWhr Inventory Fissile 0.475 0.518 Salt 0.074 0.068 Salt replacement 0.145 0.133 Fissile burnup 0.067 0.113 Total 0.761 0.831 2Plutonium typical of first LWR cycle. Atom percent 239/240/241/242: 60/24/12/4. 93¢ enriched. ¢Switched to secondary feed at four years, each cycle. dNuclear conversion ratio, not considering plutonium discard or fissile processing loss. ¢Excluding processing costs. Obtained from present-worth calculation of fissile, fertile, and carrier salt purchases and fissile sales over life of reactor, with discount rate = 0.07 year™!, compounded quarterly, and inventory charge rate = 0.132 year !, Values of 11.9 $/g 235U, 13.8 $/g 233U, and 9.9 $/¢ fissile plutonium were assumed. between the return of withdrawn uranium and the start of enriched 335U feed at the fifth year. Most of the fissile plutonium is burned out by the end of a cycle,. even in the first cycle, while a considerable amount of 242Py remains to be discarded. Some of the results of the plutonium feed case are compared with an otherwise identical enriched-uranium feed case in Table 2.5. The conditions, particularly the low first-cycle thorium concentration, are not near optimum for uranium feed. A uranium feed case given in the last semiannual report, for example, for a fuel containing 14 mole % thorium, had an average con- version ratio of about 0.95 and a fuel cost of 0.76 mill/kWhr. Nevertheless, it is interesting to compare two cases identical except for the feed material. The initial critical loading for plutonium is less than half that for uranium, because of the higher effective cross sections for plutonium in a well-thermalized spectrum. This gives a significant cost advantage for plutonium startup, holding down the fissile inventory charges at the beginning of the lifetime, when they are most important to the levelized fuel cost. Perhaps the most striking observation on these two cases is how little they differ in performance. The conversion ratios, net fissile requirements, and fuel costs are all fairly close. We speculate that, when two well-optimized cases are available for comparison, we will find even smaller differences in the performance between these two feeds. 3. Systems and Components Development Dunlap Scott During this period we narrowed the scope of the systems and components development program to strengthen the efforts in support of the experimental programs for the two salt loops now under design and construction. The coolant-salt technology facility (CSTF) will replace the Inconel PKP loop used in tests previously reported,' and the gas system test facility (GSTF) will provide a means for testing the noble-gas removal system with an MSBR type of fuel salt. The industrial study of conceptual designs of steam gen- erators for use with MSR’s is under way and will continue; however, the plans for building facilities for testing steam generator models have been delayed until a more definite program for building another molten- salt reactor is established. The study of valves for use with molten salts has also been suspended indefinitely. 3.1 GASEOUS FISSION PRODUCT REMOVAL 3.1.1 Bubble Separator and Bubble Generator C. H. Gabbard The development of the bubble separator and the bubble generator continued in the water test loop. The 1. MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 3Tl. ® SWIRL VANE GAS REMOVAL LINE l SWIRL VANES final design of the bubble separator for testing in the GSTF has a tapered casing with a 44-in. separation length between the swirl and recovery vane hubs and gas removal from both hubs. This configuration, which is shown operating in Fig. 3.1, has a gas void about % to Y, in. in diameter and has maintained a stable vortex in all normal ranges of liquid and gas flow. Figure 3.2 shows the gas removal efficiencies for operation with 31% CaCl, solution with fine bubbles that have passed through the circulating pump and with coarser bubbles injected at the pump discharge. . A second dilution experiment starting with 31% CaCl, solution was conducted to determine if the CaCl, concentration where bubble coalescence began was dependent on the type or quantity of gas. The test was conducted with helium, nitrogen, and argon at gas flow rates of 0.455, 2.8, and 5.15 scfm. The results of this experiment were consistent with the results of the previous dilution experiment,? and bubble coalescence occurred at concentrations below about 2.7 wt % CaCl,. The coalescence concentration was independent of the type and quantity of gas. However, at concen- trations greater than the coalescence point, the separa- tion efficiency was higher with the less dense and less 2. Ibid., p. 26. . PHOTO 793134 P X : S RECOVERY VANE GAS REMOVAL LINE S CENTER LINE GAS VvOID FLOW —» RECOVERY VANES] Fig, 3.1. GSTF bubble separator operating at 500 gpm and 0.8% inlet void fraction. 28 ORNL-DWG 72-7582 100 A A A A Al a a A 8 o) < 80 s 3 2 2——?————&—2 oo (o] z 8 - - - > -4 8] 60 o ¢ w a ® G 60 Fo g ® = u A W o) 31 o2 O 500 gpm g 40 ® 400 gpm — LEU A A 300 gpm @ a 500 gpm WITH GAS INJECTED 2 AT PUMP DISCHARGE © 20 0] 0 { 2 3 49 5 6 NITROGEN GAS FLOW (scfm) Fig. 3.2. Separation efficiency of GSTF bubble separator on 31% CaCl, solution. soluble gases. This higher efficiency implies a larger bubble size with the less soluble gases. Pressure-drop data were taken on the bubble sepa- rator and the bubble generator, and from this data a tentative pressure distribution was prescribed for the GSTF main piping. The data for the bubble generator indicated a larger than expected increase in the gas supply pressure required as the gas flow was increased from zero to design flow. Efforts to reduce this pressure increase to a more acceptable level are continuing. 3.1.2 Bubble Formation and Coalescence Test C. H. Gabbard The fabrication of the bubble formation and coales- cence test rig was completed and testing is in progress. A schematic drawing of the test rig is shown in Fig. 3.3. The sample capsule moves vertically with a 1-in. stroke at frequencies ranging up to about 1600 cpm. The maximum accelerations would be about 1170 ft/sec? or 36 g’s. The test procedure is to shake the capsule at the desired frequency for several seconds to ensure an equilibrium void fraction and bubble size distribution. Photographs were then taken during the agitation, when the capsule stopped, and at 5-sec intervals until the fluid became relatively clear. Table 3.1 is a summary of the tests completed to date. All the capsules were sealed with about 1 atm of helium overpressure at the operating temperature. The capsule of LiF-BeF,-ThF, (72-16-12%) MSR fuel salt contained contaminants that made the capsule opaque during and after agitation. After sitting for a few hours 29 the salt again became transparent, with the contami- nation floating on the liquid surface, deposited on the capsule wall, and suspended as particles or flocs within the salt. The contamination prevented us from taking suitable photographs of this capsule. Figure 3.4 shows the bubbles produced in the various test fluids at 1600 cpm and the bubbles remaining in the fluid after various times after the agitation was stopped. The relatively large bubbles in the LiF-BeF, photographs are attached to the wall. The following conclusions can be drawn from the tests that have been run to date. 1. In demineralized water, small bubbles coalesce immediately, and the fluid clears of bubbles in a fraction of a second after capsule motion stops. 2. The capsules of 41% glycerin—water and 31% CaCl, contained a relatively high void fraction of very small bubbles. There was little or no coalescence, and a significant void fraction of small bubbles remained 20 sec after agitation was stopped. 3. Small bubbles were produced in the LiF-BeF, capsule as evidenced by the dark appearance during agitation. There was some coalescence after the agita- tion was stopped, but significantly less than with demineralized water. A very low void fraction of small-diameter bubbles remained after 20 sec at rest. There was no foaming at the liquid surface. The bubble separator would be expected to operate at a relatively high efficiency on a salt of this type. 3.1.3 Bubble Separator Analyses T.S. Kress Analyses were started near the end of this reporting period with the objective of idealizing the swirl-flow bubble separator to develop analytical expressions that would be helpful in understanding the performance of the separator. The initial idealization essentially consisted of treat- ing the flow as inviscid. Specifically, the simplifications included: 1. energy conservation, 2. constant axial velocity, 3. free vortex tangential velocity. Based on these assumptions, separate analyses were made to determine an equilibrium cavity size and a theoretical separation efficiency. The equilibrium cavity size, r,, was determined through an integration of the expression for con- 30 ORNL-DWG 72-2178 re——-- SHAKER DRIVE | _——DRIVE TUBE ASSEMBLY | _——DRIVE TUBE BEARING _——QUARTZ SALT CAPSULE | _—— HEATING ELEMENT __—CAPSULE CAGE T REMOVABLE INSULATOR PLUG T~ 1200°F FURNACE TT——QUARTZ FURNACE LINER __—— GRAPHITE CHILL BLOCK [2) = o [m) 2 = > I a GAS OUT L S COOLING COIL N FILL LINE \ 7 W [ b 3 T | I ) » COPPER O-RING \ /‘ / v Q 1 AV N\ JOURNAL BEARING % / / / N MIXING VESSEL y N N NICKEL LINER ! ;a./m in. ID X 15 in. DEEP) 1) N, \ ! N AN L,; THERMOCOUPLE SAMPLER— | 2/ i N | H §~§' Z\l\/ IMPELLER SALT : / IIIIGII IS IIIIIIS II Fig. 3.6. Cross section of molten-salt mixer, laboratory scale. 3.5.3 Molten-Salt Mixer, Laboratory Scale The design has been completed for a small mixer to be used to prepare molten-salt mixtures for various chemical and physical properties tests. The purpose of the device is to provide more thorough mixing of the melten salt than can be accomplished by present methods, which involve bubbling a gas through the salt. The mixer, shown in Fig. 3.6, is designed to operate from 900 to 1400°F, O to 2 atm absolute pressure, and 150 to 500 rpm. The mixing vessel is joined to the bearing housing with a solid copper wire O-ring, and the bearing housing is joined to the drive coupling diaphragm with a Teflon O-ring to form a single enclosed volume. The mixer shaft is confined entirely within this enclosed volume. The torque to the mixer shaft is transmitted through the drive coupling diaphragm with two 10-pole permanent magnets. The 37 external, or driving, magnet is mounted on the shaft of a Y,-hp dc motor, and the internal, or driven, magnet is mounted on the mixer shaft. The mixer shaft is supported on a full-complement angular-contact ball bearing with Haynes alloy No. 25 balls and races. This bearing must carry approximately 75 lb of axial thrust imposed by the magnets and a negligible radial load. The Graphitar—Hastelloy N lower journal bearing will also have a low radial load. The impeller is overhung below these two bearings. The mixing vessel, made of stainless steel, has a nickel liner for containing the salt. Penetrations into the vessel through the bearing housing are (1) gas inlet, (2) gas outlet, (3) thermocouple probe, (4) fill line, and (5) access line for either a sampler or a liquid-level measuring probe to be used when filling the nickel liner with salt. 4. Instrumentation and Controls S. J. Ditto 4.1 TRANSIENT AND CONTROL STUDIES OF THE MSBR SYSTEM USING A HYBRID COMPUTER O. W. Burke The hybrid computer model! of the MSBR system (including the proposed system controllers for control- ling the secondary salt flow rate, the primary salt temperature at the reactor outlet, and the steam pressure at the throttle) has been developed, and it has been used to run a number of transients. The draft of the report covering the model development and tran- sient runs has been completed, and the final report is being published. Some of the more interesting tran- sients that were run and their results shall be discussed briefly. The severity of the transients that can be run on this simulation model is somewhat limited by the nature of the steam generator model (the calculational time step of the discrete time model is 0.5 sec). The transients were run in order to determine the system response times, the rates of change of tempera- tures, and whether the salt temperatures approached the freezing points. Steady-state runs were made for power levels ranging from 100% design power down to 30% in increments of 10%. Of most interest in these runs was whether or not the primary or secondary salt approached its respective 1. MSR Program Semignnu. Progr. Rep. Aug. 31, 1971, ORNL4728, p. 38. 38 freezing point at any of these power levels. The results showed that the minimum temperatures in both salt systems were well above the freezing points for all cases. A number of fast changes in load demand were run in order to observe the resulting system response. The rates of change of the system temperatures were of interest. The secondary salt temperature at the steam generator outlet changed at a rate of approximately 4.5°F per second for the case when the load demand was ramped from full load to 40% full load in 1%; sec. The limitations of the model precluded higher rates of change. Some cases involving changes in reactivity were run. As a rough approximation of inserting two safety rods (each worth —1.5% in 86K/K), —3% 8K/K was ramped in in 15 sec. For this case the fission power had reduced to 10% of full power in approximately 3 sec. As a rough approximation of a fuel addition accident, +0.2% 8K /K was ramped in in 1.5 sec. For this case, in the absence of a safety system, the power peaked at approximately 2300 MW(e) and was back down to normal in approximately 10 sec. The primary salt temperature at the reactor outlet peaked at approxi- mately 1390°F. Development of a model capable of simulating faster transients and of covering a wider range of operating conditions is being considered. Some accuracy would be sacrificed in the steam generator simulation, but it is hoped that multiple loop simulation will be possible. This model should be capable of operating in real time. 5. Heat and Mass Transfer and Physical Properties H. W. Hoffman 5.1 HEAT TRANSFER J. W. Cooke Previous studies of heat transfer to a proposed MSBR fuel salt (LiF-BeF,-ThF,-UF,; 67.5-20-12-0.5 mole %) have led to the suggestion that, in the Reynolds modulus range 2000 to 4000, the heat transfer coeffi- cient varies along the test section in a manner which may be related to a delay in transition to turbulent flow.! Data obtained over a salt inlet temperature range 1080 to 1390°F have confirmed that the delay is abetted by the stabilizing influence of heating for a fluid having a large negative temperature coefficient.? Additional experimental results have recently been obtained with the gas-pressurized heat transfer system at lower salt temperatures which further substantiate the effect of temperature. The data for salt inlet temperatures ranging from 964 to 994°F are presented in Table 5.1. Attempts to operate at inlet temperatures nearer the salt melting temperature (~905°F) were not successful due to freezing in colder sections of the system. In Fig. 5.1, the heat transfer function, Ng_,, is plotted as a J. J. Keyes, Jr. function of the Reynolds modulus Ny .. These data are up to a factor of 2 lower than the values predicted by the Hausen correlation,3 the greatest deviation ocur- ring at the lowest Reynolds modulus of 3048. A plot of the wall and bulk salt temperature distributions along the 2-ft heated length and 2-ft adiabatic length of the test section is shown in Fig. 5.2 for runs 5-A and 7-A. These distributions show clearly that the transition from laminar to turbulent flow has not been completed within the 4-ft length of the test section (L/D = 270). In particular, the slow decay of the wall temperature in the adiabatic length of the test section for run 7-A (no entrance length) suggests the absence of significant eddy diffusivity contribution to the heat transfer (i.e., laminar flow), resulting in a strong radial temperature gradient in the salt which 1. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL4676, pp. 64—67. 2. MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL4728, pp. 39-41. 3. H. W. Hoffman and S. I. Cohen, Fused Salt Heat Transfer — Part 111, ORNL-2433 (March 1964). Table 5.1. Experimental results of heat transfer studies empioying salt mixture LiF-BeF,-ThF4-UF4 (67.5-20-12-0.5 mole %) Heat-transfer Run 7. T AT q/A Heat Modulus? h . No. CF) P CH (105Bwhr! ft?) balance N N. & [Btu hr ! ft=2 (°F)~1] function,? Re Pr Nu S—T 1-A 9745 988.6 78.6 0.97 111 3743 220 26.9 1239 9.33 2A 9644 9779 84.6 0.86 0.97 3223 229 222 1021 7.56 3-A 9873 10028 107.6 0.87 093 3100 209 176 809 6.12 4-A 9940 10092 95.4 0.87 1.00 3264 204 19.9 915 7.02 5-A 9925 1025.4 278.9 1.8 1.09 3048 19.7 138 633 4.56 6-A 993.6 10243 195.6 1.9 110 3551 19.8 20.8 956 7.10 7-A 986.4 1019.3 214.5 1.8 1.04 3161 202 185 850 6.22 AThese are the Reynolds, Prandtl, and Nusselt moduli, respectively, calculated using the average of the local coefficients from the exit to within § in. of the entrance to the heated length. bar =N A 1/3 0.14 Ns—1=Nnu/Np'? (ulug)® 1. 40 ORNL-DWG 72-7584 . HEAT TRANSFER FUNCTION /VNu (/Vp,)'h (I_‘/#S)O,M | | 107 2 5 10 2 5 Nge » REYNOLDS MODULUS Fig. 5.1. Summary of heat-transfer measurements employing a proposed MSPR fuel salt LiF-BeF,-ThF4-UF, (67.5-20-12-0.5 mole %). The upper and lower portions of the curve are the empirical correlations of Sieder and Tate and the center section is that of Hausen, ORNL-DWG 72-7585 1400 ¢ i ‘ | : i | [ ] ‘ ‘ . + { RUN FLOW DIRECTION 1300 f—— . 20 e ! e 5A -— — —~ ':2‘. : A 7A —_— . AA_ A & ‘ ‘[““*‘ 4 '. | — BULK SALT TEMPERATURES w 1200 [— - { o — E | —t = A | a, i ! 2 | I W 100 4 - ‘ . ! waal = \ ‘ Laa - ! L 1000 * % ‘ 'tfi%'w..fl# ' ’ ! 1 le———— HEATED LENGTH———>€— ADIABATIC LENGTH | | ! | | : ! 900 i | 1 I | | | : \ 5 10 15 20 25 30 35 40 45 WALL THERMOCOUPLE NUMBER Fig. 5.2. Wall and bulk salt temperature distributions along the heated and adiabatic lengths of the test section for Reynolds modulus of 3161. equilibrates slowly in the exit length. The abrupt increase in the wall temperatures near the end of the heated test section has not yet been fully explained. The only significant change in operating conditions between the present series of runs and earlier runs is the lower salt temperature, with an accompanying increase in viscosity (u = 50 Ib ft™' hr™!) and in its first derivative [du/dt = 0.20 1b ft™' hr™' (°F)7!'}. The results (e.g., low heat transfer associated with laminar flow) demonstrate the stabilizing influence of heating a fluid with a large negative temperature coefficient of viscosity. During these runs, the wetting characteristics of the salt were measured using the wetting detection probe described previously.?® The salt was found to wet the Hastelloy N surface (contact angle = 20°). We plan to purify the salt so that the contact angle is greater than 100°. Under nonwetting conditions, a gas film might be trapped along the heat transfer surface and severely restrict the transfer of heat into the salt. 5.2 WETTING STUDIES J. W. Cooke A new technique involving measurement of bubble pressure. described previously,* is being used to study the wetting characteristics of molten salts. Two prelimi- nary experiments with samples from the same batch of the salt mixture (LiF-BeF,-ThF,-UF,; 67.5-20-12-0.5 mole %) showed the salt wetting a Hastelloy N surface (contact angle <70°) at 1500°F, and at lower tempera- tures after addition of 1 wt % of the oxidizing agent, anhydrous nickel fluoride. Such wetting behavior was attributed to a suspected large oxygen content (~400 ppm) in this bath of salt. The salt mixture was reprocessed and the oxygen content reduced to 85 ppm. The chemical composition of the salt is given in Table 5.2. Some 3800 observa- tions of the contact angle of the salt with respect to a Hastelloy N surface were made over a 43-hr period. Results are presented in Table 5.3. Initially, the salt did not wet the surface (6 = 195°); however, over a period of 18.5 hr at 1000°F the salt gradually became partially wetting and then wetting (6 = 30°). The salt was then heated to 1230°F and began to partially wet (6 = 100°) the tip surface. The salt remained partially wetting for 4. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, pp. 54—-57. Table 5.2. Chemical analysis of the salt mixture LiF-BeF ;-ThF4-UF4;67.5-20-12-0.5 mole % Element Concentration (wt %) Li 6.74 Be 2.52 Th 42.20 U 1.78 F 45.10 Ni 0.0039 Cr 0.0398 Fe 0.0082 S <0.001 0, 0.0085 41 Table §.3. Wetting of LiF-BeF,-ThF4-UF,4; 67.5-20-12-0.5 mole % on Hastelloy N Elapsed Salt 8, contact \ Number of time observations temperature angle Comments (hr) CF) (deg) 0 500 Degassing under high vacuum 17 Start 1000 145 Salt melted 19 240 1000 125 21 600 1000 145 23 960 1000 125 25 1320 1000 100 27 1680 1000 90 29 2040 1000 80 31 2400 1000 70 33 2700 1000 60 35 3120 1000 40 35.5 3200 1000 30 37 3380 1230 95 Salt temperature raised 38 3460 1230 100 40 3820 1230 9§ 43 3860 1240 90 about 6 hr at 1230°F. At this point the system was cooled down and transferred to a vacuum dry box. The initial nonwetting of the Hastelloy N surface by the salt was indication of a salt mixture of high purity and a clean, vacuum-tight system. The gradual wetting of the surface was due, we believe, to the effect of moisture in the helium gas, probably in forming an oxide film on the surface. Prior experience by other groups using titanium scrubbers to purify helium gas has shown that some moisture may be evolved from the titanium sponge during its initial operation. We plan to repeat the measurements with a new salt specimen after a complete chemical analysis is made of the helium gas supply. 5.3 MASS TRANSFER TO CIRCULATING BUBBLES T. S. Kress The mass transfer coefficients between bubbles having a limited size range and liquids flowing cocurrently in a 2-in.-diam pipeline were found experimentally to follow the correlation’ Sh/Sc1/? = 0.34 Re®-°4 (d,/D)*-° . (5.1) The Reynolds modulus exponent (0.94) in Eq. (5.1) was found to be larger than expected when compared, on an equivalent power dissipation basis, with mass transfer data for bubbles and liquids in agitated vessels.® The agitated-vessel data are correlated by use of a Reynolds modulus exponent of 0.69. Conse- quently, to assist in resolving the difference, an analysis was undertaken to provide a theoretical basis for the exponent. A considerable amount of theoretical and experi- mental mass transfer information exists for the case of bubbles moving steadily through a fluid with some distinct relative velocity. The mass transfer equations established and generally accepted are the Frossling type, which, for large Schmidt moduli, take the forms Shy, ~ Re, 1/2 Scl/2 (5.2) Shb ""I{ebll2 SCI/3 s for mobile and rigid interfaces respectively. A small bubble suspended in a turbulent field is subjected to random inertial forces created by the turbulent fluctuations. Under the influence of a given force, if sufficiently persistent, the bubble may achieve its terminal velocity and move steadily through the liquid before being redirected by another of the random forces. If an ‘“‘average’ value representing the bubble relative velocity in such a turbulent field could be determined, then a convenient formulation would be the use of this velocity as a measure of an average bubble Reynolds modulus, staying within the confines 5. MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 41. 6. T. S. Kress, Mass Transfer between Swmall Bubbles and Liquids in Cocurrent Turbulent Pipeline Flow, ORNL-TM-3718 (in press). of the well-established relative-flow Frossling-type equa- tions [Egs. (5.2)] to determine the mass transfer behavior. The movement of the bubbles through the liquid will be resisted primarily by viscous stresses. The drag force on a sphere moving steadily through a liquid at relative velocity, v, is often expressed in terms of a drag coefficient, Cy, by the equation ) CyApvy? 3 Cymu® Rey? ) 5.3 28, 8.0 (5:3) d in which the drag coefficient is itself a function of the bubble Reynolds modulus. In relative flows, however, the drag coefficient—Reynolds modulus correlation depends on the particular Reynolds modulus range. Frequently, two regimes of flow are identified, with the division occurring at Re, = 2. Common correlations for the drag coefficients in these two regimes are given below, For Rey, < 2, Cy=24/Rey and Fy = 3nu*Rey /g .p . (5.9) For 2 2 If the bubble motion were predominantly in the regime Re, > 2, the drag force would be given by Eq. (5.5). The balance, F; = F;, would then give Re;, ~ (d/D)8/4-2 Rel1/8.4 (5.10) 7. P. H. Calderbank and M. B. Moo-Young, The Continuous Phase Heat and Mass-Transfer Properties of Dispersions, Chem. Eng. Sci. 16,37 (1961). The relative-flow bubble Reynolds modulus in this regime still depends on the variables that establish the turbulence level, but the dependence is different from that of regime 1. When substituted into the Frossling equations for mobile and rigid interfaces, the results are Sh ~ Scl/2 Re0.66 (d/D)—0.2/4.2 , (5_11) Sh ~ Scl/3 Re0-66 (d/D)-0,2/4.2 , (5_12) respectively. For this regime the Reynolds modulus exponent is 0.66. The comparison of this exponent with that expected from agitated-vessel data {0.69) is interesting. However, the writer feels that the apparent difference observed between mass transfer in agitated vessels and flow in conduits is more likely due to a difference in the relative influence of gravitational forces in the two systems rather than to a difference in the controiling flow regime. Nomenclature A = bubble projected cross-sectional area C, = drag coefficient for a bubble moving through a liquid 43 d = bubble diameter d,¢ = Sauter mean diameter of a bubble dispersion D = conduit diameter ) = molecular diffusion coefficient F; = drag force on a bubble moving through a fluid F;=mean inertial force on a bubble due to turbulent fluctuations g, = dimensional proportionality constant relating force to the product of mass and acceleration k = mass-transfer film coefficient Re = pipe Reynolds modulus (VDp/u) Re;, = bubble Reynolds modulus (vydp/u) Sc = Schmidt modulus (u/p) Sh = pipe Sherwood modulus (kD/L) Shy, = bubble Sherwood modulus (kd/ ) V' =liquid axial velocity v, = mean relative velocity between a bubble and a fluid u = liquid viscosity p = liquid density Part 2. Chemistry W. R. Grimes The chemical development activities described below continue to address themselves to chemical problems posed in design and operation of molten-salt reactor systems. Experimental study of specimens and materials from MSRE has been completed except for a small number of verification or recheck analyses. The only effort devoted to the MSRE “postmortem” consists in pre- paring final reports of the complex fission product behavior in that system. This documentation will be completed in the very near future. Study of possible mechanisms of intergranular attack upon MSRE metallic components (conducted in close cooperation with the MSRP metallurgists, and reported in more detail in Part 3 of this report) engages an increasing fraction of the chemical program. Studies completed to date have indicated tellurium as the most likely offending material. Many exposures of pertinent metals to tellurium (and to other potentially harmful fission products and contaminants) are under way for various times and at varying contaminant concentra- tions. In addition, tools and techniques for more sophisticated study of exposure to tellurium and for detailed study of metal tellurides are under develop- ment. It is expected that all these efforts will be further augmented in the future. The behavior of hydrogen isotopes in molten salts and metals constitutes a substantial fraction of the chemi- cal development effort. Solubility of H, in 2LiF-BeF, has been established at 600°C; this study should be concluded within the next reporting period. Definitive data are now being obtained on permeation of metals by hydrogen isotopes at the low pressures of interest, and useful preliminary data on reduction of such permeation by coatings and films are becoming avail- 44 able. This study (which is valuable not only to MSRP but to the Controlled Thermonuclear program and to any system requiring management of tritium at elevated temperatures) will continue to have a high priority. Study of possible mechanisms capable of holdup of tritium (especially in fluoroborate systems) to allow time for its controlled recovery continues to show promise. A related effort — aimed at improved under- standing of fluoroborate chemistry and of corrosion by fluoroborates — is still under way at a modest funding level. Study of protactinium chemistry during this period has been confined to selective precipitation of oxides. Such processing (to remove a major fraction of protac- tinium from the reactor fuel without precipitation of any other species) continues to appear feasible at temperatures near 550°C. Small-scale studies of this process are now considered complete except for con- firmatory experiments and for a more accurate defini- tion of the potential of the Pa**-Pa** couple in molten fluorides. The principal emphasis of analytical chemical devel- opment programs has been placed on methods for use in semiautomated operational control of molten-salt breeder reactors, for example, the development of in-line analytical methods for the analysis of MSR fuels, for reprocessing streams, and for gas streams. These methods include electrochemical and spectrophoto- metric means for determination of the concentration of U and other ionic species in fuels and coolants and adaptation of small on-line computers to electro- analytical methods. Parallel efforts have been devoted to the development of analytical methods related to assay and control of the concentration of water, oxides, and tritium in fluoroborate coolants. 45 6. Fission Product Behavior 6.1 SOME FACTORS AFFECTING THE DEPOSITION INTENSITY OF NOBLE-METAL FISSION PRODUCTS E.L.Compere E.G.Bohlmann S.S. Kirslis The loss of noble-metal fission products from circu- lating fuel salt and their deposition on reactor surfaces have been recognized since the examination of the first specimens from the MSRE. However, a considerable uncertainty has remained concerning the factors affecting their deposition intensity. Certainly a mass transfer step is involved, but — because in the standard surveillance arrays, there were differing flow conditions for metal and graphite specimens — sticking factors, flow terms, and other factors could not readily be extricated from the data. Nor is this completely achieved in the discussion to follow. However, the final surveillance specimen array!’? (Fig. 6.1), in addition to containing four uranium capsules for neutron capture experiments, also contained sets of paired metal and graphite specimens, with differing axial positions, sur- face roughness, and adjacent flow velocities. Because flow conditions were the same or essentially so for metal-graphite pairs, the hydrodynamically controlled mass transport effects if simple should cancel in comparisons, and differences can be attributed to differences in what is commonly called sticking factor. The sample pairs are listed below in order of increasing turbulence. Flow. In the noncentral regions of the core, the flow to a fuel channel had to pass through the grid of lattice bars, and according to measurements reported on 1. C. H. Gabbard, Design and Construction of Core [rradi- ation-Specimen Array for MSRE Runs 19 and 20, ORNL- TM-2743 (Dec. 22, 1969). 2. S. S. Kirslis, F. F. Blankenship, and L. L. Fairchild, “Fission Product Deposition on the Fifth Set of Graphite and Hastelloy-N Samples from the MSRE Core,” pp. 6870 in MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622. models, the velocity in the channels was 0.7 fps with a Reynolds number of 1000. However, the flow varied with the square root of head loss, implying that nonlaminar entrance conditions extended over much of these channels. The lattice bars did not extend across the central region, which included several otherwise standard fuel channels and circular annuli with rod thimbles and surveillance specimens. Consequently, the available flow head and resultant velocities were greater in the central region than in the noncentral region. The flow through central fuel channels was indicated by model studies to be 3.7 gpm, equivalent to 2.66 {ps, or a Reynolds number of 3700; the associated head loss due to turbulent flow can thus be calculated as 0.45 ft. In the adjacent parallel channels for rod thimbles and surveillance specimens, the same driving force across the 2.6- to 2.0-in. annulus yields a velocity of 2.6 fps and a Reynolds number of 3460. These flows are clearly turbulent. Flow in the circular annulus around the surveillance specimen basket essentially controlled the pressure drops, driving the more restricted flows around and through various specimens within the basket. At the bottom of the basket cage was a hollow graphite cylinder (No. 7-3, Table 6.1) 1'% in. OD, % in. ID, containing a '%-in.-OD Hastelloy N closed cylinder (No. 7-1, Table 6.2). The velocity in the annulus was estimated as 0.27 fps, with an associated Reynolds number of DVp/u = 0.0104 X 0.27 X 141/0.00528 = 75; this flow was, therefore, clearly laminar. This value, about 10% of that originally indicated,! was obtained by considering flow through three resistances in series, respectively, 20 holes in parallel, %% in. diam by % in. long; then 6 holes in parallel, ', in.diam by '/, in. long; then an annulus % 4 in. wide by 5 in. long. A flow head loss of 0.057 ft, which should develop along the outer part of the basket, was assumed. The laminar annulus flow formula PHOTO 96504 Fig. 6.1. Final surveillance specimen array (Photo 96501, Fig. 1, ORNL-TM-2743). 46 Table 6.1. Relative deposition intensity of fission products on graphite surveillance specimens from final core specimen array Observed dpm/cmzl(MSRE inventory as isotope dpm/MSRE total metal and graphite area, cm?) Activity and inventory data are as of reactor shutdown 12/12/69 Numbers in parentheses are (MSRE inventory/MSRE total area), dpm/cm2 % . Roughness Centfri(t)n;ters Samole No 89g, 137 140, 1816, 148, 955, 95Nb 99Mo 103p., 1065, 125¢y 1321, 129m, 1314 ype (uin.) SR mpile INo. (1.37E11% (8.53E9%) (1.73E11) (1.83E1l) (8.05E10%) (1.35E11%) (1.14E11) (2.26E11) (4.48E10%) (4.47E9%) (5.63E8) (1.85E10) (1.06E11) Outside 5 -29 7-3-1-B outer 4.2 0.023 0.17 0.0039 0.0036 0.0021 0.21 0.083 0.069 0.011 0.0805 (transition flow) 25 27 7-3-1-M outer 1.6 0.022 0.13 0.0019 0.21 0.033 0.033 0.046 0.0059 125 -25 7-3-1-T outer 2.4 0.08 0.0031 0.0008 0.0018 0.18 0.035 0.035 0.029 0.0035 Outside wire 5 +8 12-1-B outer 1.9 0.17 0.0069 0.0013 0.0025 0.25 0.040 0.0001 0.012 0.0033 (turbulent flow) 25 +10 12-1-M outer 2.0 0.17 0.0090 0.0015 0.0023 0.15 0.039 0.039 0.025 0.0047 125 +12 12-1-T outer 1.8 0.16 0.0037 0.0005 0.0021 0.15 0.050 0.0050 Inside annulus ) -28 7-3-1-B inner 0.31 0.0058 0.016 0.0012 0.0006 0.0014 0.04 0.003 0.022 0.019 0.065 0.0033 (laminar flow) 5 -26 7-3-1-M inner 0.26 0.0016 0.009 0.0012 0.0006 0.0016 0.25 0.22 0.150 0.108 0.054 0.0026 125 -24 7-3-1-T inner 0.18 0.0015 0.009 0.0012 0.0011 0.0016 0.24 0.19 0.064 0.049 0.579 0.0010 Inside tube 5 +9 12-1-B inner 0.49 0.0029 0.046 0.0031 0.0009 0.0027 0.25 0.056 0.047 0.061 0.0031 (transition flow) 5 +11 12-1-M inner 0.34 0.0010 0.028 0.0018 0.0009 0.0011 0.20 0.035 0.029 0.057 0.0019 125 +13 12-1-T inner 0.39 0.0032 0.033 0.0010 0.0002 0.0018 0.09 0.084 0.065 0.050 0.0017 127 99TC Te (Inv = 2.9E9) Postmortem: MSRE 0.049 0.23 0.56 0.44 core bar segment %Inventories shown accrue from all operation beginning with original startup. To correct inventories to show the material produced during current period (runs 19 and 20) only, multiply by factors given below. To obtain similarly corrected deposition intensity ratios, divide the value in the table by the factor for the isotope. Factors are: 52-day 89gr = 0.90; 59-day My = 0.86; 40-day 103pu= 0.95; 65-day 957r= 0.84; 284-day 12%4Ce = 0.36; l-year 106pu = 0.32; 30-year 137¢s = 0.14. For isotopes with shorter half-lives, corrections are trivial. 47 Table 6.2. Relative deposition intensity of fission products on Hastelloy N surveillance specimens from final core specimen array Observed dpm/cm?/(MSRE inventory as isotope dpm/MSRE total metal and graphite area, cm?2) Activity and inventory data are as of reactor shutdown 12/12/69 i Numbers in parentheses are (MSRE inventory/MSRE total area), dpm/cm? . | Tyne Roughness Cenftri(r)nnelters Sample 895t 137¢s 140Ba 141 ce 144ce 957r ?SNb Mo 103Ru 106Ru 125gp 1327¢ 129mp. 1317 B yp (uin.) core center No. (1.37E11%) (8.53E9%) (1.73E11) (1.83E11) (8.05E10% (1.35E119) (1.14E11) (2.26E11) (4.48E10%) (4.47E9%) (5.63E8) (2.01E11) (1.06E11) ‘ Outside (transition flow) 5 +24 14-3-B 0.0016 0.0015 0.0012 0.0009 0.0004 0.0004 0.13 34 0.094 0.127 6.4 0.060 25 +25 14-3-M 0.0013 0.0056 0.0009 0.0008 0.0003 0.0003 0.12 14 0.059 0.078 1.9 0.051 125 +27 14-3-T 0.0010 0.0006 0.0007 0.0006 0.0003 0.0003 0.14 1.7 0.056 0.079 2.2 0.046 Outside wire (turbulent flow) 5 +16 13-2-B 0.0013 0.0009 0.0013 0.0009 0.0004 0.0003 0.26 0.46 0.10 0.09 1.1 0.22 25 +18 13-2-M 0.0020 0.0253 0.0018 0.0011 0.0005 0.0007 0.34 2.2 0.20 0.17 2.3 0.37 125 +20 13-2-T 0.0039 0.0022 0.0030 0.0012 0.0005 0.0004 0.49 0.32 0.10 0.08 3.0 0.60 Wire +18 13-3 wire 0.0019 0.0006 0.0015 0.0010 0.0001 0.0005 0.21 0.85 0.14 0.23 0.17 0.09 Wire +11 12-2 wire 0.0030 0.0011 0.0027 0.0016 0.0008 0.0008 0.88 1.7 0.25 0.19 0.79 0.09 Inside annulus (laminar flow) 5 —-28 7-1-B 0.0018 0.0006 0.0015 0.0011 0.0005 0.0006 0.21 1.2 0.09 0.06 0.87 0.13 125 -26 7-1-T 0.0020 0.0007 0.0017 0.0013 0.0006 0.0006 0.39 14 0.15 0.23 0.93 0.19 Inside tube [transition (?7) flow] 5 +16 13-1-B 0.0021 0.0005 0.0020 0.0014 0.0006 0.0007 0.37 3.7 0.34 0.32 1.2 0.18 5 +16 13-1-M 0.0021 0.0012 0.0018 0.0013 0.0006 0.41 4.1 0.19 0.19 1.1 0.18 125 +20 13-1-T 0.0019 0.0007 0.0015 0.0011 0.0005 0.0004 0.71 3.6 0.23 0.17 3.0 0.38 Stagnant (inside liquid region) +26 14-2-1L.2 0.0030 0.0020 0.0002 0.00003 0.00002 0.00001 0.0051 6.0 0.005 0.007 0.03 0.002 +24 14-2-L1 0.0010 0.0002 0.0002 0.00011 0.00006 0.00006 0.025 3.0 0.044 0.043 0.13 0.010 Stagnant (inside gas region) +29 14-2-G1 0.14 0.0027 0.0057 0.00001 0.000005 0.00003 0.0010 4.5 0.0012 0.0010 0.14 0.0008 +28 14-2-G2 0.47 . 0.0022 0.0077 0.00002 0.00001 0.00002 0.0012 53 0.0006 0.0006 0.03 0.0028 1 +27 14-2-G3 041 0.0014 0.0069 0.00004 0.00001 0.00086 0.0012 6.6 0.0009 0.0009 0.03 0.0005 127 991, Te (Inv = 2.9E9) Postmortem MSRE heat exchanger segment 0.20 0.55 0.13 14 1.0 MSRE rod thimble segment (core) 0.83 0.66 0.32 1.6 1.0 9Inventories shown accrue from all operation beginning with original startup. To correct inventories to show the material produced during current period (runs 19 and 20) only, multiply by factors given below. To obtain similarly corrected deposition intensity ratios, divide the value in table by the factor for isotope. Factors are: 52-day 82Sr = 0.90; 59-day °!Y = 0.86; 40-day '3 Ru = 0.95; 65-day ?5Zr = 0.84; 284-day !44Ce = 0.36; 1-year 1 9Ru = 0.32; 30-year 137Cs = 0.14. For isotopes with shorter half-lives, corrections are trivial. given in Perry’s Chemical Engineer’s Handbook, 1V, Table 5-11, was used. In the annulus between the outside of the graphite cylinder and the basket, the velocity was estimated to be about 1.5 fps; the associated Reynolds number is 2200, and the flow was either laminar or in the transition region. At the top of the cage was a Hastelloy N cylinder (No. 14, Table 6.2; No. 4 in Fig. 6.1) of similar external dimension which presumably experi- enced similar flow conditions on the outside. This specimen was closed at the top and had a double wall. Inside was a bar containing electron microscope screens. The liquid around the bar within the cylinder was stagnant, and gas was trapped in the upper part of the specimen. Below this and above the midplane of the specimen cage were located respective graphite (Table 6.1, No. 12) and double-walled Hastelloy (Table 6.2, No. 13) cylinders, with connecting '4-in.-diam bores. Flow through this tube is believed to have been transition or possibly turbulent flow, though doubtless less turbulent than around the specimen exterior. The exterior of the 1-in.-OD cylinders was wrapped with '/ 4-in. Hastelloy N wire on %, -in. pitch as a flow disturbance. Flow in the annulus between the specimen exterior and the basket was undoubtedly the most turbulent of any affecting the set of specimens. The data from the various specimens are presented in Table 6.1 for graphite and in Table 6.2 for Hastelloy N. All activity and inventory data correspond to reactor shutdown (Dec. 12, 1969). In order to provide a common basis of comparison for all observations of deposited activity — of whatever isotope on whatever surface — the observed activity of an isotope per unit area has been divided by the quotient obtained by dividing total MSRE inventory activity of the isotope by the total MSRE surface area, graphite and metal (3.1 X 10® cm?). The resulting relative deposition intensities should be the same for all isotopes depositing the same fraction of their inventory by whatever means on unit surface of whatever kind. Thus the deposition intensity of various isotopes on either graphite or metal can be compared freely. Complete and even deposition of an isotope on all surfaces would give a value of 1.0 everywhere. However, deposition intensities vary with mass trans- port and sticking factor, etc., in various regions, so that the average relative intensity (summed over all system areas) should fall between 0 and 1.0. We saw no effect of surface roughness, which ranged from 5 to 125 pin. rms on both metal or graphite, so this will not be further considered here. 48 Fission recoil. Because the specimens were adjacent to fissioning salt in the core, some fission products should recoil into the surface.®* We calculate that where the fission density equals the average for the core, the relative impingement intensity of recoiling fission frag- ments [(recoil atoms/cm?)/(reactor production/surface area)]| ranges from 0.0036 for light fragments to 0.0027 for heavy fragments. The ratio will be higher (around 0.005) where the fission density is highest. Salt-seeking nuclides. Relative deposition intensities for salt-seeking nuclides (°3Zr, ' *!Ce) are of the order of the calculated impingement intensities or less: for 95Zr, 0.001 to 0.0027 on graphite and 0.0003 to 0.0008 on metal; for ?%'Ce, 0.0010 to 0.0090 on graphite and 0.0006 to 0.0016 on metal. The 284-day 144Ce is consistent with this, on a current basis after adjustment for prior inventory as shown in the table footnotes. There also appears to be some dependence on axial location, with higher values nearer the center of the core. Thus all of the salt-seeking nuclides observed on surfaces could have arrived there by fission recoil; the fact that remaining deposition intensities on metal surfaces are consistently less than impingement densi- ties indicates that many atoms that impinge on the surface may sooner or later return to the salt. Nuclides with noble-gas precursors. The nuclides with noble-gas precursors (3?Sr, '37Cs, '*°Ba, and to a slight extent, '4'Ce) are, after formation, also salt seekers. They are found to be deposited on metal to about the same extent as isotopes of salt-seeking elements, doubtless by fission recoil. However, noble- gas precursors can diffuse into graphite before decay, providing an additional and major path into graphite. It may be seen that values for 8%Sr, '37Cs, and '*°Ba for the graphite samples are generally an order of magni- tude or more greater than for the salt-seeking elements. It appears evident that the deposition intensity on graphite of the isotopes with noble-gas precursors was higher on the outside than on the inside, both of specimen 7 and specimen 12. Flow was also more turbulent outside than inside, and atomic mass transfer coefficients should be higher. [Flows are not well enough known to accurately compare the outside of the lower specimen (No. 7-3) with the inside of the upper graphite tube specimen (No. 12-1).] Appreciably more 3.1-min ®°Kr and 3.9-min '37Xe should enter the graphite than 16-sec '*%Xe or 2-sec 3. W. C. Yee, A Study of the Effects of Fission Fragment Recoils on the Oxidation of Zirconium, ORNL-2742, Appendix C (April 1960). 141Xe, but the '37Cs values are considerably lower than for 82Sr. Only about 14% of the !*7Cs inventory was formed during runs 19 and 20. With this correction, however, !'?7Cs deposition intensities still are less than observed for strontium. It has been noted previously® that the major part of cesium formed in graphite will diffuse back into the salt much more strongly than the less volatile strontium; this presumably accounts for the lower ! 37Cs intensity. At first glance the “fast flow™ values for #9Sr on graphite appear somewhat high: similar deposition intensity on all flow channel graphite would account for the major part of the ®°Sr inventory, while salt analysis® for the period showed that the salt contained ~82% of the 39Sr. But the discrepancy is not unaccept- able, since most core fuel channels had lower velocity and less turbulent, possibly laminar, flow. On the inside of the closed tube the deposition of salt-seeking daughters of noble gases was much higher in the gas space than in the salt-filled region. This is consistent with collection of °Kr in the gas space and relative immobility of strontium deposits on surfaces not washed by salt. Noble metals: niobium and molybdenum. Turning to the noble-metal fission products we note that °5Nb deposited fairly strongly and fairly evenly on all surfaces. The data are not inconsistent with post- mortem examination of reactor components. Molyb- denum (°?Mo) deposited considerably more strongly on metal than on graphite (limited graphite data). Because the deposition intensity of molybdenum on metal is similar to that of 8°Sr on graphite, which is attributed to atomic krypton diffusion through the salt boundary layer, it may be that molybdenum could also have been transported in appreciable part by an atomic mechanism, and presumably had a high sticking factor on metal (~1 ?). Under similar flow conditions, the decomposition intensity of molybdenum on graphite is much less; hence the sticking factor on graphite is doubtless much below unity. Postmortem component examination found that the °°Tc daughter also was more intensely deposited on metal than on graphite. The widely varying ®*Mo values reported® for salt samples taken during this period, however, imply that a 4. E. L. Compere and S. S. Kirslis, “‘Cesium Isotope Migration in MSRE Graphite,” MSR Program Semiannu. Progr. Rep. Aug. 31,1971, ORNL4728, pp.51-54. 5. E. L. Compere and E. G. Bohimann, *“Fission Product Distribution in MSRE Pump Bowl Samples,” MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp. 111-18. 49 significant amount of ®?Mo occurred along with other noble-metal isotopes in pump bowl salt samples as particulates. Since molybdenum was relatively high in the present surveillance samples also, it may be that an appreciable part of deposition involved material from this pool. Because molybdenum deposited more strongly than its precursor niobium, an appreciable part of the molybdenum found must have deposited independently of niobium deposition, and niobium behavior may only roughly indicate molybdenum behavior at best. This may well be due to the relation of niobium behavior to the Redox potential of the salt, while molybdenum may not be affected in the same way. Ruthenium. The ruthenium isotopes !°3Ru and 106 Ru appear to be quite similar in behavior and did not exhibit any marked response to flow or flux variations. The ruthenium isotopes appear to deposit several-fold more intensely on metal than on graphite. The correction of inventories to material formed only during the current period will increase the '°®Ru intensity ratios about threefold, but will change the 103 Ru intensity ratio very little. On a current inventory basis the '°?Ru deposition will then be appreciably lower than that for '®®Ru. This indicates that an appreciable net time lag may occur before deposition,® and argues against a dominant direct atomic deposition mechanism for this element. Tellurium. The tellurium isotopes ' >?Te (on metals) and 12°™Te (on graphite) show an appreciably stronger (almost 40X) deposition intensity on metal than on graphite, indicative of real differences in sticking factor. Deposition intensities of tellurium were moderately higher in faster flow than in low-flow regions (2X or more), possibly indicative of response to mass transfer effects. Flux effects are not significant. There is an appreciable discrepancy in the literature concerning the fraction of '2°Sb decaying to '22™Te (as pointed out to us by A. Houtzeel and by J. R. Tallackson). In calculating inventories we used 36%, together with 129 chain yields of 2, 0.8, and 2% for 233y, 235U, and 23%Pu, respectively, from the ORIGEN code library of Bell.” Other values of the branching fraction have been given elsewhere as 24 and 16%. The relative deposition intensities of *22Te are inversely proportional to this branching fraction, and 6. E. L. Compere and E. G. Bohlmann, ‘“Noble Metal Fission Product Behavior,” MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 60—-66. 7. M. J. Bell, “Nuclear Transmutation Data,” Appendix of ORNL-TM-3053 (November 1970). use of lower branching fraction values would increase the indicated relative deposition inensity of 129" Te. However, the differences between metal and graphite relative deposition intensities remain quite sharp. Postmortem examination of MSRE components showed '?3Sb and '27Te deposition intensities which were consistent with this, except that the deposition intensity of tellurium on the core fuel channel surfaces was higher than we observe here on surveillance specimens. On balance, it appears that the sticking factor of tellurium on metal is relatively high. This might result from direct atomic deposition and/or deposition on particulate material which then deposits selectively but securely. Such strong intensity of tellurium in the deposits could be the result of direct deposition of tellurium, or similar prompt deposition of precursor antimony with retention of the tellurium daughter, or both. The data do not tell. Iodine. Iodine exhibits deposition intensities respec- tively averaging about an order of magnitude less than tellurium both on metal and on graphite; thus the iodine could have and probably did come from de- posited tellurium, with a substantial part returning to salt either by recoil or by dissolution after formation. Sticking factors. It appears evident that the sticking factor must be below unity for many of the noble-metal isotopes, either on metal or graphite, because the déposition intensity differs for different isotopes under the same flow conditions. The deposition intensity appears rather generally to be higher on metals than on graphite, and could approach unity. In terms of mechanism, low values of sticking factor could result if only part of the area was active, or if material was returned to the liquid, either as atoms via chemical equilibrium process, or by pickup of deposited particu- late material from the surface. If all of the relative deposition intensjties were known, the fraction of a given isotope to be found on given reactor surfaces could be obtained by integration over the area under consideration, and dividing by total reactor area. Tables 6.1 and 6.2 give the relative deposition intensities under some conditions along the core axis. However, in order to extend these to other reactor areas of interest in the absence of further observation or experiment, appropriate relationships to the conditions in those areas are needed. The usual method is mass transfer analysis, but to establish the driving force coupled with such analysis, an adequate statement of all significant paths and mechanisms must be established. The present data are sufficient for only part of this, and their extension will not be attempted here. 50 6.2 EFFECTS OF SELECTED FISSION PRODUCTS ON HASTELLOY N, NICKEL, AND TYPE 304L STAINLESS STEEL AT 650°C J. H. Shaffer = W.P. Teichert W. R. Grimes Samples of Hastelloy N from various sections of the fuel system revealed widespread, superficial grain boundary cracking of the alloy.® The absence of grain boundary cracks on surfaces exposed to the coolant salt mixture or to the cell atmosphere, together with supporting analytical data on surface layer materials, indicated that certain fission product elements may have contributed to the corrosion process. As part of a larger effort within the MSRP to investigate this phenomenon, a joint effort of the Reactor Chemistry Division and the Metals and Ceramics Division was launched to identify those fission products which are capable of producing grain boundary cracks in Hastel- loy N and to develop methods for evaluating remedial measures, both chemical and metallurgical, which might be prompted by this investigation. Fission products of general interest in this study are comprised of those elements whose fluorides are ther- modynamically unstable in the molten-salt fuel environ- ment. Of this group, those elements which have relatively low boiling points and which form low- melting alloys with metal components of the structural material are being investigated first. Preliminary experi- ments based on these criteria have been designed to expose metallurgical tensile specimens to various ele- mental solute vapors within quartz ampuls at elevated temperatures. Tests with tellurium, selenium, sulfur, iodine, cadmium, arsenic, and antimony on tensile specimens of Hastelloy N, nickel, and type 304L stainless steel have been conducted or are in progress. Experiments will follow in which metallographic tensile specimens will be contacted with molten salt during exposure to these elements. The first experiment consisted of five quartz ampuls, each containing four Hastelloy N tensile specimens that were secured to a Hastelloy N wire rack. Sufficient solute was added to the ampuls during their preparation to yield a concentration of 100 ppm by weight in a 5-mil layer on the metal surfaces after vapor deposition. (Samples of Hastelloy N from the MSRE had tellurium concentrations of this magnitude.) The effects of tellurium, selenium, sulfur, and iodine were examined separately in four of the test ampuls. Arsenic, anti- mony, and cadmium were combined in the remaining 8. MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 89. quartz ampul. The five ampuls, sealed by fusion after residual gases were evacuated, were heated at controlled rates to 650°C and maintained at that temperature for 1000 hr. Metallographic examinations were performed on one specimen from each group at the conclusion of this test period. The stressed specimen exposed to tellurium was visibly corroded, that exposed to iodine showed only marginal corrosion, and the specimens exposed to selenium, to sulfur, and to arsenic plus antimony plus cadmium were not detectably attacked. The second experiment provided an additional 1000-hr exposure at 650°C, with the remaining test specimens arranged to examine effects with and with- out additional solute. A new set of specimens was also exposed to tellurium in a duplicate of the first test. Results were consistent with those of the first test, but no significant effects from additional solute or the increased exposure period were evident. The third experiment, conducted in the same manner, examined the effects of tellurium, iodine, and mixtures of these two elements on tensile specimens of pure nickel and type 304L stainless steel, and of tellurium alone on Hastelloy N. Solute concentrations of 100 ppm by weight per element and a test period of 1000 hr at 650°C were again used. The results of metallurgical examinations at the conclusion of the experiment showed that only those specimens exposed to tellurium alone were attacked. Of these specimens, Hastelloy N was comparable to those of previous tests, nickel showed less attack than Hastelloy N, and the type 304L stainless steel showed no apparent corrosion. The most salient result of these experiments is that effects on Hastelloy N not unlike that found in the MSRE fuel system have been produced with relatively low concentrations of tellurium during relatively short time periods at a temperature comparable to that experienced by the MSRE. A detailed description of the metallurgical examinations and discussions of results appear in Part 3 of this report. Additional experiments are in progress to further examine the effects of these solute elements on Hastelloy N, nickel, and type 304L stainless steel. One set of experiments will examine the effects of tellurium at concentrations of 300, 5000, and 10,000 ppm by weight on tensile specimens of these three materials at 550, 650, and 700°C during a 1000-hr exposure period. A second set of experiments will examine the effects of tellurium, selenium, sulfur, iodine, cadmium, arsenic, and antimony at concentrations of 300 ppm by weight on tensile specimens of the three materials for timed exposures of 3000, 5000, and 10,000 hr at 650°C. 51 6.3 REACTION OF CoF; WITH TELLURIUM J.D.Redman C.F.Weaver The studies described in the preceding section and in Part 3 indicate that tellurium may play an important part in the surface cracking of Hastelloy N. Conse- quently, the reactions, stability, and volatility of tellurium fluorides and of the structural metal tellurides are being investigated. This work was initiated with a mass spectrometric study of the reaction of cobalt trifluoride with tellurium metal in a nickel cell, since this reaction might afford a convenient means of adding tellurium fluoride vapors to a molten-salt mixture to assess the thereby induced corrosion of metal speci- mens. The first experiment was made with CoF; and tellurium contained in a compartmented cell and unmixed except in the vapor phase. It was found that observable reaction started at 225°C, evolving TeF¢ at a pressure of 5 X 107 torr. The pressure of this species was 3 X 107 and 3 X 107% torr at 250 and 275°C respectively. At 300°C a mixture of TeF¢ and TeF, was evolved at pressures of 6 X 1072 and 5 X 1073 torr respectively. The cracking patterns for these and sub- sequent species are shown in Tables 6.3—6.5. No other Table 6.3. Cracking patterns for tellurium fluorides® Relative intensity for TeFg TeF4 TeFs 0.03 TeFs' 100 TeF,' 5 3 TeF5' 9 100 TeF, 5 18 TeF™ 6 10 Te™ 7 8 475-¢V electrons, Table 6.4. Cracking pattern for tellurium vapor® Ion Relative intensity 13 OTe+ 39 2567,," 100 386Te3+ 4 51 4Te4+ 2 640 et 0.1 475.¢V electrons. Table 6.5. Cracking pattern for CoF3* Ion Relative intensity CoF3' 20 CoF," 100 CoF* 23 Co* 43 475-¢V electrons. species was produced until the temperature reached 400°C. Tellurium vapor was observed at 400°C, cobalt trifluoride at 600°C, and nickel fluoride at 700°C. In addition, a deposit of grayish-white material formed on the lid of the effusion cell. This has not yet been identified but is likely to be a lower fluoride of tellurium.”> A pink residue of CoF, identified by x-ray diffraction’® remained in the cell. A second experiment duplicated the first except that the solid CoF5 and tellurium were mixed. This mixture, which did not react at ambient temperature, started to evolve volatile tellurium fluorides at slightly less than 52 200°C. The reaction was much more rapid than in the first experiment and continued even when the tempera- ture of the cell was lowered to 100°C. Only TeF was observed at 100°C; TeF, and F, were evolved at 150°C in addition to TeFg; tellurium metal was observed at 200°C. The occurrence of these species at lower temperatures than in the first experiment suggests that the sample of mixed solids may have been heated above the cell temperature by exothermic reactions. These results indicate that CoF; and tellurium will provide a convenient source of tellurium fluorides, but that the reactants should be mixed in the gas phase, since the reaction of mixtures of the solids, even in small amounts, will be difficult to control. 9. D. R. Vissers and M. J. Steindler, “Laboratory Investiga- tions in Support of Fluid-Bed Fluoride Volatility Processes. Part X. A Literature Survey on the Properties of Tellurium, Its Oxygen and Fluorine Compounds,” ANL-7142, p. 22 (February 1966). 10. The x-ray analysis was provided by R. M. Steele in the Metals and Ceramics Division of ORNL. 7. Behavior of Hydrogen and Its Isotopes 7.1 SOLUBILITY OF HYDROGEN IN MOLTEN SALT D. M. Richardson A. P. Malinauskas The ease with which tritium, either as T, or TH, can be removed (or lost) from the molten-salt loops depends to some extent on the solubility of the gas in the corresponding molten-salt solvents. As no data existed in this regard, and an adequate theoretical treatment, with which reliable estimates could be made, was lacking, an experimental program was developed. Measurements of the solubilities of helium and hydrogen in Li,BeF, at 600°C were completed during this period. The experiments, which were performed at saturation pressures between 1 and 2 atm, were conducted using the two-chamber apparatus described previously.' Although little difficulty was encountered with the helium measurements, the results initially obtained with hydrogen displayed a disturbing degree of scatter.' Under the assumption that the cause of the scatter was due to air inleakage over prolonged periods of time, we prefaced each hydrogen experiment with a lengthy hydrogen sparge at 700°C. This pretreatment resulted in reproducible measurements and yielded values signifi- cantly lower than those obtained previously. As a result, all of the earlier data were discarded. The experimental data obtained at 600°C are sum- marized graphically in Fig. 7.1, where the solubilities of the two gases in Li, BeF, are plotted as a function of saturation pressure. The solid lines represent the corre- sponding Henry’s law constants for the two solute- solvent systems at the temperature indicated; the constants derived from the data displayed in the figure are (8.40 + 0.16) X 10™® mole of gas per atm per cm® of salt for helium, and (4.34 + 0.20) X 107® mole of gas per atm per cm® of salt for hydrogen. The value quoted above for helium is about 40% lower than that reported by Watson et al.? Although this discrepancy is outside the limits of mutual uncer- tainty, and we are unable to account for the discord- ORNL-DWG 72-1767R 20— - E ™M § . } HEUM 2 . £ e " ____,_..—--—r""—'_'_A/ @ -4 : HYDRF)GEN 0 1.0 1.2 1.4 16 1.8 20 2.2 p (atm) Fig. 7.1. Pressure dependence of the solubilities of helium and hydrogen in a 66 mole % LiF—35 mole % BeF, eutectic at 600°C. The solid lines correspond to Henry’s law constants of 8.40 X 1078 mole of gas per atmosphere per cubic centimeter of salt for helium and 4.34 X 1078 mole of gas per atmosphere per cubic centimeter of salt for hydrogen. ance, we do not believe it to be serious enough to warrant special study. Solubility measurements with helium and hydrogen in Li, BeF, are being conducted at 700°C. 7.2 INITIAL TRITIUM CHEMISTRY IN THE CORE OF A MOLTEN-SALT REACTOR R. A. Strehlow In the analysis of tritium behavior it is necessary to consider what happens when a high-velocity triton is born in a fuel of oxidation potential fixed by the ratio U3/U*. Half of the tritium in the reference-design MSBR is produced by Lit+n—>T"'+*He+ 4.8 MeV , 1. J. E. Savolainen and A. P. Malinauskas, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 115-17. 2. G. M. Watson, R. B. Evans IIl, W. R. Grimes, and N. V. Smith, J. Chem Eng. Data 7,285 (1962). 53 54 which gives a triton with an initial energy of 2.7 MeV.? The triton will lose energy at a rate which may be approximated as one-third that of protons for collision with light elements. This gives an initial value of about 30 MeV g7! cm ™2, increasing to 200 MeV g™! cm™? at an energy of 0.05 MeV.* Integration of this stopping power yields a calculated path length of about 0.05 g/cm? or 0.025 cm in salt having a density of 2 g/cm?. This is about a tenth of the minimum dimension of fuel passages in an MSBR, so only a few percent of the initially formed tritium would be expected to reach the graphite before thermalization. Tritons from the "Li(n,an) reaction are born with generally less energy and would be thermalized in a shorter path length. Thus nearly all the tritium will appear in the salt initially as thermalized tritons. Some equations for pertinent equilibria involving the thermalized triton are: For tritium 2T° =T,(d), (D) T°+F°=TF, 0) T°+U% =T+ U*, (3) T® + Cgraphite = CTadsorbed > 4) T°® + H® = TH (if hydrogen is present), (5) T° + H, = TH + H® (if hydrogen is present) , (6) T° + FP* =T* + FP.("V* (7) where F.P. is fission product or corrosion product, T*+e =T° (8) (viewing the electron as a caged chemical species). For fluorine, additionally, F~ - F° + e (due to fission recoil) , 9) FP+U"=U+F", (10) F°+e =F, (11) 2F° =F, , (12) U4++ e‘:U3+. (13) The fluorine atom reactions have been considered by Jenks.5 The principal source of fluorine atoms which he considered is indicated as Eq. (9) — the formation of fluorine by the fission recoil atoms during their thermalization. He concluded that it is possible that in the bulk of the melt, U*" ions serve as scavengers for the radiolytic products F° and e~ [in accordance with Eq. (10) and {13)]. From Jenks’ calculation a possible steady-state concentration of F° is calculated to equal about 1.5 X 10'!/cm?, assuming a fission rate of 3 X 10'! sec™! cm ™3 (about 10 W/cm?) and a uranium(IV) concentration of 5 X 10'%/cm?. The production of F° associated with tritium production (about 10'° sec™! cm™?) would not contribute significantly to the con- centration following from the fission recoil process. Because of this low expected concentration of fluo- rine atoms and the dominance of U*, tritium atoms should be expected to enter into equilibrium with U** in accordance with reaction (3). This has been the assumption made previously in consideration of tritium core chemistry.® Detailed consideration of reaction rates will be needed to determine the extent to which these conclusions are valid. 7.3 PERMEATION OF HYDROGEN THROUGH METALS AT LOW PRESSURES R. A. Strehlow H. C. Savage Knowledge of the permeation behavior of hydrogen through iron and nickel-based alloys at low pressures is required for the prediction of tritium behavior in the molten-salt reactor. Tritium release to the environment from a molten-salt breeder would occur because the isotopes of hydrogen flow readily through hot metals. The tritium is ex- pected to follow the flow of heat through the reactor’s heat exchangers into the steam system and thence to the outside world. At high pressures the permeation rate is proportional to the square root of the pressure, because the hydrogen molecules dissociate into atoms when they enter the metal. A transition to a linear dependence has 3. D. J. Rose and M. Clark, Jr., Plasmas and Controlled Fusion p. 296, Wiley, New York, 1961. 4. Dwight E. Gray (ed.), American Institute of Physics Hand- book, pp. 8—20, 2d ed., McGraw-Hill, New York, 1963. S. G. H. Jenks, Proposal for Research to Determine Rates of CF4 and F, Production During Irradiation of Molten Fluoride Salts in Contact with Graphite, ORNL-CF-62-10-69 (October 1962). 6. R. B. Briggs and R. B. Korsmeyer, Distribution of Tritium ina 1000 MW(e) MSBR, ORNL-CF-70-3-3 (Mar. 18, 1970). generally been reported to occur as the pressure is lowered. The point at which this transition occurs, if at all, is important, since a linear dependence might predict an acceptable tritium flow at some particular pressure, while the square-root relationship would give an intolerably large discharge at the same pressure. Many workers in the field have stressed the fact that the hydrogen flow rate through materials such as Hastelloy N,” iron,® and palladium® varies with the square root of pressure at higher pressures, but begins to deviate from the square-root dependence below some low pressure. The departure has been reported in the cited work to be significant at pressures as high as 200 torr,” although departures beginning at values near 10 torr are more commonly reported. Such departures have been attributed to a reduction in apparent area due to a restricted surface coverage by sorbed hydrogen atoms or by the formation of a film of lower permeability. Since extrapolation of permeation data to the lower pressures required in a molten-salt reactor is essential in calculations of tritium management schemes, an experi- mental program was undertaken to measure the pres- sure dependence of hydrogen permeation through candidate structural materials from pressures of 30 torr down to about 107> torr. The apparatus is schematically shown in Fig. 7.2. Deuterium, rather than hydrogen, was chosen for this study to avoid problems posed by the large (107° torr) hydrogen background in the mass spectrometer. A mixture of argon—4% deuterium is maintained at a total pressure from 1 atm down to 107! torr; regulation of the pressure of this mixture produces the steady-state deuterium pressure of interest and allows accurate pressure measurements to be made readily. In order to measure the pressure of the permeated deuterium accurately and to maintain a low pressure of deuterium inside the permeation tube, an argon purge is used. Use of an absolute gage to measure the total pressure and a mass spectrometer to measure deuterium concentration of the purge allows accurate determina- tion of the back pressure and correction for back diffusion of deuterium. Flow rate of the argon purge is used in combination with the measured composition to determine the permeation rate of deuterium. Permeated 7. R. W. Webb, Permeation of Hydrogen Through Metals, NAA-SR-10462 (July 25, 1965). 