o | ' - CENTRAL RESEARCH LIBRARY 100 | DOCUMENT COLLECTION :7.253 ORNL-4676 UC-80 — Reactor Technology ‘;" “vlurlw\\ \MJ‘\JM‘ AT MOLTEN-SALT REACTOR PROGRAM SEMIANNUAL PROGRESS REPORT FOR PERIOD ENDING FEBRUARY 28, 1971 OAK RIDGE NATIONAL LABORATORY operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION Printed in the United States of America. Available from National Technical Information Service US. 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. Contract No. W-7405-eng-26 MOLTEN-SALT REACTOR PROGRAM SEMIANNUAL PROGRESS REPORT For Period Ending February 28, 1971 - M. W. Rosenthal, Program Director R. B. Briggs, Associate Director P. N. Haubenreich, Associate Director AUGUST 1971 OAK RIDGE NATIONAL LABORATORY Qak Ridge, Tennessece operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION ORNL-4676 MARIETTA ENERGY SYSTEMS LIBRARI AN 3 445k 0428254 O 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 ORNL4191 ORNL-4254 ORNL4344 ORNL4396 ORNL-4449 ORNL4548 ORNL-4622 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 Contents INT RODUCTION .ottt et et et ettt e ettt ittt ie e ee s ix QUMM AR Y . .o e e e e e e X1 PART 1. MOLTEN-SALT REACTOR EXPERIMENT 1. POSTOPERATION EXAMINATIONS . ... ...\ uttttneteeneneet et e ieaeaaes I 1.1 Outline of Program . .. .. ... i e 1 1.2 Reactor Vessel and Core .......... ST e 2 1.3 Fuel PUmp ..o et 6 1.4 Heat EXChanger .. ... ... . e e e ettt e 11 1.5 Leak at Freeze Valve FV-105 . .. . i e e e i et e 12 1.6 Other EXaminations ... ... .. ... ittt itiae iy e 12 1.7 Evaluation of Tools and Procedures .......... . ... i 13 2. FURTHER INVESTIGATIONS .. ... .. i P 17 2.1 Test of Coolant Salt Flowmeter and Conclusions . ............ .. . ... 17 2.2 Inventories of Residual Uranium and Plutonium .......... ... . .. . i 18 2.3 Search for Unrecovered 235U ..o en ettt e 20 PART 2. MSBR DESIGN AND DEVELOPMENT 3. DESIGN .. e e e e e e e 21 3.1 Single-Fluid 1000-MW(e) MSBR Design Study Report ... ... ] 3.2 Molten-Salt Demonstration Reactor Design Study ........ ... .. .. .. ... i 22 320 Introduction . ... . . s e 22 3.2.2 Addition of Third Salt-Circulation Loop ........ ... it 22 3.2.3 Salt Overflow and Gas Stripping Systems . . .. .. ... i 24 3.2.4 Primary Drain Tank . ... ... o e e 24 3.2.5 Drain Valves for Salt Service . .......... .. ... .. il 25 3.2.6 Heat Exchangers . . .. ... ..o i e e 25 3.2.7 Building and Containment . ........ ... . it e 28 3.3 Initial Temperature Transients in Empty MSBR “Reference Design” MSBR Heat Exchangers . .. . .. 32 3.4 The Consequences of Tubing Failure in the MSBR Heat Exchanger ......................... 34 3.5 Tritium Distribution and Control inthe MSBR ....... ... ... .. ... .. . i i 35 3.6 Industrial Study of 1000-MW(e) Molten-Salt Breeder Reactor ............. ... ............. 36 111 iv 37 MSBE Design . ... et 36 370 General . ... e e e e e e 36 372 MSBE Core Design . ...t e e e e 36 3.7.3 MSBE Primary Heat Exchanger Design . ............ ... . ... ... .. . .. 39 4. REACTOR PHY SICS ..ot ettt e et e e et it 41 4.1 Physics Analysis of MSBR . ... ... e 41 4.1.1 Single-Fluid MSBR Reference Design .. ..... ... ... ..o 41 4.1.2 Fixed-Moderator Molten-Salt Reactor .......... ... ... .. . . i 43 4.2 MSR Experimental Physics . . .. A 45 4.2.1 HTLTR Lattice Experiments . ......... ...ttt iieiiieaieieeenens 45 5. SYSTEMS AND COMPONENTS DEVELOPMENT ........... PN 49 5.1 Gaseous Fission Product Removal ................ e e e e e e e e 49 5.1.1 Gas Separator and Bubble Generator ....... ... ... ... ... . ... . . i 49 5.2 Gas System Test Facility ......... ... e 51 5.3 Molten-Salt Steam Generator . .... PR B 51 5.3.1 Steam Generator Industrial Program . ........ .. ... .. ool 52 5.3.2 Steam Generator Tube Test Stand (STTS) ... ... ... i i 52 5.3.3 Molten-Salt Steam Generator Technology Facility (SGTF) .......... .. ... ... ... ... 52 5.3.4 Development Bases for Steam Generators Using Molten Salt as the Heat Source .......... 52 5.4 Sodium Fluoroborate Test Loop .......... ... .. ... e 52 54.1 Pump Bowl and Rotary Element . ... ... . ... ... . .. . . 53 5.4.2 BF; Feed and Salt Level Bubbler Tube ... .. ... . . . ... . ... . i i, 55 5.5 -Coolant Salt Technology Facility ............ .. i i 57 56 MSBRPumps ............... .. .. ...... e e e e e e e 58 5.6.1 MSREMark2 Fuel Pump .. ... ... ... . . i ... 58 5.6.2 MSRE Salt Pump Inspection .......... .. .. . . e 58 5.6.3 ALPHA Pump ... ..o e e e 59 5.7 Remote Welding ... .... P 59 5.7.1 Automatic Controls . ... ... ... .t e e e 59 5.7.2 Pipe Cleaning Tests . . . .. ...ttt e 60 6. MSBR INSTRUMENTATION AND CONTROLS . ... .. e it ciee e ... 6l 6.1 Development of a Hybrid Computer Simulation Model of the MSBR System .................. 61 6.1.1 Introduction . ...... ... . . i e e 61 6.1.2 Steam Generator Model .. ... ... . . e 61 7. HEAT AND MASS TRANSFER AND THERMOPHYSICAL PROPERTIES ........................ 64 7.1 Heat Transfer . .. ... o 64 7.2 Thermophysical Properties .. ... ... ... 67 7.2.1 Wetting Studies ............ ... ... . ... ..., e S 67 7.2.2 Thermal Conductivity . ... ... ... .. e 69 7.3 Mass Transfer to Circulating Bubbles .. ... ... . .. ... . . .. .. . . 69 7.3.1 EXPeriment . .. ...ttt e e 69 7.32 Theory . ... e 70 8. FISSION PRODUCT BEHAVIOR .. ................ S 73 8.1 Determination of Tritium and Hydrogen Concentrations in MSRE ‘Pump Bowl! Gas ............. 73 8.1.1 Calibration Apparatus ..............uremimmneinnnannnnn. PP e 74 8.1.2 Analysis for Hydrogen .......................... e e e e 75 8.1.3 Tritium Diffusion-Studies . . .. ... .. e 76 8.2 Examination of Deposits from the Mist Shield in the MSRE Fuel Pump Bow! ......... e 76 B8.2.1 TritiUIN . vttt it et e e e e e e e e e e e 82 8.3 Synthesis of Niobium Fluorides . ............ ... ... .. ... ... e 85 84 Reaction Kinetics of Molybdenum and Niobium Fluoride in Molten Li, BeF, Solutions . ......... 85 8.5 Mass Spectroscopy of Niobium Fluorides . ... ... ... . i 86 9. COOLANT SALT CHEMISTRY AND TRITIUM CONTROL .. ... .. ... i 88 9.1 Studies of Hydrogen Evolution and Tritium Exchange in Fluoroborate Coolant . .. .............. 88 9.2 Reaction of Sodium Fluoride—Sodium Tetrafluoroborate with Water ........................ 90 9.3 Identification of Corrosion Products in the Bubbler Tube of the Fluoroborate Test Loop ... .. A 9.4 Mass Spectroscopy of Fluoroborate MSR Coolants . ...... ... .. .. ... ... ... . o, 93 9.5 Spectroscopic Investigations of Hydrogen- and Deuterium-Containing Impurities : in NaBF, and NaF-NaBF, Eutectics ... ... ... . i i i 94 9.6 Raman Spectra of the High-Temperature Phase of Polycrystalline NaBF, ..................... 96 | 9.7 A New Method for Synthesis of NaBF;0H ... ... . ... 98 9.8 Solubility of BF; Gasin Fluoride Melts . . .. ... ... ... . i i 98 9.9 Equilibrium Phase Relationships in the System RbF-RbBF, .. .. U 100 9.10 Activities in Alkali Fluoride — Fluoroborate Mixtures . . . ... ... ... .. ..o i ... 100 9.11 High-Temperature Crystal Structure and Volume of Sodium Tetrafluoroborate and Related Compounds . ...t e 101 9.12 Hydrogen Permeation through Oxide-Coated MELalS .« v vee e e e e e 103 10. PHYSICAL CHEMISTRY OF MOLTEN SALTS .. ... .. .. i PR 107 10.1 Thermodynamics of LiF-BeF, Mixtures from EMF Measurements of Concentration Cells .. ... ... 107 10.2 Equilibrium Phase Relationships in the System LiF-BeF,-CeF; ........... ... ... ... . 109 10.3 Electrical Conductivity of Molten and Supercooled NaF-BeF, (40-60!'M01e_ V) P 109 10.4 Glass Transition Temperatures in the NaF-BeF, System .............. N e 110 10.5 Raman Spectra of BeF, 2~ in Molten LiF and NaF to 686°C .. ......... ... . ... ........... 112 10.6 Bubble Formation by Impingement of a Jet Stream on a Fluid Surface .................... ... 115 107 The Solubility of Hydrogen in Molten Salt . . . . .. ... .. i, . 115 108 Enthalpy of UF, from 298 to 1400°K ... ... .. .....oiieeaianies, e 117 109 Absorption Spectroscopy of Molten Fluorides: The Disproportionation Equilibrium of UF3 Solutions .. ... i e 118 10.10 The Oxide Chemistry of Pa** in the MSBR Fuel Solvent Salt .. ... ........vuerinrenennnn... 119 10.11 The Redox Potential of Protactinium in MSBR Fuel Solvent Salt ................. e 120 PART 3. CHEMISTRY 11. 12. 13. 14. vi 10.12 The Crystal Structures of Complex Fluorides . . ... ... ... .. i e 122 10.12.1 The Crystal Structure of CsUgFa5 ... ..o 122 10.12.2 The Crystal Structure of a-KThgF,s . ... ..o ol 122 10.12.3 The Crystal Structure of Li,MoFg . .. ... ..o e 122 10.12.4 The Crystal Structure of RbTh3F,3 ... ... i I 124 10.12.5 The Crystal Structure of RbgZraFay ..o 125 10.12.6 DiSCUSSION ..t tv vt ee i ee et e e ee et e e e e, 125 10.13 Noncrystalline BeF, at 25°C: Structure and Vibrational Motion . ......................... .. 125 10.14 Relationship between Entropy and Sonic Velocity in Molten Salts .......................... 126 10.15 A Reference Electrode System for Use in Fluoride Melts . ............. ... ... .. ... .. .. .. 127 CHEMISTRY OF MOLTEN-SALT REACTOR FUEL TECHNOLOGY ...........iiiiiiiiiinannn, 129 11.1 Extraction of Rubidium and Cesium from MSBR Fuel Solvent into Bismuth by : Reduction with Lithiumat 650°C ... ... ... i, S 129 112 Distribution of Thorium between MSBR Fuel Solvent and Bismuth ' Saturated with Nickel and Thorium at 650°C ... ... ... ... . ... e, 130 11.3 Bismuth-Manganese Alloys as Extractants for Rare Earths from MSBR Fuel SOIVENt . . .ottt et e et e e et e e 131 11.4 Removal of Solutes from Bismuth by Fractional Crystallization ............................ 131 DEVELOPMENT AND EVALUATION OF ANALYTICAL METHODS FOR MOLTEN-SALT REACTORS ....... e e e e L. 134 12.1 Electroanalytical Studies of Titanium(IV) in Molten LiF-NaF-KF (46.5-11.542.0Mole 20) . ... oot e ettt eeeeans 134 12.2 Reference Electrode Studiesin Molten NaBF, . . ... ... . . i 135 12.3 Electrochemical-Spectral Studies of Molten Fluoride Salt Solutions ......................... 136 12.4 Spectral Studies of Molten Fluoride Salts . .. ......... ... ... . . . i 136 12.5 Analytical Studies of the NaBF, Coolant Salt . ....... ... ... ... . . . . . . . ... ... . ..., 137 12.6 In-Line Chemical Analysis .. ... ... ...ttt ittt ettt eaannns 138 PART 4. MATERIALS DEVELOPMENT EXAMINATION OF MSRE COMPONENT S . .. ..ot e et et e e .. 139 13.1 Examination of a Graphite Moderator Element ........ ... .. .. ... .. ... . ... . ... 139 13.2 Auger Analysis of the Surface Layer on Graphite Removed from the Core of the MSRE .. .. ... e e e 143 13.2.1 Auger Electron SpectrosCopY . ..ottt ittt ettt et 144 13.2.2 Results and Discussion .. ... ... ...ttt ittt et 145 " 13.3 Examination of Hastelloy N Control Rod Thimble .................cc.uuun... e 147 13.4 Examination of the Sampler Assembly . .. .. ... .. . 150 13.5 Examination of a Copper Sample Capsule ....... ... .. . . i 154 13.6 Examination of the Primary Heat Exchanger .......... ... ... .. . . .. 156 13.7 Examination of Freeze Valve 105 .. ... .. .. .. . i e 160 GRAPHITESTUDIES .................... e e PP 167 14.1 Graphite Irradiations in HFIR .. .. .. .. .. 167 15. 16. 17. vii 14.2 Graphite Fabrication ............ e e e e e 169 14.3 Graphite Development — Chemistry .. ... ot i 170 144 Graphitization Study of a Lampblack-Pitch Carbon .. ............. e 171 14.5 Reduction of Graphite Permeability by Pyrolytic Carbon Sealing ........... [ .. 173 14.6 Fundamental Studies of Radiation Damage Mechanisms in Graphite ......................... 174 14.7 Lattice Dynamics of Graphite ...........c.. i i it 176 HASTELLOY N o et e et ettt e 179 15.1 Status of Laboratory Heat Postirradiation Evaluation .............. ... .. ... ... i ... 179 15.2 Postirradiation Creep Testing of Hastelloy N ............ e e 180 15.3 The Unirradiated Mechanical Properties of Several Modified Commercial Alloys . ............... 181 154 The Weldability of Several Modified Commercial Alloys ........... ... i 185 15.5 Postirradiation Properties of Several Commercial Alloys ......... ... .. . oo 188 15.6 Status of Development of a Titanium-Modified Hastelloy N ......... e 192 157 COIToSion STUGIES . . o v ot et e e e ettt et ettt e e e e 192 15.7.0 Fuel Salts ..o oii ittt ettt e et sttt e e 194 15.7.2 Fertile-Fissile Salt . . . .. P T R RETRES 196 15.7.3 BlanKet Salt . .. v oottt it i e 196 1574 Coolant Salt .. ...t e EERTTER. 196 15.7.5 Analysis of H, O Impurities in Fluoroborate Salts ............ ... oo 197 15.8 Forced-Convection Loop Corrosion Studies ............... e 202 15.8.1 Operation of Forced-Convection Loop MSR-FCL-1 ....... ... ...t 202 15.8.2 Metallurgical Analysis of MSR-FCL-1 .. ... ... .. i 204 15.8.3 Forced-Convection Loop MSR-FCL-2 ........ e e 206 159 Retention of Tritium by Sodium Fluoroborate ....... ... ... i 210 15.10 Support for Components Development Program ........ ... ... oot 211 15.10.1 Metallurgical Examination of Inconel Bubbler Tube from PKP-1 Pump Loop .. ......... 211 15.11 Corrosion of Hastelloy Nin Steam . .. .. .ottt ittt et am e 216 SUPPORT FOR CHEMICAL PROCESSING . ... ... e 218 16.1 Construction of a Molybdenum Reductive-Extractive Test Stand . .......... ... .. .. ... ... 212§ 16.2 Fabrication Development of Molybdenum Components ............ ..ot 219 16.3 Welding Molybdenum . . ......... ..o .. e R 220 16.4 Development of Bismuth-Resistant Filler Metals for Brazing Molybdenum .................... 221 16.5 Compatibility of Materials with Bismuth ......... ... ... ... i 225 16.6 Chemically Vapor Deposited Coatings ........... oo, e 231 16.7 Molybdenum Deposition fromMoFg ... .. .o i 232 PART 5. MOLTEN-SALT PROCESSING AND PREPARATION FLOWSHEET AN ALY SIS ..ttt et ittt e it et 235 17.1 Protactinium Isolation Using Fluorination and Reductive Extraction ........................ 235 17.2 Combination of Discard Streams from the Protactinium Isolation System and the Metal Transfer System ........... ... i 237 18. 19. 20. viii 17.3 Protactinium Isolation Using Oxide Precipitation .. . ... e e e 237 17.4 Stripping of Rare-Earth Fission Products from LiCl in the Metal Transfer SyStem . . ..ottt et e e e e e e 240 17.5 Importance of Uranium Inventory in an MSBR Processing Plant . ........................... 240 PROCESSING CHEMISTRY . ... ..ot e e i e e 242 18.1 Measurement of Distribution Coefficients in Molten-Salt—Metal Systems ..................... 242 18.2 Solubilities of Thorium and Neodymiurh in Lithium-Bismuth Solutions ............. e L. 244 18.3 Oxide Precipitation Studies ............ ... ..o e 245 ENGINEERING DEVELOPMENT OF PROCESSING OPERATIONS ........ ... ... ... .. ... ..... . 249 19.1 Engineering Studies of the Metal Transfer Process for Rare-Earth Removal .................... 249 19.2 Design of the Third Metal Transfer Experiment .. .......... ... .. .. ... .., 254 19.3 Development of Mechanically Agitated Salt-Metal Contactors ............. ... ... . ... ... 255 19.4 - Reductive Extraction Engineering Studies .. ... e 256 19.5 Contactor Development: Pressure Drop, Holdup, and Flooding in Packed Columns ........... ... ... .. ... ... ... .. e e h s e e 260 19.6 Development of a Frozen-Wall Fluorinator ...... ... ... ... ... . . i i, 262 19.7 Estimated Corrosion Rates in Continuous Fluorinators . ......... ... ... ... ... ... ... .. ... 264 19.8 Axial Dispersion in Simulated Continuous Fluorinators . ....... PP 265 19.9 Engineering Studies of Uranium Removal by Oxide Precipitation ........................... 267 19.10 Design of a Processing Materials Test Stand and the Molybdenum Reductive Extraction Equipment ... .. ... ... . .. . . e 267 CONTINUOUS SALT PURIFICATION SYSTEM ... ................ JE 269 ORGANIZATIONAL CHART .. .. e i ettt et aaan 271 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 2331JF, dissolved in a salt that is a mixture of LiF and BeF,, but it could be started up with 23°U or plutonium. The fertile material would be ThF, dis- solved 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% ZrF,, 29% BeF,, and 65% " LiF; the uranium was about 33% 233U. 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. 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. ix 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 svstems for civilian power reactors. We concluded that this objective had been achieved and that the MSRE had shown that molten-fluoride reactors can be operated at 1200°F without corrosive attack on either the metal or graphite parts of the system, that the fuel is stable, that reactor equipment can operate satisfactorily at these conditions, that xenon can be removed rapidly from molten salts, and that, 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 23U was added 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 233U 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, on the chemistry of fuel and coolant salts, on fission product behavior, on processing methods, and on the development of com- ponents 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 tho- rium-bearing fertile salts. In late 1967, however, a one-fluid breeder became feasible because of the de- velopment 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. MOLTEN-SALT REACTOR EXPERIMENT 1. Postoperation Examinations A limited program of postoperation examinations was completed during this period. The work on the radio- active systems was done through the maintenance shield using tools specially developed and tested in mockups. The control rods, rod thimbles, and one graphite moderator element were removed, and the interior of the reactor vessel was viewed. The interior of the fuel pump bowl was also viewed through a hole left by excision of the sampler cage. A section of the primary heat exchanger shell was cut out, and portions of six tubes were removed. The salt leak that occurred during the final shutdown was located at a freeze valve and was cut out for inspection. The tools and procedures worked well, and conditions in the reactor were found to be generally very good. 2. Further Investigations The differential-pressure system on the coolant salt flowmeter was tested and found to be in error by 6.7%. Correction of all known errors brings the heat-balance value for full power down to 7.65 MW. Less uranium was recovered from the UF, absorbers than had been expected on the basis of salt inventory measurements at the time of the fuel processing in 1968. Careful review of all evidence indicates that about 2.6 kg of uranium (33% 23°U) was left some- where in the MSRE processing plant. A uranium search procedure based on neutron interrogation of closed vessels was tested but proved insufficiently sensitive. PART 2. MSBR DESIGN AND DEVELOPMENT 3. Design The comprehensive report on the conceptual design of a 1000-MW(e) single-fluid MSBR power station has received final editing and is scheduled for distribution in June 1971. | | Exploratory design and evaluation studies of a 300-MW(e) molten-salt demonstration reactor (MSDR) xi were continued. In these studies, the design conditions are made less stringent than those in our MSBR reference design. For example, the reactor would operate as a converter so as not to await demonstration ‘of the advanced processing system being designed for the breeder, and the power density would be reduced so that the graphite core would have a life of 30 years. The graphite would not have to be sealed to reduce’the permeation by xenon, and rapid fuel reprocessing would not be necessary. By substituting periodic salt replacement for continuous fuel processing the reactor could operate with a conversion ratio of about 0.8 until the chemical plant was fully developed. The MSDR reactor vessel design has not been revised since last reported, but the general flowsheet, the drain tank, the primary heat exchangers, steam generators and reheaters, and the cells and building have all been changed in an investigation of a salt-circulation loop interposed between the secondary system and the steam system or to otherwise modify the system parameters. The heat transport fluid used in the third loop would be a nitrate-nitrite mixture which would form water of any tritium diffusing into it from the coolant salt and would thus block escape of tritium into the steam system. Use of the nitrate-nitrite salt also makes it possible to construct ‘the steam generators and reheaters of less expensive materials and to deliver feedwater and reheat steam to the boilers and reheaters at conventional temperatures rather than at the abnormally high values specified for the MSBR. The fluid used to transport heat from the primary heat exchangers to the secondary exchangers would be changed from the previously proposed sodium fluoroborate to ’LiF-BeF,, a salt used successfully in the MSRE and one which poses few problems if it were to leak into the fuel salt. , In the MSDR the drain tank is not used as an overflow volume for the pump bowls, and only one ‘small pump is used to transfer salt from the tank when filling the primary system. The sump tanks of the main circulation pumps now provide the surge volume. A wvalve in which a thin film of salt is frozen between 2 movable poppet and the seat to effect the final leak-tight closure is proposed for use in the reactor drain line. The valve in the drain line from the cell catch basin would be sealed with a membrane that would be ruptured in the unlikely event that the line is needed. The primary and secondary heat exchangers were redesigned to use the new secondary and tertiary salts and to account for lowering the fuel salt temperature from the reactor outlet to 1250°F. As could be expected, the exchangers have more surface than those in previous concepts. The design of the heated equipment cells was changed to incorporate water cooling of the cell walls. The method of heating the cells was changed from use of radiant electrical heaters to circulation of hot nitrogen gas. The gas can also be cooled and used to cool some equipment. The temperature transients following shutdown of the reactor and draining of the primary and secondary salts were calculated for heat exchangers of the design proposed for the MSBR reference plant. Decay of fission products on the metal surfaces provides the heat source, and the heat is radiated to the surroundings, which are at 1000°F. It is conservatively estimated that the maximum temperatures would reach 2150 and 1850°F, respectively, in heat exchangers of 563- and 141-MW capacity at about 2.7 hr after shutdown. An analysis was started to determine the conse- quences of the mixing of primary and secondary salts that would result from the rupture of a tube in a primary heat exchanger of an MSBR. : Additional studies were made of methods for keeping small the amount of tritium that reaches the steam system in an MSBR plant; 0.2% or less would reach the steamn if essentially all the tritium and tritium fluoride were stripped from a side steam of 10% of the circulating fuel salt flow in a countercurrent contactor. Use of helium containing a small volume percentage of water vapor as the coolant in the secondary system in place of the sodium fluoroborate salt would inhibit the transport to the steam. Continuous addition and removal of hydrogen fluoride in the sodium fluoroborate in the secondary system would be effective in reducing the transport of tritium to the steam if the rate were more than 507 times the tritium production rate and the hydrogen fluoride did not react rapidly with the metal walls. The Ebasco Services group, consisting of Ebasco Services, Continental QOil, Babcock and Wilcox, Cabot, Union Carbide, and Byron-Jackson companies, was selected to perform the industrial design study of a 1000-MW(e) MSBR plant. A report was issued that outlined the objectives and design bases of the MSBE and provided a brief description of a reference reactor. Our reference reactor has a graphite core 45 in. in diameter and 57 in. high X11 containing 15 vol % salt. The core is centered in a 7.5-ft-ID spherical vessel, with the space between the graphite and the vessel wall filled with fuel salt. The design power is 150 MW(t), and the start of life breeding ratio is 0.96. , We continued design studies to determine the prob- lem areas and to evaluate possible solutions. Major emphasis was on maintenance of the core graphite and the primary heat exchanger. We also prepared new layouts of the cell and primary system to help indicate how the problems would be handled in the different configurations. Alternate core moderator element configurations were investigated. A cylindrical element design appears attractive except that it requires a 20.5% salt fraction, as compared with 15% for the prismatic element. Since the elements do not interlock, this concept lends itself to removal of individual elements by a handling machine. - We sized the primary heat exchanger, holding the tube length constant at the 28 ft proposed for the MSBR. This design, with salt on the tube side, utilizes 1340 tubes *% in. in diameter in a 31.5-in.-ID shell. 4. Reactor Physics Calculations of the neutronic performance of the reference single-fluid MSBR have been brought up to date by modifying fission product removal rates to conform to the recently adopted metal transfer process. In addition, a few minor data corrections and cross section revisions were included in these calculations. The results indicate a slightly higher breeding ratio for the new processing scheme (1.071 as compared with 1.063) and a very slightly lower fissile inventory (1487 kg as compared with 1504 kg). The most important differences in the neutron balance are the absence of - absorptions in plutonium, with the new process, re- duced absorptions in fission products, and a higher value of 7, because a higher proportion of the fissile material is 233 U, Studies of the possible performance of a 1000-MW(e) molten-salt converter reactor have been. continued. Some recent calculations were based on the core design of a “permanent-core” MSBR, that is, one whose peak" power density is low enough (i.e., 9 W/cm?) to permit the graphite to have a design life of 24 full-power years. In place of the breeder’s continuous, rapid chemical processing, however, we assumed the occasional discard of the carrier salt, with recovery and recycle only of the uranium in the salt. Batch cycles of six and eight full-power years and variations in salt composition were studied. Results indicate that the batch-cycle converter reactor should be operated with about the same salt composition as the breeder, for a given core design optimized for breeding. Reoptimization for different salt compositions might well reduce the apparent sensitivity of the reactor performance to salt composi- tion. The calculations indicate that a conversion ratio of 09 or higher, with a fuelcycle cost of 0.55 to 0.65 mill/kWhr(e), can be achieved with plutonium feed (60% 23°Pu, 24% 2*°Pu, 12% 2% 'Pu, 4% ***Pu) and a batch fuel replacement cycle of six full-power years. Reactor physics experiments with an MSBR lattice configuration have been initiated in the High-Tempera- ture Lattice Test Reactor at the Pacific Northwest Laboratory. These experiments include measurements of the neutron multiplication factor at temperatures from 300 to 1000°C, along with the reactivity effects of varying fuel density, of changing the lattice configu- ration, and of inserting various materials in the lattice, xiii including simulated control rods. Results of these measurements will be used to check the accuracy of nuclear data and computational models used in MSBR design studies. 