8. C. J. Smithalls and C. E. Ransley, Proc. Roy. Soc. London, Al150, 172 (1935). 9. G. Borelius and S. Lindbloom, Ann. Physik. 82, 201 (1926-27). 55 ORNL-~DWG 72- 4138 MECHANICAL * VACUUM PUMP A-4% D, PERMEATION TUBE 7~ Ry FURNACE PRESSURE GAGE Z é 7 / % ARGON PURGE - PRESSURE GAGE MASS SPECTROMETER MECHANICAL VACUUM PUMP Fig. 7.2. Low-pressure permeation rate apparatus. deuterium pressures in the range of 107® torr can be determined readily with this procedure. Only steady- state permeation has been attempted in this work. Consequently, no separate determination of the diffu- sivity has been made. Values of the permeation constant obtained for Hastelloy N fall within a factor of 2 of those of Webb.” In contrast with the earlier data, however, the pressure dependence of the permeation flow rate was found to follow a square-root relation quite precisely over the entire pressure range studied. This is shown in Fig. 7.3, where the data are presented in the form of an arithmetic plot of hydrogen flow rates vs /P, — \/P,, where P, is the permeation gas partial pressure, and P, is the “back™ or “downstream” pressure which must be used as a correction. Arbitrary permeation flow rate figures, without con- cern here for sample areas or thicknesses, are used to permit comparison with the earlier work on Hastelloy N. The only explanation offered here for the marked difference between the data obtained in this study and the earlier work'? is that, previously, pressure rise rates have been generally used at the lowest pressures studied. Since we have observed that steady state may require tens of hours to achieve at low pressures and that permeation flow rates themselves may be in the range of 7 X 107* cc(STP) hr ™' cm™2, a steady-state method such as the one used here is perhaps better. The conclusion from these experiments is that for Hastelloy N, a square-root pressure dependence is recommended for extrapolation to low pressures. 10. Only Webb’s (ref. 7) low-pressure results are discussed here. His steady-state method for permeability determination, which he conducted at higher pressures than for the work reported here, is not at all in question. Webb’s higher-pressure measurements are in very good agreement with those we have determined. ORNL-DWG 72-4139 ] 700 600 500 / THIS WORK "_” FROM WEBB 400 300 / - ; PERMEATION FLOW RATE (arbitrary units) 200 100 Fig. 7.3. Permeability of Hastelloy N for hydrogen as a function of P1/2. Comparison of results of this work with those of Webb [Permeation of Hydrogen through Metals, NAA-SR- 10462 (July 25, 1965)] at low pressure. 7.4 INFLUENCE OF FILMS OR COATINGS ON HYDROGEN PERMEATION RATES R. A. Strehlow H. C. Savage Calculations based on the results of the work reported in the preceding section show that bare metal probably does not provide adequate impedance to tritium flow at low pressures. Consequently, to determine whether options might exist to obtain the low flow rates required in a reactor becomes the question. The use of 56 oxide coatings to impede flow has been proposed,® since the permeation rate through oxides should be proportional to the first power of pressure. (It is known that hydrogen does not dissociate upon dissolving in ceramics.) An oxide such as forms on stainless steel in an oxidizing atmosphere (Py,o/Py, > 107° for Cr, 0 at 700°C) might offer, therefore, a useful resistance to permeation flow. Initial experiments were carried out with the Hastel- loy N tube which had been used in the low-pressure study. An argon—20% oxygen mixture was used to oxidize the external surface of the tube in situ at 690°C before determining the permeation behavior. Although a reduction in permeability was observed initially (about a factor of 2), at steady state the permeability had recovered to its value for the unoxidized tube. An experiment designed to deposit a carbon coating by thermal decomposition of methane yielded the same results. However, many hours were required to reach steady-state conditions for the permeation of the hydrogen. It should be noted that equilibrium oxida- tion conditions for nickel are associated with Py, 0/Pu, ratios greater than about 50 at 700°C. Even at the lowest pressures used in this study, we believe the conditions used in these experiments were still reducing rather than oxidizing. Type 304L stainless steel was selected as representa- tive of a class of metals which might behave differently from the Hastelloy N. A permeation tube specimen, cleaned by etching, was installed in the apparatus. Measurements at pressures from 30 to 1072 torr at first showed a permeability nearly equal to that of Hastelloy N, with a square-root dependence over the entire range. Subsequently, over a period of some tens of hours, the permeability was observed to decrease by a factor of about 8. This was presumed due to the formation of the oxide by trace amounts of oxidant in the gas system used in this work. Surprisingly, the pressure dependence of permeation rate still followed the square-root relation! A reasonable expectation had been that some departure from this should have been observed for any significant reduction or permeability. The logic for this expectation is fairly well seen in Fig. 7.4, which shows the logarithm (permeation flow rate) for various materials and thick- nesses as a function of logarithmic pressure of tritium. Zero downstream pressure is assumed here for illustra- tive purposes. The restriction of loss of tritium from the system must be accomplished by reducing the tritium pressure to values below the intercept of the permea- tion flow rate with the range indicated, which here corresponds to a loss of 1 to 30 Ci/day for an 8 X 107 cm? heat exchanger. ORNL-DWG 72-7590 0 V ) 2 4 R £t 1 // s - | ! -6 o ":E, F< / 3 / S 6 o -8 ~ r g 7 | 0 _/ . 4 / PERMEATION RANGE O -12 V4l / INTEREST FOR 7, / y /| MANAGEMENT IN REACTORS 48 46 44 -2 -0 -8 -6 -4 - log 2. (Torr) 2 Fig. 7.4. Tntlum fluxes as function of T, pressure in the absence of ! H, at 600°C. Line designations are as follows: 1 = Hastelloy N (or stainless steel) 1 mm thick, 2 = the vacuum impact rate limit, 3 = quartz 200 A thick, 4 = stainless steel oxidized for many hours (this work), 5§ = calculated value for tungsten 0.006 in. thick, 6 = quartz 0.006 in. thick. For simplicity in operating the tritium removal processes in a reactor, the maximum permissible pres- sure of tritium determined in this way should be as large as possible. (That is, purge gas volumes needed would then be minimized, and process rates for chemical removal and chemical trapping would then require the minimum mass handling.) Line 1 in Fig. 7.4 shows the permeability of Hastelloy or freshly etched stainless steel which intercepts the vacuum impact rate, line 2, at a very low pressure but still in the calculated range of interest. As an oxide surrogate the data for quartz were used,!' and a thickness of 200 A was selected for display as line 3. To account for the diminution of permeation flow rate by the factor observed using this model of two impedances in series with their different pressure dependences, a very nearly first-power dependence should have been evident. The observation of a half-power dependence leads to the conclusion that a film with only partial effectiveness in covering the metal had been obtained. The temperature was raised to 785°C from 700°C in order to permit the development of a thicker coating. The range of measurement was also extended to atmospheric pressure. For the stainless steel sample, after some hundreds of hours at temperature, a depar- ture from half power has been obtained with the data varying in a regular and reproducible way from about the two-thirds power at higher pressures to the three- 11. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 116. fourths power at lower pressure. This is shown in line 4. Also shown in Figure 7.4 is the expected permeation rate for a 0.006-in. tungsten coating and a 0.006-in. ceramic coating (again using the value for quartz). These are labeled as lines 5 and 6 respectively. The relative efficacies of these low-permeability coat- ings in considering the tritium management question must be viewed, however, in terms of the effect of added protium, since it probably cannot be avoided. The flow impedence offered by metal relative to that offered by oxide is decreased with addition of hydrogen to the system. Combined credit for the coating and for use of this mass action effect cannot be taken except for metal coatings. As a consequence, as shown in Fig. 7.5, the 0.006-in. tungsten coating becomes as good as or better than a ceramic one at pressures near 1 torr of 'H, in terms of the maximum tolerable tritium partial pressure. Within the assumptions made here, the possibly achievable factor of 103 decrease in tritium permeation rates for the thin ceramic coating and the larger factor for the thick coating simply specify the tolerable partial pressures of tritium both in the absence and presence of protium. The questions about coating use thus become those associated with ways to overcome the unavoidable differential thermal expansion problem and to achieve minimum porosity in the coating. 7.5 EXPERIMENTS ON HYDROGEN EVOLUTION FROM FLUOROBORATE COOLANT SALT S. Cantor R. M. Waller The objectives of these experiments are to (1) confirm the relatively high analytical concentrations of ORNL-DWG 72-759 5 =i log 0( PERMEATION $AI\(I)%E OF A MANAGEMENT -18 -6 -4 -2 -10 -8 -6 -4 -2 [¢] 2 4 log Py (Torr) Fig. 7.5. Tritium fluxes as function of HT pressure calculated at 600°C in the presence of 1 torr of 1H2. Line designations are as follows: 1= Hastelloy N (or stainless steel) 1 mm thick, 3=quartz — 200 A thick, 5= tungsten — 0.006 in. thick, 6 = quartz — 0,006 in. thick. hydrogen (20 to 50 ppm) reported for samples of fluoroborate coolant salt taken from thermal convec- tion loops and (2) measure the pressure of H, gas in equilibrium with hydrogen within the salt samples. The apparatus and gas handling techniques for these experi- ments are the same as for previous studies of hydrogen evolution and tritium exchange.! 2 Two fluoroborate samples were provided for this study from natural convention loop 14. Previous analyses of salt from this loop had shown: O, 3900 ppm; Cr, 350 ppm; Ni, <4 ppm. A small part of the first sample, when analyzed by the infra-absorption method, showed 26 ppm hydrogen. The remainder of the first sample (29.6 g) was loaded, in a dry box, along with Hastelloy N coupons into a nickel capsule. The capsule was subsequently evacuated and sealed off by welding a crimped section of an attached tube. The capsule was next placed in the quartz enclosure of the gas handling apparatus, in which it was heated to 535°C. In the apparatus the gas pressure, exerted predominantly by hydrogen diffusing out of the cap- sule, was measured with a McLeod gage; samples of gas were intermittently collected and analyzed for hydro- gen by gas chromatograph. The sample was maintained at 535°C for 61 days; in this period the equivalent of 23 ppm hydrogen escaped as gas from the nickel capsule. The experiment was terminated when the permeation rate out of the capsule had decreased to less than 0.025 std cm?® per day. Such small quantities of hydrogen are difficult to analyze accurately by gas chromatograph. The reaction producing gaseous hydrogen can be represented by the equation 1 1 1 OH~ +-§Cr "#~3—Cr3++02'+5H2T. (D) The equilibrium quotient is given by Cr3 ? (0%) Py 12 o= L0, @ ac:'’* (OH") where the quantities in parentheses are concentrations (e.g., g-ions/g of salt), ac, is the activity of chromium (~0.1) in Hastelloy N, and Py, is the partial pressure of hydrogen. Since the concentrations of Cr** and 0% at the start of the experiment greatly exceeded that produced in 12. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 88. 58 the experiment, the equilibrium quotient can be sim- plified to 3 2 350)1/ Pyt Q = 3900 (6'1‘ OH) 1/2 =59 X 10'4-}1‘2—_—. (3) (OH™) Thus one can obtain @ by measuring equilibrium pressures of hydrogen corresponding to OH™ concentra- tions calculated by mass balance from the original 26-ppm hydrogen concentration. During the experiment, hydrogen pressures were measured after allowing at least three days for the hydrogen passing out of the capsule to equilibrate with the hydrogen within the capsule. The data and calcu- lated Q are given in Table 7.1. Postexperimental analysis of the salt for hydrogen showed 23 ppm, substantially greater than the expected 3 ppm based upon the original analysis (26 ppm) minus the amount removed in the experiment (23 ppm). At present this discrepancy is difficult to understand. Using the postexperimental analysis for calculating the equilibrium quotient of reaction (1), we obtain Q ~ 0.72, a factor of 5 less than that based on the original hydrogen analysis (see Table 7.1). Another nickel capsule, containing the second sample from loop 14, has been installed in our apparatus; an experiment similar to the one detailed above is in progress. The infrared analysis of this particular sample, prior to encapsulation, showed 45 ppm hydrogen. In summary, our experiments thus far suggest: (a) Substantial concentrations of hydrogen were present in the samples supplied from loop 14. At least 26 ppm hydrogen could be accounted for in the first experi- ment; this level of hydrogen, if maintained in the Table 7.1. Equilibrium quotients for reaction (1) Hydrogen previously P R A (atm X 1075 (ug/g of salt) pp 1.61 13 3-4 0.43 18.3 3-0 0.31 20.2 3.3 0.25 21'3 3.7 . 0.20 21.8 37 0.125 22.7 37 Av 3.5 coolant loop of the MSBR, is more than adequate as a sink for tritium. (b) Relatively high oxide and chro- mium ion concentrations are beneficial for retaining hydroxide in fluoroborate. In reaction (1) and in the equilibrium quotient (2) the hydrogen pressure to the one-half power in inversely proportional to the oxide concentration; we may, therefore, expect that the hydrogen permeation rate (which probably follows a ', power law) will also be inversely proportional to oxide concentration. (¢) The relatively high concentration of hydroxide in the salt in loop 14 is almost certainly a reflection of the low permeation of hydrogen through the metal walls. Since this loop has been operating for a few vyears, it is likely that there is a fairly thick, coherent oxide coating on the outer surfaces. Hope- fully, high-pressure steam will deposit a similar coating on the tube walls of the steam-raising system in an MSBR. 7.6 APPARATUS FOR INFRARED SPECTRAL STUDIES OF MOLTEN SALTS J.P.Young J.B.Bates G.E.Boyd Spectral studies of molten NaF-NaBF, that had been contacted with D3BO; exhibited an absorption band attributable to BF;0D™ (ref. 13); these studies were carried out in SiO, cells. Since then, a small furnace assembly and an LaF;-windowed nickel cell have been designed and fabricated so that melts compatible with these materials of construction can be studied by infrared spectral techniques. The apparatus can be used for several fluoride melts of interest to the MSR project, but the first studies have been of melts of BF;O0H™ and/or BF;0D™ in molten NaF-NaBF,. The furnace and cell are shown in Fig. 7.6. The furnace was fabricated by LaMont Scientific Company (State College, Pennsylvania), from a con- ceptual design that we developed. The furnace is small enough to fit in the sample space of a Perkin-Elmer spectrophotometer or a Digilab FTS-20 spectrometer, or even within a Perkin-Elmer 4X beam condenser. It is anticipated that this apparatus can also be used in the emission attachment for the FTS spectrometer. The heated portion of the furnace consists of a boron nitride (BN) tube, % in. ID, wound with resistance heating wire. Boron nitride wool insulation surrounds the tube, and the assembly is enclosed in a metal can 13. J. B. Bates, J. P. Young, M. M. Murray, H. W. Kohn, and G. E. Boyd, “Stability of BF3OH ™ Ion in Molten and Solid NaBF4 and NaF-NaBF, Eutectics,” to be published in Journal of Inorganic and Nuclear Chemistry. 59 that is open at the ends, which are aligned with the BN tube ends. These ends are water cooled, equipped with Viton O-rings, and will accept infrared transmitting windows. At present we are using AgCl or AgBr windows. The furnace with windows in place is vacuum tight and can be operated under vacuum, or with an atmosphere of inert gas. A control box associated with the furnace is capable of regulating a preset temperature to less than 1°C. There is also a design arrangement so that a gas bubbler (or diffusion) tube can be inserted into the furnace and infrared cell so that melts in the cell can react with gases of interest. The cell, shown assembled in Fig. 7.6, consists of a nickel body designed to accept two LaF3 windows. The windows are held in place by nickel retainers, and the retainers are pressed against the windows by the small screws seen at the windowed ends of the cell. The apparatus is designed so that the windows can be used without a gasket as in the diamond-windowed cells.! ? In some of our early experiments it seemed that molten NaF-NaBF, leaked around the windows. Accordingly, gold gaskets (0.005 in. thick) are now being used between the window and the cell body. The operation of the cell and furnace has been satisfactory; however, some cell windows have cracked, probably as a result of samples having been cycled through the freezing point a number of times. (Thermal expansion on melting of a previously frozen sample could put a severe strain on crystalline windows.) The thermal characteristics of the furnace have been quite satisfactory; a regulated temperature of 450°C can be obtained rapidly (i.e., in less than 15 min) after a sample has been loaded. To load the cell into the furnace, one of the furnace windows is removed; the cell containing the sample is placed in the center of the furnace tube, and the window is resealed so that the atmosphere in the furnace can be flushed with argon. 7.7 INFRARED SPECTRAL STUDIES OF THE CHEMICAL BEHAVIOR OF BF,;0OH™ AND BF;0D IONS IN MOLTEN NaF-NaBF, J.B.Bates J.P.Young G.E.Boyd Recent infrared measurements'® with solutions of BF;O0H™ in molten and crystalline NaF-NaBF, estab- lished that this protonic species was stable with respect to unimolecular decomposition at temperatures below 14. L. M. Toth, J. P. Young, and G. P. Smith, Anal. Chem. 41, 683 (1969). 15. J. B. Bates, J. P. Young, M. M. Murray, H. W. Kohn, and G. E. Boyd, J. Inorg. Nucl. Chem., to be published (1972). 4 2 . l OAK RIDGE NATIONAL LA %l LABORATORY ] 60 PHOTO 1042-72 ———— e —— — 3 Fig. 7.6. Photograph of LaF-windowed cell and infrared furnace. 650°C. These studies also showed that reactions of D, with NaF-NaBF, melts containing BF;OH™ ions pro- duced BF;0D" ions by isotopic exchange and/or via reactions involving oxide impurities; for example, D2 +B2F602‘+M0'>ZBF30D_ - (l) Additional experiments were designed to further study the chemical behavior of BF3OH™ dissolved in molten NaF-NaBF, and, in particular, to discover the mecha- nism by which BF;0D” is formed on reaction of D, with these melts. The high-temperature microinfrared furnace em- ployed is described in detail in Sect. 7.6.'® The LaF;-windowed sample cell permitted infrared meas- urements to be made over the region from 2400 to 4000 cm™'. The silica cells and high-temperature furnace used in the earlier studies were also useful in the more-recent experiments.! $ The spectrum shown in Fig. 7.74 was obtained from an NaF-NaBF,; melt containing about 25 ppm hydrogen as BF;OH™. The band centered at 3645 cm™' is assigned to the OH stretching mode of BF; OH™ based on the previous solid-solution measurements.!® As 16. R. N. Kust and J. O. Burke, Inorg. Nucl. Chem. Lett. 6, 33-35 (1970). ORNL-DWG. 72-3734 T T T I T T T | T T T * A 2 (@] 3 7 o 3645 3600 = 0 = < @ |— 59 % 4/\/_ W | B | 3645 2690 | ] | | | | | | | 1 | ! 3800 3600 3400 3200 3000 2800 2600 FREQUENCY (cm™) Fig. 7.7. Infrared spectra of molten NaF-NaBF, contained in LaF-windowed cell. (4) Melt before treatment with D,; (B) melt after treatment with D,. observed in this case, some melts exhibited a second band at about 3600 cm™', which may be due to a species such as B3FsO3;O0H> . The spectrum of Fig. 7.7B was measured from the same melit after bubbling D, gas through it for about 5 min. The band at 2690 cm”! was previously observed in an NaF-NaBF, melt which had been spiked with D3 BO;. This band is due to the OD stretching mode of BF;0D™. Because the nickel had not been previously hydrogen fired, the BF;0D™ could have been produced either by isotopic exchange with BF;OH™ or by a reaction which depends on the reduction of Ni** jons by D,: Ni** + D, = Ni® + 2D*. (2) Following these measurements, the nickel cell and deuterium bubbler were cleaned and hydrogen fired. The initial experiment was then repeated, but the band at 2690 cm™! was not observed after bubbling D, through the melt for time periods of up to 30 min. Considerable difficulty was encountered at this point with loss of sample through the top of the cell because of a vapor lock with the D, gas. The infrared spectra of NaF-NaBF, melts contained in silica cells are presented in Fig. 7.8. The salt sample was the same as before (25 ppm H). Figure 7.84 shows the melt spectrum prior to D, treatment. The band at about 3670 cm™! is due to the hydroxyl impurities in the silica cell. After bubbling D, through the molten salt for 10 min, the band at 2690 cm™' appeared as shown. Additional D, treatment was continued for 30 61 min, but no change in absorbancy of the 2690 cm™ band was detected. To determine if reduction by D, of a metal ion is the primary step in the production of BF;0D7, a crystal of FeF, was added to the melt, and the treatment with D, was resumed. The infrared spectrum measured after about 10 min of bubbling is shown in Fig. 7.8B. Rather than an increase in the intensity of the 2690 cm™ band, an additional band at about 2750 cm ™! was observed. The above experiment was repeated with the addition of NiO to the melt in place of FeF,. The spectrum after D, treatment (Fig. 7.8C) exhibited bands at 2835 and 2750 cm™' in addition to the one at 2690 cm ™' . Some tentative conclusions may be drawn on the basis of the experimental results described above: (1) Direct isotopic exchange of BF;OH™ with D, did not occur to a measurable extent. (2) The reaction respon- sible for the production of BF;0D™ on bubbling D, into molten NaF-NaBF, involved reducing some chemi- cal species present in the melt with D,: A™+ D, > A2 1 oD* (3) As suggested previously,'® the D* ions may then react with BF;OH ™ and/or with oxides such as B3 F4 037 to ORNL-DWG. 72-3735 bt — — —y Before D2 /Affer 02 | 2690 ] 2750 2685 % TRANSMISSION —= 2835”7 , 2750 “og85 | I | I | l 1 l 3200 3000 2800 2600 FREQUENGCY (cm™) i 2400 Fig. 7.8. Infrared spectra of molten NaF-NaBF4 contained in silica cells. (4) Melt before and after treatment with D,; (B) melt after addition of FeF, and treatment with D,; (C) melt after addition of NiO and treatment with D,. 62 form BF;0D": D*+ BF;0H™ - BF,0D™ + H*, 4) F +D*+ %B,F¢0,3 > BF,0D" . (5) (3) The species reduced in Eq. (3) was not necessarily a corrosion product such as Fe?* or Ni**, since no evidence of increased production of BF3;0D~ was detected on addition of solid FeF, or NiO to the melt. Clearly, further experiments must be performed to determine if BF;0D™ ions can be produced on dif- fusion of D, through a metal tube immersed in molten NaF-NaBF,. Furthermore, additional studies will be directed toward understanding the mechanism by which BF30D"™ is produced in D,-treated NaF-NaBF,; melts. 7.8 THERMAL STABILITY OF NaNO;, KNO;, NaNO,, AND HITEC J.D.Redman C. F. Weaver A mass spectrometric study of the thermal stability of NaNO;, KNO3, NaNO,, and HITEC was made as part of an evaluation of their potential use as coolants in molten-salt reactors. In such salt mixtures as HITEC, NaNQO,-NaNO;-KNO; (40-7-53 wt %), the reaction H, + 2KNO; = 2KNO, + H,0, which has a large negative standard free energy, may afford the means to convert elemental tritium to tritiated water, and thus provide a means to control its distribution. Since nitrates and nitrites are known to decompose thermally, it was useful to determine what additional vapor species were formed. Reagent-grade NaNQO;, KNOj;, and NaNO, were dehydrated by bubbling helium gas (which had been dried by passage through activated charcoal at liquid- nitrogen temperature) through the molten salt. These dehydrations were continued for about 16 hr at temperatures of 345, 320, and 285°C for KNO;, NaNOQOj, and NaNO, respectively. The NaNO, cooled to a hard, yellow glass. The other two salts were relatively soft and nearly white. Handling and storage of the dehydrated materials were done in a dry-box atmo- sphere. Subsequent mass spectrometric studies showed the purification technique to be successful with respect to H, O removal. Small samples of each of the salts were monitored during evaporation from a nickel Knudsen cell. Vapor species of each salt began to appear 5 to 10°C above the preparation temperature. Decomposition pressures limited the upper temperature to approximately 125°C above the preparation temperatures and about 100°C below the temperature of MSR interest. All three salts evolved NO. In addition, the nitrates produced O, and the nitrite produced N,. Appearance potentials and relative intensities of the fragments confirmed the identity of these molecules. These permanent gases do not allow accurate pressure meas- urements because of background interference, but the total pressures were certainly in the range of 0.1 to 1 torr at the upper temperature limit (450°C). In addition to the NO, O,, and N,, Na and K were carried into the gas phase by some unidentified species. The appearance potentials of the Na* and K* ions make it clear that the Na and K were not present as the metal vapors. Both NaNO; and NaNO, attacked the nickel cell and produced NiO. These results are in agreement with the conclusion of Kust and Burke'® that nitrate melts decompose at lower temperatures than previously suggested,’”!'® though Kust and Burke did not determine the composi- tion of the evolved gases. The earlier higher temperature studies! 7 (600 to 780°C) produced mixtures of O, N,, and NO,, while at the lower temperatures (up to 450°C) of these experiments we saw NO and N, or NO and O, for nitrite and nitrate respectively. A batch of HITEC was prepared from the previously purified components. Dry helium gas was bubbled through the molten mixture, contained in a glass flask, overnight at 180°C. The cooled, nearly white, very hard material was ground, handled, and stored in a dry box. A sample of the ground and well-mixed material was evaporated in a nickel Knudsen cell over the tempera- ture range 200 to 600°C. Decomposition products appeared near 275°C. The gases NO, O,, N,, and possibly N, O were observed over this mixture as well as unidentified species carrying Na and K in a nonmetallic form. Total pressures were about an order of magnitude lower for the ternary mixture when compared with the components at the same temperatures. Interestingly, NiO was not detected in the residue from this experi- ment. Studies of these materials at higher temperatures will require a different approach, such as transpiration and off-gas analysis, since the pressures will be above the range acceptable with the mass spectrometer. 17. Eli S. Freeman, J. Phys. Chem. 60, 1487—93 (1956). 18. B. D. Bond and P. W. M. Jacobs, J. Chem. Soc., A, 1966, pp. 1265—68. 8. Fluoroborate Chemistry 8.1 SOLUBILITY OF BF; IN FLUORIDE MELTS S. Cantor R. M. Waller The solubility measurements of BF; in fluoride melts serve to determine the activity of fluoride ion in molten mixtures of LiF-BeF,, LiF-BeF,-ThF,, and LiF-ThF,. Since the scope of this program involves measurements in a relatively large number of melts, we have changed our method of measurement from the sparge-and-strip technique! to a new method which generates data faster and, probably, with greater accuracy. In the present method, BF; pressures are measured in a cylindrical nickel vessel containing a weighed sample of salt and a known vapor volume. Once the vapor space in the vessel is determined, the inert, virtually insoluble calibrating gas (argon) is evacuated; a meas- ured amount of BFj; is then introduced into the evacuated vessel. The fraction of BF; that does not dissolve in the sample is calculated from the pressure, volume, and temperature profile of the vapor space. Salt charges are usually sufficiently large to dissolve at least 80% of the BF; introduced into the vessel. Pressures are read off a strip-chart recorder which tracks the output from a stainless steel strain-gage transducer connected to a riser on the vessel. The vessel is positioned in a furnace mounted on a motor-driven rocker. As a test of the reliability of the new method, BF; solubilities were measured in LiF-BeF, (66-34 mole %) in the temperature range 498 to 839°C. The solubility of BF3 per unit pressure, that is, Henry’s law constant, agreed closely with that obtained in the sparge-and-strip method (see Table 8.1). Solubilities of BF; in two other LiF-BeF, melts were also measured. These data and some derived enthalpies of solution are listed in Table 8.2. Previously, we have reported? that Henry’s law constants for BF; solubility varied linearly with the thermodynamic activities> of LiF in the system LiF- BeF,. As Fig. 8.1 shows, the newer data obtained in three LiF-BeF, mixtures exhibit this linear behavior. 63 Table 8.1. Solubility of BF3 (mole %/atm) in 66-34 mole % LiF-BeF, Solubility (mole %j/atm) Temperature ©C) Present Sparge-and-strip Difference (%) method method 500 0.75- 0.77, 2.9 650 0.14, 0.14¢ 0.7 800 0.045, 0.0444 1.3 Further, the newer results are consistent with previous data? obtained in compositions with greater LiF con- centrations. Solubilities of BF; were also measured in MSBR fuel solvent (72-16-12 mole % LiF-BeF,-ThF,). The data are summarized in Table 8.2. When we apply the isotherms of Fig. 8.1 to the Henry’s law constants in this solvent, the calculated activities of LiF are 0.21, 0.225, and 0.255 at 500, 600, and 700°C respectively. The LiF activities in MSBR fuel solvent are about one-half those in 72-28 mole % LiF-BeF,.3 This trend is consistent with other observations (e.g., LiF liquidus curves?) which show that LiF-ThF, interactions are considerably stronger than LiF-BeF, interactions. 8.1.1 Reactor Applications Assuming that the BF; solubility in MSBR fuel solvent is about the same with or without 0.3 mole % 233UF,, then at 704°C (1300°F) the B/233U ratio would be 0.31 under a BF; partial pressure of 1 atm. Since the boron coefficient of reactivity is approxi- mately 0.5 multiplied by the atomic ratio, B/U,> then 1. S. Cantor and W. T. Ward, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, p. 78. 2. S. Cantor and R. M. Waller, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 78-79. 3. B. F. Hitch and C. F. Baes, Jr., Inorg. Chem. 8, 201 (1969). 4. R, E. Thoma (ed.), Phase Diagrams of Nuclear Reactor Materials, ORNL-2548, pp. 33, 72 (November 1959). 5. J. R. Engel, ORNL, personal communication. 64 Table 8.2, Solubility of BF3 in molten fluoride solvents and enthalpy of solution Henry’s law Temperature constant-temperature (mole %) range measured equation;? Kyy? in LiF BeF, ThF, °0) mole fraction BF3/atm; temperature in °K Solvent AH (kcal/mole) 66 34 498 to 839 in Ky 60 40 474 to 810 ~15.232 + 7452/T ~14.8 501 49.9 408 to 600 ~16.614 + 7771/T ~15.4 72 16 12 530 to 795 = -14.950+ 7667/T ~15.2 —14.949 + 7782/T —15.5 9] east-squares fit of the data. bExperimental error in Kpy is approximately +5%. 62 ORNL-DWG 72-7592 positive shutdown of the reactor. To obtain some idea /_—[r—_t1—+l about the rate of BF; dissolution in fuel salt, we 866 (mole % LiF) permitted BF; to flow into the vapor space of an i evacuated vessel containing 2 kg of MSBR fuel solvent 5 500°C 75 at 585°C. Except for possible convection currents, the / salt remained quiescent during the gas inflow. Under a 600 70 BF; pressure of 28 psig, the initial rate of dissolution 4 o8 was 0.63 X 107% mole of BF; per minute per cubic A e66 centimeter of salt. This rate of dissolution corresponds to a B/U atom ratio of 0.004 per minute, and if this rate prevailed in a reactor, subcriticality would have /1. 50.1 /70 occurred in 10 min. However, this time interval refers J to a stagnant salt melt and a relatively low driving force [\V] o2 [w] (@] [] O W\ ~ (6} S w 66 of 28 psig of BF3. Very simple improvement in BF; f injection, together with circulation of the salt, would / have greatly shortened the time to subcriticality. These 60e J considerations suggest that BF; gas at moderate pres- P50 sures can be effective in a backup scram system in the / | J I MSBR. The BF; could be subsequently removed from ® PRESENT DATA _| the salt by first sparging with helium, followed by V O PREVIOUS DATA neutronically “burning” any boron residue in the salt. 1 Ao The measurements of BFj3 solubility in MSBR fuel solvent are also useful in estimating certain effects of mixing fluoroborate coolant and MSBR fuel salt. If o fluoroborate coolant were to leak into the MSBR fuel 0.05 0.1 0.2 0.5 1.0 circuit, the NaBF, in the coolant would decompose to ACTIVITY OF LiF NaF and BF;; the former would dissolve in the salt, Fig. 8.1. Henry’s law solubility of BF vs activity of LiF in while the latter would distribute between the salt and molten LiF-BeF, at 500, 600, and 700°C. the vapor space. The equilibrium distribution can be calculated from the following equations: W ~ HENRY'S LAW CONSTANT (mole fraction 8F3/a?m) at this temperature and the BF; equilibrium pressure, the reactivity loss would be 15.5%; this number is about iy =hg * 1, (1) eight times greater than a 2% reactivity loss which RTK W+ V assures subcriticality. Thus fairly low pressures of BF, B _K1%H £ (2) ng Vy Vg in equilibrium with MSBR fuel salt are sufficient for where n, is the total number of moles of BF; leaking in from the coolant; Ny and ng are the number of moles of BF; in the vapor space and in the salt respectively; R is the gas constant; I7M is the molar volume of the salt in the fuel circuit; 7 is the temperature in degrees Kelvin; V¢ and V, are the volumes of salt and vapor space in the fuel circuit (to a good approximation, VeV is the void fraction in the salt); and Ky is Henry’s law constant in moles of BF; per mole of salt per atmosphere of BF3, almost the same as the units of Ky in Table 8.2. Equation (2) is readily derived from the ideal gas law in conjunction with simple dimen- sional considerations. A simple application of the above equations involves the case of a low inleakage such that Ky is not very different from the Ky of the fuel itself. Assume T = 950°K (1250°F) and a void fraction of 0.01; also R = 82.054 cm®-atm/(°’K-mole) and ¥, = 19.5 cm®/mole. Solving Eq. (2), 82.054 X 950 19.5 n -+ exp (—14.95) g 1 —— =411, 0.01 X exp (7667/950) X The magnitudes of ng and n, would then be calculated from Eq. (1). 8.2 FREE ENERGIES OF FORMATION OF NaFeF; AND NaNiF;; THEIR RELATIONSHIP TO THE CORROSION OF HASTELLOY N BY FLUOROBORATES C. F. Baes, Jr. The double salts NaNiF; and NaFeF; have been shown to be stable, relatively insoluble compounds in molten NaF-NaBF, mixtures.® A knowledge of their formation free energies would permit an estimate of the equilibrium position of such corrosion reactions as B. F. Hitch C. E. Bamberger 2HF(g) + Ni (Hastelloy N) + NaF(d)= NaNiF5(c) + Hy(g), (1) 2HF(g) + Fe (Hastelloy N) + NaF(d) = NaFeF;(c) + Ho(g) . (2) 6. F. A. Doss and J. H. Shaffer, MSR Program Monthly Rep. Dec. 1970—Jan. 1971, MSR -71-13, p. 32 (February 71); MSR Program Monthly Rep. Aug. 1971, MSR-71-81, p. 16 (Septem- ber 1971). 65 The former reaction should correspond to the most oxidizing condition which Hastelloy N can tolerate. The present work describes the experimental determi- nation of AG/ of NaNiF; and of NaFeF, based on measurements of the equilibria NaNiF;(c) = NaF(d) + NiF,(c), K3, =an,F » (3a) NaFeF;(c) = NaF(d) + FeF,(c), K3p =an,F » (3b) where ay, g is the activity of NaF, and (¢) and (d) represent, respectively, the crystalline and dissolved state. The experimental procedure consisted in equilib- rating at various temperatures the solids NiF, and NaNiF, or FeF, and NaFeF; with molten NaBF, while accurately measuring the BF; pressures. Since the solids are not very soluble in molten NaBF, + NaF, the liquid phase may be treated as a simple binary mixture, and thus the BF; pressures measured can be used to calculate the NaF content of the melt corresponding to equilibria (3¢) and (3b) above. The NaBF,-NaFeF;- FeF, mixture was contained in a 1%-in. nickel vessel fitted with a copper liner; the mixture with niobium compounds did not have the liner. Each reaction vessel was equipped with a thermocouple well and connected directly to a mercury manometer and, through a Hoke 413 valve, to a vacuum pump. All tube connections were either silver soldered or welded. Typically (where M is either niobium or iron), 0.3 mole of NaBF,, 0.1 mole of MF,, and 0.06 mole of NaMF; were loaded into the reaction vessel. The NaMF; compounds were prepared by heating stoichio- metric amounts of NaF and MF, with a small amount of NaBF, at 800°C under argon. X-ray diffraction analysis of the product indicated it to be more than 95% NaMF;. After loading, the reaction vessel was sealed and flushed with argon several times. The system was then evacuated for several hours at room tempera- ture; no detectable leaks were found. The reaction vessel was then heated to 500°C and cooled to room temperature. A residual gas pressure of 50 and 80 mm Hg remained (with iron and niobium compounds respectively), due probably to unreacted BF3 and traces of water which had reacted to produce HF. Subse- quently, the system was evacuated once more {0 a low pressure (1 to 10 u) to remove any residual gas. All the data presented here (Fig. 8.2) are from measurements made during the latter runs. At the end of each equilibration the solids were analyzed by x-ray diffrac- tion, which confirmed the presence of NiF, and NaNiF; and of FeF, and NaFeF;. ORNL-DWG 72-7593 1000 500 r 50 20 - 145 1.20 {25 4.30 1000/ oy Fig. 8.2, BF3; pressures generated by NaF-NaBF; melts saturated with (1) NiF;-NaNiF; and (2) FeF,-NaFeF;. The lines were calculated by a least-squares fit of the data according to Eqs. (6), (7), and (9) or (10). The activities of NaF,ay, p, were calculated from the measured BF; pressures by means of the equilibrium constant for the reaction NaBF,(d) == NaF(d) + BF5(g) , (4) AdNaF Ky=P,p ——— BF3 aNaBF4 which has been measured by Cantor’ as log K4 = [5.772 — 6.513 (10°/T)] £0.04 . (5) Introducing mole fractions (X) and activity coetficients (v) into the equilibrium constant expression and solving for ay, r, we obtain 1 BFa/KaYNapF, ¥ VNaF (6) ANaF =P In the present measurements, wherein the melt was always >97 mole % NaBF,, Cantor’s results® indicate that yy,pp, is very nearly 1.0 and yy,p is 1.3 at 66 1000°K. Assuming NaF-NaBF, mixtures to have an ideal entropy of mixing, we may then estimate 113.9 108 YNaF =< T >i0.04. The mole fractions of NaF at each equilibration temperature [calculated from Egs. (6) and (7)] are plotted in Fig. 8.3. The smooth curves shown in Figs. 8.2 and 8.3 were generated by assuming that log K3, and log K5 vary linearly with 1/T(°K), that is, that (7) logay,p=a+b(10°/T). (8) First, the more numerous data from the equilibrium with NiF,-NaNiF; were fitted by least squares, and then assuming that the entropies of reactions (3a) and (3b) are the same, the Py data from the equilibrium with FeF,-NaFeF; were fitted by least squares, using the value of & determined for equilibrium (3z). The resulting expressions for the equilibrium constants and free energies for reactions (32) and (3b) are log K3, = [0.58 — 2.11 (10°/T)] *0.04, (9) (10) AG [reaction (32)] = [9.66 — 2.65 (T/10%)] £ 0.2, (9) log K3p = [0.58 — 1.93 (103/T)] +0.04 AG [reaction (3b)] = [8.81 — 2.65(T/10%)] £0.2.(10) These results were then used to calculate the free energies of formation of NaNiF; and NaFeF; as follows: AGY [NaNiF;(c)] = AGS [NaF(D)] + AGS [NiF,(c)] — AG (reaction 32), (11) AGY [NaFeF;(c)] = AGY [NaF(1)] + AGY [FeF,(c)] — AG (reaction 3b). (12) 7. S. Cantor (ed.), Physical Properties of Molten Salt Reactor Fuel, Coolant, and Flush Salts, ORNL-TM-2316, p. 34 (August 1968). 8. S. Cantor, R. E. Roberts, and H. F. McDuffie, Reactor Chem. Div. Annu. Progr. Rep. Dec. 31, 1967, ORNL-4229, p. 55. ORNL—DWG 72~7594 0.05 0.02 w o P s S 0.0 (8] 9 — ] o , g 71; - K —_ 08~ 3 0.005 —— —[— G R e S 7 L ! } 8\4/0 ) i | i S _‘T—flkf s \.. _._J Ne 0000 L | 140 115 120 1.25 130 435 140 145 1000 /7 (o) Fig. 8.3. NaF mole fractions in equilibrium with NaBF4 saturated with (1) NiF,-NaNiF; and (2) FeF,-NaFeF;. The lines were calculated from Eqs. (7) and (9) or (10). The formation free energies of NiF,(c) and of FeF,(c) are known from the measurements of Blood:® AGT [NiF,(c)] = [-156.33 +37.65(7/10®)} £ 09, (13) AG/S [FeF,(c)] = [-168.62 +3298(7/10%)] £09. (14) The formation free energy of liquid NaF was obtained from the JANAF tables:1° AG/ [NaF(1)] = [—130.39 +19.42(T/103)] +1.0. (15) Combining these gives the formation free energies of NaNiF; and NaFeF;: AG/ [NaNiF;(c)] = [-296.38 +59.72(T/103)] 1.4, (16) 9. C. M. Blood, Solubility and Stability of Structural Metal Difluorides in Molten Fluoride Mixtures, ORNL-CF-61-5-4 (September 1961). 10. JANAF Thermochemical Tables, 2d ed., U.S. Department of Commerce, NSRDS-NBS-37 (June 1971). 67 AGS [NaFeF;(c)] = [-307.82 +55.05(T/10*>)) £ 1.4. (17) Combining Eq. (9) with the equilibrium constant for the reduction of NiF,(c) by hydrogen, based on measurements by Blood,? H,(g) + NiF2(c) = 2HF(g) + Ni®(¢) (18) logK,s = [8.67 — 5.67 (103/T)} + 0.04, we obtain the equilibrium constant of reaction (1): log (PHQ/PHFz aNi9 XNaF YNaF) = [-9.25 + 7.78 (10%/T)] £ 0.06 . (19) Similarly, combining Eq. (10) with the equilibrium constant for the reduction of FeF,(c) by hydrogen, based on measurements by Blood,? log Ko = [6.65 — 8.36 (103/T)] £0.02, (20) we obtain the equilibrium constant for reaction (2): log (Pyy, /Pyy5” @pe® X NaF ¥ NaF) = [-8.23 + 10.28 (10%/T)} £0.05. (21) Assuming that the activities of nickel and iron in Hastelloy N are, respectively, ~0.70 and ~0.05, and that the activity of NaF in the coolant salt is ~0.08 (for the eutectic composition), the ratios of PHF, /PH, at which nickel and iron will be oxidized, respectively, to NaNiF; and to NaFeF; are calculated as follows: Temperature Ni’/NaNiFj Fe®/NaFeF, O K19 Pup’/Pu, K20 Pur*/Pu, 400 203.0 0.09 1.10x 107 2.3x 1073 500 6.5 2.75 1.16 X 10° 22%x 1073 600 0.5 38.82 3.50X 10° 7.14 X 1072 From these estimates it would appear that relatively large HF partial pressures may be used, if hydrogen is present, without significant amounts of nickel oxida- tion in NaF-NaBF, mixtures. Under similar conditions the iron in Hastelloy N is more susceptible to oxidation by HF. The availability of formation free energies and the low solubilities of NaNiF; and NaFeF; in molten 68 NaF-NaBF,; mixtures suggest that these compounds would be promising materials for reference electrodes to be used in molten NaF-NaBF,. 8.3 PREPARATION OF FUSED SODIUM FLUOROBORATE FOR THE COOLANT SALT TECHNOLOGY FACILITY F. A. Doss W. P. Teichert Wiley Jennings, Jr. J. H. Shaffer The preparation of approximately 1550 Ib of the fused mixture NaBF;-NaF (92-8 mole %), for use in operating the Coolant Salt Technology Facility (CSTF), was begun on January 3, 1972, and completed on March 7, 1972. This production operation was con- ducted in the Fluoride Production Facility in six batch operations of about 258 ]b each. Since the CSTF is currently under construction, its detached drain tank was positioned within the production facility for direct loading from the batch processing unit. The loaded drain tank was then moved to the CSTF for installation. Sodium fluoroborate used in this production effort was part of a 6000-Ib lot purchased from the Harshaw Chemical Company as a custom preparation. Maximum concentrations of impurities in this material were specified at levels sufficiently restrictive to facilitate the production of the fused mixture. Sodium fluoride, which was added to make the fused fluoroborate mixture, was RACS grade and did not significantly affect the total impurity concentration of the mixture. With these starting materials the production procedure was reduced effectively to the removal of water vapor by evacuation while heating the powdered salts at controlled rates up to 300°C. Although materials specifications allowed 1000 ppm by weight of water in the starting materials, only negligible quantities were collected in the cold trap during each run. The salt mixture was then heated beyond its melting point to S00°C under a static atmosphere of argon and then sparged at 10 liters/min with BF; for 10 min to ensure mixing of the two salts. Residual BF; was purged from the system with argon, and a filtered sample of the salt was withdrawn for chemical analyses. The salt transfer line between the drain tank and the production vessel was connected, and the molten fluoroborate mixture was displaced into the drain tank by argon pressure. This procedure was repeated for each of the six batches of fluoroborate mixture. The results of chemical analyses from the six batch preparations are shown in Table 8.3. The averages of these values should be representative of the total of materials transferred into the drain tank. Nickel, chrom- ium, and iron contents of 16, 13, and 141 ppm by weight, respectively, are well within limits specified for molten-salt systems. The average oxide content of the melt was 319 ppm by weight and compares favorably with a specified limit of 250 ppm in the starting material. Results of proton analyses on five of the six batches correspond to 15 ppm by weight. This quan- tity, if present as the hydroxyl ion, would imply its association with 240 ppm of the total oxide found. The average values reported for the major constituents (i.e., sodium, boron, and fluorine) differ {from those calcu- lated for the mixture by quantities no greater than the error limits of the analytical determinations. Arbitrary calculations of the salt composition in mole percent yield an average value of 95.2% NaBF, and 4.8% NaF. This implied discrepancy in the salt composition is of little consequence to the operation of the CSTF, since Table 8.3. Results of chemical analyses of NaBF4-NaF (92-8 mole %) prepared for the Coolant Salt Technology Facility Major constituents Batch Impurities (ppm by weight) N (wt %) o. i Na B v Ni Cr Fe 0 H 1 21.2 9.89 69.3 21 38 223 275 NA 2 21.6 9.51 67.9 10 8 25 320 17 3 22.2 9.45 69.0 19 <10 203 306 15 4 21.7 9.93 67.8 16 8 135 267 14 5 21.4 9.77 69.3 15 6 125 437 13 6 21.9 9.66 67.7 15 6 135 310 15 Average 21.67 9.70 68.5 16 13 141 319 15 Calculated 22.03 9.53 68.4 the actual salt composition will readily adjust to an equilibrium with the BF; value of the cover gas stream. The sodium fluoroborate mixture prepared by this production effort is considerably better, with respect to its oxide content, than previous batches produced by 69 generally similar procedures for loop operations. This improvement is probably the direct result of the “best effort” provided by the Harshaw Chemical Company in their preparation of the sodium fluoroborate starting material. 9. Protactinium Chemistry 9.1 OXIDE CHEMISTRY OF PROTACTINIUM IN MSBR FUEL SALT R.G.Ross C.E. Bamberger C.F.Baes, Jr. Studies of the precipitation of protactinium from molten LiF-BeF,-ThF, (72-16-12 mole %), described in several previous progress reports,!” have been com- pleted, and the results are summarized here. Under sufficiently oxidizing conditions (Fig. 9.1), a very insoluble compound of protactinium(V) is formed in which, judging from the stoichiometry of the precipitation, the O/Pa ratio is 2.5. It is thought that this compound is a fluoride addition compound, prob- ably of LiF, because: (1) the entropy change deter- mined for the following equilibrium is about 23 eu higher than expected for pure Pa, O;: Pa02_ 5 '”LiF(C) + S/4ThF4(d) = PaF¢(d)+ nLiF(d) + % ThO,(c), (1) log (Xpar o)/ (XThE,)*/? = [4.49 - 8.66(10°/T)] £0.2. (As wusual, the letters g, d, ¢, and ss will denote, respectively, components in the gaseous, molten fluo- ride, crystalline, and solid solution states.) This suggests that when the Pa,O5 phase reacts with the molten fluoride, it releases more species to the solution than would be the case for pure Pa,O5. (2) Despite several attempts, we were unable to identify pure Pa,Og in oxides separated from equilibrated mixtures. If the 1. C. E. Bamberger, C. F. Baes, Jr., R. G. Ross, and D. D. Sood, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 62—66. 2. C. E. Bamberger, R. G. Ross, and C. F. Baes, Jr., MSR Program Semiannu. Progr. Rep. Feb. 28, 1971 ORNL-4676, pp. 119-22. 3. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr., MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp- 92-95. 70 Pa, Os phase does contain added fluoride, it is probably LiF, since lithium ion is the most basic cation present. With UF, in the molten fluoride at concentrations typical of an MSBR fuel (~0.3 mole %), the ThO, phase in reaction (1) is replaced by a solid solution UO,-ThO, (~95 to ~5 mole %): P302 .5 'nLiF(C) + 5/4 UF4 (d) =PaF(d) + nLiF(d) + %, U0, (ss), (2) log (XpaF ) (XU0,7U0,)°*/(Xur,)® = [4.49 — 5.69 (10°/T)] £0.2. The equilibrium quotient for this reaction was meas- ured directly and found to agree well with values predicted from reaction (1) and the previously deter- mined quotient for the U*-Th** exchange between fluoride and oxide phases (see the following sections). The resulting values indicate that under sufficiently oxidizing conditions and sufficiently low temperatures a large fraction of the protactinium in an MSBR fuel may be efficiently isolated as a pure phase without precipitation of the UQO, solid solution or any other phase. Under reducing conditions, the Pa,0Os solubilized as PaF,: phase is PaO, 5-nLiF(c) + % UF4(d) + "5 H,(g) = HF(g) + PaF,(d) + nLiF(d) + %, U0, (ss), (3) log (Xpar,) (Xv0,700,)°/* (Pup)/(Xyg,)*’* X (Py,)'/? = 8.41 — 11.44 (10°/T)] £ 0.5 . Combination of reactions (2) and (3) gives the equi- librium HF(g) + PaF4(d) = PaFs(d) + ", H,(g) , 4) log (Xpar,) (Pu,)' */(Xpar,) (Pur) = [-3.92+5.16 (10°/T)] £ 0.5 . 71 ORNL-DWG 72-7588 4 1 . | o 3 o o x b3 ™ l ™ —_ » Ilunlr Ilum___g g| 3 X | X - g ! [ Pofs () I 2 PO, & n LiF (¢) S " | € |~0.95U0,~0.05 ThO,-Pa0,] 7 0 i (ss) — | e - ] —~ o | W N - —5 {: - XpaFg/XpaFa= ! I : D 5 LS % | “Pory=3x107 —Hax g -3 Xpa0, ¥ 1.3x1073 ——— g | _XPaF5=3x10_4 13 =3 -5 _fifi—L-—‘ -4 Xpar, = 3X10 XP002% 1.3x1072 — — 2 -5 Paf, () 4, ~0.95 U0, ~0.05 Th0,-Pa0; -6 (ss) 7 — 0 -7 -8 -4.5 -4.0 -3.5 -3.0 log on_ Fig. 9.1. Pourbaix diagram for protactinium species in molten LiF-BeF,-ThF,-UF, (72-16-11.7-0.3 mole %) at 600°C. This reaction is of special interest, since such reactions, involving the HF/H,,F~ couple, have been used to establish a set of E© values for various couples in MSR molten fluorides.* From the equilibria measured and other thermo- chemical data, the following formation heats and free energies were estimated for protactinium compounds: AH 398 AG67300-1000 Compound (kcal/mole) (kcal/mole) Pa0,(c) (-270) ~268.5 + 41.3 (T/103) PaF4(d) —~469.3 +61.1 (T/103) PaF 5 (d) —~558.1 + 78.0 (7/103) “Pa,05(c)” —697 —693.5 +147.6 (T/103) The AHf, 45 value for PaO, was estimated by interpo- lation from the values for ThO, and UQ,. The values for “Pa, 05" were calculated on the assumption that the Pa,Os phase in the present system contained no fluoride. If, as we suspect, it does, then these AH' and AGS values are more negative than those for pure Pa,0q. A Pourbaix diagram, representing the behavior of protactinium in an MSBR fuel as a function of Redox potential and the oxide concentration, is shown in Fig. 9.1. The striking features are the strong dependence of 4. C.F.Baes, Jr.,, “Nuclear Metallurgy,” p. 617 in Symposium on Reprocessing of Nuclear Fuels, vol. 15, ed. by P. Chiotti, USAEC-CONF-690801 (1969). the solubility of the protactinium on the oxidation potential (i.e., the U*/U>" ratio) of the fuel and the very low solubility of the Pa,Os phase. The potential of the Pa®"/Pa** couple will determine the maximum U%*/U*" ratio in the fuel necessary to prevent inad- vertent precipitation of protactinium oxide. It will also determine the minimum U%/U3 ratio in the fuel necessary to provide the desired insolubility of the Pa,Os phase in a separation process. Because of the importance of the Pa®*/Pa** couple, it is in need of more accurate determination. Plans are being made for such measurements which will employ volumetric and/or spectrophotometric methods. 9.2 BINARY SOLID SOLUTIONS OF PaO, AND OTHER ACTINIDE DIOXIDES AND THEIR EXCHANGE EQUILIBRIA WITH MOLTEN-SALT REACTOR FLUORIDES® C.E. Bamberger R.G.Ross C.F. Baes, Jr. We have previously reported® on measurements of the equilibrium quotient of the reaction ThO, (ss) + PaF,4(d) = ThF,(d) + PaO5(ss), (1) Pa _ Q1% = XPa0, XThF,/XThO, XPaF,, - We found that the distribution of Pa*" favors the oxide phase, although not as strongly as U*' in a similar system.” Additional data have been obtained from a new experiment where the amount of ThO, added was significantly increased in order to decrease the ratio ThO, dissolved/ThO, added. This, in turn, reduced the uncertainty in the composition of the precipitated phase when calculated by material balance. Previous and present data are shown in Table 9.1. The values obtained for the distribution quotient Q%i have been correlated with the previous determinations of Qgh (ref. 3) and a value of Q,l;‘f] derived indirectly from previous measurements® on the basis of the following considerations: 1. Hietala® has successfully accounted for the ob- served heats of mixing of binary alkali halide solid solutions with a common ion in terms of a simple model which calculates the displacement of the common ion in a fixed lattice of the randomly mixed counter jons. For 1 mole of solution the result is of the approximate form dy —d\? AH,, =C g X, X,, 2) 1 72 where d, and d, are the cation-anion distances for the two pure salts. For solid solutions with the NaCl structure, C is a constant. It contains no adjustable parameters, being a function only of the molar volumes and compressibilities of the pure salts. While an analogous treatment has not yet been completed for substitutional solid solutions with the fluorite structure, such as those of the actinide diox- ides, it is clear that the result will be of the same form. Hence for 1 mole of solution we may write XTho,AM0O, s (3) aTh02 —aMO2 2- AH, ~ A aTho, where ap g, is the lattice parameter for the pure oxide. If we assume random mixing, we obtain for the activity coefficients in the binary solid solutions 2 4 /2ThO, — aMO, e X2 Inymo, = &= 2o, > ThO,» (4) 2 A aThO2—aM02 ln7Th02=§“T' Tho. X’Mo, - () 2. Considering now the exchange reaction (1), we would expect from electrostatic considerations that the enthalpy of reaction could be approximated by 1 1 AH® =B — : aMQ, 4ThO, (6) Taking AS? for this reaction to be zero, as has been found to be the case for the U*-Th*" exchange, we obtain for the equilibrium constant mgM =B 1 13 RT\@M0, 4Tho, (7) 5. Abbreviated version of a paper presented at the Solid State Chemistry Symposium, Gaithersburg, Md., Oct. 18-21, 1971. 6. C. E. Bamberger, R. G. Ross, C. F. Baes, Jr., and D. D. Sood, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 62. 7. C. E. Bamberger and C. F. Baes, Jr., J. Nucl. Mater. 35, 177 (1970). 8. C. E. Bamberger, R. G. Ross, and C. F. Baes, Jr., J. Inorg. Nucl. Chem. 33,767 (1971). 9. I. Hietala, Ann. Acad. Sci. Fenn., Ser. A6: Ph. 121-23 (1963). Table 9.1. Equilibrium data obtained for the reaction ThO,(ss) + PaF4(d) = Pa0O,(ss) + ThF4(d) Temperature Sample °0) Xzo” Xpar, Xpa0,” log 072 a (A) x 1073 X 107* A 567 0.35 4.38 (0.324) 2.12 £ 0.48 B 567 0.58 4.10 (0.291) 2.08 + 0.26 C 567 1.08 3.40 (0.292) 2.16 + 0.15 D 663 2.78 3.38 (0.167) 1.85 +0.13 E 727 2.70 4.16 (0.109) 1.55 + 0.08 F 567 2.83 2.71 0.175 0.02"d 1.97 + 0.06 5.5926 + 0.0003 0.073 = 0.004 1.69 + 0.05 G 3 14. . . + 0. 730 88 191 (0.050) 1.52 + 0.05 5.5910 + 0.0002 H 567 8.88 0.746 (0.062) 2.02 £0.10 I 567 8.88 0.686 (0.063) 2.07 + 0.10 4Total moles of oxide added per mole of solvent (LiF + BeF, + ThF,). bNumbers in parentheses were calculated by material balance. “Determined by x-ray fluorescence. dDetermined by gamma spectrometry. ORNL-DWG 71-11946 3. Since it has been found experimentally that the 10° —————— ——— = —— ratio of activity coefficients in the fluoride phase . — ——— 7 YThF,/YMF, is a constant 1ndependent of melt compo- I S R TS // sition, and may be taken as unity, QTh and KTh are ) (@=s3%6 B} related by NpC, l { 10 (0 =5.425 A)— o — e V4 M _ M — ——|—— KTh - QTh7M02/7ThO2 . (8) § T ) R / o Combining Egs. (4), (5), (7), and (8), the following 2 / - expression for Qrp, as a function of oxide composition S103 / and temperature is obtained: G S S W2 ?/ b GEEE E 5 (0 =5.4704 A)/ i POINTS ] »= Pa0, / ] In QY =R£< L1 > SE L0 . ... |le=55094) 7[7 T'\amo0, 9Tho, . é/ 55(}— 102 3/ 2 ) . SRS S ) 4 — = 7;7; 4 [Tho, — aMO, B = 2X -1. 9 ° R\ armon 2XMo, - D). ) - _ 2 / S o B - This expression is compared with the measurements 10! y ] (normalized to 600°C) in Fig. 9.2. The values of 4 and i T B B, > 7/ , / = + 2 ——— Th02 —_ A =2440 + 300 kcal/mole , : /7(1 5 5955 &) - 10 ¢ B =2620% 53 kcal, A, mole™ | 0.78 0.180 0.182 0.184 0.186 1/a have been chosen to reproduce the hne previously generated by least squares to fit the Q Th values vs Xvo, and T. As can be seen, the consistency with the other QTh values is qu1te satisfactory. The large uncertainty assigned to QT reflects the large uncer- tainty in the free energy of dissolution of UF4(c) in Fig. 9.2. Exchange quotient QThM as a function of the composition of MO,-ThO, solid solutions (in mole fractions X MOg) at 600°C. The points are measurements in which the composition of the oxide phase was determined directly (e) or by material balance (o). The lines were calculated by means of Eq. (7). LiF-BeF, (67-33 mole %), assumed to be equal to that of PuF,(c¢), presently unknown. Values of KTh’ calcu- lated from measured values of QT by means of Eqs. (4), (5), and (8), are shown in Fig. 9.3, together with a line calculated with Eq. (7). We hope in the future to obtain more data on Pu0O,-ThO, solid solutions, probably by direct meas- urements of the equilibrium involving PuF,. The large values of K,I;‘;1 indicate that the equilibrium oxide phase precipitated from melts containing ThF,, and suffi- ciently oxidized to contain appreciable amounts of PuF,, should consist of nearly pure PuO,. ORNL-DWG 71-11947R 10° — i 7 5 — A Pu0,-ThO, | 2 V/// o [/ // AV x N Q X'_ 2 \v A 4 POINTS S 17"&02_“02 x 5 —— -' s sE - - o — 2 B - Pa0,-Tho, ] 5 =— | = 1 2 10! 0 0.2 0.4 0.6 08 1.0 XMO Fig. 9.3. Equilibrium constant for exchange reactions as a function of the reciprocal of the lattice constants (1/a) of the pure actinide dioxides at 600°C. The points are derived from those in Fig. 2.1 using Egs. (4), (5), and (8). 74 9.3 TERNARY SOLID SOLUTIONS OF ThO,, PaO,, AND UO, C. F. Baes, Jr. In the previous section, the exchange of two tetra- valent actinide cations, M** and N**, between a binary MO,-NO, solid solution and a molten fluoride solution containing MF, and NF,; was discussed in terms of the exchange equilibrium constant M :XM02XNF4 MO, N XNo,XMF, TNO, This constant and the activity coefficients, ypmo, and YNO,>, Wwere (see previous section) represented by expressions [Egs. (4), (5), and (7)] which involve the lattice parameters of the oxides, the temperature, and two empirical constants (4 and B). In an MSBR fuel, three tetravalent actinide ions — Th**, Pa*", and U* — will be present under normal conditions. The oxide phase which could be precipitated inadvertently by oxide contamination or intentionally for the purpose of fuel reprocessing will, therefore, be a ternary ThO,- Pa0,-UO, solid solution. In order to employ the equ111br1um constants KP?I and KUh (or KU , which is equal to K /KP ) to determine the composmon of this oxide phase as a function of the fuel composition, it is first necessary to estimate activity coefficients in such a ternary oxide solid solution. This was done as follows: Again it is assumed that such a solid solution has an ideal entropy of mixing. The problem of estimating activity coefficients then reduces to estimating the heat of mixing of the three components, since SAH,, on: i RTInvy,; = (1) The heat of mixing of n, + n, + n3 moles of the three oxides can be approximated by imagining that first the following binary solutions are made: ny tysand n, + (n3 — y3), y3 being a portion of n3, chosen such that these two solid solutions have equal lattice parameters. Assuming that the lattice parameter -of each solution varies linearly with mole fraction (Vegard’s law), the value of 3 which will yield the same lattice parameter for both solutions is given by (ay —ax)n, +(a, —as)n, (@ —az)ns +(ay; —az)n,’ Y3 =hy 2) 75 where the a’s denote the lattice parameters of the pure K22 = exp [3784} KU oxides. Th ~ T Th From Hietala’s model® it is expected that the heats of 5473 U 1689 mixing for each of these solutions will be, to a good = exp [—T—J Kpa = [——J , approximation, N YTho, = eXP{(Xpa0, * Xuo,) AH[ny + 93] =A n\ys , 3) 2 { 2 U0, o CIARE « [397 633 58 [7: Pa0, * —T‘XUOJ — 7 XPa0,Xv0, AH[n, + (n3 — y3)] =A (X2 + X3) (@, —a2)? X salt containing RTa 1 X =0.12, X =0.003, X =0.0003 , +@r —a3) Xa] — (@ —a3)* XaX5] 5 (7) Thta VUra Fata the oxide phase at equilibrium with it (at 600°C) will analogous expressions are obtained for the activity have the composition coefficients vy, and v;3. - - - Introducing the numerical values of 4 and B deter- X1ho, =0.036. Xyo, =095, Xpso, = 0013 mined in the previous section and the lattice parameters ~ This example serves to show that the oxide solid for ThO,, Pa0,, and UO,, solution precipitated from an MSBR fuel will contain little protactinium. Specifically, the ratio of Pa/Uin the oxide phase should be approximately one-seventh of A = 2440 kcal/mole; B = 2620 kcal, A, mole ™ ; that in the salt phase. For oxide solutions so dilute in aTho, = 5.597 Asapao, = 5.509 Asayp, = 5.4704 &, PaO, the values of yyo, and y1ho, (Fig. 9.4) will not be significantly different from those already determined for binary ThO,-UQO, solutions.” The estimates of we may write the following expressions for the various YPa0, for low concentration of Pa0,, varying from 1.0 exchange equilibrium constants and activity coefficients to ~1.4, ought to be accurate enough for present for ternary ThO,-Pa0,-UO, solid solutions: purposes. ORNL-DWG 72-7589 20N - [ 3 &(gfé Pa0, 7 £l //f/// - = 76 Fig. 9.4. Estimated activity coefficients in ternary oxide solid solutions of ThO,, Pa0,, and UO, at 600°C. 10. Development and Evaluation of Analytical Methods for Molten-Salt Reactors A. S. Meyer 10.1 IN-LINE CHEMICAL ANALYSIS OF MOLTEN FLUORIDE SALT STREAMS J.M.Dale A.S.Meyer Automated analyses for U(1II) in LiF-BeF,-ZrF,-UF, (65.4-29.1-5.0-0.5 mole %) in the NCL-21 thermal convection salt loop were continued. The percent of the total uranium present as U(III) for the first 3600 hr of loop operation is shown in Figs. 10.1—10.3. These analyses were largely made by unattended operation of the computer-controlled voltammetric system described in the previous report.’ The first analyses, made 70 hr after the salt was loaded into the loop, showed that the concentration of 1. J. M. Dale and A. S. Meyer, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL4728, p. 69. U(11l) was about 0.02%. The U(IIl) concentration started increasing as the chromium from the metal loop dissolved into the salt and reduced the U(IV). The irregular appearance of the data at 600 hr was due to large temperature fluctuations in the electrode tank. The first measurements with the shielded electrode were made at 1200 hr and are plotted as squares in Figs. 10.1 and 10.2. The shielded electrode is an assembly designed to eliminate the interference of material from the surface of the melt which apparently deposits on the electrode and changes its area and electrical characteristics. The electrode is surrounded by an open-ended nickel tube which can be periodically purged with helium to provide a clean melt surface. At about 1250 hr it was first noted that analyses with the shielded and unshielded electrodes gave different values for the U(III) concentration. This was later ORNL- DWG. 74-14377R —_ T "7 T L ]7 T l T [ 1 030 - B 14 0.25 -3 I ] 5 i ] 2 o020l ] + o 4 [19] . - | - 4 S o4s5f o = T L _ o N E ] & o010l Fow ] _ fa’ : A ] C Pa ] 0.05- Mj"‘#‘"‘ — i o ] L wad "‘WM A 0 i i | 1 | \ ] | ] ) | 0 200 400 600 800 1000 1200 LOOP OPERATION, hours Fig. 10.1, U({II) in MSRE fuel salt, 7/20/71 to 9/8/71. 77 shown to be an effect of the nickel tube around the shielded electrode and not to a difference in the platinum electrodes. At 2350 hr the nickel tube shield was raised out of the melt, and the measurements with the two electrodes were in agreement as shown in Fig. 10.2. Fortunately, the surface material has apparently been removed from the melt by continuous operation of the loop, and the unshielded electrodes are operating satisfactorily. It will probably be necessary to use shielded electrodes for other systems, however, and 78 further investigation is needed to determine the reason for the difference in the two electrodes. A possible explanation is that the difference results from a cooling of the melt by the shield tube. During the first 3600 hr of loop operation the Hastelloy N corrosion specimens were inserted into the loop at three different times (140, 1316, and 3200 hr).2 Except for the first time, when the U(III) 2. J. W. Koger, Sect. 13.5.1, this report. ORNL-DWG. 71-14378R T T + 0.30 0.25 N R % @ 8 0.20 PER CENT U>7Uy o o o o 0.05 lllllllllllllfllll[[ll 1 I I I 5%:*# + tE e+ o+ o4 I T [ ! I T llllllllll.l]llll + o, o+ lllJIlIIJlllJlI 1 l L i L 0 1200 1400 1600 | 1800 2000 2200 2400 LOOP OPERATION, hours Fig. 10.2. UQII) in MSRE fuel salt, 9/8/71 to 10/28/71. DWG 0.30 0.25 o N (@) W 0.15 PER CENT U3*/U; O o ll[l]lII]IIllI]II+TI|IITII|IIIITI o) 1 | | | ] ORNL- . 71-14379R T T A { Illl[llll]llllllIllllllllllLlIJl | 1 I 1 | 1 2600 2800 3000 3200 3400 LOOP OPERATION , hours Fig. 10.3. U(II) in MSRE fuel salt, 10/28/71 to 12/17/71. concentration was already at a low value, the insertion of the metal specimens appeared to introduce an oxidant which caused a decrease in the U(III) concen- tration. Special precautions are being taken for the next specimen insertion to determine if this operation can be performed without oxidizing the U(III). At 1725 hr the shielded electrode, which had previously been removed for examination, was put back into the loop without immersing it into the salt. As shown in Fig. 10.2, the U(ITI) concentration decreased and then rapidly re- covered, probably due to dilution effects caused by the salt flow. The voltammetric technique will also be used to determine other melt constituents such as corrosion product ions. In the reducing melts that have been present in the loop the equilibrium concentration of Fe(ll) and Ni(Il) is negligible, while all of the chromium is present in the ionic form. In fuel melts, Cr(Il) is the most difficult of the corrosion products to measure because its reduction potential is so close to that of U(IV) (see the next section of this report). One of the proposals which we have made but not demonstrated is that chromium can be determined by a stripping technique. This technique is based on the fact that metallic chromium, the reduction product of Cr(lI), is deposited on the electrode, whereas U(III), the reduc- tion product of U(IV), is soluble and diffuses back into the meit. Thus if the electrode is held at a potential that is sufficiently cathodic to reduce both U(IV) and Cx(11), the U(III) will rapidly approach a steady-state concen- tration at the electrode surface, while chromium metal is continuously deposited. When the electrode is elec- trolytically stripped by an anodic scan, the contribution of U(Ill) should be independent of reduction time, while the stripping current from the chromium metal should increase with plating time. Figure 10.4 shows some typical stripping curves for chromium at the 100-ppm level in the salt. Chromium was plated on the working electrodes for different lengths of time at —360 mV vs £, and then stripped from the electrode at a scan rate of 0.1 V/sec. When these stripping currents were integrated with the volt- ammeter and plotted vs plating time, an excellent straight-line relationship was obtained. Extrapolation to zero plating time showed that about 4.5 mC of the current, a reasonable value, was due to reoxidation of U(III) to U(IV). In theory, the slope of the curves of integrated current vs plating time should be propor- tional to the concentration of chromium ion in the melt. Although gradual increases in the slope of such curves which may correspond to increases in the concentration in the melt have been observed, we have 79 ORNL-DWG. 71-14374R f T T T 71 T _Ij Cr = 106 ppm Electrode Areo = 0.418 cm? 24|— e} E 18T_ E. [w) g S:J 15(‘ . r o = 32 chr:J I w a [ 3 1 oF G) Ll E s 3 [, ] | ] -0.4 Q 02 04 06 08 we VS EEQ , volts PLATING TIME, min Fig. 10.4. Stripping curves for chromium in MSRE fuel. not accumulated sufficient analytical data to establish the analytical validity of this technique. In summary, the results of the ratio measurements have, in general, shown excellent reproducibility and have been consistent with the known factors of loop operation. Although several problems (such as the offset in the potential of shielded electrodes, smaller offset voltages observed between the reference and working electrodes, and the establishment of accurate calibrations for the determination of corrosion prod- ucts) remain to be solved, our experience gives us considerable confidence that this relatively simple transducer system can be widely applied to other test systems and ultimately to in-line reactor streams. The measurements also provide an explanation for the lower-than-expected results obtained in attempts to apply this technique to the hot-cell measurement of U(IV)/U(I1T) ratios in MSRE samples.® Despite all our precautions, the contamination introduced during the sampling and transfer of the MSRE samples must have been at least equivalent to that observed during the introduction of corrosion specimens to the NCL-21 loop. It should be noted, however, that although these MSRE analyses provided little relevant data as to the ratio in the fuel, they indicated no evidence for interference to the method from the radioactivity of the samples. 3. J. M. Dale, R. F. Apple, and A. S. Meyer, MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL4548, p. 180. 10.2 THEORETICAL CONSIDERATIONS OF THE VOLTAMMETRIC IN-LINE DETERMINATION OF URANIUM(III) J. M. Dale The present method for determining the uranium(111) concentration in MSR fuel involves the measurement of the potential difference between the equilibrium poten- tial of the melt and £, /5, the voltammetric equivalent of the standard potential of the U(IV)/U(III) couple.* The equilibrium potential of the melt, EEQ, is the potential of an inert platinum wire immersed in the melt and depends upon the U(IV)/U(III) ratio. The standard potential, £y /5, is the potential on the U(IV) reduction wave at which the concentrations of U(IV) and U(III) are equal. The relationship between Egq, E 15, and U(IV)/U(II) is given by RT U(1V) Even =E,, +—In —-=. EQ 712 T pF T un) Because the voltammetric circuit uses E, as a refer- ence potential, we can consider it to be zero and _RT_UQIV) “E2 T E Uy Evaluation of £, ,, with respect to Egq then permits the U(IV)/U(11I) ratio to be calculated. Nicholson and Shain® have reported a solution to the boundary value problem that can be applied to this system. They found for a reversible couple where both species are soluble that Ey,, corresponds to the potential on a theoretical reduction wave where the current is 85% of the peak current. The experimental U(V) reduction wave, however, from the NCL-21 thermal convection loop also involves the reduction of chromium, which could adversely affect the potential at which the 85% point occurs. For this reason it was of interest to compare the theoretical reduction wave with the U(IV) reduction wave from the experimental system. The potential, E, of the working electrode during a voltammetric scan is represented by (E = Eyj2)n = (RT/F) (In¥0 — ar), 4. H. W. Jenkins, D. L. Manning, G. Mamantov, and J. P. Young, MSR Program Semiannu. Progr. Rep. Feb. 28, 1969, ORNL-4396, p. 201. 5. R. S. Nicholson and I. Shain, A nal. Chem. 36, 706 (1964). 80 where Y= VDo/Dr 5 0 = exp [(nF/RT) (E; — E | )))] , at =nfvt/RT . The term 0 is a constant, where D is the diffusion coefficient of the oxidized or reduced species and £ is the initial starting potential of the voltammetric scan. The term at is a variable, where v is the voltage scan rate and ¢ is the time elapsed after the scan is started. All other terms have their usual significance. The current, i, through the working electrode during a voltammetric scan is represented by i=nFAC, \/nDya x(at) . The term x(at) is a function of at, 4 is the electrode area, and C, is the concentration of the U(IV) in the melt. The expression k — 2\/5{x(1)\/7€_+ L‘l vk =i [x(@i+1) — x(i)]} i=1 = 1/[1 + v0 exp (—a?)] , where & = (at)/k, defines the relationship between the values of chi and the variable at, where k is the serial number of the particular value of chi being evaluated. In order to get a true theoretical current-potential curve it is necessary to solve for the values of chi at small voltage intervals, as each successive value of chi depends upon all of the previously determined values. A program was written for the PDP/81 for the numerical evaluation of the chi values over an 800-mV range at 2-mV intervals. This gave § a value of 0.025. The initial potential, £, was 500 mV anodic to £, /2 making In 0 =6.29. Evaluation of the current expression, where A 0.418 cm?, C, = 0.282 mole/liter, Dy = 5.62 X 107° cm?/sec, and v= 0.1 V/sec, gives i =53.69 x(at) mA for the potential defined by at. The derivative current expression is di/dt = Ai/ At = 53.69 Ax(at)/ At mA/sec, where the values of Ax(at)/Ar are the incremental slopes between the 2-mV intervals of the successive 81 ORNL- DWG, 71-14376R T T T U%*t= 0.282 moles /liter Electrode Area = 0.418 ¢cm? D=5.62 x 1075 cm? / sec Scan Rate = O volt /sec - [T [MEs Temp. = 650°C - _é o Z & o - E — =z ~ L wl 2 g © i ~ -’ = O el | i [ ] 1l . ] ] c -01 02 -03 -04 -05 -06 -0.7 -0.8 -09 £ volits we VS Egq s Fig. 10.5. Reduction of UIV) in MSRE fuel. values of chi. By proper substitution of the values of x(at) and Ax(at)/At, these two expressions were used for the construction of the theoretical current-potential curve and its derivative. In Fig. 10.5 the experimental waves are represented by the solid lines, and the theoretical waves are represented by the points plotted at 10-mV intervals. There is the expected deviation in the region where Cr(II) is reduced (—0.4 to —0.5 V), both on the normal and derivative waves. The experimental derivative peak is lower because the Cr(Il) reduction causes the normal wave to rise less sharply in this region. However, the potentials at which the two derivatives show a maxi- mum agree to within 5 mV, and there is good agreement between the experimental and theoretical values of Ey,. It is concluded that the chromium reduction does not affect the current-potential curve for the U(IV) reduction at the £y, potential and that the method for determining £, /5 is valid. 10.3 ELECTROANALYTICAL STUDIES OF TITANIUM(IV) IN MOLTEN LiF-BeF,-ZrF, (65.4-29.6-5.0 MOLE %) F.R.Clayton® D.L.Manning Gleb Mamantov’ Since reporting on the voltammetry and chrono- potentiometry of titanium(IV) in molten LiF-NaF-KF 3 we have continued these studies in molten LiF-BeF,- Z1F,. The initial voltammetric studies were undertaken using K, TiFg4 as the solute; however, the results of titanium(IV) reduction were greatly complicated by the observed instability of Ti(IV) in the melt. White deposits collected in the cooler part of the electrolytic system after K,TiF; was added to the melt. The volatilization of titanium tetrafluoride from the melt was confirmed in a separate experiment in a closed stainless steel vessel provided with a cold trap; the volatile product collected was identified as TiF; by x-ray analysis. Voltammetric studies of the oxidation of tita- nium(II) in molten LiF-BeF,-ZrF, at 500° were then initiated, since TiF; was expected to be stable in the melt at this temperature. (The sublimation point of TiF5 is 930° in vacuo.) Well-defined linear-sweep voltammograms for molten LiF-BeF,-Z1F, containing Ti(Ill) were obtained at a sheathed glassy carbon electrode. Similar but less-weli- defined waves were obtained at an unsheathed platinum electrode. The wave corresponds to the anodic oxida- tion of Ti(Ill) to Ti(IV). This was the only wave observed upon addition of TiF; within the potential limits of the melt, +1.5 to —1.5 V measured at the platinum vs nickel(Il)(saturated)/nickel reference elec- trode (LaF; membrane type). These potential limits correspond to anodic dissolution of platinum and the reduction of Zr(IV) respectively. A plot of i, vs vl/2 resulted in a straight line which is indicative of a simple charge transfer reaction. The concentration dependence of i, was also linear. The standard electrode potential £° for the Ti(IV)/Ti(III) couple may be estimated from the £, for the process Ti(II[) = Ti(IV) + e in molten LiF-BeF,-ZrF, at 500°C. The E |, is +0.380 V with respect to the Ni(Il)(saturated)/Ni(ll) reference elec- trode. However, an extrapolation of a Nernstian plot of the nickel couple to unit mole fraction of nickel(II) gives a potential that is 183 mV more anodic than the potential corresponding to the saturation point of nickel(Il). Applying this correction of —183 mV to relate the potential of the reference electrode to the potential of a unit fraction Ni(II)/Ni couple, £° for the Ti(AV)/Ti(III) couple in molten LiF-BeF,-ZrF, at 500°C is estimated at +0.197 V (vs a unit mole fraction Ni(II)/Ni electrode). The chronopotentiograms recorded at a sheathed glassy carbon electrode were reasonably well defined. 6. Student Participant, University of Tennessee, Knoxville, 7. Consultant, Department of Chemistry, University of Tennessee, Knoxviile. 8. F. R. Clayton, D. L. Manning, and G. Mamantov, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL 4676, p. 134. The product jo7!/? at current densities from 0.04 to 0.16 A/cm? was found to be reasonably constant at 0.060 + 0.002 A sec'/? cm™. Transition times were in the range 0.1 to 2.3 sec. Further verification of n = 1 for the Ti(Ill) oxidation was accomplished using the ratio of the voltammetric ip/vl/ 2 to the chronopotentiometric i7!/2. This rela- tionship for the determination of » is given at S00°C as ip/Vl /2 =1.96n'/? 1/2 An n value of 0.9 was determined by this method, which is independent of other parameters such as concentration of the electroactive species, diffusion coefficient, and electrode area. For chronopotentiometry, the half-wave potential for a reversible charge transfer corresponds to the potential at one-fourth the transition time. This value was found to be about +0.40 V vs an Ni(II)(saturated)/Ni refer- ence electrode, and agrees reasonably well with the +0.380-V voltammetric value. From our voltammetric and chronopotentiometric studies, we believe this elec- trode reaction conforms to a reversible Ti(Ill) - Ti(IV) oxidation process and that this reaction could be used to monitor any buildup of titanium(IIl) in a flowing salt stream. 10.4 ELECTROCHEMICAL STUDIES OF BISMUTH(III) IN MOLTEN LiF-BeF,-ZrF, AT 500°C J.S. Hammond® D. L. Manning Bismuth(1II) being of importance in fuel stream purification systems is, therefore, a possible impurity in the reactor fuel salt. Voltammetric and chronopoten- tiometric studies were initiated to characterize the reduction behavior of this substance. The linear sweep voltammograms were well defined for the reduction of bismuth at pyrolytic graphite, platinum, iridium, and silver indicator electrodes. Re- verse scans indicated possibly alloy formation between bismuth and platinum and also between bismuth and silver. No evidence of alloy formation was observed at pyrolytic graphite and iridium electrodes. Of the indicator electrodes tested, the results were most reproducible at the iridium electrode. This material appears to be an excellent inert electrode for electro- analytical studies in corrosive melts. 9. GLCA Student, Denison University, September—December 1971. Granville, Ohio, 82 The peak-shaped voltammogram corresponds to the reduction of Bi** to metallic bismuth at approximately —0.05 V vs a platinum quasi-reference electrode. Plots of iy, vs v!/2 were linear to about 5 V/sec, with slightly downward curvature at the faster scan rates. This suggests that the electrode reaction is becoming quasi- reversible at the faster scan rates.'® The concentration dependence of i, was linear over the concentration range ~10 to 200 mM Bi(III). It was also discovered that for appreciable periods of time (days), stable solutions of Bi(Ill) in molten LiF-BeF,-ZrF4 could not be maintained in either graphite or copper cells. However, the bismuth solutions appeared to be more unstable in graphite. The mechanism for the bismuth instability is not yet resolved. Chronopotentiograms recorded at an iridium elec- trode (~0.1 cm?) were well defined. The ratio of forward to reverse transition times was unity, indicating the reversible deposition of an insoluble substance.'! The values for the diffusion coefficient evaluated from voltammetry and chronopotentiometry were 1.08 and 1.03 X 107 cm?/sec respectively. Verification of n = 3 for the bismuth reduction was carried out using the ratio of the voltammetric ip/v” 2 to the chronopotentiometric io7!/2. This equation for the determination of n at 500°C is given as cu1/2 ',”l— =2.69n'/? . igT1/2 An n value of about 2.9 was obtained, which is independent of other parameters such as the concen- tration of the electroactive species, diffusion coeffi- cients, and electrode area. 10.5 VOLTAMMETRY OF CHROMIUM(IHI) IN MOLTEN NaBF,-NaF (92-8 MOLE %) D. L. Manning A voltammetric study is under way on the reduction characteristics of chromium(lll) added as Na;CrFg4 to molten NaBF4-NaF. The melt is contained in a graphite cell enclosed in a nickel apparatus to maintain an inert atmosphere. A cover gas of helium is maintained under static conditions at approximately 5 psi. Enough Na;CrFg was added to give a chromium(III) concen- tration of approximately 410 ppm if all the reagent dissolved. Chemical analysis of melt samples taken at 10. Paul Delahay, J. Phys. Coll. Chem. 54, 630 (1950). 11. W. H. Rienmuth, Anal. Chem. 32, 1514 (1960). 440 and 500°C revealed a chromium(I1I) concentration of 200 and 300 ppm respectively. The chromium reduction wave (Cr** - Cr%) was observed at approximately —1.0 V vs a platinum quasi-reference electrode at platinum and palladium indicator electrodes. The waves were reasonably well defined; limiting current values at platinum (~0.1 cm?) were S00 and 1100 pA at 440 and 500°C, respectively, at a scan rate of 0.1 V/sec. Scan rate studies revealed that plots of peak current (4A) vs scan rate, (V/sec)'/2, were linear to about 10 V/sec. From the slope of the line, the diffusion coefficient of the Cr(III) species can be evaluated, and at 440°C was found to be about 2 X 1078 cm?/sec. Attempts to record chronopotentiograms were not successful; apparently the potential of the chromium reduction is too close to the melt limit to record meaningful transition times. Additional experiments will be carried out to ascer- tain the linearity of the peak current vs chromium concentration plots. 10.6 VOLTAMMETRIC AND HYDROLYSIS STUDIES OF PROTONATED SPECIES IN MOLTEN NaBF, D. L. Manning A.S. Meyer We are continuing our investigation of the electrolysis of hydrogen from NaBF, melts at evacuated palladium electrodes.'? For the reduction of hydrogen, most electrode materials yield ill-defined and ragged volt- ammetric waves characteristic of gas film formation. However, at the palladium electrode the deposited hydrogen rapidly dissolves into the electrode to elimi- nate the film formation and gives well-defined volt- ammograms. Moreover, if the electrode is held at a sufficiently cathodic potential, some of the deposited hydrogen enters the evacuated portion of the electrode to yield a measurable pressure. Both of these techniques offer promise for a sensitive method for the in-line determination of protons in the coolant salt, with the pressure measurement technique offering the advantage of specificity. During this period we have changed our experimental conditions by enclosing the melt in a nickel vessel as opposed to the quartz and Pyrex system that we usually use to protect molten fluoride salts. In both cases the actual melts are contained in a graphite liner under a static pressure of helium and make no direct contact 12. D. L. Manning and A. S. Meyer, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL4728, p. 74. 83 with the protective envelope. In contrast with the instability of the protonated species observed in the glass-enclosed systems,!'? the reduction waves in the metallic container were stable indefinitely in melts containing from 14 to 40 ppm hydrogen, according to infrared analyses. The effect of glass containers on the stability of these melts is surprising, because the only communication to the container walls is through the gas phase. This behavior is consistent with the observations that the absorption peak of the BF;0D™ ion faded during measurement on melts contained in quartz cells.?? We are now attempting to calibrate both of these types of measurements for analytical applications. This work is complicated by variations in the rate of diffusion through the electrode material and also by a low correlation of concentration between peak reduc- tion currents, assuming a reasonable diffusion coeffi- cient value and the analyses of the melts by the infrared pellet technique. Currently, we are investigating elec- trodes fabricated from silver-palladium alloys and are considering the possibility that the observed voltam- metric wave results from the reduction of only one of multiple protonated species in equilibrium in the melt. 10.7 DETERMINATION OF HYDROGEN IN NaF-NaBF, SALTS J.P. Young A.S.Meyer Based on a calibration factor derived from an intimate physical mixture of NaBF;OH and NaF-NaBF, powders,'* infrared (IR) determinations of protons, as NaBF3;O0H, in samples of NaF-NaBF, salts have been carried out by J. R. Lund (Analytical Chemistry Division). It can ordinarily be expected that standard additions of a species of interest to a sample matrix are a perfectly reliable way to generate samples for a calibration curve; however, the labile nature of NaBF3;OH with respect to even moderate increases in temperature,' > coupled with the expected instability of protons in any form in molten NaF-NaBF, !¢ called for rather stringent tests to obtain independent proof that 13. 1. B. Bates, H. W. Kohn, J. P. Young, M. M. Murray, and G. E. Boyd, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 96. 14. J. P. Young, J. B. Bates, M. M. Murray, and A. S. Meyer, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL- 4728, p. 73. 15. L. Kolditz and Cheng-shou Lung, Z. Chem. 1, 469 (1967). 16. S. Cantor and R. M. Waller, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 80. the IR method gave reliable results. The first two reported attempts to verify the IR method by the mass spectral method using D, as an isotopic diluent'* were subject to an exorbitant blank which resulted from D, exchange with OH™ in the Pyrex experimental appa- ratus. Some or all of the measured hydrogen attributed to the sample may have resulted from increased exchange of the OH™ in the Pyrex ampuls due to etching. The only eutectic salt samples available at that time contained 20 to 30 ppm protons (IR); the only significantly lower proton concentration was in a sample of recrystallized NaBF,. It is conceivable that a fortuitous difference in etching rates between the two samples could have yielded the excellent agreement obtained. During this period these Pyrex blanks have been reduced from 8 micromoles in a typical sample to 4 micromoles of H, (these values correspond to 10 and 5 ppm protons respectively). This reduction in blanks was accomplished by prolonged heating of the Pyrex ampuls at a temperature near their softening point in-a stream of helium prior to the addition of any sample. Data collected on the correlation of mass spectral and IR determinations of protons in NaF-NaBF, samples are given in Table 10.1. Most of the results of this comparison are in acceptable agreement between the two methods. Occa- sional high results by the mass spectral method are not unexpected in view of the experimental difficulties of handling of samples and ampuls. Moreover, no sys- tematic variation is observed with improvements in the blank. We therefore conclude that the results of the IR analyses do indicate that a significant concentration of hydrogen is present in the samples submitted from corrosion loops. Precise confirmation of the IR calibra- Table 10.1. Determination of hydrogen in NaBF 4 samples by infrared and mass spectral methods H, blank Hydrogen found (ppm) Sample . (micromoles) Mass spectral Infrared NaBF, 8 84 74 NaF-NaBF, 8 224 244 S 20 16 S 27 22 6 31 10 4 7 9 8 b 28 12 aReported previously: J. P. Young, J. B. Bates, M. M. Murray, and A. S. Meyer, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 73. bBlank lost. 84 tion curves will require additional comparisons with samples of low or no proton content and with samples in which standard additions of protons have been made. From further experimental work with SiO, ampuls, we have concluded that they are unsatisfactory due to excessive diffusion of H, through the walls of the ampul. From the results given in Table 10.1 it can be noted that the proton concentration, by IR analysis, varies from 9 to 24 ppm. Until this period, not much variation was seen in the proton concentration, particularly low concentrations, in any NaF-NaBF, sample that had been melted. The 7-ppm value reported last period was for purified NaBF, that had never been melted. During this period, attempts have been made both to obtain melts of very low and relatively high proton (as BF3;O0H") concentrations. In this respect, by carefully controlling experimental conditions and proper pre- treatment of apparatus, it has been possible to prepare melts of NaF-NaBF, that had 6.5 to 7 ppm protons, based on IR analysis. Under our experimental condi- tions we have not yet been able to prepare melts of lower proton content. For several reasons we desire a melt which has no detectable proton concentration, and this work is continuing. Attempts have been made to make standard additions of protons to NaF-NaBF, melts for use in the analysis evaluation. Using melt containers that are in an inert atmosphere, but exposed to thermal gradients within the system, apparently only 25%, in the best cases, of the protons added, as HBO,, NaOH, NaHSO,4, or KHSO,, remain in the melt. Containers which will be sealed and isothermal are being designed for further standard addition work. In summary, from our work it would appear that the IR analyses given by the analytical service laboratories are reasonably accurate for the pellet as pressed for the spectral measurement. Whether this value is the same as the proton concentration when the sample was molten remains to be proven. Certainly the IR analyses show the variation in proton concentration in the direction expected for a variation of a given experimental parameter. In line with the pellet work, we are also engaged in IR spectral studies of NaF-NaBF, melts in cooperation with J. B. Bates and G. E. Boyd (Director’s Division). This work is reported in Chap. 7. 10.8 SPECTRAL STUDIES OF MOLTEN SALTS J.P. Young Miscellaneous spectral studies of solute species in molten LiF-BeF,-based solvents and molten LiCl have been carried out in cooperation with, and as an aid to, personnel from other divisions who are studying melts of interest to the program. Several techniques have been devised; several apparatus modifications have been designed, and some built; and both new and previously obtained spectral data have been applied to these areas of investigation. Based on the spectral results from Pa(IV) in molten BeF ,-based solvents obtained with the hot-cell spectro- photometer,'? the spectral determination of Pa(IV) in equilibrium with Pa(V) and Fe(Il) offers a method for determining the potential of the Pa(IV)-Pa(V) couple in MSR solvent. Knowledge of this couple is important in assessing a possible fuel reprocessing scheme involving the precipitation of Pa, Q5. In cooperation with C. F. Baes, C. E. Bamberger, and R. G. Ross (Reactor Chemistry Division), we plan to undertake such a study. The ability to use small (~300-mg) samples in the spectral technique has led to the proposal and approval to perform these experiments with palladium out of the hot cell. Techniques for sample transfer and treatment are much simpler if the experiment can be performed in the laboratory. A new sample transfer apparatus has been designed and will be constructed which will ensure no contact of the palladium sample with the laboratory atmosphere when it is transferred from the palladium glove box to the spectral furnace. The same general design will likewise ensure transfer of other samples from inert-atmosphere boxes to the spectral furnace without any exposure of the sample to the atmosphere. The transter container makes use of two Cajon O-ring fittings arranged so that one O-ring seals the bottom of a sample spectral cell, windowed or windowless. The salt sample is within the cell, and the cell is attached to a rod within a cavity that serves as the transfer container. The rod protrudes through and is sealed by the second O-ring fitting. The bottom of the transfer container can be sealed to the spectral furnace in the normal fashion. After the transfer container with a sample, loaded in a glove box or inert-atmosphere box, is sealed to the spectral furnace, the sample can be introduced into the furnace by loosening the two Cajon fittings; they will remain gas-tight during this process, however. For possible use in the palladium experiment as well as in other experiments, it is advantageous at times to bubble gases into melts used in spectral studies. It has previously been thought that gases could not be bubbled through liquid contained in a windowless cell, because the agitation would cause the liquid to run out of the cell. By using tubing of small inside diameter 17. 1. P. Young, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 114. 85 (e.g., 0.006 in.) and outside diameter (<0.062 in.), it has been found that gas can be passed through liquids without spillage. At least 4 ml of helium gas per minute can be bubbled through 0.1 to 0.2 ml of liquid contained in a windowless cell. Although liquid in the cell is agitated and stirred by the gas contact, the stability of the total liquid shape is unaffected because of the small size of the bubbles and because the bubbles tend to rise in a region immediately surrounding the purge tube. This procedure has been checked with both water and molten LiF-BeF, in windowless cells. In experiments to date, it is noted that the tip of the bubbler tube plugs within 1 hr if helium gas is bubbled through molten LiF-BeF, ; the reason for the plugging is unknown, but several explanations are possible. Further work is planned to bubble HF-helium and/or H, -helium mixtures through melts to establish a particular redox level for pretreatment of melts. The use of these gas mixtures may also prevent tip plugging. In cooperation with C. E. Bamberger and C. F. Baes, the solubility of CuO in LiF-BeF, was studied in a silica cell by a spectral study of the dissolved Cu(Il). After several days equilibration at $25°C the intensity of the Cu(ll) spectral peak at 880 nm remained constant. Based on the molar absorptivity found by Whiting'® for Cu(Il) in LiF-NaF-KF, approximately 2000 ppm Cu(ll) dissolved in the melt. We hope to expand and combine work of this type with the thermodynamic data of Baes and Bamberger for the system BeF,(d) and Si0, (s) to study CuF, and CuF in LiF-BeF, melts. The possible chemical reactions of copper containers with various fluoride melts seem to be a fruitful and partially neglected area of investigation. Several spectral studies have been carried out with personnel in L. M. Ferris’ group (Chemical Technology Division), in connection with fuel salt makeup and reductive extraction. Several attempts were made to transfer, melt, and spectrally observe a sample of U(V) in MSR salt prepared by M. R. Bennett by reacting UF ¢ with U(IV) in the melt. Only a weak peak was seen at 1465 nm, the wavelength of a sharp U(V) absorbance peak, and this peak was only seen if the samples submitted by Bennett were loaded in gold-plated cells. No spectral indication of U(V) was observed if the samples were melted in graphite or copper cells; rather, in these cells as the sample melted, there was visual indication of gas bubbles and resultant melt agitation, with the melt immediately turning green and exhibiting the spectrum 18. F. L. Whiting, G. Mamantov, and J. P. Young, submitted for publication in Journal of Inorganic and Nuclear Chemistry. of U(1V). In previous work with G. I. Cathers (Chemical Technology Division), spectral studies of U(V) in LiF-BeF, melts were undertaken; the solutions were prepared by adding Na, UFg to the molten solvent in a graphite windowless cell. During this period this solu- tion preparation was repeated. Once the U(V) solution was prepared, the melt was frozen and then remelted. On remelting, no evidence of the U(V) spectrum was seen. These results suggest that something in the thermal cycling of a sample containing U(V) may cause a loss of U(V). Spectra of solutions of LizBi in molten LiCl have been observed during this period. If a sample is melted directly in a molybdenum windowless cell under our experimental conditions, the melt immediately de- colorizes. However, if the sample is melted in the presence of some lithjum-bismuth alloy, there is an initial loss of absorbance for several hours; then the spectrum of the yellow melt is fairly constant. The spectrum we see for LiyBi in LiCl is similar to that 86 reported for LizBi in LiCl-LiF melts and consists mainly of a strong absorbance in the ultraviolet region; we do see a shoulder at 330 nm, however, which was not previously reported.!® It is probable that we are observing the spectrum of <150 ppm Li; Bi in the melt. Some preliminary work with related species has also been carried out. L. M. Ferris’ group has provided samples of LiCl which have been in contact with lithium-lead alloys. No positive results have been obtained as yet, but it is interesting to note that, analytically, much more lithium is found in the melt than would be accounted for in the compound LisPb. The spectrum of lithium was not observed in these melts; however, in an earlier study of lithium in pure LiCl;,%2° no measurable absorbance was observed at similar concentrations. 19. M. F. Foster, C. E. Crouthamel, D. M. Gruen, and R. L. McBeth, J. Phys. Chem. 68, 980 (1964). 20. J. P. Young, J. Phys. Chem. 67, 2507 (1963). 11. 11.1 THE OXIDE CHEMISTRY OF NIOBIUM IN MOLTEN LiF-BeF, MIXTURES Gann Ting" C.F.Baes,Jr. G.Mamantov? The chemistry of niobium in molten fluorides is of interest to the MSBR program for several reasons. In the MSRE it was observed that the appearance of fission product ?SNb in the fuel seemed to be a sensitive function of the state of oxidation (the U**/U%* ratio) of the fuel.> This effect, which presumably involves the oxidation of the metal to a lower valence state in solution, might be a useful indicator of the state of oxidation of the fuel. Niobium pentoxide (Nb, Os), like protactinium pentoxide (Pa, Os), is expected to be sparingly soluble in molten fluorides, and it has been proposed that Nb(V) might be used as a stand-in for Pa(V) in studies of fuel reprocessing methods involving oxide precipitation. There may be some important differences, however, between the chemistry of Nb(V) and that of Pa(V). It seems likely that Nb(V), unlike Pa(V), forms oxyions such as NbOF,*"3)~ in molten fluorides.* In the system NiO-Nb,Os,%°® at least two intermediate compounds, NiNb, Og and NigNb, O, are known, and since NiQ is also a sparingly soluble oxide, precipitation of these nickel niobates may be expected to complicate the chemistry of Nb(V) in the presence of 0%~ and Ni** ions. Weaver et al.” equilibrated niobium metal and a lower-valent niobium fluoride in Li, BeF; with hydro- gen. Their results suggested the reaction 4HF(g) + Nb°(c) = NbF,(d) + 2H,(g) , (1) Q1 = XNvFy P, /Pur® ~ 1012 (500°C). Senderoff and Mellors® report potentials, obtained chronopotentiometrically, for the formation of Nb(l), Nb(IV), and Nb(V) in molten LiF-NaF-KF (46.5-11.5-42 mole %) at 750°C. These indicate, in agreement with the results of Weaver and Friedman, that: Nb(IV) should be stable in the presence of nickel 87 Other Molten-Salt Researches or nickel-base alloys, Nb(l) should disproportionate to the metal and Nb(IV), and Nb(V) should oxidize nickel to form appreciable concentrations of NiF, in solution. The purpose of the present study has been to explore further the chemistry of niobium in molten fluorides by means of equilibria involving oxide phases of nio- bium(V). Two series of equilibrations have been com- pleted thus far. In the first series, performed in a nickel vessel, NiNb, O¢ and NizNb, Oy were formed. In the second, performed in a graphite-lined vessel under circulating mixtures of CO and CO,, the equilibrium solid phase was Nb, Os. We will describe the second series of measurements first. 11.1.1 Equilibrations of Nb, O and BeO with Molten LiF-BeF, Mixtures The equilibrations were carried out in a welded cylindrical nickel vessel (2% in. in diameter, 12 in. long) with a graphite liner. Initially, 500 g of LiF-BeF, (67-33 mole %) which had been purified by the usual HF-H, treatment was placed in the vessel. Beryllium oxide (2.41 g) was added along with 3.59 g of 1. Alien Guest from Institute of Nuclear Energy Research, Republic of China. 2. Consultant from the University of Tennessee. 3. R. E. Thoma, Chemical Aspects of MSRE Operation, ORNL-4658, pp. 94—99 (December 1971). 4. J. S. Fordyce and R. L. Baum, J. Chem. Phys. 44, 1166 (1966). 5. H.J. Goldschmidt, Metallurgia 61,211 (1960). 6. E. V. Tkachenko, F. Abbattista, and A. Eurdese, [norg. Mater. (USSR) 5, 1671 (1969). 7. C. F. Weaver and H. A. Friedman, MSR Program Semi- annu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp. 124-25; C.F. Weaver, H. A. Friedman, and J.S. Gill, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622,p.71;C. F. Weaver and J. S. Gill, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL4676, pp. 85—86. 8. S. Senderoff and G. W. Mellors, J. Electrochem. Soc. 113, 66 (1966). B-Nb, Os,” labeled with ®>Nb. Agitation of the molten mixture was provided by means of a vigorous flow of CO-CO, or CO-CO,-Ar gas mixtures through the melt. The gas was recirculated by means of a finger pump acting on a length of flexible Tygon tubing. The gas circuit also included a copper filter, a copper cold trap for collecting any gaseous Nb(V) species which might be evolved, and an infrared (IR) cell for examination of the circulating gas. Filtered samples of the melt were taken with copper filter sticks (25 to 50 u) as a function of time in order to follow the approach to equilibrium. The time required for equilibrium varied from 20 to 100 hr, decreasing with increasing tempera- ture. Samples of the equilibrated oxide phases were occa- sionally collected by suction on the tips of filter sticks. After washing with hot water to remove the solvent salts, these oxides were examined by various methods. Emission spectroscopy revealed beryllium and niobium to be the only metallic elements present in appreciable amounts. X-ray powder diffraction showed BeO to be present in all samples, -Nb,Os in two early samples, and y-Nb,O5° in samples taken near the end of the series; the other samples gave an unidentified pattern, perhaps for an unreported polymorph of Nb,Os. The scanning electron microscope (SEM) showed well- formed crystals containing major amounts of niobium. From this evidence we conclude that BeO and Nb, O were the equilibrium phases. Only traces of niobium were found deposited in the cold trap or on the KBr windows of the IR cell. This is consistent with the calculated'® equilibrium constant for the reaction 1/2 Nb2 05 (C) + 5/2 Ber(d) =NbFs(g) + % BeO(c), (2) log (Pypp,)=2.05—-11.76 (10°/T), which predicts that the partial pressures of NbF; generated in these equilibrations will be in the range 107'% t0 107%* atm. From the niobium content of filtered samples (Fig. 11.1) it appears that there was no effect of the CO,/CO ratio, which was varied from 0.1 to 1.8. Since variation of this ratio should have caused changes in the NbF, 9. A Reisman and F. Holtzberg, ‘“Nb, Qs and Ta,Os Struc- ture and Physical Properties,” p. 217 in High Temperature Oxides, Part 11, ed, by A. M. Alpov, 1970. 10. C. F. Baes, Jr., Reprocessing of Nuclear Fuels, Nuclear Metallurgy, vol. 15, 617 (1969), USAEC-CONF 690801. 88 ORNL-DWG 72-7595 -1000 -500 -200 | —100 ~ — € 2 Py s p x -50 2 : = {-20 5 F) ® INCREASING TEMPERATURE, RADIOCHEMICAL _—1 T ANALYSES ] o DECREASING TEMPERATURE, RADIOCHEMICAL -10 [ ANALYSES — A INCREASING TEMPERATURE, CHEMICAL 2 fi— ANALYSES — A DECREASING TEMPERATURE, CHEMICAL ANALYSES 10”6 L 1 | ] L 1 0.95 1.05 145 1.25 1000/, 0y, Fig. 11.1. Concentration of niobium(V) in LiF-BeF, mix- tures at equilibrium with BeO and Nb,Os under circulating CO,-CO mixtures. The numbers indicate the ratios Pcg, /Pco. content of the melt because of changes in the oxidation potential of the system (cf. reaction 3 below), the amount of NbF, present evidently was small. Since, moreover, no visible or ultraviolet absorption spectrum corresponding to niobium(IV) was detected in filtered samples,’! it is concluded that only niobium(V) was present in appreciable amounts in solution. The CO,/CO ratio was always higher than the values at which Nb, O is expected to be reduced to NbQ, .! 2 The conclusion that Nb(IV) was not present in solution is confirmed by the following calculation. From the estimate of Weaver and Friedman’ of the equilibrium constant for the reduction of NbF, in Li, BeF4 by hydrogen, one can calculate AG! 730k [NDE4(d)] = —306.4 (kcal/mole) . 11. Performed on remelted samples by J. P. Young, Analyt- ical Chemistry Division, ORNL. 12, J. F. Elliott and M. Gleiser, Thermochemistry for Steelmaking, American Iron and Steel Institute, Addison- Wesley, Reading, Mass., 1960. This, combined with AG values for Nb,Os, BeO, BeF,, CO, and CO,, gives for the reaction ', Nb, 05 (c) + ", CO(g) + 2BeF,(d) == NbF4(d) + 2BeO(c) + 2CO(g) (3) the very small equilibrium constant XNoF, (Pco,/Pco)'/* ~1X 10727 Hence, the amount of NbF, in the solutions is expected to be quite negligible. With only Nb(V) present in solution in equilibrium with Nb, O5(c) and BeO(c), we find that its concentra- tion (Fig. 11.1) is far greater than would have been predicted from the reported stabilities in molten fluo- rides of NbF,7'® and NbFs® and the available free- energy data for Nb, Os(c), BeO(c), and BeF,(d).!® In particular, if NbFs is the component in the present solutions we may calculate that NbF,;(d) should dispro- portionate completely to the metal and NbF5(d). The evident stability of Nb(V) in the present system is most plausibly explained by the formation of one or more oxygen-containing species which, for the present, we will assume to be of the form NbOF,*3)~ The data in Fig. 11.1 then may be represented by the equilibrium ', Nb, Os(c) + % BeF,(d) = NbOF;(d) +%BeO(c), (4) where, as usual, we represent the species in solution as neutral components. The equilibrium quotient for the reaction is defined as Qs =XNbOF; - The results in LiF-BeF, (67-33 mole %) give log Q4 = 1.70 — 5.27 (103/T) . In LiF-BeF, (52-48 mole %) they give log Q4 = 0.90 — 4.81 (10%/7) . The expression for Q, in LiF-BeF, (67-33 mole %), along with the estimate for Q;, and AGf values for NbFs, Nb,Os, NbO,, BeO, and BeF, were used to generate the Pourbaix diagram shown in Fig. 11.2. The variation of Xp,op, (ie., of Q4) with melt composition (Fig. 11.3) may be used to estimate the 89 activity coefficient y of the component NbOF; as follows: K, may be written 3/2 Ks =XNb0F3 7NbOF3/(‘1BeF2) / > wherein ag, F, is the activity of BeF, in the solution. Hence Ks4 - 3/2 YNbOF3 "E("Ber) /2 As has been customary in previous studies, we define standard states such that ynpor, =1 and ager, = 1 in Li, BeF, (33 mole % BeF,); hence K4 =(Q4)XBBF2 =0.33. Introducing values for ag.y,, available from previous measurements,! > we obtain the activity coefficients for NbOF; shown in Fig. 11.4. The rapid rise of YnboF, with Xpef, in these melts suggests that NbOF; competes strongly for the limited supply of fluoride ions, forming an anion with at least a —2 or —3 charge, that is, NbOF5 %~ or NbOF4>". 11.1.2 Equilibrations Involving Nickel Niobates in Molten Li, BeF, In the first series of equilibrations, Nb, Os, BeO, and NiO were equilibrated in Li, BeF,; in a nickel container under circulating argon. The apparatus and procedures were otherwise essentially the same as described in the previous section. Some difficulty was encountered in obtaining filtered samples free of solids. As judged by the reproducibility of the *Nb and nickel content of successive samples, this was corrected by use of a finer porosity filter (maximum pore size 10 u). Equilibration times were from 20 to 120 hr, decreasing with increasing temperature. At the outset of these experiments, only Nb, O5 and BeO were introduced; however, the nickel content of filtered samples quickly rose to values expected for NiO saturation, and, as will presently become clear, Nb, Os was also being converted to NiNb, Og¢. The required NiO presumably was supplied by inleakage of air early in the run. In discussing the results (points numbered in chro- nological order in Fig. 11.5), it will be convenient to refer to the phase diagram in Fig. 11.6. Here the boundary between the BeO and Nb, Qs regions (re- 13. B. F. Hitch and C. F. Baes, Inorg. Chem. 8, 201 (1969). 90 ORNL~-DWG 72-7596 4 , _ ! Nb,Os (¢} _| NbFg (g) 5 NbOF (d) s X ° I b9 » - © | 2 < << - I S J[’l ° 2 E PnbFg/XNbF, = ! ” 33 | Z _ % | o .}IN 9‘?; I Q\m NbF (d) % Py ' -~ [ ~ Q o < 2 XNbFg = \ N \ 4 X i Nb (c) \' NbOp (c) -6 :czlé o — 3'% B < w -8 -16 —12 -8 -4 0 Iog/Xo-z Fig. 11.2. Pourbaix diagram for niobium in molten Li, BeF,4 at 500°C. action 1, Table 11.1) is derived from the previously described measurements of Q4. The boundary between BeO and NiO (reaction 4, Table 11.1) is derived from available thermodynamic data.'® The other boundaries were generated from the data in Fig. 11.5 as follows: Points 1 to 3. Since the nickel content in solution indicated NiO saturation in these initial samples, and the x-ray patterns of an oxide sample taken at point 5 indicated the presence of NiNb, O4 and BeO as well, it was presumed that the dissolved Nb(V) concentration corresponds to point A4 in Fig. 11.6, at which all three oxide phases are present and which is metastable with respect to the precipitation of NigNb,O,. With the position of this point presumably established, the line corresponding to the boundary between BeO and NiNb, O¢ (reaction 2, Table 11.1) was drawn with the required slope of —% . After point 3, the apparatus was cooled to room temperature, and NiO was added. Points 4 and 5. The nickel content of the melt appears somewhat lower and the niobium content somewhat higher than previously. This may reflect the conversion of small amounts of NiNb, Og to NigNb, Oy according to reaction 6, Table 11.1. In the phase diagram (Fig. 11.6) this would correspond to a move- ment from point 4 toward point B. Point 6. Here the nickel content of the melt has fallen well below the value corresponding to NiO saturation; however, the niobium content has not risen appre- ciably. Hence, even though NiyNb, O, was detected in minor amounts with the SEM, it is thought that the system was still, in effect, near point A4. GRNL-DWG 72-7597 N4 Xnp (v) ¥1073 S 0.32 0.36 0.40 0.44 0.48 0.52 X BeF, Fig. 11.3. Effect of melt composition on the concentration of niobium(V) in LiF-BeF, mixtures at equilibrium with BeO and Nb, Q5 at 606°C. ORNL-DWG T72-7598 | Y, 20 // i Neos i b 2 - 7 5] [ A @ 1 178 o :_’ 5 /| | ; [ /0'33 YBer_ e / b4 I < /f S~ - 0.32 0.36 0.40 0.44 0.48 0.52 XBeF, Fig. 11.4. Activity coefficients of assumed component NbOF; and BeF, in LiF-BeF, mixtures at 606°C. 91 Point 7. Here the SEM suggested major amounts of NiyNb, Oy, the nickel content of the melt remained low, and the niobium content began to rise with time. However, final equilibrium with respect to NiNb, O¢, Nig4Nb, Og, and BeO (point B) was not established. The dashed boundaries in Fig. 11.6, therefore, are located only tentatively, being based on the amount by which the nickel content of the melt was depressed at point 7. On the basis of these results it is clear that the solubility of niobium can be quite low (<10 ppm) in melts containing appreciable amounts of 0%~ and Ni%* ions. We plan next to equilibrate molten Li, BeF,4 with NigNb, Oy and NiO. This should better locate one of the dashed boundaries in Fig. 11.6, and hence the other two as well. Also, there is an incentive for determining experimentally the slope of the boundaries between the NiO-Nb, Os phases, since this is related directly to the formula of the niobium component in solution. In particular, it may be shown that if the component has the general formula NbyxOyFsxay , then the slope of the boundaries between the NiO- Nb, Os phases will be 5x — 2 slope =———. Thus further measurements will not only define more precisely the solubility of niobium(V) in the presence of 0% and Ni** ions, but also test the validity of the assumption that the niobium(V) species in solution may be represented as the component NbOF;. 11.2 THE REACTION OF MoF4; WITH NIOBIUM J.D.Redman C. F. Weaver The mass-spectrometric investigation of niobium fluo- rides and oxyfluorides was continued. Previous studies include the vaporization of NbFs,'*>'® the reaction of niobium with low-pressure fluorine,'® the dispropor- tionation of NbF,,!7 the thermal decomposition of 14. C. F. Weaver and }. D. Redman, Reactor Chem. Div. Annu. Progr. Rep. Dec. 31, 1968, ORNL-4400, pp. 37-38. 15. C. F. Weaver and J, D. Redman, MSR Program Semiannu. Progr. Rep. Feb, 28, 1969, ORNL-4396, p. 161. 16. C. I. Weaver and J. D. Redman, MSR Program Semiannu. Progr. Rep. Aug. 31, 1969, ORNL4449 pp. 119-20. 17. C. F. Weaver and J. D. Redman, MSR Program Semiannu, Progr. Rep. Aug. 31, 1970, ORNL4662, pp. 73—-74. 92 ORNL- DWG 72-7599 -4 -3 10 10 T - — r ] A — I 5 F————T‘.{ 5 N J“ | X O e e Lfisr ‘ [ $) 2 3\ _ > CALCULATED FOR SATURATION | | \ — 50 WITH NiO AND BeO 5 — —5 | I\ g ?'f_ -4 2 107 — - T 20 3T % — | \ - = S~ P N L_ 2. 20 | _ M — \\ T S 5 5 JT 5 3 I 'l T 4 - '\1 — 10 - i ol — J_r 2 1 Lt e k | () (6) 10”8 ’ 1o=5 L= J _ L 0.9 1.0 14 {.2 0.96 1.04 {42 1.20 1OOO/T(°K) 100Cyr(°+<) Fig. 11.5. Concentrations of (7) niobium(V) and () nickel(II) in molten Li; BeF, equilibrated with NiO-Nb, O; phases and BeO. The numbers indicate the order in which points were determined. The straight line in b represents calculated values corresponding to saturation (reaction 4 in Table 11.1) with NiO and BeO, Table 11.1. Reactions in Li, BeF, represented by the boundaries in Fig. 11.6 Saturating oxide phases Reaction Equilibrium constant expression (1) NbyOg + BeO (2) Nle206 + BeO 3) Ni4 Nb2 09 + BeQ % Nb,05(c) + %BeF,(d) =NbOF3(d) + % BeO(c) XNbOF, /yNiNb,y O4(c) + 2BeF,(d) = NbOF3(d) + ¥ NiF,(d) + 2BeO(c) %y NigNby Og(c) + % BeF,(d) =NbOF3(d) + 2NiF,(d) + % BeO(c) 1/2 XNbOF; (ANiF,) / 2 ANbOF; (ANiF,) (4) NiO + BeO NiO(c) + BeF, (d) = NiF,(d) + BeO(c) XNiF, (5) Nb,Os + NiNb,Og 2Nb; 05(c) + %, NiF(d) =NbOF 3(d) + % NiNb, 0 (c) XNbOF3/(XNiF,) > (6) NiNbyOg + NigNb20g %NiNbyO4(c) + %NiF2(d) =NbOF3(d) + % NigNby0o(c) XNbOF,/ (XNiF,) /2 (7) NigNb,yOg + NiO " NigNb,09(c) + % NiF,(d) = NbOF5(d) + 1, NiO(c) XNbOF,/(XniF,) /2 93 ORNL-DWG 72-7600 10 T T T I ] {l ‘[ '/ ! {I fI T T }]rll T ! / l I I T + 5 L NiNb,Og yARR T . T T I \\ / i | 1 4 > \\ / s /l ! . , \\ / ’1 | :/ N ’ / -5 \( / 10 . N . / 4 T ‘ , N\ 4 4 SRAS / R — N T ' g \highe,0 1 5 > torrs) Species o 25°C 50°C 100°C 150°C 200°C 250°C 300°C 350°C MoFg 5.7 39.6 26.2 13.2 52.5 15.2 41.2 15.6 Mo, Fio 0 0 0 0 0 0 9.2 24 MoF; 0 1.4 1.8 3.8 10.5 33.6 5.2 MoF, 0 0 0 0 0 0 0 NbF5 0 0 0 0 0 16.3 45.5 64.0 NbF; 0 0 0 0 0 0 3.7 7.1 MoOF, 0 0 0 0 12.7 4.3 0 4.5 Nb,OF,4 0 0 0 0 0 0 3.4 0 Pressure (10_3 torrs) Species " R - 400°C 450°C 500°C 550°C 600°C 650°C 700°C 750°C MoFg 2.0 1.6 3.4 5.1 5.7 0.9 0.9 1.1 Mo, F; ¢ 2.4 0 0 0 7.4 0 0 3.4 MoF; 24.0 4.3 9.2 30.4 2.9 7.6 7.9 5.5 MoF, 0 0.9 0 6.8 13.1 11.6 14.5 17.5 NbF5 68.0 114.5 125.0 54.0 22.5 62.0 26.0 8.5 NbF3 7.3 0 38.7 31.8 34 15.6 7.7 19.4 MoOF,4 8.4 1.7 0 1.1 0 0 0 0 Nb,OF,4 0.2 1.5 5.0 4.2 6.2 2.7 4.2 15.4 Table 11.3. Partial pressure of molecules above MoF4 plus molybdenum Pressure (1073 torrs) Molecule 75°C 100°C 150°C 200°C 250°C 300°C 400°C MoFg 14.2 8.0 6.2 8.0 5.3 5.0 5.3 Mo,F; ¢ 0.5 0.9 1.7 5.9 3.5 8.8 46.7 MoF,4 0 0 0 0 0 0 0 MoOF,4 0.6 1.0 0.6 0.3 0.09 0.18 0.9 Pressure (10> torrs) Molecule 450°C 500°C 550°C 600°C 650°C 700°C 750°C MoFg 2.9 1.5 1.4 1.6 1.2 1.1 3.9 Mo, F; g 254 12.7 16.4 4.4 8.7 6.7 14.3 MoF, 0 0.6 0.4 1.1 1.9 3.3 6.8 MoQF, 0.7 0.4 3.8 3.9 3.0 0 0.8 Table 11.4. Cracking pattern for MoF ;“ Ion Relative intensity MoFs" 0 MOF4 100 MoF5 11 MoF," 27 MoF™ 22 Mo* 12 275V electrons. Table 11.5. Cracking pattern for Nb, OF ;¢ Ton Relative intensity Nb,OF," 0 Nb,OF 53 100 Nb,OF," 1 Nb,OF* 6 Nb, 0" 2 Nb, 5 4752V electrons. fluorides reacted with structural tantalum in the ap- paratus, producing both the previously reported MoTaF, 23 and the new NbTaF,,. Previously, the determination of the cracking pattern for MoFs has been impaired by simultaneous presence of much Mo, F,q. In this study the small fraction of Mo, F,, allowed the determination of a tentative cracking pattern for MoFs, which is given in Table 11.4. The oxyfluoride Nb, OF,; was seen previously only in the Nb, Os fluorination study.?® In the present work the absence of monatomic niobium oxyfluorides allowed the cracking pattern of the Nb, OF, to be determined. It is shown in Table 11.5. In this study the ratio of MoF, to MoFs was greater than found in the reduction of MoFg with molyb- denum, possibly a reflection of the greater reducing power of niobium compared with molybdenum. X-ray diffraction analysis identified the residue in the cell at the conclusion of the study as being MoF; and molybdenum. The niobium was completely consumed. 11.3 THERMODYNAMICS OF LiF-BeF, MIXTURES D.D.Sood J. Braunstein Excess chemical potentials of LiF in molten LiF-BeF, mixtures, calculated from accurate liquidus data,?8-2° 95 showed that the limiting chemical potentlal interaction parameter, lim (NELiF/XZ) s X-0 where uf LiF 18 the excess chemical potential and X is the mole fraction of beryllium fluoride, was correlated satisfactorily with the analogous parameters for other alkali—alkaline-earth fluoride mixtures in terms of the conformal ionic solution theory. On the other hand, as seen in Fig. 11.7, the liquidus data are reproduced well, up to 0.25 mole fraction BeF,, if the liquidus tempera- tures are calculated from the known heat of fusion and heat capacities of LiF under the assumption of an ideal mixture of LiF and Li, BeF,. These liquidus tempera- tures were obtained as part of a detailed study?® of the LiF-BeF, system, and the revised phase diagram is shown in Fig. 11.8. Accurate data on mixing thermo- dynamics for other alkali fluoride—beryllium fluoride mixtures are needed to elucidate whether the ionic model or the complex ion model is the more useful. In 28. K. A. Romberger and J. Braunstein, MSR Program Semiannu, Progr. Rep. Feb. 28, 1970, ORNL4548, p. 161; K. A. Romberger, J. Braunstein, and R, E. Thoma, J. Phys. Chem. 76, 1154 (1972). 29. J. Braunstein, K. A. Romberger, and R. Ezell, J. Phys. Chem. 72,4383 (1970). ORNL-DWG 71-8375 900 \ ‘ T 800 ' \?\\.,*\I\\ }mem_ LiF- BeFZT — S 700 Lo - | - .\ *, 3 ; Nt g 600 - —Y— | \ 1 . = \ : l} © - wl _ o : . W 500 [— — : .._Jfi LN m r | IDEAL LiF- JaeF J/ “. o - ‘L l ! 400 0.33 | A 300 L. - - = 200 f ) 1 Y O.15 020 0.25 0.30 0.35 XBeF, (mole fraction) 0 0.05 0.10 Fig. 11.7. Calculated liquidus temperatures based on hy- pothetical ideal LiF-BeF, and hypothetical ideal LiF-Li, BeF, compared with the measured liquidus. 96 ORNL-DWG 71-5270R2 [ | ] 900 500 — ; 7 848 /x (EUTECTIC) =0.3280 + 0.0004 Tmax =459.1 £0.2 %o 800 |— 450 LIQUID — X (EUTECTIC = 0.531 £0.002 700 — 400 — LipBeF, + - 5 LiF + LIQUID & LIQUID & 350 X 600 555 £ 030 0.35 0.40 045 0.50 0.55 @ wl a / & 500 B ————— L - — = 458.9 +0.2°C | BeF, (B-QUARTZ TYPE) | + LIQUID | 400 | — | 363.5 +0.5°C : | S — o LipBefy + BeF, (8-QUARTZ TYPE) LiBeFs + Befp e _ 300 LizBefa | ETTaT R 28000 (a-QUARTZ TYPE), Q m | LipBeFa+ W[ LiBeF, + BeF, (8-QUARTZ TYPE) e 7 LiBeFs @ 227°C 200 1 - \ 1 | i 0 0.4 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 XBeF2 (mole fraction) Fig. 11.8. Revised phase diagram of the LiF-BeF, system. Data from K. A. Romberger and J. Braunstein, MSR Program Semiannu, Progr. Rep. Feb, 28, 1970, ORNL-4548, p. 161; K. A. Romberger, J. Braunstein, and R. E. Thoma, J. Phys. Chem. 76, 1154 (1972). a mixture containing mole fraction X BeF,, the calculated mole fraction of M,BeF,, where M is an alkali metal, is y = X/(1 + 2X). The excess chemical potential of MF, based on mixing with M, BeF, rather than with BeF, , becomes “E,MF ‘_‘“EMF tRTIn 1—yp and the transformed interaction parameter becomes lim (flE'M 1:‘/)’2) s y=0 which is zero if MF and M, BeF, mix ideally. However, it is readily shown that if lim (WE'y g/¥?)=0, y=0 then the untransformed interaction parameter must have the value fim (uEy, p/X?)=2RT X-0 (nRT if the compound formed is M,,BeF,, ). Thus if the alkali fluoride—beryllium fluoride mixtures form ideal solutions of M,BeF, in MF, their limiting interaction parameters should all be about 4 kcal/mole. For LiF-BeF,, the interaction parameter is close to this, 5 kcal/mole. However, in order to be consistent with the application of conformal ionic solution theory to other alkali—alkaline-earth fluorides, the interaction parameters would be expected to exceed 10 kcal/mole for RbF and CsF. This seems to be the case,?? as shown in Fig. 11.9, although the conclusion is based on data which had not been intended to provide accurate chemical potentials. However, the interaction param- eters are not likely to be in error by 6 kcal/mole. It is striking that the ionic model leads to useful predictions, even though strong Raman bands have been reported for BeF,?% ion in LiF-BeF, and NaF-BeF, melts.?° 11.4 ELECTROCHEMICAL MASS TRANSPORT IN MOLTEN BERYLLIUM FLUORIDE—-ALKALI FLUORIDE MIXTURES H. R. Bronstein J. Truitt J. Braunstein The uncommon unicationic conduction process in molten LiF-BeF, mixtures,®>! whereby electrical cur- 30. A. S. Quist, J. B. Bates, and G. E. Boyd, J. Phys. Chem. 76, 78 (1972). 31. K. A. Romberger and J. Braunstein, /norg. Chem. 9, 1273 (1970). 97 ORNL-DWG 69-8542R2 i6 i} Nu,Bo’ I Na;,nCaj l Na,S K, Sr 12 La r _',L K,Ca TAA o K,Ba /’I i 'N 1 8 p T L . — I N 0 _—" FLUORIDES < CHLORIDES / 0 wg 4 = o Na, Ba Na, Ca Li, Be Ng, Be K,Be| Cs,Be Na, Sr Na Rb,B 4 L 14 Jl 1 |91 “ tu ef -0.08 -0.04 0] 004 008 042 046 020 024 0.28 8y (A7) Fig. 11.9. Application of conformal ionic solution theory to the chemical potential interaction parameters in alkali—alkaline earth fluoride mixtures. 5, is the difference of reciprocal cation-anion distances, 1 1 dys —p— dg?_p- rent is carried only by lithium ions (relative to fluoride as the reference constituent) over a wide range of composition, suggests a number of interesting con- sequences®>? of possible theoretical and practical im- portance. These are being tested by electrochemical scanning methods including voltammetry, chrono- potentiometry, potential step methods, etc. Such methods have been useful previously in characterization and analysis of electrolytes, including molten salts, but primarily with solutions dilute in electroactive species. Here the electroactive constituent [Be(Il)] predomi- nates. An electrochemical cell consisting of a small beryllium electrode and a large inert electrode in an alkali fluoride—beryllium fluoride melt may be expected to show appreciable difference in electrical conductivity for anodic or cathodic current at the beryllium elec- trode. Anodically, a film of virtually pure (highly resistive) BeF, should be formed at the beryllium electrode as the current-carrying alkali cations migrate away. On reversal of polarity, BeF, is reduced at the beryllium electrode, its supply then being diffusionally replenished by the large concentration gradient. The system should thus behave as a two-state device of high or low resistance. Furthermore, the relaxation time of the high-resistance film of BeF, on current reversal should depend on the beryllium-alkali ratio in the melt and could conceivably provide the basis for an analyti- cal method for this ratio. Preliminary chronopotentiometric measurements®3 made at a beryllium electrode in an NaF-BeF, melt containing approximately 75 mole % BeF, appear to be qualitatively consistent with the predicted behavior. The extremely well-defined transition times encourage further study. Figure 11.10 shows a measurement of emf of the beryllium indicator electrode relative to a beryllium reference electrode in NaF-BeF, (~75 mole % BeF, at ~490°C) during passage of a constant current, first anodically and then cathodically, between the beryllium indicator electrode and a large nickel counterelectrode. On passage of constant anodic cur- rent at the beryllium electrode, the measured potential first rises rapidly to a nearly constant value as, presumably, Be(Il) is formed in the melt. A rather abrupt rise in electrode potential is then required to sustain the constant current, possibly associated with the formation of a resistive BeF, film. On reversal of polarity, the potential drops very rapidly to a nearly constant value, suggesting rapid reduction of the BeF, film and indicating a much smaller concentration polarization cathodically than anodically. 32. J. Braunstein and K. A. Romberger, “High Temperature Ionic Switching Device,”” CNID-2739. 33. We wish to acknowledge T. R. Mueller’s valuable sugges- tions and aid in initiating the chronopotentiometric measure- ments, ORNL-OWG 72-7604 600 | 500 - N l POTENTIAL AT WHICH POLARITY WAS REVERSED 400 4* S {— R E 300 I S b — 2 | | — 5 200 S U S T \OPEN -CIRCUIT POTENTIAL 0 {00 200 300 400 TIME (sec) Fig. 11.10. Typical chronopotentiogram in molten NaF-BeF, (~25 to 75 mole %) at 520°C, current density ~40 mA/cm?2. Potential of beryllium working electrode, relative to a beryllium reference electrode, vs time. These measurements should prove useful also in testing for the formation of subvalent beryllium species. 11.5 ELECTRICAL CONDUCTANCE IN BERYLLIUM FLUORIDE RICH NaF-BeF, MIXTURES G.D. Robbins J. Braunstein Continuing the investigation of transport properties in molten alkali fluoride—beryllium fluoride sys- tems,>#3¢ electrical conductance in the NaF-BeF, system has been measured at compositions 90, 85, 80, 75, and 70 mole % beryllium fluoride and at tempera- tures between 485 and 600°C. The aim of these experiments had been to determine the manner in which melt composition affected the increase in resis- tivity and the change of activation energy in the region between the normal molten salt and the glass transition temperature. It was not possible, however, to avoid nucleation during extensive slow supercooling of the melts, and a cell is being designed for conductance measurement during warming of quenched glasses. 34. G. D. Robbins and J. Braunstein, MSR Program Semi- annu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 81-8S. 35. G. D. Robbins and J. Braunstein, MSR Program Semi- annu., Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 98—100; ORNL4676, pp. 109-10. 36. K. A. Romberger and J. Braunstein, Inorg. Chem. 9, 1273 (1970). 98 The data are shown in Fig. 11.11 as Arrhenius plots of the logarithm of resistance against the reciprocal of absolute temperature. These data were obtained with the previously described®® all-metal conductance cell and a series RC bridge. Starting materials were distilled beryllium fluoride glass and single-crystal sodium fluo- ride, handled under an inert atmosphere and introduced into the conductance cell in a ratio to produce a 90 mole % BeF, melt. Changes of composition were made by NaF addition. The high viscosity of the BeF,-rich mixtures pre- cluded mixing by gas bubbling, and extremely long equilibration times were required. With the 90 mole % BeF, mixture, a systematic decrease of resistance with time was observed during the first half of a 60-day period, followed by slowly varying apparently random fluctuations that probably reflected bubble migration in and out of the interelectrode region. Thus the absolute values of conductance may be no more accurate than +10% at this composition; however, the temperature dependence of the data taken over a short time interval should be more accurate. The frequency dispersion of the measurements showed no consistent pattern and led to corrections of 0.3 to 5% in the resistance. Preliminary values of Arrhenius energies, d logR E, =4.575 a7 obtained graphically at 525°C from the curves shown in Fig. 11.11 are 10.7, 1l.5, 13.7, 155, and 154 kcal/mole at increments of S mole % from 70 to 90 mole % beryllium fluoride. The Arrhenius energies appear to vary less with temperature than previous results with LiF-BeF, mixtures. The apparent slight decrease of Ej with decreasing temperature for the 90 mole % BeF, mixture, in contrast to the usual increase for most molten salts and mixtures, requires verification in view of experimental uncertainties, but had been observed also with pure BeF,. 11.6 THE DISPROPORTIONATION EQUILIBRIUM OF UF; SOLUTIONS L.M. Toth L. O. Gilpatrick In the previous report,>” measured equilibrium ratios of UF3/(UF; + UF,) were reported for the reaction of 37. L. M. Toth and L. O. Gilpatrick, MSR Program Semi- annu, Progr. Rep. Aug. 31, 1971, ORNL4728, p. 77. ORNL-DWG 72-7602 500 475 ' [\ E_ 1000 90 mole % - fli > Bef, — 500 1 q 200 100 12 N p — ~ j —110 g K - = c I | 1s 70 N . o —2 l { 124 126 128 130 132 134 136 1000/ (k) Fig. 11.11. Resistance vs temperature for molten mixtures of BeF, and NaF. The line drawn through the points at 70 mole % Bel, corresponds to a previous set of measurements. dilute UF; solutions with graphite at 550°C as shown in Eq. (1), 4UF; + 2C(graphite) = 3UF,; + UC, . (1) Equilibration was followed by measuring UF3 and UF, concentrations in a diamond-windowed graphite spec- trophotometric cell as a function of time using a Cary 14-H absorption spectrophotometer. X-ray powder pat- tern analyses were used to identify the uranium carbide formed. However, some uncertainty in the carbide identification was caused by the small amount of material which was sampled. In addition, the UF,/ (UF; + UF,) values reported at equilibrium and recalcu- lated in Table 11.6 as equilibrium quotients, Q = (UF3)*/(UF,)?, were unusually low when compared to Long and Blankenship’s previous data.?® To resolve these difficulties it was necessary to study the back reaction of Eq. (1) using pure uranium carbides to reduce UF, solution in LiF-BeF, (66-34 mole %) at 550°C, 38. G. Long and F. F. Blankenship, The Stability of Uranium Trifluoride, Part II, ORNL-TM-2065 (November 1969). Table 11.6. Equilibrium UF3/(UF3 + UF,4) ratios, R, for LiF-BeF, mixtures in graphite Taken from ref, 34 and recalculated as equilibrium quotients, Q = (UF3)*/(UF,4)3, where concentrations are expressed in mole fractions. Solution 650°C 550°C (mole %) LiF-BeF, R Q R Q 66-34 0.025 2.73x107'% 0004 1815x 10713 48-52 0.13 3.03x10°7 0.03 6.21 x 10710 UC + 3UF, = 4UF, + C, (2) Y, U,C3 + 3UF, =4UF, + %,C, (3) UC, + 3UF, =4UF; + 2C , (4) where the expected product of the reaction would be UF; and graphite. The relative magnitudes of the equilibrium quotients, Q = (UF3)*/(UF,)?, for reactions (2) through (4) can be predicted using previously established free energies of formation for the uranium carbides and assuming the activity coefficients for UF; and UF,; do not change for the three equilibria. The free energies of formation for UC, %, U,C5, and UC, at 550°C are reported to be —24.83° —24.81,*° and —22.36%° kcal/mole respec- tively. It is expected that Qgq.(3) < C@Eq.2) < QEq.(4); that is, either U;C3 or UC should be the stable carbide phase in contact with the UF;-UF, solution at 550°C. To experimentally test these predictions, very pure samples of UC, U,C5, and UC, were obtained from Los Alamos Scientific Laboratory sealed in glass ampuls and used to measure Q values as described for Egs. (2)—(4). Although for the more active carbides a process such as shown in Eq. (1) would be simultaneously occurring, equilibria involving these more active carbides can be measured by adding excess reagent carbide and taking advantage of the fact that the reaction rates of Egs. (2)—(4) are fast with respect to Eq. (1). Equilibrium quotients measured spectrophotomet- rically in the graphite cell for Eqs. (2)—(4) are shown'in Table 11.7. These indicate that the stable carbide phase is UC, and not U,C; or UC. Furthermore, UC, was identified as a product of reaction (2). Equation (2) was then modified to include the product UC;: 2UC + 3UF,; = 4UF; + UC, . (2a) The other two reagent carbides produced no detectable carbide products and so remain with only graphite as the reaction product. Equilibrations similar to these were performed in sealed evacuated nickel capsules in the presence of an atmosphere of 4% hydrogen—argon gas. The same results were found for these experiments as those conducted in graphite cells. Using the measured Q values in Table 11.7 at 550°C and the previously established AG°g,5 for U,C; as —49.62 kcal/mole,*' AG°s,3 for UC and UC, can be calculated assuming the activities of the pure phase carbides are unity and that the activity coefficients for UF; and UF, do not change from Eq. (2) to (4). These new values for UC and UC, are found to be —24.03 and —28.0 kcal/mole respectively. ) The equilibria of Eqgs. (22), (3), and (4) lead to the conclusions: (1) the stable carbide phase in contact with LiF-BeF, (66-34 mole %) solutions of UF;-UF, is UC,; (2) the equilibrium quotient for Eq. (4) is much in agreement with the conclusions of Long and Blank- enship, even though the value of AG®g,; (UC,) which they used from Rand and Kubachewski*? (—20.1 kcal/mole) was different from either the currently accepted value or the one reported here; and (3) the 100 Table 11.7. Measured equilibrium quotients, 0 = (UF3)*/(UF4)3, for dilute UF4+ UF3 in LiF-BeF, (66-34 mole %) at 550°C Concentrations expressed in mole fractions Eq. (2) Eq. (3) Eq. (4) Reagent carbide uC U,C3 uC, 0 x 10* 28.11 1.362 0.194 equilibrium quotient for Eq. (4) is considerably larger than was given in the preliminary report and repro- duced in Table 11.6. This last conclusion suggests that the previously reported UF3/(UF; + UF,) ratios in Table 11.6 did not involve equilibria with pure carbide phases but probably involved an oxycarbide phase of fixed uranium activity. However, this “oxycarbide” phase has not been identi- fied. The equilibrium involving the “oxycarbide’ phase does point to a lower stability of UF; in graphite. As long as pure carbides and/or graphite is present, the UF5/(UF; + UF,) ratio in solution is quite large (approx 0.30 at 550°C), but when the system is contaminated by oxygen the UF;3/(UF; + UF,) ratio will reach still lower values, represented by those in Table 11.6. 11.7 THE RAMAN SPECTRA OF Be,F,*>” AND HIGHER POLYMERS OF BERYLLIUM FLUORIDES IN THE CRYSTALLINE AND MOLTEN STATE L.M.Toth J.B.Bates G.E.Boyd The tendency of BeF, to form extensive three- dimensional networks of | | —B'e—F—Ble—F— chains in its crystalline and glassy state has been shown by x-ray diffraction measurements.*> Presumably, mol- 39. E. K. Storms, The Refractory Carbides, vol. 2, pp. 171-213, Academic, New York, 1967. 40. W. K. Behl and J. J. Egan, J. Electrochem. Soc. 113, 376 (1966). 41. The value of AG°g,3 for U,C3 was chosen as fixed because previous data existed at this temperature region and the influence of atmospheric contaminants was expected to be minimal. 42. M. H. Rand and O. Kubaschewski, The Thermodynamic Properties of Uranium Compounds, p. 71, Interscience Pub- lishers, John Wiley and Sons, Inc., New York. 43, A.H. Narten, J. Chem. Phys. 56,1905 (1972). 101 ten BeF, also is highly associated, as may be inferred from its Raman spectrum*®* and from its very large viscosity. According to Baes’ polymer model,*® the addition of basic fluoride such as LiF to molten BeF, is believed to cause breakage of these network links by supplying extra F: | | —Ble—F—BIew +F~ IS If an excess of fluoride ion is added, a complete disruption of the network results, and free BeF4 2™ ions are formed. Although Raman spectra for the two extremes in the BeF, system have already been presented,****5 that is, that of pure molten BeF, and that of BeF,?", no evidence has been given to support the polymer model mechanism of Eq. (1) which occurs in the intermediate region. This region is the subject of the following discussion, which includes Raman spectra in support of the polymer formation mechanism. The approach taken was to identify simple species occurring during initial polymerization stages by com- 1/2- 1/2- (1) paring the Raman spectra of melts with those of known species found in solid crystalline compounds. The ion Be, F, 3", representing the first step in the polymeriza- tion process, was sought in a number of alkali metal fluoride—BeF, mixed-salt cdbmpounds.*” Single crystals of the congruently melting compound Na,LiBe,F, were grown, and, from an x-ray structure determina- tion,*® the presence of Be, F,3~, consisting of two Be-F tetrahedra sharing a corner and having approximate C, symmetry, was verified. The Raman spectrum of polycrystalline Na, LiBe, F,%8 at 77°K is shown in Fig. 11.12. The strong band at 525 cm ™! is assigned to the symmetric stretching mode involving both the bridging Be-F and the terminal Be-F bonds. This mode is related to the symmetric stretch of BeF4%™ at 550-560 cm ™' but, it should be noted, is displaced to lower frequencies. 44, A. S. Quist, J. B. Bates, and G. E. Boyd, Spectrochim. Acta 28A (1972) (in press). 45. C.F. Baes, Ir., J. Solid State Chem. 1, 159 (1970). 