5. Systems and Components Development The construction of the water test loop for testing MSBE-scale gas separators and bubble generators was completed, and operation was begun. The loop was operated primarily on demineralized water at liquid flow rates of 200 to 550 gpm and at gas flow rates of O to 2.2 scfm. The loop was also operated with water containing small amounts of n-butyl alcohol and so- dium oleate and with a 41.5% glycerin-water mixture which is hydraulically similar to fuel salt. The bubble generator operated satisfactorily, and the bubble sepa- rator . operated satisfactorily with water. An unex- plained reduction in bubble size was observed when the test fluid was changed from demineralized water to the other fluids, and the separator was unable to remove the small bubbles at the required rate. Tests were started to determine if the production of small bubbles was influenced by the pump efficiency or only the pump head. It is believed that the small bubbles are a characteristic of the test fluid and that they will not be produced in salt. The conceptual design of a molten-salt loop for testing gas systems was completed, and the conceptual system design description was written. Work is now beginning on the preliminary design. The facility will be used for developing the technology of the fuel salt for MSRs and in particular for tests of the bubble generator and separator. The facility is scheduled for initial operation in early FY 1973. | A program plan for obtaining reliable steam genera- tors for the MSRP was outlined, and activities in the first of three phases were started. We received affirma- tive responses to an inquiry of interest in participating in a conceptual design study from eight of nine industrial firms contacted, and we are proceeding to obtain proposals and to contract with one firm for the studies. The conceptual systems design description of the 3-MW test facility to be built as part of the molten-salt steam generator program was completed, and further work was suspended until late in FY 1972. Work was begun on preparation of a development basis report for molten-salt steam generators. In this report we expect to evaluate the elements of the LMFBR and other programs which have a bearing on the moltensalt technology and then to point out problem areas which need further study and outline a program for such studies. This report should be finished in the second quarter of FY 1972. Following completion of the fluoroborate test pro- gram in the PKP test loop, the salt pump rotary element, the bubbler tube for BF; feed and salt level indication, and other items of hardware were removed for examination. The appearance of the pump rotary element indicated that the fluoroborate service did not cause excessive corrosion damage to the Inconel system. Deposits of Na;CrFg and NaNiF; found in the bubbler tube were attributed to reaction of the fluoroborate salt with moisture introduced in the gas feed. A deposit of mefallic nickel which blocked the mouth of the bubbler tube was -probably formed by transfer of corrosion product nickel from the bulk salt. The condition of the gas pressure control valve was found to be like new. The conceptual system design description for the coolant salt technology facility was completed, and the detailed design of the facility and components was begun. We make maximum use of the drawings from the PKP-1 test stand and of the components and materials to be salvaged from the MSRE. The expected completion date of the facility is late December 1971. The salt was drained satisfactorily from the main loop of the MSRE Mark 2 pump test stand after replacing the plugged drain line which connects the loop piping to the storage tank. : The rotary element and pump tank of the MSRE coolant salt pump were inspected visually as the pump was removed from the coolant salt system. Except for evidence of leakage oil from the lower shaft seal on shield plug and tank surfaces, the pump appeared to be in very satisfactory condition. The water test program to qualify the ALPHA pump for the hydraulic conditions required in the MSR-FCL-2 test facility was concluded satisfactorily. The design and fabrication of the remaining parts needed for the pump for the facility were then completed. Most of the work previously supported by the MSRP remote welding program has been transferred into an automated welding program sponsored by the LMFBR program to meet their needs in reactor pipe construc- tion and maintenance. We are, however, completing a small program to develop and test weld-torch position- ing mechanisms and control circuitry; to define remote maintenance inspection, viewing, and alignment criteria; and to investigate pipe cleanliness requirements for maintenance welding in salt systems. Preliminary tests. have shown that small pipe filled with solid molten salt can be welded after a rather simple cleaning procedure and that these welds satisfy nuclear code x-ray inspec- tion standards. | 6. MSBR Instrumentation and Controls A hybrid computer simulation model of the reference 1000-MW(e) MSBR is being developed. The steam generator is modeled mathematically on the hybrid machine in continuous space and discrete time using sets of differential equations derived from the conserva- tion of momentum, energy, and mass. The integrations are performed by the analog computer, while the digital computer calculates the terms of the derivatives of the differential equations and provides storage and control for the calculations. The thermodynamic properties of water are stored in the digital computer as two- dimensional tables. The model of the reactor, primary heat exchanger, piping, etc., is a continuous-time model similar to those traditionally used on analog computers and is time scaled to 0.01 of real time. The discrete-time steam generator calculations are stored and sampled at 1-sec intervals, representing 0.01 sec in simulation time, then smoothed and applied to the continuous-time analog model. The hybrid program for the steam generator has been written and nearly debugged. The analog model has been developed, but has not yet been patched. Integra- tion of the two models will require some additional time, and the total simulation is expected to be in operation during the next reporting period. 7. Heat and Mass Transfer and Physical Properties Heat transfer studies using the inert-gas-pressurized flow system have shown that the average heat transfer coefficient for a proposed MSBR fuel salt at 1070°F and a Reynolds modulus of 3300 is 15% higher with a Xiv hydrodynamic entrance length than without. These results are difficult to explain in terms of commonly accepted theories of the combined development of hydrodynamic and thermal boundary layers, but can be explained by a recent theory which suggests that the flow of a fluid whose viscosity has a large negative temperature dependence will be stabilized by heating. The present results tend to substantiate this theory. A new technique for determining the wetting charac- teristics of liquids was used to study wetting behavior of the molten salt LiF-BeF,-ZrF,-ThF,-UF; (70-23- 5-1-1 mole %) on a Hastelloy N surface at 700°C (1292°F). It was found that the typical nonwetting condition of this salt could be changed to a wetting condition within several minutes by the introduction of a zirconium rod. Several hours were required to change from the wetting back to the nonwetting condition by the addition of 1 wt % nickel fluoride. Water calibration of an improved variable-gap thermal conductivity apparatus designed for use with molten salts at 830°C gave results in excellent agreement with the specialized room-temperature measurements pub-. lished in the literature. Thermal conductivity measure- .. ments are being made for the molten fluoride salt system LiF-BeF,. g , , MSBR-related mass transfer experiments involving diffusion of oxygen dissolved in glycerin-water solu- tions into helium bubbles have been extended to include the case of horizontal flow. The volume fraction of bubbles, which is needed to determine the interfacial area per unit volume, was found to correlate with the ratio of axial to thermal velocity of bubble rise. By making use of the interfacial area derived from the bubble volume fraction correlation, overall mass transfer coefficients (including the separator) have been extracted from the measured concentration decay rate of dissolved oxygen for a Reynolds modulus range from 26,000 to 66,000 at one value of the Schmidt modulus, 1228. When these recent results are compared with " earlier results for vertical flow, it is found that, at a sufficiently high Reynolds modulus for gravitational forces.to be negligible compared with inertial forces, the mass transfer coefficients for vertical and horizontal flow become identical. Mass transfer coefficients for the horizontal test section, corrected by subtracting the measured mass transfer which occurs in the separator, are correlated with Reynolds modulus for three values of mean bubble diameter. A theoretical description of mass transfer from a turbulent liquid to bubbles moving at the local liquid velocity has been developed which includes variation in eddy diffusivity with concentration gradient near a bubble interface, velocity, and frequency components of the turbulence. A computer program has been written to solve the pertinent equations. PART 3. CHEMISTRY 8. Fission Product Behavior Laboratory apparatus for determination of the con- centration of tritium and hydrogen concentrations in the MSRE fuel pump bowl gas was constructed and calibrated. Examination of the sampler cage and mist shield excised in January 1971 from the MSRE pump bowl revealed that all surfaces were covered with a deposited film, generally gray-black, over unattacked metal. These films contained carbon, lithium and beryllium fluorides (and, doubtless, uranium and zirconium fluorides), structural metals, fission products dominated by noble metal isotopes and '37Cs, and tritium. Much, but probably not all, of the structural metals could be debris from cutting operations. Tritium/carbon atom ratios of ~1 X 107* are consistent with nontrivial exchange with hydrogen in tars resulting from the cracking of lubricating oil which leaked into the pump bowl. A very high proportion, relative to inventory, was observed for all noble metal fission isotopes: °°Nb, 99Tc, '93Ru, '%%Ru, 12°8b, and 27" Te. The relative proportions of '®*Ru and '®°Ru isotopes and the quantities of deposit suggest that these deposits were accumulated over a long period and that several percent of the reactor inventory of noble metal isotopes is deposited in the pump bowl. Thickest deposits were noted on the sample cage rods and inner mist shield walls below the surface, suggesting the steady agglomer- ation and deposition of suspended material in the less-agitated regions. Attempts to synthesize niobium fluorides of lower oxidation numbers provided additional quantitative information concerning their disproportionation and fractional sublimation. Studies of the stability of molybdenum and niobium fluorides in molten Li, BeF, were extended to temperatures as high as 700 and 900°C respectively. Evidence is presented which indi- cates that the mechanism of removal of Mo from molten Li,BeF; changes as temperatures exceed 500°C. Both increased nobility of niobium metal and instability of NbF, solutions were observed in the 700-t0-900°C range as contrasted with the range 500 to 700°C. Mass spectroscopy of the niobium fluorides yielded new information on the pentafluoride poly- mers, the fluorination of niobium metal, the dispropor- tionation of NbF,, and the behavior of the associated oxy fluorides. XV 9. Coolant Salt Chemistry and Tritium Control The results of experiments performed with nickel as the only metal in contact with molten sodium fluoro- borate indicated that adequate but small concentrations of chemically bound hydrogen can be retained in the salt as a potential means for controlling the distribution of tritium in molten-salt reactors. Preliminary values for the solubility of HF in the NaF-NaBF, eutectic mixture were measured in the temperature ra‘nge 400 to 600°C. Analysis of materials removed from an engineering test loop showed that the corrosion experienced during operation originated with the introduction of water vapor as a contaminant in the cover gas. Mass spectroscopic studies of the vapors over molten fluoroborate melts were initiated, examining first the effects of NaOH and moisture as impurities in fluoro- borate systems. Evidence for the existence of hydrogen-containing impurities in NaBF,; was obtained from near-infrared spectra of the molten salt and from mid-infrared spectra of pressed pellets of the pure material. Raman spectra were measured with polycrystalline NaBF, at just above and below the dimorphic crystal transition temperature. The results rationalize anom- alies in x-ray diffraction data for NaBF, and related structures. Laboratory studies were continued in attempts to synthesize hydrogen-containing species for the reten- tion of tritium in molten fluoroborates. The solubility of BF; was determined in five molten- salt solvents composed of LiF and BeF,. For concentra- tions of BF; in the melt below 1 mole %, Henry’s law is obeyed. An investigation of the equilibrium phase diagram of the system RbF-RbBF,; was completed. Component activities at the liquidus in alkali fluoride— fluoroborate systems (Na, K, Rb) were computed and compared with the ideal liquidus. In each case the liquidus shows positive deviations at low concentration of fluoroborates but exhibits negative deviation as it approaches the eutectic. The relative volume increase of the alkali fluorobo- rates and perchlorates and alkaline-earth sulfates were correlated for a range of temperatures up to and through that of the dimorphic transition as a function of cation size. An experimental program was initiated to study hydrogen permeation of metals under conditions closely analogous to those which might be expected in a molten-salt reactor steam generator heat exchanger. 10. Physical Chemistry of Molten Salts Excess chemical potentials and partial molar enthal- pies of mixing in molten LiF-BeF, were derived from emf measurements using concentration cells. An investigation of the equilibrium phase relation- ships in the system LiF-BeF,-CeF; was continued. Preliminary electrical conductance measuréments of supercooled NaF-BeF, melts were obtained with an all-metal conductance cell. Glass transition tempera- tures were determined in the glass-forming region of the NaF-BeF, system to permit comparison with theo- retical glass transition temperatures obtained from the temperature dependence of activation energies of elec- trical conductance in BeF, -containing systems. The Raman spectrum of BeF,?” was measured in melts containing excess F~ ion; the results indicated that in the molten mixtures the BeF4?™ anion retained tetrahedral symmetry under a wide variety of composi- tions and temperatures as in crystalline Li; BeF,. Parameters affecting bubble formation were studied experimentally by impingement of a liquid jet on a fluid surface. Apparatus was constructed and and used in prelimi- nary measurements of the solubilities of helium and hydrogen in molten salt. The enthalpy of UF, was measured from room temperature to temperatures above the melting point. In contrast with recent reports in the Soviet literature, the compound was found not to be dimorphic. Factors affecting the stability of uranium trifluoride as a dilute species contained in graphite were examined. The stability, as measured using absorption spectros- copy, was found to be less than predicted; stability was also found to exhibit a marked solvent composition dependence. Because of the promising simplicity and efficiency of a protactinium removal process based on its precipita- tion as Pa,Og, work has continued on the oxide chemistry of protactinium in molten fluorides. The solubility of a Pa**-oxide phase was found to be approximately four times higher than that of Pa;Os. Together with measurements of the redox potential of the couple Pa**/Pa®*, this leads to the prediction that Pa, O; will not precipitate in the core of an MSCR or MSBR in the event that accidental contamination with oxide occurs. The crystal structure of several complex compounds of the alkali fluorides and heavy metal fluorides were determined employing automated collection of X-ray diffraction data from single crystals. Structural charac- teristics of CsUgF,5, «KThgF,5, Li;MoFg, RbTh;F,;,and RbsZr, F,, were established. Xvi A program of investigation of molten salts by x-ray and neutron diffraction was resumed; initial studies were devoted to fluoride glasses and SiO, . The characteristic velocity at which a disturbance (e.g., a thermal spike) is propagated in a fluid (molten salts included) is approximately equal to the sonic velocity in that fluid. A new method, based on the Debye theory of specific heats, was developed which allows sonic velocity to be estimated reliably from the entropy (either measured or estimated) of molten salts. In continuing efforts to develop an Ni-NiF, reference electrode for use in fluoride melts, it was shown that beryllium is very slightly soluble in LiF-BeF, melts and that conductance in LaF; crystals in such melts has a small electronic component. 11. Chemistry of Molten-Salt Reprocessing Technology Distributions of rubidium and cesium in the fluo- ride/bismuth extraction at 650°C were measured; the order of extractability of the alkali metals from the salt was shown to be Cs, Na, Rb, K. The distribution of thorium between MSBR fuel solvent and bismuth saturated with nickel and thorium at 650°C was measured; the results indicated that the distribution of soluble thorium was unaffected by the presence of ThNiBi, in the bismuth phase. The potential application of bismuth-manganese al- loys as extractants for rare earths from MSBR fuel solvent was assessed experimentally. The results were sufficiently encouraging to warrant further investi- gation. The removal of solutes from bismuth by fractional crystallization was examined in attempts to rationalize the behavior occasionally observed in metal samples removed directly from liquid bismuth mixtures. 12. Development and Evaluation of Analytical Methods for Molten-Salt Reactors An investigation of the electroanalytical behavior of titanium (a constituent of Hastelloy N) in several molten fluoride salts was initiated. Initial observations in LiF-NaF-KF show that Ti(IV) is reduced reversibly to Ti(lll) and, at a more cathodic potential, to the metal. Evidence also suggests that Ti(IV) is reduced to Ti(Ill) by graphite or nickel. The behavior of titanium in Li, BeF, and NaBF, is presently being studied. The Ni/NiF, (LaF3) reference electrode, which performs satisfactorily in MSRE-type salts, has been found unsuitable for molten NaBF, . This appears to be due to the low solubility of Ni(II) in NaBF,. Other electro- chemical couples are being investigated for use .as a reference system for molten NaBF, . In an investigation of the electrochemical generation and spectral characterization of solute species in molten fluorides, Mn(IIT), Cu(Il), Co(III), U(III), and CrO,;*" were generated electrolytically. Also, superoxide ion was spectrally characterized in fluoride melts. A con- tinuing study has shown that several other oxygenated species, including VO,%, NO,~, and —OH can be observed in coolant and/or fuel melts. These observa- tions have led to a cooperative spectral study of —OH and —OD in NaBF,. The generation of CrO,* is of particular interest because it offers a means for the spectrophotometric determination of traces of chro- mium and possibly oxide in molten fluorides. In LiF-BeF, and NaBF,, hexavalent chromium was found to exist as Cr,0,2". The generation of Cr,0,% by chemical oxidation of CrF, has been demonstrated in LiF-BeF, melts but has not yet been achieved in NaBF, . By careful purification of NaBF,, melts transparent to wavelengths as short as 200 nm were obtained. Absorbance measurements below 300 nm were found to provide a sensitive method for the detection of iron as well as other impurities. An evaluation of the azeotropic distillation method for the separation of water from NaBF, showed that significant positive errors can be introduced by high background (blank) titrations and possible contamina- tion during sample additions. With an improved proce- dure, previously undetectable concentrations (<50 ppm) of distillable water were found in samples from coolant salt test loops. The azeotropic distillation method is under development for measurement of hydrolysis products in coolant cover gas as well as for salt analysis. ‘ Components for the first in-line applications of electroanalytical methods to fuel salt are now being tested. Voltammetry will be used to establish the U(IIT)/U(IV) ratio and to determine the concentrations of certain corrosion products in the flowing fuel in a thermal-convection loop. A reference electrode system will be used for continuous measurement of the redox potential. A PDP-8/I computer coupled with a voltam- meter of improved design will be used for the unat- tended measurement of U(III)/U(IV) ratios. Before installation, this system will be tested on a melt in a grounded cell. PART 4. MATERIALS DEVELOPMENT 13. Examination of MSRE Components Several components from the MSRE primary circuit were examined. The graphite moderator element was in xvii excellent condition, with no detectable dimensional changes. There was no metallographic evidence of corrosion. There was a shallow surface layer about 2 mils deep having a modified structure that we attribute to surface working. All metal surfaces exposed to the fuel salt were embrittled to a depth of 5 to 10 mils as evidenced by grain boundary cracking to a depth of 5 to 10 mils. The failure was located in freeze valve 105 and was attributed to thermal fatigue. A copper sample capsule was retrieved and found to be extremely brittle; all surfaces were coated inhomogeneously with Ni, Fe, Cr, and Mo. 14. Graphite Studies The routine evaluation of vendor-furnished graphites has been essentially completed, and the irradiation program is rapidly shifting to the investigation of experimental graphites exhibiting the desired structural variations. In support of this, the fabrication effort on graphite has expanded. In the study of precursor chemistry, the effect of hetero atoms (N, S, O) is being investigated. These materials lead to a deterioration in crystal perfection, but also lead to the desired iso- tropicity of the cokes and graphites. Raw cokes and blacks are currently being fabricated into isotropic graphites with the required monolithic structure. This work is still in its early stages, but the fabricated bodies look most promising. The black-based graphites discussed in the previous semiannual report continue to be irradiated. Heat- treatment series have been prepared and characterized. At the level of fundamental understanding, the distortions around an interstitial cluster have been calculated from elastic continuum theory, and the phonon dispersion curves have been measured by neutron scattering. 15. Hastelloy N Postirradiation tests on laboratory melts of modified Hastelloy N have shown that acceptable properties can be obtained with additions of 1.5 to 2.5% Ti, at least 0.5% Hf, and multiple additions of Ti, Hf, and Nb. Small commercial melts have confirmed the beneficial effects of Ti and Nb but have not reproduced the beneficial effects of Hf. The creep strength of all of the modified alloys is higher than that of standard Hastel- loy N. Weld metal cracking was encountered with high Hf and Zr concentrations, but suitable welds were made using standard Hastelloy N as a filler metal. The evidence is quite encouraging that an alloy with 1.5 to 2.5% Ti will have adequate postirradiation properties. Corrosion tests of Hastelloy N in LiF, BeF,, UF,, ThF, salts continue to give very low corrosion rates, <0.1 mil/year. Tests in sodium fluoroborate now involve four thermal convection loops and two pump loops. The corrosion rates vary from 0.1 to several mils/year, depending on salt purity. Evidence indicates that the impurities most affecting the corrosion are water and oxygen. A test to study tritium retention in sodium fluoroborate by exchange with H showed that the tritium was released quickly by the salt, indicating a lack of stable hydrogen in the salt. Hastelloy N specimens exposed to steam at 538°C continue to show a metal loss rate of <0.25 mil/year. 16. Support for Chemical Processing We have begun fabrication of components for a molybdenum test stand. A 5-ftlong, 1'%-in.-diam by 0.080-in.-wall-thickness section of molybdenum piping was fabricated by forward extrusion. Several 37%-in.- diam by 0.125-in.-wall closed-end, cylinders were back extruded and machined for welding studies. Procedures have been developed for electron beam welding a girth joint between two of the 3%-in.-OD molybdenum headers and for electron beam welding tube-to-header joints between Y%-, Y%-, %-, %-, and 1%-in.-OD tubes and the 37%-in.