46. A. S. Quist, J. B. Bates, and G. E. Boyd, J. Phys. Chem. 76, 78 (1972). 47. The authors acknowledge valuable suggestions from Max Bredig, Chemistry Division, in this matter. 48. G. Brunton, ‘“The Crystal Structure of Na,LiBe,F;,” to be published (1972). ORNL-DWG. 72-2918 | T | ' | | POLYCRYSTALLINE Na,LiBe,F, A) 77°K I Slit 2¢m™! 1 x 103 ¢/s 520+ 298°K 890 Slit 33cm™ seq ® 3 2 x10%¢ss 5761 ’ 352 206 > 370 270 |.__- (7)) Z Ll - 443 - 779 352 W* ~ e tia 1 l 1 [ 1 1 ) l | l 1 l | L 900 800 700 600 500 400 300 200 FREQUENCY (cm™') Fig. 11.12, Raman spectrum of polycrystalline Na, LiBe, F5 at to show weaker modes. 77°K. Curve A4, full scan at normal scale; curve B, expanded scale Extensive usage of this frequency shift will be made in characterizing the beryllium fluoride species as the melt compositions change based on the interpretation that increasing length of the Be-F network is represented by progressively lower frequency values for the 525-cm ™ band. When Na, LiBe, F, is melted, an additional feature at 550 cm™! appears as a shoulder on the side of the 525-cm ™! band in Fig. 11.134. This band is identified as arising from the symmetric stretching mode of BeF,%", which is produced by a dissociation process which occurs on melting the compound: xBe, F73 = BeF, 2 + Beyy 1 Frypa3¥2) 7, (2) where a similar band due to the larger component, Beyy_ 1 Frx_aCG¥~2)7 is expected to lie under the 525-cm ™' band and is not resolved. Molten Na, LiBe, FF; thus represents a system of various Be-F species which combine during crystal growth to form essentially pure Be, F,3 ions. The BeF,? and Be, F,3" can be explicitly identified in the melt because the polarized bands at 555 and 525 cm™' for each, respectively, are resolved. ORNL-DWG. 72-2486 T T T T T T T T MELT SPECTRA NoF LiF BeFp A 40 20 40 mole % 478°C B 286 429 28.5 mole % 617°C 5504 522 INTENSITY ~—= 900 700 500 300 100 FREQUENCY (cm™') Fig. 11.13. Raman spectrum of (4)molten Na,LiBe,F, and (B) NaF-LiF-BeF, (28.5-42.7-28.6 mole %). The effect of adding excess fluoride ion, curve B, causes the 525-cm ! band to disappear and the 550-cm ' band, attributed to BeF42_, to remain, 102 A more complex situation exists in melt mixtures of LiF and BeF,, because bands of individual species other than BeF4* and Be,F,;>” cannot be resolved. In Fig. 11.14, only a shift in the strong polarized band envelope to lower frequencies is evident as the F~ concentration is decreased. The shift extends to 480 cm™" for the 48-52 mole % LiF-BeF, composition and, in the limit of pure BeF,, should reach 282 cm™ as observed previously.** At present, it is believed that the shift to lower frequency results from a change in form of the symmetric mode from a stretching (BeF;?) to a bending mode (BeF,) as the extent of polymerization increases. Furthermore, examination of this mode for the Li,BeF, composition (Fig. 11.14, curve B) indicates that it is not entirely due to BeF,* but already contains contributions from modes of ORNL-DWG. 72-2916 IIIIlllllllllllllllllllllllll 550 540 (25 >_ = wn & — Mole Percent = LiF BeF, A 75 25 B oo 34 C o0 40 D 54 46 E 48 52 IlllllllllllllllllIllll_ljllll 700 600 500 400 FREQUENCY (cm™) Fig. 11.14. Effect of varying LiF/BeF, melt composition on the strong polarized band in the Raman spectrum of the LiF-BeF, system at approximately 600°C. higher analogs. The symmetric mode of BeF, 2™ is then more exactly represented by curve 4 with a peak still at 550 cm ™! but with a smaller band half width. 11.8 RAMAN SPECTRA OF MOLTEN AND CRYSTALLINE POTASSIUM DICHROMATE*? J.B.Bates L.M.Toth A.S.Quist G.E.Boyd Raman spectra of the room-temperature phase of crystalline K, Cr, O, and spectra of aqueous potassium dichromate have been reported recently 5951 References to earlier spectroscopic studies with dichromate are cited in refs. 50 and 51. A more comprehensive investigation of the Raman spectrum of K,Cr, O; was undertaken because: (1) The room-temperature (triclinic) phase is known to have a unit cell structure in which the Cr,0,%" ions occupy two sets of nonequivalent sites. Multiple- site effects have been proposed as the primary cause of mode splitting in spectra of several crystalline materials (MoFs and Na,CO,) studied in this Laboratory.5 253 It was therefore of interest to investigate a material in which a “two-site” effect could be well established. (2) The Cr,0,% ion has a C,, molecular structure in the molten and crystalline state isomorphic with that of the Be, F,* ion; it was hoped that the present study would aid in interpreting the spectrum of the latter species. Single-crystal Raman spectra were recorded at 77 and 300°K using the 6328-A line of a helium-neon laser for excitation.>* Low-temperature spectra at frequencies above 300 cm™ are well represented in Fig. 1 of ref. 50. Hence, only the spectra observed at 77 and 300°K in the region below 300 cm ™' are shown in Fig. 11.15. At ca. 269°C, K,Cr,O; undergoes a phase transition from the triclinic form to a structure reported to be monoclinic.®® Spectra were obtained at temperatures above and below the transition point from a single crystal placed in the high-temperature Raman furnace. Results obtained in the v, and v, regions of dichromate are presented in Fig. 11.16. Band frequencies obtained from low- and high-temperature solid-state spectra of 49. Abstracted from a paper to be submitted for publication. 50. W. Scheuermann and G. J. Ritter, J, Mol. Struct. 6, 240 (1970). 51. M. S. Mathur, C. A. Frenzel, and E. B. Bradley, J. Mol. Struct, 2,429 (1968). 52. 1. B. Bates, Spectrochim. Acta 27A, 1255 (1971). 53. M. H. Brooker and J. B. Bates, J. Chem. Phys. 54, 4788 (1971). 54, Additional experimental techniques employed in these measurements have been detailed in earlier reports. 55. L. A. Zhukova and Z. G, Pinsku, Sov. Phys. Crystallogr 9, 31 (1964). 103 ORNL-DWG. 72~ 3312 z{x})y 300°K i lJ_llgLIllllIIl{llilliLllLil[llllJ; =300 250 200 150 100 50 = = z(x2)y = 77°K IAI;I.J;IIl_‘;jllllgilllLIJ_LlLll‘LL 250 200 150 100 50 0 FREQUENCY (cm™') Fig. 11.15. Low-frequency Raman spectra of crystalline KQCI'Q_O'? at 300 and 77°K. K,Cr, O, are collected in Table 11.8. The Raman spectrum of molten K,Cr,0, (mp = 398°C) was measured at 435°C from a sample contained in a Pyrex capillary tube. The spectrum of a saturated aqueous solution of dichromate at 25°C was also obtained. The melt and solution spectra are presented in Fig. 11.17, and the frequencies observed from these experiments are given in Table 11.8. The low-temperature (<269°C) structure of crystal- line K, Cr, 0, is triclinic with four Cr, 0,2 jons in the primitive unit cell. The lattice is centrosymmetric and belongs to the space group C;' . The four Cr, 0, ions occupy two sets of nonequivalent sites. The observed splitting of the vibrational modes of Cr, 07 in the low-temperature solid phase is thus due to a two-site effect. Each anion mode gives rise to a single A, component (Raman active only) and a single A4, component (infrared active only) as a result of dynamic coupling between two equivalent anions. The other equivalent pair of anions in the unit cell also gives rise to Ag + A, crystal states. Thus, for example, v, - Al +Ag° + A4,% + AP, and the 914-909 cm ™! pair of Raman bands (Fig. 11.15) indicates a two-site splitting of 5 em™! for v;. Similar assignments can be made for the B; and A, modes observed in the 750- and 560-cm ™" regions respectively (Table 11.8). In other regions of the Raman spectrum of K, Cr, 0, (triclinic), the over- ORNL-DWG. 72-3313 Ifilr]l[llfi T]ll]'] 280°C ._J\,__ 268°C >._ = % & A~ '._ = 241°C _z\,_,.«/\\'- -196°C LAA_+___JL l||l|||IlJ_|]l||||l| 1000 950 900 600 550 FREQUENCY (cm) Fig. 11.16. Raman spectra of the »v; and v; regions of K,Cr, 07 recorded at temperatures above and below the solid-solid phase transition point of ca. 267°C. 104 ORNL-DWG. 72-3344 895 MOLTEN K,Cr,0; 435°C ) 138 ZXxy 3?0 268 3 > A B 5 s leea o leaaa byl eea v v by clsianl = (1000 950 9Q0Q 850 400 350 250 200 150 100 w - z 905 AQUEOUS KyCry04 25°C 1 b l 1 J_L 1 1 ]. LJ; 1 L 200 T U G I S 1200 1000 800 600 400 FREQUENCY {(cm™) 0 Fig. 11.17. Raman spectra of molten and aqueous K, Cr, 0. lapping of mode frequencies precludes a unique assign- ment. It is interesting to note that measurements of polarized Raman spectra with single crystals are of small value in assigning the observed bands, since all Raman-active phonons have the same symmetry, that is, Ag. The spectra presented in Fig. 11.16 show that the two-site splitting of v, and »; can be detected at temperatures just below the transition point. In the high-temperature phase, single bands were observed for both these modes. The change in the crystal structure results in a change in symmetry of the factor group states, while the number of k = O states is the same in both phases. In the monoclinic phase, each molecular state gives rise to four factor group states. For example, vi(d,)>Ag + B, + A, + B, . The single symmetric bands observed for v, and v; were assigned to A, components because the intensity of the B, component is expected to be much less than, that of the 4, component. Furthermore, the A,-B, splitting which arises from correlation field coupling may be much less than the Ag-4, splitting observed in the triclinic phase as a result of the two-site effect. Particular attention was given to the measurements of the low-frequency region in the molten and crystalline phases of K, Cr,0,. The molten-salt spectra (Fig. 11.17) revealed a band between 80and 130 cm™ which is tenta- tively assigned to an 4, torsional mode denoted by v_. Table 11.8. Frequencies (cm_l) and assignments of the bands observed in Raman spectra of aqueous, molten, and crystalline K, Cr, 04 105 Aqueous solution (25°0) Melt? (435°C) Solid? 320°C 280°C 268°C 241°C -196°C Assignmentb 940 dp°© 905 p 556 368 217 928 dp 895 p 545 380 368 224 138 949 939 907 562 385 369 365 282 220 950 940 907 561 110 950 940 906 561 546 959 947 938 908 905 562 550 125 113 184 172 166 969 964 954 946 939 930 923 914 909 575 568 394 390 386 384 248 237 222 157 147 142 136 130 123 116 107 100 95 88 80 77 74 71 64 61 380 375 370 364 361 59 57 54 56 28 A2’B15B2 P L z | L Al:AZ:Bl’B2 ] L A4, B, Yr, VL | L vy - “mp = 398°C; solid-solid phase transition at ca. 269°C. Assigned according to symmetry species of Cqy point group; v, denotes torsional mode, vy denotes external modes. “p = polarized, dp = depolarized. 106 The frequency of this mode appears to undergo large shifts with temperature in the solid phases. At liquid- nitrogen temperatures (Fig. 11.15), v, also exhibits a fine structure which may be a result of interaction between this mode and the lattice phonons of dichromate. A similar interaction may also occur between v, and low-frequency hindered librations of dichromate ions in the molten salt. A better understanding of internal- external mode coupling of species in molten salts is important in interpreting the results obtained in melts in which polymerization can occur, such as the LiF-BeF, system (see sect. 11.7 of this report). 11.9 NONIDEALITY OF MIXING IN THE SYSTEMS Li2 BeF4'LiI, Naz BeF4-NaI, AND C52 BeF4'CSI A.S.Dworkin M. A. Bredig Effects of ion size, charge, and polarizability in systems with a complex anion, BeF,?", were studied by 650 -, 600 | NazBeF4—NoI e 550 7 (°C) 500 400 100 mole Y% MI MI measuring the phase diagrams of the systems M, BeF,- MI (M = Li, Na, Cs). These are shown in Fig. 11.18, together with ideal diagrams where the MI and M, BeF, liquidus are calculated for one particle per solute molecule (undissociated BeF,*™ or 17, respectively;n = 1). The calculated effect of complete dissociation of BeF,? to one Be* and four F~ ions (n = 5) is also shown. The experimental iodide liquidus is similar in the three systems. The BeF,*™ seems to show a high degree of dissociation only in very dilute solution. At the lowest concentration measured, about 3 mole % M, BeF,, only about 10 to 15% dissociation is indi- cated. The rate at which the degree of dissociation decreases with increasing M, BeF, concentration is difficult to estimate because of the nonideal mixing of I~ and BeF,% ions evident at and near the eutectic composition. The Csl liquidus appears slightly less curved to and beyond the ideal one than the Nal or Lil ORNL-DWG 74—~ 438894 80O 750 CszBeF;—CsI N 700 650 \ \ \ 600 \ \ —to— & A 0“__——— —————‘J— 550 500 0 20 40 60 80 100 CszBeF4 mole Y% Csl Csl Fig. 11.18. The systems Li, BeF4-Lil, Na; BeF4-Nal, and Cs, BeF4-Csl. liquidus. The positive deviation from ideality for the iodide liquidus, that is, excess partial free energy of mixing u¥ 1 > 0, at concentrations greater than about 20 mole % M,;BeF; (and probably at even lower concentrations) may be compared with the LiF liquidus in the LiF-Li,BeF, system, which is ideal up to approximately 45 mole % Li, BeF,.5% In mixtures with a common anion, the polarization energy usually makes the overwhelming contribution to the interaction po- tential. On the other hand, the much smaller interaction potentials for mixtures with a common cation result from a delicate balance between the positive contribu- tions of the Coulomb and van der Waals energies and the negative contributions of the repulsion and polariza- tion energies. It is tempting to offer an explanation for the more positive deviation from ideality when Li, BeF, is added to Lil than to LiF by juggling the changes in magnitude of the four effects above with a change from I” to F~. However, we must also remember that the differences reflected in the phase diagrams are 56. M. A. Bredig, Chem. Div. Annu. Progr. Rep. May 20, 1971, ORNL4706, pp. 115-56. 107 due to partial excess free energy of mixing and not necessarily to the enthalpy of mixing, which determines the interaction potential. The explanation for differ- ences in degree of nonideality must then include differences in the entropy of mixing for which little data exists for these systems. The M, BeF, liquidus lines all show positive deviation from ideality, although to a much lesser extent than the MF liquidus in MF-MI mixtures,>® especially in the dilute MI region. The sodium system shows the greatest nonideality, followed by the lithium system and then the cesium system, with only a very small deviation from ideality evident above 20 mole % Csl. Again, an explanation for the differences in nonideality will depend on a knowledge of the entropies of mixing. Fusion enthalpies of Na,BeF, and Cs,BeF; were calculated from the initial liquidus slopes to be 6.5 and 11.0 kcal, with a probable error of *5%. Fusion entropies, 7.5 and 10.5 cal deg™ mole™!, are consider- ably less than the AS = 14.5 cal deg™ mole™ for Li, BeF,. This is as expected, considering that Na, BeF, and Cs,BeF, are isostructural with the corresponding sulfates, while the Li,BeF, is isostructural with Be')_SiO‘p Part 3. Materials Development H. E. McCoy J. R. Weir Our materials work is currently involved with several important areas including (1) gaining a better under- standing of the intergranular cracking of Hastelloy N that occurred in the MSRE and a method of controlling this cracking in future reactors, (2) developing a graphite with improved dimensional stability under radiation, (3) development of surface sealing methods for reducing the permeability of graphite to ' 35 Xe, (4) modification of the composition of Hastelloy N to obtain an alloy that is more resistant to neutron irradiation, (S5) evaluation of Hastelloy N for use in steam generators, (6) construction of a molybdenum system for use with bismuth and fluoride salts, and (7) the evaluation of other materials for use as structural materials in the chemical processing plant. The first area has occupied a high priority; additional parts of the MSRE have been examined, and a rather extensive program of laboratory tests has been initiated. The laboratory tests are directed at determining the cause of the cracking, the rate of progression with time and temperature, and a reasonable solution to the cracking problem. The lifetime of a breeder core will be determined by the dimensional stability of the graphite. Dimensional changes can be accommodated to some extent, but volume expansion is usually accompanied by increasing permeability to xenon, which can become intolerable. Thus the problem of dimensional stability cannot be separated from the requirement of low permeability to 135Xe. We are evaluating graphites made by commer- cial vendors and locally in an effort to understand what types of graphite have the best dimensional stability. This information is being fed back to commercial 108 vendors and into our own experimental fabrication program. Pyrolytic carbon coatings derived from propene are currently being studied as a means of reducing the permeability to ! *° Xe. The program to develop a modified Hastelloy N with improved resistance to radiation indicated that alloys with additions of 1.5 to 2.0% Ti were most promising, and recent efforts have concentrated on these. Material has been obtained from three vendors that has been made by two basic melting practices. The evaluation includes weldability tests, unirradiated mechanical property tests, and postirradiation mechanical property tests. A corrosion facility in TVA’s Bull Run Steam Plant is being used to evaluate the corrosion of Hastelloy N in steam. Both unstressed and stressed samples are in- cluded in the tests. Most of the work on chemical processing materials is going into the construction of a reasonably complex test facility constructed of molybdenum. Many new techniques have been developed which overcome or circumvent the basic difficulties of molybdenum fabri- cation. (Welds in molybdenum are inherently brittle, and fabrication into large sizes is hampered by the need for high temperatures and large forces to fabricate large parts.) The mutual requirement of compatibility with bismuth and salt narrows the choice of materials for the processing plant, but our screening tests offer encour- agement that besides molybdenum, both tantalum and graphite will be compatible. Experiments are being started to determine whether these materials can be used and under what operating conditions. 12. 109 Intergranular Cracking of Structural Materials Exposed to Fuel Salt H. E. McCoy Examination of Hastelloy N components removed from the MSRE has shown that all surfaces that came in contact with fuel salt had intergranular cracks to a depth of 1 to 13 mils. Some of the cracks were visible when the parts were removed from service, whereas others did not appear until after the parts were deformed. Our observations and experiments indicate that this cracking is probably associated with fission products, although we have not ruled out the possibility that some yet-undetected mode of corrosion may cause the cracking. Our work involves (1) further studies of the materials from the MSRE, (2) corrosion experiments to deter- mine whether intergranular attack will occur under certain conditions, (3) studies of alloys containing or exposed to numerous fission products, (4) numerous experiments with tellurium to determine its effects on Hastelloy N, and (5) limited experiments with other alloys such as type 304L stainless steel and nickel-200 to determine their susceptibility to cracking. Our objectives are to determine the cause of the cracking; the dependence of the rate on temperature, time, and concentration; and a reasonable solution to the prob- lem. 12.1 EXAMINATION OF HASTELLOY N COMPONENTS FROM THE MSRE B. McNabb H. E. McCoy Several Hastelloy N components from the MSRE have been examined. Partial presentations of our findings were made previously,'*? and additional results will be presented in the current progress report. Some work is still in progress, and it is likely that further observations and interpretations of prior observations will continue to be made for some time. 12.1.1 Freeze Valve 105 The freeze valve that failed during the final shutdown of the MSRE (FV 105) was examined further. The 1. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 147-66. 2. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 89-106. failure was attributed to fatigue from a modification that had been made. The valve was made by flattening a section of 1'%-in. sched 40 Hastelloy N pipe and welding an air cooling shroud around the flattened section. This valve was used to isolate the drain tanks and was frozen only when the salt was in the drain tanks. The salt was static except when the reactor vessel was being filled. It was filled with salt and maintained above 500°C for about 21,000 hr. The valve was filled with salt from the drain tanks, so the fission product concentration that the valve was exposed to was considerably lower than that seen by components in the primary circuit. Thus the freeze valve was exposed to another set of conditions involving fuel salt and fission product concentration and may help in separating the effects of each on the behavior of Hastelloy N. Three rings % ¢ in. wide were cut from the pipe away from the flattened section and were pulled in tension in the same manner as previously described for the control rod thimble rings.®> One rectangular piece was cut and bend-tested with the ID of the pipe in tension. Table 12.1 is a tabulation of the observed mechanical prop- erties. The yield stress was essentially unchanged, and the ultimate stress was reduced about 15% from the vendor’s certified properties. The elongation was re- duced considerably but was still greater than 25%. A gage section is difficult to define in a ring test, so crosshead travel and reduction in area are reported for 3. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 149. Table 12.1. Results of mechanical property tests on specimens from FV 105 (heat 5094) at 25°C and a deformation rate of 0.05 in./min Stress (psi) Crosshead Reduction Type of test ] Ultimate travel in area Yield tensile (in.) (%) Vendor’s, tensile 45,800 106,800 52.6 Ring, tensile 45,800 89,700 0.72 25 Ring, tensile 483900 90,100 0.59 29 Ring, tensile 41,900 90,300 0.73 37 Wall segment, bend 71,300 0.41 334 2Maximum strain in outer fibers. the rings from FV 105 rather than percent elongation in 1 in. The bend test was discontinued due to strain limitations of the bend fixture after 0.41 in. crosshead travel, which corresponds to 32.7% strain in the outer fibers of the specimen and to a 90° bend angle. The yield stress calculated from the forces on the bend specimen is too high because elastic formulas were used in the calculation and the specimen deformed plas- tically, but this calculated quantity is useful for 110 comparison with other bend tests such as those re- ported previously for the mist shield.* Figure 12.1 is a macrophotograph of the tension side of the bend specimen from FV 105. Some very fine shallow cracks are visible on the tension surface, and cracking is visible at the edges in the burrs remaining 4, B. McNabb and H. E. McCoy, MSR Program Semiannic. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 90. rR-56852 ® g ’ e T | i W 1] “p gn T - b ity i iy P, £ - !:_1._'-"“ fl-"‘*; - Fig. 12.1. View of tension side of a bend sample from FV 105. The tension side was exposed to static salt. 111 Fig. 12.2. Composite of photomicrographs of a ring from FV 105 that was pulled in tension. The upper surface was exposed to salt and the lower surface to the cell environment of N, plus 2 to 5% O,. Only the surfaces were photographed. from the remote cutting operation. One of the three tensile-tested rings was examined metallographically. Figure 12.2 is a composite of photomicrographs of a section through the specimen showing the inside sur- face, which was exposed to fuel salt (top), and the outside, which was exposed to the cell environment of nitrogen plus 2 to 5% O,. The reason for the uneven nature of the oxide on the outside is not known. Possibly it was due to corrosion after the leak, although the rings were cut approximately 4 in. away from the nearest visible residue from the salt leak. Figure 12.3 is 500X photomicrographs of the oxide in one of the worst areas. The cracks tend to blunt and do not penetrate into the metal beyond the oxidized surface. Figure 12.4 is 500X photomicrographs of the inside of the pipe exposed to fuel salt. There is about 1 crack per grain, or 240 cracks per inch, but the cracks are shallow and blunt, having an average depth of 0.75 mil and a maximum depth of 1.5 mils. Figure 12,5 is a 40X photomicrograph of the fracture and shows that a large amount of strain occurred before fracture. From these tests we see that the mechanical proper- ties of the Hastelloy N in the freeze valve were not degraded seriously by the exposure to fuel salt with some fission products for a long period of time. Numerous intergranular cracks were present in the surfaces exposed to the salt. These cracks were similar to those in the surfaces from the primary circuit but were shallower. 12.1.2 Control Rod Thimble The control rod thimble was near the center line of the MSRE core for the duration of operation. Because it was exposed to the peak neutron fluence of any component and to a relatively high concentration of fission products, it has been of much interest and has been studied rather thoroughly.®'® The additional studies during this reporting period investigated a possible effect of salt flow rate on the severity of cracking. The details of construction of the thimble are shown partially by Fig. 12.6. The thimble was made of 2-in.-OD, 0.065-in.-wall tubing with occasional spacers. The spacers had small ribs machined on them to ensure that salt could flow between the thimble and the adjacent graphite. The spacers were attached to the thimble by a small weld bead on the thimble that was made through a clearance hole in the spacer. Thus the spacer was restrained vertically but could move some radially. The shop drawings allowed a maximum diam- etral clearance of 15 mils between the spacer and the thimble. It is likely that most of the annulus was filled with salt, but the flow rate should be very slow. Thus a comparison of the cracking tendencies under the spacer, where flow was restricted, and outside the spacer should give an indication of the sensitivity of the cracking to flow rate. A composite of photomicrographs of a deformed ring from the thimble that was exposed to flowing salt is shown in Fig. 12.7. The inside of the thimble was exposed to the cell environment of N, plus 2 to 5% O, and was oxidized. Cracks formed in the oxide but did not penetrate the metal. The outside of the thimble, which was exposed to fuel salt, had 192 cracks per inch, having an average depth of 5.0 mils and a maximum depth of 8.0 mils. A similar ring was cut from the thimble under the spacer. It was deformed and ex- amined metallographically. A composite of photo- micrographs is shown in Fig. 12.8. This sample had 257 cracks per inch with an average depth of 4.0 mils and a maximum depth of 8.0 mils. A ring was also cut from the spacer for testing and examination. This sample was exposed to flowing salt on one side and restricted salt flow on the other side. A composite of photomicro- graphs of this sample is shown in Fig. 12.9. The side "5, B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 147-51. 6. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 89-106. 112 o1 in. 6.0 N 0OX 0.003 n. 0.007 INCHES (a) f———— c.007 m | [F 005 n. [0.001 in. e T 500X 10.003 in. 0.007 INCHES i 10.005 in. i, . o 3 o | . e e A“ . - * . ‘y * * R v ; - . N 1 L4 - ! -~ ..,w_“wqg.\‘ . # E * Yo, e T Ty . L b o N‘M - . . o - - - ——— > S "F # - bnt— 10.007 in Fig. 12.3. Photomicrographs of a tensile sample of FV 105 showing the oxide that formed on the outside of the pipe. The cracks penetrated only the depth of the oxide. (@) As polished. (b) Etched with lactic, HNO3, HCL 113 R-56856 lO.O"fl n. HES D.C0O7 1M 10005 in. T e IC.007 in. et | 10.001 in. T 10.003 in. 500X Q.007 INCHES T 10,005 in. — 10.007 in, Fig. 12.4. Photomicrographs of the salt side of a tensile specimen from FV 105. (a) As polished. (b) Etched with lactic, HNO3, HCL 114 R R B ~ R-56865 Fig. 12.5. Photomicrograph of the fracture of a tensile specimen from FV 105. The left side was exposed to fuel salt and the right side to the cell environment. 40X. Etched with lactic, HNO3, HCL R-54116 Fig. 12.6. Portion of Hastelloy N control rod thimble removed from the MSRE. The cut end was near the axial center of the reactor. The bottom end was near the bottom of the core. The sleeves are Hastelloy N spacers and were held in place by a small weld bead. Fig. 12.7. Deformed ring from the MSRE control rod thimble. This sample was exposed to flowing salt. Fig. 12.8. Deformed ring from the MSRE control rod thimble. This sample was under the spacer and was exposed to restricted salt flow, 17X. . R-55474 Fig. 12.9. Control rod thimble spacer exposed to flowing salt on one side and to almost static salt on the other side. Deformed at 25°C. exposed to flowing salt had 178 cracks per inch with an average depth of 3.0 mils and a maximum depth of 7.0 mils. The side exposed to restricted salt flow had 202 cracks per inch with an average depth of 3.0 mils and a maximum depth of 5.0 mils. Although there are differences in the crack numbers and depths on the various surfaces, we do not feel that they are signifi- cant. Duplicate samples from the same locations as the mechanical property samples were dissolved and ana- lyzed for fission products. The concentrations of tellurium in atoms/cm? X 10'7 are given with each sample location: (1) thimble, flowing salt — 2.9, 2.9; (2) thimble, restricted flow — 0.95, 0.74; and (3) spacer — 1.6, 4.5. The tellurium concentration under the spacer, where the flow was restricted, is lower than on the bare thimble by a factor of 3 to 4. The concen- tration on the spacer sleeve is not detectably different from that noted on the bare thimble. The other fission products that were analyzed showed similar trends. Thus the severity of cracking was not influenced appreciably by fission product concentrations that varied by factors of 3 to 4. 12.1.3 Sampler Cage Rod The sampler cage rod from the pump bowl that was tensile tested to failure at room temperature showed marked variations in the severity of cracking at surfaces exposed to fuel salt and those in the gas space above the salt.” Pan photomicrographs were made of the surfaces of the rod exposed to liquid fuel salt and to the gas above the salt. Figure 12.10 shows the variations of cracking along the length and on opposite sides of a %,-in.-long segment of the rod from the fracture (right), 7. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 96-97. extending toward the bottom of the pump bowl (left). The fracture occurred in the area of the largest deposits on the rod, which was probably the average liquid level in the pump bowl.® Unfortunately, the radial orienta- tion of the rod was not maintained during mounting. Additions of beryllium and uranium by means of the sampler were made occasionally to adjust the chemistry of the fuel salt, and local conditions in the sampler cage could have been quite different from those in the bulk salt stream. This might account for some of the variations of cracking around the circumference of the rod. Figure 12.11 is a pan photomicrograph of a %;-in. segment of the same rod starting % in. above the fracture (left) and extending farther up into the gas region (right). The surface cracking is diminishing in severity, but the frequency is still almost one crack per grain. The sampler cage rod represents about the worst conditions of attack of any of the components ex- amined in the MSRE, with cracks opening up to a maximum depth of 13 mils. However, the mechanical properties were not degraded seriously, and the ob- 116 served small property changes may have been due to the long time that the rod was held at high temperature. Examination of this component shows clearly the severe cracking that occurred in the pump bowl. It also shows that the cracking diminishes in traversing from the liquid into the sheltered gas region inside the mist shield. 12.1.4 Mist Shield The mist shield was a spiral baffle of ';-in. Hastelloy N sheet whose purpose was to keep salt spray from the region where fuel salt samples were taken in the pump bowl. Four bend test specimens were cut from portions of the shield that were exposed to different conditions of fuel salt flow or spray. Macrophotographs and the mechanical properties of the bend specimens from these various locations appeared in the last semiannual 8. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 7681, 150-53. Fig. 12.10. Photomicrographs of a sampler cage rod deformed to failure at 25°C. The fracture is on the right, and the sample extends farther into the salt from right to left. 16.3X. R—56470 Fig. 12.11. Photomicrographs of a portion of the sampler cage rod that was fractured at 25°C. The left end was s in. above the fracture, and the sample extends farther into the vapor space from left to right. 14.3X. 117 Fig. 12.12. Bend specimen S-62 from the outer gas region of the mist shield. The top was in tension and the bottom portion in compression. report.” At that time only twg of the specimens had been examined metallographically, but it was evident that there were considerable differences in the fre- quency and severity of the cracking in the specimens exposed to the restricted salt flow and to the gas inside the mist shield. During this report period we examined the two samples that were taken from the outer end of the spiral shield at top and bottom. Bend specimen S-62 came from the outer top portion of the mist shield, where it had been exposed to fuel salt mist. This specimen was bent, with the surface that had been on the outside in tension, until it fractured. Figure 12.12 shows the tension (top) and compression (bottom) surfaces of the specimen. The depth of cracking was only about 5 mils (except for the fracture), but there was a tendency for grains to become dislodged near the fracture (right). Cracking was confined to a relatively small area around the fracture, probably due to the relatively low fracture strain of 11%. Figure12.13 shows photomicrographs of the tension () and compression (b) sides of this bend specimen. Bend specimen S-68, from the bottom outer region of the mist shield, had been immersed in salt that was agitated and may have contained bubbles and materials from the surface of the salt pool that were carried under by the xenon stripper jets. A composite view of this specimen is shown in Fig. 12.14. The tension side is shown at the top and the compression side at the bottom. The cracking on the tension side is spread over a slightly larger area than in specimen S-62, and there is less tendency for grains to become dislodged. The 9. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 89-95. cracking depth is about the same except for one crack about 12 mils deep. Figure 12.15 shows photomicro- graphs of the tension (2) and compression () sides of bend specimen S-68. The outside of the spiral was the tension side of the bend test, but the compression side was also exposed to agitated fuel salt. Comparison of the observations on these two speci- mens from the outer portions of the mist shield spiral and those reported previously from the inner end of the spiral’® show that the cracking severity was greatest in the outer gas sample, next most severe in the liquid samples from inside and outside the shield, and least severe in the sample from the inner gas region. 12.2 AUGER ANALYSIS OF THE SURFACE LAYERS ON GRAPHITE FROM THE CORE OF THE MSRE R. E. Clausing We continued to use Auger electron spectroscopy to analyze deposits of fission products and/or corrosion products on graphite surfaces from the core of .the MSRE. This technique offers unique capabilities for analyses of the first few atomic layers of a surface. The principles of construction and operation of the equip- ment have been described,'® and results of analyses of two samples from a core moderator element have been reported.’! In these samples, relative concentrations of the deposited elements were determined as functions of depth below the original surface. Molybdenum, nio- 10. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 143—45. 11. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 107-10. 118 R-56759 [ w" !J“"\j ; 9 I g 9 o o 3 0 P L . 1 R-5_6?64 T 0.035 INCHES N 100X |7 Fig. 12,13. Photomicrographs of bend specimen $-62 from the outer gas region of the mist shield. (¢) Tension side. (b) Compression side, Etched with lactic, HNO3, and HCL Fig. 12.14. Photographs of bend specimen S-68 from the outer liquid region of the mist shield. Top portion was in tension, and the bottom was in compression. As polished. bium, rhodium, and technetium were present in amounts estimated to exceed 5% of the exposed surface layer. Tellurium and ruthenium analyses were difficult because of interference from other elements, but both elements appeared to be present in significant amounts. Neither iron nor nickel contents were determined for these first two samples. Analysis of these samples and of several others has now been completed. All of the results are summarized in Table 12.2. Sample No. 6, the first analyzed, was taken from a core surveillance specimen removed from the MSRE in April 1968 at the conclusion of operation with 235U fuel. The other samples were all taken as 0.3-in.-diam plugs by core drilling various parts of the graphite moderator element 1184-C-19, which was removed from the MSRE after shutdown in December 1969.11 The data in Table 12.2 are normalized relative to a carbon peak of 100, and peak heights are given in arbitrary units. The data are not yet quantitative due to lack of suitable standards, and it should be noted that because the sensitivity is different for different ele- ments and for different peaks for the same element, peak height cannot be equated to concentration. (We intend to make our data quantitative by use of suitable standards that are now being prepared.) Samples 5, 7, and 10 were taken from the flow channel: sample 5, 3 in. from the top; sample 7, in the center; and sample 10, 3 in. from the lower end of the element. Samples M6 and 13 were taken from the edge of the graphite moderator element (outside the main flow channel) with sample 13 about 3 in. from the bottom end of the element and sample M6 about halfway along the element length. Sample 10 was examined in the scanning electron microscope after sputtering approximately 16 atom layers from the surface, and typical photographs are shown in Fig. 12.16. Note that the sample surface is fairly smooth but has a few small particles adhering to it. These particles appear to be graphite dust, still adhering from the core drilling operation, which should not interfere with the Auger analysis. The smoothness of the surface should ensure the uniformity of removal of layers from the surface by sputtering. All of the samples examined had substantial amounts of molybdenum, technetium, sulfur, and rhodium. Iron, nickel, and chromium were present on most of the samples, with the two samples from the bottom of the moderator element (samples 10 and 13) having the largest concentrations. Tellurium is probably present on all of the samples and is discussed briefly below. Peaks which may be associated with antimony were detect- able on samples 10 and 13, although the peaks are not listed in Table 12.2. An unresolved pair of peaks tentatively identified as due to palladium were detected at 330 eV on nearly all of the samples. Tellurium, chromium, and ruthenium concentrations are difficult to determine because of interference from strong lines 120 0.035 INCHES N WO | ‘] 4 P S o oS Tro 0.035 INCHES 100X |17 Fig. 12.15. Photomicrographs of bend specimen S-68 from the outer liquid region of the mist shield. (a) Tension side. (b) Compression side. Etchant: lactic, HNO3, and HCL Table 12.2. Comparison of relative Auger peak heights obtained from surfaces of graphite samples removed from the MSRE core Estimated Technetium, Technetium Niobium Rhodium Palladium Sample depth Tellurium molybdenum, molvb denur;1 (199) (303) (331D Nitrogen Oxygen Iron Nickel No. (atom (20 eV) sulfur (1}2132 V) (202) (308) (336) (383 eV) (510eV) (650eV) (859 ¢V) layers) (150 eV) (200eV) (300 eV) (330 eV) 6 0 na 29 9 2 na na na na na na (surveillance specimen) 10 na 5 23 4 na na na na na na 25 na 4 16 3 na na na na na na 7 0 na na na na na na na na na na (flow channel, center) 15 na 19 40 nd 9 na 12 25 na na 30 na 26 50 nd 11 na 17 26 na na 5 0 na 13 5 w 4 w 2 14 nd w (flow channel, top) 15 1,600 35 16 w 5 1 13 37 2 1 10 0 na na na na na na na na na na (flow channel, bottom) 16 4,000 63 13 w 2 w 13 96 3 2 M6 0 na 15 21 w 1 w 8 24 nd nd (edge, center) 15 nd 24 15 w 2.5 w 5 19 nd nd 13 0 4,300 13 14 2 6 2 15 64 5 1 (edge, bottom) 15 16,000 131 24 5 8 w 17 97 10 11 w = weak. nd = not detected. na = not analyzed. 1¢1 Ty Y - 1344 122 Fig. 12.16. Scanning electron micrographs of the surface of a sample from an MSRE moderator element after sputtering approximately 16 atom layers from the surface. (a) 20X. (b) 500x. of oxygen, carbon, and molybdenum. The largest amounts of tellurium and chromium are probably similar in quantity to the largest amounts of iron. As of now, we believe that ruthenium is probably present, but the lines are not nearly as strong as those for molyb- denum. Low-energy Auger peaks like that at 20 eV for tellurium are frequently very strong; thus, the large amplitudes listed in Table 12.2 need not be associated with high surface concentrations of tellurium. Pure tellurium would have several strong peaks between 400 and 500 eV, but our data do not disclose strong lines on any of the recorded spectra at these energies. 12.3 AUGER ANALYSIS OF THE SURFACE OF A FRACTURED HASTELLOY N SAMPLE The surface of a Hastelloy N foil that was attached to a retaining strap on a group of surveillance specimens in the MSRE is being analyzed. After being exposed in the core at operating temperature for 7203 hr the foil fractured intergranularly with little or no ductility. Part of this strap is being examined extensively by other techniques. Figure 12.17 shows scanning electron micrographs of the fractured foil. The surface has the faceted appearance of an intergranular fracture and shows some discrete particles in the boundaries. These particles probably are not responsible for the brittle behavior; a very thin layer of brittle material spread more uniformly over the grain boundaries is a more likely cause of a brittle fracture like that observed. An attempt is being made to determine if the agent responsible for the brittle behavior can be identified on the fracture surface with the use of Auger spectroscopy. Table 12.3 shows some preliminary Auger data from the flat surface of the 0.004-in.-thick foil and from the fracture surface. The data shown were obtained on surfaces sputtered only very lightly so that only about two atomic layers were removed from the as-received surface. The major difference seems to be the substan- tially larger portion of molybdenum on the fractured — e — — 123 Y-—112253 T Y-142254 S e RPN Fig. 12.17. Fractographs of Hastelloy N foil fractured after exposure to the MSRE core for 7203 hr at operating conditions. (a) 500X. (b) 1000X. Table 12.3. Auger electron peak intensities? from a edge, which is likely due to the molybdenum-rich Hastelloy N fracture surface carbides that form along the grain boundaries of Hastelloy N. No tellurium was detected on either Intensity surface. Blement eflfgjf& | Fractured Flat surface The conditions under which this sample was fractured surface of foil and subsequently handled, including decontamination by ultrasonic solvent cleaning and examination in the Tellurium 20 nd® b relatively poor vacuum of the oil-pumped scanning hipkel: o 34 £39 microscope, make it quite possible that material Technetium, 150 100 160 = : Saotvkdeniin: strongly concentrated within the first few atomic layers sulfur of the original fracture surface was no longer present at Molybdenum 222 25 35 the time of the Auger analysis or was covered with Technetium, 182 43 65 other material so that it was no longer detectable. molybdenum Therefore, we plan to fracture another piece of the g;f:;l;m ggg 2? 2; same foil in the Auger system under ultrahigh vacuum Rhodiun 300 b b conditions and analyze it immediately. This technique Palladium 330 b b has been used to identify embrittling agents in the grain Nitrogen 383 3 10 boundaries of tungsten and should be useful for Oxygen 510 10 54 Hastelloy N if the fracture is through the embrittled :i(:iel ggg 32 22 region, as it appears to be. Shtoniletn 2 > s 12.4 INTERGRANULAR CORROSION OF Chromium 569 c ¢ HASTELLOY N @Peak intensiti in arbit it J. W. Koger €aK Intensities are in aroitrary units. ; p Dnd = not detected. We are now using loop NCL-16 [Hastelloy N circu- ‘w = weak. lating LiF-BeF,-UF4(65.5—-34.0-0.5 mole %)] in the study of intergranular cracking of Hastelloy N in molten salts. Specifically, we are investigating the possibility that the attack is related to the localization of normal corrosion processes to grain boundaries. In any solid solution alloy where there is a difference in nobility of the constituents, oxidation-reduction reactions may result in removal of the least noble constituent, with attack being preferential along grain boundaries. In time, given a continuing electrochemical process, this will lead to crevices in the grain bound- aries. Diffusional processes within a crevice may lead to its broadening and ultimately to the formation of pits. However, if the root of the crack is anodically polarized relative to the walls, knifeline attack will continue. Such a condition may arise if the walls of the crevice become covered with a very noble material (nickel or molybdenum). This covering by a noble constituent can occur either by the noble material remaining on the wall when the least noble constituent is removed or by dissolution of all the alloy constituents with subsequent precipitation of the more noble constituents, Prior to its use in the cracking studies, loopNCL-16 had operated for 29,500 hr with a fuel salt circulating in the system. The maximum weight loss after this period was 2.9 mg/cm?, and the largest weight gain was 1.7 mg/cm?. Assuming uniform loss, the maximum corrosion rate was 0.04 mil/year. The chromium con- tent of the salt had increased 500 ppm, and the iron had decreased about 100 ppm in 29,500 hr. Titanium- 124 modified Hastelloy N specimens (Ni—12% Mo—7% Cr—0.5% Ti) had smaller weight losses than standard Hastelloy N specimens (Ni—16% Mo—7% Cr—5% Fe) under equivalent conditions. For our study of cracking we initially added 500 ppm FeF, to the loop. Specimens were removed, weighed, and portions of specimens examined metallographically 450 and 1100 hr after the first addition. Then an additional 500 ppm FeF, was added. Specimens were then analyzed 800 and 1800 hr after this second addition, with the total exposure to the highly oxi- dizing salt being 2900 hr. After each removal we found weight changes typical of all our temperature-gradient mass transfer systems, with weight losses in the hot section and weight gains in the cold section. Figure 12.18 shows the weight changes of selected specimens as a function of time, and Fig. 12.19 shows the changes completely around the loop. Note that the balance point (the point at which there is no weight change) did not shift. Note, also, that the weight changes after the FeF, additions were relatively large. The changes during the 450 hr after the first addition equaled those during the previous 10,000 hr. Weight changes during the next 650 hr were 2 or 3 times those for the first 450 hr and were larger than those obtained during 29,500 hr of operation before any additions. Metallographic examination after the initial FeF, additions disclosed grain boundary attack which altered ORNL-DWG 72-2021R 10 | 7 :, 545°C y S 540°C & o % : 2 — £ /- - o ’ \| \+ 6 5 e76°C” |/ AT g 650°C*] | 5 698°C |_ \ I S 0 b — ] J = ADDED 500 ppm FeF, l\ e socmn e TN ‘ | A -15 i T3 * STANDARD ALLOY, ALL OTHERS MODIFIED WITH 0.5% Ti & oo L L | l | | 0 8000 16,000 24,000 32,000 OPERATING TIME (hr) Fig. 12.18. Weight changes of Hastelloy N specimens exposed to LiF-BeF,-UF, (65.5-34.0-0.5 mole %), with FeF, added, in NCL-16 as a function of time and temperature. ORNL-DWG 72-2020 s ‘ | T T — = \ | | I | | | 10— : L ; ? J—_L S - B T * ; HEATED AND INSULATED —~ 5 }wlgl : ‘ S S S < ] T o . S — = | E g9 - w l ‘ ‘ ¢ 9086 h o % ‘ ‘ | o 19656 hrr | ‘ _L . AN xI : ! 5 _s J(' | . ®29509hr __ J‘*_ ) Y300 - | ‘ FeF, ADDED . ‘ ~ 2 \ } l o %99%7 hhr } | ; L ! < 30606 ; > = -0 —T —j— FeF, ADDED —] ;—F— | 100 ! \ * 31445 nr ‘ > L 832396 hr | { | ! g ~15 - ( “ SO f — [ - = —TéJr 100 % ! | : ! ! . — 0 10 20 30 a0 50 60 70 80 90 100 DISTANCE (in.) — e - UPPER COLD LEG BEND LOWER BEND HOT LEG CROSSOVER {VERTICAL) CROSSOVER (VERTICAL) Fig. 12.19. Weight changes of Hastelloy N specimens exposed to LiF-BeF,-UF, (65.5-34.0-0.5 mole %), with FeF, added, in NCL-16 as a function of position and time. Y-110485 i_. Tro Tox 0.007 INCHES 1>7500x% Tom B Fig. 12.20. Hastelloy N exposed to clean LiF-BeF,-UF, (65.5-34.0-0.5 mole %) salt at 700°C for 29,509 hr, to salt containing 500 ppm FeF, for 1100 hr, and to salt containing another addition of 500 ppm FeF, for 1700 hr. The weight loss was 19 mg/cm?. As polished. 500X. the polishing characteristics of the specimen, but no cracks were visible. Examination of the hottest speci- men 800 hr after the second addition revealed more grain boundary attack but still no cracks. The surface of the specimen was “lacy’” due to severe corrosion by the salt and chromium removal from the alloy. This specimen was bent, and some cracking was induced in the depleted area, but no cracks penetrated the matrix. Specimen examination after the total 2900 hr exposure disclosed that the weight losses were six times greater than in the previous 29,500 hr operation of the loop, and cracks were now visible to a depth of 0.5 mil. The cracks (Fig. 12.20) seemed to be similar to those seen in the MSRE samples but were much shallower. We have also examined pieces from loop 1249, which "was operated about 12 years ago to determine the diffusion coefficient of chromium in Hastelloy N at high temperature. It operated with an NaF-ZrF, salt for 792 hr, and then 3720 ppm FeF, was added. Operation continued for 264 additional hr. We took two pieces of the loop piping, one from the hottest position (927°C wall temperature and 860°C bulk fluid temperature) 126 and one from the cold leg (682°C), and bent them to determine their cracking tendencies in the chromium- depleted regions. The cold-leg specimen did not crack, but cracks formed in the hot-leg specimen. The longest extended 3 mils into the sample (Fig. 12.21), and several small cracks about the depth of a grain were also visible. 12.5 TUBE-BURST EXPERIMENTS H. E. McCoy J. W. Koger Tube-burst experiments at 650°C were run in three environments to determine whether stress or environ- ment had a detectable effect on the fracture character- istics of Hastelloy N. Two sets of specimens were tested: one with clean surfaces, the other with 0.01 mg/cm? of tellurium electroplated on the surfaces. The tubular specimens were made from 1-in.-OD, 0.065-in.- wall tubing with a gage section 3 in. long and 0.020-in. wall. The material was from heat Y-8488 and was similar chemically to heat Y-8487, which was used to fabricate the control rod thimble. The samples were Y-109638 0.018 INCHES f Fig. 12.21. Hastelloy N exposed to equimolar NaF-ZrF, at 860°C for 792 hr before and 264 hr after an addition of 3720 ppm FeF,. Specimen was then bent so this surface was in tension. Etched with glyceria regia. 200X. fabricated and given a 1-hr anneal at 1180°C before insertion into the test chambers. The test environments were helium, a fuel salt of composition LiF-BeF,- ZrF4-UF, (65.4-29.1-5.0-0.5 mole %), and the fuel salt with an addition of 500 ppm FeF, (300 ppm Fe). Posttest analyses of the salt charges showed that the clean fuel salt contained 110 ppm Fe and 89 ppm Cr and that the fuel salt with FeF, contained 284 ppm Fe and 120 ppm Cr. The rupture lives of unplated tubes in the various environments are compared in Fig. 12.22. The rupture life is independent of environment, except perhaps at the lowest stress level. Even at this stress, the rupture lives vary by only a factor of 2, and additional observations would be necessary to establish whether the difference is reproducible. The tubes were exposed to a two-dimensional stress, and the effective-stress— effective-strain criteria would predict that for the same maximum principal stress, the time to rupture for a specimen under biaxial (2:1) stress should be about double that for a specimen under uniaxial stress.!?>!'3 The data in Fig. 12.22 indicate the opposite trend. There is also a large difference in slope. The uniaxial data were obtained primarily from heats from another vendor, and there were differences in chemical compo- sition that may account for the differences in proper- ties. The uniform strains of the unplated tubes were divided by the rupture life to obtain effective minimum creep rates, shown in Fig. 12.23. There are no ob- servable effects of environment outside of what is considered reasonable data scatter. The creep rate of a biaxially (2:1) stressed sample should be about 0.4 times that of a uniaxially stressed sample. The results in Fig. 12.23 show the opposite trend and indicate that the tubing is weaker than the bar and plate material used to obtain the uniaxial data. The fracture strains of the unplated tubes are sum- marized in Table 12.4. The most important observation is that there is no detectable effect of environment on the fracture strain. Generally, the fracture strain de- creases with increasing rupture life. This is the normal trend, but the range of 10 to 2% for rupture lives of 20 to 1000 hr is larger than that noted for bar and plate stock.!® The differences between the uniform and total 12. C. R. Kennedy, “The Effect of Stress State on High- Temperature Low-Cycle Fatigue,” Amer. Soc. Testing Mater. Spec. Tech. Publ. 338,92—107 (1963). 13. H. E. McCoy, Jr., and J. R. Weir, Jr., In- and Ex-Reactor Stress-Rupture Properties of Hastelloy N Tubing, ORNL- TM-1906 (September 1967). 127 ORNL-DWG 72-3165 70 e ~. T ~ AVERAGE UNIAXIAL DATA : \\‘ (AIR ENVIRONMENT) I | |oH A i | 40—-»—-*—-k**—¢, “|': . il H 3 A 50 ——-»vxw\r ‘L_\ '4 T}Tfi m\ i (] & He - | la SALT Do f “f‘ 20 .. O SALT + FeF, N R — 'OPEN POINTS - CLEAN TUBES o { ‘CLOSED POINTS Te COATED MAXIMUM PRINCIPAL STRESS (1000psi) 2 5 10! 2 5 102 2 5 10 2 RUPTURE TIME (hr) Fig. 12.22, Stress-rupture properties at 650° of INOR-8 tubes in various environments, ORNL-DWG 72-7664 ~ (@] o] (@] AVERAGE UNIAXIAL DATA ¢)] o D o W O n O MAXIMUM PRINCIPAL STRESS (1000 psi) s} LT.\TA ] w* ‘_-SZLT i-SALT+FeF4 i 0% 2 5 10°2 5 1G22 5 15" 2 10 | il ~TUBE RESULTS ‘ ‘ #%M( | fifp: = A CREEP RATE (%/hr) Fig. 12.23. Creep rates at 650°C of unplated Hastelloy N tubes in various environments. elongations are small for most samples and indicate very little third-stage creep. This was also indicated by the fact that the failures in all but one of the tubes were “pinhole” fractures that were difficult to locate and exhibited very little bulging at the fracture. Metallographic examination of the tubes revealed that the fractures were entirely intergranular (as is normal at 650°C). The metallographic features seemed inde- 14. H. E. McCoy, “Variation of the Mechanical Properties of Irradiated Hastelloy N with Strain Rate,” J. Nucl. Mater. 31, 72 (1969). Table 12.4. Fracture strains of unplated tubes tested at 650° C? Stress (psi) Helium Salt Salt plus FeF, 47,000 9.50 (9.92) 11.4 (11.8) 9.98 (11.0) 40,000 5.98 (6.12) 6.44 (7.39) 6.47 (6.56) 35,000 4.05 (10.2) 3.78 (3.81) 2.99 (3.24) 30,000 1.86 (1.90) 1.82 (4.33) 1.48 (1.61) @First number is uniform strain, and numbers in parentheses are maximum strain at the fracture. Strains are given in percent. pendent of the test environment. Typical photo- micrographs of the 30,000-psi test sample from each environment are shown in Figs. 12.24-12.26. The stress-rupture properties of the tellurium-plated tubes are shown in Fig. 12.22. The data from these tubes exhibit more scatter, but the rupture lives are all shorter than those noted for tubes not plated with tellurium. Test environment did not have a detectable effect on the rupture life. The posttest examination of these tubes has not been completed, but two of the tubes tested in fuel salt have been examined metal- lographically. A typical photomicrograph is shown in Fig. 12.27. Numerous intergranular cracks extend from one-third to one-half through the tube wall. This cracking occurred consistently around the tube. 12.6 CRACKING OF SAMPLES ELECTROPLATED WITH TELLURIUM B. McNabb H. E. McCoy We have developed a method of coating surfaces with tellurium by electroplating. The tellurium is dissolved in hot concentrated nitric acid, evaporated to dryness, and ammonium hydroxide added to obtain the desired volume. Platinum gauze is used as the anode, and the sample to be plated is used as the cathode. The plating potential used is 24 V dc. Plating current is dependent on several factors, such as specimen size, anode size, and distance between the electrodes, but typically it is about 1 mA on a specimen with a surface area of 6 cm?. A specimen of vacuum-melted Hastelloy N (heat 2477) was plated with 1 mg/cm? of tellurium under these conditions. The weight gain was linear with time at a rate of 0.01 mg cm™ min~!. After being plated, the specimen was annealed 65 hr at 650°C in argon, then bend tested at room temperature. Figure 12.28 is a photomicrograph of the center of the tension side of the bend and shows numerous cracks extending to a 128 maximum depth of about 3 mils and an average of about 1.5 mils. This demonstrated that tellurium could cause cracking under these conditions, with a fairly high concentration of tellurium but short exposure time. Approximately 40 sheet specimens were prepared for plating to determine which alloys were susceptible to cracking by tellurium. The materials included (1) standard and modified Hastelloy N; (2) Hastelloys B, C, X, and W; (3) Inconels 600, 601, and 718; (4) Incoloy 800; (5) stainiess steel types 304, 310, 316, 406, 410, 502, and 17-7 PH; (6) nickel; (7) copper; (8) colum- bium and Cb + 1 Zr; (9) Mo—0.5 Ti; and (10) René 62. These specimens were plated with approximately 0.01 mg/cm? of tellurium (equivalent to about 100 ppm uniformly distributed throughout a layer S mils deep over the entire surface area). They were then encapsu- lated in stainless steel and annealed 207 hr at 650°C in argon. After annealing, they were bend tested at room temperature to a bend angle of 90°. None of the standard or modified Hastelloy N specimens showed any cracking under these conditions. Figure 12.29 is a photomicrograph of the same heat 2477 that exhibited cracking at the higher concentration of tellurium (1 mg/cm?) but did not exhibit cracking at 0.01 mg/cm?. Some materials did crack under these conditions. Hastelloy W, type 406 stainless steel, and Mo—0.5 Ti cracked completely through the specimen. Control specimens with no plating are being prepared for comparison with these, since it is likely that these materials may crack with no tellurium present. Speci- mens are being prepared with higher concentrations of tellurium and will be annealed for longer times to determine the threshold concentration that will cause cracking in Hastelloy N. 12.7 CRACKING OF HASTELLOY N BEING CREEP TESTED IN TELLURIUM VAPOR H. E. McCoy B. McNabb A sample of Hastelloy N was stressed at 30,000 psi at 650°C in an environment of argon containing a small partial pressure of tellurium. The tellurium vapor came from a small vial of tellurium metal at 550°C. When the specimen was removed for examination after 900 hr, the quartz vial, which had initially contained 300 mg Te, had only 75 mg remaining, and the specimen was coated with fine whiskers of unidenti- fied material. (We did not collect enough of this material for identification.) The stressed portion of the sample had cracks that were visible to the naked eye, and a metallographic section revealed numerous inter- granular cracks, with some extending to a depth of 25 129 Y-110880 0.035 INCHES N 100X Y-110883 ) Y-110882 500t m 10.003 in. 3.007 INCHES 500X T ‘[0.005 in. ; { g (c) ~ T [0.GQ7 in. Fig. 12.24. Photomicrographs of unplated Hastelloy N tubing stressed at 30,000 psi in a helium environment at 650°C. (a) Fracture, as polished. (b) Fracture, etched with glyceria regia. (¢) OD near fracture, as polished. 130 Y-110888 = 215 Y-110889 ol o2 (o] B Y~ 110891 fi__ 5 ji=] : < [S] 0.007 INCHES 500X 1 Fig, 12.25. Photomicrographs of unplated Hastelloy N tubing stressed at 30,000 psi in clean salt at 650°C. (¢) Fracture, as polished. (b) Fracture, etched with glyceria regia. (¢) OD near fracture, as polished. The salt composition was LiF-BeF ;-Z1F 4-UF, (65.4-29.1-5.0-0.5 mole %). 131 Y- 110896 = 0.035 INCHES I 100X Y-110897 I ,,,,, N | \ Y-11089¢8 T0,00{ in T 16,003 in. 500X T - 0.007 INCHES 10.005 in. T (c) [0.007 in. Fig. 12.26. Photomicrographs of unplated Hastelloy N tubing stressed at 30,000 psi in salt containing FeF, at 650°C. (a) Fracture, as polished. (b) Fracture, etched with glyceria regia. (¢) Outside edge at fracture, as polished. The salt composition was LiF-BeF,-Z1F4-UF4 (65.4-29.1-5.0-0.5 mole %) with the addition of 500 ppm FeF,. 132 0.035 INCHES o 100X Jen Fig. 12.27. Photomicrograph of Hastelloy N tube plated with tellurium and stressed in fuel salt for 800 hr at 650°C and 25,000 psi in sait. Y-110916 F 10.001 in, + '10.003 in, 0.007 INCHES 500X T 15.005 . = f0.007 in. Fig, 12.28. Hastelloy N (heat 2477) plated with 1.0 mg/cm2 of tellurium, annealed 65 hr at 650°C in argon, and bent 90° at 25°C. Tension side. As polished. 133 Y-112658 — 1 0.035 INCHES IN 100X Jou - ____mu Fig. 12.29. Hastelloy N (heat 2477) plated with 0.01 mg/cm? of tellurium, annealed 207 hr at 650°C inargon, and bent 90° at 25°C. Tension side, as polished. Y —-H2176 T 0.035 INLHED v 100X IBY Fig. 12.30. Photomicrograph of a Hastelloy N creep specimen stressed at 30,000 psi at 650°C for 900 hr. The sample strained 2.5%, and the test was discontinued prior to failure. The environment was argon plus a small partial pressure of tellurium. 100X. As polished. mils (Fig. 12.30). No cracks were present in the unstressed portion of the specimen. The sample had strained only 2.5%, and such cracks would not be present in material tested without tellurium present. Thus this experiment showed that stress seems to aggravate intergranular cracking in Hastelloy N. The stress of 30,000 psi used in this test is higher than would normally be encountered in service, so a test is being started at a lower stress of 21,500 psi. Tests of type 304 stainless steel and pure nickel have also been started to investigate the cracking tendencies of these materials in a tellurium-containing environment. 12.8 INTERGRANULAR CRACKING OF MATERIALS EXPOSED TO SULFUR AND SEVERAL FISSION PRODUCT ELEMENTS H. E. McCoy As described in Sect. 6.2, Shaffer et al. developed techniques for exposing small metal tensile specimens to vapors of S, Se, I,, Te, and a mixture of As, Cd, and Sb. These elements have sufficient vapor pressure to transfer when the tensile samples and small amount of 134 these elements are sealed together in quartz and placed in a furnace for annealing at 650°C. The amount of each element has been small, being enough to result in a concentration of 100 ppm in the outer 5 mils of exposed metal surfaces (~0.01 mg/cm?). The first set of experiments involved only Hastelloy N samples that were annealed 1000 hr in each environ- ment. The samples were then strained to failure at 25°C, and sectioned metallographically for viewing. There were no detectable effects on the mechanical properties, but numerous intergranular edge cracks were formed in the sample exposed to tellurium. The fracture and the sample edges near the fracture are shown in Fig. 12.31. However, the statistics on crack frequency and depth given in Table 12.5 for the first sample reflect the fact that the cracks were present throughout the deformed section of the specimen. Cracks were not formed in the specimens exposed to the other environments. In a second experiment, Hastelloy N specimens that had already been exposed for 1000 hr were exposed to a new aliquot of the same element for 1000 hr at 650°C. The samples were strained to failure at 25°C. Y- 4109917 Fig. 12.31. Hastelloy N specimen exposed to 0.01 mg/cm2 of tellurium vapor for 1000 hr at 650°C. Strained to failure at 25°C. As polished. 33X. Again, only the samples exposed to tellurium exhibited significant intergranular cracking. A photomicrograph showing a section at the fracture is shown in Fig. 12.32. The data in Table 12.5 (specimen 5) show that the additional exposure did not increase the crack fre- quency but did increase the depth. Another Hastelloy N specimen from the first experi- ment that had been exposed to tellurium for 1000 hr at Table 12.5. Cracking in Hastelloy N strained at 25°C after exposure to tellurium vapor Specimen Exposure Cracks Depth (mils) No. condition? ~ Total ~ Per “C age Maximum counted inch 1 A 191 168 0.9 2.7 2 A 225 157 1.5 6.6 3 A 187 135 0.9 24 4 A B 133 111 1.0 39 5 AA 209 164 1.4 4.5 135 9A = 1000 hr at 650°C in 0.01 mg/cm?® Te; B = 1000 hr at 650°Cin Ar. 650°C was annealed for 1000 hr at 650°C in argon. The crack depth increased slightly by the additional anneal- ing (specimen 4, Table 12.5). One sample was included in the second experiment that had not been exposed previously and was exposed to 0.01 mg/cm? of tellurium at 650°C for 1000 hr. It was strained to failure at 25°C. The crack statistics (sample 2, Table 12.5) and the metallographic section (Fig. 12.33) show clearly that the cracking was more severe than in sample 1, which was exposed to similar conditions. We have no explanation for this observa- tion. A third experiment was run in which samples of Ni-200 and type 304L stainless steel were exposed to 0.01 mg/cm? each of I,, tellurium, and combinations of I, and tellurium for 1000 hr at 650°C. Hastelloy N was exposed to tellurium for 1000 hr in the same experiments. Only the Hastelloy N samplie showed significant intergranular cracking (sample 3, Table 12.5). The samples of nickel and type 304L stainless steel did not crack after exposure to any of the environments. Photomicrographs of the fractures of Hastelloy N, nickel, and type 304L stainless steel are shown in Fig. 12.34. Y-1440255 Fig. 12.32. Hastelloy N specimen exposed to 0.01 mg/em? of tellurium vapor for 1000 hr at 650°C, exposed to another 0.01 mg/cm? of tellurium vapor for 1000 hr at 650°C, and strained to fracture at 25°C. As polished, 33X. 136 Y-110248 Fig. 12.33. Hastelloy N sample exposed in the second experiment to .01 mg/cm2 of tellurium at 650°C for 1000 hr. Strained to failure at 26°C. As polished. 33X. These experiments demonstrate clearly that tellurium will cause intergranular cracking in Hastelloy N and that exposure to Se, S, I,, Cd, As, and Sb for up to 2000 hr does not cause cracking. Type 304L stainless steel and Ni-200 seem resistant to cracking after exposure to tellurium for 1000 hr at 650°C. 12.9 MECHANICAL PROPERTIES OF HASTELLOY N MODIFIED WITH SEVERAL ELEMENTS H. E. McCoy Several elements are formed as fission products that may diffuse into the structural metal and alter its mechanical properties. Sulfur is also of interest because it is a residual impurity in the salt and may have been introduced through oil that leaked into the pump bowl. Alloys have been made with nominal additions of 0.01% of several of these elements, and some test results have been obtained. The alloys prepared to date are listed in Table 12.6. All except alloys 361 and 363 have the composition of standard Hastelloy N, namely, Ni—16% Mo—7% Cr—4% Fe—0.5% Mn—0.5% Si—0.05% C. Alloys 361 and 363 did not contain chromium, to determine whether sulfur and tellurium had different effects whether chromium was or was not present. Ru, Sn, Sb, Te, S, and As have been added successfully, but Sr, Cd, and Cs have not been retained in measurable concentrations. The test program for these alloys is to obtain tensile data at 25 and 650°C in the solution-annealed con- dition. The alloys will then be aged for various times at 650°C and some of the testing sequence repeated. Limited tests have been run to date, and the data are summarized in Table 12.6. Several trends already seem obvious. 1. There were variations in the yield stresses of the solution-annealed alloys at both 25 and 650°C. These variations were small and were likely related to the carbon content rather than the small alloy addition. (Several of the alloys had only 0.02 to 0.03% C and had lower strength.) 2. The fracture strains at 25°C were quite high for all alloys. The fracture strains in tensile tests at 650°C were generally above 25%. Alloys 352, 359, 361, and 363 had fracture strains slightly below 25%. 3. Several of the alloys had lower stress-rupture proper- ties than the undoped alloy (351), but the re- ductions of only a factor of 2 or less were likely due to the lower carbon concentrations of some alloys. 137 Y~{11464 [ Y-1114i8 (b )il Y-1{1434 Fig. 12.34. Photomicrographs of specimens exposed to 0.01 mg/cm? of tellurium for 1000 hr at 650°C and fractured at 25°C, (@) Hastelloy N. () Nickel-200, (c) Type 304L stainless steel, As polished. 