-OD headers. For field welding molyb- denum we have modified a commercial orbiting-arc welding head and developed procedures for making tube-to-tube welds. We continued studies on three iron-base brazing alloys for molybdenum (Fe—15% Mo—5% Ge—4% C—1% B, Fe—15% Mo—4% C—1% B, and Fe-25% Mo—4% C—1% B). We determined the optimum clearance for back brazing welded and me- chanical joints in the molybdenum test stand and also found that the braze alloys could be used to repair cracked welds. The shear strength and percent elonga- tion were measured for the Fe—15% Mo—5% Ge—4% C—1% B and Fe—15% Mo—4% C—1% B compositions at both room temperature (30,000 psi and 10%, respec- tively) and at 650°C (18,000—29,000 psi and 40% respectively). Molybdenum metal seal couplings were procured and evaluated by helium leak checking before and after thermally cycling the joints from room temperature to 650°C. Two techniques, magneforming and roll bonding, have been investigated for mechani- cally joining molybdenum. Several helium-leak-tight joints (<5 X 107% std cc/sec) have been made by roll bonding. We continued to evaluate the compatibility of several - container materials and brazing alloys in pure bismuth and bismuth-lithium. solutions. In quartz .thermal- convection loop tests (700°C maximum temperature, Xviii 100°C temperature difference) three different grades of graphite showed no significant attack except that the more open grades contained bismuth in their pores. Four iron-base filler metals (Fe-C-B) that contained O, 15, or 25% Mo, and in one case 5% Ge, were also tested in pure bismuth under similar conditions. No increase in iron concentration of the bismuth was found, but the alloys picked up significant amounts of bismuth and exhibited surface layers that were high in iron. These alloys showed no attack when exposed to an LiF-BeF, - ZrF,-UF,;-ThF, mixture for 1000 hr in a 304L stainless steel thermal-convection loop operating at a maximum temperature of 690 and a 100°C temperature difference. A quartz thermal-convection loop contain- ing molybdenum and TZM samples was operated for 3000 hr with a Bi—100 ppm Li solution (700°C maximum temperature and 100°C temperature dif- ference), and only very slight attack of the very fine grains along the surface was found. We have continued to coat stainless steels and nickel-base alloys with tungsten and molybdenum by hydrogen reduction of WF, and MoF. Both coatings were found to be adherent to the base materials during thermal cycling between 25 and 600°C, and tungsten coatings exhibited tensile bond strengths of 20,000 to 35,000 psi. Objects of various size and shape were coated with tungsten to demonstrate the applicability of the process. We have found that we can coat cylindrical shapes of up to 4 in. in diameter and 48 in. long with little difficulty. The optimum conditions for applying smooth, adherent molybdenum coatings were found to be about 900°C and an H, /MoF ratio of 3 to 6. Bend tests have shown that the molybdenum coatings are more ductile than the tungsten coatings. Studies have also continued to develop a technique for depositing molybdenum on iron-base substrates from MoFg in a molten fluoride salt mixture. Corrosion reactions resulting from too high a concentration of MoF¢ in the salt have limited the success of experi- ments conducted thus far. PART 5. MOLTEN-SALT PROCESSING AND PREPARATION 17. Flowsheet Analysis An improved flowsheet was developed in which protactinium is isolated from the fuel salt of an MSBR and held for decay in a secondary salt stream that is physically and chemically isolated from the reactor. A processing plant based on this flowsheet should be much easier to control than one based on the earlier. flowsheet, and a considerable saving in capital equip-. ment cost should result. A method for combining and fluorinating the various waste streams from the metal transfer process and the protactinium isolation system was developed. This will eliminate several potential routes for loss of fissile material from the system. Oxide precipitation is being considered as an alterna- tive to the fluorination—reductive-extraction method for isolating protactinium and for subsequently re- moving uranium from the fuel salt of an MSBR. Two possible flowsheets -based on oxide precipitation were developed, and the effects of several parameters on operation of the processes were investigated. In the first flowsheet, the isolated protactinium is held in a secondary salt from which the uranium is removed by fluorination. In the second flowsheet, the isolated protactinium is dissolved in processed fuel carrier salt from the metal transfer process, and the resulting stream is recycled through a protactinium decay tank to the protactinium oxide precipitator. The precipitator efficiency required for these flowsheets is 60 to 80% for the first flowsheet and about 96% for the second. Although the MSBR processing flowsheets considered thus far have uniformly resulted in very low uranium inventories in the processing plant, several potential processing systems may result in uranium inventories as large as 5 to 10% of the reactor inventory. A processing plant uranium inventory of 5% of the system fissile inventory would increase the fuel cycle cost by about 0.015 mill/kWhr and would increase the system dou- bling time from 22 to 23.1 years. While there is incentive for maintaining a low uranium inventory in the processing plant, it does not appear that a uranium inventory as high as 5 to 10% of the system fissile inventory would rule out an otherwise attractive proc- essing system. 18. Processing Chemistry Studies in support of the development of the metal transfer process for removing rare-earth and other fission products from MSBR fuels were continued. The distri- bution of several actinide elements between molten LiCl and liquid bismuth solution was determined at 640 and 700°C. Additional information was obtained at 640°C on the effect of LiF concentration on the distribution of several solutes between LiCl-LiF solutions and liquid bismuth. This information confirmed earlier indications that LiF in concentrations of less than about 4 mole % has little effect on the behavior of di- and trivalent solutes but that the thorium—rare-earth separation factor decreases with increasing LiF concentration. Xix Measurements of the solubilities, both individual and mutual, of thorium and rare earths in lithium-bismuth solutions were also continued. At each temperature in the range 400 to 700°C, the solubilities of both thorium and neodymium increased as the lithium concentration in the solution increased from 0 to 25 at. %. The mutual solubilities of thorium and neodymium were measured in lithijum-bismuth solutions at 640°C and appeared to be much higher than those required in the stripping of rare earths from the LiCl acceptor salt into lithium-bismuth solutions. Investigation of oxide precipitation as a means for isolating protactinium from MSBR fuel salt was con- tinued. Two methods were used to estimate the solubilities, at various temperatures, of Pa, Qs in mol- ten LiF-BeF,-ThF,-UF, (71.8-16-12-0.2 mole %) that was saturated with UO,. These solubilities define the lowest protactinium concentrations attainable without attendant precipitation of uranium and thorium oxides. 19. Engineering Development of Processing Operations The second engineering experiment on the metal transfer process for the removal of rare earths from single-fluid MSBR fuel salt has been completed. The experiment operated satisfactorily for about three months. before it was shut down for inspection. During that period, more than 85% of the lanthanum and more than 50% of the neodymium originally in the fuel carrier salt were removed and deposited in a lithium- bismuth solution. There was no measurable accumula- tion of thorium (<10 ppm) in the lithium-bismuth solution, thus demonstrating that the rare earths can be - removed without significant removal of thorium. The thorium-lanthanum decontamination factor was about 10%. The distribution coefficients for lanthanum and neodymium between the fluoride salt and the thorium- saturated bismuth were relatively constant and in agreement with expected values. The distribution coef- ficients for lanthanum and neodymium between LiCl - and thorium-saturated bismuth were somewhat higher than expected. When the vessel was disassembled for in- spection after the run, the condition of the interior of the vessel was generally good; however, some corrosion had occurred on the components made of thin carbon steel (the lithium-bismuth container and the sparge tubes). The third engineering experiment for development of the metal transfer process is presently being designed. This experiment will use flow rates that are 1% of the estimated flow rates for a 1000-MW(e) reactor. Mechan- ical agitators will be used to promote efficient contact- ing of the salt and metal phases. The three vessels required for the experiment will be made of carbon steel; one of the vessels has already been fabricated. A test vessel has been fabricated that will be used with salt and bismuth for testing the proposed agitator drive unit, the shaft seal, the performance of a vapor- deposited layer of tungsten on the vessel interior, and a new technique for applying nickel aluminide coatings to the exterior of carbon-steel vessels. A program has been initiated for the development of mechanically agitated salt-metal contactors as an alter- native to packed columns presently considered for the MSBR processing system. Studies to date have been concerned primarily with selection of a contactor design for the third metal transfer experiment. Tests with mercury and water using a four-bladed flat-paddle agitator located in the mercury-water interface fre- quently resulted in a stable dispersion of very small mercury droplets and entrainment of water in the mercury. We have begun studies of a contactor which has a paddle operating in each phase, well away from the interface, at a speed that does not result in the dispersion of either phase. We have continued to study the extraction of uranium from molten salt by countercurrent contact with bismuth containing reductant in a packed column. Two successful runs, which included six periods of steady-state operation covering metal-to-salt flow ratios ranging from 0.75 to 2.05, have been made. The data could be correlated by an HTU model based on the assumption that the uranium transfer rate was con- trolled entirely by the diffusive resistance in the salt phase. The observed HTU values ranged from 0.77 to 2.1 ft and were inversely proportional to the metal-to- sait flow rate ratio. Flooding data obtained during countercurrent flow of salt and bismuth in the packed column have continued to show good agreement with predictions based on studies with a mercury-water system. . Preparations were begun for mass transfer experi- ments in which the rate of exchange of zirconium isotopes will be measured between salt and bismuth phases otherwise at equilibrium in a packed column. The first experiment of. this type had to be terminated because of a salt leak, which necessitated replacement of the salt feed-and-catch tank. Studies of hydrodynamics in packed columns during the countercurrent flow of high-density liquids are being made in order to evaluate and design contactors for processing systems based on reductive extraction. XX Data obtained by an MIT Practice School group show that the slip velocity with nonwetted packing is dependent on the —0.167 power of the continuous- phase viscosity rather than being independent as was previously assumed. Knowledge of the dependence of slip velocity on the continuous-phase viscosity allowed calculation of the power dependence of slip velocity on the difference in densities of the phases. A power depen- dence of 0.5 was calculated rather than the previously assumed dependence of 1.0. Studies with packing wetted by the metal phase indicate a substantial reduction in interfacial area between the phases and an increase in the slip velocity. Twentyseven runs were made with the simulated continuous fluorinator to determine heat generation rates in a column- of nitric acid (used as a stand-in for molten salt), in the pipe surrounding the acid column (representing the fluorinator wall), and in induction- coils of three different designs. Air was bubbled through the nitric acid at rates up to 2.16 scfh, resulting in bubble volume fractions as high as 18%, to determine the effect of bubbles on the heat generation rate in the acid. Efficiencies predicted for heating molten salt in a fluorinator using the best coil design tested to date ranged from 37.0% with no bubbles to 32.9% with a bubble fraction of 15%. These efficiencies are suffi- ciently high to allow operation of a fluorinator having a 4.5-in.-diam molten zone and a 1.5-in.-thick frozen salt layer by using an available rf generator. Rate constants for the corrosion of Ni-200 and Ni-201 in gaseous fluorine at l-atm pressure were calculated from literature data, assuming that the reaction follows a parabolic rate law; the calculated rate constants were correlated with temperature, assuming an Arrhenius temperature dependence. The resulting data were then used to estimate average corrosion rates when the NiF, protective film is periodically destroyed. Average corrosion rates of 2.9 and 0.97 mils/year for Ni-200 and Ni-201, respectively, were estimated at 450°C (a typical fluorinator wall temperature) when the NiF, film was assumed to be destroyed 52 times annually. These corrosion rates are acceptable for long-term fluorinator operation. Studies of the effect of column diameter on axial dispersion in open bubble columns were extended to include a 6-in.-diam column, and additional data were obtained on the effect of the viscosity of the liquid. The dispersion coefficient values obtained with a 6-in.-diam, 72-in.-long column were about three times the values measured at the same superficial gas velocity in a 3-in.-diam column; the data show little dependence on superficial gas velocity. The dispersion coefficient in a 6-dn.-diam column is not noticeably affected by changing the viscosity of the liquid from 0.9 to 12 cP. The same change in viscosity in a 2-in.-diam column results in a 20% decrease in dispersion coefficient. However, since the diameter of continuous fluorinators will be 6 in. or larger, the viscosity of the liquid will not affect the dispersion coefficient. An engineering-scale experiment to study the precipi- tation of uranium oxide from MSBR fuel salt, from which the protactinium has been previously removed, is being designed. In this experiment, uranium will be precipitated from 2 liters of fuel salt in a single-stage batch operation by contact with an argon-steam mix- ture. The design of the precipitator vessel is described. Very little additional design work on the processing materials test stand and the molybdenum reductive- extraction equipment has been done pending develop- ment of molybdenum fabrication techniques. The X X1 plastic model of the head pot was tested and modified, an improved design is ready for final testing. 20. Continuous Salt Purification System To date, 11 iron fluoride reduction runs have been carried out in order to study the countercurrent contact of molten salt with hydrogen in a packed column. In the first three runs, iron fluoride mass transfer coeffi- cients averaged 0.016 ft/hr. Coefficients for the suc- ceeding runs are questionable because of inconsistent iron analyses. Oxide was removed from the salt by contacting the salt with H,-HF, first by countercurrent contact in the column and second by bubbling the gas through a static column of salt. Corrective measures taken to alleviate the increased pressure drop in the packed column are also described. Part 1. Molten-Salt Reactor Experiment P. N. Haubenreich The postoperation examinations were completed during this report period, and the plant was secured to await ultimate disposal of the salt. This section outlines the examination program, describes the on-site work, and presents some of the results. Results of some of the detailed analyses made on items removed from the MSRE are described in Parts 3 and 4 of this report. 1. Postoperation Examinations R. H. Guymon The purpose of the postoperation examinations was to round out the MSRE experimental program by providing information not available during operation, such as the condition of materials throughout the system (for comparison with specimens from the core), the condition of key components, the location and nature of deposits in the pump bowl, and the exact location and nature of a leak in the fuel salt drain line. 1.1 OUTLINE OF PROGRAM Within a few weeks after the end of nuclear opera- tions in December 1969, the salt was allowed to freeze in the drain tanks, and other system conditions were established so that routine surveillance could be main- tained from the ORNL Central Waste Monitoring Facility.! The operating crews were disbanded, and the maintenance planning staff (one engineer and two technicians) began development of the special tools and detailed procedures that would be required for the examination tasks that had been selected.? Specimens of the coolant piping and radiator tubes were cut out in July 1970.> but it was mid-October before craft support was available for the final testing of the tools for the fuel system. ' 1. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp. 24 -25. 2. P. N. Haubenreich and M. Richardson, Plans for Post- Operation Examination of the MSRE, ORNL-TM-2974 (April 1970). 3. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL+4622, pp. 119-33. P. N. Haubenreich The special tools included abrasive cutters with vertical and horizontal wheels and a hydraulically operated shear. These were intended for use on small piping, and tests consisted simply of cutting pieces of pipe in the maintenance practice cell, with the tools operated from overhead through the portable mainte- nance shield. The job of cutting into the pump bowl was more complicated, and the preparations included moving the MSRE prototype pump from the pump development facility into the practice cell and going through the operations of cutting away interfering structure and attaching the bowl-cutting tool. This tool was tested by cutting inclined ' -in. plates simulating the top of the pump bowl. Preparations for the heat exchanger work involved development by welding specialists " of techniques for remotely operating a plasma torch for cutting the shell and a Heliarc torch for welding on a patch. The tools and fixtures for this work were tested on a partial mockup of the heat exchanger in the maintenance practice cell at the MSRE. Final preparations were completed by November 23, and on that date the reactor cell was opened to begin work on the radioactive systems. The main line of activities, a sequence of operations involving use of the portable maintenance shield, is outlined in Fig. 1.1. Concurrently the coolant pump and the 5-in. coolant piping outside the reactor cell were cut out and removed for use in MSR technology development facilities. Descriptions of the major tasks are given in the sections which follow. " ORNL- DWG T1-6975 OPEN REACTOR CELL REMOVE CONTROL ROD DRIVES, RODS REMOVE STANDPIPE REMOVE CORE ACCESS PLUG AND THIMBLES } CORE VIEW INSIDE REACTOR VESSEL REMOVE CORE GRAPHITE ELEMENT CLOSE REACTOR VESSEL TEST REMOTE MAINTENANCE PROVISIONS CUT SAMPLER TUBE DEC 1970 FUEL \ A CUT AROUND SAMPLER CAGE _ " PUMP REMOVE SAMPLER CAGE VIEW INSIDE PUMP BOWL RETRIEVE CAPSULE PATCH PUMP BOWL 5 3 P-d g CUT HOLE IN HX SHELL | HEAT REMOVE & TUBES EXCHANGER WELD PATCH ON HX SHELL REMOVE THERMAL SHIELD SLIDE } 5%’5‘&8'* OBTAIN INSULATION SAMPLE INSULATION SEAL HEAT EXCHANGER PATCH FIND LEAKED SALT CUT DRAIN LINES - LEAK DELIVER FREEZE VALVE FVv-105 FEB 1971 PLUG SALT TRANSFER LINES SEAL CONTAINMENT CELLS . Fig. 1.1. Outline of postoperation examinations. 1.2 REACTOR VESSEL AND CORE Before the large access flange on the reactor vessel could be opened, equipment above it had to be removed. The first items, the three control-rod drives, were removed quickly and uneventfully. The drives were not inspected but were placed in the equipment storage cell. Next the three control rods were removed and viewed in the reactor cell. They appeared to be in good condition, still flexible, with no perceptible distortions, surfaces smoothly oxidized with only minor marks from rubbing inside the thimbles. The poison section of one rod was severed, put in a lead-shielded pipe, and placed in the storage cell. (Gamma radiation was 7 R/hr at the outside of 2 in. of lead.) The 20-in.-diam standpipe attached to the core specimen access flange came out only after some flexible air lines were cut with a torch to clear the way. The standpipe, reading 10 R/hr at 3 ft, was bagged and buried. After electrical connections and the standpipe ventilation line were detached, the lead which had protected the rod drives was removed. The bolts in the 10-in. access flange came out without excessive torque. (The removal tool suffered a broken universal joint pin but worked satisfactorily after repair.) The access plug (which includes the control rod thimbles) lifted out freely and was laid down on top of the thermal shield, as shown in Fig. 1.2, for viewing. The surfaces of the rod thimbles were uniform in appearance, with a dull-gray surface. The lower 42 in. of thimble 3 was severed with a Heliarc torch and delivered for hot-cell examination (see Sect. 13.3). In the strainer basket section that had extended into the top head of the reactor vessel, the edges of the perforations .were sharp, with no visible erosion. The only salt evident on the whole assembly was small blobs visible through a few of the perforations. There was no heavy deposit in the region of the salt-gas interface, but farther up there was a blackened area (visible near the flange in Fig. 1.2). Inspection of the inside of the access nozzle with a periscope showed a darkened area near the top matching that on the plug. The area was opposite the air inlet into the cooling jacket, suggesting that the dark material was a surface deposit. Otherwise the inside of the access nozzle and the upper head of the reactor vessel appeared to be in excellent condition, with only a few scattered droplets of frozen salt in evidence. The first view down through the access opening was - as shown in Fig. 1.3. Several irregular chunks up to 2 in. in size were seen lying on top of the core structure in this area. Close examination with a right-angle scope showed them to be broken pieces of graphite. Figure 1.4 is a typical view through this scope, looking westward at a large chunk inside the metal centering bridge and a small chunk resting on top of the bridge. (These chunks are visible in Fig. 1.3 at the lower right.) A wide look at the top of the core through a fisheye lens disclosed six or seven more chunks of graphite lying about. Although no gaps were noticed in the core array, the characteristic pyramid shape on at least two of the chunks showed that they were the upper ends of vertical graphite bars. Figure 1.5 is a photograph through the fisheye lens, blurred by vibration of the long scope during the exposure. The direct view was sharper, and gaps in the array would probably have been noticed. The elements were not counted, however, and the absence of some around the periphery could easily have been overlooked. PHOTO 101607 Y 1 CORE_ACCESS PLUG S8 N . T “ ¥ CONTROL RODf == i 3 Fig. 1.2. Top of thermal shield during core inspection. PHOTO 101622 Fig. 1.3. View of top of core through access opening. PHOTO 101619 Fig. 1.4. Closeup view of broken pieces of graphite on top of core. One piece of graphite visible in Fig. 1.3 was definitely established as coming from the periphery of the core. This was the one lying partly across the rod thimble hole toward the bottom of the figure. (In this view the lighting makes the flat surface which faces upward appear bright.) The angles on the machined end of this piece coincide with those on construction drawings for one of the elements that was partly cut away for the retaining ring that circled the core at the top. It is the mirror image of the piece pointed out in Fig. 1.6. The most likely explanation for breaking of the graphite is that it occurred during cooldown. After salt trapped between the Hastelloy N retaining ring and the graphite froze, further thermal contraction of the ring could produce bending loads on the upper ends of the bars. (This is basically the same chain of events that broke many of the graphite pieces in the first specimen array exposed in the core.)® The chunks that got loose, to float up to the outlet screen on a fill and subside somewhere atop the core on a drain, were broken off above the retaining wire. The larger pieces (see Fig. 1.5) 4. MSR Program Semiannu. Progr. Rep. Aug. 31, 1966, ORNL4037, pp. 97-102. PHOTO 101615 Fig. 1.5. View of top of core through a fisheye lens. must have come from the rows inside the centering bridge, where the graphite extended almost 3 in. above the wire. An attempt was made to retrieve some of the pieces lying directly below the access opening, but all were accidentally dislodged and fell down through the thimble holes into the lower head where they would have been very difficult to recover. Except for the broken graphite (which could not have caused any significant perturbations during operation), the conditions in the upper head looked good, with clean surfaces and sharp edges. An example is the weld at the extreme right of Fig. 1.4. The east removable graphite stringer was marked on top to show its orientation and then was removed without difficulty into a shielded carrier for delivery to the HRLEL. (It was found to be in excellent condition, as described in Sect. 13.1.) Another stringer was being lifted out to provide a wider view of the lower head when an improper movement of the crane broke the graphite and the lower part fell back into the core. There was sufficient clearance for a light and a periscope to be let down into the lower head, but the view was rather restricted because of the metal grid structure. The bottom surface appeared to have a sandy texture, but if there was a deposit, its depth was very small compared with the dimensions of the drain line entrance, which could be clearly seen. No foreign objects were seen in the lower head. After viewing inside the reactor vessel was finished, one of the control rods was placed in a core channel and the access was closed with a gasketed blank flange. 1.3 FUEL PUMP The preliminary steps in cutting into the fuel pump bowl involved testing some of the built-in remote maintenance devices on the pump. The mechanism for compressing the bellows in the sampler tube spool piece was rusty, and a drive chain broke when its use was attempted, so the spool piece had to be removed by an improvised procedure. On the other hand, bolts in the PHOTO 70660 Fig. 1.6. Core graphite during assembly, before installation of retaining rings and centering bridge. Arrow points to piece similar to one found broken on top of core. large flanges on the motor and the rotary element were loosened without difficulty. An oxyacetylene torch was used to cut away the pump support plate, the cooling air shroud, and other obstructions over the pump bowl. The sampler tube was then cut off short, and the latch stop was tapped for attachment of the bowl-cutting tool shown in Fig. 1 This special tool consisted of an air-motor-driven abrasive cutter swivel-mounted on a bracket that fastened between the sampler tube stub and a bolt hole in the pump flange. Manual swiveling of the tool produced a trepan cut like that shown in Fig. 1.7 in the simulated pump bowl top. The actual cutting of the pump bowl was delayed by cutting wheels breaking, and when the fourth wheel broke after the cut was nearly complete, a chisel was used to finish freeing the sampler mist shield and cage. This assembly was then removed for detailed examination in the HRLEL, as described in Sects. 8.3 and 13.4. Visible on the bottom of the pump tank in the area that had been enclosed by the mist shield were the copper bodies of the two 10-g sample capsules that had been dropped in August 1967 and March 1968. The capsules seemed to be coated with an irregular dark deposit, neither had any sign of its steel cap, and one had been mashed rather flat. Subsequently the flattened sule was retrieved for inspection, but the other capsule was accidentally knocked out of reach. Evi- dently the recovered capsule was the older of the two, whose corroded steel cap had been retrieved in May 1968 and whose body had been bent by the impact of the heavy magnets used at that time.® In the same area as the capsules there was a pile (S0 to 100 cm?®) of loose dark material, some as small particles, some as lumps up to 1 cm, and some as thin S. MSR Program Semiannu. Progr. Rep. Aug. 31, 1968, ORNL-4344, pp. 26-29, 113-14. proTo_tor336 Fig. 1.7. Tool for excising fuel pump sampler cage, mounted on mockup of top of pump tank. sheets up to 2 cm across. On contact, the larger pieces crumbled easily, and heat from the viewing light caused the material to smoke. A sample was scooped up and delivered with the capsule. Analysis of the loose material and the deposits on the mist shield and cage are described in Sect. 8.3. Results of the inspection of the capsule are given in Sect. 13.5. A periscope was used to view the interior of the pump bowl, where throughout the years of operation oil had decomposed, salt mist had floated about,® and noble metals had concentrated.” Considering all this, surpris- ingly little material was found deposited on the various surfaces of the pump tank that were visible from the opening (see Fig. 1.8). The top head of the tank was covered with a dull, dark, rather uniform film, seemingly like that on the outside of the mist shield near the top. Here and there thin sheets of this material had peeled and hung down. 6. 1. R. Engel, P. N. Haubenreich, and A. Houtzeel, Spray, Mist, Bubbles and Foam in the MSRE, ORNL-TM-3027 (June 1970). 7. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, pp. 2—4. Some of the sheets could be seen swaying, presumably in the thermal currents from the hot lamp. Quite different were the deposits on the upper surface of the sloping baffles in the upper part of the bowl. These were nonuniform, rather rough or angular, and reflected more light than did the deposit on the top head. This type of deposit is illustrated in Fig. 1.9, which is focused on the edge of the "s-in.-thick baffle that slopes down from the volute support cylinder. (There is some blurring in the image due to motion during the exposure. The overflow pipe and its junction with the lower head, seen in the background, are also out of focus.) The same kind of deposit is seen in Fig. 1.10. This is a closeup of the end of the spray ring and attached baffle. The baffle itself is "-in.-thick metal, and the deposit seen here in edge view is less than half as thick. The ripples in the weld on the end of the tube are visible through the deposit. In the patch seen in Fig. 1.10 near the juncture of the baffle and the tube, where the deposit had evidently flaked off, the exposed metal was clean and smooth. Since it seemed that this kind of deposit was also present on the excised mist shield, no effort was made to obtain a sample from surfaces inside the bowl. SAMPLE CAPSULE CAGE OVERFLOW PIPE ORNL-DWG 69~ 10172 OFFGAS LINE BUBBLER T Z (“Amn\\\ 40| J 20 ot . Z LeveL = DISCHARGE o e JSALT <= RADIOACTIVE GAS <= CLEAN GAS SUCTION Fig. 1.8. MSRE fuel pump internals. Fig. 1.9. View inside fuel pump bowl showing edge of spray baffle, with scattered deposits. The overflow pipe is visible in the background. 10 puoro_iorees 101098 Fig. 1.11. View in lower part of pump bowl showing ports into volute suction. The portion of the spray ring below the baffle was clean, and the spray holes that were visible were sharp and clear. One of the brackets supporting the spray ring appeared white and seemed to be detached from the wall. Nowhere to be seen in the pump bowl were salt droplets or the globular deposits often associated with frozen mist, and nothing was visible in the salt-gas interface zone to suggest that there had been any foam. Except for the debris around the sampler region, the bottom of the pump bowl was relatively clean, with only a few areas showing a thin dark film or deposit. The metal surfaces that had been submerged in the salt were generally smooth dull gray. One exception was on the lower extension of the volute where there was some discoloration and minor roughness around the ports. This is evident in Fig. 1.11. On the bubbler tube there was a white area just above its penetration of the baffle extending out from the volute, where gas that collected beneath the baffle bubbled up through the clearance gap around the tube. After the viewing was concluded, the hole in the pump bowl was sealed. A steel patch with silicone rubber sealant around the periphery, held in place by a jack bolt extending up to the rotary element flange, was installed and proved leak-tight in a soap test at 5 psig. 11 1.4 HEAT EXCHANGER Two heater units, HX-1 and -2, had to be removed to permit cutting into the heat exchanger. Both had been removed and reinstalled in 1968 without particular difficulty,® but this time HX-1 hung on a freeze flange clamp during removal and was pulled out of shape. Both units were placed in the equipment storage cell. A section of the '-in-thick shell of the heat exchanger was cut out, using a plasma torch device that had been developed and tested on a mockup of the heat exchanger shell. Figure 1.12 is a view through the 10- by 13-in. oval opening. The dull appearance of the tubes is due to a thin coating which wiped off easily during handling. The oily spot on the tubes at the center of the opening is lubricant used in tapping a hole drilled in the shell for attachment of the torch fixture. The abrasive cutoff tool shown in Fig. 1.13 was used to remove sections of six tubes for the detailed examina- tions described in Sect. 13.6. Several other tubes were cut and raked out into the reactor cell to permit viewing inside the heat exchanger. The periscopic view showed the inside of the shell, baffles, tubes, and 8. MSR Program Semiannu. Progr. Rep. Aug. 31, 1968, ORNL4344, p. 31. Fig. 1.12. Opening in primary heat exchanger, showing cut made with plasma torch and exposed tubes. PHOTO 101333 Fig. 1.13. Abrasive cutoff tool for removing heat exchanger tube section. cross-lacing. There seemed to be a slight coating on everything, but otherwise nothing unusual was ob- served. The opening in the heat exchanger shell was closed by welding on a patch of %-in.-thick type 304L stainless steel. A sound weld was obtained around most of the periphery, but at some spots there was porosity. After repeated attempts to obtain a good seal by adding filler metal were unsuccessful, iron-filled epoxy putty was applied. A soap test at 5 psig then disclosed no leaks. 1.5 LEAK AT FREEZE VALVE FV-105 When the leak appeared during the final shutdown on December 12, 1969, evidence pointed to freeze valve FV-105 as the probable location.” This was confirmed during the postoperation examination as soon as the heater-insulation units were removed from the salt lines. None of the units was especially radioactive except for the one covering FV-105, which was significantly contaminated, and a visual scan of the lines showed a blob of salt in the suspected area, immediately adjacent to the cooling air shroud. Close inspection through a periscope revealed the knobby lump shown in Fig. 1.14, suggestive of a slow leak in a freezing environment. On the other side of the pipe, there was, in addition to more knobby material, a thin sheet of salt that had run out on the flat base under the removable insulation unit. The total amount of salt that had leaked was estimated at between 2 and 3 in.% . The section of 1%-in. pipe containing the freeze valve was cut out with an abrasive cutoff tool and delivered to the HRLEL. Examinations (described in detail in Sect. 13.7) showed the leak to be a crack in the pipe just outside the shroud, where large cyclic stresses were produced by differential thermal expansion of the shroud and pipe during operation. These stresses were higher than in the development model that had been tested because a late modification to increase cooling air flow'® had the unintended effect of greatly strengthening the shroud. This had been overlooked, and the failure occurred after many fewer cycles than had been considered allowable. 1.6 OTHER EXAMINATIONS One of the uncertainties in the tritium balance in the MSRE'' was the amount of tritium produced from lithium in the thermal insulation around the reactor vessel. In order to measure the lithium content, samples of the insulation were obtained by taking out one of the removable slides of the thermal shield. The cooling water lines were cut with a torch, and the slide was lifted out without difficulty. Working through the maintenance shield, we chiseled the metal can open and took three samples of the block insulation. Lithium analyses ranged from 12 to 38 ppm, which meant that 9. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL4548, pp. 3, 14. 10. MSR Program Semiannu. Progr. Rep. Feb. 28, 1965, ORNL-3812, p. 28. 11. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp. 9-10. PHOTO 023071 Fig. 1.14. Closeup view of salt near freeze valve FV-105. The drain line and %-in.-diam thermocouple sheaths loom at upper left. only a small fraction (less than 5%) of the 3 to 5 Ci/day of tritium appearing in the cell atmosphere could have been produced in the insulation; nearly all must have diffused out through the walls of the fuel system. Tritium in the samples indicated that the total retained in the 500 kg of insulation was only 2 to 3 Ci. The outside of the reactor vessel was examined, using a periscope inserted through the thermal shield slide opening. All visible surfaces of the vessel were smooth and seemed to have remarkably little oxide, being noticeably lighter in color than the 5-in. salt pipe into the vessel, which had a uniform black surface. The drain line, between the reactor vessel and its penetration of the thermal shield, showed no visible effect of opera- tion. The same was true of the heater sheaths and thermal insulation cans. Several sheathed thermocouples were removed from various locations in the MSRE for testing to determine effects of radiation and/or high temperature. Thermo- couple TE-R52, which had been in a well in the core access plug, was the most highly irradiated of those that were recovered. Two thermocouples were chiseled off the core inlet line and four off the heat exchanger, but attempts to detach one of the thermocouples welded to the outside of the reactor vessel were unsuccessful. 1.7 EVALUATION OF TOOLS AND PROCEDURES It was not required that the postoperation examina- tions leave the fuel system in operating condition or able to be restored. The patches on the openings in the fuel system were required to be reasonably leak-tight but not necessarily resistant to very high temperature and radiation. This simplified the tools and procedures but limited the demonstration of radioactive mainte- nance capability. Even so, the experience was valuable. The technique used in the examinations was basically the same as had been used for maintenance of the radioactive systems: long-handled tools, operated through a portable work shield set up over a hole in the cell roof. Most of the tasks were new, however, and required specially designed tools. In addition to the usual socket wrenches, hooks, etc., there were air- motor-driven abrasive wheels in several configurations, a hydraulically operated shear, oxyacetylene cutting torches, a ‘plasma torch, and Heliarc welding torches. Most of the special tools are shown arrayed on top of the drain tank cell in. Fig. 1.15. Not shown are the periscopes and the plasma torch with its fixture for mounting on the heat exchanger. The removal of the control rods, rod thimbles, and core graphite had been provided for in the reactor design, and few difficulties were encountered. The only one that would have been troublesome if the aim had been to resume operation was the unexpected inter- ference of some flexible air lines with the removal of the containment standpipe. The remote removal and replacement of the fuel pump motor and rotary element had been tested in 1965, and the relative ease with which the flange bolts were removed during the postoperation work suggested that this task was still practical. Replacement of the pump. bowl and piping was a job that had been anticipated but not tried. The failure of the built-in devices for removing the sampler tube spool piece would have added to the already formidable propor- tions of this task. 14 that the pipe ends were left clean and square, ready for tapping or plugging. The oxyacetylene torch proved quite satisfactory for rough cutting, including the 2-in. steel support plate for the fuel pump, the cooling air shroud, and the water lines to the thermal shield slide. The plasma torch used on the heat exchanger shell was a commercially available unit: only the device for moving it over the surface was specially designed for this job. Development tests on a mockup proved the device capable of maintaining a constant separation distance and established the proper control settings. In the reactor cell, however, considerable difficulty was met in starting the cut. On the first five attempts, backspatter of molten metal ruined the torch nozzle, which had to be replaced. This was complicated because of radioactive contamination on the torch. Finally the . gap between the torch and the shell, which had been set Cutting Hastelloy N with long-handled tools was | something new. This alloy is characteristically difficult to cut, and under the conditions in the reactor cell, the positive advance of the cutter necessary to avoid work ~ hardening could not be guaranteed. Abrasives work satisfactorily on Hastelloy N, however, so the rotating grinder (Fig. 1.7) was chosen over a hole saw for cutting out the fuel pump sampler cage. When the ground-off latch stop was tapped for attachment of the tool, even though lubricant was applied, the torque required was so high that elastic torsion of the pipe on which the tap was mounted made the job difficult. During the trepanning operation, grinding wheels broke more frequently than in the mockup tests. It appeared that bits of slag from the torch cuts on the obstructions overhead were being dislodged by the vibration and falling into the trepan groove. The cut took about 8 hr grinding time, but replacement of three broken wheels extended the working time to several days. The grinding wheels worked well in cutting the sampler tube, heat exchanger tubes, and the fuel lines around the freeze valves. Two of the cuts in the last area were made without difficulty through a pipe filled with frozen salt. An advantage of cutting this way was at "¢ in., was reduced to % in., and the cut was made with no difficulty. The jet from the torch extended far enough to sever a few tubes. Spatter was confined to a zone near the cut, but there was a widespread, nonuniform light coating inside the heat exchanger that was probably condensed material vapor- ized by the torch (see Sect. 13.6). Patching the heat exchanger shell was a challenging task. The original plan was to cut a ' -in.-thick piece of curved plate fitting into the hole left by the plasma torch. Then a Heliarc torch, guided around the path followed by the cutting torch, would fuse the patch and shell. After mockup tests showed that it was very difficult to get an adequate fit, however, it was decided to try an overlapping patch. The overlap unfortunately ruled out use of the existing torch guide to maintain a fixed gap, but successful welds were made by a welder, watching through a periscope and controlling the gap by a rope attached to the torch. Although this worked well in the mockup, troubles were encountered in the reactor. The shell around the hole was prepared with brushes and sanding disks mounted in an electric drill, and the patch was set in place without difficulty. When welding was started, however, the edges of the patch tended to melt and run off without adhering to the shell. Changing the angle of the torch to put more heat on the shell helped, and a good-looking fusion weld was obtained around most of the periphery despite the awkward working arrangement. In places, however, the patch had melted back so far that gaps were left. Attempts to close these up using welding rod were not completely successful as shown by a pressure test. After more fruitless efforts at weld repair, a good seal was made with iron-filled epoxy, backed up by a “dam” cut from a stainless steel plate that had been rolled to fit on the heat exchanger shell around the patch. The radia- tion level on the shell was 630 R/hr, which would not cause deterioration of the epoxy for several years. The severed fuel drain lines were plugged with devices like the one shown in Fig. 1.16. A conical plug of soft copper was forced into the end of the 1%-in. pipe by a jack bolt through a bracket clamped to the pipe. A tube connected to a hole through the plug extended up to the top of the cell, where temporary connections could be made for gas addition. Each of the three lines leading to the salt tanks extended down into the frozen salt and 5o could be pressurized to test the plugs. There were some problems in installation due to brackets breaking, but satisfactory seals were obtained in every case. The hydraulically operated shear worked well in tests but was not used in the reactor because the anticipated need did not materialize. PHOTO 0227-71 B WAL LABORATORY Fig. 1.16. Device for plugging severed ends of salt lines. 2. Further Investigations During operation of the MSRE there arose two discrepancies which we hoped to resolve during the postoperation examination. One was the discrepancy between the reactor power indicated by the heat balance and that indicated by the changes in ratios of nuclides in the fuel. The other was a difference between the amount of uranium that was stripped from the fuel salt in 1968 and the amount that was later recovered from the UF4 absorbers. 2.1 TEST OF COOLANT SALT FLOWMETER AND CONCLUSIONS C. H. Gabbard " P, N. Haubenreich The power level of the MSRE was routinely obtained from a heat balance on the reactor cell by the on-line computer. Using the finally accepted value of the specific heat of the coolant salt, this computation indicated that the maximum power (heat removal) was about - 8.0 MW. Serious doubt was cast on this value, however, when analysis of the changes in uranium and _ plutonium isotopic ratios over long periods of operation indicated that the power had been 7 to 8% less than that indicated by the computer heat balance. Heat removal in the coolant salt was the preponderant term in the heat balance (98% of the total at full power), and suspicion was directed at the coolant salt flow measure- ment. _ : The volume flow rate of coolant salt was measured by a venturi flowmeter in the 5-in. pipe near the radiator inlet, There were two readout channels, each consisting of two pressure taps in which molten salt transmitted pressure through metal diaphragms to NaK-filled lines leading out of the heated zone to a differential pressure cell. Associated electronics produced a signal that ranged from 2 to 10 V in proportion to the square root of the pressure differential. In the computer this voltage was converted to flow rate by a multiplicative factor based on the manufacturer’s calibration of Ap vs flow in the venturi and a voltage-Ap relation for the instrument. that had been determined before operation. Near the end of nuclear operation, a review of the flow-Ap calibration disclosed a mistake that was in- troducing an error of —2.9% in the indicated flow, and 17 K 'whose correction only increased the discrepancy be- tween heat balance power and the nuclide indications.’ The Ap-voltage calibration of the flowmeter readout instrumentation could not be checked until the salt piping could be cut. During the postoperation examinations, the 5-in. coolant pipe was removed on either side of the venturi, and a plug was installed in the converging section and one upstream so that pressure could be applied to the upstream taps. The pressure differential between the upstream and throat taps was raised and lowered through the normal range while Ap and voltages were accurately measured. It was found that the voltage which had been interpreted as 850 gpm was produced by a pressure differential equivalent (according to the corrected flow-Ap calibration) to a flow of only 793 gpm, or an error in the Ap-voltage calibration of +6.7%. In addition to the errors of —2.9 and +6.7% in the measured salt flow rate, an error of +0.4% in the salt temperature rise had been indicated by an experiment on the effects of isothermal system temperature changes on thermocouple biases.? The net effect of correcting - for these errors was to change the heat balance power from 8.0 to 7.65 MW. . Ragan had calculated nominal full power values of 7.30 + 0.10 MW from the 23°U depletion and 7.45 * 0.18 MW from the 2°® U buildup for a weighted average of 7.34 + 0.09 MW.? Thoma and Prince inferred a value of 7.41 MW from the changes in 2*°Pu/23°Pu ratios and observed that changes in the 234 U/?33 U ratio were in excellent agreement with this value.* The 7.65-MW value from the finally corrected heat balance is higher than the nuclide values by about 3%. Uncertainties in cross sections, fission energies, salt density, salt specific heat, venturi calibration, and d/p instrument calibration are enough to account for the difference. 1. MSR Program Semiannu. Progr. Rep. Aug 31, 1969, ORNL-4449, p. 12. _ 2. C. H. Gabbard, Reactor Power Measurement and Heat Transfer Performance in the MSRE, ORNL-TM-3002 (May 1970). 3. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp. 65-66. 4, Ibid., pp. 98—-102. It is unfortunate that throughout the operation of the MSRE, changes from time to time in the best estimate of the power level required reworking of detailed analyses of reactivity, heat transfer performance, and fuel chemistry. There appears to be no prospect of further reducing the uncertainty that still remains. The nominal full power of the MSRE was probably between 7.4 and 7.6 MW. Analyses and reports that have been made since the last few months of operation have generally used a nominal full power of 7.34 or 7.4 MW, and there is insufficient cause to redo this work. 2.2 INVENTORIES OF RESIDUAL URANIUM AND PLUTONIUM R. E. Thoma As mentioned in the preceding section, we deduced from comparisons of observed changes in the isotopic composition of plutonium with nominal values that the -maximum power output of the MSRE was 7.4 MW(t). After obtaining this value, we computed a material balance on the 228U + 23°U that was in the fuel salt from 1965 to 1968. This balance indicated that the amount that left the fuel salt during the chemical processing operations in 1968 was some 4 kg more than the amount that was recovered from the UFq ab- sorbers.® As will be described in a forthcoming report summarizing experience with the MSRE.® refinement of that assessment resulted from isotopic-dilution anal- yses, which showed that the amount of 23®U loaded into the MSRE was actually about 2 kg less than was credited in on-site records. The resultant material balance is presented in Table 2.1. Recovery of uranium was carried out by Goodyear Atomic Corporation, Piketon, Ohio, by dissolution of the NaF pellets in which the UF¢ was absorbed. Special precautions were taken to minimize measurement un- certainties,” and a review® by Goodyear of the oper- ations did not reveal any reason for revising the originally reported amount of uranium recovered. We have reexamined the possibilities of such errors as might be ascribed to misestimates of power output of the reactor and the implications of reactivity anomalies and 5. R. E. Thoma, internal correspondence to M. W. Rosenthal, Sept. 14, 1970. 6. R. E. Thoma, Chemical Aspects of MSRE Operations, ORNL-4658 (in press). 7. Letter GAT-532-70-212 from W. B. Thompson, Goodyear Atomic Corp., to R. B. Lindauer, ORNL, Sept. 2, 1970. 8. Letter GAT-510-70-153 from C. D. Tabor, General Manager, Goodyear Atomic Corp., to R. V. Anderson, Manager, Portsmouth Area, AEC, Nov. 11, 1970. 18 Table 2.1. Material balance for uranium in the fuel sait, 1965—-1968 Kilograms of uranium Charge at initiation of power operation 227.020 Additions as fuel replenishment +2:461 Transferred to flush salt -6.272 Removed in samples? ~0.256 Consumed in 9005 EFPH (equivalent full -3.594 power hours) of operation? I Net at end of 235U operation 219.359 Retained and mixed with 233U charge€ —1.935 Removed during processing 217.424 Recovered from UF¢ absorbers (max) 214.776 Disparity 2.648 4]. R. Engel, MSRE Book Uranium Inventories at Recovery of 235U Fuel Charge, internal memorandum MSR-68-79 (May 1968). bMSR Program Semiannu. Progr. Rep. Aug 31, 1969, ORNL-4449, p. 25. ‘ “MSR Program Semiannu. Progr. ORNL-4396, p. 131. Rep. Feb. 28, 1969, short-term trends in the results of chemical analyses; their possible contribution to the disparity indicated in Table 2.1 is negligible. Analysis of scrubber solutions during the processing showed less than 1 g of uranium discarded. The conclusion, therefore, is that the ma- terial balance affords unequivocal evidence that some 2.65 kg of uranium (33.08 wt % 23°U) remains at the MSRE, possibly in the chemical processing plant. Although there was no direct evidence to support uranium retention in the processing equipment, two components have been identified as conceivable sites for such retention. Engel noted that the particle filter (a 9-ft? filter designed to remove corrosion product solids from the fluorinated salt before its reuse in the reactor) in the line between the fuel storage tank and the processing tanks could, after treatment of the flush salt was completed, have contained an unknown amount of zirconium metal, delivered to this location as the processed flush salt was returned to the reactor system. It is difficult to assign high probability to the events which could have reduced the uranium from the fuel charge as, subsequently, it passed through this filter in such a.way that some 2.5 kg of uranium remained in the filter; however, the possibility cannot be excluded. Another possible site where uranium may have been retained, as suggested by R. B. Lindauer, is the high-temperature sodium fluoride absorber bed, which is positioned between the fuel storage tank and the NaF absorbers. The design temperature for operation of this 19 Table 2.2 Inventory of residual uranium and plutonium in the MSRE# Uranium inventory 233U 234U 235U 236U 238U sU Fuel circuit inveritory, run 20-1, kg 28.568 2.526 0.869 0.036 2.020 34.019 Total inventory, run 20-1, kg 31.052 2.746 0.945 0.039 2.196 36.978 Drain tank inventory, run 20-1, kg 2.484 0.220 0.076¢ - 0.003 0.176 2.959 Fuel circuit inventory, run 20-F, kg 28.406 2.533 0.866 0.036 2.011 33.852 Transfer to flush salt, run 20-F, kg 0.411 0.037 0.013 0.061 0.029 0.491 Charged into drain tank, run 20-F, kg 27.995 2.496 0.853 0.035 1.982 33.361 Drain tank residue, kg 2.484 0.220 0.076 0.003 0.176 2.959 Final drain tank inventory, kg 30.479 2.716 0.929 0.038 2.158 36.320 U/zU, wt % 83.918 7.478 2.558 0.105 5.941 Plutonium inventory 239Pu 240py 238,241,242Pu > Pu Fuel circuit inventory, run 20-1, g 625.8 61.81 2.39 690.0 Total inventory, rtun 201, g 680.2 67.19 2.60 749.99 Drain tank inventory, run 20-1, g 54.4 .5.38 0.21 59.99 Fuel circuit inventory, run 20-F, g 615.6 65.43 2.37 683.4 Transfer to flush salt, run 20-F,P g 61.8 6.57 0.23 68.6 Charged into drain tank, run 20-F, g 553.8 58.86 2.14 614.8 Drain tank residue, g 54.4 5.38 0.21 59.99 Final drain tank inventory, g 608.2 64.24 2.35 674.79 Pu/ZPu, wt % 90.13 9.52 0.35 Weights are based on comparisons of analytical results and computed values. These comparisons indicate maximum power output as 7.4 MW(t). Final estimates assume 4167 EFPH at 7.4 MW(t). bThis item makes the simplifying assumption that the total amount of plutonium estimated to be transferred to the flush salt was transferred during the final flush of the fuel circuit. Table 2.3. Composition of fuel salt stored in the MSRE drain tanks TLiF BeF, ZrF, 233.42yF, 239.11pyp, Composition, mole % Nominal 64.50 30.180 5.199 0.132 . 