33X. 138 Table 12.6. Influence of various alloying additions on the mechanical properties of Hastelloy N Solution-annealed material Aged 1000 hr at 650°C Some of the more important property changes are shown in Fig. 12.35. Tellurium and sulfur have delete- rious effects, with tellurium being more detrimental. The effects of both elements are accented by the absence of chromium. Metallographic samples have been viewed of alloy 359 after various types of tests. All samples were annealed 1 hr at 1180°C before testing. The fracture of a sample tensile tested at 25°C is shown in Fig. 12.36. The fracture is primarily transgranular, and no intergranular cracks are visible. A specimen tensile tested at 650°C is shown in Fig. 12.37, and numerous intergranular cracks Alloy N Yield Total Yield Total Rupture life Fracture Yield Total Rupture life Fracture No. Addition stress strain stress strain at 40,000 psi strain stress strain at 40,000 psi strain at 25°C at 25°C at 650°C at 650°C and 650°C at 650°C at 25°C at 25°C and 650°C at 650°C (psi) (%) (psi) (%) (hr) (%) (psi) (%) (hr) (%) 351 None 49900 546 41,100 25.7 198.5 9.5 55,800 55.0 307.0 22.6 352 500 ppm Se 47,900 53.1 35,500 23.5 61.4 4.1 §2,800 57.5 62.5 5.8 353 S ppmSr 58,900 55.3 33,300 26.2 68.8 5.9 55,600 525 202.0 12.8 354 50 ppm Tc 55,000 544 32,000 27.9 (IOO)b (10) 52,800 9.5 228.5 14.7 355 200 ppm Ru 40,900 56.5 27,800 30.2 77.1 8.6 56,800 58.0 267.4 14.0 356 <40ppmCd 53,700 50.8 33,000 25.4 89.5 5.6 54,400 52.8 185.7 10.2 357 200ppmSn 46,700 56.8 33,000 32.5 82.8 7.3 54,900 55.5 157.1 11.3 358 200 ppm Sb 46,800 70.0 28,600 35.5 55.7 6.6 48,500 61.5 312.8 12.8 359 600 ppmTe 49,600 50.8 39,400 19.9 (16) 2) 53,000 52.0 16.8 3.8 360 <0.7ppm Cs 78.3 6.3 56,000 57.0 234.9 16.0 361€ 400 ppm Te 41,000 60.6 30,600 22.9 (2) (5) 46,900 52.5 20.1 6.7 362 90 ppm S 44,500 64.6 29,400 35.0 39.0 0.91 52,300 56.2 364.2 12.5 363¢ 970 ppm S 51,300 51.6 40,800 18.9 (10) 2) 58,200 50.0 22.9 4.5 364 960 ppm S 53,700 53.5 37,000 27.3 61.7 4.9 60,100 49.4 216.7 9.7 365 40 ppm As 47,500 63.7 77.4 5.2 54,900 58.5 291.1 12.6 %Contains 30 ppm S. bEstimates based on interpolation of data from other stress levels. “Does not contain Cr. However, alloys 359, 361, 362, and 363 had 50 | TN ST ORNL-DWS 72;.’11‘66 reductions in stress-rupture properties and fracture T~ 7:\‘“\6 W : | } strain outside the range attributable to carbon 26 7 T \\é\ IR i | f content. z | \\\JD:\\\\\E\\ Al 4. Aging for 1000 hr at 650°C caused a slight increase 8 30 L 24\\\$?\1\7\~§z\:\1§3€0 . in the yield stress at 25°C of most alloys but had no - o iee 7o) N 2\\\71\ i,\%l;%wss detectable influence on the fracture strain. The & ,o| o 362 (90ppm$) Yo, ‘e\K%Te | i stress-rupture properties and fracture strains were — © © 364 (960ppm S)| | N TeNocr | | ’ . K . ® 363 (970 ppm S, NoCr) S NoCr : : generally improved by aging. The properties of 10 |- & 361 (Te, NoCr) | }.14 ‘ L alloys 352, 359, 361, and 363 were not worsened [} i i | R : appreciably by aging. o i } ‘ ’ S 10° 10! 102 103 10% RUPTURE TIME (hr) Fig. 12.35. Stress-rupture properties at 650°C of Hastelloy N modified with various additions. Numbers indicate the fracture strain, are present. Photomicrographs of a specimen tested at 650°C at a stress of 30,000 psi are shown in Fig. 12.38. The sample failed in 354 hr with 1.6% strain. The fracture is intergranular, and there are numerous inter- granular separations throughout the stressed portion of the specimen. 139 Y-114554 toG0om. 0.035 INCHES 100X Moo {6.030 . Fig. 12.36. Photomicrograph of the fracture of alloy 359 after straining to failure at 25°C. As polished. Y-111562 - T0.0iO n. 0.035 INCHES 100X 10.030 in. Fig. 12.37. Photomicrograph of the fracture of alloy 359 after straining to failure at 650°C. As polished. 140 i 0.010 in. 0.035 INCHES 100X 10.030 in. Y-144559 1500t m, 500X '0.003 in, 0.007 INCHES 16,005 =, ity e T == 10,007 in. Fig. 12.38. Photomicrographs of alloy 359 after creeping at 30,000 psi and 650°C. Failed after 354 hr with a strain of 1.6%. (a) Fracture. 100X. (b) Typical view of edge. 500X. As-polished condition. The alloying additions that show the largest influ- ences on the mechanical properties are those of group VI-B — sulfur, selenium, and tellurium. These elements have been shown to reduce the ductility of nickel in the temperature range of 500 to 800°C but have no effects on the properties up to 500°C.!° This pattern is consistent with that noted with the more complex alloy, Hastelloy N. 12.10 STATUS OF INTERGRANULAR CRACKING STUDIES H. E. McCoy Several important observations have been made con- cerning the intergranular cracking of Hastelloy N in the MSRE. Cracks were visible in many of the samples after they were removed from the MSRE, but the number and visible depth of the cracks increased in most instances with further straining. A heat-exchanger tube was one notable exception where the number of visible cracks was about equivalent before and after straining. The cracking statistics on various components that were exposed throughout the operation are summarized in Table 12.7. The surveillance samples that were exposed for different periods of time in the MSRE core gave some indication of the time dependence of the cracking (Table 12.8). The number of cracks increased, but it is not at all apparent that the depth of cracking increased with time over the range studied. Laboratory corrosion experiments in which Hastelloy N was exposed to fluoride salts thus far have not reproduced the cracking that was observed in the MSRE. Selective intergranular attack has been produced by making the salt very oxidizing, but the maximum depth of attack in 3000 hr was 0.5 mil, compared with several mils in the MSRE. Experiments in which Hastelloy N specimens were exposed to low concentrations of vapor of S, Se, Te, I,, As, Sb, and Cd for 1000 or 2000 hr at 650°C and then strained at 25°C resulted in intergranular cracking only in those specimens that had been exposed to tellurium vapor. The cracks in these specimens were similar in depth and appearance to those in materials from the MSRE. Type 304L stainless steel and nickel-200 did not crack when tested under these same conditions. Hastelloy N tube-burst specimens and a creep speci- men that were stressed in the presence of tellurium had extensive intergranular cracks. Special alloys of Hastelloy N that contained small additions of Se, Tc, 15. C. G. Bieber and R. F. Decker, “The Melting of Malleable Nickel and Nickel Alloys,” Trans. AIME 221, 629 (1961). 141 Ru, Sn, Sb, Te, S, and As have been prepared and tested under a variety of conditions. The alloys containing sulfur and tellurium had reduced rupture lives and fracture strains at 650°C. Thus the laboratory tests to date leave no question that intergranular cracking of the type noted in the MSRE can be produced by tellurium and possibly by sulfur. However, the evidence does not seem in hand to show that the cracking in the MSRE resuited from tellurium or sulfur. Sulfur is not a fission product but was introduced by inleakage of oil from the pump bearings and possibly as a contaminant in the initial salt charge. The probable amount of sulfur introduced by oil inleakage was 27 g, and the maximum amount of sulfur in the initial fuel charge was 24 g, or a total concentration in the fuel salt of 10 ppm. A similar concentration of sulfur was present in the coolant circuit, but no cracking was observed. The coolant loop was not exposed to as intense radiation as the primary circuit, and this may have been a factor. However, our laboratory experiments do not strongly support the ability of very low concentrations of sulfur to embrittie Hastelloy N, and we currently discount the possibility that the cracking is due to sulfur. Qur surface chemical studies on material removed from the MSRE showed that all fission products with sufficient half-lives to be detected were present to shallow depths in the alloy. Close metallographic examination has shown that many of the intergranular cracks are present in samples removed from the MSRE, so the apparent penetration of the metal by fission products may not represent much solid-state diffusion along the grain boundaries, but rather the coating of the surfaces of cracks. The cracks may be due to the formation of brittle or very low-melting compounds along the grain boundaries. The element responsible for the behavior has not been isolated, but a rather strong circumstantial case has developed for its being tellurium. Laboratory experi- ments have concentrated on tellurium, but some work on other fission products is in progress. On the assumption that tellurium is associated with the cracking, it is important to speculate about the time and temperature dependence of the cracking. The accumulated cracking statistics on samples exposed to tellurium are summarized in Table 12.9. The various nuclear experiments vary in the time of exposure and the concentration of tellurium, so it is impossible to separate the effects of the two variables. The first three experiments were for relatively short times, had low tellurium concentrations, and did not show detectable cracking in the available photomicrographs. The surveil- lance samples and components from the MSRE were 142 Table 12.7. Crack formation in various samples from the MSRE after straining at 25°C Over 500°C for 30,807 hr Cracks Depth (mils) 12714 Total Te Sample description Counted Per inch Average Maximum atoms/cm atoms/cm?® x 10'° x 1017 Exposed thimble 91 192 5.0 8.0 1.8, 1.8 29,29 Thimble under spacer sleeve 148 257 4.0 8.0 0.59, 0.46 0.95, 0.74 imbl 88 1 .0 7. Thfmb e spacer, OD 78 3 0 } 1.0,2.8 1.6, 4.5 Thimble spacer, ID 106 202 3.0 5.0 Mist shield, inside vapor 47 192 1.0 2.0 Mist shield, inside liquid 33 150 4.0 6.5 Mist shield, outside vapor 80 363 4.0 5.0 Mist shield, outside liquid 54 300 3.0 5.0¢ 0.55 0.89 Sampier cage rod, vapor 100 143 2.5 5.0 Sampler cage rod, vapor 170 237 3.2 10.0 Sampler cage rod, liquid 102 165 3.7 10.0 Sampler cage rod, liquid 131 238 7.5 12.5 Freeze valve 105 131 240 0.75 1.5 0.04 0.06 Heat-exchanger tube (unstrained) 100, 135 228, 308 2.5 3.8 Heat-exchanger tube (strained) 219 262 5.0 6.3 IMeasured. bCalculated. “One crack was 12 mils deep; next largest was 5 mils. Table 12.8. Crack formation in Hastelloy N surveillance samples strained to failure at 25°C Time of a Cracks Depth (mils) Heat Exposure exposure (hr) Counted Per inch Average Maximum 5085 Control 5,550 1 1 5.7 5.7 5085 Core 5,550 24 19 2.5 8.8 5085 Control 11,933 0 5085 Core 11,933 178 134 1.9 6.3 5065 Control 11,933 3 3 1.0 2.0 5065 Core 11,933 277 230 1.8 3.8 5085 Control 19,136 4 3 1.5 2.8 5085 Core 19,136 213 176 5.0 7.0° 5085 Core 19,136 140 146 3.8 8.8 5065 Control 19,136 3 3 2.5 4.0 5065 Core 19,136 240 229 5.0 7.5 “Control specimens were exposed at 650°C to static unenriched fuel salt for the indicated time. Core specimens were exposed at 650°C to MSRE core. 5One crack was 15 mils deep; next largest was 7 mils. exposed for various times to different levels of tellu- rium. The frequency of cracking increases with time and tellurium content, but the depth of cracking does not increase detectably. Several samples have been electroplated and vapor plated with tellurium. One sample was electroplated with 47 X 10'7 atoms/cm? and annealed for 65 hr at 650°C. The sample had numerous cracks after bend testing. Another sample was electroplated with less tellurium, annealed 200 hr at 650°C, and did not have any detectable cracks after a bend test. These two samples show a definite effect of tellurium concentration. Several samples have been vapor plated with tellurium and held at 650°C for 1000 and 2000 hr. These samples cracked more severely than 143 those from the first group of surveillance samples, which had been at temperature for 5550 hr but had only 0.1 X 10'7 atoms of tellurium per square centimeter. Again, this indicates an effect of concen- tration. It is impossible to determine the effects of temperature from the available data. The temperature variation in the MSRE was quite small, and other effects masked any effects of temperature. Thus the available data offer evidence that tellurium concen- tration is important but give no indications about the effects of time and temperature. The superior resistance of type 304L stainless steel and Ni-200 offers encouragement that all materials are not affected adversely by tellurium. Table 12.9. Cracks produced in various samples after straining at 25°C Time at Tellurium produced temperature dul;ing o;?era tion Cracks Depth (mils) Source of samples :fiei flSSI(:In d (atoms per square Counted Per inch Average Maximum proguc (s};))r oduce centimeter of metal) x 10'7 Trauger, MTR 44-1 682 0.2 MTR 44- 766 0.2 Compere and Bohlman 1,366 0.2 Surveillance samples, 5,550 0.1 24 19 2.5 8.8 group 1 Surveillance samples, 11,933 0.8 178, 277 134, 230 1.9, 1.8 6.3, 3.8 group 3 Surveillance samples, 19,136 1.2 213, 140,240 176,146,229 5.0,3.8,50 7.0,8.38,7.5 group 4 MSRE, end of 30,807 1.4 143-363 0.75-17.5 1.5-12.5 operation Electroplated sample 65 47 196 273 0.8 1.5 Electroplated sample 200 0.5 0 Vapor-plated sample 1,000 0.5 191, 225,187 168,157,135 0.9,1.5,0.9, 2.7,6.6,2.4 Vapor-plated sample 2,000 1 209 164 1.4 4.5 13. Graphite Studies W. P. Eatherly INTRODUCTION The objectives of the MSRP graphite program con- tinue to be the development of improved radiation- resistant experimental graphites and the development of techniques to seal the graphite against '?°Xe. As indicated in the previous semiannual report,! the emphasis of the MSRP work has been shifted from the bulk graphite to the sealing problem. A significant amount of experimental graphite fabrication is con- tinuing, however, but it is now largely funded from other sources and is directed at applications other than nuclear. Because much of the work is concerned with relating physical properties to microstructure and such relations contribute to our understanding of radiation damage, the results are reported here. 1. W. P. Eatherly, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 111-24. 13.1 GRAPHITE DEVELOPMENT C.R.Kennedy W.P. Eatherly The program to fabricate advanced types of graphite for extended reactor life continued through this report period. The graphite desired is one having a monolithic structure with very strong grain boundaries to resist the shearing deformation caused by irradiation growth. We are attempting to obtain this structure by the use of green cokes to reduce or eliminate the binder shrinkage cracks during fabrication. We have been working exten- sively with Robinson coke (an air-blown isotropic coke), Santa Maria coke (a less isotropic coke with fairly high impurities), and recently with two acicular cokes: coke A (a very anisotropic coke) and coke SA (similar to coke A but more isotropic). The thermal- gravimetric results for these cokes are shown in Fig. 13.1 for comparison. It is readily evident that the Robinson, Santa Maria, and SA are similar in this regard ORNL-DOWG 72-7666 ] I J SANTA MAR% WEIGHT LOSS (%) )] yas i | f ’_\A - rfi.‘< e T | 100 200 300 400 P 500 600 700 800 900 1000 TEMPERATURE (°C) Fig. 13.1. Thermal-gravimetric results for four green cokes. Heating rate, 3°C/min. 144 145 with about 11 to 12% volatile, while the A coke has only 7% volatile. We have been using 15V coal tar pitch, 240 petroleum pitch, furan resins, and furan—petroleum-pitch mixtures as binders to fabricate the graphites for evaluation. The filler particles have all been ground to fairly fine sizes (mean particle size <10 um), and the large surface area requires large binder contents. The filler and binders have been slurry mixed with benzene, dried, and remicronized before molding. The optimum combina- tion of molding temperature, pressure, and binder content for each of the above binder and filler combinations is different; therefore, direct comparison of component raw materials cannot be made using identical fabrication techniques. We have found, how- ever, that for each filler, there exists a reasonable relationship between green and graphitized density which is independent of binder, binder content, and molding conditions. This relationship is given in Figs. 13.2 and 13.3 for Robinson and Santa Maria fillers. It is evident from these figures that there are coke-filler— thermal combinations which improve packing and so ORNL-OWG 72-7667 ‘-9 ‘ ° / A 1.8 A s/& [ ) o° V4 A7 0’9( a a - t.6 O- J [ ] o] 15 75 GRAPHITIZED BULK DENSITY { g/cm3) O ALL ROBINSON COKE 65% ROBINSON COKE 35% THERMAX o MODIFIED ROBINSON COKE 13 — FILLED PQINTS INDICATE NITROBENZENE ADDED | 12 0S8 1.0 A4 1.2 1.3 1.4 GREEN (AS-MOLDED ) BULK DENSITY {g/cm®) Fig. 13.2. Green density vs graphitized density for Robinson coke graphites. give higher green densities; however, the basic coke filler density proportionality still exists. Comparison of the filled and open points in Figs. 13.2 and 13.3 shows that nitrobenzene is significant in increasing both the green and graphite densities. To examine whether the effect of nitrobenzene is a result of its plasticizing or of its lubricating the mix, a series of moldings with Robinson filler and water, oil, and nitrobenzene additions were made. The results of this series, shown in Fig. 13.4, revealed that additions of oil ORNL-DWG 72-7668 2.0 T T T O ALL SANTA MARIA & A 65% SANTA MARIA € 359% THERMAX 'Y R 49 2 FILLED POINTS INDICATE A /{ > NITROBENZENE ADDED AA Aa = R ) 518 g / X 8 9. ® [m) 1.7 >, N o © - Q = (o) a o® g o g 1.6 / 15 1.0 14 1.2 1.3 1.4 1.5 GREEN (AS-MOLDED) BULK DENSITY (g/cm®) Fig. 13.3. Green density vs graphitized density for Santa Maria coke graphites. ORNL-DWG 72-7669 2.0 -—-—-"‘—-‘— s 18 — ] e o el e e e e, B -_o-"'-'. s\ \i % 16 & o ROBINSON COKE WITH O TO 10% OIL OR HQ " a SANTA MARIA COKE £ 4 65% SANTA MARIA COKE i g 14 35% THERMAX & ALL WITH 30 pph 45-V PITCH-MOLDED % WITH 1300-psi PRESSURE AT 55°C D — @ 12 10 ! 0 2 a4 6 8 10 2 NITROBENZENE ADDED (%) Fig. 13.4. The effect of nitrobenzene on molded graphites. or water up to 10% have no effect on the density of moldings with nitrobenzene contents from 0 to 8%. Therefore, we concluded that the densifying effect of nitrobenzene is in plasticizing the mix and not due to lubrication effects. Although the data in Fig. 13.4 indicate that about 8% nitrobenzene gives the maxi- mum density, other considerations also affect the optimum amount. For example, one difficulty in the use of nitrobenzene is a strong tendency towards agglomeration which yields a very large pore size. During this report period we began to make 3%-in.- diam blocks of the experimental graphites for evalua- tion. These blocks are large enough to obtain bend, electrical resistivity, sonic modulus, x-ray parameters, metallography, thermal expansion, thermal conduc- tivity, and HFIR irradiation samples. The testing is still in progress, but some values are available and are listed in Table 13.1. The sonic moduli are shown in Fig. 13.5 as a function of the accessible porosity. For comparison, the upper line represents the behavior of Poco graphite as deter- mined by LASL and confirms the suggestion of Armstrong® that the modulus is independent of binder. The comparison of these results with similar data® on uranium nitride suggests that the decrease in modulus is simply a geometric effect of increasing porosity. These 2. H. L. Whaley, W. Fulkerson, and R. A. Potter, “Elastic Moduli and Debye Temperature of Polycrystalline Uranium Nitride by Ultrasonic Velocity Measurements,” J. Nucl. Mater. 31,345-50 (1969). 3. P.E. Armstrong, Effects of Porosity on Graphite Profiles, CMF-13, Research on Carbon and Graphite, report No. 15, LA-4631-MS. ORNL-DWG 72-5713 ! . . , o ROélNSON—szzs © ROBINSON-PLASTICIZED (SMALL CRACKS) i a ROBINSON-240 PITCH ¢ SANTA MARIA-QX229 * 80% ROBINSON-20 % THERMAX-QX229 «L 4 SEMI-ACICULAR-QX229 1 J | . T AT MODULUS {psi) o ~ LASL POCO DAT | | 30 20 25 35 40 45 POROSITY (%) Fig. 13.5. Dynamic modulus of ORNL graphites. 146 graphites tested are all very isotropic and do not reflect the large effects of orientation on the elastic constants. Over the years we have accumulated bend test results of a large number of both commercial and experimental graphites. These results are given in Fig. 13.6 for comparison to the ORNL graphites. These results are very informative in classifying the graphites by their pore morphology. There appear to be two definite classes of graphites which have a constant strength-to- modulus ratio, indicating similar pore morphology or defect structures with like stress intensification. The highest class is the Poco and ORNL grades, with similar microstructures having a more equiaxial pore structure. The second class, with a lower strength-to- modulus ratio, is the conventional graphites made from calcined filler and several of the ORNL grades which have a more platelike pore structure. This second class includes graphites made from graphite filler particles obtained from grinding Poco grades of graphite. Also included are seven grades of graphite having strength-to-modulus ratios toward the bottom class. In every case, these very low strength-to- modulus ratios can be attributed to abnormally severe defect structures relative to density. In three of these ORNL-DWG 72-5744 (x103) I T T | 0 ORNL GRAPHITES A POCO GRAPHITES . 14 |—*® IMPREGNATED POCO GRAPHITES © CONVENTIONAL GRADES .A o ® CARBON BLACK . / GRADES / 12 A [ ] : Yo T g o/ T G a o °o6 5 8 / o> o N 4 B a § %00 Q g 4 - 4 P4 .o o ‘/ & ° o ° ° 4 0%90 o o o 2 / 0 0 1 2 (x108) MODULUS OF ELASTICITY (psi) Fig. 13.6. Bend strength as a function of Young’s modulus. Table 13.1. Physical properties obtained on ORNL graphites Sonic Sonic Bi . Bulk Electrical resistivity Modulus of Fracture Young’s i Sonic Bacon ock Filler . . Sicm) . longitudinal transverse . , . No. coke Binder densit (u rupture strain modulus Youns's modulus shear modulus Polssc_)n s anisotropy Remarks . . g (g/cm™) Axial Transverse (psi) (%) (psi) (psi) (psi) ratio factor X108 x10° x10° 10 Robinson QX229 1.55 0.90 0.40 0.06 Small samples 15 Robinson QX229 1.60 1.14 0.49 0.13 Small samples 20 Robinson QX229 1.70 1.39 0.60 0.17 Small samples 25 Robinson QX229 1.77 1.71 0.74 0.15 Small samples T-12 Robinson QX229 1.79 1410 2.04 0.92 0.25 1.02 Shrinkage cracks T-61 Robinson 240 Pitch 1.84 1162 13,180 0610 2.81 2.40 1.00 0.30 1.01 X-58 Robinson Nitrobenzene 1.92 1405 8,610 0.383 2.72 1.97 0.85 0.20 1.02 Fine network of very small cracks 11-16 Robinson QX228 1.77 1662 8,800 0.380 2.61 1.84 0.85 0.19 Molding cracks 11-23 Robinson QX229 1.84 1500 11,100 0.520 2.69 2.10 1.01 0.26 Molding cracks 22-40K Robinson QX302-2 1.77 1480 T-32 Robinson, QX229 1.75 2340 8,100 0411 2.32 1.79 0.84 0.21 20% Thermax 11-24 Robinson, QX234 1.73 1774 1.03 Small samples 15% nat. flake 11-25 Robinson, QX234 1.78 1629 1.02 Small samples 10% nat. flake 11-26 Robinson, QX234 1.78 1679 1.04 Small samples 20% nat. flake 10-24 Santa Maria QX229 1.79 1583 10,620 0.669 2.07 1.68 0.71 0.18 1.03 10-25 Santa Maria QX229 1.83 1473 10,620 0.613 2.26 1.86 0.81 0.21 1.02 X-2 Santa Maria QX229 1.80 1.84 0.82 0.22 1.03 Shrinkage cracks 11-1 Santa Maria QX229 1.85 1356 11,320 0.601 2.51 1.87 0.87 0.20 1143 SA QX229 1.78 1070 990 1.67 0.77 0.20 1.07 Shrinkage cracks 11-44 SA QX229 1.82 950 Shrinkage cracks 11-46 A QX229 1.71 980 872 1.11 Shrinkage cracks 22-32K A 240 Pitch 1.62 1608 4130 0.487 1.19 0.72 0.38 ~0 12-1 A, QX234 1.63 0.67 0.35 ~0 1.13 Shrinkage cracks 10% nat. flake 12-2 QX234 1.74 2076 4130 0.365 143 0.98 047 0.04 A, 10% nat. flake A4 148 materials where the fabrication schedule is known, the large linear defects are a result of mix agglomeration into high-density regions. In the others, the micro- structure revealed evidence of a heavily impregnated graphite with abnormally large filler particles dispersed in a mixture of fine particles. Associated with each large particle was an equally large defect. Many of the graphites made by conventional fabrica- tion are heavily impregnated and appear to retain a constant strength-to-modulus ratio independent of the degree of impregnation. This strongly suggests that the pore morphology is not affected by the impregnant. However, in several of the Poco grades, impregnation was attended by significant increases in the strength-to- modulus ratio, implying an improvement in the reduc- tion of stress intensity factors. This suggests that the better ORNL grades could similarly be improved by impregnation. The binder carbon in the ORNL graphite cannot be resolved in the microscope. We do, however, see some differences in the electrical resistivity due to binder as well as filler material. The pitch-bindered Robinson graphite has a lower resistivity than the furan-pitch- bindered graphites. The Robinson and Santa Maria fillers appear to be very similar and have a lower conductivity than the A and SA fillers. It was expected that Thermax would increase the electric resistivity, as was observed. The natural flake additions, on the other hand, were expected to decrease the resistivity; instead they also resulted in increases. It appears likely that this increase by natural flake additions is a result of the poor bindering characteristics of the natural flake. The Bacon anisotropy factors (BAF) in Table 13.1 show that both the Robinson and Santa Maria graphites are very isotropic. (The BAF is 1.00 for material having completely random crystalline orientation.) The SA filler graphite has some degree of anisotropy, and the A filler graphite is more anisotropic. Addition of natural flake intensifies the anisotropy of all of the graphites. During this report period, samples of ORNL graphites which had completed one irradiation in the HFIR were examined and replaced in HFIR for further irradiation. The observed dimensional changes are shown in Figs. 13.7 and 13.8 for Robinson and Santa Maria graphites. The dimensional changes in the Santa Maria grades show a slight anisotropy, while the Robinson grades behave very isotropically. Additions of up to 35% Thermax in either the Robinson or the Santa Maria graphites have no clearly distinguishable effect. In summary, the significant points in our graphite development program during this report period are as follows: LENGTH CHANGE (%) ORNL-DWG 72-7672 4 - 1 T -1 T T o ALL ROBINSON COKE a 35% THERMAX “ 3 OPEN POINTS-PARALLEL TO MOLDING CLOSED POINTS-NORMAL TO MOLDING A L 1 TSR D AXF POCO 0 : /‘ S ‘x\ ™~ A - a \T__ - N ! o | | | > | | 0 5 10 15 20 25 20 35 (x10%) FLUENCE (neutrons /cr? ) £>50 keV) Fig. 13.7. HFIR irradiation results for Robinson grades at 715°C. ORNL-DWG 72-7673 3 ] | | S { o ALL SANTA MARIA — 2| # 35% THERMAX 4 < OPEN POINTS-PARALLEL TO MOLDING H CLOSED POINTS-NORMAL TO MOLDING | / z t | ] <1 ’ a I — AXF POCO [&] 5 o ° ‘__/ < \. ° 9 ‘ [F9} . | .\\fi—w ] ~3 J \ _2 i | 0 5 10 15 20 25 30 35 (x10%) FLUENCE (neutrons/cm?) ( E>50keV) Fig. 13.8. HFIR irradiation results for Santa Maria grades at 715°C. 1. We are beginning to recognize and understand the salient factors which control the quality of green- coke molded graphites. 2. We have obtained graphites with an unusually high strength-to-modulus ratio, indicating measurable suc- cess in optimizing pore morphology. 3. The modulus of elasticity for well-bindered isotropic graphites was found to be porosity dependent and independent of binder or filler coke. 4. Electrical resistivity was found to be dependent on both filler and binder. 5. Both Thermax and natural flake additions reduce the electrical conductivity. 6. The irradiation results show that the ORNL grades are paralleling the behavior of the best commercial grade (Poco) to the highest fluence obtained to date (1.5 X 10%? neutrons/cm?). 13.2 PROCUREMENT OF VARIOUS GRADES OF CARBON AND GRAPHITE W. H. Cook W.P. Eatherly We have received a set of rough-machined carbon materials from Poco Graphite, Inc., which are precur- sors of their grade AXF, annealed at various tempera- tures from 1400 to 2500°C. We have also received an unmachined block fired to 2500°C, which we have refired to 3000°C. These provide us with stock for test specimens of grade AXF precursor materials fired at 1400, 1800, 2000, 2200, 2500, and 3000°C. We shall make the final, finished machining of the test specimens, characterize them, and integrate them into our graphite irradiation program. The purpose in irradiating the AXF precursors is to supplement the information obtained from irradiation of a series of carbon-black grades.* For the carbon- black grades a very sharp decrease in volume occurred early in the irradiation, which then transformed into a slow linear expansion. The expansion rate was un- affected by the heat-treatment temperature, whereas the sharp contraction was strongly temperature depen- dent, ranging from 15% after 1000°C treatment to less than 1% at 2400°C and above. Further, the carbon- black materials fired at high temperatures showed very little damage effect until a fluence of 1.5 X 10?2 neutrons/cm? was obtained. The purpose of the present experiment is to determine whether similar effects occur in the more stable AXF-type materials. We have acquired a plate, 1 X 3% X 8% in. (nominally), of grade JA-5 manufactured by Airco Speer Carbon Products. Although it has a bulk density of only 1.68 g/cm?, it is nearly isotropic and has a special type of filler material which makes it of interest to our overall irradiation studies. 13.3 TEXTURE DETERMINATIONS 0O.B.Cavin D.M. Hewette 11 The degree of anisotropy is an important parameter in the development of graphites both at the Y-12 Plant and at ORNL. We have continued to determine the x-ray anisotropy of materials made at both installations and are now extending our capabilities to include an optical technique which can be applied to an area as small as 10 um in diameter. This will allow us not only to determine the anisotropy of the bulk material but 4. C. R. Kennedy and W. P. Fatherly, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL4728, p. 113. 149 also to independently determine the anisotropy of thin coatings used for surface sealants. During this report period a number of graphite samples were analyzed for the Y-12 Plant.> These samples, all fabricated by molding, ranged from near isotropic (BAF = 1.04) to highly anisotropic (BAF = 2.01). (A Bacon anisotropy factor of 1.00 indicates completely random crystalline orientations.) The x-ray anisotropy of a number of the ORNL materials de- scribed in Sect. 13.1 has also been determined, and the values are shown in Table 13.2. Most of these graphites are close to being isotropic, with BAF values ranging from 1.003 to 1.125 for samples T-30 and 12-1 respectively. In many instances, we have determined anisotropy values in the parallel and perpendicular directions by cutting samples whose axes were parallel with these directions in the body. The values for the two directions agree to within experimental error. Recently, we obtained a Leitz microscope photo- meter to be used in determining the optical anisotropy factor (OPTAF) of graphites of interest in the Gas Cooled Reactor Program, and its use is now being extended to include the thin experimental surface sealants being placed on molten-salt graphites. The reflectivity (r) of materials having a high degree of crystalline anisotropy (such as graphite) is much greater when the plane of vibration of a linearly 5. Submitted by L. G. Overholser of the Chemical Engineer- ing Development Division, Y-12. 13.2. X-ray anisotropy values of experimental graphites Sample No 4 Rb RO BAF 10-24 0.660 0.670 1.030 10-25 0.663 0.669 1.017 T-12 0.662 0.669 1.021 T-30 0.666 0.667 1.003 T-61 0.664 0.669 1.012 X-58 0.661 0.670 1.027 X-2 0.661 0.669 1.025 11-24 0.659 0.670 1.034 11-25 0.662 0.669 1.022 11-26 0.657 0.672 1.044 11-43 0.651 0.674 1.071 11-46 0.643 0.678 1.109 12-1 0.640 0.680 1.125 dgee sect. 13.1 br )} and R are the anisotropy values in the parallel and perpendicular directions, respectively, of the fabricated body. Many of the R values were determined independently from a second sample. R + 2R = 2. polarized monochromatic light beam is parallel with the a axis than when it is parallel with the ¢ axis. The values of r, and r, have been reported to be 22.5 and 5% respectively.® A metallographic sample whose surface is parallel with a predominantly a crystallographic direc- tion will exhibit maxima and minima in the reflected polarized beam as it is rotated through 360° on the microscope stage. The ratio of the maximum to the minimum intensity is then a measure of the optically measured preferred orientation (OPTAF) of the mate- ria. An OPTAF of 1.0 is obtained on perfectly isotropic material. We have on hand a series of specimens that have been shown to have a relatively large range of preferred orientations by OPTAF measurements made at another laboratory. These samples have been evaluated on our equipment, and we have obtained a reasonable correla- tion between our values and those obtained elsewhere on the same samples. It is difficult if not impossible to determine anisot- ropy of thin coatings with conventional x-ray tech- niques. Relatively thick coatings are being prepared from which both x-ray and optical anisotropy values can be determined to arrive at a correlation between the techniques. We will then be able to determine not only the substrate texture but also the local texture that occurs in the pyrocarbon sealants. 13.4 THERMAL PROPERTY TESTING J.P.Moore D.L.McElroy T.G.Kollie Thermal conductivity (A\) measurements on an irradi- ated sample of AXF-5Q graphite were completed in the range 300 to 800°K (20 to 500°C) using a guarded longitudinal heat-flow technique. This sample had been irradiated in HFIR at a nominal temperature of 835°K (560°C) to a fluence of 2.6 X 10%! neutrons/cm?(E > 50 keV). The X results on this irradiated sample are compared with values for the unirradiated sample in Fig. 13.9. Neutron irradiation reduced A at all tempera- tures, with the ratio of A (irradiated) to A (unirradiated) ranging from 0.25 at 300°K to 0.4 at 800°K. Electrical resistivity (p) and Seebeck coefficient (s) values were obtained during the A measurements and show signifi- cant increases due to neutron irradiation. The room- temperature p increased from 1120 to 2300 uf2-cm, and the room-temperature Seebeck coefficient in- creased from 2 to 25 uV/deg. 6. H. B. Gruebmeier and G. P. Schneidler, “An Optical Method for the Determination of the Local Anisotropy of Pyrolytic Carbon Layers and Graphite,” ORNL-tr-2127. 150 ORNL— DWG 72-753 1.2 10 \ummmmzo -~ 0.8 \fiL g \ © L & TOP \Q\ N o MID ~§ o BOTTOM 0.6 f 0.4 IRRADIATED T = = —d M—-D—— e - e — 0.2 300 500 700 900 TEMPERATURE (K} Fig. 13.9. The thermal conductivity of unirradiated and irradiated AXF-5Q graphite. Irradiated at 825°K to a fluence of 2.6 X 102! neutrons/cm? (£ > 50 keV). A second AXF-5Q graphite sample, irradiated to a somewhat higher fluence, has been installed in the apparatus for measurements from 300 to 900°K (20 to 600°C). After irradiation in HFIR at a nominal temper- ature of 925°K (650°C) to a fluence of 4.2 X 102! neutrons/cm? (£ > 50 keV), the specimen had a room-temperature p value of 2265 uSk-cm, which is a preliminary indication that the neutron irradiation produced changes similar to those cited above. A second thermal property of considerable engi- neering and scientific interest is the linear thermal expansion coefficient. This property is also known to be sensitive to irradiation. A new apparatus has been constructed, employing a quartz differential dila- tometer. The first series of calibration tests has been completed using a copper specimen. This dilatometer, shown schematically in Fig. 13.10, is designed to also measure the temperature dependence of the tempera- ture coefficient of thermal expansion (CTE) to 2% uncertainty of AXF-5Q and H-337 graphites. Various aspects of the dilatometer have been interfaced with a computer-operated data acquisition system (CODAS). Length changes are measured to 2.5 X 107® c¢cm by an electronic micrometer and input to CODAS via a BCD reader. Sample temperature control is provided by CODAS using a three-action control algorithm which 151 CONTROL AND ORNL-DWG 72-794 READOUT CABLES ™~ ELECTRONIC HELIUM GAS COUNTER MICROMETER DIAL DRIVE MOTOR— L ELECTRICAL Le—WINDOW INSULATION— O |l —-MICROMETER SCREW CONTACT SENSOR | _pLEXIGLAS ASSEMBLY —— SPRING GROUND STRAP +— SAPPHIRE SPHERE SAPPHIRE DISK e WATER~- COOLED v INVAR BASE PLATE -- A er AN _ar ARGON GAS LTHERMOCOUPLES QUARTZ SUPPORT TUBE QUARTZ PROTECTION TUBE QUARTZ PUSH ROD QUARTZ DISK NICKEL CYLINDER SPECIMEN OLIITITIETINELIAT IR EER IR LTI DTSSR Fig. 13.10. Schematic drawing of the 1200°K quartz differ- ential dilatometer designed to measure coefficient of thermal _expansion of graphites. varies the output of the power supply for the sample furnace. Since each data point requires 4 to 8 hr for temperature stability, use of CODAS more than triples the data collection rate. Although the first CTE values must be considered preliminary until the system has been calibrated with NBS certified thermal expansion standards of quartz and copper, these initial results were smooth and encouraging close-to-literature values for the CTE of copper. 13.5 NOMINAL HELIUM PERMEABILITY PARAMETERS FOR VARIOUS GRADES OF GRAPHITE W.H. Cook J. L. Griffith We have a limited characterization program in pro- gress for some special grades of graphite that are too anisotropic to have potential value in applications involving large fast-neutron fluences at 715°C but are of interest as potential structural and containment mate- rials in the chemical processing of the MSBR fuel. Some of these are liquid-hyrdocarbon-impregnated grades of graphite that were subsequently heat treated to convert the hyrdocarbon impregnants to carbon or to graphite. Table 133 is a summary of some of the helium permeability parameters for these grades. Grade 2020 is a fine-grained, unimpregnated material that has an accessible porosity of 17% and for which pore entrance diameters are <3 um. Grades 2044 and 1226 are also fine-grained graphite bodies that were given special surface impregnations with liquid hydro- carbons and heat treated. This produced a graphitic impregnation in their pores from Y through 3% in. below their exterior surfaces. The first row of data in Table 13.3 for each of these is from specimens taken from the core (the unimpregnated zone) of the stock, and the second row of data for each is for specimens taken from the impregnated zones. The unimpregnated core of both grades had approxi- mately 16% accessible voids. Nominally, 65% of the accessible voids had pore entrance diameters of 1 to 2 um for grade 2044 and 1 to 3 um for grade 1226.7 The microstructures of the impregnated zones of grade 2044 indicated a uniform impregnation, but that for grade 1226 was nonuniform. In the grade 1226 impregnated zone there were tunnels of poorly impreg- nated voids. These were single, exploratory impregna- tions for each grade, so it is not surprising that the permeabilities are relatively high for the impregnated zones. Grades Graph-i-tite “G” and “A” are products of a commercially established process in which the stock has been impregnated throughout the accessible void spaces. Graph-i-tite “G” has been fired to graphitizing temperatures. Graph-i-tite “A” was fired at a lower temperature and is described as an amorphous carbon- filled material. None of the above grades approach the gastightness of <1078 cm?/sec for helium at STP, required for MSR core applications. Additional tests, such as compati- bility and pore entrance diameter spectra, should be made to better evaluate these as structural materials for the fuel salt processing systems, which do not require the low gas permeabilities. 13.6 REDUCTION OF HELIUM PERMEABILITY OF GRAPHITE BY PYROLYTIC CARBON SEALING C. B. Pollock The breeding performance of a molten-salt reactor is significantly affected by the extent to which '*®Xe can 7. These were determined with a mercury porosimeter. 152 Table 13.3. Nominal helium permeability parameters for various grades of graphite

- Kyt = Bo——+ %Ko, where Ky = permeability coefficient for helium at 28°C (cm2 [sec), Bg = viscous permeability (cm2),

= average pressure across specimen (dynes/cmz), n = viscosity of gas, helium (poises), K¢ = slip coefficient (cm), V= average molecular velocity of gas helium (cm /sec). Specimen . Buk By Ko L He Grade Source? number Orientation density (em?) (cm) (em? fsec) (g/cm®) X10~ x107° 103 2020 Stackpole KG1-1 WG 1.71 224 64.7 166 KG11-1 AG 1.72 156 493 123 2044 Stackpole Al4M-319 AG 1.72 123 45.6 108 A141-12¢ AG 1.79 7.14 6.04 12.0 1226 Stackpole A-2M-149 AG 1.75 33.2 20.6 43.2 A-21-2¢ AG 1.81 6.99 6.19 12.2 Graph-i-tite “G” Carborundum 52 WG 1.88 5.62 242 5.53 62 AG 1.88 1.87 0.856 1.92 Graph-i-tite “A” Carborundum 51 WG 1.90 0.152 0.177 0.337 61 AG 1.89 1.19 0.553 1.23 62 AG 1.90 0.074 0.087 0.167 4Stackpole Carbon Company and Graphite Products Division of Carborundum Company. bThe specimen is a disk, nominally 1.000 in. in diameter and 0.500 in. thick, and the direction of helium flow is parallel with the axis of rotation of the disk. ‘WG = with grain, parallel with the general g-axis orientation; AG = across grain, perpendicular to the general ¢-axis orientation. dSpecimen from unimpregnated core of the stock. €Specimen from the liquid-hydrocarbon-impregnated zone of the stock. be stripped from the fuel salt and prevented from entering the graphite moderator. Effective sealing of graphite surfaces is, therefore, highly desirable. Pyro- lytic carbon is the desired sealing material because of its low neutron cross section and its compatibility with the fuel salts. We have developed two techniques for sealing with pyrolytic carbon. The first is designed to plug surface pores with pyrolytic carbon using a vacuum- pulse impregnation process, and the second is a macro coating process in which a continuous layer of pyrolytic carbon is deposited on the surface of the graphite. During this report period we did not do any additional work on the impregnation process other than the continued examination of previously prepared speci- mens with the aid of scanning electron microscopy (see Sect. 13.7). We continued to fabricate and study the properties of surface coatings. The furnace that we are using was described in the last semiannual report. The samples are of conventional HFIR geometry (see below), but all sharp corners have been rounded. The samples were fixed in place in the furnace using an arrangement shown in Fig. 13.11. The vertical position of the samples can be adjusted in the center of the furnace, and the supports are anchored at the top and the bottom of the reaction tube. This scheme is amenable to scale-up and changes in sample geometry. We have conducted coating experiments varying conditions of gas supply rate, gas mixtures, tempera- ture, and time. Early experiments were shotgunned in order to determine operating ranges that are practical. In general, the sequence of operation is to heat the 153 Y-111076A FO B INCHES Fig. 13.11. Sample holder and sample for pyrolytic carbon coating studies. Table 13.4. Description of typical molten-salt graphite samples sealed by coating with pyrolytic carbon Coating Coating Coating Helium Y —-442598 Sample temperature time thickness permeability (OC) (min) (mils) (cmzlsec) MS-80 1100 10 1 1x 10! MS-81 1100 10 1 1 x10*° MS-83 1100 15 2 1x 10710 MS-87 1100 20 3 LY 10F2 13-1 1150 10 3 1x107'° 17-1 1200 8 4 1x107° sample to reaction temperature in the presence of a substitute gas (helium) that levitates a quantity of small carbon beads around the sample. Then a carefully controlled amount of the gaseous hydrocarbon (C3Hg) mixed with a diluent gas (helium) is allowed to enter the reaction chamber, and pyrolysis occurs. After the designated coating time has elapsed, the coating gas is replaced with an inert substitute gas, and the sample temperature is increased to greater than 1800°C in order to stress-relieve the coatings. The sample is then INCHES slowly cooled to room temperature. We have coated specimens to temperatures in the Fig. 13.12. A view of a graphite sample coated with range 1100 to 1400°C. Table 13.4 describes some pyrocarbon. 0 OGN N02. ¥ 0.4 L | o S coating parameters and helium permeabilities after coating. The effects of gas supply rate and the mixtures of the hydrocarbon and a diluent gas are still unre- solved, but in the lower temperature ranges a large excess (200% by volume) of diluent gas works very well, and, of course, as the hydrocarbon supply rate is decreased the carbon deposition rate also decreases. 154 L.OOT INCHES n, ", 500X Fig. 13.13. The microstructure of a sample coated with pyrolytic carbon at 1150°C. The upper photomicrograph is a bright field Coating time is used to determine coating thickness, which has ranged from 3 to 7 mils. ' Figure 13.12 shows a sample coated with 1 mil of pyrolytic carbon at 1150°C. Figure 13.13 shows the ~ microstructure of a coated sample at 500X. The upper Y-112902A view of the microstructure at 500X, and the lower view is a polarized light view of the same area. 155 photomicrograph shows a bright field view, and the lower view shows the response to polarized light. The physical characteristics of the coating are being examined, but only partial results are available. The density of material deposited at 1150°C was 2.10 g/cm®, while material deposited at 1200°C had a measured density of 2.08 g/cm?. Visual examination of the samples indicates that the coatings are isotropic, but at least one sample did respond to polarized light in an anisotropic manner. The bond between the coating and the graphite substrate appears to be quite good, and to some degree the surface pores have been impregnated with pyrolytic carbon. 13.7 CHARACTERIZATION OF PYROCARBON SEALANTS FOR GRAPHITE USING REFLECTED LIGHT AND SCANNING ELECTRON MICROSCOPES® W. H. Cook We have continued our detailed characterization of graphite sealed with pyrocarbon.® Our objectives are to learn more about the pyrocarbon sealing techniques and, in particular, to determine what produces a pyrocarbon seal that has maximum resistance to damage by fast neutrons. The basic specimens being studied are hollow cylin- ders of pyrocarbon-sealed grade AXF graphite, as described in the preceding section. We have been examining both unirradiated and irradiated specimens with reflected-light and scanning electron microscopes. The scanning electron microscope (SEM) is a recently acquired one'® that has improved resolution. Most of the specimens that we have examined have been sealed with a 1,2-butadiene source of pyrocarbon at 700 to 750°C in fluid-bed coaters that were originally designed to coat uranium oxide or uranium carbide fuel particles 300 to 400 um in diameter. For specimens coated in this way we found that 8. The scanning e¢lectron microscope work was done by R. S. Crouse and D. R. Cuneo, and the reflected-light microscopy work was done by M. D. Allen, all of the Metals and Ceramics Division. 9. W. H. Cook, MSR Program Semiannu. Progr. Rep. Aug. 31,1971, ORNL-4728, pp. 120-21. 10. Model JSM-U3 manufactured by the Japan Electron Optical Laboratory. Y- 110050 i Fig. 13.14. Scanning electron photomicrographs showing cracks and carbon debris in 0.001-in.-thick pyrocarbon coating deposited from 1,2-butadiene, C4Hg, at 700°C onto grade AXF graphite, 156 1. there is a lack of control and uniformity in the pyrocarbon sealants, 2. some coatings had cracked during their deposition, 3. sharp corners on the specimens tended to cause flaws in the pyrocarbon sealing, which acted as crack propagation centers during irradiation, 4. there is evidence that surface cracks were created or existing surface cracks enlarged during irradiation. The evidence for these observations is briefly described below. Figure 13.14 is a series of SEM photomicrographs showing cracking and carbon particle debris in an unirradiated specimen. In Fig. 13.14q there is a carbon particle in the center of the photomicrograph with a family of cracks radiating out from it. Figure 13.14 b, ¢, and d shows the crack and carbon debris at increasingly greater magnifications. Figure 13.15 is a transverse fractograph through this specimen. One can clearly see the base stock graphite (grade AXF) and the pyrocarbon coating. There is an indication that the PYROCARBON COATING GRADE AXF latter may be a duplex type of coating in which the final deposition was a thin layer of pyrocarbon differ- ent from the major part deposited in and on the base stock. A transverse view through two cracks in the coating is shown in the reflected light photomicrograph in Fig. 13.16, which is a polished section of the specimen. The pyrocarbon coating in this region is approximately 0.001 in. thick, and the cracks are almost completely through the pyrocarbon coating. Barely discernible in Fig. 13.16 is a thin final pyrocarbon coating over the surfaces of the cracks and the main coating. At higher magnifications with polarized light or a rotatable sensitive tint plate, this is resolved and verifies the SEM indication on this. This thin coating and the cross section of the carbon debris particle are both weakly anisotropic, while the bulk of the pyrocarbon coating is an amorphous, isotropic material. The final thin film plus the fact that the cracks stopped short of pene- trating the coating probably accounts for this unirradi- ated specimen having a relatively low helium per- meability of 2.9 X 10™® cm?/sec. R-57567 A 0.0025 in. Fig. 13.15. A scanning electron micrograph of the transverse fractured surface of a pyrocarbon coating deposited from 1,2-butadiene, C4Hg, at 700°C onto grade AXF graphite. 1000X. 157 PC COATING GRADE AXF Y-140038 L 1 10.001 in. T 10.003 in. 0.007 INCHES 500X 10.005 in. + [0.007 in. Fig. 13.16. Reflected-light micrograph of the polished surface of a transverse section of a pyrocarbon coating deposited at 700°C from 1,2-butadiene, C4Hg, onto grade AXF graphite. As polished. 500X, The nature of cracks in an irradiated specimen was distinctly different, as can be seen in Fig. 13.17. The helium permeability of this specimen, which was 4.4 X 107'® cm?/sec in the unirradiated condition, had increased to 7.2 X 107° ¢m?/sec after being exposed to a fluence of about 1.7 X 10®? neutrons/cm? (£ > 50 keV) at 715°C. Although the cracks are numerous, they do not appear to completely penetrate the pyrocarbon. Whether the cracks were created during the irradiation or are modifications of V-shaped cracks that were in the pyrocarbon coating prior to the irradiation, the square- bottom shape of the cracks probably indicates shrink- age and creep of the pyrocarbon under the fast-neutron irradiation. Figure 13.18z illustrates a typical defect in the pyrocarbon sealant that occurs on the sharp comers of a specimen, and Fig. 13.18b shows how this type of fault can be enlarged and propagated under fast-neutron irradiation. In an attempt to eliminate these faults at sharp edges, we are now rounding the edges of specimens prior to sealing them with pyrocarbon. The other changes that we have made in efforts to improve the coatings are to change to the furnace described in Sect. 13.6 and to turn to propene as the source of the pyrocarbon. (There is evidence that pyrocarbon from propene has superior resistance to neutron damage.! ) A comparison of the pyrocarbons derived from 1,2-butadiene (C4Hy) and propene (C3Hg) is shown in Fig. 13.19. The 1,2-butadiene was deposited at 700°C in the fluid-bed fuel-particle coating furnace, and the propene was deposited in the new furnace at 1200°C. (It was found, after this specimen was prepared, that depositing pyrocarbon from propene at 1150°C was more effective in sealing the base stock, but we expect that the geometric structure of the pyrocarbon de- posited from propene at 1150 and 1200°C will be essentially the same.) Note that the pyrocarbon from the 1,2-butadiene tends to be in the form of smooth spheroids and that from the propene is in smaller, irregular-shaped particles. From both sources, the larger pyrocarbon particles are usually made up of agglom- 11. D. M. Hewette II and C. R. Kennedy, MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL4548, pp. 215—18. 158 Fig. 13.17. A scanning electron photomicrograph of a cracked pyrocarbon coating on grade AXF graphite after an exposure at 715°C to an accumulated fluence of 1.7 X 10?2 neutrons/cm? (>50 keV). (a) 100X. (b) 500X. (¢) 1000x. (d) 3000X. R58083 Fig. 13.18. Scanning electron microscopy photomicrographs of the outside edges of pyrocarbon-coated grade AXF graphite specimens. (¢) A typical as-coated edge with some cracking and spalling; (b) a specimen has accumulated a fluence of 2.1 X 1022 neutrons/cm? (£ > keV) at 715°C and has relatively more spalling and cracking. 100X. 159 erates of smaller ones; however, this appears to be more frequently the case for that from the propene. Because we only recently acquired the capability for making very sharp SEM pictures of graphite surfaces, we have not yet examined a specimen, irradiated it, and then reexamined the same spot. This should provide more accurate information about the pyrocarbon coat- ings and the neutron irradiation effects than can be obtained from the comparison of irradiated specimens with other unirradiated specimens that we have made to date. 160 Fig. 13.19. Scanning electron microscopy photomicrographs of pyrocarbon coatings deposited on grade AXF graphite. (a) Pyrocarbon deposited from 1,2-butadienc, C4Hg, at 700°C and () pyrocarbon deposited from propene, C3Hg, at 1200°C. 500X. 14. Hastelloy N H. E. McCoy 14.1 DEVELOPMENT OF A TITANIUM-MODIFIED HASTELLOY N H.E.McCoy B. McNabb Our initial work to improve the resistance of Hastel- loy N to embrittlement by neutron irradiation involved alloys with titanium additions up to 0.5%.!°? These alloys had good postirradiation properties when irradi- ated at 650°C, but the properties deteriorated rapidly as the irradiation temperature was increased.® The good properties were associated with the formation of fine MC-type carbides and the poor properties with the formation of coarse M,C-type carbides. Studies of a series of laboratory melts containing up to 3% Ti showed that the titanium concentration for optimum properties was about 2%.* Above this level the brittle intermetallic NizTi formed, and the ductility de- creased.’ During this report period, studies of small commercial melts have confirmed our work with small laboratory melts. These test results will be described in some detail. One necessary experimental modification in our postirradiation creep program is to measure the strain that occurs on loading. Since many tests are run above the yield stress, this strain is often quite large. Experi- mental techniques have been established to measure this strain, but the need exists for a method to correct the strains for tests run before these new techniques were established. Figure 14.1 was prepared from some of our 1. H. E. McCoy et al., “New Developments in Materials for Molten-Salt Reactors,” J. Nucl. Appl. Technol. 8(2), 156—69 (February 1970). 2. H. E. McCoy and J. R. Weir, ASTM Special Technical Publication 457, pp. 290311 (1969). 3. M. W. Rosenthal et al., “Recent Progress in Molten-Salt Reactor Development,” At. Energy Rev. 9(3) (August 1971). 4. H. E.McCoy and C. E. Sessions, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 192. S. R. E. Gehlbach and S. W. Cook, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 129. 161 latest test results. The stress has been “normalized” by dividing it by the yield stress of the particular alloy at 650°C. The results seem rather consistent and define a line that can be used to predict the loading strain on tests run previously. Some of the results on heat 66-548 presented in the following figures were corrected by this correlation. The test results from the four small commercial melts described in Table 14.1 will be compared in the next several figures to show the improvement in properties as the titanium level is increased. The main element that varies is titanium, but the variations in carbon level are also likely significant. The stress-rupture properties of the modified alloys and standard vacuum-melted Hastelloy N are compared in Fig. 14.2. The alloy containing 0.45% Ti has lower stress-rupture properties than the standard alloy, pri- marily because of the reduction of molybdenum in the modified alloys. The two alloys with 1.1% Ti have improved stress-rupture properties, and the alloy with 2.1% Ti has even better properties. The stress to produce rupture in 1000 hr varies from 34,000 psi for the alloy with 0.45% Ti to 53,000 psi for the alloy with 2.1% Ti. ORNL-DWG 72-7674 [ M T=—F | [ : i L H ~H " - JF oe T SPEC. FROM CUT BLANK + 11— | T @ ; R 5, ) L — ‘ o - & b “W % g 100 _ e . e » o - = A— , o 1o e 7 i O : i DS ST | | | = H T n ‘ N o B & ] | g e Q i oo : | oot ‘ ! ! o ! L | 102 2 5 ot 2 5 o 2 5 10! STRAIN ON LOADING (%) Fig. 14.1. Relation between creep stress and strain on loading at 650°C. STRESS (1000 psi}) Table 14.1. Chemical compositions of titanium-modified commercial alloys Alloy Concentration (wt %) number Mo Cr Fe Mn Si Ti Zr Hf Nb B C 66-548 124 7.7 0.03 0.14 0.05 045 <0.001 a 0.0003 0.00007 0.040 70-785 123 7.0 0.16 0.30 009 1.1 0.012 <«0.003 0.097 0.0020 0.057 67-548 12.0 7.1 0.04 0.12 0.03 1.1 0.002 a 0.0005 0.0007 0.082 70-727 13.0 7.4 0.05 0.37 <0.05 2.1 0.011 <0.01 <0.01 0.0008 0.044 4Not analyzed for, but should be <0.01%. 70 ORNL-DWG 72-3296R ’ ‘ | | HTT o ORNL-DWG 72-3297R LT 1] 60 50 — 40 - = 3 S — 30 = . STANDARD VACUUM J\ 8 O 66-548(0.45%Ti) | MELTED ALLOY o & 70-785 (1.4% Ti) 20 ¢ o508 oasmTh b e 0 67-548 (1.1%Ti) o] - .4 o 11 7/ - o, H & S0 res (D% 2 A 0 70-727 (2.1 /°T|) 0 0 67-548 (14 % Ti) “~1-STANDARD vacuum @ 774 (.75%Ti) 0 70-727 (24 %WTi) MELTED ALLOY ® 717583 144 % T ® 7i- 114 (1.75%Ti) UL " 783 UaamT) | | [Hl 10 - o L RETN LU L1 2 To} 102 10° 104 RUPTURE TIME (hr) o LU T LTI L LPLI 1 5 10°° 10°2 Tomk 10° Fig. 14.2. Stress-rupture properties of titanium-modified MINIMUM CREEP RATE (%/hr) alloys at 650°C. The creep rates of these same alloys are compared in Fig. 14.3. The creep strength is about equivalent for the standard alloy and the alloy modified with 0.45% Ti. Further increases in titanium increase the creep strength. The stress to produce a creep rate of 0.001%/hr ranges from 37,000 psi for the alloy with 0.45% Ti to 55,000 psi for the alloy with 2.1% Ti. The fracture strains of the modified alloys are compared in Fig. 14.4. For rupture lives of about 100 hr the alloys with 1.1 or 2.1% Ti have higher fracture strains than those of the alloy with 0.45% Ti. However, for rupture lives in excess of 1000 hr, the fracture strains appear to be approaching each other. Samples of the four modified alloys have been irradiated at various temperatures between 650 and 760°C to thermal fluences of about 3 X 102! neu- trons/cm?. The stress-rupture properties of these alloys Fig. 14.3. Creep properties of titanium-modified alloys at 650°C. are shown in Fig. 14.5. The contrast between the curves for alloy 66-548 (0.45% Ti) irradiated at 650 and at 760°C shows the very important influence of irradia- tion temperature. The other alloys were not irradiated over such a large temperature span, and it is not clear that the variation of irradiation temperature from 700 to 760°C had much of an influence on the properties. When irradiated at 760°C, the postirradiation stress- rupture properties improve in the order of 66-548, 70-785, 67-548, and 70-727. The postirradiation creep properties of these alloys are shown in Fig. 14.6. The higher irradiation tempera- ture has a very detrimental effect on the creep strength of heat 67-548. The creep strengths after irradiation at 760°C improve in the order 66-548, 70-785, 67-548, and 70-727. 163 ; ORNL-DWG 72-3298 The parameter of most importance is the postirradia- 0 "] oe6-5as0as%Ti) ||| | | |11l tion fracture strain, shown in Fig. 14.7. Because the 5 70-785(1.1% Ti; number of points is so small, the lines that have been O67-548 (1.1%Ti oy : . 60 4 70-727 (24 % Ti) - drawn are at best only indicators of relative ductility. The results for alloy 66-548 illustrate the large effect of irradiating at 650°C compared with 760°C. The maxi- mum fracture strain observed for alloy 66-548 after irradiation at 760°C was 0.5%. Alloy 70-785 had fracture strains in the range of 1 to 4%. Alloy 67-548 had fracture strains in the range of 1 to 4% out to rupture lives of 100 hr, and then the strains increased to 8% in 2000 hr. The behavior of alloy 70-727 is difficult to define. The data cover a band from 3 to 9% with no apparent dependence on irradiation temperature or rupture life. Although these results are rather limited, we feel that 10 they support the conclusion that an alloy modified with 2% Ti with adequate resistance to embrittlement can be o l | | \Ufl | ] | ||Ul ] | J_LH developed. Small commercial melts from several ven- 10’ 102 103 104 dors are being evaluated. RUPTURE TIME (hr) FRACTURE STRAIN (%) 14.2 ALLOYS WITH EXCEPTIONAL STRENGTH Fig. 14.4. Fracture strains of titanium-modified alloys at 650°C. H.E. MCCOY All of our modified alloys are as strong as or stronger than standard Hastelloy N. However, three alloys having 20 ORNL-DWG 72-3299 AL I 1Y A S R IO A VAL O 66-548(0.45% Ti) A 70-785(1.4% Ti) 60 |— 66-548 0 67-548(1.1% Ti) 760 (650°C) “\_654 ~.7!8 ¢ 70-727(2.4% Ti) STANDARD ALLOY, 50 L AIR MELTED, 40 4 30 STRESS (1000 psi) 20 ~_ T 70-785 - 66-548 STANDARD ALLOY, ~ Pa VACUUM MELTED, 760°C R R N T 107! 100 10! 102 103 104 RUPTURE TIME (hr) Fig. 14.5. Postirradiation stress-rupture properties at 650°C of titanium-modified alloys after irradiation at the indicated temperature to a thermal-neutron fluence of 3 X 102° neutrons/cm?. 164 ORNL-DWG 72-3300R IO T T I T I T I ' STANDARD ALLOY o | © 66-548(0.45%Ti) / AIR MELTED, 760°C A 70-785(11% Ti) 718 D 67-548(1.1% Ti) o~ 5o L0 70-727 (2.4 % i) /' 66-548 _ et 650 (650°C) o 760 650 760,/ A-87-548 g 718 == 760663 760 L~ =~ o// j’wso AP % 0 70-727" 650/"730/; =/ 118 @ G /n%fégfla /0760 4% ?go/ 20 " 70-7857 8 “\ 760 STANDARD ALLOY 780 80 10 ™ vACuUM MELTED, ‘ reoe 66-548 S v 1 O S M O A Y 1074 1073 1072 107" 100 10 MINIMUM CREEP RATE (% /hr) Fig. 14.6. Postirradiation creep properties at 650°C of titanium-modified alloys after irradiation at the indicated temperature to a thermal-neutron fluence of 3 X 10%° neutrons/cm?. ORNL-DWG 72-3301 4 14 T o) L PTTINIT Fesa [T T gesl [TH] ] IHTWH R 0650 0 66-548(0.45% Ti} . a 70-785(11%Ti) | O 67-548 (1.1% Ti) 623 0 70-727 (2.4% Ti) — 10 : 5 718 g EBO 0 67-548 5 S~ 650 760 8 ~ - 0 —0— e = S~ 746 760 L 6 § 0650 > 718 I\ [T 4 A - S ‘ 718 ) —— o S i : — 76058 — , A 70-785 760 760 ) 760 ol 86548 | | LUl il 107! 10° 101 102 103 104 RUPTURE LIFE ¢(hr) Fig. 14.7. Postirradiation fracture strain at 650°C of titanium-modified alloys after irradiation to a thermal fluence of 3 X 102° neutrons/cm? at the indicated temperature. 165 nominal additions of 0.5% Ti and 2% Nb have excep- tionally high strength at 650°C. These alloys were small (50 to 100 1b) commercial melts that were obtained as 1 -in.-thick plate. The chemical compositions are given in Table 14.2. Alloys 69-648 and 70-835 were prepared by one vendor and were double vacuum melted. Alloy 69-344 was prepared by another vendor by the electro- slag remelt process. The high silicon content of this alloy resulted from the slag. The stress-rupture properties at 650°C of the three alloys are compared with those of standard Hastelloy N in Fig. 14.8. The allowable stress for a rupture life of 1000 hr ranges up to twice that for standard Hastelloy N, varying concurrently with the combined concentra- tions of niobium and titanium in the three alloys. The minimum creep rates at 650°C are shown in Fig. 14.9. The allowable stress for a minimum creep rate of 0.001%/hr ranges to more than twice that for the standard alloy. The creep strength does not vary in the same order as the rupture strength. Very limited creep testing has been done at 704°C, and the results are presented in Table 14.3. There is a marked increase in the strength of the modified alloys ORNL-DWG 72-7676 STRESS (1000 psi}) %0 T T T 1] N T : | } .)/ 80 I F 0,6'5 - 70 : : V. gl Lol Pt A e ,/c A A/( a /XT | 60 gl - ! o, Peq W d G ! LA - le JA* //4//Ma /,/ 50 Paley d fe” ] ML aa"" 0 // Pl / /’»/ / I 0»9\9// 40 / /. o 'K o WL <_jjpfi/ 1 / »V(?\Syr/ 30 / 4_/TT ! /( ‘ , Vd ' ! - § ¢ 1 L DT ot 2 5 0° 2 5 02 2 5 10! 2 5 MINIMUM CREEP RATE (% /hr) Fig. 14.8. Stress-rupture properties at 650°C of three alloys with nominal modifications of 2% Nb plus 0.5% Ti. over that of standard Hastelloy N at the test conditions used. At the upper end of the temperature range that we are considering for MSBR structural materials, the superiority over standard Hastelloy N is less. Figures 14.10 and 14.11 present the results of creep tests at this temperature (760°C). Alloys 69-648 and 70-835 have creep strengths about 25% greater than those of standard Hastelloy N, while the strength of alloy 69-344 is not significantly above that of standard Hastelloy N. The increased strength of these alloys seems to result from a very finely dispersed precipitate. Photomicro- graphs of alloys 69-648, 70-835, and 69-344 are shown in Figs. 14.12, 14.13, and 14.14 respectively. The precipitate is visible on the strained portion of the specimen and is so fine that it cannot be resolved by optical microscopy. Electron micrographs of heat 69-648 have been reported previously.® Even at magni- 6. R. E. Gehlbach and S. W. Cook, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4729, pp. 125-32. ORNL-DWG 72-7675 STRESS (41000 psi} 30 — H Y i < ‘ i \~ | N \ l DL I 2 5 102 2 5 105 3 RUPTURE TIME (hr) 20 L \ J\i‘ Fig. 14.9. Creep properties at 650°C of three alloys with nominal modifications of 2% Nb plus 0.5% Ti. Table 14.2. Compositions of experimental alloys Alloy Concentration (wt %) No. Mo Cr Fe Mn Si Ti Zr Hf Nb C 69-648 128 6.9 0.3 0.34 0.05 092 0.005 <0.05 1.95 0.043 69-344 13.0 74 4.0 056 054 077 0019 <01 1.7 0.11 70-835 125 7.9 0.68 0.60 0.05 0.71 <0.005 0.031 2.60 0.052 166 Table 14.3. Creep-rupture properties at 35,000 psi and 704°C of standard Hastelloy N and several alloys modified with 0.5% Ti and 2% Nb Heat Rupture life Minimum creep Fracture strain Reduction in No. (hr) rate (%/hr) (%) area (%) 5065 (std) 28.4 0.23 14.8 14.2 69-648 732.2 0.025 21.6 18.0 69-344 181.6 0.048 41.6 39.0 70-835 898.1 0.032 62.3 51.0 ORNL-DWG 72-7677 ORNL-DWG 72- 7678 50 T T TTTT0 50 [T 71T O 69-648 O 69-648 40 A 70-835 4017 5 70-835 oy 4 _ = _ e & 30 N H :\ 0 69-344 2 5o | O 69-344 | Qo N \\ Q D/ V /’ 3 NN \m\ 8 /z fiv @ 20 ~\\- N 2 20 #L/ ’% w NN N ”n /:fl 2 \fi\\ ‘\ W e P n NN :.n\ 'ON 5 ) /F/ = r ! STANDARD HASTELLOY N H \\ ‘// §/< STANDARD HASTELLOY N S L N 0 Pad IR 10! 102 103 104 1073 1072 107! 10° RUPTURE TIME (hr) Fig. 14.10. Stress-rupture properties at 760°C of three alloys with nominal modifications of 2% Nb plus 0.5% Ti. fications of about 50,000, it is difficult to resolve the strain fields of individual particles. The precipitate has not been identified, but we suspect that it is basically an Ni3(Nb,Ti) compound. Future studies will involve determining (1) the mechanical-thermal conditions required to form the precipitate and (2) the postirradiation mechanical prop- erties of the alloy with the precipitate present. 14.3 WELDABILITY OF COMMERCIAL ALLOYS OF MODIFIED HASTELLOY N B.McNabb H.E.McCoy Two small commercial heats of modified Hastelloy N were procured from the Stellite Division of Cabot Corporation. The starting materials were consolidated as a single vacuum melt of 120 b and cast into two electrodes. One of these was consumable vacuum remelted (heat 71-114), and the other was electroslag remelted (71-583). The vendor’s analysis of heat 71-114 in weight percent was 78.1% Ni, 12.03% Mo, 7.2% Cr, 0.06% Fe, 0.01% Mn, 0.05% C, 0.04% Si, 0.12% Al, 1.96% Ti, 0.002% B, 0.002% P, and 0.0055% S. The MINIMUM CREEP RATE (%/hr) Fig. 14.11. Creep properties at 760°C of three alloys with nominal modifications of 2% Nb plus 0.5% Ti. analysis of heat 71-583 was 78.34% Ni, 12.16% Mo, 7.2% Cr, 0.06% Fe, 0.01% Mn, 0.05% C, 0.05% Si, 0.12% Al, 1.79% Ti, 0.001% B, 0.002% P, and 0.004% S. These heats were received as '%-in -thick plates about 1 ft square. Strips % in. square were sawed from these plates and swaged into Y%- and %,-in.-diam. welding wire. The remaining plates were beveled for welding with a 100° included angle V groove between the plates to be welded. The plates were welded to a steel strongback in pairs for a fully restrained weld. These were then welded, filling the V groove with weld wire identical to the base metal in composition. Both sets of plates welded very well with no cracking during or after welding by visual, dye penetrant, and x-ray inspection. Heat 71-583 appeared to weld a little more easily, and the weld metal flowed slightly better than for heat 71-114. Side-bend specimens % in. thick were sawed trans- verse to the welding direction and bent 180° around a %-in.-radius mandrel. No cracks were evident by visual or dye-penetrant inspection, as shown in Fig. 14.15. (Some darker areas in the figure are due to differences in the thickness of the developer used and some due to 167 Y —-100081 0.035 INCHES [N 100X [ | i __Y—100082 .o htd e < £ES 100X T 0.035 117 H Fig. 14.12. Photomicrographs of alloy 69-468 annealed 1 hr at 1180°C and tested at 55,000 psi at 650°C. Failed after 1759 hri with 20.5% strain. (¢) Fracture, ({b) unstressed shoulder. Etchant: glyceria regia. 168 Y-413619 0.035 INCHES i ST g ‘ ‘ Y -143347 g W - o L [e St v L ! ¥ ‘; ~ ” i i :w ," i sy // . R L lt ; 2 e - P 2 e [%2] . g - - ~ Sl - < ", 2io : R " . N . SE : v § R, t Y & “' K b = ¥* 7 < 3 - T e 13 o 2 w .o s e i HE ¥ (C) - ¢ RN Sy — o L i Fig. 14.13. Photomicrographs of alloy 70-835 annealed 1 hr at 1180°C and tested at 55,000 psi at 650°C. Failed after 2592 hr with 30.6% strain. () Fracture — 100X, () typical of stressed section — 500X, (c¢) typical unstressed portion — 100X. Etchant: glyceria regia. 169 Y-113298 0.035 INCHES ™~ 100X fod Fig. 14.14. Photomicrograph of the fracture of a sample of heat 69-344 that was annealed 1 hr at 1180°C and tested at 40,000 psi at 650°C. Failed after 2410 hr with 13.9% strain. Etchant: glyceria regia. 