2.38 x 1073 Analytical® 64.53 30.43 4.90 0.137 Composition, wt % Nominal 41.87 35.44 21.67 1.0195 0.0177 Analytical® 41.37 35.27 20.19 1.06 2Current calculations do not include corrections for transfer of carrier solvent residues to flush salt nor flush salt residues to fuel. Disparity between nominal and analytical values for zirconium will be reduced by introduction of this correction factor. bRun 17-20, average of 33 samples. absorber is 750°F, based on previous laboratory studies.” The laboratory studies indicate that this absorber would not retain UF4 at the operating temperature; only the possibility that temperature gradients prevailed within the absorber at periods near the end of fluorination operations and allowed the retention of some uranium within the absorber gives 9. MSR Program Semiannu. Progr. Rep. Aug. 31, 1966, ORNL4344, p. 321. - any credence to the likelihood that uranium wou}d be retained here. Attempts to locate uranium in the processing plant are described in Sect. 2.3. While confirmation of the amount of uranium in the processing plant by direct experiment would be de- sirable, the isotopic dilution analyses and the data which were used to monitor the transfers of uranium and plutonium within the reactor system appear to be sufficiently reliable for estimating the amounts of 23°U and 238U retained in the reprocessing system and for computation of final inventory distribution. " In preparation for phase Il of the program for decommissioning of the MSRE, inventories of uranium and plutonium in the stored salt charges were esti- mated. Weights and isotopic composition of uranium and plutonium in the drained fuel and flush salts are listed in Table 2.2. Using these weights and a value of 4707.5 kg as the weight of the fuel carrier salt, the composition of the fuel salt was computed. Nominal values are compared with analytical data in Table 2.3. 2.3 SEARCH FOR UNRECOVERED %3°U J. R. Engel The data and analysis reported in the preceding section clearly showed that the removable absorbers in which the original charge of uranium was recovered as UF¢ contained significantly less uranium than had been in the fuel salt at the end of nuclear operation with 2357U. We sought to determine more accurately how much uranium had been left behind and where it was located. As a first step we hoped to examine the particle filter and the hot NaF trap. Since it was necessary to leave the processing plant in an operable condition for ultimate disposal of the fuel and flush salts, a nondestructive technique was required to look for the uranium. Neutron interrogation has been developed as a highly useful technique for assaying the fissile-material content of closed containers. Nor- mally the technique is applied to relatively portable 20 objects that can be examined under carefully controlled conditions with high-sensitivity neutron detectors and intense sources. For this particular application, the interrogation would have to be performed in situ under relatively crowded conditions in a concrete cell. Never- theless, it appeared to offer the only hope for a nondestructive search. Preliminary investigations indicated that useful in- formation might be obtainable with a relatively simple experimental arrangement. Therefore it was decided to proceed with an experiment using an isotopic neutron source and a single neutron detector. The neutron source to be used contained ~14 ug of 232Cf, which produced about 3 X 107 neutrons/sec. The neutron detector was a boron-lined chamber with an absolute efficiency of 15 counts per nvt. Tools were developed and built to permit positioning and manipulating the source and detector in the processing cell near each of the two components. Because of the complexity of the geometries involved and the expected importance of extraneous scattered neutrons from the source on the detector response, a simplified mockup of the salt filter was built to measure the absolute sensitivity of the interrogration equipment. A clean ORR fuel element was used as the uranium- bearing target in the mockup. Measurements with this system revealed that the sensitivity was too low to provide reliable information about the components in the processing cell, and the investigation was dis- continued. Consideration will be given to other methods of locating and recovering the unaccounted-for uranium during the final phase of the MSRE de- commissioning, ~Part 2. MSBR Design and Development '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 done a conceptual design for a 1000-MW(e) plant, and the report describing the plant is in press. A contract is now being negotiated with an industrial group to do a conceptual design of a 1000-MW(e) MSBR plant using the ORNL design for background and incorporating 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 are doing studies of plants that could be built as the next step in the development of large MSBRs. 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 alternative is the Molten-Salt Demonstration Reactor, 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 minimum 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. In addition to these general studies of plant designs, the design activity includes the assessment of the safety of molten-salt reactor plants. Some studies related to safety are in progress preliminary to a comprehensive review of safety based on the design of the 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-salt 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. Work is progressing on the design of one loop facility for 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. Design is nearing completion for a second loop facility for studies of equipment and processes and of the 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 basi¢ design data are few and the potential problems are many. A program involving industrial participation is being un- dertaken to provide the technology for designing and building reliable steam generators for molten-salt re- actors. 3. Design E. S. Bettis 3.1 SINGLE-FLUID 1000-MW(e) MSBR DESIGN STUDY REPORT Roy C. Robertson The report' covering the design and evaluation studies of a 1000-MW(e) molten salt thermal breeder reactor power station in which the fissile and fertile - materials are incorporated in a single fluoride salt has 21 received final editing and approval and is now in the process of being published. Distribution is scheduled for June 1971. 1. Molten-Salt Reactor Program Staff, Conceptual Design Study of a Single-Fluid Molten-Salt Breeder Reactor, ORNL-4541 (in press). 3.2 MOLTEN-SALT DEMONSTRATION REACTOR DESIGN STUDY E. S. Bettis H. A. McLain J. R. McWherter H. L. Watts C.E. Bettis C.W. Collins W. K. Furlong 3.2.1 Introduction Design and evaluation studies of a 300-MW(e) molten- salt demonstration reactor (MSDR) were continued. This power station would be a first-of-a-kind prototype to demonstrate the feasibility and delineate the prob- lems of construction of a large-scale molten-salt reactor power station. The prototype studies have concentrated on concepts which would require a minimum of development and would permit construction of the demonstration plant in the near future. This aspect led to our decision to design the MSDR as a low-power- density converter rather than a breeder, since this could be done without a significant penalty on the short-term fuel cycle costs, yet would eliminate the need for core graphite replacement during the 30-year life of the " plant and for sealing of the graphite against gas permeation to reduce the '*%Xe poisoning. The con- verter could also substitute periodic salt replacement for continuous fuel salt processing until a suitable chemical plant was fully developed. The conversion ratio without processing would be about 0.8. The MSDR general flowsheet, plant layouts, reactor vessel, primary heat exchangers, and drain tanks have been described previously.? During the past report period we decided to make several revisions to the concept, however, the most notable of which was to interpose an additional salt circulation loop between the primary heat exchangers and the steam generators in order to assure confinement of the tritium formed in the fuel salt. The building structure was altered to accommodate the new loop, the drain tank and gas-handling systems were modified, and the systems for heating the cells were revised. The revised flowsheet is shown in Fig. 3.1. No revisions have been made in the reactor concept itself since last reported.? The reactor core is 21 ft in diameter and 21 ft high and is surrounded by a 2Y, -ft-thick graphite reflector. The all-welded reactor vessel is fabricated of Hastelloy N fortified with additives to improve the resistance to radiation damage (see Sect. 15). 2. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548. 22 3.2.2 Addition of Third Salt-Circulation Loop Although studies of the amounts of tritium that could diffuse from the fuel salt into the coolant salt and steam system are not complete and the various meas- ures that could be used to mitigate the migration have not been fully evaluated, it was decided for the present to incorporate into the MSDR concept an assured method of tritium confinement. Chemical considera- tions indicate that an efficient tritium trap can be obtained by use of a nitrate-nitrite salt mixture in a loop between the secondary system and the steam system, as shown in the flowsheet, Fig. 3.1. The salt would oxidize the tritium to water and prevent its reaching the steam systemn. The nitrate-nitrite salt is believed to be unsuitable for use in the secondary system because it decomposes rapidly as the tempera- ture is raised above 1100°F and it would react vigorously with the graphite if it were to leak into the primary system. Addition of another set of heat exchangers, circu- lating pumps, and connecting piping has the obvious disadvantages of added complexity and cost, but there are desirable features of the new arrangement besides trapping the tritium, and some of the aspects tend to partially offset the capital cost of the equipment needed for the third loop. With the steam-raising equipment in a thlrd loop the equipment in the secondary system can be closely coupled, the inventory of secondary salt can be - considerably reduced, and use of more expensive coolants with higher melting temperatures can be considered. The most obvious of these is the ’ LiF-BeF, (66-34 mole % with 99.99% 7Li) salt mixture which performed well as a coolant in the MSRE. Although it has a high melting peint of about 850°F and a relatively high cost of about $11.40 per pound (based on a "Li cost of $120 per kilogram), it has the decided advantage that a leak of the coolant into the fuel salt (LiF-BeF,-ThF4-UF,) becomes of considerably less significance. On this basis, " LiF-BeF, was selected as the fluid for the MSDR secondary circulation system shown in the revised flowsheet (Fig. 3.1). The tertiary circulation loop, which transports heat from the secondary heat exchangers to the steam generators and reheaters, uses a commercially available nitrate-nitrite heat transfer salt, KNO;-NaNO,-NaNO,, with the eutectic composition 44.2-48.9-6.9 mole %. In addition to being an effective chemical trap for tritium, the salt has a desirably low liquidus temperature of 288°F, a relatively low cost of about 15¢/Ib, and good heat transfer and flow properties (see Table 3.1). There are three outstanding advantages to its use in transporting ENTRAINMENT SEPARATOR TERTIARY SALT PUMP SECONDARY SALT PUMP 1100°F 23,000 gpm 1000°F 27,180 gpm HEAT REJECT TANK ORNL—DWG 71—3633 GAS SEPARATOR REHEATER PRIMARY SECONDARY HEAT HEAT EXCHANGER EXCHANGER ~_700°F 1050° F 900°F I__J'_"‘I | | | GAS CLEANUP - | |~ SYSTEM | . ) N > - TO CHEMICAL | ¥|| - «— PROCESSING SYSTEM H | i1l PRIMARY 1 {| DRAIN TANK ip3 [} Fig. 3.1. Preliminary flowsheet for MSDR power plant. 900°/ 900°F . 2400 psi N HIGH PRESSURE TURBINE SUPERHEATER MOISTURE SEPARATOR INT. AND LOW PRESSURE TURBINES !-_'__'I L FEEDWATER CHAIN GENERATOR ] i | CONDENSER €T heat to the steam generators and reheaters: (1) in the event of small steam leaks, there is no chemical reaction with water, (2) corrosion rates at high temperature are very low with materials which are less expensive than Hastelloy N, and (3) the low melting temperature permits use of feedwater and cold reheat steam at conventional temperatures. These three advantages are discussed below. . The steam pressure will exceed the pressure in the tertiary heat transport loop; thus tube failures or leakage would cause water to mix with the coolant salt. At the operating temperatures and pressures in the nitrate-nitrite loop, however, essentially all the contami- nating water would exist as steam above the surface of the salt, and the vapor could be purged, taking care, of course, to prevent the escape of tritiated water. The salt would not react with the water to produce highly corrosive conditions. Hastelloy N has been chosen as the material of" construction in all previous designs of steam generators and reheaters for molten-salt reactor power systems. Use of the nitrate-nitrite heat transport salt would make it possible to use less expensive materials. For example, the data available to date indicate that the salt would be compatible with Incoloy 800, a material that has proven acceptable for service in high-temperature steam systems. The substitution of Incoloy for Hastel- loy N could make significant savings in the capital costs of equipment. The previous MSBR conceptual design employing sodium fluoroborate coolant salt to transport heat to the steam generators provided a minimum feedwater temperature of 700°F. Although this minimum has not been established experimentally, the 725°F liquidus temperature of the salt is likely to dictate feedwater temperatures well above those in conventional regenera- tive feedwater heating systems. In the steam system flowsheet for the reference design MSBR, the cold reheat steam is preheated from 550 to 650°F by heat exchange with 3500-psia 1000°F steam taken from the throttle supply, and the feedwater is heated to 700°F by direct mixing with the exit heating steam from the reheat steam preheater. If the nitrate-nitrite salt, with its liquidus temperature of 288°F, were used to transport the heat, conventional feedwater and cold reheat steam temperatures could be used. This would effect important savings through elimination of the reheat steam preheater, the pressure booster pumps, and the mixing chambers employed in the special MSBR feedwater circuit. Since the aforementioned direct-mixing arrangement for feedwater heating fa- 24 vored use of supercritical pressure steam,’ elimination of this aspect allows consideration of a 2400-psia cycle on a more equal basis with the 3500-psia system, although the latter is still probably preferred for the MSDR. Although the startup of an MSDR has not been studied to date, it seems probable that the less- restrictive feedwater conditions of the nitrate-nitrite coolant system would also make it possible to use a less expensive type of startup boiler than was assumed in the reference design MSBR.! In summary, addition of a third salt-circulation loop in an MSDR or an MSBR provides tritium control and other important advantages. In particular, it could simplify operation of the MSDR steam system. Were it not for the tritium problem, however, the additional loop probably would not be recommended because of the added capital expense. A more definitive study is now under way to evaluate the costs. 3.2.3 Salt Overflow and Gas Stripping Systems A further major change in the flowsheet from that previously reported? is elimination of the drain tank as an overflow tank for the primary circulation system. All salt volume changes in the circulation system are now accommodated in the pump tanks. This revision was made in the plant, because the jet pumps used to return the overflow salt to the circulation systems would have had to pump a large flow of salt against a head now considered too large to be practicable in the MSDR. Each primary pump tank has sufficient free volume to accommodate a 10% total change in system salt volume. In addition, each pump tank has an overflow line connected to the reactor outlet plenum. During steady- state operation, the cooling and fountain flows in the pumps are returned to the circulation system via this same route of low flow resistance. 3.2.4 Primary Drain Tank With elimination of the continuous salt overflow from the pump bowls, the drain tank could be modified to provide only one entering salt line. This nozzle connects with both the reactor drain line and with the reactor cell catch basin drain line through valves that are discussed below. A jet pump (activated by a salt flow from a small salt-circulation pump located in the chemical processing cell) is used for returning salt from the drain tank to fill the primary system for startup. The size of the drain tank, the drain tank crucible, and - the arrangement of the cooling system are as previously reported.? 3.2.5 Drain Valves for Salt Service The drain valves in the main reactor drain line and in the catch basin drain line are located in a small cell between the reactor cell and the drain tank cell, as shown in Fig. 3.2. The main drain valve for the reactor is now visualized as a combination mechanical and freeze valve. A bellows-sealed poppet, or plug, ap- proaches a hard-faced surface but does not seat tightly against it. A cooling liquid is then circulated through the poppet to freeze an annular ring of stagnant salt between the two faces to effect a tight shutoff. The circulating liquid can be heated to thaw the salt to open the valve. ' Two valves are used in parallel in the line from the reactor cell catch basin to the drain tank. These valves would not be opened except in the unlikely event of a salt spill from the primary system and are sealed by a thin membrane. If salt entered the pipe upstream of the valves, a signal from an electrical conductivity probe would cause a spring-loaded actuator to rupture the membrane and permit the collected salt to enter the drain tank. A normally open mechanical-type valve is installed downstream from each of the rupture-type 25 valves to isolate them from the drain tank in the event it becomes necessary to replace the rupture disks. The bottom of the above-mentioned valve pit is also provided with a catch pan, drain line, and rupture-disk- type drain valve, 3.2.6 Heat Exchangers Addition of the third salt-circulation loop, as dis- cussed in Sect. 3.2.2, not only required a conceptual design for the new secondary heat exchangers but, through changes in the working fluids, made it neces- sary to redesign the primary heat exchangers and the steam generators. Also, in order to stay closer to MSRE experience, we decided to reduce the maximum salt- temperature in the MSDR from 1300 to 1250°F. A further aspect was a change in the maintenance philoso- phy to one in which we propose to plug failed tubes . rather than to replace an entire tube bundle as was previously planned. The properties of the fuel salt and the two coolant salts used in the heat exchanger calculations are given in Table 3.1. The design data for the primary heat exchanger are given in Table 3.2, and a sketch of the Table 3.1. Physical properties of the fuel and coolant salts used in the MSDR Fuel salt Composition Density, Ib/ft> Viscosity, Ib hr 1 ft~! Specific heat, Btu 1b~! (°F) -1 Thermal conductivity, Btu hr~! ft -1 (°F)~! LiF-BeF,-ThF 4-UF 4 (71.5-16.0-12.0-0.5 mole %) 236.3 — 2.33 X 107 2T (°F) 7362 0.2637 exp —————— 637 eXP 9.7+ T CF) 0.324 0.75 LiF-BeF;, coolant salt Compeosition Density, 1b/ft> Viscosity, Ib hr ™1 ft~! Specific heat, Btu Ib~! (°F)"! Thermal conductivity, Btu hr~! ft ! (OF) -1 TLiF-BeF; (66-34 mole %) (99.99+% "Li) 138.68 — 1.456 x 10727 (°F) 6759 0.2806 exp —— 2 —— *P 459.7 + T °F) 0.57 0.578 Nitrate-nitrite coolant salt Composition (eutectic) Density, 1b/ft> Viscosity, 1b hr ! ft! Specific heat, Btu Ib~! (°F)-! Thermal conductivity, Btu hr ! ft~1 (°F) 1 KNO3-NaNQ;-NaNQ 3 (44-49-7 mole %) 130.6 — 2.54 X 1027 (°F) 3821.6 0.1942 exp ————>— P 4597+ TP 0.373 0.33 ORNL-DWG 71-5034 26 16-in. SLEEVE PUMP COOLING, 2in. SYSTEM DRAIN, 6in. SYSTEM FILL, 2in. OFF-6AS, 1Y in. VALVE- CELL, 64-in. DIAM {7 PaCiy! % 0 EMERGENCY DRAIN VALVES 4... a%eled 22 1 4 ! f L r I ! A (Y g KO N s AN - ——— MECHANICAL VALVE {NORMALLY CPEN) EMERGENCY DRAIN, 8in. _-.h THERMAL INSULATION -—\J SYSTEM DRAIN VALVE | CATCH PAN <> XK A A RIRIESK TG FROM FILL PUMP \ TO DRAIN TANK Fig. 3.2. Containment cell for valves in salt drain lines. 27 Table 3.2 MSDR primary heat exchanger design data Type Rate of heat transfer per unit MW Btu/hr Tube-side conditions Hot fluid Entrance temperature, °F Exit temperature, °F Pressure drop across exchanger, psi Mass flow rate, Ib/hr - Shell-side conditions Cold fluid . Entrance temperature, F Exit temperature, °F Pressure drop across exchanger, psi Mass flow rate, Ib/hr Tube material ‘Tube OD, in. Tube thickness, in. Tube-sheet-to-tube-sheet distance, ft Shell material Shell thickness, in. Shell ID, in. Tube sheet material Number of tubes Pitch of tubes, in. Total heat transfer area, ft2 Basis for area calculation Type of baffle Number of baffles Baffle spacing, in. Disk OD, in. Doughnut ID, in. Qverall heat transfer coefficient, U, Btu hr™! ft72 Volume of fuel salt in tubes, ft3 U-tube, U-shell, countercurrent, one-pass shell and tubes with disk-and-doughnut baffles 125 4.2687 x 108 Fuel salt 1250 1050 127.4 6.588 x 106 Coolant salt (2LiF-BeF,) 900 1100 114.7 3.744 x 10° Hastelloy N 0.375 0.035 29.96 Hastelloy N 0.5 26.32 Hastelloy N 1368 0.672 (triangular) 4023.6 Qutside of tubes Disk and doughnut 47 7.66 19.0 18.6 700.7 20.78 exchanger is shown in Fig. 3.3. Use of the " LiF-BeF, coolant salt in the shell side and the reduced tempera- ture difference between the fuel salt and the coolant salt altered the heat exchanger dimensions somewhat from those of previous concepts. In order to keep the shell diameter small, for ease of afterheat removal and maintenance, we found it desirable to use two primary heat exchangers in parallel in each of the three fuel-salt circulation loops. The exchangers have a bolted, in- - verted dished head over a seal-welded membrane. A metal ring gasket in the head flange provides backup containment in the event of a leak in the membrane. This design makes maintenance easier than in the previously reported concept that involved plasma arc cutting of a 1-in.-thick heat exchanger head. The secondary heat exchanger design data are given in Table 3.3. The configuration is like that of the primary heat exchanger. Since the LiF-BeF, heat transport circuits will become much less radioactive than the primary salt circuits, the remote maintenance require- ments for this exchanger are not nearly as restrictive as for the primary units. The different parameters that can now be considered in design of the steam generators were mentioned in Sect. 3.2.2. We are investigating conceptual designs for ORNL- DWG 71-5032 | SECONDARY SALT PRIMARY SALT INLET 1J\10-in. PIPE } —== | ft Bin, e - 16 ft 73 in. T~2ft 3%in. 13 ft ' - HEAT TRANSFER RATE: 125 MW TUBES: 1368 ’ TUBE SIZE: ¥g-0D x 0.035-in. WALL TUBE PITCH: 0,67 4 Fig. 3.3. MSDR 300-MW(e) primary heat exchanger. both 2400- and 3500-psi steam systems but as yet have no results to report. 3.2.7 Building and Containment The structures housing the MSDR are basically the same as previously reported,? but the main contain- ment building has been increased from 92 to 112 ft in diameter and the heat exchanger cells from 23 to 33 ft in diameter and to 38 ft in depth to accommodate the third salt-circulation loop, as shown in Fig. 3.4. The control room has been moved to outside the contain- ment area, and some adjustments have been made in the dimensions of other cells. The revised building plan and elevation are shown in Figs. 3.5 and 3.6. The reactor cell wall construction has been revised and the cell heating system has been changed. The new construction has % -in.-thick 304 stainless steel walls with about 5 in. of blanket-type insulation between the wall and the concrete biological shield, as shown in Fig. 3.4. The flat tops of the cells are hung by Unistrut hangers from the concrete. This design simplified the contain- ment construction to provide a maximum of freedom for thermal expansion and to present a smooth surface on the inside of the cell. The integrity is more assured both from the standpoints of inspection and access for repair. As in the previous design, the cell will operate at a slight negative pressure, but the structure would be designed for a positive pressure of 50 psi. Table 3.3. MSDR secondary heat exchanger design data Type Rate of heat transfer per unit MW Btu/hr Tube-side conditions Hot fluid Basis for area calculation Type of baffle U-tube, U-shell, counter- current, one-pass shell and tubes with disk-and- doughnut baffles 125 42687 X 108 . 2LiF-BeF, salt Entrance temperature, °F 1100 Exit temperature, °F 900 Pressure drop across exchanger, psi 80 Mass flow rate, Ib/hr 3.744 x 10° Shell-side conditions Cold fluid Hitec Entrance temperatgre, °F 700 Exit temperature, F 1000 Pressure drop across exchanger, psi 80 Mass flow rate, 1b/hr 3.815x 10 - Tube material Hastelloy N Tube OD, in. 0.375 Tube thickness, in. 0.035 Tube-sheet-to-tube-sheet distance, ft 37.5 Shell material Hastelloy N Shell thickness, in. 0.5 ‘Shell ID, in. 30.5 Tube sheet material Hastelloy N Number of tubes 1604 Pitch of tubes, in. 0.7188 (triangular) Total heat transfer area, ft? 5904 Outside of tubes Disk and doughnut Number of baffles 52 Baffle spacing, in. 8.65 Disk OD, in. 22.0 Doughnut ID, in. 21.6 Overall heat transfer coefficient, ' U, Btu hr ! ft =2 - 501 Volume of 2LiF-BeF, salt in tubes, ft3 30.5 Three external electrical furnaces and a circulating hot gas system are used to heat the reactor, heat exchanger, drain tank, and drain valve cells rather than the electric resistance-heated thimbles formerly em- ployed. The furnaces are located adjacent to the heat exchanger cells, as shown in Fig. 3.5. The design data for the heating system are shown in Table 3.4. The new BUILDING CONTAINMENT\ COOLING WATER PIPES——__ | L § [ ] CONCRETE : COOLING WATER OUTLET INSPECTION ACCESS PLUG PUMP SLEEVE PUMP SLEEVE \ ( ,/ ' ORNL—-DWG 71—5036 [CONTNNMENT HEAD FLEXIBLE SUPPORT RODS TO STEAM GENERATING SYSTEM N XA & 3 P % 4— CELL COOLING AND HEATING RETURN DUCT THERMAL {NSULATION CELL CONTAINMENT [~ COOLING WATER QUTLET HEAT EXCHANGER CELL | /PIPE SLEEVE REACTOR CELL ) CATATAN SN HEAT EXCHANGER SUPPORT. ) ;/ . N COOLING WATER PLENUM COOLING WATER INLET COOLING WATER INL ET e £ COOLING WATER CUTLET 4= ™ CELL CONTAINMENT REACTOR DRAIN LINE DA SO ST O TATAAA. ST T Z / \\ DEFLECTOR EMERGENCY DRAIN PORTS COOLING e WATER sl PIPES ¥ 4 REACTOR SUPPORT THERMAL INSULATION -—*" 4 / > il N AVAVAV. O ATAVAVLN. 90 6 6 0 6 8 44 AX A X P AT AVAV.N DX XK ~ XX k(ZOOLING WATER PLENUM Fig. 3.4. MSDR 300-MW(e) reactor and heat exchanger cells — élevafior_l. | INSPECTION ACCESS PLUG (3) 4— COOLING WATER INLET {3) CELL COOLING —+—— AND HEATING - _\SUPPLY DUCT (3) = COOLING WATER OUTLET (3) d___TO VALVE ANC DRAIN TANK CELLS 62 30 ORNL-DWG 71-3632 CELL HEATING SYSTEM PRIMARY AND SECONDARY HEAT EXCHANGER CELL NoK-WATER HEAT EXCHANGER CONTROL ROOM OFFICES AND AUXILIARY EQUIPMENT BELOW STEAM GENERATING CELL Fig. 3.5. MSDR 300-MW(e) reactor complex — plan section A-A. 31 ORNL-DWG 71-5035 [P 12 ft CONTAINMENT : | 16 ft - 176 ft Fig. 3.6. MSDR 300-MW(e) reactor complex — vertical section X-X. Table 3.4. Heater design data for MSDR containment cells Estimated normal containment heat loss (total), kW Design heat loss, kW Number of heater cells Capacity of each heater cell, kW Circulating gas Gas inlet-outlet temperatures at heaters, °F Gas flow rate through each heater cell, cfm Heater element Heater length, ft Number of heater elements per heater cell Heater element arrangement per heater cell Pressure drop in circulating gas, in. H,O Heater cell width and depth, ft 600 1200 3 400 - N2 1000-1100 32,500 (at 1050°F) 1-in.