170 Y-110352 Y-110351 0 0.5 1.0 N R T T INCHES Fig. 14.15. Side-bend specimens of Hastelloy N weld specimens of two heats modified with 2% Ti. (z) Heat 71-1114, consumable vacuum remelted, (b) electroslag remelted. Dye penetrant has been applied, and no cracks are visible. the developer being rubbed off in handling before photographing.) Some transverse tensile specimens were machined from the weld area with a gage or reduced section (Y, in. in diameter) containing about half weld and half heat-affected zone and base metal. Some of these specimens were creep tested in the as-welded condition at 650°C and 55,000 psi and some at 760°C and 20,000 psi. Specimens of the base metal of each heat were tested at the same conditions, and the creep rupture properties are tabulated in Table 14.4. The rupture life, minimum creep rate, and fracture strains were reduced by welding. The lower minimum creep rates for the welded specimens indicate higher strengths than the base metal specimens. Table 14.5 is a tabulation of the 171 tensile properties of these two heats at 25, 650, and 760°C. The yield strengths of the welds were consider- ably higher at each of these temperatures, the ultimate strengths were reduced, and the fracture strains re- duced. The fracture strains probably can be increased by postweld annealing, and this is being investigated. In summary, these results confirm our previous observa- tions that the 2% Ti-modified alloy has good weld- ability. 14.4 ELECTRON MICROSCOPE STUDIES R. E. Gehlbach S.W. Cook Our microstructural studies to characterize precipita- tion in Hastelloy N alloys have primarily involved Table 14.4. Comparative creep-rupture properties of base metal and transverse weld specimens of modified Hastelloy N (2% Ti) temg::; ture Stress Specimen® Rupture life Minimum creep Fracture strain Reduction in C0) (psi) (hr) rate (%/hr) (%) area (%) 650 55,000 71114-BM 119.0 0.140 443 36.8 71114-TW 68.0 0.034 6.1 31.5 71583-BM 63.1 0.111 314 26.0 71583-TW 6.8 0.055 34 9.9 760 20,000 71114-BM 281.5 0.078 44.2 41.9 71114-TW 142.8 0.0037 3.9 7.1 71583-TW 232.1 0.109 529 41.5 71583-TW 148.8 0.0044 33 7.8 2Number refers to the heat; letters indicate a transverse specimen from the weld area (TW) or a specimen of base metal (BM). Table 14.5. Tensile properties at various temperatures for base metal and transverse weld specimens of 2% titanium-modified Hastelloy N Heat Type Test Yield stress Ultimate stress Fracture strain Reduction in . g temperature . . No. specimen €0 (psi) (psi) (%) area (%) 71-114 BM 25 44 800 120,700 72.6 59.0 W 25 80,000 112,300 30.8 52.1 BM 650 29,100 87,000 56.2 39.7 ™wW 650 55,700 66,800 11.5 294 BM 760 28,700 64,300 52.8 504 W 760 52,700 66,600 10.3 29.0 71-583 BM 25 45,900 121,900 73.3 60.2 T™W 25 84,600 108,600 27.6 545 BM 650 30,200 86,000 49.0 384 ™ 650 52,300 65,500 11.7 36.1 BM 760 29,500 64,400 45.0 47.2 TW 760 50,400 60,100 5.3 16.0 2BM — base metal; TW — weldment, as welded. observations made on NizTi and strain-induced precipi- tation in several laboratory and small commercial heats. Studies are under way to characterize two titanium- modified and two hafnium-modified heats recently received from Cabot Corporation. In addition, we have modified our x-ray diffraction techniques to greatly improve detectability and precision of phase analyses. 14.4.1 Intermetallic Precipitation in Hastelloy N As previously reported,® precipitation of the len- ticular Ni3Ti occurs in small laboratory heats of Hastelloy N when the titanium concentration is ap- proximately 2% and above. Recently we have observed this precipitation also in a 1% Ti heat. Although Ni;Ti precipitation is very profuse and homogeneous in the 2.9% Ti alloy, it nucleates at MC carbides present in the stacking-fault morphology for lower titanium concen- trations. (This carbide morphology is obtained during aging after annealing at 1260°C, rather than the usual 1177°C, and forms around primary MC particles which are partially dissolved during the anneal.) The inter- metallic distribution in alloys containing 2.4 and 0.98% Ti is shown in Figs. 14.16 and 14.17 respectively. Both specimens were annealed 1 hr at 1260°C prior to exposure at 700°C for 1000 and 10,000 hr respectively. A different morphology of intermetallic precipitation, shown in Fig. 14.18, was observed in the highest (2.9%) ' YE-10614 Fig. 14.16. Transmission electron micrograph showing pre- cipitation of Ni3Ti on MC carbides in Hastelloy N containing 2.4% Ti. The carbides, in a stacking-fault morphology, result from annealing at 1260°C prior to aging 1000 hr at 700°C. 12,000x. 172 titanium-modified heat. The specimen was aged 10,000 hr at 700°C following a 1-hr anneal at 1260°C. This morphology occurred as rather extensive sheets in the vicinity of primary carbide particles and appeared to Fig. 14.17. Transmission electron micrograph showing small amounts of Ni3Ti precipitating on MC carbides in Hastelloy N containing 0.98% Ti. The material was aged 10,000 hr at 700°C after a 1260°C anneal. 12,000X. Fig. 14.18. Precipitation of Ni3Ti in two morphologies around primary and thermally induced MC carbide precipitates in Hastelloy N containing 2.9% Ti. The sheetlike phase may be the stable eta phase. Annealed at 1260°C and aged 10,000 hr at 700°C. 12,000x%. Fig. 14.19. Strain-induced precipitation along grain boundary, in commercial 2.1% Ti-modified Hastelloy N. The precipitate may be eta phase. 12,000X. extend in the same direction [although not necessarily on (111) planes] as the MC carbides which precipitated in the stacking-fault morphology. We have not positively identified these intermetallic phases but have indications that the fine lenticular phase is the metastable gamma-prime (y") Ni; Ti, where- as the second type is the stable n-Ni; Ti phase. Examination of a commercial heat of 2.1% Ti- modified Hastelloy N (70727) revealed precipitation of a noncarbide phase in a sheetlike morphology (Fig. 14.19). This precipitate appeared to nucleate at grain boundaries in the gage section of creep samples which were stressed at elevated temperatures. It was not present in unstressed material. Electron diffraction patterns indicated that it may be eta phase (Ni;Ti);its morphology is similar to that observed both in the laboratory heat containing 2.9% Ti and in the stressed titanium-niobium commercial alloys discussed. pre- viously. Other studies’ indicate that nucleation of eta phase is enhanced by deformation. Future studies will deal with specific identification of intermetallic-phase precipitates and the relationship between strain-induced and intermetallic precipitation in the modified alloys. 14.4.2 Precipitation in New Commercial Alloys We have examined material from several new titanium-modified and hafnium-modified commercial alloys. The titanium-modified heats were prepared both 173 by electroslag remelting and vacuum arc remelting and contained 1.5 to 2.0% Ti. Primary MC carbides (a = 4.30 A) were present in the alloys after solution annealing at 1180°C. Grain-boundary carbide precipitation after aging 200 hr at 704°C was quite similar to that which occurred in the Alvac series of alloys®*® with carbides in irregular grain boundaries and extending into the matrix 0.5 to 1.0 um. Fine MC carbides precipitated around the primary particles in the matrix. Although most of the primary precipitates were MC carbides, some appeared to be NizTi. NiyTi precipitation during aging was not observed in the 200-hr specimens. We will be examining material aged for longer times in the temperature range of 650 to 760°C. The hafnium-modified heats contained large quan- tities of primary carbides. We expect these carbides to be extremely rich in hafnium, based on the high lattice parameter of these carbides (4.62 A, compared with 464 A for HfC). We have not examined any aged material by transmission electron microscopy. 14.4.3 Modification of X-Ray Diffraction Techniques A considerable effort has been expended in the refinement and modification of x-ray diffraction tech- niques to improve detectability, resolution, accuracy, and ease of data interpretation for phase-analysis studies. We have installed a graphite-crystal-diffracted- beam monochromator on a Norelco diffractometer which greatly improves peak resolution and line-to- background ratios. We are also using silicon single crystals, cut so that no diffraction peaks occur, as substrates for extracted precipitates. As a result, we have decreased the background by 1.5 to 2 orders of magnitude. We are now able to detect weak diffraction peaks not observed previously and can resolve peaks with differences in interplanar spacings down to 0.005 A in the 2.0- to 2.5-A range. This capability is important for detection of carbides with very small differences in lattice parameters. Data are collected on punched tape, using slow goniometer scanning speeds to provide good counting statistics, and computer processed for ease in data handling. A semiquantitative technique for using a 7. J. M. Oblak et al.,, “Heterogeneous Precipitation of Metastable v Ni,Ti in a Nickel-Base Alloy,” Acta Met. 19, 355-63 (1971). 8. R. E. Gehlbach and S. W. Cook, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 164—65. 9. R. E. Gehlbach and S. W. Cook, MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp. 131-38. computer to estimate relative amounts of phases pres- ent in a sample may be modified for use with pre- cipitates encountered in Hastelloy N. 14.5 SALT CORROSION STUDIES J. W. Koger The success of an MSBR is strongly dependent on the compatibility of the container materials with the molten salts used in the primary and secondary circuits of the reactor. Because the products of oxidation of metals by fluoride melts are quite soluble in the corroding media, passivation is precluded, and the corrosion rate depends on other factors, including the thermodynamic driving force of the corrosion re- actions.’® Design of a practicable system utilizing molten fluoride salts, therefore, demands the selection of salt constituents that are not appreciably reduced by available structural metals and alloys whose compo- nents can be in near thermodynamic equilibrium with the salt medium. Nickel-base alloys, more specifically Hastelloy N and its modifications, are considered the most promising for use in molten salts and have received most attention. Of the major constituents of these alloys, chromium is the least noble and forms the most stable fluoride, so that corrosive attack is normally manifested by the selective removal of chromium. Stainless steels, having more chromium than Hastelloy N, are more susceptible to corrosion by fluoride melts but can be considered for some applications. Several different oxidizing reactions may occur, de- pending on the salt composition and impurity content. Among the most important reactants are UF,, FeF,, and HF. Reaction of chromium with the latter two tends to proceed to completion at MSR temperatures; reaction with UF, to form CrF, and UF; soon reaches equilibrium in an isothermal system. The equilibrium constant has a small temperature dependence, however, which means that a mechanism exists for continued attack in nonisothermal systems. Corrosion by molten salt under a temperature gradient involves material removal from hot surfaces and material deposition on cold surfaces, with a steady-state amount of corrosion products remaining in the salt. This phenomenon, called temperature gradient mass transfer, proceeds in the following manner. At the beginning of operation of a temperature-gradient system, the least-noble constit- uent of the container alloy (chromium in our case) 10. W. R. Grimes, “Molten-Salt Reactor Chemistry,” Nucl. Appl. Technol. 8, 137 (1970). 174 will be oxidized at all surtaces and go into solution until the increasing corrosion-product concentration in the salt comes to “equilibrium” with the alloy at the lowest temperature point of the loop. Corrosion at the higher-temperature surfaces continues, however, causing the salt’s corrosion-product concentration to increase. As the concentration begins to exceed the equilibrium concentration at the lower-temperature surfaces, the metal begins to deposit there. The rise in corrosion- product concentration in the circulating salt will con- tinue until the rate at which metal is retumning to the walls at low temperatures balances the rate at which it is entering the salt in the hot-leg regions. Thereafter, there is continuing removal and deposition with no overall change in corrosion-product concentration in the salt. At steady state the rate of corrosion is generally limited by the rate of diffusion of the chromium through the alloy to the hot surface where it is being removed. Corrosion by temperature-gradient mass transfer in- volving selective removal of chromium occurs in all of our nonisothermal systems. Under abnormally oxidizing conditions (as in fluoroborate systems containing water) most constituents of Hastelloy N may be removed. The experiments discussed in this section were con- ducted primarily to determine quantitatively the amounts of corrosion in various salt-alloy systems designed and operated to show the effects of variables such as alloy constituents, temperature, salt impurities, salt velocities, and exposure times. These experiments are operated under design parameters based on the molten salt breeder reactor, primarily in nine thermal- convection loops which provide nonisothermal, dyna- mic conditions. Eight of the loops are constructed of Hastelloy N, one of stainless steel. The salts of interest either are LiF-BeF,-based with UF, (fuel), ThF, (blanket), or ThF, and UF, (fertile-fissile), or are an NaBF4-NaF mixture (coolant salt). Pumped loops containing NaBF,-NaF are covered separately (Sect. 14.6). The status of the thermal-convection loops in opera- tion at the end of this period is summarized in Table 14.6. 14.5.1 Fuel Salt The termination of loop 1255 and a discussion of the failure that occurred due to external corrosion under a ceramic bushing were previously reported.!! Detailed 11. J. W. Koger, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 139-43. 175 Table 14.6. Status of MSR program thermal convection loops through February 29, 1972 Loop . . Salt_ . Max. AT Operating number Loop material Specimens Salt type composition temp. °0) time (mole %) °C) (hr) 1258 Type 304L Type 304 L stainless Fuel LiF-BeF,-ZrF 4-UF4-ThF 4 688 100 74,951 stainless steel steeldD (70-23-5-1-1) NCL-13A Hastelloy N Hastelloy N; Ti-modified Coolant NaBF4-NaF (92-8) plus 607 125 29,163 Hastelloy N controls?:¢ tritium additions NCL-14 Hastelloy N Ti-modified Hastelloy Nb¢ Coolant NaBF 4-NaF (92-8) 607 150 37,738 NCL-15A Hastelloy N Ti-modified Hastelloy N; Blanket LiF-BeF,-ThF4 (73-2-25) 677 55 31,000 Hastelloy N controls?¢ NCL-16 Hastelloy N Ti-modified Hastelloy N; Fuel LiF-BeF,-UF, 704 170 35,352 Hastelloy N controls?-¢ (65.5-34.0-0.5) NCL-17 Hastelloy N Hastelloy N; Ti-modified Coolant NaBF 4-NaF (92-8) plus 607 100 23,400 Hastelloy N controls?-¢ steam additions NCL-19A Hastelloy N Hastelloy N; Ti-modified Fertile- LiF-BeF,-ThF4-UF, 704 170 17,787 Hastelloy N controls?:€ fissile (68-20-11.7-0.3) plus bismuth in molybdenum hot finger NCL-20 Hastelloy N Hastelloy N; Ti-modified Coolant NaBF4-NaF (92-8) 687 250 19,281 Hastelloy N controls?-¢ NCL-21 Hastelloy N Hastelloy No-¢ MSRE fuel LiF-BeF,-ZrF4-UF, 650 110 5,377 (65.4-29.1-5.0-0.5) 9Hot leg only. bRemovable specimens. “Hot and cold legs. analyses of the behavior of standard Hastelloy N, a 2% Nb-modified Hastelloy N, and appropriate welds after 9.2 years (80,439 hr) exposure to LiF-BeF,-ZrF 4 -UF,- ThF, (70-23-5-1-1 mole %) salt and air at temperatures from 560 to 700°C are under way. It is evident from the examinations that temperature gradient mass trans- fer did occur. Attack in the hot section was manifested by formation of voids having a maximum depth of about 4 mils. Deposits less than 1 mil thick were noted in the cold regions. The actual void formation and chromium depletion agreed favorably with those pre- dicted from calculations based on the rate of chromium diffusion. No differences in corrosion were seen for standard Hastelloy N, modified Hastelloy N, and welded areas. Figure 14.20 shows typical attack in each of the above materials. Where exposed to air, the materials formed two-layer oxide having a maximum thickness of about 2 mils. Over all, Hastelloy N is quite suitable for long-term use as a container material for a molten salt of the type used in this test and has acceptable air oxidation resistance at the temperatures tested. Loop 1258, constructed of type 304L stainless steel, has operated 8.5 years with the same salt as loop 1255. The maximum corrosion rate over the last 8395 hr was 1.0 mil/year. The chromium content of the salt is 580 ppm, an increase of about 500 ppm during operation. The loop continues to operate satisfactorily. NCL-21 is a Hastelloy N thermal-convection loop, with removable Hastelloy N specimens in each leg, containing salt of the same composition as the MSRE fuel. As discussed in some detail in Sect. 10.1, this loop is equipped with electrochemical probes to measure the U3*/U* ratio. Such probes had been used successfully in small static systems, but this experiment is designed to evaluate their possible use for on-stream analysis in a large system. The specimens from NCL-21 were removed, weighed, and examined three times. Weight losses and gains, as expected in a temperature-gradient mass transfer sys- tem, were seen in the hot and cold legs respectively. The maximum weight loss in the hot-leg specimens after 4193 hr was 0.23 mg/cmz, which corresponds to a corrosion rate of 0.02 mil/year, assuming uniform 176 Y—109570 Y0.001 in. Q.007 INCHES 0.007 INCHES () \ o s Y —{44366 HES 0.CO7 1t T Fig. 14.20. Portions of loop 1255 exposed to LiF-BeF,-Z1F,4-UF4-ThF, (70-23-5-1-1 mole %) for 9.2 years. (a) Hastelloy N insert specimen — 675°C. Etched with glycena regia, S00X. (b) Hastelloy N—2% Nb insert specimen — 695°C. As polished, 500X. (¢) Hastelloy N—-Hastelloy N—2% Nb weld — 665°C. Etched with glyceria regia, 500X. material removal. The chromium content of the salt has increased 40 ppm. The U3*/U* ratio gradually in- creased with time between specimen insertions. This presumably reflects the corrosion reaction of UF, with chromium, and we are correlating these ratios with our weight-change data. As indicated in Sect. 10.1, on at least two of the three occasions when specimens were inserted, some oxidant appeared to be introduced, as indicated by a decrease in the U>*/U*" ratio. After these initial measurements, further steps will involve addi- tions of oxidants and reductants to study their in- fluence on the measurements of the U*3 /U™ ratio and the corrosion rate. 14.5.2 Fertile-Fissile Salt A fertile-fissile MSBR salt has circulated for over 17,000 hr in Hastelloy N loop NCL-19A, which has removable specimens in each leg. The test has two purposes: (1) to confirm the compatibility of Hastelloy N with the salt and (2) to determine if bismuth will be picked up by the salt and carried through the loop. The bismuth is contained in a molybdenum vessel located in an appendage beneath the hot leg of the loop. Assuming uniform loss the maximum weight loss has been quivalent to only 0.02 mil/year. A modified Hastelloy N alloy (Ni—13.0% Mo—8.5% Cr—0.1% Fe—0.8% Ti— 1.6% Nb) has lost less weight than a standard alloy at the same position. We attribute this difference to the low iron in the modified alloy, compared with about 5% in the standard alloy. Principal corrosion reactions in this loop appear to be 2UF,(d) + Cr(s) 2 2UF3(d) + CrF,(d) and FeF,(d) + Cr(s) Z Fe(s) + CrF, (d), where s and d refer to crystalline solid and dissolved states respectively. The chromium content of the salt has increased 178 ppm in 15,000 hr, and there is no detectable bismuth in the salt. During the last exposure period, the specimens in the hot leg were inadvertently placed 4 in. lower than usual. This put the specimen at the bottom of the stringer into the molten bismuth. On removing the stringer, this bottom specimen was miss- ing. The end of the stringer appeared to have been in contact with the bismuth but was not dissolved. Thus it is likely that the very small specimen pins dissolved during the run, and the specimen fell off the stringer. The bismuth in contact with the salt seems to have had no effect on our mass-transfer results. 177 14.5.3 Blanket Salt Loop NCL-15-A, constructed of standard Hastelloy N and containing removable specimens in each leg, has operated over three years with the LiF-BeF,-ThF, blanket salt proposed for a two-fluid MSBR. Mass transfer, as measured by the change of chromium concentration in the salt, has been very small. The previously reported “glaze” or coating (probably a high-melting thorium compound) on specimen surfaces has now disappeared, and significant weight changes are measurable. The maximum corrosion rate of 0.06 mil/year (assuming uniform removal) is quite acceptable for an MSBR. 14.5.4 Coolant Salt Loops NCL-13A and NCL-14, constructed of stan- dard Hastelloy N and containing removable specimens in each leg, have operated for 3.3 and 4.3 years, respectively, with the fluoroborate mixture NaBF,-NaF (92-8 mole %). The maximum corrosion rate (assuming uniform removal) at the highest temperature, 605°C, has averaged 0.7 mil/year for both loops. Corrosion has generally been selective toward chromium by the reaction FeF,(d) + Cr(s) < CrF,(d) + Fe(s). However, there have been short periods when gaseous impurities entered the salt to cause general attack of the Hastelloy N, for example, by the reaction 2HF(d) + M(s) Z H, (g) + MF,(d), where d, s, and g refer to dissolved state, crystalline solid, and gas, respectively, and M may be Cr, Fe, Ni, or Mo. These periods resulted from leaks in the gas lines or in ball-valve seals. The valves are exposed to a mixture of helium and boron trifluoride gas and not to salt. We attribute the leaks of the ball valves to corrosion induced by air and moisture inleakage from lines into the gas mixture. Although the overall corrosion rates in these loops are not excessive, the rates observed in the absence of leaks have been an order of magnitude lower than the average rate. In NCL-13A the maximum corrosion rate has been 0.3 mil/year over the last 3170 hr. Loop NCL-17, constructed of standard Hastelloy N and containing removable specimens in each leg, is being used to determine the effect of steam on the mass-transfer characteristics of the fluoroborate salt mixture. After 1000 hr of normal operation, steam was 178 injected into the salt.!? The loop has now operated 22,400 hr following that injection. There was a large increase in weight change during the first 239 hr after steam injection, and another abrupt change in the rate of weight change occurred at 10,178 hr because of a leak in the cover gas system; however, the overall mass-transfer rate decreased steadily between these two events. Specimens were recently removed after 7000 hr of continuous exposure, and the average corrosion rate at the maximum temperature, assuming uniform dissolu- tion, was 0.7 mil/year. The rate is equal to that seen in the loop before a leak in the cover gas system that was noted after 10,178 hr exposure. Once again the corrosion rate does appear to be decreasing. Loop NCL-20, constructed of standard Hastelloy N and containing removable specimens in each leg, has operated for over 14900 hr with the fluoroborate coolant salt at the extreme temperature conditions considered for the MSBR secondary circuit (687°C max and 438°C min ). Forced-air cooling of the cold leg is 12. J. W. Koger, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 170. required to obtain this A7. For the first 11,900 hr operation, the maximum corrosion rate was 0.2 mil/year. Then a regulator failure allowed some mois- ture leakage into the loop and increased the maximum corrosion rate to 0.7 mil/year. During the last 4797 hr, this rate (assuming uniform loss) was 0.4 mil/year and decreasing. An experiment designed to evaluate the corrosion properties of eight brazes for Hastelloy N in NaBF,- NaF (92-8 mole %) at 607°C for 4776 hr was conducted. The test was isothermal and utilized samples of the lap-joint geometry. The brazes included Au-Ni, Ag-Cu, Cu, and several types of Ni-Cr-Fe alloys. Over the whole test, all specimens gained weight. Figure 14.21 shows specimens before and after the test. The difference in appearance is attributable to deposits on the surface of the Hastelloy N. However, during two time periods, some weight losses were evidenced, indicating highly corrosive conditions in the melt. On the basis of these weight losses and metallographic observations, we tentatively ranked the brazes accord- ing to corrosion resistance, and the Cu and Ag-Cu appear to be the best. To further aid us in our evaluation, we are using the microprobe to determine Y-4109689A BOTTOM Fig. 14.21. Hastelloy N specimens brazed with Ni-P-Cr alloy. Top not tested, bottom exposed to NaBF4-NaF (92-8 mole %) for 4776 hr at 607°C. compositional changes in the braze caused by exposure to the salt. 14.6 FORCED-CONVECTION LOOP CORROSION STUDIES W. R. Huntley J. W.Koger Operation of forced-convection loop MSR-FCL-1A continued during the past six months, providing infor- mation on the compatibility of standard Hastelloy N with NaBF,-NaF (92-8 mole %) coolant salt at tempera- tures similar to those expected in the MSBR secondary circuit. Maximum and minimum salt temperatures in the loop are 620 and 454°C (1150 and 850°F). Hastelloy N corrosion test specimens are exposed to the circulating salt at 620, 548, and 454°C and a nominal velocity of 5% fps. In this loop, the tubing must be cut to remove the specimens, and welding is required for specimen replacement. A second forced-circulation loop of improved design, MSR-FCL-2, also operated throughout this report period. This loop is also of standard Hastelloy N and circulates sodium fluoroborate at 454 to 620°C. Veloci- ties are much higher, however, nominally 10 and 20 fps in the test sections. Three sets of corrosion specimens are exposed to salt at 620, 537, and 454°C. The FCL-2 design permits these specimens to be easily removed and inserted, with a minimum of contamination of the loop. 14.6.1 Operation of Loop MSR-FCL-1A Loop MSR-FCL-1A, which started in August 1971, operated throughout this report period except for a six-week scheduled shutdown for examination of corro- sion specimens after 2000 hr of loop operation. Normal loop operation was resumed and by the end of the period had reached a total of 3700 hr at design conditions. Following established practice, during the shutdown the LFB pump bearings and oil seals were replaced. During cold shakedown, several seal failures occurred. Because of these problems the specified flatness of the seals was increased, a new pump assembly procedure was prepared incorporating additional inspection and cleaning steps, and other changes were made to improve the reliability of operation. 14.6.2 Results from Loop MSR-FCL-1A When the specimens from the three points in the loop were weighed after about 2000 hr of loop operation, all 179 were found to have lost weight. Amounts ranged from —2 mgf/cm? for the lowest-temperature (454°C) speci- mens to —19 mg/em? for those at the highest tempera- ture (620°C). The corrosion rate at the highest tempera- ture was approximately 4 mils/year. This contrasts with the pattern of moderate temperature-gradient mass transfer (the FCL-2 results, for example), in which weight losses occur in the hot sections accompanied by weight gains in the cooler sections and increases in corrosion products in the salt. Analyses of salt samples taken from the loop during the interval indicate little overall change in nickel and molybdenum and an increase of about 150 ppm in chromium, with a decrease in the concentration of iron in the salt. The salt analyses indicate that there was at least a small amount of rapid, general corrosion at the very beginning of loop operation. Results of a sample taken 90 hr after the new loop was first filled (during which time the salt circulated isothermally at 540°C) indi- cated that chromium, iron, and molybdenum concen- trations in the salt had all increased sharply. Chromium had jumped from 36 to 275 ppm; iron had gone from 350 to 472 ppm; molybdenum, which is presumed to have been around 10 ppm or less in the salt at the beginning, was now 175 ppm. Nickel in the salt had not changed appreciably. Analyses of subsequent salt samples for corrosion products indicate that during 1984 hr of operation at design conditions prior to the removal of the specimens, chromium came down and leveled off at around 200 ppm, iron decreased (rapidly at first then more slowly) to about 25 ppm, and molybdenum quickly fell to less than 10 ppm. Oxygen analyses indicated an upward trend, from about 500 ppm to approximately 1000 ppm. The concentration of hydrogen as BF;OH™ was indicated by infrared analyses to be 63 ppm in the sample at the beginning of design operation and 26 to 30 ppm in the four samples taken between 222 and 1984 hr. It appears that when the salt was first circulated in the new loop, there was rapid, general attack which removed not only chromium but also iron and moly- bdenum (and probably nickel) from all portions of the loop. Subsequent mass transfer removed more material from hot surfaces and deposited material on cold surfaces, producing the spread from —2 to —19 mg/cm? observed on the metal specimens. Salt sample analyses during operation at design conditions after the specimens were removed and replaced gave no indication of abnormally corrosive conditions. 14.6.3 Operation of Loop MSR-FCL-2 Corrosion loop MSR-FCL-2 began routine operation on September 1, 1971, and during this report period accumulated about 3900 hr at design conditions. The salt has not been drained from the piping system since the initial fill. No serious operational problems were encountered, but six unscheduled, automatic shut- downs occurred during the six months. The automated protective circuits functioned as planned on each occasion to stop the salt pump and cooler blowers and to place the loop in an isothermal “standby” condition. This safety action was necessitated twice by a low flow alarm on the lubrication oil, once by a temporary electric power outage, once by a brush failure on the Adjustospede motor that drives the salt pump, once by a brush wire failure on the pump motor, and once by a failure of a vacuum tube in the pump speed control. After each of these incidents, normal operation was resumed following a few hours delay. , On several occasions, severe electrical noise appeared on some of the sheathed insulated-junction loop ther- mocouples. The cause was found to be arcing of the slip rings on the clutch of the pump motor and was eliminated by honing the slip rings. The design of the corrosion specimen removal sta- tions, which permits rapid removal of each of the three stringers via a salt freeze valve, proved to be very effective during the five scheduled shutdowns in which the specimens were removed. Downtime for specimen removal and examination was only two days, compared with several weeks and much more manpower in MSR-FCL-1A. The performance of the ALPHA salt pump in MSR-FCL-2 and a description of proposed engineering improvements to the pump are described in Sect. 3.5.2 of this report. 14.6.4 Results from Loop MSR-FCL-2 The specimens were weighed after 450, 914, 1780, 2774, and 3846 hr of routine operation at design conditions. Cumulative weight changes subsequent to the beginning of design operation are shown in Fig. 14.22. (Changes during 500 hr of startup testing,'? which ranged from +0.9 to —1.5 mg/cm?, are not included.) Each point is the average for more than one specimen exposed to a particular combination of temperature and velocity. These results indicate 13. W. R. Huntley et al., MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 151-53. 180 ORNL-DWG 72—52486 20 [ VELOCITY & o 20.8 fps £ e 10.9 fps 210 o £ o] ~ ° °}as4 oc b=4 8 3 e _ g Oe 537 °C S o T — .\. ° . . z — “}620°c o— 10 = -20 0 1000 2000 3000 4000 5000 6000 TIME (hr) Fig. 14.22. Weight changes of Hastelloy N specimens exposed to NaBF4-NaF (92-8 mole %) in FCL-2 as a function of time, temperature, and velocity. temperature-gradient mass transfer at a steadily de- creasing rate, with a significant velocity effect. Speci- mens in the hottest position have remained fairly bright, with the specimens in the middle temperature position darker and the specimens in the coldest position darkest. Differences in appearance such as these have been previously observed.'* Corrosion-product concentrations in the salt showed no anomalous behavior, either during the startup or in subsequent operation. As shown in Fig. 14.23 the chromium concentration increased by about 60 ppm during the startup period, rose another 70 ppm or so during the first few hundred hours of routine operation, and thereafter changed very little. The iron concentra- tion decreased. Molybdenum and nickel remained low. In sets of specimens exposed at the same temperature to the same superficial salt velocity, the upstream specimen generally showed the greatest weight change, the second specimen in line showed less, and the downstream specimen showed the least. The differences (on the order of 10% between the first and last specimen) were clear early in the operation, when weight changes were relatively large, but become practically indistinguishable later, when all weight changes approached the limits of precision of measure- ment. Changes of this nature are quite common in liquid metal systems (“downstream” effect) and are due to a concentration gradient driving-force effect between the wall and the fluid. Turbulence may also be responsible for the changes. 14. J. W. Koger, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 175 -76. 181 ORNL-DWG 70-4932RBR 10 T I 7 9 I - - 8 | _ /% —— ~ 7 — 5 2%, Cr STEEL—538V L g 6 i . .'2 — o ! mpy_~- S 5 - P4 / ’,1 =t A 5 ° - e 4 7 —~ N, AIR —] T ¢ L MELTED 593°C | , | .a % 3 h -~ 0.5 mpy L o 3 —1 4 > el ; 2 i _— ‘? N, VACUUM ~ ) [— MELTED 593°C ' == e = 0.25 mpy | et A ——t | fi":{:fl-—-"fl'“"‘b 538°C, N, AIR AND W~ =<2 VACUUM MELTED 0 2000 4000 6000 8000 10,000 12,000 14,000 16,000 TIME (hr) Fig. 14.23. Results of FCL-2 salt analyses. Table 14.7. Weight losses of high-temperature (620°C) samples in NaBF4-NaF in loop MSR-FCL-2 and equivalent corrosion rates Operating time (hr) Weight loss (mg/cmz) Equivalent corrosion Period end Interval At 10.9 fps At 20.8 fps rate? (mils/year) At 109 fps At 20.8 fps 450 450 3.13 4.40 2.7 38 914 464 1.40 1.76 1.17 146 1780 866 1.11 1.53 0.49 0.68 2774 994 0.28 0.58 0.11 0.23 3846 1072 0.29 0.38 0.10 0.14 3846 3846 6.21 8.65 0.62 0.87 9 Assuming uniform material removal of all constituents. The dependence of the rates of weight change on time and velocity that is evident in Fig. 14.22 is revealed more sharply in Table 14.7, which shows weight losses of the high-temperature specimens from one measure- ment to the next. For the sake of comparison with other corrosion data, the weight losses are also ex- pressed as an equivalent corrosion rate during the interval. For specimens having a weight loss, the effect of the velocity was to the 0.5 power. The corrosion rates of the hottest specimens are approaching the same values, which is an indication that the controlling mechanism of corrosion has now changed from a solution rate control to the velocity-independent solid- state chromium diffusion control. Experience indicates that corrosion in fluoroborate systems is due largely to impurities, particularly water. Reactions that are believed to be involved include the following. [Although chromium is the most readily oxidized constituent of Hastelloy N, reactions similar to (3) involving other constituents can be important if the concentration of HF is high. | H,0 + NaBF, 2 NaBF;OH + HF, (1) NaBF;0H > NaBF, O + HF, (2) 6HF + 6NaF + 2Cr 2 2Na; CrFg + 3H,. (3) These may be combined to give 6NaBF;0H + 6NaF + 2Cr 2 6NaBF, 0 + 2Na;CrF4 + 3H, (4) and 3H20 + 3N3BF4 +.6N3F +2Cr P BNHBon + 2Na3CTF6 + 3H2 . (5) Thus both the NaBF;OH initially in the salt and any H, O that may be admitted later result in corrosion and the appearance of hydrogen, which can diffuse through the hot metal walls and be lost from the system. Removal of hydrogen from the system by diffusion through the metal would drive reaction (4) and (5) to the right. The results’® of the tritium addition to NCL-13A supported the contention that little if any hydrogen-containing impurities existed in the molten fluoroborate. Because of the connection with corrosion 15. J. W. Koger, MSR Program Semijannu. Progr. Rep. Feb. 28, 1971, ORNL4676, pp. 210-11. 182 (and, as discussed below, with tritium transport), samples of salt from FCL-2 were routinely analyzed for oxygen and for BF3OH™. (The infrared-absorption technique for the latter analysis is discussed in Sect. 7.6 and 10.7 of this report.) The oxygen results shown in Fig. 14.23 do not reveal any trend with time. The measured concentrations of hydrogen as BF;O0H™, on the other hand, clearly indicate a downward trend for about 2800 hr, then little change over the next 1000 hr. The hydrogen behavior in the fluoroborate salt in FCL-2 is of great interest because of its implications for tritium transport in an MSBR. The concentrations shown in Fig. 14.23 indicate a much less rapid loss of hydrogen than would be expected on the basis of the interrelations of BF3;OH™ concentration, hydrogen pressure, and diffusion through Hastelloy N observed in other experiments (Chap. 7 of this report). 14.7 CORROSION OF HASTELLOY N IN STEAM B. McNabb H. E. McCoy The unstressed Hastelloy N specimens exposed to steam in the Bull Run facility at 538°C continued to ORNL-DWG 72-7740 OXYGEN (ppm) OXYGEN | l N CORROSION PRODUCTS °© Cr 7 e Fe € 300 — {\ T T a g , = , 2 200 7 g /// o ° o ° o 100 |— - . 7 7 o ] o P 1 / e £ 40 y T ] g HYDROGEN AS BF0H™ & 20 j | S . T r . ° P (o) >— T o 800 —— STARTUP 160G OPERATING TIME (hr) 2400 Fig. 14.24. Weight changes of Hastelloy N and 21/4% Cr steel after exposure to steam in the unstressed condition. show very low generalized corrosion rates of less than 0.25 mil/year, assuming uniform corrosion. The actual weight gains for standard Hastelloy N, both air melted and vacuum melted, were about 0.43 mg/cm? for 11,000 hr exposure to steam at 538°C and about 3.45 mg/cm? for 15,312 hr exposure to steam at 593°C. Figure 14.24 compares the weight changes of Hastelloy N with those of the 2Y,% Cr steel that is commonly used in modern power plants. There was no evidence of spalling on any of the specimens. Although the results are not shown in Fig. 14.24, several modified Hastelloy N specimens are also in- cluded in the facility at 538°C. The weight changes were slightly higher than those for the standard alloy, but none of the modified specimens had weight gains greater than 0.7 mg/cm? after 11,000 hr exposure to steam at 538°C. Some of the specimens were removed for detailed examination after 10,000 hr exposure. Table 14.8 shows the compositions of these alloys. Metallographic examination showed standard air-melted Hastelloy N (heat 5065) to have some small nodules of oxides. Figure 14.25 shows one of the deepest nodules, having a penetration of about 0.4 mil. The oxide is generally 183 very thin and adherent, with occasional patches of oxide penetration. The vacuum-melted heat of standard Hastelloy N (heat 2477) also had a very thin, adherent oxide, but the oxide nodules were more frequent and deeper (0.8 mil) than for the air-melted heat 5065. Figure 14.26 is a photomicrograph of one of the worst areas in the vacuum-melted alloy and shows the thin, adherent oxide in some areas and the maximum penetrations in other areas. The lower manganese and silicon (~0.05 wt % each) in the vacuum-melted heat probably account for its slightly lower corrosion resistance. One of the modified Hastelloy N heats, 21546, with low Fe, Mn, and Si (Table 14.8) had weight gains of about 0.44 mg/cm®. However, the grain boundary penetrations were about 10 mils deep in some areas. Figure 14.27 shows one area with a thin, adherent oxide at the surface and one of the deep grain boundary penetrations. Microprobe analysis indicates the outer surface of the grain boundary penetration to be slightly enriched in iron. The grain boundary penetrations appear to be slightly enriched in chromium, slightly depleted in nickel, and little different in the mo- lybdenum concentration. Y-144522 T 500X 10.003 in. 0.007 INCHES —F {5.005 . Fig. 14.25. Photomicrograph of air-melted Hastelloy N (heat 5065) after exposure to steam for 10,000 hr at 538°C. As polished. 184 Y-111528 10.001 in. '0.003 in. 0.007 INCHES o005 ™. 500X f I1C.007 in. Fiy. 14.26. Photomicrograph of vacuum-melted Hastelloy N (heat 2477) after exposure to steam for 10,000 hr at 538°C. As polished. A specimen of a modified alloy containing 2.1% titanium (heat 70-727) was removed after 7330 hr at 538°C. Figure 14.28 shows the thin, adherent oxide with metallic-appearing particles above the surface. Microprobe analysis indicated that these were enriched in iron, and, as previously reported, it appears that iron oxide particles are entrained in the steam’ ® and stick to the surface of specimens. The 2.1% titanium content of this heat appears to stabilize the grain boundaries and prevent the penetrations observed in heat 21546. The heat 5065 specimen gained 0.33 mg/cm?®, heat 2466 gained 0.26 mg/cm?, and heat 21546 gained 0.44 mg/cm?, all in 10,000 hr at 538°C. In 7330 hr at 538°C, heat 70-727 gained 0.37 mg/cm*. The corrosion was uniform in some specimens, and the weight changes are indicative of the depth of attack. In other specimens the corrosion was not uniform, and the weight changes do not indicate the depth of attack. 16. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 178-81. There appears to be an effect of surface preparation on the weight gain of standard Hastelloy N in steam. Figure 14.29 is a plot of weight gain time for heat 5065 specimens with several surface preparations. All material was rolled to 0.035-in. sheet, with intermediate anneals at 871°C and a final anneal of 1 hr at 1180°C. Specimens 9 and 10 were exposed in this condition, specimens 11 and 12 were surface abraded with 400-grit paper, and specimens 13 and 14 were electropolished. The surface-abraded specimens have the highest weight gains, as-received (rolled) specimens lower, and the electropolished specimens the lowest weight gains. There are several possible reasons for the observed effect. One is simply the differences in actual surface areas. The abraded specimens would have the greatest surface area, the as-rolled specimens next, and the electropolished specimens the least surface area. The surface area was calculated from micrometer measure- ments and would not represent the true surface areas. Greater surface roughness would also give more sites for particles of oxides entrained in the steam to stick, with 185 Y-111542 L o lo.octin, | st pati 0 0.007 INCHES [0.005 in. 500X — 10.007 in. Fig. 14.27. Photomicrograph of modified Hastelloy N (heat 21546) after exposure to steam for 10,000 hr at 538°C. As polished Y-111463 | Tocot e HES 10,003 in, ~ 0007 IN to.008 500X Fig. 14.28. Photomicrograph of modified Hastelloy N (heat 70-727) after exposure to steam for 7330 hr at 538°C. As polished. Table 14.8. Compositions of several heats of standard and modified Hastelloy N Alloy Concentration (%) No. Mo Cr Fe Mn C Si Cu Co Y W Al Ti B Cb Hf Zr Mg 5065 160 7.1 4.0 0.55 0.06 0.57 001 007 023 01 <003 <001 0.001 <0.05 <0.1 <0.1 0.02 2477 160 69 4.1 0.055 0.057 0047 001 005 <001 003 003 002 0.0002 <0.0005 <0.001 <0.001 <0.005 21546 123 7.3 0.046 0.16 0.05 0009 001 <0.10 <0.10 <0.10 002 0.10 0.0002 0.005 70727 130 7.4 0.05 0.37 0.044 <0.05 <0.01 <0.01 <0.01 <0.01 <0.03 2.1 0.00006 <0.01 <0.01 <0.001 0.015 981 187 05 ORNL-DWG 72- 7680 a greater sticking probability on the rougher surfaces. ' T This increased surface area would also provide added SURFACE TREATMENT sites for oxide nucleation on the metal. These speci- -~ 0 AS ROLLED - . - . o . o8 & AS ROLLED mens are continuing in the facility to see if these trends v figggggg continue to larger weight gains. S — a4 ELECTROPOLISHED In addition to the unstressed specimens described o ELECTROPQLISHED above, the Bull Run facility also contains stressed tubular specimens. Stresses are maintained by having plant steam inside the tubes and a low-pressure annulus ,) around them. When first installed, this facility included four tubes with capillary connections to sense leakage into the annuli.! 7 During this report period, in order to increase the rate of accumulation, four additional tube-burst specimens were installed in the facility without the capillary tubing extending outside the ; chamber to indicate rupture. Internal diametral strains ) will be measured periodically, and rupture times will be ?‘/ estimated by comparison with the instrumented tests. 06 |- = WEIGHT CHANGE (mg/cm?) The specimen design is identical to the other double- wall tube-burst specimens, except that the annulus between the tubes is not connected to the outside of © 2000 4DD$IME (hjooo Be00 %99 the chamber by capillary tubing for indication of ) ) ) _ o failure. Fig. 14.29. Comparative weight changes in steam at 538 C of Hastelloy N (heat 5065) having various surface treatments. All samples rolied to 0.035 in. and annealed 1 hr at 1180°C prior to 17. B. McNabb and H. E. McCoy, MSR Program Semiannu. receiving different surface treatments. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 216. Y—-111910 INCHES Fig. 14.30. Photograph of tube-burst sample that was stressed at 58,000 psi in steam at 538°C and failed in <1000 hr. The smaller tube was initially pressurized, failed, and pressurized the annular region, and the smaller tube was collapsed when the plant steam pressure decreased rapidly. 188 i 0.035 INCHES " 100X 1 1 0.035 INCha3 I 100x = 0.035 INCHES M 100X Fig. 14.31. Photomicrographs of the inner tube shown in Fig. 14.30. Intergranular fractures occurred on both sides of the tube because of the unusual loading sequence. (¢) Cracks from the OD, (b) cracks from the ID, (¢) cracks from the ID. Etched with glyceria regia. Reduced 33%. Two tube-burst specimens of heat N15095 have accumulated 5000 hr exposure to steam at 40,250 psi and 28,000 psi and have diametral strains of 0.71 and 0.19% respectively. Two tube-burst specimens of heat N25101 have 4000 hr exposure at 56,000 and 50,000 psi with diametral strains of 0.57 and 0.33% respec- tively. Heat N25101 thus appears stronger than N25095 under these creep conditions, although the certified tensile properties were equivalent. The four new speci- mens were heat N15095. The specimens with the highest stress (58,000 psi) failed sometime during the first 1000 hr of exposure. A small crack developed in the inner tube and pressurized the annulus between the tubes. When the plant steam pressure was reduced, the pressure in the annular region collapsed the tube, as shown in Fig. 14.30. The specimen had a hairline crack extending almost the entire length of the reduced section of the tube wall. Numerous cracks formed on the OD and the ID. Plant records at the Bull Run plant have been examined in an effort to determine the time of failure more closely. Failure could have occurred on February 20, when a pressure excursion to 3750 psig momen- tarily occurred and, subsequently, the pressure dropped to zero in about 30 min. This 3750 psig would have been 62,000 psi on the specimen. This would have meant that the failure occurred after 792 hr. As shown in Fig. 14.31 numerous intergranular cracks were formed in the failed sample. The other three new specimens after 1000 hr of exposure have diametral strains of 0.25% at 53,000 psi, 0.24% at 42,502 psi, and 0.14% at 42,400 psi. 189 14.8 EVALUATION OF DUPLEX TUBING FOR USE IN STEAM GENERATORS B.McNabb H. E.McCoy One type of duplex tubing proposed for steam generators is nickel 280 (Ni + 0.05% Al) on the outside for salt corrosion resistance and Incoloy 800 (Fe—34% Ni—21% Cr) on the inside for resistance to steam corrosion. Some mechanical property tests on a 10-ft length supplied by the International Nickel Company were reported previously.!® Tube-burst tests at 538°C, in an argon atmosphere with argon internal pressure, continued during this report period. One specimen, with a pressure corresponding to a stress of 46,000 psi on the Incoloy 800 (or 28,720 on the entire wall), ruptured at 3263 hr with a diametral strain of 3.04%. The nickel 280 exterior was cracked profusely. A tube-burst specimen was stressed at 40,000 psi on the Incoloy 800 and was discontinued at 7075 hr with a diametral strain of 1.14%. Figure 14.32 is a photograph of this specimen with dye penetrant applied to delin- eate the cracks in the nickel 280 exterior. It did not crack as extensively as the higher-stressed specimen, but it was discontinued before rupture or equivalent dia- metral strains were reached. Figure 14.33 shows photo- micrographs of a typical cross section and shows the extensive cracking in the nickel 280. It is evident that cracking does occur in this nickel 280 at relatively low strains. 18. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 156—-62. Y—1411626 INCHES Fig. 14.32. Photograph of duplex tube-burst specimen discontinued after being stressed for 7075 hr at 40,000 psi at 538°C. The tube had a diametral strain of 1.14%, and the dye penetrant clearly reveals cracks in the outer nickel 280 layer, 190 LT ———— e T Y -112660 0.035 INCHES ™ 100X T (a) A Y-112659 0.035 INCHES ™ 100X o Fig. 14.33. Photomicrograph of the cross section of the specimen shown in Fig. 14.32. The nickel 280 is on the outside, and numerous cracks are present. The Incoloy 800 is on the inside and is not cracked. As polished. ORNL DWG ?2 7681 40 ‘ /?_5 000 psi ‘ | | ‘ ST e i | A | 20 } /’\/ } T 15000/ | )/ — 200 |7 2000 A T | A b | 15,000 ‘7 15 — | olls B : — | - S0 R S S N A S B O 200 400 600 800 1000 1200 1400 1600 1800 2000 TIME (br) Fig. 14.34. Creep curves of nickel 280 sheet at 538°C in argon. 191 The base properties of nickel 280 sheet look quite good. This test material was 0.06 in. thick and had been heavily worked. Figure 14.34 shows several creep curves for nickel 280 sheet at 538°C in argon. There appears to be no clear-cut effect of rolling direction on creep properties, as the longitudinal specimens had longer rupture times at the two higher stresses and the transverse specimen had a longer rupture life at 15,000 psi. All of the sheet specimens tested had rupture strains greater than 20%, indicating much greater ductility than that shown by the nickel 280 on the duplex tubing. Thus it appears likely that the duplex tubing can be produced with a nickel layer that has good ductility. The International Nickel Company is preparing another length of Incoloy 800—nickel 280 duplex tubing for further testing and evaluation. 15. Support for Chemical Processing J. R. DiStefano A reductive-extraction process for protactinium isola- tion and a metal transfer process for rare-earth removal are being developed for molten-salt breeder reactors, and this chapter deals with the materials development in support of these chemical processes. A principal requirement of materials for this application is compati- bility with molten Bi-Li-Th solutions at 500 to 700°C. Molybdenum appears quite promising, and currently our efforts are concentrated on the fabrication of a reductive-extraction test stand of molybdenum. Al- though compatible with bismuth-lithium solutions and molten-salt mixtures, molybdenum is difficult to weld and, therefore, to fabricate into complex equipment. This test stand, probably the most difficult all- molybdenum system ever attempted, has involved the development of many fabrication procedures not pre- viously used on this material, such as back extrusion, orbiting arc welding, and roll bonding. With this unit we will obtain metallurgical data on the test equipment and chemical engineering process data under a variety of conditions. In addition, we are continuing our program to evaluate the compatibility of molybdenum, TZM, tanta- lum, T-111, Ta—10% W, brazing alloys, and various grades of graphite with bismuth-lithium solutions under reprocessing conditions. 15.1 CONSTRUCTION OF A MOLYBDENUM REDUCTIVE-EXTRACTION TEST STAND J. R. DiStefano A. J. Moorhead We are constructing a molybdenum test stand for metallurgical and chemical engineering evaluation. It consists of a 5-ft-long, 1%-in.-OD packed column and associated pots and piping through which molten salt and bismuth will circulate in the temperature range 550 to 650°C. Details of the design of this test stand have been reported previously.'** Construction involves (1) fabrication of containers and column, (2) machining of 192 subassembly components, (3) construction of an un- joined mockup from test stand components, (4) fabrica- tion of head pot and column subassemblies, and (5) interconnection of the subassemblies to complete the test stand. [tems 1 and 2 above have been completed. All sizes of tubing (Y4, %, %, and 7% in. OD) have been obtained and inspected by fluorescent-dye-penetrant and ultra- sonic techniques. Using the full-size wood-aluminum- steel test-stand mockup (Fig. 15.1) as a guide, we have selected the specific tubes to be used in fabrication. Previously, we reported brittle behavior of the Y- and %-in.-OD tubing at room temperature.? We traced this behavior to the presence of a brittle layer on the inner surface and found that the tubing could be made ductile by removal of 0.001 to 0.002 in. from the wall. To accomplish this, the bore of the Y-in.-OD tubing was etched at ORNL. However, more %-in.-diam tubing was needed than was on hand. Therefore, we returned 50 ft of three different heats of %-in.-diam tubing to the vendor and purchased 100 ft of a single heat produced according to specifications that ensured the removal of any brittle inner surface layers. To evaluate this new material, ring-shaped samples were squashed at room temperature at a constant displacement rate of 0.005 in./min. Total displacement (original ring dia- meter minus minor axis of elliptical shaped specimen after testing) was approximately 0.11 in., and under these conditions no cracks developed in any of the samples. Similar tests were then conducted on samples from the Y%-, %-, and 7%-in.-OD tubes, and no cracking occurred. 1. E. L. Nicholson, Conceptual Design and Development Program for the Molybdenum Reductive Extraction Equipment Test Stand, ORNL-CF-71-7-2 (July 1, 1971). 2. J. R. DiStefano, “Construction of a Molybdenum Reductive-Extraction Test Stand,” MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 163-69. Fig. 15.1. Mockup of molybdenum reductive-extraction test stand. €61 To gain practice in subassembly fabrication, we are making a prototype of a head pot using molybdenum components similar to those that will be used in construction of the test stand. This prototype involves all of the operations required to fabricate an actual component: (1) electron-beam tube-to-header welds, (2) roll bonding %-, %-, and %-in-OD tubes, (3) tungsten coating the inside of each roll-bonded joint by chemical vapor deposition, (4) electron-beam welding two half sections together, (5) back brazing the outside of the tube-to-header electron-beam-welded joints and the roll-bonded joint, and (6) brazing a ring around the 3%-in.-OD girth weld joining the two half sections. More detailed information on the design and purpose of the molybdenum test stand is presented in Part 4 of this report. Progress on welding, brazing, and fabrica- tion of molybdenum components is reported in this chapter. 15.2 FABRICATION DEVELOPMENT OF MOLYBDENUM COMPONENTS R. E. McDonald A.C. Schaffhauser We have completed fabrication of the 37%-in.-OD closed-end molybdenum back extrusions required for the head pots and disengaging sections of the molyb- denum test stand. The back extrusion process and tooling we developed to fabricate components having an internal length up to 11 in. have been described previously.®> The feed pots, lower disengaging vessel, 3. A. C. Schaffhauser and R. E. McDonald, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 166-69. 194 and upper disengaging vessel each require two back extrusions with internal lengths of 3.5, 7.75, and 9 in., respectively, which are joined by girth welding (see Fig. 15.2). The results of the back extrusions are given in Table 15.1. All of the required back extrusions have been fabricated and machined to final dimensions. However, the spare back extrusions for the disengaging vessels contained cracks in the wall that make them unusable unless they can be repaired by welding. We are presently back extruding additional spare parts for these vessels. ORNL- DWG 72- 7665 Y32 in. DIAM PIN S— 0.096 DIMENSIONS IN INCHES Fig. 15.2. Molybdenum half sections joined by electron-beam girth weld. Pins through the step joint (arrow) held the halves in contact for welding. Table 15.1. Molybdenum back extrusions fabricated for molybdenum test stand 3 Extrusion Internal Part E::;;:;n temperature length Results ‘0 (in.) Feed Pot 1197 1600 4 No cracks Feed pot 1250 1600 4 End cracks, 3.5 in. usable Feed pot 1257 1600 45 End cracks, 3.5 in. usable Feed pot 1259 1600 4.5 End cracks, 4 in. usable Spare 1251 1600 4 End cracks, 3.5 in. usable, surface cracks on top Spare 1256 1600 4 Surface cracks on top Lower disengaging 1258 1600 8 No cracks Lower disengaging 1260 1600 8 No cracks Spare 1261 1600 8.5 Cracksin wall Upper disengaging 1286 1700 9 No cracks Upper disengaging 1290 1700 11 End cracks, 9.5 in. usable Spare 1288 1700 9 Reextruded, crack in wall 15.3 WELDING OF MOLYBDENUM A.J.Moorhead During this report period we have continued our efforts to develop reliable procedures for joining com- ponents of the molybdenum test stand. The full-scale mockup shown in Fig. 15.1 was constructed of alumi- num, stainless steel, and wood to aid in design of the loop. (Proposed changes were incorporated in the mockup as they were suggested in order to ensure that they were compatible with the rest of the system.) After the design of the test stand was finalized, we began using the mockup to make certain that our welding and brazing fixtures will fit in the allotted spaces. This type of check is necessary since the distances between lines and other components have been held to a minimum. Additionally, the mockup has been invaluable in determining the step-by-step se- quence that will be used in construction of the test stand. Major progress was made during this period in three areas that are crucial for fabrication of the loop: (1) girth joint welding, (2) tube-tube welding, and (3) vent-tube welding. A decision was made to use electron beam welding to join the half sections to form the four pots required for the test stand. This process has the advantage of allowing a self-aligning step type of joint to be used and minimizes distortion and abnormal grain growth. One disadvantage of this process is that the joint fit-up must be very tight. In some of our earlier developmental welds, the two half sections were held together by a spring-loaded rod passing axially through the center. This approach is not possible in some of the test-stand vessels because of their internal configura- tion. Therefore, we made a prototype part with the halves held together by three Y4, -in.-diam molybdenum pins passing radially through the step joint. This technique proved successful for making the girth weld. Fluorescent dye-penetrant inspection revealed no de- fects in this weld, which is shown in Fig. 15.2. The technique was repeated using potheads more nearly similar to actual loop components, and it again proved successful. A commercially obtained (Rytek Corporation) orbiting-arc weld head is being used to develop proce- dures for joining the %- and 1Y%-in.-diam tubes for the test stand. Although this head is heavier than the Astroarc head and is water cooled, the 125 A required to fuse this relatively heavy wall molybdenum tubing is at the upper limits of its capability. Because of this limitation, we have reduced the wall thickness of both the 7%- and 1%-in.-OD tubing to 0.050 in. We feel there 195 is considerable advantage in using this tool compared with manual welding, because it eliminates having to rotate large sections of the loop inside the dry box. To date, we have encouraging results when welding the g-in.-diam tubing using a weld insert, but equipment problems have hindered our effort on the 1%-in. sizes. Two electron-beam welds are required to attach the vent tube to the bottom half of each feed pot. The first attaches a nominal '%-in.-diam “washer” to the end of the tube. Subsequently, this subassembly is welded to a trepanned joint inside the half section. On an earlier attempt to weld this latter joint, the protruding tube was impinged on and melted by the electron beam. The joint was redesigned with an increase in the diameter of the washer (to 0.585 in.) and a tighter fit between the washer and the pothead. Both changes were made to ensure that the tube would not be hit by the beam. These changes resulted in a successful weld on our second attempt. This weld, which is shown in Fig. 15.3, was not only leak-tight when bubble tested under alcohol with an argon pressure of 7 in. Hg, but also helium leak-tight as well. The two feed pot bottoms are presently being remachined to incorporate this design change. 15.4 DEVELOPMENT OF BRAZING TECHNIQUES FOR FABRICATING THE MOLYBDENUM TEST LOOP N. C. Cole As was previously reported,* we have brazed mock- ups of the feed pots and disengaging sections of the chemical processing test loop. We used an iron-based filler metal (Fe—15% Mo—5% Ge—4% C—1% B), which was developed to meet the requirement of this loop. Brazing was accomplished in a vacuum furnace (<1073 torr) by heating at a rate of 5°C/min until flow of the brazing alloy occurred. The parts were positioned in the furnace so the braze alloy fillet could be observed visually. Figure 15.4 shows an example of a 3%-in.- diam molybdenum part in which short lengths of roll-bonded tubing were back brazed and a split ring was brazed around the girth weld. Sections of the 17-ft-long test loop will not fit into our vacuum furnace, and considerable effort has been expended to design and build portable furnaces that will fit around the part to be brazed and will heat only 4. N. C. Cole, “Development of Brazing Techniques for Fabricating the Molybdenum Test Loop,” MSR Program Semi- annu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 172-73. 196 Y-111689 0 1 2 | R | et INCHES Fig. 15.3. Molybdenum half section with vent tube welded into place. The weld joint is indicated by an arrow. Y-111080 0 1 2 e e | INCHES Fig. 15.4. Mockup of a molybdenum feed pot, showing molybdenum tubing back brazed to the bosses of the feed pot and a split sleeve brazed over its girth weld. the immediate braze area. We have developed two types of portable heating sources, one heating by resistance and the other by induction. Each has advantages and disadvantages depending on the size and location of the particular joint to be brazed. The resistance furnaces have tantalum heating ele- ments. Two types of furnaces have been built. One consists of a continuous helical coil that will fit around the joint to be brazed. The other is a split tantalum sheet heater which can be opened and placed over the welded joint in situ and then removed the same way after brazing. This feature is extremely valuable for sections where the furnace cannot be slipped over a large or complex section to reach the braze region. Using these heaters, we have been able to reach brazing temperature on a mockup that consisted of a sleeve around a 7-in. tube with heat sinks similar to that on the actual test loop. We are in the process of deter- mining power output, time to brazing temperature, and other parameters needed for brazing sleeves on tubes of the sizes %, ‘%, %, and Y in. diameter. We have also developed induction heating techniques for use either inside or outside an atmospheric chamber. In the past we have experienced problems with arcing in a dry box under argon or helium atmospheres as well as in vacuum when the brazing alloy binder (used for preplacing the filler metal) volatilized.* We have over- come this difficulty by installing an auxiliary trans- former that changes the high voltage from the induction machine to low voltage—high amperage at the coil. With this attachment we have been able to braze inside the dry box in argon or helium without arcing problems. In addition, by changing the size and shape of the copper leads (from thin-walled tubing to thick bus bars), for the first time we have been able to achieve brazing temperature using split coils. Using a l-in. split coil either inside or outside the dry box, we have reached brazing temperature on a '/4-in. tube and matching split-sleeve assembly. Split coils have an advantage in removability, as discussed above for the split-resistance heater. Unfortunately, the 1-in. split coil does not couple well enough with the other sizes of tubing. As a result, we have ordered additional split coils for the other sizes, both smaller and larger. Both types of heating techniques can be used inside an atmospheric chamber, but some joints will have to be made with portable atmosphere-protection devices. With induction heating we will be able to braze by protecting the molybdenum part from oxidation. If we use the refractory-metal resistance furnace, we will have to devise a technique to protect the heater as well. 197 15.5 COMPATIBILITY OF MATERIALS WITH BISMUTH B. W. McCollum L. R. Trotter 0. B. Cavin J. L. Griffith We are studying the compatibility of potential struc- tural materials with bismuth and bismuth-lithium solu- tions at 700°C. Three different experimental techniques are being used: (1) static capsule tests, (2) quartz thermal convection loops for dynamic tests on samples in up to 0.01 wt % (0.3 at. %) lithium in bismuth, and (3) all-metal thermal convection loops for dynamic testing in up to 3 wt % (48 at. %) lithium in bismuth. 15.5.1 Tantalum and T-111 We previously reported® that T-111 alloy (Ta—8% W-2% Hf) showed excellent resistance to mass transfer but lost its room-temperature ductility while being tested in either bismuth or Bi—0.01 wt % Li at 600 to 700°C for 3000 hr. Tantalum, under similar conditions, showed greater weight changes but no change in ductility. We postulated that the loss of ductility in T-111 could have been .caused by an intergranular hafnium-bismuth reaction. Recently, however, Inouye and Liu,® in studying the brittle behavior of T-111, have found that relatively small additions of oxygen in T-111 can cause room-temperature embrittlement when oxygen is added below 1000°C. For example, the addition of 800 wt ppm oxygen to T-111 at 1000°C will cause embrittlement, but at 815°C it takes only 400 wt ppm. Extrapolation of these data suggests that oxygen concentrations as low as 100 to 200 ppm might induce room-temperature embrittlement if the oxygen is added at 600 to 700°C, which is in the range of our loop operating temperatures. There is evidence that oxygen reacts with hafnium to form hafnium oxide and that the morphology and concentration of this oxide at grain boundaries control the degree of embrittlement. Our samples did pick up oxygen during exposure in the quartz loop, reaching approximately 150 wt ppm oxygen, an increase of about 120 ppm during loop operation. Because the oxygen pickup in T-111 ap- peared to have been associated with the use of quartz as the loop material, we built and operated an all-metal T-111 alloy loop. This loop, which contained Bi—2.5 wt 5. O. B. Cavin and L. R. Trotter, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 173. 6. H. Inouye and C. T. Liu, private communication. % (44 at. %) Li, is presently being examined after operating 3000 hr. If, as seems to be indicated, the embrittlement of T-111 is associated primarily with oxygen and hafnium, it appears that we can circumvent this problem by eliminating the hafnium as an alloying addition. To further check this reasoning, we are testinga Ta—10 wt % W alloy in a quartz thermal-convection loop con- taining high-purity bismuth. 15.5.2 Graphite One of the concerns in the consideration of graphite for processing equipment is the extent to which bismuth-lithium alloys intrude into the graphite pores. Tests with graphite specimens in a loop circulating Bi—100 ppm Li, reported previously,” showed varying amounts of intrusion. During this report period we conducted tests with pure bismuth and with high- lithium alloy (3 wt %, 48 at. %, Li) in crucibles made of three grades of graphite. One similar test, under only slightly different conditions, was conducted in the Chemical Technology Division. As described below, significant differences in intrusion were observed be- tween pure bismuth and the high-lithium alloy, among different grades of graphite, and between the Chemical Technology Division test and our test with the same grade of graphite. Our tests used crucibles 4 in. long and 0.75 in. OD with a wall thickness of 0.19 in., machined from AT]J, AXF, and Graph-i-tite “A”. These three grades, ob- tained from Carbon Products Division of Union Carbide Corporation, from Poco Graphite, Inc., and from Graphite Products Division of Carborundum Company, respectively, differed in pore size and bulk density, which was 1.79, 1.8, and 1.86 g/cm?