-OD cartridge with Incoloy 800 cladding 3 150 12rowsof 12 or 13 elements on-3-in. A pitch 9.6 3.25%X 2.6 Thermal conductivities for cell wall (k), Btu hr™! ft™! (°F) ! : 1/2 -in. stainless steel cell liner 5-in.-thick fiber glass cell insulation Prestressed concrete with 2-in. sched 40 carbon-steel water cooling pipes on 6 in. centers located 4 in. from inside concrete face 12.4 0.034 1.12 (concrete) 25.9 (steel) Assumed heat transfer coefficient in water pipes, Btu hr ™! ft =2 (°F)~! 7.9 Maximum concrete temperature, °F 150 arrangement reduces the number of cell wall penetra- tions and gives better access for repair of the electric heaters. The heater elements are inserted into double- walled thimbles welded into the furnace duct and are therefore easily removed for replacement. The same gas circulation system can be provided with cooling coils to remove excess heat from the cell. 3.3 INITIAL TEMPERATURE TRANSIENTS IN EMPTY MSBR “REFERENCE DESIGN” MSBR HEAT EXCHANGERS J. R. Tallackson Additional calculations® to estimate the initial after- shutdown temperature transients in these heat ex- changers were completed.® As in the earlier analyses, 40% of the noble metal fission products (Nb through Te) were assumed to be deposited uniformly on the heat exchanger tubes. Figure 3.7 shows the estimated 3. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622. 4. J. R. Tallackson, Thermal Radiation Transfer of Afterheat in MSBR Heat Exchangers, ORNL-TM-3145 (March 1971). after-shutdown temperature growth in the 563-MW(t) “reference design” heat exchanger. If used as a design guide, the estimated peak transient temperature, slightly less than 2150°F, is on the high side and should be regarded as conservative. This is because, during the initial phase of the transient, the elements comprising the exchanger, the tubes and inner shell, the inter- mediate shell, etc., were considered to be isolated and insulated from each other and from the outside. The method used to calculate this transient is described in detail in ref. 4. The curve in Fig. 3.8 applies to the 141-MW(t) unit. This curve was inferred from Fig. 3.7 and drawn by inspection. The peak temperature during the transient is expected to be about 1800°F. These peak temperatures, which take into account the effects of heat distribution and heat capacity, supersede the higher steady-state peak temperatures reported earlier.? The Hastelloy N heat exchanger shells, if not highly stressed, will safely withstand temperatures up to 2150°F for short periods. If the current estimates of fission product deposition remain unchanged or de- crease, it can be concluded that heat exchangers, not dissimilar to the “reference design” and with ratings in the 500-MW(t) region, can be designed to accommodate safely this worst-case afterheat situation. 33 ORNL-DWG 71-576A Curve A: Peak steady-state temperature computed for type 1 afterheat rates at the indicated times and with the emissivity.of all internal surfaces = 0.2 and the emissivity of the outer surface of the outer shelf = 0.8. Curve B: Temperature growth in the inner shell and the tube annuius computed as if; (1) the annulus and shell are perfectly insulated, (2) have a total heat capacity of 129 Btu/°F per foot of height,and (3) generate 77% of the total afterheat. Curve C: Temperature growth in the intermediate shell computed as if; (1) the shell is per- fectly insulated, (2) has a heat capacity of 287 Btu/°F per foot of height, and (3) generates 23% of the total afterheat. TN | R AU R CURVE B CURVE C 2500 \\\CURVE A 2000 : /E;:;;ATED o 3 TRANSIENT Y L 1 o l— L= o wl a. s i 1500 ve 1000 _TEMPERATURE OF INFINITE "BLACK" SURROUNDINGS, 100C°F 2.78 hr 27.8 hr 11.6 days S T T A Y O AR R A 10° 10° 10* 10° 108 ELAPSED TIME AFTER SHUTDOWN (sec) Fig. 3.7. Estimated initial temperature transient caused by noble metal afterheat in an empty 563-MW MSBR heat exchanger. TEMPERATURE (°F) 34 ORNL-DWG 71-577A Curve A: Peak steady-state temperature computed for type 2 afterheat rates at the indicated times and with the emissivity of all internal surfaces = 0.2 and the emissivity of the outer surface of the outer shell = 0.8. Curve B: Temperature growth in the inner shell and the tube annulus computed as if: (1) the annulus and shell are perfectly insulated, (2) have a total heat capacity of 32 Btu/°F per foot of height,and (3) generate 7 0% of the total afterheat. _ Curve C: Temperature growth in the intermediate shell computed as if: (1) the shell is per- fectly insulated, (2) has a heat capacity of 72 Btu/°F per foot of height, and (3) generates 23% of the total afterheat. 3000 ' / CLTTHHE T DTy TP rry 1T T TIT CURVE B CURVE C 2500 2000 \ ESTIMATED TRANSIENT 1500 7 . — 1000 TEMPERATURE OF INFINITE "BLACK" SURROUNDINGS, 1000°F 2.78 hr 27.8 hr 14.6 days SO I 1 A A A 1 O R W B A 11 102 10° 10* 10° 10° ELAPSED TIME AFTER SHUTDOWN (sec) Fig. 3.8. Estimated initial temperature transient caused by noble metal afterheat in an empty 141-MW MSBR heat exchanger. 3.4 THE CONSEQUENCES OF TUBING FAILURE " An analysis has been started to determine the consequences of tubing failure in one of the main MSBR heat exchangers. The following four cases -are being considered: R. P. Wichner =+ IN THE MSBR HEAT EXCHANGER 1. double-ended rupture near the fuel outlet, 2. double-ended rupture near the fuel inlet, 3. small leak into the primary system, 4. small leak into the secondary system. For case 1, it is estimated that 3.5 lb/sec of NaBF, will leak into the primary system until protective action, such as pump cutoff, is taken. It is not known if BF; gas will evolve in this case; though sufficient solubility of BF; in fuel exists to maintain BF; in solution, the kinetics of the mixing process are insuffi- ciently known to preclude this possibility. Hence case 1 is subdivided into (@) assuming no BF;(g) evolution and (b) assuming complete BF,(g) evolution. For case 15, prompt pump cutoff is essential to prevent excessive primary loop pressures and pump damage due to circu- lating voids. It is estimated that the poison effect of NaBF, is —0.18% &k/k per pound of NaBF, smeared over zones I and II of the MSBR core. This is overwhelmingly due to boron, the effect due to sodium being approximately 0.06% as large. | In addition, for case 1, an estimated 0.96 Ib,, [sec of fuel will leak into the secondary system until protective action is taken. For case 2, an estimated 7.4 Ib,,,/sec of fuel will initially leak into the secondary system, and there will be no coolant flow into the primary system. 3.5 TRITIUM DISTRIBUTION AND CONTROL IN THE MSBR R. B. Korsmeyer The transport behavior of tritium in a 1000-MW(e) MSBR was examined previously, and various hypothet- ical methods for its control and containment were considered.® In that study the tritium concentrations in the fuel and coolant salts were assumed constant around each circulation loop — a condition that is closely approximated only when the fraction of tritium that is removed per pass is small. Since then this restriction was removed, and the efficacy of preventing tritium from appearing in the steam system by purging it from side streams of fuel and coolant salts was examined, along with the effect of a circulating bubble fraction in both fuel and coolant loops.® It was found that a bubble fraction in the range of 0.01 to 0.1 circulating in the salt streams had 35 negligible effect on the tritium transport. Increasing the - purge gas flow rates increased the tritium removal from the side streams and markedly reduced that reaching the steam system for all side stream fractions. A purge gas flow sufficient to remove essentially all the tritium from a side stream of 10% of the salt flow through the reactor reduced the amount of tritium reaching the steam to 0.2% of the production. 5. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, p. 53. 6. R. B. Korsmeyer, The Effect of Purge Rate on Tritium Distribution in the MSBR, ORNL-CF-70-11-5 (Nov. 16, 1970) (for internal use only). In all cases where the tritium appearing in the steam was negligible, almost all of it was removed from the fuel as TF, which is assumed not to diffuse through the metal walls. Since these results depend strongly on a high UF,/UF, ratio to convert T, to TF according to the mass action relationship used and on the rates of reaction, of TF with the walls being low, it will be necessary to reexamine the supporting chemistry before reliance is placed on these results. In another study’ the effect of using wet helium under pressure as the secondary coolant was examined with reference to the distribution of the tritium flow in the purge gas systems, the cell enclosures, and the steam system by suitable modification of the transport equa- tions used in the TRIPOR program.’> The results of some 18 typical runs showed that: 1. With helium at high pressure and containing water vapor ‘at a partial pressure of 1 atm, the tritium reaching the steam is negligible provided the water vapor is changed at a rate equal to or greater than 0.2 Ib/hr. If radiolysis were to increase the hydrogen concentration in the steam by a factor as great as 100, the water purge rate would have to be increased to 20 Ib/hr to maintain the low tritium loss. It would be desirable to keep the purge rate below about 1 Ib/hr, but rates as high as 75 lb/hr might be acceptable. : 2. If the water vapor partial pressure is as low as 6 or 7 torrs, water purge rates as low as 0.02 lb/hr are effective in preventing significant quantities of tritium from reaching the steam. If radiolysis were to multiply the hydrogen concentration by a factor of 100, the purge rate would have to be increased to about 2 1b/hr. A radiolysis factor as high as 100 is considered to be very unlikely. In a third study® the effect of adding HF to the sodium fluoroborate coolant salt was examined with reference to the distribution of tritium flow in the coolant purge gas system and the steam system, utilizing a simplified model of the secondary coolant and steam systems instead of the more complete model and transport equations of the TRIPOR program. Use of the simplified model appears justified in that the calculated fraction of tritium transported to the steam in the absence of HF is comparable with that calculated for the same conditions in the TRIPOR program. The results of nine runs showed that: 7. R. B. Korsmeyer, Capture of Tritium in Helium—H,0 Coolant for MSBR's, ORNL-CF-70-12-9 (Dec. 11, 1970) (for internal use only). 8. R. B. Korsmeyer, Suppression of Tritium Transport to Steam in an MSBR by!HF Addition to the Coolant Salt, ORNL- CF-71-1-19 (Jan. 15, 1971) (for internal use only). 1. A ratio of HF addition to HT diffusion into the coolant salt of at least 50 is required to hold the / tritium reacliing the steam to 1% or less of that entering the coolant, the balance being carried off by the purge gas. 2. For large ratios of HF/HT addition the tritium fraction reaching the steam is inversely proportional to the ratio. : 3. Increasing the purge gas flow from 20 to 100 cfm, or by a factor of 5, has a negligible effect on the tritium transport, at least for large HF/HT addition ratios. The mean residence time for fluoride gases (HF + HT) in the circulating coolant salt is an inverse function of the purge gas flow and is independent of the HF feed rate inasmuch as an increased feed rate at constant purge rate merely increases the concen- tration proportionally. For all cases studied, a purge rate of 20 cfm established a residence time of 62.5 hr, and a rate of 100 cfm reduced the residence time to 12.5 hr. For HF/HT = 50 the corresponding HF concentrations in the salt were 0.60 and 0.12 ppm. 5. For an assumed corrosion rate equal to half of the HF feed rate to the system and HF/HT = 5000, the calculated uniform rate of metal removal is about 0.001 in./year. 3.6 INDUSTRIAL STUDY OF 1000-MW(e) MOLTEN-SALT BREEDER REACTOR M. 1. Lundin J. R. McWherter Proposals were solicited from a number of industrial firms to perform design studies of a 1000-MW(e) molten-salt breeder reactor. An evaluation team visited each group that submitted a proposal. The Ebasco Services group, consisting of Ebasco, Conoco, Babcock and Wilcox, Cabot, Union Carbide, and Byron-Jackson companies, was selected as the one that could most nearly meet our objectives. A subcontract is being negotiated with them. They will initially develop their concept of a molten-salt breeder reactor plant. Using this concept as a base, trade-off and parametric studies of the nuclear steam supply system, the energy conversion system, and the fuel processing system will be made. After incorpo- ration of the results of these studies in the reference concept, they will estimate the plant capital and fuel-cycle costs. A review of the research and develop- ment program will be made. An independent assessment of chemical processing and a safety review of the proposed plant will be conducted. Technical liaison is 36 being furnished by ORNL. All the work will be reported. 3.7 MSBE DESIGN M. 1. Lundin J. R. McWherter 3.7.1 General J. R.McWherter W, Terry. The design studies reported previously’ were con- tinued. A report'® was issued outlining the objectives of the MSBE and establishing the design bases of the plant. The reference core configuration consists of a cylindrical array of graphite bars. The core is 45 in. in diameter and 57 in. high, with 15% of the horizontal cross section open for salt flow. The core is in a 7.5-ft- ID spherical reactor vessel. That part of the vessel between the core and vessel wall, with the exception of the graphite alignment structure, is filled with salt. This configuration achieves a start of life breeding ratio near 1 (0.96) and the desired fast (£ > 50 keV) neutron flux of 5 X 10'* neutrons cm > sec ™' at a thermal power of 150 MW. Design data of the MSRE and MSBR and ‘that proposed for the MSBE are compared in Table 3.5. We are conducting studies to determine the problem areas .and to evaluate possible solutions. Layouts of the reactor cell and the primary system are being made to indicate how the operational and maintenance problems might be handled in different configurations. The current reactor cell concept is shown in Figs. 3.9 and 3.10. Removal of the graphite array as a unit and removal of individual graphite bars are being considered. The reactor vessel closure problems associated with the graphite maintenance are being evaluated. It is desired that the graphite maintenance approach be similar to that proposed for the MSBR. The thermal and hy- draulic problems of the core are being examined for use of cylindrical and prismatic graphite bars. The support and alignment structure of the graphite bars is being studied. 3.7.2 MSBE Core Design W. K. Furlong W. Terry A brief investigation was made of alternatives to four previously published”+'® MSBE design concepts, speci- 9. MSR Program Semiannu. Progr. Rep. Feb. 28, 1969, ORNL-4396, pp. 71-75. 10. J. R. McWherter, Molten Salt Breeder ExperzmentDeszgn Bases, ORNL-TM-3177 (November 1970). fically: (1) the prismatic moderator elen..nts, (2) the removal and replacement of all the elements as a unit, (3) the unreflected core, and (4) the three-pass flow pattern. The results of these investigations are sum- marized in the following paragraphs. A cyhndrlcal graphite moderator element design was evolved. A 4-in.-diam cylinder with a 1-in. central hole appears acceptable, based mainly on temperature con- 37 siderations. The cylinders are separated by ", ¢-in.-thick collars at three axial locations. This eliminates the low-flow cusp region at points of tangency (except at the narrow collars) and permits interchannel mixing of the salt. The center hole is required to have acceptable temperatures with a 4-in. element. The salt fraction is 20.5% in an array of such elements with 4 in. triangular pitch. Table 3.5. Comparison of design data for the MSRE, MSBE, and MSBR - MSRE MSBE MSBR Reactor power, MW(t) 7.3 150 2250 Breeding ratio 0.96 1.06 Peak graphite damage flux (£, > 50 keV), neutrons cm ™2 sec ™! 3x 1013 5 x 1014 3x 1014 Peak power density, w/cc, core including graphite 6.6 114 70 Volume fraction of salt in core 0.225 0.15 0.13 Primary salt Composition, mole % LiF 65 71.5 71.7 BeF, 29.1 16 16 ThF,4 None 12 12 UF, 0.9 0.5 0.3 Z1F,y 5 None None Liquidus, ° 813 932 932 Density, 1b /ft at 1100°F 141 211 210 Viscosity, lb ft™ hr! at 1100°F 19 29 29 Heat capacity, Btulb 1ept 047 0.32 0.32 Thermal conductivity, Btu hr ™! ft™1 ¢F)™! 0.83 0.71 0.71 Temperature, °F ) Inlet, reactor vessel 1170 1050 1050 Outlet, reactor vessel 1210 1300 1300 Circulating primary salt volume, ft3 70 266 1720 Inventory, fissile, kg 322 3962 1470 Power density, primary salt, circulating, average, W/cc 4 20 46 Number of primary loops 1 1 4 Primary pump capacity, gpm 1200 5400° 16,000 Secondary system salt LiF-BeF, " NaBF4-NaF NaBF;-NaF Composition, mole % 66-34 92-8 92-8 Liquidus temperature, °F 850 725 725 Density, Ib,, /ft® at 1000°F 124:1 117 117 Viscosity, i, /ft ™1 hr ™! at 1000°F 28.7 3.4 3.4 Heat capacnty, Btulb,, ™' CF)™! 0.57 0.36 0.36 Thermal conductmty, Btuhr™! ft™! CF)! 0.58 0.23 0.23 Temperature, °F Heat exchanger inlet 1015 850 850 Heat exchanger outlet 1075 1150 1150 Number of secondary pumps 1 1 -4 Secondary pump capacity, gpm 850 5300 - ‘ 20,000 Tertiary system Air _ Steam Steam Inlet temperature, °F ~70 700 700 Outlet temperature, °F ~180 1000 1000 Outlet pressure, psia 14.7 3600 3600 9MSR Program Semiannu. Progr. Rep. Feb 28, 1 970 ORNL-4548 pp. 42-45. 2334 initial. “For 200°F AT; 4300 gpm required at 250°F AT, - 38 ORNL~DWG 71-6T710 90-1t DIAM - wa v i L AIR LOCK PRIMARY PUMP — SECONDARY PUMP } i i A HOT CELL STEAM o BLOCK VALVES SUSZS R . [Ei I-lllr- 2 STEAM GENERATOR 60 ft Oin NN\ ———— N —— PRIMARY SALT DRAIN TANK . k‘.".'.v'.‘v.'[ viwieratalsl | REACTOR VESSEL HEAT EXCHANGER SECONDARY SALT DRAIN TANK Fig. 3.9. Molten-salt breeder experiment, reactor building section B-B, 150 MW(t). A major feature of cylindrical elements is that they do not interlock (as the prismatic elements do) and hence can be moved sideways relative to one another by a handling machine stationed in the center of the core. This idea was pursued through the conceptual design of a moderator element handling machine. The machine has two telescoping arms operating on the jackscrew principle. It can rotate azimuthally, move up and down, and can operate such that elements are removed and replaced either along a radius or offset 4 in. from a radius. The latter position is required for a few elements that will not move along a radius without forcing or jamming them. Exact indexing in all degrees of freedom is obtained by use of worm and sector gears. They are driven either by cables or shafts from electric motors located in a relatively cool, shielded area. The machine is inserted into the core after first rcmoving the center 16 elements plus three control rod sleeves. These are removed through a 20-in. flanged access port by a straight lift using an expandable tool. The flange serves to support and align the machine, which then removes each element, including the noncylindrical “‘fence” pieces, to a point where it can be grasped and lifted clear of the vessel by an expandable tool. Further refinements of the machine design may incorporate an elevator mechanism in place of the “manual” expand- able tool. Several reactor and vessel layouts were studied using the cylindrical elements both with and without the need for the handling machine. In the latter case a large (~48 in.) removable head is required. One such layout - used a cylindrical vessel with a 1-ft graphite reflector. It appeared feasible to build such a reactor, although, once installed, the reflector would not be removable. Because of this and several other considerations, the decision was made to use an unreflected reactor in a spherical vessel.! ° One-pass flow through the reactor vessel was used in this most recent study. This has advantages over the previously proposed three-pass (up through blanket, down through outer part of core, up through center of core) arrangement because (1) downflow in the three- pass concept requires a velocity of 2.4 fps to overcome the buoyancy effect for a salt AT of 125°F, (2) the PRIMARY SALT DRAIN TANK 90 ft Qin. REACTOR VESSEL 39 —— ORNL-DWG 71-6711 GAS PROCESSING CELL STEAM CELL STEAM BLOCK VALVES | - HEAT EXCHANGER SECONDARY SALT DRAIN TANK Fig. 3.10. ' Molten-salt breeder éxperiment, reactor building plan A-A, 150 MW(t). resulting mechanical layout is simpler, and (3) the reactor pressure drop is lower. A drawback is the lower velocity and correspondingly lower heat transfer coef- ficients for cooling the vessel wall and the moderator elements. ' 3.7.3 MSBE Primary Heat Exchanger Design H. A.McLain C.E.Bettis W. Terry The heat transfer area required in the primary heat exchanger was determined for core inlet and outlet temperatures of 1050 and 1300°F respectively. It was assumed that the primary salt was in the tubes and that a secondary salt was on the shell side. A number of tube and shell configurations are being considered in the studies of the primary cell arrangement. In the heat exchanger studies, the primary sait inventory is being kept as small as practicable. " It was reported previously that spirallyindented heat exchanger tubes would reduce the inventory of primary salt.' ' However, they are not being used in the MSBE design for’ the following reasons: (1) there are no experimental data for heat transfer to molten salts flowing in spirally indented tubes at Reynolds numbers of interest'and (2) initial fabrication work on enhanced tubes of “standard” Hastelloy N was troubled with cracking resulting from the carbide stringers present in the metal.!? - B The principal data for a 150-MW MSBE shell-and-tube primary heat exchanger are given in Table 3.6. 11. MSR Program Semiannu. Progr. Rept. Feb. 28, 1969, ORNL4396, p. 57. 12. MSR Program Semiannu. Progr. Rep. Aug. 31, 1968, ORNL-4344, p. 289. Table 3.6. Principal design data for 150-MW MSBE primary heat exchanger Thermal rating, MW Tube-side conditions Fluid Tube OD, in. Tube wall thickness, in. Tube length, ft Number of tubes Temperatures, °F Inlet Qutlet Mass flow rate, 1b/hr Pressure drop due to flow, psi Volume of fluid in tubes, ft3 Shell-side conditions Fluid Shell ID, in. Baffle type Baffle spacing, in. Baffle cut, % Tube pitch, in. Temperature, °F Inlet Outlet Mass flow rate, Ib/hr Pressure drop due to flow, psi Tube natural frequency, cps Vortex shedding frequency, cps Frequency ratio Approximate overall heat transfer coefficient, Btu hr™! ft 2 (°F) ! 150 Primary salt ¥ 8 0.035 . 28.0 1340 1300 1050 6.32 X 10° 111 19.0 Secondary salt 31.5 Disk and doughnut 7.3 40 1316 triangular Considerable attention was given to reducing the possibility of having a tube vibration problem. There- fore, one design criterion was that the ratio of the natural frequency of the tube to the shell-side vortex shedding frequency must be 2 or greater. Because of this, the tube pitch is greater and the baffle spacing is smaller than are normally used in commercial heat exchangers. The 28-ft tube length is about the same as is proposed for the MSBR primary heat exchangers. Use of the same tube length is desirable to permit investiga- tion of such factors as corrosion, fission product deposition on the tubes, bubble behavior, and transport of tritium through the tube walls. Layout studies of this unit are in progress, with some consideration being given to the fabrication, reliability, and maintenance. The maintenance approach is to be able to locate a leaking tube and plug the tube with the heat exchanger in place in the reactor cell. In addition, it-is to be possible to remove the entire heat exchanger for repair or replacement. These studies so far indicate that the best arrangement is to have a U-shaped horizontal primary heat exchanger located near the top of the reactor cell. 4. Reactor Physics A. M. Perry 4,1 PHYSICS ANALYSIS OF MSBR 4.1.1 Single-Fluid MSBR Reference Design H.F.Bauman ' The nuclear performance of the reference-design single-fluid MSBR has been calculated again to take into account the effects of the recently adopted metal transfer process on the concentrations of various nuclides in the fuel salt. The last set of calculations for the reference design (case CC-120), reported a year ago,! was based on a reductive-extraction process which removed protactinium on a 3-day cycle, europium on a 500-day cycle, and other rare earths on a 50-day cycle; plutonium was assumed not to be removed, though in fact it probably would be removed (in a hydrofluori- nation step in the process). Thus the assumption is equivalent to assumning recycle of plutonium as well as uranium. The new set of calculations assumed that the fuel salt circulates through the processing plant on a 10-day cycle, with removal efficiencies and effective removal times that depend on the element. In addition, a few minor data corrections and cross-section changes were included in the later calculations. In calculating the concentrations of noble gases and daughters, we assumed, as in the earlier calculation, that xenon and krypton are removed by gas stripping on a 50-sec cycle. Also, as before, a fixed poison fraction of 0.005 was assumed to allow for absorption of ! 3° Xe by the core graphite. Except for the elements listed in Table 4.1, there is complete removal of fission products from the salt as it passes through the processing plant on a 10-day cycle. Also removed completely are plutonium and protactinium. Two important neutron poisons, 237"Np and 23°U, were assumed not removed in processing but were assigned 16-year removal times to simulate their average concentrations over the lifetime of the reactor. As in previous calculations the 1. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL4548, p. 58. 