® respectively. One set of nine crucibles (three of each grade) was filled with high-purity bismuth and another set with Bi—3 wt % Li alloy. Prior to filling, the graphite crucibles were ultrasonically cleaned in absolute alcohol for 30 sec, then dried for 16 hr at 190°C. High-purity bismuth was added to the crucibles in the form of solid cast sticks. In preparation for the alloy tests, sufficient bismuth- lithium alloy to fill nine crucibles was first made in a molybdenum-lined type 304L stainless steel pot by adding small pieces of purified lithium to molten purified bismuth. After each lithium addition, the melt was agitated with a molybdenum rod to ensure com- plete mixing and alloying. The melt was then cast into a 7. O. B. Cavin and L. R. Trotter, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 175-76. 198 thin sheet, and pieces of this alloy were used to fill the graphite crucibles, which were being held above the melting temperature of the alloy. Each crucible was then placed inside a stainless steel capsule and the end cap welded on inside an argon-filled atmosphere cham- ber. Each combination of graphite and metal was tested at 700 * 5°C for 500, 1000, and 3000 hr. As of this writing, only the 500-hr specimens had been examined. Metallography of the crucible cross sections after the 500-hr tests indicated that the high-purity bismuth did not penetrate the open porosity of any of the grades of graphite tested. Conversely, in the Bi—3 wt % Li tests, the melt did intrude into crucible walls. Radiographs of the three graphite grades after 500-hr exposures to the bismuth-lithium alloy are shown in Fig. 15.5. One can see that the greatest amount of penetration occurred in the ATJ graphite, the lowest-density grade tested. This penetration was confirmed by metallography, and in some regions of extremely low density the melt had completely penetrated the wall. Figure 15.6 shows a radiograph of a longitudinal half section of ATJ which Y-110855 Fig. 15.5. X-may radiographs of three grades of graphite crucibles containing Bi—3 wt % (48.2 at. %) Li and tested at 700 + 5°C for 500 hr. Y-112034 Fig. 15.6. X-ray radiograph of a half section of the ATJ graphite crucible shown in Fig. 15.5. The crucible was probably rotated from the previous radiograph. illustrates the variable degree of intrusion. This kind of variation is not unusual considering the inherent varia- tions in porosity of the graphite bodies from which the crucibles were made. The Chemical Technology Division used an ATJ graphite crucible (1% in. OD, 1% in. ID, and 6 in. long) to test the stability of lithium-bismuth solutions in contact with graphite.® The crucible was first degassed by heat treating for seven days at 1000°C in a flowing argon atmosphere. It was then cooled to room temperature, and sufficient amounts of solid lithium and bismuth were added to produce 300 g (1% in. deep) of a Bi—2.2 wt % (40 at. %) Li alloy. The temperature of the crucible was raised to 650°C in a flowing argon atmosphere and held for 30 days, during which time the melt was periodically sampled to determine the lithium concentration in the melt. 8. F. J. Smith and C. T. Thompson, private communication. 199 Chemical analyses of filtered samples showed that the lithium concentration was 2.05 * 0.03 wt % and remained constant with time. Metallographic examina- tion of the cross section of this crucible did not indicate any significant metal intrusion, as shown in Fig. 15.7. There is, however, a thin layer of as yet unidentified nonmetallic material along the surface. It is difficult to reconcile the absence of intrusion with the large open porosity near the surface unless some surface layer existed to seal off the pore entrance. At present, discrepancies observed in the two differ- ent static crucible tests can only be ascribed to variations in experimental technique, such as crucible degassing and alloy preparation. A program has been initiated to investigate the effect of these and other possible variables and also to determine how lithium additions to bismuth increase the penetrating character- istics of the melt. 15.5.3 Tungsten-Coated Hastelloy N Both refractory metals and graphite require that external surfaces of equipment be protected from oxidation at temperatures of interest for MSBR fuel reprocessing. One possibility for overcoming this restric- tion would be to coat the inner surfaces of equipment made of a conventional material with a thin layer of a material that is compatible with the bismuth-lithium alloys. To determine the feasibility of a protective coating on such a material, the inner wall of a Hastelloy N thermal convection loop was tungsten coated by chemical vapor deposition.® Despite repeated attempts to recoat,’® the tungsten had at least one small crack when we decided to proceed with loop testing. The loop operated for only about 24 hr at approximately 700°C when the bismuth started coming through the wall in many spots, as shown in Fig. 15.8, Analysis of the failure is in progress. 15.5.4 Molybdenum A molybdenum all-metal loop scheduled for a 3000-hr run is now operating at a maximum tempera- ture of 696°C and a AT of 116 = 5°C. It contains molybdenum tensile samples in Bi—2.5 wt % (44 at. %) Li. 9. J. 1. Federer, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 231-32. 10. J. 1. Federer, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 176-77. 200 Y-109808 10.010 in. 0.035 INCHES 100X 10.030 in. Fig. 15.7. Photomicrograph of cross section of ATJ graphite crucible tested by the Chemical Technology Division. (¢) Small niece of metal adhering to the inner wall of the crucible. (b) Thin surface layer on graphite. 201 PHOTO 0530-72 | Fig. 15.8. Tungsten-coated Hastelloy N loop tested for approximately 24 hr at 700°C and a AT of 100 + 5°C. Note the small particles of bismuth on the top horizontal portion and the metal that has flowed down each vertical leg. Clamshell heaters are around the high-temperature portion. 15.6 MOLYBDENUM BRAZE ALLOY COMPATIBILITY J. W. Koger The compatibility of braze alloys for molybdenum is being determined in environments similar to those that will be encountered during reprocessing. Molybdenum braze specimens previously exposed for 109 hr to H,—20 vol % HF have been subsequently exposed to LiF-BeF,-ZrF4-UF,; (65.4-29.1-5.0-0.5 mole %) salt in a capsule at 650°C for 115 hr. The specimens with 35M 202 braze alloy (Fe—15% Mo—4% C—1% B) lost 0.1118 g, or approximately 0.2%, while a specimen with 42M braze alloy (Fe—15% Mo—-4% C—1% B—5% Ge) lost 0.0462 g, or approximately 0.1%. After the initial exposure to the gas mixture only, both specimens had gained 0.01 g, probably due to the formation of metal fluoride reaction products. These products are likely soluble in the salt and, therefore, contributed to the overall weight loss noted after exposure to the salt mixture. Since molybdenum is known to be rather inert in molten fluoride salts, it is suspected that most of the weight loss may be attributed to the iron in the braze. Part 4. Molten-Salt Processing and Preparation L. E. McNeese Part 4 deals with the development of processes for the isolation of protactinium and the removal of fission products from molten-salt breeder reactors. During this period we continued to evaluate and develop a flow- sheet based on fluorination—reductive extraction for protactinium isolation and the metal transfer process for rare-earth removal. Work was initiated on a com- puter program that can be used for calculating steady- state concentrations and heat generation rates in an MSBR processing plant. The behavior of 687 nuclides in 56 regions that represent the processing plant is considered. The program will be used for carrying out parametric studies involving the fluorination—reductive- extraction—metal transfer flowsheet and for making comparative studies of flowsheets based on other processing methods such as oxide precipitation. Studies related to the chemistry of fuel reconstitution were continued. It is believed that absorption of gaseous UF¢ into molten salt containing UF, results in the formation of UFs, and gold apparatus was found to exhibit to exhibit satisfactory resistance to gaseous UF and to UFs dissolved in MSBR fuel carrier salt (72-16-12 mole % LiF-BeF,-ThF,) at 600°C. Under certain conditions, UFs disproportionated slowly, with the rate of disproportionation being second order with respect to UFs concentration. Studies were continued on the equilibrium distribu- tion of lithium and bismuth between liquid lithium- bismuth alloys and molten LiCl over the temperature range 650 to 800°C. Data from these studies are consistent with the observed behavior of lithium in the second metal transfer experiment (MTE-2) completed previously. The data are also consistent with the observed behavior of lithium in an engineering experi- ment involving the metal transfer process (MTE-2B). In this experiment the rate of transfer of lithium to LiCl from lithium-bismuth solutions containing 3.5 to 15 at. % lithium has been measured. The results indicate that the concentration of lithium in LiCl that is in equilib- rium with a 5 at. % lithium-bismuth solution, which will 203 be used for removal of trivalent rare earths from LiCl, will have a negligible effect on the metal transfer process. The concentration of lithium in LiCl that is in equilibrium with 50 at. % lithium-bismuth alloys, which are proposed for removal of divalent rare earths from the LiCl, is about 500 wt ppm; however, only about 2% of the LiCl is fed to the divalent rare-earth removal step, and the transfer of lithium will occur at an acceptably low rate. Installation of the third experi- ment (MTE-3) for development of the metal transfer process was completed. The system has been leak checked and treated with hydrogen for oxide removal, and the salt and bismuth phases are being purified and transferred to the system. The experiment will use salt flow rates that are 1% of the estimated flow rates required for processing a 1000-MW(e) MSBR. Design was initiated for a facility in which we will carry out the fourth metal transfer experiment (MTE-4). This experiment will use salt flow rates that are S5 to 10% of those which will be required for processing a 1000-MW(e) MSBR and will yield information on the rate of transfer of rare earths in equipment of a design suitable for a processing plant. Our work on contactor development was continued successfully during this report period. Mass transfer experiments were carried out in which the rates of transfer of zirconium and uranium from molten salt to bismuth were measured in a 24-in.-long, 0.82-in.-ID column packed with ¥;-in. molybdenum Raschig rings. The measured HTU (height of a transfer unit) values range from 2.3 to 4.4 ft, which indicates that packed column contactors can be used successfully in MSBR processing systems. Studies were continued on mechani- cally agitated salt-metal contactors that are of particular interest in the metal transfer process. We have continued studies of oxide precipitation as an alternative to the fluorination—reductive extraction method for isolating protactinium and to fluorination for subsequently removing uranium from MSBR fuel salt. Additional data were obtained to define more accurately the conditions required for the precipitation of protactinium from an LiF-BeF,-ThF, (72-16-12 mole %) solution containing UF, by sparging the salt with H, O-HF-Ar gas mixtures. Operation of a small- scale engineering facility was continued for investiga- tion of the precipitation of UO,-ThQ, solid solutions from molten fluoride salt by contacting the salt with Ar-H, O mixtures. The precipitates have been observed to settle rapidly, and the salt has been separated from the oxide by decantation; minimal entrainment of oxide has been observed. Samples of the salt and oxide 204 precipitate have shown that the two phases are not in equilibrium. A precipitation process model in which the solids, once formed, do not equilibrate with the salt has been found to agree quite well with experimental data. It appears that this effect can be exploited in order to remove most of the uranium from MSBR fuel salt, without the attendant removal of significant quantities of ThO,, in a single-stage system rather than in a batch countercurrent system containing three or more stages (as we had thought would be required previously). Results of these experiments are encouraging. 16. Flowsheet Analysis The final report was written for a design study and a cost estimate for a fluorination—reductive-extraction— metal transfer processing plant that continuously pro- cesses the fuel salt from a 1000-MW(e) MSBR. The design study pointed out the need for additional information in three important areas: (1) finding materials of construction suitable for containing molten bismuth and bismuth-salt mixtures, (2) determining the chemical behavior of noble metals in an MSBR, and (3) preventing entrainment of bismuth in salt leaving bismuth-salt contactors. A computer program that can be used for calculating steady-state concentrations and heat generation rates in an MSBR processing plant is being developed. The behavior of a total of 687 nuclides in 56 regions that represent the processing plant is presently treated by the code. 16.1 DESIGN STUDY AND COST ESTIMATES OF A PROCESSING PLANT FOR A 1000-MW(e) MSBR W. L. Carter E. L. Nicholson A design study and a cost estimate of the fluorina- tion—reductive-extraction—metal transfer processing plant for a 1000-MW(e) MSBR were concluded with the writing of a final report.! Most of the results from the study were summarized previously.? The estimated direct cost of the plant for a ten-day processing cycle is $20.6 million, and the indirect cost is $15 million; the 1. W. L. Carter and E. L. Nicholson, Design and Cost Study of a Fluorination—Reductive Extriction—Metal Transfer Pro- cessing Plant for the MSBR, ORNL-TM-3579 (May 1972). 2. M. W. Rosenthal, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL4728, pp. 178—83. total investment required is $35.6 million. Lowering the processing rate reduces the capital cost, and the total investment was estimated to fall to $25 million at a 37-day cycle time. The study, which consisted of a critical design analysis of the flowsheet for the processing plant, not only gave capital and operating costs but also identified several areas where additional information is required. Laboratory and engineering data show that the chemi- cal basis of the fluorination—reductive-extraction— metal transfer process is fundamentally sound; however, further development work is required in the areas discussed below. The most basic problem is to find a suitable material of construction for containing molten bismuth or bismuth-salt mixtures. Molybdenum and graphite have exhibited excellent corrosion resistance to the process fluids. However, molybdenum is expensive and ex- tremely difficult fo fabricate, and the technology for fabricating the shapes and sizes required for a process- ing plant has not yet been developed. The use of graphite components in an otherwise all-metallic system introduces design problems, and additional data are needed on the compatibility of graphite with bismuth containing reductant. Each of these materials needs further evaluation and development. In the design study, we assumed that noble-metal fission products, that is, fission products whose free energies of formation as fluorides are more positive than the free energy of formation of CrF,, would be in a reduced state in the reactor and would be removed rather quickly after formation by adhering to reactor and heat exchanger surfaces. This assumption consider- ably diminished the heat load in the processing plant caused by the decay of fission products, particularly in the gas recycle system. Although experience in MSRE operation indicates that noble-metal fission products were removed from the salt, the behavior of these fission products is not adequately understood; thus additional data on their behavior in an MSBR are required. Since nickel-base alloys (of which the reactor would be fabricated) are rapidly corroded by molten bismuth, it is important that bismuth be contained in the areas of the processing plant where it is used. Consequently, if bismuth is entrained in salt leaving a salt-bismuth contactor, adequate measures for its removal to accept- able levels must be taken. The problem not only consists in removing bismuth to low concentrations but also in detecting and measuring extremely small amounts of bismuth entrained in salt. Current develop- ment work on salt-bismuth contactors should provide a better understanding of the extent of entrainment and will allow testing of bismuth removal devices. 16.2 MULTIREGION CODE FOR MSBR PROCESSING PLANT FLOWSHEET CALCULATIONS C.W.Kee M.J.Bell L.E.McNeese We are developing a computer code that can be used for calculating steady-state concentrations and heat generation rates in an MSBR processing plant. The behavior of a total of 687 nuclides in 56 regions is treated by the code. Consideration of this number of nuclides is necessary, because many important heat sources result from the decay of nuclides having short half-lives, while nuclides that are present in significant concentrations are normally long-lived or stable. The program can be easily expanded to include as many as 250 regions if needed. Each region can consist of two phases that are in equilibrium, and a given region can communicate with any other region by specifying a flow of either of the phases that are present or by rate-limited transfer of material from one of the phases. The performance of a molten-salt breeder reactor is represented by the computer code MATADOR, which has been described previously.! Since there is no net accumulation of any nuclides in a given region at steady state, a material balance on nuclide i provides one equation for each of the regions considered. The concentrations of nuclides in the second phase of a region that represents an equilibrium contact are related to those in the first phase by a set of distribution coefficients. For region n, a material balance on nuclide i yields the following relation: 205 0=} NG iXin (Vs KV p) i JFi + Z F.S'm,nXi,m+ E FBm,nKi,mXi,m m m m#*n m¥*n + Z (kla)m,nXi,m - 7\i(VS,n +Ki,n VB,n) Xi,n m m¥n ‘Xi,n< E FSn,m+Ki,n Z FBn,m> , (D m m m#*n m¥+¥n where fj i = fraction of decays of nuclide j which give nuclide 7; Fgm n = flow rate of second phase from region m to region n, cm>/sec; Fgp = flow rate 0f3first phase from region m to region n, cm” [sec; K; ,, = equilibrium constant for nuclide i in region n, (moles/cm® second phase/(moles/cm? first phase); (k;@)p,, ,, = first-order rate constant (usually the prod- uct of a mass transfer coefficient and an interfacial area) for transfer of nuclide i from region m to region n, cm?/sec; Vg ,, = volume of second phase in region n, cm’; Vg , = volume of first phase in region n, cm’; X i n = molar concentration of nuclide i in region n, moles/cm?. Approximately 38,000 simultaneous algebraic equa- tions result from consideration of 687 nuclides in 56 regions of the processing plant. The simultaneous solution of this number of equations, plus those required for representing the behavior of nuclides in the reactor, would normally be a formidable task. However, if the processing plant is considered separately from the reactor, and if the concentrations of each nuclide in the processing plant are calculated before the concentra- tions of daughters of the nuclide are calculated, the set of 38,000 equations can be divided into 687 subsets, each of which consists of 56 equations. Direct solutions can then be obtained for each of the subsets, and, in this manner, the concentrations of all nuclides in all regions of the processing plant can be calculated directly for a specified set of concentrations in the reactor. The results of this calculation are then used with MATADOR for obtaining an improved estimate for the concentrations of nuclides in the reactor. Subsequently, the concentrations of nuclides in the processing plant are recalculated. This sequence of calculations provides for rapid convergence to the steady-state concentrations in both the reactor and the processing plant; these concentrations are then used to calculate heat generation rates in each of the regions. Convergence can be obtained with 18 iterations or less, requiring 7 min or less of CPU time on an IBM 360/91 computer. The code has been used for calculating heat genera- tion rates and concentrations of all materials in a 206 processing plant whose operation is based on the fluorination—reductive-extraction—metal transfer flow- sheet, and a copy of the results has been sent to the group of Continental Oil Company employees currently engaged in a design study for an MSBR processing plant as a part of the Ebasco Services subcontract. The present work on the code is aimed at improving the representation of process steps and minimizing the effort necessary for specifying a flowsheet for which calculations are desired. In the immediate future, attention will be given to parametric studies of the fluorination—reductive-extraction—metal transfer flow- sheet, to a more complete representation of the flowsheet, and to comparative studies of flowsheets based on other processing methods such as oxide precipitation. 17. Processing Chemistry L. M. Ferris Several chemical aspects of the metal transfer process'*2 for the removal of rare earths and other fission products from MSBR fuel salt received further study. This work included measurements of the equilib- rium distribution of lithium and bismuth between liquid lithium-bismuth alloys and molten LiCl, measure- ment of the solubility of europium in liquid bismuth, and calculation of the integral heats of lithium-bismuth solutions. Studies of the precipitation of Pa,Og from MSBR fuel salt by sparging with H,O-HF-Ar gas mixtures were also continued. Equilibrium quotients for the reaction PaF¢(d) + % H,0(g) = ", Pa, O5(s) + SHF(g) were determined at 600 and 650°C. Studies related to the chemistry of fuel reconstitution were continued. This work involves investigation of the reaction of gaseous UF, with UF, dissolved in MSBR fuel salt. 17.1 DISTRIBUTION OF LITHIUM AND BISMUTH BETWEEN LIQUID LITHIUM-BISMUTH ALLOYS AND MOLTEN LiCl L. M. Ferris J. F. Land In the metal transfer process,’:? rare earths and the attendant small amount of thorium would be stripped from the LiCl acceptor salt into lithium-bismuth solu- tions having lithium concentrations of S to 50 at. %. In preliminary work,> we showed that at 650°C, both lithium and bismuth distributed between liquid lithium- bismuth alloys and molten LiCl and that the extent of the distribution to the LiCl increased with increasing lithium concentration in the liquid alloy. Furthermore, since the ratio of the “free” lithium to bismuth present in the salt phase was 3, it was suggested that the data could be interpreted in terms of the distribution of the saltlike species Li; Bi between the two phases. M. A. Bredig*® has proposed a model to describe the distribution of Li;Bi between a liquid lithium-bismuth alloy and molten LiCl. In this model, the activity of Li; Bi dissolved in LiCl is defined as ayi.Bi(d) = VLizBi(d) TLisBi(d) » (1) in which d denotes dissolved species in the salt phase, N is mole fraction, and < is an activity coefficient. Equation (1) can also be written as ayi,Bi(d) = VBi(d) YLi; Bi(d) = [VLiey/31 YLigmiey» (@ in which Ng; 4 and N 4y are the measured mole fractions of bismuth and “free” lithium in the LiCl. Concentrations in the alloy phase are defined by assuming that the following reaction occurs when lithium is added to bismuth: 3Li® + Bi® - 3Li* + Bi>" . (3) The ion fractions of Li* and Bi®™ are defined as Li+ e @ Lit " "Bio0 and -2 (s) Bi3~ ’ Ngi3- T g0 1. L. E. McNeese, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 234. 2. D. E. Ferguson and staff, Chem. Technol. Div. Annu. Progr. Rep. Mar. 31, 1971, ORNL-4682, p. 2. 3. L. M. Ferris and J. F. Land, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 191. 4. M. A. Bredig, personal communication. 207 208 in which n denotes the number of moles. It should be noted that ng;, is the number of moles of Bi® present in the alloy after reaction (3) has occurred. The activity of Li; Bi in the alloy phase is defined as 21 i3Bi(m) =X+ Xgio- YLi3Bi(m) (6) in which m denotes dissolved species in the lithium- bismuth alloy. Since the activity of Li; Bi is the same in both phases, at equilibrium and at a given temperature, Eqs. (2) and (6) can be equated. After appropriate substitutions and rearrangements, we obtain, in log- arithmic form, log Ng;(q) = log [NVLicay/3] Ny ;® 1 (3~ N, 3-3N.,) = log +logl, (7) in which &, is the mole fraction of lithium in the alloy. If the ratio of the activity coefficients, L i, Bi(a)/ YLi3Bi(m)> Were constant at a given temperature, I’ would be a constant; in this case, a plot of log Npj(ay OF log [NLi(d)/3] vs the logarithm of the bracketed term on the right-hand side of Eq. (7) would give a straight line of unit slope. During this report period, we made additional meas- urements of the equilibrium distribution of LizBi between liquid lithium-bismuth alloys and molten LiCl in the temperature range 650 to 800°C. The apparatus and general procedure have been described elsewhere.® After each equilibration period, samples of the salt phase were removed by means of molybdenum pipets. Each salt sample was hydrolyzed in water, and the hydrogen that was evolved was collected; then the quantity was determined by gas chromatography. The quantity of hydrogen evolved was assumed to be equivalent to the amount of “free” lithium dissolved in the salt. The hydrolysis residue was acidified to dissolve any precipitated bismuth, and an aliquot of the resulting solution was used for bismuth analysis. When the bismuth concentration in the salt was greater than about 100 wt ppm, a colorimetric analytical method was used. An inverse-polarographic method was used to determine bismuth at lower concentrations. The lithium concentrations in the liquid alloys were determined by flame-photometric analysis. 5. L. M. Ferris, J. C. Mailen, J. J. Lawrance, F. J. Smith, and E. D. Nogueira, J. Inorg. Nucl. Chem 32,2019 (1970). Equilibrium data obtained at selected temperatures in the range of 650 to 800°C obeyed the relationship represented by Eq. (7) over a wide range of alloy compositions, justifying the assumption made regarding the activity coefficient ratio. This is illustrated in Fig. 17.1, using data obtained at 650°C. The placement of the line (which has a slope of unity) was established primarily by the bismuth concentrations because these could be determined analytically with greater accuracy than the lithium concentrations in the salt. As seen, the lithium concentration in the alloy was varied from about 10 to 50 at. %, and the equilibrium lithium and bismuth concentrations in the LiCl changed by about three orders of magnitude. Similar isotherms were obtained at 700, 750, and 800°C. Values of I obtained from the respective isotherms were as follows: Temperature r O 650 0.32 700 0.37 750 0.49 800 0.55 The estimated uncertainty in each value of I' is £0.03. These values can be represented by log I' (0.03) = 0.7256 — 1086/T(°K). Combining this expression with ORNL-DWG 71-128414A LITHIUM CONCENTRATION IN Li-Bi PHASE (mole fraction) 0.05 0.4 0.2 0.3 0.4 0.5 T T T T T "% S g —D - o LITHIUM ® s N o BISMUTH b id — 100 = o 52 2 pd | © z o Q — 4 > = °© 3 g ! = x = — —t & 20 2 4] (o] bd (& % : L~ ] 0-1 E S -1 = - - ( | L | 001 -5 -4 -3 log [ N (3-N ¥ (3-3N.)) Fig. 17.1. Equilibrium distribution of lithium and bismuth between liquid lithium-bismuth alloys and molten LiCl at 650°C. Eq. (7) yields: - 10 ‘]\[Li4 BlG N, (3-3N.) +0.7256 — 1086/ T(°K) . (8) The estimated uncertainty in log Ngicqy 1 +0.05. Equation (8) correlates the data obtained for the equilibrium distribution of lithium and bismuth be- tween liquid lithium-bismuth alloys and molten LiCl in the temperature range 650 to 800°C. 17.2 SOLUBILITY OF EURQPIUM IN LIQUID BISMUTH F.J.Smith C.T. Thompson No measurements of the solubility of europium in liquid bismuth have been reported in the literature. We made measurements of this type over the temperature range 325 to 550°C and obtained results that can be expressed as log Sg, (wt %) = 4.4823 — 2973/T(°K). These solubilities are considerably higher than those reported® for the other lanthanides over the same temperature range. Interestingly, the heats of solution of all the lanthanides in liquid bismuth appear to be about the same. In our previous studies,” we found evidence for the mutual interaction, in liquid bismuth solution, of thorium with the trivalent lanthanides neodymium and lanthanum. Under appropriate conditions, compounds of the apparent composition ThLnBi,, were formed. In recent work, we found that thorium and europium also interact in bismuth solution to form a thorium- and europium-containing solid. However, the mutual solu- bilities of thorium and europium in liquid bismuth are more than adequate to satisfy metal transfer process conditions. 17.3 INTEGRAL HEATS OF LITHIUM-BISMUTH SOLUTIONS L. M. Ferris At various points in the fluorination—reductive- extraction—metal transfer flowsheet! ? for processing MSBR fuel salt, either lithium is added to a bismuth stream or lithium-bismuth solutions having different lithium concentrations are mixed. Also, in an actual plant, mixing of lithium-bismuth solutions will occur in dump tanks that will be provided for collecting various 209 lithium-bismuth soiutions in the event of an emergency or scheduled shutdown. Knowledge of the integral heats of solution allows estimation of the heats of the various reactions involved. We used the emf data of Foster, Wood, and Crouthamel® to calculate partial molar enthalpies for lithium in lithium-bismuth solutions. Their data for solutions containing up to about 55 at. % lithium are presented as: RTIn vy ;= [9397 + 18.16T — 0.010972] — [7103 — 19.44T+ 0.0068T2] N, (9) over the temperature range of about 600 to 800°C. This equation, when used in conjunction with Inagy;=InNp;+1ny (10) and _ , 0 lnay ; AH, ;= —-RT , (11) oT Ny, yields the following expression for the partial molar enthalpies of lithium in lithium-bismuth solutions at 650°C: AH(cal/mole) = —18,683 — 1310V, ; . (12) Here, N;; is the mole fraction of lithium in the lithium-bismuth solution. Integral heats of solution were obtained using the following form of the Gibbs- Duhem equation: - NLi A 5F AHfsoln_NBiIO (AHy/Ng;*)dNy; . (13) Substitution of Eq. (12) into Eq. (13) yields an equation of the form som _ )fx NLi ((1 + b.X)dx (14) x)2 ’ 6. D. G. Schweitzer and J. R. Weeks, Trans. ASM 54, 185 (1961). 7. F. J. Smith, C. T. Thompson, and J. F. Land, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 193, 8. M. S. Foster, S. E. Wood, and C. E. Crouthamel, /norg. Chem. 3, 1428 (1964). which, on integration, gives: AHT = (a+ BN +b(1—Np)In(1=Nyp) . (15) Using the values of ¢ and b from Eq. (12), we obtain the following expression for the integral heats of formation of lithium-bismuth solutions at 650°C: AHT soln (kcal/g-atom soln) = —19.993/N; , To a first approximation, the integral heats of solution can be expressed as AHT soln (kcal/g-atom soln) = —19.1N, ; . (17) It is readily deduced from Eq. (16) or Eq. (17) that the formation of a lithium-bismuth solution of high lithium concentration is a strongly exothermic reaction. For example, the formation of 1 g-atom (about 100 g) of lithium-bismuth (50-50 at. %) from the elements results in a heat release of about 10 kcal. The calculated heats of mixing or dilution of lithium-bismuth solutions based on Eq. (16) are quite small and are endothermic. 17.4 PROTACTINIUM OXIDE PRECIPITATION STUDIES O. K. Tallent L. M. Ferris Studies in support of the development of oxide precipitation processes for isolating protactinium and uranium from MSBR fuel salt have been continued. We are investigating a process in which salt from the reactor would be treated with the appropriate HF-H, O-Ar gas mixture to convert practically all of the protactinium to Pa>* and to precipitate a large fraction of the protactin- ium as Pa, O without precipitating uranium oxide. Protactinium has been systematically precipitated from molten LiF-BeF,-ThF,-UF, solutions at 600 and at 650°C by equilibrating the salt with various HF-H, O- Ar gas mixtures; previously described equipment and experimental procedures were used.’ The data obtained were considered in terms of the equilibrium PaFs(d) + % H,0(g) = 4Pa; O5(s) + SHF(g), (18) 9. O. K. Tallent and F. J. Smith, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 196. 210 for which the equilibrium quotient at a given tempera- ture can be written as 5 Pur 5/2 Pr1,0°"* Vpar, 0, = (19) In the above expressions, d, g, s, &, and p denote dissolved species, gas, solid, mole fraction, and partial pressure respectively. If the ratio py, /Py is fixed at some value 4, Eq. (19) can be written, in logarithmic form, as log Npap, = 2.5log (pyp/A4) —log Q, . (20) At each temperature, log-log plots of protactinium concentration in the salt vs pyp/A should be linear with a slope of 2.5. In a previous report’ we presented data (from experiments 1 and II) obtained at 600°C with salt containing 0.08 and 0.25 mole % UF, respectively. These data obeyed the relationship represented by Eq. (20). However, the protactinium concentrations as determined by gamma spectrometry were usually higher than the protactinium concentrations determined by the alpha-pulse-height method. In examining the gamma-spectrometric method, we found that 2'2?Pb was contributing to the apparent 2°3Pa count rate. After correcting for the contribution of the 2! 2Pb, we obtained excellent agreement between the gamma- spectrometric and the alpha-pulse-height analyses. Recently, we completed two more experiments (ex- periments III and IV) using salt in which the uranium concentration was 2 * 0.2 mg/g (about 0.05 mole %). Data from these experiments, which were conducted at 600 and 650°C, respectively, are incorporated in Table 17.1 with corrected data from our earlier experiments. The protactinium concentrations listed were deter- mined by gamma spectrometry. Log-log plots of the equilibrium protactinium concentrations in the salt vs Py /A gave lines of slope 2.5 at each temperature (Fig. 17.2). This behavior supports the assumption that essentially pure Pa,Oj is the solid phase at equilibrium. From these plots we get Q, values of 3.9 £ 0.5 at 600°C and 21 * 4 at 650°C. In each experiment, the salt samples were also analyzed for uranium. The general behavior of uranium at 600°C is illustrated in the upper part of Fig. 17.3. As seen, the uranium concentration in the salt remained constant, within analytical error, at its initial value while the protactinium concentration decreased with decreasing py; /A4 as the result of the precipitation of 211 Table 17.1. Equilibrium protactinium concentrations obtained by sparging LiF-BeF,-ThF 4-UF, solutions with HF-H, O-Ar gas mixtures under various conditions Composition of carrier salt: LiF-BeF,-ThF, (72-16-12 mole %) Sample Experiment Temperature p pHF/A Concentration in salt CO (atm}) U {mg/g) Pa (wt ppm) 1 I 600 3 0.0103 2.41 9.4 2 111 600 3 0.0104 a 8.1 3 ol 600 3 0.0111 1.92 20.2 4 I 600 3 0.0133 2.92 13.2 5 I 600 3 0.0140 2,59 28.1 6 I 600 3 0.0143 a 18.3 7 I 600 3 0.0143 a 28.1 8 I 600 3 0.0167 a 37.4 9 1 600 3 0.0173 2.62 32.0 10 I 600 3 0.0173 2.63 28.0 11 II 600 3 0.0196 9.07 315 12 i1 600 3 0.0206 1.83 54.9 13 I 600 3 0.0206 2.08 47.5 14 I 600 3 0.0206 1.74 45.1 15 I 600 3 0.0210 3.34 63.0 16 i 600 3 0.0211 2.06 47.7 17 I 600 3 0.0217 3.32 55.6 18 I 600 3 0.0220 9.22 66.1 19 11 600 3 0.0221 2.16 54.5 20 1 600 3 0.0233 2.82 58.0 21 m 600 3 0.0237 2.06 76.5 22 11 600 3 0.0240 9.31 84.0 23 v 650 1.1 0.0200 a 13.0 24 v 650 1.1 0.0262 a 16.7 25 v 650 1.1 0.0342 2,29 29.6 26 v 650 1.1 0.0355 1.80 384 27 v 650 1.1 0.0427 2.19 66.8 28 v 650 1.1 0.0444 2.48 78.9 @gample not analyzed for uranium. ORNL- DWG 72-94RA 100 - o I ra III g 50 Lo v a E . z 5 | — < o '_ Z 20 - i ] z O S + © & T T T 1 ¥ FEXPERIMENT TEMPERATURE e /) ~n o / - [ ] o @ I P9 [e)] Po3* CONCENTRATION (mole ppm) 1o 1 | 0,005 0.0 0.02 0.05 Ryl (A= PHZO/PHF) Fig. 17.2. Precipitation of protactinium from LiF-BeF;- ThF4-UF4 (71.9-16-12-0.05 mole %) according to the reaction PaF5(d) + ,H,0(g) = hPa, 05(s) + SHF(g). Pa, O5. The protactinium concentrations shown in the lower part of Fig. 17.3 were calculated using @, = 3.9 at 600°C. As the value of py /A4 was decreased, a point was finally reached where the uranium concentration in the salt also began to decrease. This point was depend- ent on the uranium concentration in the salt. We interpret this behavior to mean that, at this point, the salt became saturated with both Pa, 05 and a UQ,- ThO, solid solution. Thus, we were able to estimate values for the equilibrium quotient for the reaction YoPay 04(s) + % UF4(d) = PaF(d) + % U0, (ss) , (21) for which 5/4 0= — ) UF, The mole fraction of UQ, in the UO,-ThO, solid solution, Nuoz(ss), was calculated from the reported 212 uranium concentration in the salt and the equilibrium quotient for the reaction ThF4(d) + UO,(ss) = UF4(d) + ThO,(ss) (23) reported by Bamberger and Baes.'® Our estimated values of Q, are given in Fig. 17.4, where they are compared with the values reported by Ross, Bamberger, and Baes.!! The agreement is quite good, considering the uncertainties involved in the methods utilized. If anything, we would expect our values of g, to be high, since we used the highest values of Np,y, and Nuyr, indicated by our data. ORNL~DWG 72~ 97RA 10 T INITIAL U CONCENTRATION . a—=a (mg/q) ,/ A 9.2 / 8 ———— o0 2.8 f C) ® 2.0 / ~ o / E TEMPERATURE 600°C / A / g ¢ r = / 0<: / ! — / & I S 4 7 8 / 3 2t o - 2 Op” [0 © 2 ”40 . [" 2—.—..-_4 P?Ojll /0' /] A / / /, / / 0] - 80 ” yal PG5t CONCENTRATION {wt ppm) i / 0 0.005 0.010 0.015 0.020 0.025 PHF/A (A=PH20/PHF) Fig. 17.3. Estimation of points at which both Pa;0s and UQ,-ThO, (ss) precipitate from LiF-BeF,-ThF, (72-16-12 mole %) at 600°C. ORNL-DWG 72-3218A TEMPERATURE (°C) 700 650 600 o T T 1 0.08 0.06 ~ ~ \ %g 0.04 ~ S >~ Nghy *I e =] ~ T i g \ J. ™ g S * 2 \\ n ”\\ ROSS, BAMBERGER, 1r 0.02 b——— BAES“ \ | 0.04 10.0 10,5 1.0 1.5 10.000/7 (e Fig. 17.4. Estimated equilibrium %uotients for the reaction "Pay05(s) + %4 UF4(d) = PaFs5(d) + “4U0;(ss). The above data strongly indicate that it will be impossible to precipitate a large fraction of the protac- tinium without precipitating some UO, if the initial protactinium concentration is 100 wt ppm (the protac- tinium concentration in an MSBR from which protac- tinium is removed on a five-day cycle). As a con- sequence, flowsheet variations involving coprecipitation of protactinium and uranium will receive further evaluation. 17.5 CHEMISTRY OF FUEL RECONSTITUTION M. R. Bennett L. M. Ferris In the current flowsheet'’> for the processing of MSBR fuel, the fuel reconstitution method involves, first, absorbing the UF,4 evolved in the fluorination step in salt containing dissolved UF, and, then, reducing the resultant higher-valent uranium species to UF, with 10. C. E. Bamberger and C. F. Baes, Jr., J. Nucl. Mater. 3§, 177 (1970). 11. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr., MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 64. gaseous hydrogen. The expected sequence of reaction is: UFs(g) + UF,4(d) = 2UF(d) , (24) 2UFs(d) + Hy(g) = 2UF4(d) + ZHF(g) . (25) Studies of the chemistry involving the fuel reconstitu- tion step, initiated prior to the last reporting period,!? have continued. Originally,' * scouting experiments were conducted primarily to find a container that was inert to UF; dissolved in molten fluoride salts. In these experiments, it appeared that both gold and type ATJ graphite were stable at 600°C to LiF-BeF,-ThF, (72-16-12 mole %) that contained 6 to 12 wt % uranium as UFs. The results of more recent experiments show that, under most conditions, graphite is not suitable for the containment of salts containing dissolved UF;. In two of these experiments, sufficient UF, was added to LiF-BeF,-ThF, (72-16-12 mole %) containing dissolved UF, to produce solutions in which the UF concentra- tions were about 2 and 4 wt % respectively. Analyses of salt samples taken at various times after the addition of UF; to the system showed that the total uranium concentration in the salt remained constant but that the UFs concentration decreased with time according to second-order kinetics. As the UFs concentration de- creased, a corresponding increase in the pressure of the system occurred. Mass-spectrographic analyses of sam- ples of the gas showed that they contained significant amounts of CF, and C,Fg, in the ratio of about 6:1. These results indicated that UFs was disproportionating to UF, and UF,. Since the total uranium concentration in the salt remained constant, we postulate that the UF¢ produced by disproportionation of UF; reacted with graphite to yield dissolved UF,; and gaseous fluorocarbons. The apparent stability of graphite to salt containing UFs noted in our earlier tests'? may be related to the very high UFs concentrations in the salt in those tests, as discussed later. Our most recent series of experiments was conducted using the same type of apparatus and procedure as described previously;'? however, most portions of the system that were exposed to salt or gaseous UFg (the crucible, sparge tube, thermowell, and sampler) were fabricated of gold. In the first experiment in this series (experiment 15-UR), 200 g of LiF-BeF,-ThF; (72-16- 12 mole %) containing 1.13 wt % uranium as UF,; was 12. M. R. Bennett, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 190. 213 first sparged for 48 hr at 600°C with HF-H, (50:50 mole %). Then, sufficient UF4 was bubbled into the salt at 600°C to convert 75 to 80% of the uranium to UF;. The system was subsequently left under a slight argon pressure, and samples of the salt were removed periodi- cally. Analyses of these samples showed that both the U°* and the total uranium concentrations in the salt decreased with time (see Table 17.2), presumably due to the disproportionation of dissolved UF; [the reverse Table 17.2. Data obtained in studies of the reaction of gaseous UF¢ with UF4 dissolved in LiF-BeF,-ThF, (72-16-12 mole %) at 600°C Concentration Time in salt Percer.lt Experiment Sample of uranium (h) Totalu U™ as US* (Wt %) (Wt %) 15-UR 1 0 1.13 0 0 1t 0 201 1.54 76.6 2 1 2.03 1.60 78.8 3 3 1.99 1.58 79.4 4 6 1.90 1.46 76.8 5 24 1.88 1.08 57.4 6 72 1.75 0.90 51.4 7 120 1.67 0.78 46.7 8 168 1.59 0.64 40.2 9 192 1.48 0.58 39.2 16-UR 0? 0 1.69 0 0 12 0 312 2.68 85.9 2 24 3.11 2.68 86.2 4 96 3.17 2.24 70.7 5 120 2.56 2.04 79.7 6 168 2.48 1.94 78.2 17-UR o? 0 257 0 0 15 0 511 452 88.4 30 0 5.08 4.56 89.8 4? 0 493 446 90.5 5 0 5.00 4.46 89.2 6? 0 489 4.68 95.7 7€ 0 512 4.70 91.8 8¢ 0 5.04 472 93.6 9¢ 0 5.30 4.92 92.8 10¢ 0 498 4.72 94.8 11¢ 0 5.36 4.88 91.0 12°¢ 0 5.07 4.60 90.7 13 24 4.75 4.20 88.4 14 96 431 3.76 87.2 15 120 4.28 3.74 87.4 16 144 4.07 3.54 87.0 17 168 3.97 3.34 84.1 4Sample taken after hydrofluorination of the salt. Al uranium present as UF,. bSample taken immediately after addition of UFg to the salt. ¢Sample taken immediately after excess UFg was bubbled through the salt. of reaction (24)]. The disproportionation appeared to be second order with respect to UFg concentration, as evidenced by the linearity of a plot of the reciprocal of the UFs concentration vs time (Fig. 17.5) and the fact that the net decrease in UFs concentration was about twice that of the total uranium concentration (Table 17.2). Salt samples from experiment 15-UR were also analyzed for gold. Each analysis showed that the gold concentration in the salt was less than 200 wt ppm and that corrosion, if it occurred, essentially all took place during the 48-hr hydrofluorination period. No further increase in gold concentration was detected either after the UFs was introduced into the melt or during the subsequent eight-day period in which UFs was present in the salt. These results confirm our earlier indica- tions' 2 that gold is suitable as a container for molten fluoride salts containing dissolved UFs. In experiment 16-UR, UF4 was added to salt contain- ing 1.69 wt % uranium as UF, to produce a salt in which the total uranium concentration was 3.1 wt %; analyses (Table 17.2) showed that 86% of the uranium was present as U" after addition of the UF4. As in other experiments, the UF4 was absorbed very rapidly by the salt. The color of a quenched sample of the original salt that contained uranium only as UF, was light green, but the color of a similar sample taken immediately after addition of UF4 was almost white. As seen in Table 17.2 and Fig. 17.5, both the total uranium and the US" concentrations decreased very slowly with time and did not follow second-order kinetics. These results suggest that some UFg was present in the vapor phase and, therefore, that the system was close to steady-state conditions. Results obtained in experiment 17-UR were similar to those of experiment 16-UR. In the first part of this experiment, we attempted to add to LiF-BeF,-ThF, (72-16-12 mole %) containing 2.57 wt % uranium as UF, the amount of UF4 required to convert all the uranium to UFs. The average of the analyses of six samples taken immediately after addition of the UFg showed that about 91% of the uranium was present as UF. (Table 17.2). As before, the color of the salt after the addition of UF, was nearly white. After these samples had been removed, UF¢ was bubbled through the salt until its sorption on an NaF trap in the exit line was detected. Six samples of the salt were removed immediately after this treatment. Analyses of these samples showed that both the total uranium and the UFs concentrations in the salt had not changed significantly. Quenched samples of the salt after this treatment were practically snow-white. The system was 214 ORNL~DWG 72-1442A 1.8 - - EXPERIMENT 15-UR /0" (wt %) EXPERIMENT 16-UR UR g— " EXPERIMENT 17- ] 100 200 TIME (hr} 50 150 Fig. 17.5. Variation of the U™ concentration in LiF-BeF;- ThF, (72-16-12 mole %) with time at 600°C, starting with different UFs concentrations. Data from experiment 15-UR obey a second-order rate expression. left at 600°C, and the samples of the salt were withdrawn periodically. As found in experiment 16-UR, both the total uranium and the U* concentrations decreased only very slowly (Table 17.2, Fig. 17.5), suggesting that the system was nearly at steady state. We conclude from the above series of experiments that gold is inert at 600°C both to gaseous UF¢ and to UF; (in concentrations up to at least 5 wt %) dissolved in LiF-BeF,-ThF, (72-16-12 mole %). The results obtained also show that UF¢, when added to salt containing dissolved UF,, reacts very rapidly with the UF, to form UFjs according to reaction (24). When excess UF¢ was bubbled through the salt in experiment 17-UR, the oxidation state of the uranium that was dissolved in the resultant salt did not exceed 5+, suggesting that the solubility of UF4 in the salt is low at 600°C. In experiment 15-UR, in which the U** concen- tration in the salt was 1.6 wt % or lower, UF; disproportionated by a second-order process. However, the results obtained in experiments 16-UR and 17-UR, in which the U concentrations were initially greater than 2.5 wt %, suggest that sufficient UF, was present in the vapor phase to retard the disproportionation of UFs. It is probable that no detectable disproportiona- tion would have occurred in these experiments if the entire system had been constructed of gold. We speculate that near-equilibrium conditions were estab- lished initially but that gaseous UF¢ was slowly consumed by reaction with the nickel containment vessel. This effect would be less apparent with high concentrations of UFg and, undoubtedly, was responsi- 215 ble for our earlier indications' ? that graphite was stable in salts containing 6 to 12 wt % UFs. Preliminary experiments relative to the reduction of dissolved UFs with gaseous hydrogen have been con- ducted. No quantitative results are available. It appears, however, that UFs is easily reduced by hydrogen but the hydrogen utilization (in our apparatus, at least) is quite low. 18. Engineering Development of Processing Operations L. E. McNeese Studies related to the development of a number of processing operations were continued during this report period. Additional information on the behavior of lithium during metal transfer experiment MTE-2 was obtained. The results are consistent with a recently developed correlation of data concerning the distri- bution of lithium and bismuth between LiCl and lithium-bismuth solutions. Operation of experiment MTE-2B was continued in order to further study the transfer of lithium from lithium-bismuth solutions containing lithium at concentrations of 3.7 to 16 at. %. Installation of equipment for the third engineering experiment for development of the metal transfer process (MTE-3) has been completed. The equipment has been leak tested and treated with hydrogen for the removal of oxides, and the salt and bismuth phases for the experiment are now being purified and transferred to the system. Design for a facility that will be used for the fourth engineering experiment on the metal transfer process (MTE-4) has been initiated. The experiment will use salt flow rates that are 5 to 10% of those that will be required for processing a 1000-MW(e) MSBR. Overall mass transfer coefficients were obtained with a water- mercury system in a stirred interface contactor of the type being used for developing the metal transfer process. The experimentally determined mass transfer coefficients are quite close to those predicted by extrapolation of a literature correlation that is based on data from organic-solvent—water systems. Experiments were continued in which the rates of transfer of 7 Zr and **7U from molten salt to bismuth were measured during the countercurrent contact of salt with bismuth in a packed column. Design and development work were initiated for the reductive-extraction process facility which will allow testing and development of all steps of the reductive-extraction process for isolation of protactinium with salt flow rates as high as 25% of those required for processing a 1000-MW(e) MSBR. Studies of methods for generating heat in molten salt 216 were continued in order that nonradioactive tests of a frozen-wall fluorinator could be carried out, Tests were made using both induction and autoresistance heating. We have continued to operate a small-scale engineering facility in order to investigate the precipitation of UO,-ThO, solid solutions from molten fluoride mix- tures by contacting the salt with mixtures of argon and water. Designs of the components for the processing materials test stand and of the molybdenum reductive extraction equipment were continued, and fabrication of some of the structural parts of the test stand was started. An eddy-current-type bismuth-salt interface detector was tested at temperatures in the range of 550 to 700°C, and the probe appears to be a sensitive and practical indicator for determining the bismuth level or for locating the salt-bismuth interface. 18.1 LITHIUM TRANSFER DURING METAL TRANSFER EXPERIMENT MTE-2 E. L. Youngblood L. E. McNeese During metal transfer experiment MTE-2, the lithium concentration in the lithium-bismuth phase that was used to extract rare earths from the LiCl decreased from an initial value of 0.35 mole fraction to 0.18 mole fraction after 570 liters of LiCl had been contacted with the lithium-bismuth solution.! Only a small fraction of this decrease can be accounted for by the reaction of rare earths with lithium. The major portion of the decrease is believed to be associated with the circulation of LiCl, since little or no decrease occurred during periods of noncirculation. Previously, the rate of decrease of the lithium concentration in the lithium- bismuth solution in experiment MTE-2 was compared with information that was calculated using equilibrium 1. L. E. McNeese, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 249-53. ORNL -DWG 71-13984RA METAL TRANSFER EXPERIMENT MTE-2 0.4 ] @ :L b d . [ J N [ & 0.3 i ;’ CALCULATED Lot = CURVE e = FOR F=0.95 e g ~. = \ 3 [ ] x 0.2 | ;—:\ Y g o- DATA POINTS FROM MTE-2 oK | 0 100 200 300 400 500 600 VOLUME OF LiCi PUMPED (liters) Fig. 18.1. Lithium concentration in the lithium-bismuth solution. data available at that time.> The experimental and calculated values were in good agreement. Since that time, additional data have been obtained, and a theoret- ical correlation has been developed (see Sect 17.1) for the concentration of metallic lithium and bismuth in LiCl that is in equilibrium with lithium-bismuth solutions. The variation of the lithium concentration in the lithium-bismuth solution during experiment MTE-2 has been recalculated using the latter correlation for determining the concentration of lithium in the LiCl after contact with the lithium-bismuth solution. A comparison of the experimental values with calculated values is shown in Fig. 18.1. The best agreement between the calculated and measured values was ob- tained by assuming that the concentration of lithium in the LiCl after contact with the lithium-bismuth solution was 95% of the equilibrium value. The calculated values based on the new correlation are in better agreement with the data from experiment MTE-2 than are the calculated values based on the earlier equilibrium data. 2. E. L. Youngblood and L. E. McNeese, Molten-Salt Reactor Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 2024, 217 18.2 OPERATION OF METAL TRANSFER EXPERIMENT MTE-2B E.L. Youngblood L. E. McNeese We are continuing to operate metal transfer experi- ment MTE-2B (which was described previously') to measure the rate at which metallic lithium is transferred by circulation of LiCl between alithium-bismuth solu- tion (containing from 3.7 to 16 at. % lithium) and a thorium-bismuth solution (initially containing about 0.02 mole % thorium and 0.2 mole % lithium). Lithium-bismuth solutions containing 5 to 50 at. % lithium are used for removing rare earths from LiCl in the metal transfer process. The concentrations of metallic lithium and bismuth in LiCl that is in equilib- rium with lithium-bismuth solutions containing from about 10 to 50 at. % lithium have been determined in laboratory studies. However, because of the low con- centration (<1 ppm) of metallic lithium in LiCl that is in equilibrium with lithium-bismuth solutions con- taining less than 10 at. % lithium, it is very difficult to determine the lithium concentration by direct analysis of the LiClL. The rate of transfer of metallic lithium by circulation of LiCl in experiment MTE-2B can be determined by several methods that do not require direct analysis of the LiCl; this allows study of the transfer of lithium from lithium-bismuth solutions containing lithium concentrations of less than 10 at. %. The equipment used in experiment MTE-2B has been described previously' and is shown schematically in Fig. 18.2. All components that contact salt and bismuth are fabricated of carbon steel. The main vessel is constructed of 6-in.-diam sched 40 pipe and is divided into two compartments by a partition that extends to within % in. of the bottom of the vessel. The two compartments are interconnected by a pool of bismuth containing reductant. One compartment contains fluo- ride salt (67-33 mole % LiF-BeF,) to which 11 mCi of '47NdF; and sufficient ThF,; to produce a concen- tration of 0.19 mole % had been added. The other compartment contains LiCl. A separate electrically insulated vessel containing a lithium-bismuth solution is connected to the LiCl compartment with a '/;-in.-diam sched 80 pipe. During operation, molten LiCl is circulated between the lithium-bismuth vessel and the compartmented vessel at the rate of about 25 ¢cm?/min; by pressurizing the lithium-bismuth container with argon, the LiCl is forced to flow back and forth between the main vessel and the vessel containing the lithium-bismuth solution. Gas-lift sparge tubes are used to improve the contact of the salt and metal phases. The experiment is being operated at 645°C. The experiment is designed in a manner such that data on lithium transfer can be obtained by the following independent methods: 1. Direct determination of lithium in the lithium- bismuth solution used for extraction of rare earths from the LiCl. Direct determination of the lithium and thorium concentrations in LiCl in equilibrium with the lithium-bismuth solution. . Determination of the rate at which lithium is transferred from the lithium-bismuth solution to the main bismuth pool, as indicated by changes in the distribution coefficients for thorium and '*7Nd. The composition of the fluoride salt and the relative volumes of the fluoride salt and bismuth were chosen so that the maximum thorium concentration that can be obtained in the bismuth is only one-half the solubility of thorium in bismuth. This prevents ARGON INLET AND VENT Pz 6-in. CARBON STEEL PIPE ——al LEVEL ELECTRODES 4-in. CARBON STEEL PIPE V7772 218 the bismuth phase from becoming saturated with thorium. If saturation occurs, the thorium and neodymium distribution coefficients will not be sensitive to the transfer of lithium into the main bismuth pool. . Measurement of the voltage that is developed when the two bismuth phases containing lithium are connected by the LiCl. It has been found that the developed voltage can be interpreted in terms of a concentration cell involving bismuth phases that contain lithium at different concentrations. Data from the initial operation of the experiment have been reported previously.' These data, as well as data obtained during this report period, are summarized in Figs. 18.3, 18.4, and 18.5. The concentration of lithium in the thorium-bismuth phase as determined by direct analysis is compared in Fig. 18.3 with values for the lithium concentration that were calculated from ORNL -OWG 71-6199A 24-i LiF~BeFp— ThFy, 3-in. CARBON STEEL PIPE 5at % Li IN Bi (66.8-33- 0.19 mole %) ALUMINA INSULATORS Fig. 18.2. Schematic diagram of equipment used for metal transfer experiment MTE-2B. emf measurements. The lithium concentration in the thorium-bismuth phase as determined from the distri- bution of thorium between the fluoride salt and the thorium-bismuth solution is compared in Fig. 18.4 with lithium concentration values that were calculated from emf measurements. Values for the lithium concen- tration in the thorium-bismuth solution as indicated from neodymium distribution data are compared in Fig. 18.5 with values for the lithium concentration that were calculated from emf measurements. The values for the lithium concentration in the thorium-bismuth solu- tion as determined by direct analysis, from thorium distribution data, and from emf measurements are in 102 LITHIUM CONCENTRATION (mole %) 0] 400 80O 1200 _'_fi METAL TRANSFER EXPERIMENT MTE- zsj_: —F 219 good agreement. The values based on the distribution of neodymium are somewhat lower and show more scatter than those obtained with the other methods. During the first four months of LiCl circulation (in which 2307 liters of LiCl was circulated), the lithium concentration in the lithium-bismuth solution was maintained at values in the range of 3.7 to 6.0 at. %. The lithium concentration in the lithium-bismuth solu- tion was initially 4.5 at. %; however, it decreased to a value of 3.7 at. % after 1589 liters of LiCl had been circulated. Sufficient lithium was then added to the lithtum-bismuth solution to produce a lithium concen- tration of 6 at. %. During the first four months of ORNL-DWG 71-13978RA RS — I 2000 VOLUME OF LiCl CIRCULATED (liters) 1600 2400 Fig. 18.3. Lithium concentrations determined by direct analysis and by voltage measurements. ORN L-DWG 71-13979RA fi:fl{-f ‘#'T'*'Ififl' ADDED F N LJ,44¥ LITHIUM CONCENTRATION {mole %) 0 400 800 1200 I : === 1 :F 1 L CONCENTRATION IN Bi-Th CALCULATEH: FRQM SMOOTHED EMF DATA . A 1600 2000 2400 2800 VOLUME OF LiCl CIRCULATED (liters) Fig. 18.4. Lithium concentrations calculated from thorium distribution data. 220 ORNL-DWG 71-13980RA 102 .- TR T IR P DS W — - METAL TRANSFER EXPERIMENT MTE-2B ; — ¢ BASED ON FLUORIDE SALT DISTRIBUTION COEFFICIENT ‘ON LiCl DISTRIBUTION COEFFICIENT ‘ | —_— S LITHIUM CONCENTRATION (mole %) 1200 Z CALCULATED FROM SMOOTHED EM J— ——— . 1600 2800 2000 2400 VOLUME OF LiCi CIRCULATED Uliters) Fig. 18.5. Lithium concentrations calculated from neodymium distribution data. operation, the lithium concentration in the lithium- bismuth solution decreased steadily, at a rate equivalent to a lithium concentration of 0.2 ppm in the LiCl after its contact with the lithium-bismuth solution. It would be expected that this lithium would be transferred to the thorium-bismuth solution and that the concen- tration of reductant in the thorium-bismuth phase would increase steadily. However, during this period, the concentrations of both the lithium and the thorium in the thorium-bismuth solution decreased; the rate at which reductant was lost from the thorium-bismuth solution was twice the loss rate from the lithium- bismuth solution. After 1021 liters of LiCl had been circulated through the system, 8.4 g of thorium was added to the thorium-bismuth phase in order to increase the reductant concentration to near its original value. Additional '*7Nd tracer and LiCl were also added to the system at that time. The decrease in the reductant concentration in the thorium-bismuth solu- tion probably resulted from the reaction of reductant with impurities introduced into the system by the argon purges or by the removal of samples of salt and bismuth. If the loss of reductant was due entirely to reaction with oxygen in the purge gas, an oxygen concentration of about 10 ppm would be required. Extrapolation of laboratory data on the concentration of lithium in LiCl that is in equilibrium with a 5 at. % Li-Bi solution indicates a lithium concentration of about 0.02 ppm. Thus the mechanism for the removal of most of the lithium from the lithium-bismuth solution must have been the reaction of lithium with impurities in the LiCl. After 2307 liters of LiCl had been circulated through the system, the lithium concentration in the lithium- bismuth solution was increased to 15.6 at. %. At this lithium concentration, the rate of transfer of lithium by the LiCl was sufficient to cause the concentrations of lithium and thorium in the thorium-bismuth solution to increase at a rate equivalent to a lithium concentration of 0.2 ppm in the LiCl after contact with the lithium-bismuth solution. If it is assumed that loss of reductant by its reaction with impurities in the system continued at a rate equivalent to a lithium concen- tration of 0.6 ppm, the resulting lithium transfer rate (0.8 ppm) is in reasonable agreement with the concen- tration of lithium (1.2 ppm) that would be present in LiCl in equilibrium with a 15 at. % lithium-bismuth solution. The inventory of '*7Nd tracer in the system de- creased more rapidly than expected during operation of the experiment, presumably because of reaction of the neodymium with impurities in the system. The neo- dymium transferred to the lithium-bismuth phase as expected during the experiment; however, because of the decrease in neodymium inventory, the rate of transfer could not be accurately determined. Only a negligible amount of thorium (<10 ppm) has been transferred to the lithium-bismuth solution thus far. A lithium concentration of about 15 at. % in the lithium-bismuth solution is being used in the continued operation of the experiment. The information obtained to date is consistent with the experimentally deter- mined results for the distribution of lithium between LiCl and lithium-bismuth solutions and indicates that the rate of transfer of lithium from a 5 at. % lithium-bismuth solution in the metal transfer process will be negligible. The rate of transfer of lithium from a 50 at. % lithium-bismuth solution will be appreciable; however, such transfer will not necessitate changes in the present processing flowsheet or increases in the rate at which reductant is added to the processing system. 18.3 INSTALLATION, TESTING, AND CHARGING OF MATERIALS TO THE THIRD METAL TRANSFER EXPERIMENT E.L. Youngblood H.O. Weeren L. E. McNeese Equipment for the third metal transfer experiment (MTE-3) has been installed, and the system is currently being prepared for operation. Details of the main process vessels, which are constructed of carbon steel, have been presented previously.®> The experiment is shown schematically in Fig. 18.6 and after installation in Fig. 18.7. The basic equipment consists of a 14-in.-diam surge tank that contains fluoride salt (72-16-12 mole % LiF-BeF,-ThF,), a 10-in.-diam salt- metal contactor, and a 6-in.-diam vessel that contains a 5 at. % lithium-bismuth solution and LiCl. The con- tactor is divided into two compartments by a partition 3. E. L. Youngblood and L. E. McNeese, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL4728, pp. 204-17. 221 that extends to within % in. of the bottom of the vessel. The two compartments are interconnected by a pool of bismuth containing thorium and lithium. The bismuth solution in the contactor also provides a seal between the two compartments. During operation, the fluoride salt will be circulated at the rate of about 33 cm®/min from the surge tank through one side of the contactor by a carbon-steel pump that uses moiten bismuth as check valves.? Lithium chloride will be forced to flow back and forth between the vessel containing the lithium-bismuth solution and the salt- bismuth contactor at the rate of about 1.25 liters/min. The salt flow rates to be used in the experiment are about 1% of those that will be required for processing a 1000-MW(e) MSBR. The system will be operated at about 650°C. Mechanical agitators are used in both compartments of the contactor and in the vessel containing the lithium-bismuth solution in order to improve the contact between the salt and the bismuth phases. The main objectives of the experiment are to deter- mine the rate at which rare earths and alkaline-earth elements will be transferred from the fluoride salt to the lithium-bismuth solution and to determine the dependence of the mass transfer coefficients at the salt-bismuth interfaces on agitator speed. The agitators (see Fig. 18.8) are constructed of molybdenum and use blades having a pitch of 45° in both the salt and the bismuth phases. The agitators are designed to vigorously ORNL-DWG 71-147RA AGITATORS VENT LEVEL ELECTRODES LEVEL ELECTRODES |~ FLUORIDE VENT SALT PUMP [- ] ; —— — ARGON ‘ SUPPLY L ARGON SUPPLY 33 cm¥/min - Bi-Ih Li—Bi 72-16-12 mole % LiF ~BeF,~ThF, FLUQORIDE SALT SALT-METAL RARE EARTH RESERVOIR CONTACTOR STRIPPER Fig. 18.6. Flow diagram for metal transfer experiment MTE-3. 222 PHOTO 2884-718B 7.2 ¢ q. "‘( 'i'!rl \“ _ | AGITATOR DRIVE UNIT o JIR L M () =, - | TR ‘ - FLUORIDE SALT PUMP !|J} e LiCl TRANSFER LINE b ' CONTACTOR 28 Fig. 18.7. Equipment for metal transfer experiment MTE-3 after installation. 