41 Table 4.1. Removal efficiencies and effective removal times for rare earth and active metal elements in the metal transfer process Effective Element Removal 16 efficiency removal time (days)? Rb 1.0 10.0 Sr 0.595 16.8 Y 0.333 30.0 Cs 1.0 10.0 Ba 0.595 16.8 La 0.452 22.1 Ce 0.602 16.6 Pr 0.333 30.0 Nd 0.333 30.0 Pm 0.340 294 Sm 0.372 26.9 Eu 0.195 51.2 Gd 0.333 30.0 9For a ten-day processing cycle. effects of deposition of noble metals in the core and the burnout of boron in the graphite were neglected, since the latter slightly more than compensates for the former. Carrier salt was discarded on a 4200-day cycle. The description of the reactor given in Table 4.2 is the same as in the last set of calculations (CC-120) except for the absence of plutonium in the fuel. There were several differences from the earlier cal- culation besides the change in processing. A key-punch error was corrected in the 233Pa decay constant (from 2.6279 X 1077 sec™! to 2.9279); this correction makes a negligible change in the neutron balance. The same spectrum-weighted cross-section set was used for the annulus as for core zone 2; this set matches the neutron spectrum in this narrow region better than the set weighted for an infinite 100% salt region used pre- viously. The same cross-section set was used for the reflector as for core zone 1. This set matches the spectrum in the reflector reasonably well and avoids an error in the thorium resonance group cross section which was uncovered in the set previously used for the Table 4.2. Characteristics of the single-fluid MSBR reference design with processing by the metal transfer process on a ten-day cycle A. Description Identification Power MW(e) MW(t) Plant factor Dimensions, ft Core zone 1 Height Diameter Region thicknesses Axial Core zone 2 Plenum Reflector Radial Core zone 2 Annulus Reflector Salt fraction Core zone 1 Core zone 2 Plenums Annulus Reflector Salt composition, mole % UF,4 ' PUF3 ThF, BCF2 LiF B. Performance Conservation coefficient, [MW(t)/kg] ? Breeding ratio Yield, % per annum (at 0.8 plant factor) Fissile inventory, kg Specific power, MW(t)/kg System doubling time, years Peak damage flux, £ > 50 keV, neutrons cm™? sec”! Core zone 1 Reflector Vessel Power density, W/cm?3 Average Peak Ratio 16.2 1.071 3.63 1487 1.51 19 3.5 x 1014 2.9 x 1013 3.3x 10! 22.2 70.5 3.18 ° Table 4.2 (continued) Fission power fractions by zone Core zone 1 Core zone 2 Annulus and plenums Reflector ~ C. Neutron balance Absorptions 232Th 0.9968 233p, 0.0045 233y 0.9242 234y 0.0819 235y 0.0758 236y 0.0074 237Np 0.0064 238py 0.0001 239py 0.0 240py 0.0 241py : 0.0 242py 0.0 Li : 0.0032 TLi 0.0161 Be 0.0070 19g 0.0203 Graphite 0.0510 Fission products 0.0150 Leakage® 0.0221 ne 2.2318 D. Fuel-cycle costs® (mills/kWhr) Inventory Fissile Salt Replacement salt Processing Fissile production credit Total 0.791 0.150 0.050 0.009 Fissions 0.0030 0.8245 0.0004 0.0618 0.00444 0.382 0.060 0.034 0.293 -0.104 0.665 %(n,2n) reaction. bInciuding delayed neutron losses. €At 13.2%/year inventory charge on materials, 13.7%/year fixed charge rate on processing plant, $13.8/g 233U, $11.9/g 235y, $12/kg ThO,, $120/kg Li, $26/kg carrier salt (including - 7Li). Excluding graphite replacement cost. reflector. Both these cross-section changes would be expected to enhance the absorptions in thorium relative to uranium and hence to give a higher breeding ratio. However, these changes were in minor regions of the reactor, and the effect on the calculated reactor performance is estimated to be small relative to the effect of the change in processing. The results indicate a higher breeding ratio, a lower fissile inventory, and a higher conservation coefficient for the reference MSBR with the metal transfer process, compared to the same reactor with the reductive- extraction process (1.071 vs 1.063, 1487 kg vs 1504 kg, and 16.2 [MW(t)/kg]? vs 14.1 [MW(t)/kg]? respec- tively). The most important differences in the neutron balance are the absence of absorptions in plutonium, reduced absorptions in fission products, and a higher value of 7ne, because a higher proportion of the fissile material is 233U, The fuel-cycle cost in Table 4.2 is on a different basis from the earlier calculation. The fissile material charge is based on 235U valued at $11.9 per gram. The inventory charge is taken as 13.2% rather than 10%, and no allowance is made for graphite replacement [this allowance has previously been estimated to be about 0.1 mill/kWhr(e)]. On the same basis, the fuel-cycle cost for case CC-120 is 0.70 mill/kWhr. 4.1.2 Fixed-Moderator Molten-Salt Reactor H. F. Bauman The use of the ROD and the HISTRY codes for the calculation of molten-salt reactors with batch proc- essing was described in the last semiannual report.? An iterative process is used, in which reaction rates per atom for each type of nuclide are obtained from ROD and supplied to HISTRY, where nuclide concentrations as a function of time are calculated over one or several batch processing cycles; averages of these concen- trations over the reactor life are then computed and returned to ROD for another calculation of reaction- rate coefficients. The process is continued until ROD and HISTRY converge on a common set of average concentrations. This process has now been automated by including the HISTRY code as a subprogram in ROD. The average performance of a reactor with batch processing may now be calculated in a single case, rather than in four to six separate cases required by the old method. The results obtained are the same, but the chance of error in transcribing data in the iterative 2. MSR Program Semignnu. Progr. Rep. Aug. _31, 1970, ORNL4622, p. 28. : ’ 43 Table 4.3. Fixed-moderator 1000-MW(e) molten-salt breeder reactor Core graphite life, 30 years Effective processing cycle time: protactinjum removal, 10 days; rare earth removal, 25 days Plant size, MW(e) 1000 Thorium concentration in fuel salt, mole % 14 Case identification (SCC series) 201 Volume fraction salt in core Zone 1 0.137 Zone 2 0.111 Zone 3 0.127 Thickness of core zones ? ft Zone 1 8.46 Zone 2 3.64 Zone 3° 1.97 Annulus 0.14% Core diameter, overall, ft 28.5 Fissile inventory, kg 2220 Breeding ratio 1.069 Conservation coefficient, [MW(t)/kg] 2 7.0 Fuel yield, %/year (at 0.8 plant factor) 2.34 Fuel-cycle cost,© mills/kWhr 0.85 9Three-zoned core with zone thicknesses and volume fractions optimized for maximum conservation coefficient by ROD calcu- lations in spherical geometry. bNot optimized. ©Based on 235U at $11.9 per gram and an inventory charge of 13.2%. process has been eliminated. The cases in the following section were calculated with the new ROD-HISTRY program. | A study of the fixed-moderator MSBR with con- tinuous processing by the metal transfer process was reported in the last semiannual report. The best 1000-MW(e) case from this study was SCC-201, a three-zone-core reactor with 14 mole % thorium fuel salt. The characteristics of this reactor are given in Table 4.3. We have made a brief study of the performance of this same reactor operated as a con- verter with batch processing. We assumed plutonium feed of a composition typical of first-cycle discharge from a light-water reactor. We examined thorium concentrations lower than used for the continuous processing case, anticipating that this would give lower fissile inventory and fuel-cycle cost. Since there were no processing restrictions on the salt composition, we elected to use the minimum-liquidus-temperature com- positions given in Table 4.4. Batch processing times of six and eight years were used; we know from earlier cases that the optimum range of batch cycle time is about four to eight years. The results of the four cases in this study are shown in Table 4.5. The fuel-cycle costs are based on a fuel and salt inventory charge of 13.2%, rather than the 10.0% used in previous cal- culations. Cases A-13 and -14 show the effect of the batch- processing cycle time, with the salt composition held constant. The reactor lifetime is assumed to be 30 years at 0.8 plant factor, or 24 fuil-power years. It is convenient to assume a refueling program of either three 8-year cycles or four 6-year cycles. The 6-year cycle appears at first glance better than the 8-year; it has a higher conversion ratio, a lower fissile inventory, and essentially the same fuel-cycle cost. However, it has Table 4.4. Fuel carrier salt compositions for minimum liquidus temperature .Salt number 14/17 12/20 10/23 8/27 Composition, mole % ThF4 14 12 10 8 BeF, 17 20 23 27 LiF ‘ 69 68 67 65 Liquidus temperature,? °C 495 485 475 465 9R. E. Thoma (ed.), Phase Diagrams of Nuclear Reactor Materials, ORNL-2548, p. 80 (November 1959). 44 a higher peak damage flux and therefore a lower core life (0.73 relative to case SCC-201), although this apparent disadvantage could probably be overcome by a slight change in salt volume fractions in the three core zones. Cases A-14 through A-17 have the same cycle time but different thorium concentrations. The conversion ratio, fissile inventory, fuel-cycle cost, and relative core life all increase with increase in thorium concentration. The lifetime fuel composition for the 12% thorium case is given in Table 4.6. The base reactor (SCC-201) was optimized with a salt containing 14 mole % thorium so as to have a flat flux distribution in core zone 1. When other salt compo- sitions are substituted in this design, the flat flux distribution is no longer obtained, and for this reason the cases with lower thorium concentrations tend to have higher peak fluxes and therefore shorter-lived cores. We believe that this penalty on core life can be largely avoided by reoptimizing the core design for the desired salt composition; our purpose in this study, however, was to calculate the effect of changing the fuel composition in a reactor of fixed core design. We conclude from this study that the fixed-moderator MSBR design is attractive when operated as a converter Table 4.5. Lifetime averaged performance of a 1000-MW(e) fixed-moderator MSBR operated as a converter reactor with batch processing and plutonium feed Base reactor design: SCC-2019 Identification Thorium concentration in fuel salt, mole % Carrier salt? ‘Processing cycle time, batch, years Average conversion ratio® Specific fissile inventory, kg/MW(e) Peak damage flux (£ > 50 keV), 10!3 neutrons cm 2 sec™! Relative core lifed : Fuel-cycle cost, mills/kWhr Inventory, at 13.2% Fissile Salt Replacement® Processin, Fissile feed® Total fuel-cycle cost A-13 A-14 A-16 A-15 A-17 10 10 8 12 14 10/23 10/23 8/27 12/20 14/17 8 6 6 6 6 0.88 0.90 0.85 0.92 0.93 1.64 1.60 1.32 1.89 2.09 5.74 6.15 6.27 5.29 4.73 0.78 0.73 0.72 0.85 0.95 0.317 0.322 0.249 0.394 0.443 0.048 0.045 0.044 0.048 0.049 0.060 0.080 0.075 0.083 0.087 0.033 0.036 0.036 0.036 0.036 0.015 0.093 0.147 0.059 0.043 0.573 0.576 0.551 0.620 0.658 2Refer to Table 4.3. bRefer to Table 4.4 €Assuming that the plutonium discarded with the fuel salt is not recovered. 4Based on the 30-year core life of the base reactor = 1.0. €Replacement includes thorium and carrier salt. fNormalized to 0.1 mill/kWhr for a 300-MW(e) plant on a 4-year cycle; unit cost assumed to vary inversely as the 0.3 power of the average processing rate. 1 8First-cycle plutonium; composition 60% 239Pu; 24% 24%Pu; 12% 241 Pu; 4% 242 Pu. Valued at $9.88 per gram fissile. 45 ‘Table 4.6. Lifetime fuel composition and conversion ratio for a fixed-moderator 1000-MW(e) MSR with plutonium feed for four 6-year cycles Case A-15, 12 mole % thorium Time (full-power years) Inventory (kg) Conversion Percycle Cumulative 233pa 233y 234y 235y 236y 237Np 239py 240py 241p, 242py, ratio? 0 0 0.0 0.0 0.0 0.0 0.0 0.0 6835 2745 1378 . 46.1 0.768 1 1 58.7 4419 118 0.3 0.0 0.0 4752 4877 336.6 146.5 0.861 2 2 63.0 804.0 375 2.0 0.1 0.0 250.7 4049 3715 246.1 0.919 4 4 733 1291.1 1243 124 0.8 0.0 107.5 1946 229.0 390.4 0.909 6 6 75.0 1526.7 230.7 325 3.6 0.1b 89.8% 131.7% 141.5% 434.5% 0.880 0 6 0.0 16019 230.7 325 3.6 0.0 29.0 11.6 5.8 1.9 1.063 1 7 © 89.5 16062 2928 47.0 6.6 0.2 53.5 64.0 49.4 25.1 0.980 2 8 88.5 1659.4 3504 619 106 0.5 33.1 51.2 50.7 45.4 0.970 4 10 85.2 1699.2 4503 912 21.3 1.4 345 446 43.4 74.2 0.938 6 12 81.5 1698.1 527.6 1169 349 2.6b 41.7% 5220 4872 97.0% 0.913 0 12 0.0 17799 5276 1169 349 0.0 - 00 0.0 0.0 0.0 1.020 1 13 91.9 17135 563.5 1293 433 1.6 19.2 18.7 13.1 5.9 0.993 2 14 89.1 1726.7 5954 1403 521 3.0 18.8 22.8 19.6 ' 14.0 0.974 4 16 84.7 1725.1 6473 1588 703 5.9 27.2 32.3 28.9 33.4 0.941 6 18 80.8 1708.0 685.1 173.3 888 8.6° 37.4b 45.1b 407 5745 0.915 0 18 0.0 1789.1 6851 1733 888 0.0 0.0 0.0 0.0 0.0 1.009 1 19 90.7 - 1719.4 7023 1797 990 3.7 16.0 14.5 9.8 4.2 0.992 2 20 87.9 1730.3 718.3 1854 109.0 6.8 17.7 20.2 16.8 11.1 0.972 4 22 83.6 1726.7 7439 1948 . 1282 11.8 27.1 31.7 27.9 29.7 0.939 6 24 79.9 1709.0 761.6 202.0 146.6 1580 379Y% 453> 407P s54.0% 0913 4Not adjusted for discard of plutonium, bQuantity discarded at end of cycle. with batch processing. The results suggest that if the reactor is to be designed for dual-mode operation (e.g., initially as a converter with batch processing and subsequently as a breeder with continuous processing), the salt composition should be essentially the same for either mode of operation. In particular, the thorium concentration cannot be changed significantly without altering the peak flux and the core graphite life. 4.2 MSR EXPERIMENTAL PHYSICS 4.2.1 HTLTR Lattice Experiments G.L.Ragan- O.L. Smith The High-Temperature Lattice Test Reactor (HTLTR) is being used to study the reactor physics characteristics of a lattice that simulates a typical molten-salt reactor. Reactor measurements began in February and are to be completed in May. The experiments are being spon- sored by the Physics Branch of DRDT and performed by Pacific Northwest Laboratory (PNL). They were planned by ORNL, in consultation with PNL; data reduction and analysis will be similarly shared. Sub- sequent use of the experimental results to test MSR computational methods is the responsibility of ORNL. Gulf General Atomic has cooperated with PNL on similar lattice experiments related to high-temperature gas-cooled reactor (HTGR) work. Although the HTLTR lattices related to HTGR differ from those for the MSR in several important details, they are closely related, and we plan to study them carefully. To this end, there are continuing consultations between Gulf General Atomic and ORNL personnel. The HTLTR consists of a 10-ft cube of graphite surrounded by a gas-tight insulating oven. The central 5-ft-square by 10-ft-long section can be removed and loaded with as much as necessary of the test lattice of interest. The driver fuel and the control system are loaded outside the test lattice. One measures the reactivity changes aécom‘panying a ‘series of pertur- bations made in a 7.5 X 7.5 X 24 in. test cell volume at the center of the lattice. These and other data are used to determine certain lattice characteristics, especially a suitably defined lattice multiplication factor k. Meas- urements can be made from room temperature to 1000°C. ' An extensive set of measurements is first made at room temperature. The cadmium ratio of gold is measured at several positions in and near the test cell volume, as an indicator of spectrum matching. Loading adjustments are made until the spectrum matching is satisfactory. Foils of other materials are then activated in various positions to obtain data needed in inter- preting later measurements. One result of these acti- vation measurements, coupled with a measurement of k. for the main lattice, is an experimental determi- nation of the effective value of 7(??3U). Because of its importance in evaluating the breeding ratio of MSBR design proposals, 7(33U) will be measured as precisely as possible. However, the accuracy with which n can be determined from these measurements is intrinsically not very high, and it is not clear, at present, whether the resulting value will be any more reliable than one obtained from basic cross-section data and calculated neutron spectra. No attempt was made in designing the HTLTR experiments to achieve an exact mockup of details of current MSBR designs. Instead we use a simple lattice having neutronic characteristics comparable with those of current designs and make certain parametric vari- ations on this basic lattice. Thus we obtain information applicable to possible future changed designs and on the sensitivity of the reactor- physics characteristics to such parametric changes. For each configuration, we will test our calculational techniques and cross-section data by making critical comparisons between calculated and measured quantities. These tests of calculational ability constitute the most important objective of the ex- periment. Although the lattice finally adopted is neutronically similar to that of a typical MSR, it is quite different physically. Fuel channels are simple 0.786-in.-diam rods spaced 1.875 in. apart in a square lattice array. Instead of using molten salts as fuel, we use an appropriate mixture of solid materials: pyrocarbon-coated particles (containing ?*3UQ, and some ThO, ), powdered ThO,, and powdered graphite. These are the same materials that were used for the HTGR experiments but with proportions chosen to give the 222 U-to-moderator ratio and 233U-to-Th ratio typical of MSR designs. Com- pared with the use of actual molten salts in the test lattice, the advantages in availability and ease of handling are evident. Equally evident are shortcomings in simulating MSR fuels: the much lower thermal expansion coefficients of the solid fuel and the omission of significant absorbers such as Li, Be, and F. To compensate for the omission of absorbers, the effective cross section for each omitted nuclide will be determined from central reactivity worth measurements made on a sample containing that nuclide. We can thus evaluate each absorber separately, instead of lumping it with others. Fuel expansion differences are evaluated as 46 follows. By comparing the worth of a special test block, containing fuel of the same composition but more densely packed, with the worth of a test block containing standard fuel, we will be able to determine the fuel density coefficient of reactivity at each temperature. This is useful information in itself and also permits evaluation of the temperature coefficient of reactivity that would have been obtained using a molten salt fuel, with its known higher expansion coefficient. Test temperatures selected are room temperature, 300°C, 627°C, and 1000°C. At each temperature the reactivity worths outlined in Table 4.7 are to be measured. Series A includes insertion of a standard block into the central void (to get k), replacement by a copper foil (for calibration), and replacement by blocks having various perturbations in fuel density, fuel channel geometry, and fuel volume fraction. Other blocks simulate control rods of the “displacement” type (graphite displaces fuel) and of the poison- shutdown type (Hastelloy N plates). In series B, material worth measurements are to be made on samples containing Cu (for calibration); Be (as solid rod); F (in Teflon rod, empirically CF,, at T < 300°C); "Li (in solid " LiF salt, at T < 627°C); Hastelloy N (as a foil); and ThO,, 233U, ?3°U, and 23°Pu (as dilute dispersions in graphite powder). The HTLTR experimental design is based upon a series of XSDRN®>* neutronic calculations in which the central portion of the HTLTR was mocked up in one-dimensional cylindrical geometry. The results of these calculations are given in Table 4.7 in terms of the “expected” reactivity of the corresponding experi- mental measurement. It must be emphasized that these calculated reactivities are approximate and were in- tended only as a guide in choosing experimental design parameters. Numerous known approximations were made. For example, fuel and ThO, particle self- shielding effects were neglected. Though these effects are small, they are not negligible and must be included in the final analysis of the experiments. Special ana- lytical tools are now being developed to treat the double heterogeneity of particles in a fuel lump.. There are two series of experiments listed in Table 4.7. In both series the worth of the 7.5 X 7.5 X 24 in. test cell (see Fig. 4.1a) is to be determined for a variety 3. XSDRN is used both to obtain group cross sections for each region and also to perform multigroup one-dimensional neutronics calculations with the S, transport method. 4. N. M. Greene and C. W. Craven, Jr., XSDRN: A Discrete- Ordinates Spectral Averaging Code, ORNL-TM-2500 (July 1969). of test cell compositions and temperatures. (The test cell layout for item B1 is shown in Fig. 4.1 and will be discussed in more detail later.) In series A, various neutronic properties specifically applicable to the MSBR are to be determined. In item Al, the test cell will contain the same compesition and arrangement as the main lattice fuel, which is designed to approximate the principal properties of an MSBR fuel cell, that is, the same spectrum, k,, C/U ratio, resonance absorp- tion, etc. The overall cell temperature coefficient will be determined from this item. The remaining items of 47 series A will simulate other important features of the MSBR. For example, item A4 will be used to measure the salt density coefficient. Items A5 and A7 will help determine the fuel lumping effect on resonance absorp- tion. Items A8, A9, Al10, and All will be used to determine control-rod worth. Series B consists of material worth measurements for the principal MSBR materials. From both series A and B, we will obtain information to test the basic nuclear data and the reactor-physics computational techniques used for MSR analysis. Table 4.7. Summary of HTLTR reactivity measurements required at each temperature Series A: Worths of blocks as described (7.5 X 7.5 X 24 in. volume) to be measured relative to same volume voided [tem Descriptive L Precalcul?)ted No. block title Block description? worth (¢ at 900°K) 1 Standard Standard fuel in 16 holes 0.786 in. diam, 13.64 vol % 2 Calibrating copper A standard copper foil, lining cavity 3 V replacing Th Vary 1: V,03; replaces ThO, in fuel 4 Dense fuel Vary 1: fuel more densely packed 5 Coarse holes Vary 1: 4 holes, 1.572 in. diam, 13.64 vol % 6 37.1vol % fuel Vary 1: 4 holes, 2.578 in. diam, 37.1 vol % 7 Cruciform Retain 13.64 vol % fuel, geometrically rearranged 8 Solid graphite Contains no fuel +15 9 4-in. control rod Vary 8: one 4-in.-diam fuel rod, 22 vol % +3 10 6-in..control rod Vary 8: one 6-in.-diam fuel rod, 50 vol % -13 11 Shutdown rod Vary 8: insert four Hastel-loy N blades (2.5 X 0.25 X 23 in.) —88 Vary n: means “like item n, except as follows.” bworths based on 0.018¢/g Cu per measurements at room temperature. Series B: Worths of material samples to be measured relative to graphite rod (at center of Fig. 4.1) ltem o Precalculated No Description of sample?® worth? ’ (¢ at 900°K) 1 " Graphite rod, normal density (1.62) 0 2 Graphite powder, density ~1.0 -0.3 3 Void -0.8 4 Copper foil (10-mil, on 1.96-in.-diam graphite rod) -3.6 5 Beryllium rod . +1.7 6 Teflon rod (empirically CF;) at T < 300°C -0.7 7 7LiF (1.375-in.-diam rod; Li is 78 ppm ®Li) at 7 < 627°C -2.0 8 ThO, (250 g) dispersed in item 2 type powder —-4.4 9 233y + Th coated particles (7.5 g 233U) in same +1.9 10 235y coated particles (10 g 235U) in same +1.4 11 Pu coated particles (4.6 g 23?Pu) in same +2.0 12 Hastelloy N foil (11-mil) on item 4 type rod —-1.8 13 Main-lattice fuel mixture -6.1 9All samples 23 in. long and (unless otherwise noted) 2 in. in diameter. bworths based on 0.018¢/g Cu per measurements at room temperature. ORNL-DWG 71-6712 MAIN L+ ATTICE CELL | _——GRAPHITE 4 FUEL TEST SAMPLE ~_E L:’).TS in, ——a—t—— 3 75 in, —= (o} | | % > /////‘, ///’ Z 2, , ‘ 72}% , ' Z ;flf?’; WHITE BOUNDARY 7 g CONDITION ‘.43/!“;..{{, D R % ” TEST SAMPLE B GRAPHITE (&) Fig. 4.1. HTLTR central core geometry and composition. (a) Midplane, (b) as simulated in XSDRN calculations. As an example of the precalculational methods used to generate the approximate values given in Table 4.7, we will describe item B1 in some detail. In Fig. 4.1a is shown one quadrant of the central portion of the HTLTR, including the test cell and one square ring of adjacent main lattice cells. Zone 5 contains the material whose reactivity worth is to be determined. In item B1 the material is graphite. The geometry of Fig. 4.1a is 48 represented in XSDRN as shown in Fig. 4.1b, where zones 1 through 4 are homogenized circular annuli representing the actual geometry. For example, zone 1 represents the main test lattice. The white boundary condition at the outer edge of zone 1 simulates the rest of the HTLTR. This is a good approximation since the fluxes in the central portion of the HTLTR are nearly flat. ‘ The direct result of the XSDRN calculation is the multiplication factor k& for each of the assemblies described by Fig. 4.1b. Since the reactivity of the HTLTR with the various test samples listed in Table 4.7 (series B) will be measured relative to its reactivity with item Bl in the configuration of Fig. 4.1z, we have expressed our calculated multiplication factors as re- activity changes relative to item B1. In item B2 of Table 4.7 we introduced a 5-mil-thick copper foil at the interface between zones 4 and 5 and calculated the difference in multiplication factor between items 1 and 2, as given by XSDRN for the geometry of Fig. 4.1b. Using the experimentally determined worth of copper at the center of the HTLTR (0.018¢/g), we normalized our calculated reactivity worths for the remaining samples, as listed in Table 4.7. Status of the experimental and calculational program at the end of February was as follows. Precalculations, of approximate nature, had been made as a basis for design of the experiment. PNL had prepared all lattice blocks and all materials test samples, except for a few special ones that were supplied by ORNL: "LiF, Teflon (CF,), Hastelloy N foils and plates,-and some special carbon powder for the densely packed fuel. PNL had completed the room-temperature measurements and had begun on those at 300°C. These measurements indicated that good spectral matching was achieved and - that the specified design gave k. and measured re- activity worths in satisfactory agreement with the approximate precalculated values. ORNL was preparing to make more precise calculations on the as-built lattice. Codes were being developed to permit inclusion of grain self-shielding effects, neglected in precalcula- - tions. Cross sections were being processed for the other temperatures needed, the precalculations having been done at 900°K (627°C). 5. Systems and Components Development Dunlap Scott 5.1 GASEOUS FISSION PRODUCT REMOVAL C. H. Gabbard 5.1.1 Gas Separator and Bubble Generator Construction of the water test loop' for testing MSBE-scale gas separators and bubble generators was completed, and the loop was put into operation. Figure 5.1 shows the test loop with the venturi-type bubble generator and the gas separator installed. The loop has been operated primarily on demineralized water at liquid flow rates of 200 to 550 gpm and at gas flow rates of O to 2.2 scfm. It has also been operated with water containing small amounts of n-butyl}alcohol and sodium oleate and with a 41.5% glycerin-water mixture which is hydraulically similar to fuel salt. The performance of the 4-in.-ID gas separator was greatly improved by installation of the annular takeoff port in the recovery vane hub.? Straightening vanes were added within the recovery hub to reduce the pressure drop associated with the flow of gas and liquid through the hub. Figure 5.2 shows the final design operating on demineralized water containing about 0.3% voids at the entrance to the separator. The vortex is firmly attached to the central cone within the Plexiglas recovery hub, and only a trace quantity of very small bubbles escapes the separator at any normal operating condition within the gas flow range of O to 2.2 scfm. The improvement in performance can be seen by comparing Fig. 5.2 with Figs. 5.1 and 5.2 in the previous semiannual report.> The .performance of the venturi-type bubble generator has been satisfactory when operated with the axis vertical. The bubbles were adequately dispersed and were 0.020 in. in diameter or less at flow rates of 400 gpm or greater. There were no strong pulsations in the bubble output, as had been observed in some of the 1. MSR Program Semiannu. Progr. Rep. Aug 31, 1970, ORNL-4622, p. 39. 