223 PHOTO 1829-Tf{ Fig. 18.8. Agitators used for promoting mass transfer between salt and bismuth phases in metal transfer experiment MTE-3. contact the salt and bismuth phases without dispersing either phase. If adequate contact of the salt and bismuth can be achieved without dispersion, this will be a more desirable operating mode, since entrainment of bismuth in the fluoride salt will be much less likely to occur than in systems where the bismuth is dispersed in the salt phase. The agitators are driven by variable- speed motors, which will permit determination of the effect of agitator speed on the mass transfer coefficients at the three salt-metal interfaces. Data on mass transfer coefficients obtained in the experiment will be com- pared with values predicted from correlations that were developed for liquids having physical properties appreci- ably different from those of molten salt and bismuth. Preoperational testing of the equipment, purification of the salt and bismuth, and charging of the salt and bismuth to the system have begun. Initially, the system was determined to be leak-tight at room temperature by use of a helium leak detector. The main process vessels were then heated to 650°C, and a pneumatic pressure test was carried out at 12.5 psig (1.25 times the maximum operating pressure). The leak testing was followed by treatment of the system with hydrogen for 12 hr at 650°C in order to reduce iron oxide that was present on the internal surfaces of the equipment. The fluoride salt, bismuth, and LiCl are also being treated for removal of oxides and other impurities before being charged to the equipment. The quantities of salt and bismuth that will be used in the experiment are shown in Table 18.1. Bismuth for the contactor was treated with hydrogen at 600 to 650°C for 6.5 hr in a carbon-steel treatment vessel and was filtered before being transferred to the contactor. The fluoride salt for the experiment was previously purified by the Reactor Chemistry Division; however, in small-scale metal trans- fer experiments, we have observed that, when salt is contacted with bismuth containing reductant, there is an initial decrease in the quantity of reductant in the bismuth phase. The reason for the decrease has not been determined. In order to minimize this effect in experiment MTE-3, the fluoride salt was contacted (in two batches) with bismuth initially containing 1.9 wt % Table 18.1. Materials to be used in metal transfer experiment MTE-3 . Quantity Volume Weight Material (g-moles) (liters) (kg) Fluoride salt 1733.7 33.6 111.6 (72-16-12 mole % LiF-BeF,-ThFy) Bismuth (containing 304.8 6.6 63.4 about 60 ppm of Li and 1700 ppm of Th) LiC1 237.1 6.7 10.1 Li-Bi (5 at. % Li) 222.0 4.6 44.2 thorium for 48 or 72 hr before the salt was filtered and transferred to the fluoride salt surge tank. The filter, which was made of porous iron, had a mean pore size of about 30 u. We are now making preparations for charging the lithium-bismuth solution and the LiCl to the system. The bismuth for the lithium-bismuth solution will be treated with hydrogen in a second treatment vessel at 600 to 650°C, and the 5 at. % lithium-bismuth solution will be prepared by adding lithium metal to the bismuth. The lithium-bismuth solution will then be filtered and transferred to the system. Thorium metal will be added to the lithium-bismuth solution heel that is left in the treatment vessel, and the resulting solution will be contacted with the LiCl before the latter is filtered and charged to the system. After the salt and metal phases have been charged to the system, the first run will be carried out by observing the rate at which radium, present in the system as a decay product of thorium, transfers from the fluoride salt to the lithium-bismuth solution. Europium-154 (half-life, 16 years) fluoride, in tracer quantities, and lanthanum fluoride will be added to the fluoride salt for subsequent runs. 18.4 DESIGN OF THE METAL TRANSFER PROCESS FACILITY W. L. Carter E. L. Nicholson E. L. Youngblood L. E. McNeese We have begun the design of the metal transfer process facility (MTPF) in which the fourth metal transfer experiment (MTE-4) will be carried out. This experiment, which will use salt flow rates that are S to 10% of those required for processing a 1000-MW(e) MSBR, has several primary purposes: 224 . demonstration of the removal of rare-earth fission products from MSBR fuel carrier salt and accumu- lation of these materials in a lithium-bismuth solu- tion in equipment of a significant size; 2. determination of mass transfer coefficients between mechanically agitated salt and bismuth phases; . determination of the rate of removal of rare earths from the fluoride salt in multistage equipment; evaluation of potential materials of construction, graphite in particular; testing of mechanical devices, such as pumps and agitators, that will be required in a processing plant; and . development of instrumentation for measurement and control of process variables such as salt-metal interface location, salt flow rate, and salt or bismuth liquid level. A schematic flow diagram for the MTPF is shown in Fig. 18.9. The principal equipment items are the fluoride salt surge tank, which has a volume of about 300 liters and will consist of a carbon-steel liner in a stainless steel vessel; a three-stage salt-metal contactor made of graphite and enclosed in a stainless steel containment vessel that may have a carbon-steel liner; a stainless steel vessel having a graphite or carbon-steel liner in which rare earths will be accumulated in a lithium-bismuth solution having a volume of about 100 liters; and a hydrofluorinator that has a volume of about 150 liters and consists of a graphite crucible enclosed in a stainless steel vessel. The conceptual designs of the three-stage salt-metal contactor and its containment vessel (see Figs. 18.10 and 18.11) have been completed. Fuel carrier salt (72-16-12 mole % LiF-BeF,-ThF,) that contains rare- earth fluorides will be circulated between the fluoride salt surge tank and three compartments that are on one side of the central partition in the salt-metal contactor. The fluoride salt will be fed to one of the compart- ments and will flow through the remaining compart- ments in sequence. At the same time, LiCl will be circulated between the vessel containing the lithium- bismuth solution in which the rare earths will be accumulated and the remaining three compartments of the salt-metal contactor. The LiCl will be fed to one of the compartments and will flow countercurrent to the fluoride salt in the three adjacent compartments. The two salt streams are prevented from mixing by a captive pool of bismuth in the bottom of each pair of compartments which constitutes a physical mass trans- fer stage. 225 COMPRESSED AIR XX AR S GON METERING L 12.6 liters/min POT 7 A A 4 N CALIBRATION TANK FREEZE ACCUMUL ATORS ORNL-DWG 71-6227RA COMPRESSED AIR ARGON liters/min —ARGON SPARGE FREEZE VALVE CALIBRATION TANK TN b e AN . ________ X - FLUORIDE > 1 N SALT S == :—:_?| U U J/ H PUMP - Jab - SALT- METAL CONTACTOR ( \ Li-Bi STRIPPER —NH LiCl PUMP FLUORIDE SALT STORAGE WASTE TANK SALT-METAL HYDROF{LUORINATOR Fig. 18.9. Flow diagram for metal transfer process development facility. In each of these stages, the bismuth, which will contain reductant, will be circulated from one side of the central partition to the other. However, no mixing of bismuth between any of the physical stages will occur. At operating temperature (640°C), the vapor pressure of LiCl is appreciable (about 0.5 mm Hg), and a means for preventing transfer of LiCl to the fluoride salt by vapor transfer is required. It is also necessary to prevent the mixing of the two salt streams by entrain- ment of salt mist in the gas space above the salt-metal contactor. Both of these requirements are met by the use of a 4-in.-deep slot around the LiCl compartments that will contain molten bismuth. A metal skirt will be attached to the upper part of the containment vessel and will extend into the bismuth pool in order to provide a seal between the gas space above the fluoride and chloride salts. The hydrofluorinator will be used periodically to purify the fluoride salt and to return rare earths that have accumulated in the lithium-bismuth solution to the fluoride salt. This will allow periodic adjustment of the lithium concentration in the lithium-bismuth solu- tion and will increase the range of process conditions that can be covered. We are presently discussing the conceptual design with a graphite manufacturer in order to ensure the optimum design from the standpoints of ease of fabrication and accepted design technology for graphite vessels. Additional information is required concerning the compatibility of various grades of graphite and bismuth solutions containing reductant. After these questions have been satisfactorily answered, we will continue work on the design of the MTPF. Operation of the facility is expected to begin in FY 1974 and to continue for one to two years. 18.5 DEVELOPMENT OF MECHANICALLY AGITATED SALT-METAL CONTACTORS H. O. Weeren L. E. McNeese Experimentally determined mass transfer coefficients for the stirred-interface type of contactor cell such as that used in metal transfer experiment MTE-3 have been reported in the literature and correlated in various ways. Most of this experimental work has been done with partially miscible solvent-water systems that have been operated at Reynolds numbers almost an order of magnitude lower than those expected in the MTE-3 system. Under such circumstances, the extrapolation of available correlations becomes a somewhat doubtful exercise. A review of the suggested correlations indicated that the variables of greatest significance in correlating mass transfer between different fluids were probably the surface tension and the kinematic viscosity (u/p). These properties of the water-mercury system are generally 226 ORNL-DWG 71-13977RA Fig. 18.10. Three-stage graphite salt-metal contactor for use with metal transfer experiment MTE4. 227 ORNL-DWG 71-14308RA Fig. 18.11. Containment vessel for three-stage graphite con- tactor for metal transfer experiment MTE-4. close to the same properties of the molten salt—bismuth system; thus the water-mercury system was chosen for experimental mass transfer determinations in the belief that extrapolation of the values obtained would be more believable than with any other system. The overall mass transfer coetficient was determined for the transfer of silver from a dilute AgNQ; solution into the mercury phase. A simplified form of the stirred-interface contactor was used in the first experi- ments. The equipment used consisted of a 6-in.-diam circular cell in which sufficient water and mercury to produce a 2.2-in.-deep layer of each of the phases were present. The agitator consisted of two 3-in.-diam paddles (each having four straight blades) that were located on a common shaft about 0.75 in. from the water-mercury interface. The AgNQOj concentration was determined at intervals by titration of samples against an NaCl solution. The rate of change of the AgNO; concentration was then used to calculate the overall mass transfer coefficient. The individual mass transfer coefficients were calculated from the overall coeffi- cient. The experimental results are plotted in Fig. 18.12 in the form suggested by Lewis.* The general range of experimental values reported by several investigators using various solvent-water systems is indicated, and the experimental values obtained from the water-mercury system are shown at the upper right. This plot indicates that extrapolation of the Lewis correlation is a valid 4. J. B. Lewis, Chem. Eng. Sci. 3, 248-59 (1954). 3 ORNL-DWG 72— {446A e U s S B s T I O R R - {WATER TO - / MERCURY| - 7 MERCURY TO | / WATER - RANGE OF EXPERIMENTAL // - VALUES FROM SOLVENT-— WATER SYSTEMS / - / 7 — / ] - / ] —1 LJd L 9 | o : 58 = 5 52 3 — = = fi w 89 — @ - B ° 55 N FLUORIDE - TO 552 BISMUTH 3 o - | | I [ Lo J o] T I A I O S | A B | 10 3 4 5 10 2 5 40 2 5 10 2 5 K2 Re + REE—I_—LI— Fig. 18.12. Correlation of mass transfer data from solvent- water and mercury-water systems. procedure, even over such a wide range of Reynolds numbers. The figure also shows the Reynolds numbers that will probably be obtained at the various salt-bismuth inter- faces in experiment MTE-3. These values are not appreciably different from the Reynolds numbers of the water-mercury system. This suggests that the mass transfer coefficients that will be experimentally deter- mined in the MTE-3 system will be reasonably close to the previously used coefficients that were obtained by extrapolation of the Lewis correlation. We believe that " the mass transfer performance will be strongly depen- dent on cell geometry, and we plan to carry out additional studies in order to optimize the cell design for future salt-metal contactors. 18.6 REDUCTIVE EXTRACTION ENGINEERING STUDIES B. A. Hannaford W. M. Woods D. D. Sood We have continued mass transfer experiments in which the rate of transfer of *7Zr and ?*?U from molten salt to bismuth is measured by adding these materials to the salt phase prior to contacting the salt with bismuth containing reductant in a packed column. Three experiments were carried out in which only °7 Zr was used. A total of 19 to 46% of the *7Zr was observed to transfer to the bismuth, and the resulting HTU values ranged from 4.3 to 4.7 ft. In the final experiment, both *7Zr and %2°7U were used. The fractions of these materials that transferred to the bismuth were 14 and 15% for °7Zr and 237U, respectively, and the resulting HTU values were 3.6 and 4.6 ft respectively. System Modifications and Maintenance During the preparations for additional ®°7Zr tracer experiments, a series of minor leaks led to the replace- ment of several transfer lines and the removal of the filters from the bismuth and salt transfer lines exiting from the treatment vessel. A leak that occurred downstream of the salt filter appeared to be due to external air oxidation. All of the vessel-connecting lines that were part of the original installation were replaced. The dip tube attached to the bismuth transfer line was found to be plugged with an iron deposit and was replaced. Makeup salt was added to the treatment vessel, and most of the salt was transferred to the salt feed tank. The remaining salt and bismuth were treated with 30% HF-H, in order to oxidize the iron present in the bismuth into the salt heel overlying the bismuth. The sait heel was sampled and discarded. Both the bismuth and the salt filters were permanently removed from the system when cracks were observed in the weld joints attaching them to transfer lines. A small salt leak occurred in the line connecting the salt head tank to the column, and a similar failure appeared below the flowing bismuth sampler. All of the lines connecting the salt head tank, the column, the specific gravity pot, and the flowing bismuth sampler were replaced with steel tubing coated with nickel aluminide in order to evaluate the performance of the coating for protecting the steel from air oxidation. Concurrently with the maintenance work, several attempts were made to determine the concentrations of reductant in the bismuth phase in the treatment vessel through the use of a beryllium or uranium electrode. The results were only marginally successful; the beryl- lium had a tendency to reduce uranium from the salt, and a uranium electrode using a uranium-bismuth solution and a salt bridge exhibited excessive scatter in the variation of emf with changes in reductant concen- tration, as determined by reductant addition and chemical analyses of the salt and bismuth. Mass Transfer Runs Using * 7 Zr Tracer Several changes were made to reduce the scatter in #7Zr counting data reported previously.® These in- 228 clude: (1) the use of a deeper submergence (about 1 in.) of the sample capsules within the flowing bismuth sampler and (2) a modification of the sampler housing to maintain a leak-tight seal on the sampler capillary tubing at all times as a means of prohibiting the entry of air to the housing. In order to buffer the system with respect to the presence of trace oxidants in the argon cover gas and possible release of HF from the graphite crucible, the uranium inventory of the system was increased from 0.135 to 0.315 g-mole. After bismuth and salt had been circulated through the system to remove oxide from the new transfer lines, the bismuth and salt were treated with a 25 mole % HF-H, mixture for 15 hr. This treatment was followed by hydrogen sparging at 12 scfh for 8 hr and argon sparging for 60 hr. Lithium-bismuth alloy containing 1 g-equivalent of lithium was then added to the treatment vessel. Uranium analyses of bismuth and salt showed a uranium distribution coefficient of 0.30, which indi- cated that about two-thirds of the lithium in the lithium-bismuth alloy was oxidized before being added to the treatment vessel and/or was consumed by a side reaction in the treatment vessel. Additional lithium- bismuth alloy was added to the treatment vessel, and a nontracer run was made in order to bring bismuth and salt in all parts of the system to near-equilibrium conditions. About 15 mCi of *7 ZrQ, was added to the salt feed tank, and mass transfer run ZTR-7 was carried out. Flowing bismuth samples showed remarkably little scatter in °7Zr activity, with six of seven samples varying less than 5% from the average of all samples. This indicates that earlier difficulties with sampling the flowing bismuth stream have been alleviated. The material balance on ®7Zr tracer was excellent: 95% on flowing stream samples and 99% on the total inventory of ®7Zr from the beginning of run ZTR-7 to the end of the subsequent run (ZTR-8), which was carried out with feed streams that were essentially in equilibrium. The calculated HTU for zirconium in ZTR-7 was 4.3 ft, as shown in Table 18.2. In order to increase the zirconium distribution coeffi- cient (Dz,) to a suitable value for run ZTR-9, a total of 500 g of lithium-bismuth alloy (~1.25 g-equivalent of lithium) was added to the treatment vessel. Operation of the equipment was smooth during the run, although an increased pressure drop in the salt overflow line required that the top of the column be operated at an argon overpressure equal to 12 in. H,O. In order to obtain precise values for the net sample weights, the samples were cut open and the contents were drilled 5. B. A. Hannaford et al., MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 21214, 229 out on a small lathe. Salt was quantitatively removed from the samples by this technique; complete removal of the bismuth required a brief exposure to 6 N HNO, after most of the bismuth had been drilled from the sampler. The ®7Zr tracer balance in ZTR-9 was 127% based on flowing stream samples and 104% based on tank samples. The ®7Zr activities in the flowing stream samples were constant to within 8% of their average for salt and within 2% for bismuth. The calculated HTU value was 4.6 ft. Preparatory to the next tracer experiment, a 1.05-g bar of beryllium was dissolved in the treatment vessel in order to increase Dz,. Following a routine operation in which the bismuth and salt were circulated through the equipment in order to bring the salt and bismuth in the entire system to the same concentrations, °7Zr was added to the salt phase and tracer experiment ZTR-10 was carried out. The flowing stream samples were constant to within 8% of their average for salt and to within 7% for the bismuth. The calculated HTU was 4.7 ft, as shown in Table 18.2. Temperature Dependence of Dz, Dy Uranium analyses for the flowing stream samples taken during experiments ZTR-9 and ZTR-10 showed that about 25% of the uranium transferred from the bismuth to the salt during the experiments. Since pre-equilibrated (at 600°C) phases were contacted in the column, such transfer was not expected. Exami- nation of the temperature data from the run showed that column temperatures were generally about 50 to 75°C higher than the treatment vessel temperature. Calculations of Dy and Dz, made for various reductant concentrations and temperatures ranging from 550 to 700°C indicate that, at reductant concentrations used for ZTR-9 and ZTR-10, the uranium distribution coefficient decreases significantly at the higher tempera- tures but that the change in Dy, is less than 10% for a 50°C change in temperature. An experiment was performed in order to verify the calculated effect of temperature variations on the distribution coefficients for uranium and zirconium. Prior to the experiment, the salt and bismuth were hydrofluorinated to remove oxide impurities and were then sparged for 24 hr with hydrogen and for 60 hr with argon. Reductant was subsequently added to the bismuth electrolytically by using a consumable beryl- lium anode and a 1.5-V potential that was applied across the anode and the bismuth pool. After dissolu- tion of 1 g-equivalent of beryllium and following a 20-hr equilibration period, analyses of the phases indicated that no uranium had transferred to the bismuth and that the added reductant had been consumed by impurities (probably FeF,). A second addition of beryllium was made, followed by the addition of about 10 mCi of ®7Zr to the treatment vessel in order to determine the zirconium distribution coefficient and to permit the measurement of Dz, as a function of temperature. Zirconium distribution coeffi- cient values obtained at 600, 620, 640, and 680°C were 0.88, 0.89, 1.05, and 1.06 respectively. The weak dependence of Dz, upon temperature indicated that the nonisothermal condition that existed in the column during runs ZTR-9 and -10 introduced a negligible error in the calculated mass transfer performance. Mass Transfer Experiments with 2°7U and ° 7 Zr Tracers Preparatory to carrying out mass transfer experiment UZTR-1, °7Zr tracer was added to the treatment vessel in order to permit determination of the zirconium distribution coefficient. The resulting salt and bismuth samples indicated a Dz, of about 0.45, which was satisfactory. Following the usual equilibration run, Table 18.2. Summary of mass transfer results from experiments with °7Zs tracer Salt (72-16-12 mole % LiF-BeF,-ThF,4) and bismuth contacted at 650 to 670°C in an 0.82-in.-ID by 24-in.-long packed column Bismuth Salt Flow Zirconium Fraction flow rate, flow rate, rate distribution 97 HTU Run . . of 7" Zr Vg Vs ratio, coefficient, transferred (ft) (ml/min) (ml/min) VplVs Dy, ZTR-7 181 105 1.72 0.98 0.46 4.3 ZTR-9 45 277 0.162 39 0.19 4.6 ZTR-10 46 283 0.163 4.4 0.22 4.7 230 Table 18.3. 237U and ?77r mass transfer results from experiment UZTR-1 Column temperature, 605 to 617°C; bismuth flow rate, 143 ml/min; salt flow rate, 161 ml/min Fraction of tracer Distribution HTU Tracer officient transferred (£t) coetlicie to bismuth Uranium (*370) 0.44 0.15 4.6 Zirconium (°7Zr) 0.46 0.14 3.6 about 5 mCi of 237U and 10 mCi of 7 Zr were added to the salt feed tank, and the experiment was carried out. Data from the run are summarized in Table 18.3. A post-run measurement of Dy, indicated that no change in the zirconium distribution coefficient had occurred during the run. The reported HTU values were calcu- lated from tank samples, for which the tracer balances ranged from 100 to 106%. The HTU value for zirco- nium was significantly higher than the value of 1 ft reported earlier® for a ®7Zr tracer experiment made under similar conditions. The recent results are proba- bly more reliable, since the distribution coefficient was measured both before and after the experiment. Foliowing the post-run equilibration of bismuth and salt in the treatment vessel and sampling to obtain * 7 Zr and 227U distribution data, a 1.93-g beryllium rod was electrolytically dissolved in the salt to increase Dy to near 1.0. Periodic sampling of the equilibrated phases was begun in order to monitor the uranium distribution coefficient over an extended period of time. The samples were analyzed by counting the 0.208-MeV gamma emission from 2*7U (ry,, for 37U = 6.75 days). This allowed more information to be obtained on the rate at which reductant is lost from the bismuth in the treatment vessel. 18.7 DESIGN OF THE REDUCTIVE-EXTRACTION PROCESS FACILITY W. M. Woods W. F. Schaffer, Jr. L. E. McNeese We have initiated design and development work for the reductive-extraction process facility (REPF), which will allow testing and development of equipment suitable for use in a full-scale protactinium removal process based on fluorination—reductive extraction. A preliminary examination and conceptual design for the 6. B. A. Hannaford et al., MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 213. system have been partially completed. The facility will allow operation of all steps of the reductive-extraction process with salt flow rates as high as about 25% of those required for processing a 1000-MW(e) MSBR. A flow diagram for the facility is shown in Fig. 18.13. MSBR fuel carrier salt (72-16-12 mole % LiF-BeF,- ThF,) containing ®? Zr tracer and 0.003 to 0.01 mole % UF, (the uranium concentration expected after fluori- nation of MSBR fuel salt) will be withdrawn from the salt surge tank and will be fed countercurrent to a bismuth stream containing 0.002 mole fraction of lithium in a 2-in.-ID, 6-ft-long packed column, where part of the uranium and zirconium will be extracted into the bismuth phase. The salt and bismuth streams leaving the column will then be fed to a continuous hydrofluorinator, where the uranium, zirconium, thor- ium, and lithium in the bismuth will be converted to fluorides, which will subsequently transfer to the salt phase. Thus the salt stream leaving the hydrofluorinator will have the same composition as the salt entering the extraction column. The bismuth stream leaving the hydrofluorinator will be combined with the desired amount of reductant and returned to the extraction column. Provision will be made for sampling the salt and bismuth streams throughout the facility. The REPF will operate continuously, in contrast to the semicon- tinuous mode of operation for the present reductive extraction test facility (see Sect. 18.6). The facility will have the capability for reaching the flooding capacity of the packed column throughout a range of salt-bismuth flow rate ratios from 0.1 to 15; this includes salt and bismuth flow rates ranging from 0.5 to 3.25 liters/min and 0.2 to 2.5 liters/min respectively. The facility will be used to develop multistage salt-bismuth contactors, salt-bismuth hydrofluorinators, and other equipment items required for the removal of protactinium from MSBR fuel salt by reductive extraction. Data will be obtained on materials of construction, corrosion, and the rate of mass transfer of wuranium, zirconium, thorium, and lithium between salt and bismuth phases; other information required for the evaluation and design of protactinium removal systems will also be obtained. The materials of construction for the facility will consist largely of graphite and molybdenum and will be enclosed in steel for protection from oxidation. 18.8 FROZEN-WALL FLUORINATOR DEVELOPMENT J. R. Hightower, Jr. A nonradioactive demonstration of the use of a frozen wall to protect against corrosion in a continuous 231 ORNL-DWG 71-13789A SALT METERING POT Li-Bi Li-Bi STORAGE -8B "'fl AND 1.6 cm?/min : METERING SALT | PUMP | SALT ' I JACKLEG J SALT I Y SURGE e\ A TANK i _____ Bi—-MIXING i TANK "‘j | =7 -y SALT-Bi I ‘ HYDROFLUORINATOR | u Bi METERING I — POT T | ¥ I | [ | | | } | SALT-Bi l CONTACTOR Bi PUMP ) | 1 — Bi SURGE SALT, 850 cm3/min TANK 3 BISMUTH, 340 cm3/min Fig. 18.13. Flow diagram for the reductive extraction process facility. fluorinator requires a heat source in the molten salt which is not subject to attack by gaseous fluorine. We have previously’” simulated high-frequency induction heating of molten salt in a frozen-wall fluorinator by making measurements of heat generation in aqueous electrolytes contained in equipment similar to a fluori- nator. The measurements indicated that sufficient heat could probably be generated in molten salts if a suitable means could be found for introducing an induction coil 7. 1. R. Hightower, Jr., MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 214. into the vessel and supporting it during operation. An experiment was designed to test one means for intro- ducing the coil. Operational difficulties (which are described later) and the narrow range of acceptable operating conditions have prompted reexamination of autoresistance heating using 60-Hz power. Two series of experiments for study of autoresistance heating were performed and are also described. Molten-Salt Induction Heating Experiment The equipment for the molten-salt induction heating experiment was described previously.” Before the test 232 vessel was heated and filled with salt, preliminary experiments were necessary to determine if 85 A (the current required for operation under the design condi- tions) could be driven through the coil. The induction coil was connected directly to the coaxial transmission cable, which was connected to the generator through a transformer that reduced the terminal voltage of the generator (~19,000 V, peak) to an estimated peak voltage of 3000. In the first attempt to operate the system, a current of 50 A was driven through the coil; however, arcing occurred between the 5-in. coil lead and a %-in. compression fitting, at the point where the high-voltage coil lead penetrated the vessel. Although Teflon insulators had been used to prevent actual contact of the tubing with this fitting, a gas-filled gap existed between the tubing and the fitting. Since the arcing destroyed the lower Teflon insulator, the insulators were replaced by longer ones which completely filled the space between the tubing and the grounded vessel with Teflon (a dielectric); this measure served to prevent arcing at this point. After the Teflon insulators had been replaced, a second attempt to drive the required current through the coil was made. In this test, a current of 90 A was attained; however, arcing occurred between the first turn of the top coil and the high-voltage lead at the point of closest approach, partially melting the coil. No arcing occurred at points protected by the new Teflon insulators. The arcing in both preliminary runs can be attributed to the use of argon as the inert gas in the vessel. Argon has a dielectric strength only about 28% of that of nitrogen. The spacing at the points of arcing was such that, if 3000 V were impressed on the coil leads, electric field strengths high enough to cause dielectric breakdown of argon would indeed be present. If nitrogen had been used as the cover gas in the vessel, no arcing would have occurred. Fluorine should have a high dielectric strength, since highly electronegative gaseous materials quench the electron avalanche that precedes dielectric breakdown. Although the difficulties experienced thus far show nothing fundamentally wrong with induction heating as a method for generating heat in the molten salt of a frozen-wall fluorinator, the complexity of the required electrical equipment and the narrow range of acceptable operating conditions are distinct disadvantages. There- fore, we are reconsidering autoresistance heating and have performed several experiments to study this method of heating. Autoresistance Heating The advantages of autoresistance heating over induc- tion heating include a less complicated power supply, an easier means of controlling heat generation rates, a wider range of acceptable operating conditions, and a simpler method for introducing the flowing salt stream into the fluorinator. The disadvantages stem from the introduction of electrodes into the salt, since these electrodes are subject to corrosion from electrical processes. However, the problems associated with corro- sion are not believed to be severe. Two series of autoresistance heating experiments have been performed. A photograph of the test vessel used in the first series is shown in Fig. 18.14. This vessel was made from 2%-in. sched 40 nickel pipe and had a 3-ft-long vertical test section. The high-voltage electrode (50 to 100 V) was located in a side arm that branched from the top of the test section. The grounded electrode was the portion of the vessel wall below the test section. In an actual fluorinator, molten salt would be fed into the side arm containing the high-voltage electrode. A 6-in.-diam gas-liquid disengagement section was provided above the test section. The equipment was designed to operate with a current of about 15 A through the salt. The current was held constant by manual adjustment of an autotrans- former. Under design conditions, about 860 W would have been generated in the molten salt (corresponding to a specific heat generation rate of 9.8 kW per cubic foot of molten salt). Cooling of the test section was accomplished by conduction through the thermal insu- lation and natural convection to the cell air. These conditions should have allowed a ' -in.-thick salt film to be maintained in the test section with the vessel wall temperature about 20°C below the liquidus tempera- ture of the salt (458°C for 66-34 mole % LiF-BeF,). Three types of operation were studied with the equipment described above. In the first type, the equipment was operated as designed. In the second and third types, an electrode was inserted through a nozzle at the top of the gas-liquid disengaging section. In the second type of operation, the salt level was in the vertical test section below the upper side arm junction; in the third type of operation, the salt level was raised into the gas-liquid disengaging section. The same procedure was followed during each run. The temperature of the test section was adjusted to a temperature just above the liquidus temperature (458°C); the external heaters on the test section were turned off, and the salt was allowed to freeze on the vessel wall in the test section. Current was then passed through the salt to heat the salt internally. In each run the measured resistance of the current path was of the order of 0.1 to 0.3 €2 during periods of high heat generation. The calculated resistance for the proper current path is greater than 1 £2. During formation of a frozen film, the resistance (as measured during short periods with low current and voltage) rose as high as 0.8 Q but would drop to the lower values under prolonged periods with high current. In runs with ! PHOTO 1909 - 74 Fig. 18.14. Test vessel for autoresistance heating test No. 1. 233 the salt level raised into the gas-liquid disengaging section, heat generation rates up to 600 W were achieved, all of the heat was apparently generated in the vicinity of the upper electrode. In order to determine the reason why the current was not traveling the desired path, a new vessel that would allow easier observation of the electrode and the frozen salt was designed and built. This test vessel (see Fig. 18.15) was made from 6-in. sched 40 nickel pipe and had a 4-ft vertical test section. No upper side arm was provided for the high-voltage electrode; instead, the electrodes passed directly into the test section through a nozzle in the top flange. A lower side arm made from 4-in. sched 40 nickel pipe was placed near the bottom of the test vessel to aid in adjusting the liquid level at the top of the test section. Cooling was accomplished by conduction through ' in. of insulation and natural convection to the cell air. A flange with two sight glasses was installed in order to permit observation of the interior of the vessel during operation. Two experiments have been carried out in the second test vessel. In the first experiment, the vessel was cooled (by natural convection to the ambient air) for 1 hr in order to freeze salt on the vessel wall in the test section. Resistance measurements made during the cooldown period indicated resistances of 0.05 2 with no frozen material on the walls and 0.1 £ after 1 hr of cooling. The minimum calculated resistance, assuming complete coverage of the vessel wall by frozen salt in the 4-ft test section, is 3.7 £2. In the second test, the vessel was cooled for 1 hr, drained, and allowed to cool to room temperature. The resistance measured just prior to draining was 0.2 Q. The top flange was removed in order to visually examine the frozen salt layer. The frozen salt layer had a mean thickness of approximately % ¢ in.; however, its crystalline texture was unlike the frozen layers observed in earlier frozen- wall studies with a gas-salt contactor or with electro- lytic cells. Rather than the smooth solid-liquid interface seen in these studies,®:? the frozen salt had dendrites up to % in. long over the surface of the test section; no portion had a smooth appearance. With low heat fluxes, it is believed that the frozen material forms slowly at relatively few nucleation sites and large crystals grow from these sites. The liquid left behind in the vicinity of the crystals would have a slightly lower liquidus temperature than material ini- 8. B. A. Hannaford and L. E. McNeese, MSR Program Semiannu. Progr. Rep. Aug. 31, 1968, ORNL-4344 p. 305. 9. M. S. Lin et al., MSR Program Semiannu. Progr. Rep. Aug. 31, 1969. ORNL-4449 p. 232. PHOTO 0347-72 Fig. 18.15. Test vessel for autoresistance heating test No. 2. 234 tially present. This would allow electrically conducting liquid to be interspersed between the crystals. The resulting salt layer would thus not be electrically insulating. The appearance of the frozen salt left on the vessel walls is in agreement with this hypothesis. With higher heat fluxes, the wall temperature would be low enough to freeze the lower-melting liquid remaining after initial precipitation of some solid. Also, the higher temperature gradient associated with the higher heat fluxes should prevent formation of a dendritic structure. Both effects should lead to a completely solid (and therefore electrically insulating) frozen layer. After a few additional tests, we will modify the system to allow forced cooling and will install a larger autoresistance power supply in order to operate the test equipment with higher heat fluxes. 18.9 ENGINEERING STUDIES OF URANIUM OXIDE PRECIPITATION M.J.Bell D.D.Sood L.E.McNeese Operation of a facility for performing engineering studies of uranium oxide precipitation was continued during this report period.!© Eleven experiments have been performed in the facility, and the oxide precipita- tion process continues to appear to be an attractive alternative to fluorination for removing uranium from fuel salt that is free of protactinium. In these experi- ments, we observed the effects of the salt temperature, the gas feed rate, and the water concentration in the gas on the oxide precipitation rate. The sequence of operations during an experiment was as follows: 1. Contact of the salt containing UF; with an argon- water gas mixture in the precipitator in order to precipitate oxides of uranium and thorium. Samples of the oxide and salt phases were obtained during and after this step. 2. Separation of the oxide from the salt by allowing the oxide to settle to the bottom of the precipitator. 3. Transfer of about 90% of the salt slowly from the precipitator to the salt feed tank in order to minimize the quantity of oxide transferred. 10. M. J. Bell, D. D Sood, and L. E. McNeese, MSR Program Semiannu. Progr. Rep. Aug. 31,1971, ORNL-4728, pp. 220-22. 235 Table 18.4. Summary of data from oxide precipitation runs OP-1 through OP-8 Final uranium concentration L. Composition of Argon-water Water HF wt % . Ilzun Temgerature composition utilization evolved W% SOll.d 0. O Measured solution (percent H,0) (%) (moles) Calculated? Vessel Feed tank (percent UO2) OP-1 600 10-15 0.95 0.95 96 oPp-2 600 10-15 0.60 0.44 0.53 91 OP-3 600 10 0.84 0.36 0.43 91 orP-4 600 25 14 0.85 0.33 0.38 0.44 89 OP-5 600 35 12 1.0 0.21 0.25 0.29 89 OP-6 540 10--15 10 0.76 0.41 0.41 0.46 92 OP-7 600 35 9 1.4 0.16 0.10 0.13 91 OP-8 630 35 11 1.7 0.16 0.033 0.065 89 2A ssuming equilibrium between salt and oxide. 4. Hydrofluorination of the salt and any oxide present in the salt feed tank. The salt was sampled after this step. 5. Transfer of the salt back to the precipitator and hydrofluorination of the salt and oxide in prepara- tion for the next experiment. The salt was sampled during this step in order to ensure complete hydro- fluorination of the oxide. Results of the first eight experiments are summarized in Table 18.4. Experiments were conducted at temperatures from 540 to 630°C, and the composition of the argon-water mixture was varied from 10 to 35% water. Only a slight increase in reaction rate with an increase in temperature was observed; however, the rate of precipitation appears to vary in direct proportion to the rate at which water is supplied to the system. The values for the water utilization observed to date have been uniformly low (about 10 to 15%) and do not vary with the composi- tion of the gas stream. The utilization appears to be simply a function of the residence time of the gas in the salt; thus higher utilization values should be obtained by increasing the contact time of the gas with the salt. The initial uranium concentration in the salt was about 1 wt % in the first seven experiments, and 50 to 90% of the uranium was precipitated as oxide in most of the experiments. Table 18.4 shows the quantity of HF gas collected in each experiment, which is directly related by the stoichiometry of the reaction to the quantity of oxide formed. From the quantity of oxide formed, the equilibrium data of Bamberger and Baes!! were em- 11. C. E. Bamberger and C. F. Baes, Jr., J. Nucl. Mater. 35, 177 (1970). ployed to compute the composition of the oxide and the uranium concentration in the salt, assuming equilib- rium between the salt and oxide. The equilibrium relationship places a lower limit of about 1600 wt ppm on the concentration of uranium in the salt that can be achieved in a single equilibrium stage without the precipitation of large quantities of ThO,. When this concentration is reached, further addition of oxide results mainly in the solid becoming richer in thoria, and in minimal precipitation of uranium. In several experiments, the measured uranium concentrations in the salt samples agree fairly closely with the values calculated by assuming that the salt and oxide phases are in equilibrium. There is a trend in some of the early experiments for the uranium analyses to show slightly higher uranium concentrations than are calculated by assuming equilibrium; and, in the last two experiments, the measured uranium concentrations are below those calculated by assuming equilibrium. The salt samples obtained from the feed tank showed uranium concentrations that are only slightly higher than those measured in the precipitator vessel. Suffi- cient data are not available at this time to determine whether the higher concentrations in the receiver tank were the result of entrainment of a small amount of oxide or due to the salt heel in the receiver vessel (which contains some UF,), but information will be obtained in the next few runs to resolve this uncer- tainty. Also shown in Table 18.4 are the compositions of the UO,-ThO, solid solutions that were observed at the conclusion of each experiment. All of the samples contain about 90% UQ, even though, at the lower uranium concentrations in the salt, the solid in equilib- rium with the salt would contain 50% UQO, or less. Typical results of analyses of oxide samples obtained through experiment OP-8 are shown in Table 18.5. In 236 Table 18.5. Composition of oxide samples obtained in experiments OP-2 to OP-8 Principal phase Other Percent UO, Sample Temperature Salt composition phases — identification CC) (mole % UFy) Percent Percent (percent Experimental Equilibrium of total U0, of total) oxide 2-8D 600 0.16 >90 93.1 1-5 88-92 92 3-3A 600 0.073 80-95 91.4 5-15 73-87 84 4-2A 600 0.088 8095 89.8 5-15 7285 86 4-3A 600 0.067 60-90 89.6 10-40 54--81 82 5-1 600 0.048 >90 89.2 2-10 80-87 75 6-1 540 0.12 >90 92.0 2-10 83-90 93 6-2 540 0.11 >90 92.3 1-5 88-91 93 7-2 600 0.027 75-95 90.6 5-15 68—86 54 8-3 630 0.012 70-90 89.1 10-30 62-80 25 general, the oxide samples contain more than one ORNL-DWG 71— 13599RA face-centered cubic phase — a principal phase that is e | rich in UO, and one or more additional phases that are R N 1212 moe % rich in ThO,. In many cases, more than 90% of the 14 LiF - BeF, - ThF, solid consists of a phase that is about 90% UO,, and less than 10% of the solid consists of a thorium-rich \ phase. Apparently, a minimum of 5 to 10% of the 12 - precipitate will consist of a phase that is rich in ThO,, which is believed to be the result of inadequate mixing of the salt in the present equipment. The experimen- 10 tally observed oxide compositions are compared with . the composition of the solid solutions calculated to be \SALT_OXIDE EQUILIBRIUM AT 600 °C in equilibrium with the UF; concentration in the salt in each experiment. The average UO, content of the precipitates is 80 = 10%, even in those samples where the oxide in equilibrium with the salt is much lower (experiments OP-7 and OP-8). These results are plotted in Fig. 18.16, which shows the ThO, /UO, ratio in oxide samples taken during the first eight experiments as a function of the uranium concentration in the salt. The data are plotted along with the values given by the equilibrium relation of Bamberger and Baes, which predicts that, at low UF, concentrations in the salt, the ratio of ThO, to UO, in the solid becomes quite large. The data of Table 18.5 (the points shown with error bars) agree fairly closely with the equilibrium calculation at high concentrations of UF, in the salt but fall well below the equilibrium curve at low uranium mole fractions. The results for a number of other samples, in which only one face- centered cubic solid could be identified, are also shown in Fig. 18.16. These samples are of interest because, in most cases, the principal solid phase contains a much higher fraction of UO, than is calculated from the equilibrium expression. These results have led to a nonequilibrium precipitation model in which UO,- 08 06 MOLES ThO, PER MOLE OF UO, PRECIPITATED N e N \ b \Ei\ \..‘_I . 1 ———— NO EQUILIBRATION OF SALT AND OXIDE 0 500 1000 1500 2000 (x4076) MOLE FRACTION UF, IN SALT 2500 Fig. 18.16. Compositions of salt and oxide samples obtained during precipitation experiments OP-1 through OP-8. ThO, solid solutions are precipitated that are in equilibrium with the salt at the moment of formation, but in which the solid solutions, once formed, do not rapidly reequilibrate. Thus, solid solutions that are formed early in the precipitation process, and that contain 90 to 95% UO,, are still present during the final stages of precipitation when the solid solutions 237 being formed contain much less UO,. A curve repre- senting this model of the precipitation process is also shown in Fig. 18.16. Based on this model, 99% of the uranium could be precipitated from the salt in one stage as a solid containing 85% UO, (which is acceptable). The experimental data indicate that the oxides actually observed have a slightly lower UO, content, which is believed to be the result of precipitation of a thorium- rich phase due to inadequate mixing of the salt phase. However, the compositions of the solid solutions which have been observed are still well within the range of those required for operation of a flowsheet using a single-stage UO, precipitator. Table 18.6 shows the uranium material balances that were made for experiments OP-2 through OP-7. The inventory of uranium remaining in the salt is calculated using analyses of salt samples and the known inventory of salt in the system. The uranium concentration in the solid is obtained from x-ray analysis of oxide samples, and the total amount of oxide in the system is calculated from the amount of HF evolved. The uranium inventory calculated by the material balances for these experiments was found to agree quite well with the known uranium inventory in the system, which was 0.290 mole. This agreement is considered to be a reasonable check on the consistency of the experimental procedures and the analytical techniques. Two experiments (OP-10 and OP-11) were performed in which a large fraction of the uranium in the system was precipitated and in which the precipitate was allowed to remain in contact with the salt for a period of about one week in order to observe the rate of equilibration of the two phases. Experiment OP-10 was performed with the salt temperature at 550°C, and experiment OP-11 was carried out with the salt temper- ature at 620°C. In each experiment, gas was circulated through the draft tube in order to promote contact of the salt and oxide. Samples of salt and oxide were obtained at intervals during the experiments. No detect- able increase was noted in the uranium concentration in the salt in either case, and little or no equilibration of the two phases occurred. Samples of oxide obtained during experiment OP-11 have been washed free of salt and examined petrograph- ically. Thus far, the samples examined reveal that particles with a high UO, content average 40 + 10 u in size and that these particles are consistently larger than particles with a high ThO, content. More detailed information concerning the size distribution of the precipitate will be obtained from future samples. The slow rate of reequilibration and the particle sizes observed in experiments OP-10 and OP-11 continue to make the oxide precipitation process appear attractive. Future experiments dealing with uranium oxide precipi- tation will be designed to obtain further information concerning the size distribution of the solids, to determine the quantity of oxide that is entrained during. decantation, and to study the behavior of the system using hydrogen-water gas mixtures as the source of oxide. The present facility also lends itself to the initial investigation of the behavior of protactinium and rare earths in systems of engineering interest. An experiment is planned in which niobium will be used as a substitute for protactinium and in which ??>Pa tracer will also be present. It is expected that such an experiment will provide information on the suitability of niobium as a stand-in for protactinium and that it will be possible to demonstrate the selective precipitation of niobium and protactinium oxides in systems where uranium is present at MSBR concentrations. Experiments are being considered in which rare earths will be added to the system; in these experiments, we will attempt to demonstrate that UOQ, can be precipitated from fuel salt without the attendant precipitation of rare earth oxides. Table 18.6. Material balances for experiments OP-2 through OP-7 Total uranium Experiment UF4 in salt UQ, in solid accounted for Percent {moles) (moles) of inventory (moles) OP-2 0.154 0.136 0.290 100 OP-3 0.124 £ 0.010 0.142 £ 0.038 0.266 + 0.048 92 + 16 OP-4 0.110+ 0.015 0.143 + 0.029 0.299 + 0.044 103 £ 15 OP-5 0.073 £ 0.014 0.204 + 0.012 0.277 £ 0.026 96 £ 9 OP-6 0.124 + 0.009 0.170 £ 0.009 0.294 + 0.018 101 £ 6 OP-7 0.029 + 0.010 0.267 + 0.031 0.296 + 0.041 102 + 14 238 18.10 DESIGN OF A PROCESSING MATERIALS TEST STAND AND THE MOLYBDENUM REDUCTIVE EXTRACTION EQUIPMENT E. L. Nicholson = W. F. Schaffer, Jr. Design of the loop components continued, and fabrication of some of the structural parts of the test stand was started. Specific accomplishments include: design of the expansion loops in the molybdenum tubing; completion of preliminary piping drawings and the construction of a full-size mockup of the loop; design of the molybdenum equipment support system; design of a field assembly jig and a handling system so that the loop can be field assembled in Building 4508 and transported to Building 4505 for operation; design of the containment vessel, the seal-welded flange, the freeze valve, and the transition joint nozzles. Calculations were completed and the design was prepared for the loops to accommodate thermal expan- sion in molybdenum tubes for the instrument purge, gas-lift supply, and transfer lines. Tests, by the Metals and Ceramics Division, of '-in. molybdenum tubes brazed into a stainless steel socket and loaded as small cantilever beams showed that the molybdenum ex- hibited brittle failure at a stress of 67,500 psi. An allowable design tensile stress of one-tenth of this value (6750 psi) was used for calculating the size of the expansion loops and for all other stress calculations involving molybdenum. The following conditions must be met to ensure satisfactory performance of the expansion loops: 1. The gas entering the molybdenum tube must be preheated to at least 300°F. This preheating also ensures that all the molybdenum tubing will be heated above the brittle-to-ductile transition temper- ature. 2. The molybdenum lines inside the containment vessel must be coated with a high thermal emissivity coating to a point beyond the expansion loop to promote rapid heating of the inlet gas in the molybdenum tube to the loop operating tempera- ture, thus minimizing the amount of thermal expan- sion accommodation required for each line. 3. The gas flow to each line must be limited to about 1.5 times the normal maximum flow. Alternative designs were attempted in which none of the preceding three restrictions were imposed (i.e., the maximum accidental misoperation case). However, the required expansion loops were too large to be accom- modated in the available space in the containment vessel. Preliminary piping drawings were prepared, and the Metals and Ceramics Division constructed a full-scale mockup of the molybdenum equipment and piping. The mockup has been quite useful for improving the piping arrangement and for ensuring that sufficient space for the fabrication operations is provided around each shop- or field-welded and -brazed tubing joint. We are now preparing detailed piping drawings from the mockup arrangement. The overall height of the molyb- denum system from the underside of the containment vessel flange to the lowest point on the molybdenum tubing is slightly greater than 16.7 ft. An equipment alignment column, attached to the underside of the containment vessel flange and extend- ing down the full height of the loop, is required to brace the equipment and to position guide rings for insertion of the loop in the containment vessel. A molybdenum rod was considered for this alignment column to eliminate any differential thermal expansion and, consequently, any binding of the equipment supports. However, a rod of sufficient rigidity to prevent excessive bending of the molybdenum compo- nents during insertion of the loop in the containment vessel would have been very expensive. A bar having a diametér greater than or equal to 2.1 in. and a length of 17 ft would have been required; a single piece of molybdenum tubing of sufficient rigidity was consid- ered to be completely out of the question. Stainless steel pipe (2% in. sched 80) was selected for the alignment column. The molybdenum equipment will be hung on small-diameter molybdenum rods that will extend to the containment vessel flange and will be braced to the alignment column by horizontal molyb- denum sheet braces that will be anchored to clamps on the column. During fabrication and the subsequent erection of the loop to the vertical position, these clamps will be locked on the alignment column. The clamps will then be loosened before final installation of the loop by removing cylindrical shims so that the stainless steel column can expand freely in the axial direction without binding on the clamps and thereby stressing the molybdenum components. Drawings have been prepared for the equipment support system. The original plan was to fabricate the molybdenum vessels and piping subassemblies in electron-beam vac- uum chambers and inert-gas welding glove boxes in Building 4508 and to carry the fabricated items to Building 4505, where the field assembly work would be done. An existing 16-ft-tall multilevel platform was to be converted to a vertical assembly area in Building 4505. The problems associated with this method for field assembly were numerous, and a more satisfactory 239 assembly concept has evolved and has been adopted. A rigid transportable field assembly jig, which will allow the field assembly work to be done in Building 4508, has been designed; the jig will be used in the horizontal position, and all of the molybdenum components will be at essentially tabletop level. Temporary tubing and equipment supports and protective covers will be installed; and the jig, after assembly of the molyb- denum equipment, will be transported to Building 4505 and erected to the vertical position over cell 4A. The molybdenum equipment will then be released from the jig and lowered into the containment vessel. Rigging procedures have been reviewed to ensure that several handling problems can be solved. Fabrication of the jig, lifting frame, trunnion bars and sockets, and the special hatch cover for the cell has been initiated. When these parts are available, a run of the transport and erection procedure will be conducted, probably using the piping mockup as a stand-in for the molybdenum system. The seal-welded flange and containment vessel designs have been reviewed by the Pressure Vessel Committee and approved on a preliminary basis. Detailed design of these components is under way, and procurement of the necessary materials is in progress. Design of the molybdenum-to-stainless-steel transition joint nozzles for the containment vessel is complete, as are the designs of the freeze valve, the containment vessel support, and the auxiliary platform required inside cell 4A. We expect that most of the design work for the system will be completed in the next six-month period. 18.11 DEVELOPMENT OF A BISMUTH-SALT INTERFACE DETECTOR H. O. Weeren E. L. Nicholson C. V. Dodd An eddy-current type of detector is being devel- opedl? to allow detection and control of the bismuth- salt interface in salt-metal extraction columns or me- chanically agitated salt-metal contactors. The probe con- sists of a ceramic form on which bifilar primary and secondary coils are wound. Contact of the coils with molten salt or bismuth is prevented by enclosing the coils in a molybdenum tube. In operation, a high-fre- quency alternating current is passed through the pri- mary coil, and it induces a current in the secondary coil. The induced current is dependent on the conduc- tivities of the materials located adjacent to the primary and secondary coils; since the conductivities of salt and 12. H. O. Weeren et al., MSR Program Semiannu. Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 222-25. bismuth are quite different, the induced current reflects the presence or absence of bismuth. The principal problem associated with this type of detector stems from the high conductivity of molybdenum, the fabri- cation material of the protective sheath. Two ap- proaches for obtaining an output from the detector are being pursued. The first is based on measuring changes in the magnitude of the induced current; the second is based on measuring the phase shift that occurs between the voltage imposed on the primary coil and that which is induced in the secondary coil. The completed probe has been installed in a carbon- steel test vessel; both the probe design and the test vessel design were described previously. The test vessel has three chambers; the upper chamber is a reservoir for molten bismuth, the middle chamber contains the sheathed probe, and the lower chamber, which simu- lates the interior of the high-temperature containment vessel for the molybdenum reductive extraction equip- ment, contains 13 ft of high-temperature electrical cable in an inert atmosphere. Bismuth can be trans- ferred via pressure between the upper and middle chambers to vary the level around the probe; this level can be measured with a bubbler system and compared with probe readings. The test assembly is shown schematically in Fig. 18.17. During this report period, the test vessel was installed and leak tested. The two upper compartments were filled with argon, and a vacuum was maintained in the lower compartment by means of a vacuum pump. The vessel was heated to 625°C and treated with hydrogen for 16 hr to reduce oxides. Bismuth (21.65 kg) was added and treated with hydrogen for 16 hr at 625°C. ORNL—DWG 72—-1441A VENT ARGON SUPPLY * § j‘i 5 < N et BISMUTH RESERVOIR —MOLYBDENUM SHEATH MANOMETER PROBE TERMINAL—_ Fig. 18.17. Flow diagram of probe test system. The lower compartment was then pressurized with argon to 14.7 psia. The level of bismuth around the probe (measured with a mercury manometer) was varied at approxi- mately 1-in. intervals and was compared with the output from the level detector. This procedure was repeated at operating temperatures ranging from 550 to 700°C in increments of 25°C. A straight-line plot was made at each temperature, and the standard deviation of the readings was 0.05 in. The average measured sensitivity of all the readings was 0.866°/in., as com- pared with the calculated value of 0.827°/in. The average temperature coefficient over the entire tempera- ture range varied from —0.0075°/°C (+0.009 in./°C) at a level of 1 in. to —0.0024°/°C (+.003 in./°C) at a level of 12 in. A plot of the results obtained at 550 and 700°C is shown in Fig. 18.18. Assistance with the measurements and data analysis was obtained from the Nondestructive Testing Group of the Metals and Ceram- ics Division, who also assisted in the design of the level detector. We plan to make measurements of the change in magnitude of the induced current that is caused by a change in the liquid level in the near future, and will repeat some of the measurements with the phase-shift technique in order to check for long-term drift in the readings. 240 ORNL-DWG 72-1445A : Vi . V4 BISMUTH LEVEL (in.) [o)] N 8 © Nj N \. (S, (4] O [:] O 35 37 39 44 43 45 47 PROBE READING (degrees phase shift) Fig. 18.18. Comparison of phase-shift readings and bismuth level. 19. Continuous Salt Purification R. B. Lindauer Studies of the reduction of FeF, in molten salt by contact of the salt with hydrogen in a packed column were continued after the packed column was replaced. The new column is packed with ¥ X % X %, in. wall Raschig ring packing, which has a 32% greater void volume than the original Y ¢-in.-wall packing. Other changes, as described in the previous report,’ were an enlarged, packed liquid-deentrainment section and modifications to the piping to improve salt flow and to reduce the possibility of salt plugs in the vent lines. Before the iron fluoride reduction runs were resumed, four argon-salt tests were made to determine the column throughput. Although the installed rotameters had insufficient capacity to actually flood the column, we were able to attain flow rates that were three times the maximum values possible with the initial column. The pressure drop across the column was somewhat lower than that observed during some runs with the earlier column, indicating that still higher flow rates are possible. In order to evaluate the use of *® Fe tracer rather than colorimetric iron analyses, about 1 mCi of tracer was added to 600 ml of molten salt to which sufficient FeF, had been added to produce an iron concentration of 209 ppm. The salt was then diluted, first to 1000 ml and then to 1750 ml. After the first dilution, the calculated iron concentration in the salt was 132 ppm. Data obtained by counting the 5°Fe tracer showed 137 ppm, while colorimetric analysis indicated an iron concentration of 176 ppm. The second dilution should have reduced the iron concentration to 82 ppm. Tracer counting data showed a concentration of 78 ppm, and a colorimetric analysis showed 122 ppm. Part of the iron fluoride was then reduced by sparging the salt with hydrogen. Tracer counting data indicated that less than 5 ppm of iron remained in solution, while colorimetric analysis indicated iron concentrations greater than 20 1. R. B. Lindauer, MSR Program Semiannu. Progr. Rep. Aug. 31,1971, ORNL-4728, p. 226. 241 ppm. This test demonstrated that the data obtained from colorimetric analyses are less accurate than those obtained by 3°Fe tracer counting, especially at very low iron concentrations. Following these tests, a larger amount (~15 mCi) of S?Fe tracer was added to the salt in the feed tank of the continuous salt purification system after three iron fluoride reduction runs had been carried out using the new column. In 14 succeeding experiments on iron fluoride reduction, the variation between duplicate samples of salt containing iron fluoride was *1.8% by >?Fe tracer counting and *4.2% by colorimetric analysis. Since the average reduction per run was only 4.8%, it is apparent that the tracer method is more suitable for obtaining meaningful data from the new column. Data from the runs are summarized in Table 19.1. The average mass transfer coefficient, ks, for the new column was 7.8 X 107® mole sec ™' ecm ™3, which is lower than the value of 3.2 X 107 mole sec”' ¢cm™ for the first three runs with the old column having the thicker-wall packing. The average mass transfer coeffi- cient, kja, is defined by the following expression: where kja = product of the mass transfer coefficient and the interfacial area per unit column volume, moles sec ! cm’?, L = salt flow rate, g-moles/sec, A = cross-sectional area of column, cm?, H = height of column, c¢m, x; = iron concentration in the inlet salt, X, = iron concentration in the exit salt. The mass transfer coefficient, kjz, is reported instead of k; since the interfacial area between the salt and gas is Table 19.1. Summary of iron fluoride reduction runs at 700°C R Hydrogen Salt flow Iron analysis? H, Perfif’;“_ of Massf;‘ra‘mfter N‘;Tl Concentration Flow r3ate Colorimetric 59Fe tracer utilizationb (;;11;1 ICIO:CU;: 3;2 (::;ZS’ (atm) (liters/min) (cm”/min) (ppm Fe) (dis min "} g_l X 10—5) (%) off-gas sec Vem ™3 x 106) 17 1.0 40.6 420 649° 0.028 2.8 5.6 18 0.20 6.3 393 625 0.24 10.2 7.6 19 1.0 43.4 387 618 0.01 1.0 2.3 5009 6.06 20 1.0 45.0 439 499 5.66 0.06 6.3 154 21 0.19 6.2 390 488 542 0.24 10.7 8.7 22 1.0 52.5 220 489 5.45 3.3 23 0.15 5.7 211 444 5.11 (av of 22, 23) 24 1.0 46.5 373 454 4.92 0.024 2.7 7.3 25 0.16 5.6 291 478 4.53 0.32 14.4 12.4 26 1.0 54.7 322 436 4.30 0.022 2.5 8.6 27 0.16 5.2 459 394 4.12 0.25 12.1 10.1 28 1.0 22.0 160 429 3.95 0.02 2.4 3.5 3889 3.94 29 1.0 41.1 390 384 3.91 0.005 0.56 1.5 30 0.16 4.9 325 370 3.79 0.13 6.4 5.2 31 0.13 2.4 177 337 3.34 0.53 24 11.5 32 1.0 16.1 155 325 3.31 0.001 0.15 0.7 33 1.0 21.7 592 302 3.04 0.12 159 26.0 e aSamples taken after the run — usually the average of two samples. bBased on *%Fe activity. “Samples (9) before run 17 average 66 ppm. %When elapsed time between runs was large, the salt was resampled before the run. not known. Since the salt does not wet the packing, it did not seem reasonable to assume the interfacial area to be equal to the surface area of the packing as is normally done in systems where the packing is wetted by the dispersed phase. About half of the reduction runs were made with dilute hydrogen (13 to 19 vol %). Changes in the hydrogen partial pressure caused no decrease in the rate of reduction; this validates our previous assumption, that the rate of reaction is controlled by the rate at which iron fluoride is transferred from the bulk of the salt to the salt-gas interface rather than being controlled by the rate of reaction at the interface. A total of 18 runs were made to obtain salt holdup data during the countercurrent flow of argon and salt in the column. The data were obtained by observing the 243 amount of salt that drained from the column after salt flow was stopped at the end of a run. Additional data points were obtained by changing the salt or gas flow rate during a run and observing the change in salt inventory in the feed and receiver tanks. Most of the runs were made with a range of salt flow rates and a constant (2-liter/min) argon flow rate. Data from the best runs showed a linear increase in salt holdup from about 5% of the column void volume at 100 cm?® /min to about 11% at 500 cm®/min. Several runs were also made with a range of argon flow rates and a constant (250-cm® /min) salt flow rate. In these runs, the salt holdup appeared to decrease about 25% at the maxi- mum gas flow rate from the maximum salt holdup, which was observed at an argon flow rate of about 7 liters/min. N S w OAK RIDGE NATIONAL LABORATORY MOLTEN-SALT REACTOR PROGRAM FEBRUARY 29, 1972 M.W. ROSENTHAL, DIRECTOR R R.B. BRIGGS, ASSOCIATE DIRECTOR D P.N. HAUBENRE{CH, ASSOCIATE DIRECTOR R H.R.BEATTY,"” BUDGET B DESIGN STUDIES MATERIALS COMPONENTS & SYSTEMS DEVELOPMENT INSTRUMENTATION & CONTROLS PHYSICS MSBR PROCESSING DEVELOPMENT CHEMISTRY E.S.BETTIS R s . . J. R.WEIR M&C L. E. McNEESE** CcT DUNLAP SCOTT R S.4.DITTO 1&C A.M.PERRY R -E. . M. l. LUNDIN*® R H. E. McCOY M&C W. R. GRIMES RC CHEMICAL DEVELOPMENT MSR PHYSICS ANALYSIS HASTELLOY N STUDIES — . H.E. McCOY MEC L. M. FERRIS cT MSBR DESIGN STUDIES PUMP DEVELOPMENT PROCESS INSTRUMENTATION R. E. GEHLBACH M&C M. R. BENNETT cr REACTOR CHEMISTRY . W. R. HUNTLEY** R F.J.SMITH cT ) L. G. ALEXANDER R A.G.GRINDELL" R R.L.MOORE 1&C J.W. KOGER** M&C O.K. TALLENT cT C. F. BAES, JR. RC C. E.BETTIS* GE W. R, HUNTLEY** R P.G. HERNDON" 1&C MSR PHYSICS METHODS B. McNABB M&C J.F. LAND cT C. F. BAMBERGER RC H. A McLAIN"" R L. V. WILSON"* R O.L.SMITH R G.M.SLAUGHTER" M&C C.T. THOMPSON cT E. G. BOHLMANN" RC R.C. ROBERTSON R | E WOLFE® R 1 W. WoOoDS* M&(C S.CANTOR RC 1. R. TALLACKSON 1&C ENGINEERING DEVELOPMENT E. L. COMPERE RC L. V.WILSON** R S.S. KIRSLIS RC H. L. WATTS R MSBR CORE DESIGN GRAPHITE STUDIES L. E. McNEESE*" cT R.G. ROSS RC NUCLEAR INSTRUMENTATION AND WP EATHERLY" MG J.H.SHAFFER RC CONTROLS ANALYSIS H. F. BAUMAN® R P Z'-:- EiiLNLNAFORD S'Tr W. JENNINGS, JR. RC 0.B.CAVIN** M&C A W. P. TEICHERT RC COMPONENTS & SYSTEMS 3 -P. W. H. COOK M&C é 3} ';'(’S:TOWER' JR. g R. M. WALLER RC . . R. L. HAMNER*® M . DUNLAP SCOTT R W. H.SIDES 18C MSBR EXPERIMENTAL PHYS!CS » &G R.B. LINDAUER fos C 1 CLAFFEY® R 0. W. BURKE* 1&C C. R. KENNEDY M&C . ) T MSBE DESIGN STUDIES L CROWLEY R G.L. RAGAN R T.G.KOLLIE® M&C D.D.S00D" cr PHYSICAL AND INORGANIC CHEMISTRY C.E BETTIS® e C 1 GABBARD 8 R.W. McCLUNG" M&C J.S. WATSON cT W, COLLINS S D. L. McELROY* M&C H. 0. WEEREN cT J.BRAUNSTEIN* RC S . R -MWINN R J.P. MOORE" M&C E. L. YOUNGBLOOD cT G.D. BRUNTON" RC W. K. FURLONG R R. H. GUYMQN R ¢ . C.B. POLLOCK MEC W. M. WOODS cT £.A.DOSS RC H. A. McLAIN R R. E.HELMS R J.BEAMS cT L. 0. GILPATRICK RC J.R. MCWHERTER” R R.J. KEDL* R CHEMICAL PROCESSING MATERIALS ) : e . W. TERRY R A. 1. KRAKOVIAK R H. A.CONNELLY cr A. P. MALINAUSKAS RC : = ; V. L. FOWLER cr .D. D. WILSON A R. FEHD R FUNDAMENTAL RADIATION DAMAGE 4. R. DISTEFANGD MEC R 0. PAYNE pd :) DM RREICD:r::DSON zg H. A. McLAIN R IN GRAPHITE - 0.8.CAVIN®* M&C e : e A.N.SMITH* R N.C. COLE" M&C T. 0. ROGERS cr G.D. ROBBINS RC i’ . D. K. HOLMES® ss e - J.S. TAYLOR cT R. A. STREHLOW RC F.C.ZAPP R J.W. KOGER M&C UM TOTH R. E. McDONALD* M&C M. RC HEAT TRANSFER AND PHYSICAL PROPERTIES A § MOORHEAD® &G €. F. WEAVER RC T.S.NOGGLE" sS - . H. W. HOFFMAN®* R S M. OHR* ss A.C. SCHAFFHAUSER M&C PROCESS DESIGN CONSULTANT TECHNICAL SUPPORT E. L. NICHOLSON cT 1. 3. KEYES* R J.W. COOKE*® R F.N.PEEBLES ut M.D. ALLEN" M&C W.L.CARTER cT T.S. KRESS* R N. M. ATCHLEY™ M&C W. F.SCHAFFER, JR. cT ICAL CHEMISTRY PMENT $. J. CLAIBORNE" R E.R.BOYD" M&C ANALYTICAL DEVELO B.J. SUTTON" R B. R. CHILCOAT® M&C J.W. CHUMLEY M&.C MSRE SURVEILLANCE S.W. COOK M&C © E. DUNN* M&C RESEARCH AND DEVELOPMENT e .E. R. H. GUYMON R J.C. FELTNER M&C A.S. MEYER* AC J.W. GREER" M&C R. F. APPLE AC J.L. GRIFFITH M&C J. M. DALE AC C. W. HOUCK” M&C J.P. YOUNG AC F.P. JEFFERS® M&C H. KEATING M&C ANALYSES E.J. LAWRENCE M&C W. J. MASON® M&C L.T.CORBIN" AC B.W. McCOLLUM® M&C C.E.LAMB® AC R.S. PULLIAM® M&C L. G. RARDON* M&C W. H, SMITH, JR.” M&C G. 0. STOHLER" M&C AC ANALYTICAL CHEMISTRY DIVISION H.R. TINCH M&C B BUDGET AND PROGRAM PLANNING OFFICE L. R. TROTTER" M&C C CHEMISTRY DIVISION J.F. WILLMERING* M&O CT CHEMICAL TECHNOLOGY DIVISION J. J. WOODHOUSE" M&C SPECIAL ASSISTANCE D DIRECTORS DIVISION . GE GENERAL ENGINEERING DIVISION M.A.BREDIG® I 1&C INSTRUMENTATION AND CONTROLS DIVISION H. R. BRONSTEIN c M&C METALS AND CERAMICS DIVISION A.S.DWORKIN c R REACTOR DIVISION RC REACTOR CHEMISTRY DIVISION $S SOLID STATE DIVISION UT UNIVERSITY OF TENNESSEE PART TIME ON MSRP DUAL CAPACITY ON ASSIGNMENT FROM TVA GUEST SCIENTIST FROM INDIA cCoO-OP