2. Ibid., p. 38. 3. Ibid., pp. 36 and 37. 49 previous tests of smaller-scale bubble generators, but hydraulic turbulence in the diffuser cone produced small fluctuations in the bubble concentration. A straight cylindrical mixing section about 2.5 throat diameters in length was incorporated in the design to permit the bubbles to be formed and partially dispersed in a high-velocity region prior to reaching the diffuser. However, in practice, the gas traveled the length of the mixing section in relatively straight plumes of large elongated bubbles. The small bubbles were actually formed and dispersed by the turbulence in the diffuser cone. Maintaining a balanced gas feed around the periphery of the bubble generator has not presented a problem with the bubble generator operating in the vertical position and has not required a significant pressure drop across the gas feed holes. The loop has been operated for short periods of time on water containing, separately, n-butyl alcohol, so- dium oleate, and 41.5% glycerin. The n-butyl alcohol was used in concentrations up to 160 ppm to study the effect of a surfactant that would not otherwise alter the properties of the liquid. The sodium oleate was used in concentrations up to 40 ppm with 4 ppm of an antifoaming agent to determine the effect of reducing surface tension (from 72.5 dynes/cm for demineralized water to about 45 dynes/cm). The 41.5% glycerin-water mixture is hydraulically similar to the fuel salt. The addition of n-butyl alcohol and sodium oleate may have caused a slight decrease in the size of bubbles produced by the bubble generator, but the most obvious effect was to stabilize small bubbles, inhibit their coalescence, and promote clouding of the loop fluid by bubbles 0.001 to 0.002 in. in diameter which were formed by the pump. These small bubbles were not effectively removed by the separator, and several minutes of loop operation without gas input were required to clear the fluid. Operation with the 41.5% glycerin-water mixture gave similar results except that the small bubbles were removed much more slowly. A small amount of foaming occurred at the free liquid surfaces with water containing the n-butyl alcohol; the sodium oleate produced extensive foaming that was greatly reduced but was not eliminated by addition of the antifoaming 50 PHOTO 78448 Fig. 5.1. Gas separator and bubble generator water test loop. PHOTO 78450 GAS TAKEOFF LINE RECOVERY VANES FLOW i Fig. 5.2. Four-inch-ID pipeline gas separator operating on demineralized water. ORNL-—-DWG Ti— 6713 4 \ -3 \ £ \ \ 1.5% GLYCERIN -WATER I : N\ 4 » ./ MSBR FUEL SALT g WA - = : 2 \ X \ Ty o 0 0.004 0.008 0.012 0.016 0.020 RUBBLE DIAMETER (in.} Fig. 5.3, Calculated separation length vs bubble diameter for 4-in.-ID gas separator and liquid flow of 400 gpm. agent. No foaming occurred with the glycerin-water mixture. Loop and pump modifications are in progress to determine whether fewer small bubbles are produced by operating the pump near its best efficiency point and at a lower speed. We infer that the circulating bubbles in the MSRE and in the MSRE prototype pump test loop were not so small because they separated out relatively quickly under low or zero flow conditions. Attempts to produce foams in molten salt* by bubbling gas through a coarse glass frit produced bubbles 2 to 3 mm or larger in diameter under normal salt conditions, and a foam was generated only when moisture or hydrates were added to the salt. Therefore we do not expect to encounter significant quantities of the very small bubbles in a salt system. A computer program was written to calculate the required length for separating bubbles in a pipeline separator. The mixture of bubbles and liquid enters the separator with specified axial and rotational velocities, and the required length is the axial distance that a bubble of a specified diameter will travel as it is caused to migrate from the wall of the pipe to the central vortex by the centrifugal forces. Calculations were completed for a 4-in.-ID separator operating on water, -salt, and the 41.5% glycerin-water mixture. The results of these calculations are shown in Figs. 5.3 and 5.4. Figure 5.3 shows the rapid increase in the required separation length with decreasing bubble size and shows 4. MSR Program Semiannu. Progr. Rep. Feb. 28, 1969, ORNL-4396, p. 137. 51 ORNL—DWG 71— 6714 6 z° N £ a \\\BUBBLE DIAM =0.005 in. pd oE ~ N — e ] '."‘--..____ [ —— g ~— T————_0.007 r 2 | B =— : ———t— oo — [ ——— P =] 0.015 0.020 o} | 200 300 400 500 600 LIQUID FLOW RATE (gpm) Fig. 5.4. Calculated separation length vs fuel salt flow rate for 4-in.-1D gas separator. that the separation length is greater with salt than with _ water. Figure 5.4 shows the effect of flow rate on the separation length for several bubble sizes. 5.2 GAS SYSTEM TEST FACILITY W. K. Furlong The conceptual design of a molten-salt loop for testing gas systems was completed, the conceptual system design description was written, and work is now beginning on the preliminary design. The facility will be used for developing the technology of the fuel salt for MSRs and in particular for tests such as the perfor- mance of the bubble injection and separation compo- nents and the off-gas handling system. In addition, several other important tests will be run over the life of the loop. These include measurement of surface tension as a function of UF,/UF; ratio, the study of noble gas and possibly tritium distribution, the study of bubble dynamics, the measurement of heat transfer coeffi- cients, and the measurement of the product of mass transfer coefficient and bubble surface area. The facility is scheduled for initial operation in the first quarter of FY 1973. 5.3 MOLTEN-SALT STEAM GENERATOR J. L. Crowley R. E. Helms Our program plan for obtaining reliable steam genera- tors for the MSRP consists of the following three phases. Phase 1 — preliminary development of molten-salt steam generator technology. Some conceptual designs and proposals for development requiréments will be prepared by industrial manufacturers of steam genera- tors.> We will independently prepare a molten-salt steam generator development basis report which will also show which elements of the LMFBR program will be of use. In addition, some preliminary information will be obtained from some small-scale tests. Phase I is expected to allow us to proceed with the engineering development, procurement, and testing program for the MSBE steam generator. Phase II — completion by the industrial mamifac- turers of a preliminary design of a steam generator based on the conceptual design and completion of the engineering tests defined in phase I These tests will include the operation of some full-size steam tubes to be supplied by the designer in a molten-salt steam generator tube test stand (STTS). The completion of phase II should place the manufacturers in position to proceed with the detailed design of our MSBE steam generator. Phase [l — detailed design, fabrication, and operation of a prototype steam generator in the MSBE or in a test facility with a very large heat source. With the experience gained from this phase, the manufacturer should be capable of supplying reliable steam generators for future needs of the MSBR program. A discussion of the present activities of phase I follows. 5.3.1 Steam Generator Industrial Program Efforts continued toward obtaining conceptual de- signs and development program proposals from a qualified industrial manufacturer of steam generators. A package consisting of a summary of the 1000-MW(e) MSBR reference design, a proposed scope of work for the industrial program, and a cover letter requesting an expression of interest in receiving a request for proposal was sent to nine industrial firms. Eight firms responded affirmatively, and the request for proposal package was sent to them. We hope to enter into a CPFF contract with one firm during this fiscal year to begin work on the four tasks of the industrial study.’ 5.3.2 Steam Generator Tube Test Stand (STTS) The conceptual system design description (CSDD) was completed and issued as an internal ORNL memo. This CSDD describes the 3-MW test facility to be built as part of phase II of the molten-salt steam generator 5. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, pp- 39-40. 52 program plan previously discussed. Further work in this area has been suspended until late in FY 1972, 5.3.3 Molten-Salt Steam Generator Technology Facility (SGTF) The steam generator technology facility is being considered for use in testing sections of steam generator concepts or portions of single-tube heat exchangers while generating steam at heat rates of 150 kW or less. As an aid in preparing the requirements for such tests we are examining several combinations of tube-in-tube test sections to determine what would be most useful. Calculations indicate that these small tests can be useful for tubes up to % in. in diameter and 33 ft long producing steam at 1000°F and up to 3600 psi pressure. A conceptual design of the equipment to do these tests is being prepared. 5.3.4 Development Bases for Steam Generators Using Molten Salt as the Heat Source Work has begun on preparation of a development basis report for molten-salt steam generators. This report will review the present technology of similar systems using sodium as the heat source and will point out elements of the LMFBR program which will be of use in the molten-salt technology program. This report will identify molten-salt steam generator problem areas needing further fundamental study and development and will outline a program for the development efforts needed for designing and fabricating reliable molten-salt steam generators. 5.4 SODIUM FLUQROBORATE TEST LOOP | A. N. Smith The test program was completed during the previous report period.® Following the final shutdown of the loop, the pump rotary element, the bubbler tube which served as the BF; feed line and as an indication of the salt level in the pump bowl, the spark plug salt level probes, and the pump bowl pressure control valve were removed from the system for inspection. The general appearance of the pump impeller and adjacent surfaces indicated that the 11,567 hr of fluoroborate salt circulation did not cause excessive corrosion of the Inconel system. Deposits of salt were found in the upper part of the rotary element near the thermal barrier, about 4 in. above the normal level of 6. Ibid., p. 41. salt in the pump bowl. We concluded that the nature and cause of these deposits were the same as for similar deposits observed in May 1968.7 The BF; feed and salt level bubbler tube (Inconel) was found to contain deposits of several different corrosion products, includ- ing one at the mouth of the tube which had severely restricted gas flow. Metallographic and chemical analy- ses were made to determine the nature of the deposits and the extent of damage due to corrosion. Metallurgi- cal analysis of the Hastelloy N spark plug salt level probes is incomplete. Visual examination of the stem and seat of the off-gas pressure control valve revealed no evidence of corrosive attack. The stainless steel valve was in service throughout the entire test program, operating at room temperature with a C, 0of 0.01,a AP of 24 psi, and a gas flow of 1.5 liters/min consisting of helium with 3.5% BF;. The material below presents a more detailed discussion of the observations and con- clusions resulting from the examination of the pump and BF; feed and salt level bubbler tube. 5.4.1 Pump Bowl and Rotary Element® The pump rotary element was removed from the pump bowl in September 1970. The appearance of the PHOTO 75584 AFTER 187 hr SALT CIRCULATION (APRIL 1968 ) 53 pump bowl as viewed from the north side is compared with the same view taken in April 1968 after only 187 hr of salt circulation (see Fig. 5.5). This time consider- ably more debris, presumably corrosion products, was present on surfaces visible through the fountain flow windows. The material probably fell from the thermal barrier region when the rotary element was removed. The sharp edges on the metal surfaces at the entrance to the volute, which is adjacent to the impeller tip (tip speed of the 9-in. impeller was about 70 fps), is interpreted as an indication that the combined corro- sive-erosive attack was slight. The as-removed appearance of the pump rotary element is shown in Fig. 5.6. Here again, the sharp edges at the impeller discharge openings are an indica- tion of insignificant attack. The debris on the horizon- tal surface just above the impeller probably fell to that point from the vicinity of the thermal barrier. The presence of the white cake of salt on the thermal barrier surface (an annulus about Y, ¢ in. wide is formed by this 7. MSR Program Semiannu. Progr. Rep. Aug. 31, 1968, ORNL-4344, p. 78. 8. Reactor Handbook, vol. 1V, Engineering, 2d ed., p. 824, Wiley, 1964. AFTER 11,567 hr SALT CIRCULATION (SEPTEMBER 1970) 54 PHOTO 78058 THERMAL BARRIER GREEN SALT DEPOSITS THERMAL RADIATION BAFFLES GREY-BLACK MATERIAL HAVING HIGH NICKEL CONTENT Fig. 5.6. Post-test appearance of salt pump rotary element, NaBF circulation test, PKP loop, 9201-3. surface and the pump bowl neck) is an indication that salt was forced into this region, probably during the ingassing transients.” The dark material in the openings immediately above the heat baffles is primarily green salt similar to that observed in May 1968.7 The chemical analyses of the two green salt deposits (Table 5.1) show the later deposit to have a higher iron and chromium content, a lower uranium and thorium content, and a significantly different U/Th ratio. The region of the green salt deposits is about 4 in. above the normal level of salt in the pump bowl. The appearance and composition of the deposits seem to argue against formation from vapor or mist. The inference is that the green salt is a relatively insoluble phase formed by combination of the flucroborate salt with residue of the fuel (BULT-4) salt formerly used in the loop and that the deposits were formed during periods of abnormal salt level when the bulk salt came into contact with the cool surfaces in the thermal barrier region. The only known periods of abnormally high salt level were during the several brief ingassing transients which occurred during the operation with flush salt. We conclude that the deposits found in September 1970 must have been formed sometime between May 16, 1968, when opera- tions were resumed after inspection of the rotary element, and June 24, 1968, the date of the last ingassing transient. A gray-black piece of material found outside the fountain flow window (Fig. 5.6) was magnetic and had the following chemical analysis: Element Weight % Na 5.0 B 3.0 Fe 2.5 Cr 0.2 Ni 66.3 F Assumed remainder We think that this deposit -precipitated from the fountain flow (upper labyrinth seal leakage) stream at a point near the impeller hub where the temperature was below the bulk salt temperature. We can speculate that the circulating salt stream was saturated with nickel at normal circulation temperatures, since it has been reported by others'® that the nickel ion has a low solubility in the fluoroborate salt and tends to deposit in low-temperature regions. The nickel-bearing corro- sion products would thus tend to separate out at the 9. MSR Program Semiannu. Progr. Rep. Aug 31, 1968, ORNL-4344, p. 76. ' 10. MSR Program Semiannu. Progr. Rep. Féb. 28 1970, ORNL-4548, p. 242, ' 55 Table 5.1. Chemical analyses of green salt deposits Analysis (wt %) Element September 1970 May 1968 Na 17.8 11.0 B 2.6 3.9 Li 0.17 0.21 Be 0.10 0.04 U 0.35 12.2 Th 13.9 26.5 Fe 9.45 1.5 Cr 8.9 0.27 Ni 0.08 0.03 F 45.3 43.5 first convenient cool surface external to the isothermal circulating stream. None of the green salt was found in this area, which implies that the bulk salt at 1025°F was not saturated with respect to the green salt. 5.4.2 BF; Feed and Salt Level Bubbler Tube The gas bubbler tube which was used for BF; addition and salt level indication was removed from the pump bowl in October 1970. The %-in.-OD by % -in.- ID Inconel tube had been in service for the entire test program from March 1968 through April 1970, during which time salt circulated 11,567 hr. The tube was 13 in. long, and the bottom end contained one triangular notch Y% in. wide by % in. deep. Although operating conditions were varied briefly for special tests, during most of the test program the total gas flow was 370 cm® /min, and the gas was helium containing 13.5% BF; by voklu,me. During periods when the loop was shut down ‘and drained, the BF; flow was stopped, argon was used in place of the helium, and the flow rate was somewhat reduced. The total volume of gas which passed through the tube was about 0.26 X 10° liters during salt circulation and about 0.4 X 10° liters during the entire test program. Note that inert gas was routinely added to the pump bowl through other lines, so that the total pump bowl gas flow was normally 1500 ¢cm® /min. The loop was filled and drained a total of 16 times during the test program, and the pump bowl salt level varied somewhat from one operating period to the next. Also, the level varied whenever the salt temperature was changed. The average salt level was 2Y, in. above the bottom of the tube during the flush salt operating period and 4% in. above the bottom during the clean salt operation. In each case there were variations of not more than *1 in. . ' During the final 20 days of loop operation, the salt level instrument showed a steady increase of about 0.06 PHOTO 78258 INCHES Fig. 5.7. Interior view of BF 3 feed and salt level bubbler tube, NaBF; circulation test, PKP loop, 9201-3. in./day. This was interpreted as an apparent increase, probably caused by an increase in flow resistance (plug) in the bubbler line. During the final salt drain, when the salt level was decreasing, the indicated salt level rose and went off scale (100% = 20 in. H,0), an indication of a plugged line. The appearance of the exterior of the tube was as follows: the opening at the bottom end was completely filled with a silver-gray metallic-appearing magnetic deposit; the lower 4 in. of the exterior surface had a bright silvery-grainy appearance and was very rough to the touch; the next 3 in. was dark, not as rough as the lower region, somewhat grainy in the lower half, and had a thin coating of bright green solid on one side; the upper 4 in. was gray-black to black with a greenish cast. The tube was cut into four pieces by cuts at 1, 4%, and 7% in. from the lower end. The three lower pieces were then slit axially. The inside of the tube appeared as follows (Fig. 5.7). The plug at the bottom end was %, to % in. thick; for about % in. above the plug, the wall was covered with a thin black film; for the next % in. there was a thin deposit which was mostly light gray with a sprinkling of light green particles. From the 1%-in. level to the 4%e-in. level, bright green material was deposited and from 4% to 7% in. there appeared to be the same green material covered with a loose black magnetic powder. The deposit was %, to % ¢ in. thick for the most part, except that it was somewhat thicker at about the 2-in. level and filled about 90% of the tube cross section at the 5%-in. level. In cross section, the plug at the bottom end of the tube had a white, mottled appearance as though the metallic deposit was interspersed with lumps of salt. The green deposit had a yellowish cast. Metallographic examination of the tube, under the direction of Koger,'' produced the following conclu- sions. About 5 mils of material was removed from the Inconel tube. As expected, attack by salt and by vapor was evident, but no void formation occurred; corrosion product was left on the inside of the tube from vapor attack and was deposited on the lower opening by mass transfer from the bulk salt. Identification of the various materials found in the tube was made under the direction of Cantor.'? Analyses by x-ray diffraction and electron microprobe indicated that the green salt was Na; CrFg, the yellow salt was predominantly NaNiF; , and the black, powdery magnetic material was a mixture of nickel and iron in a ratio stoichiometrically equivalent to the alloy Ni; Fe. The plug at the bottom of the tube was a mixture of metallic nickel and NaBF, in approximately equal parts. Cantor surmised that the NazCrFg and NaNiF; were produced by reaction of the salt with moisture intro- duced by the various gas streams. The reaction products 11. J. W. Koger, part 4, this report. 12. S. Cantor, sect. 9.3, this report. probably accumulated as a scum on the surface of the salt and were subsequently transferred to the inner walls of the tube by some mechanism, most likely in liquid droplets formed by the continuous breaking of gas bubbles. Routine surveillance and control were not maintained over the water content of the incoming helium and BF;, so we do not have an accurate measure of the quantity of water which was added to the system in the gas feed. However, based on limited observa- tion,'® the water content of the BF; is thought to have been less than 50 ppm by volume. The helium was supplied from a source which is known to be of high purity (less than 1 ppm H,O). Brief tests with a moisture monitor indicated that the helium stream at the loop contained about 20 ppm, so we believe that it certainly contained less than 50 ppm H,O. The calculated rate of addition of water to the system would be about 0.8 g of H, O per 10° liters of gas per ppm H,0 by volume. Therefore, if we assume a maximum concentration of H,O in the feed gas of 50 ppm by volume, the maximum amount of water added during salt circulation is estimated to be 40 g total injected into the pump, with 10 g entering through the bubbler tube and the remainder through the shaft purge. These numbers represent water input rates of about 1 and 3.6 mg/hr, respectively, when averaged over the 11,567 hr of operation but do not include the 10 g of water injected instantaneously in a.special test.'* We do not know the weight of material which was deposited in the tube, nor what fraction of the total corrosion product this represented. However, by making rough assumptions regarding volume and den- sity, we estimate that the total amount of material in the tube was probably less than 10 g, and on this basis it appears possible that water in the gas feed could have been an important contributing factor in the plugging of the tube. 5.5 COOLANT SALT TECHNOLOGY FACILITY A. 1. Krakoviak A conceptual system 'design description (CSDD) for the coolant salt technology facility (CSTF) has been prepared (ORNL-CF-70-12-18), and preliminary design of the facility and components has begun. 13. R. W. Apple to A. N. Smith, personal communication, Mar. 5, 1971. 14. MSR Program Semiannu. Progr. Rep. Aug 31, 1970, ORNL-4622, p. 41. 57 We plan to use the piping (5 in. sched 40), the pump, and other components from the MSRE coolant system to upgrade the existing sodium fluoroborate test loop (constructed of Inconel) to a facility constructed of Hastelloy N. Where applicable, the existing drawings of the PKP-1 pump test stand, service piping, containment enclosure, cover-gas supply, and off-gas disposal facili- ties are being reused. : o The proposed salt loop will have a pump, a’salt-to-air heat exchanger (to dissipate pump power), a drain tank, and the necessary connecting pipe. An orifice will limit the loop flow to ~600 gpm at a pump discharge pressure of ~100 psig and a suction pressure of ~25 psig. The loop will be fabricated with matching sets of nozzles on the discharge and suction lines to permit later attachment of parallel salt circuits. These side loops will be used to provide the additional information needed to assure that the sodium fluoroborate—sodium fluoride eutectic does indeed meet the requirements of a secondary fluid for molten-salt reactors. The new facility, with its side loop, will supply additional information on . 1. boron trifluoride cover-gas addition to and removal from the pump bowl, chemical removal of water from the fluoroborate salt at rates typical of small steam inleakage, 3. detection and removal of corrosion products that would result from small steam leaks into the salt, . control of the salt chemistry so that the coolant may also serve as a sink for the tritium inherently produced by lithium-based reactor fuels, 5. operation of a steam generator test heated by molten salt, . heat and mass transfer near the cold wall of a steam generator, 7. operation of valves in molten salt. In some cases, further fundamental chemical and/or metallurgical studies are needed before an engineering- scale test can be made. Although the main facility is designed primarily for testing with sodium fluoroborate salt, it would also be suitable for testing other fluoride salts. At the end of this report period, the coolant system piping had been removed from the MSRE and the mechanical design of the loop was ~10% completed. The main loop is expected to be complete at the end of December 1971. . 5.6 MSBR PUMPS L. V. Wilson A. G. Grindell H. C. Savage H. C. Young 5.6.1 MSRE Mark 2 Fuel Pump As previously reported,' during the shutdown of the MSRE Mark 2 pump, we found that the system salt could not be drained into the storage tank because the connecting line was plugged. Various attempts to promote draining by increasing the temperature of the salt piping system including the freeze valve and by increasing the differential pressure between the pump and storage tank were unsuccessful. We then replaced the drain line by use of a procedure in which all the salt was frozen except that in the pump tank and in three pressure transmitters. After the new installation was inspected, the salt was melted carefully by heating the pump suction and then the pump discharge piping progressively outward from the hot salt in the pump tank. The salt drained satisfactorily into the storage tank, where it was frozen, and the facility was secured. The entire length of the plugged drain line was x rayed, and suspect sections were removed, separated into two symmetrical longitudinal pieces (shown in Fig. 5.8), and examined by the Reactor Chemistry Division. The plug was located approximately 32 in. downsfream of the junction of the drain line with the main loop piping and 44 in. upstream of the freeze valve. A considerable amount of relatively large (0.5 mm long) zirconium oxide crystals,'® which may be seen in 15. MSR Program Semiannu. Progr. Rep. Aug 31, 1970, ORNL-4622, p. 6. 16. R. E. Thoma, Examination of Specimens from the MSRE Fuel Salt Pump Test Stand, MSR-7061 (Nov. 17, 1960) (internal correspondence). SECTION I-J Fig. 5.8, was found in section I-J of the drain line. The zirconium oxide crystals, section I-J, are compared with normal salt, section K-L, in the figure. We hypothesize that the contaminants needed to produce the plugging material were introduced into the salt system by the argon purge gas and during the approximately 60 times the system was opened to perform various maintenance activities and that they precipitated in a cool section of the drain line. Additional details of the draining operation are presented in Incident Report ORNL- 70-36-Final. 5.6.2 MSRE Salt Pump Inspection A visual inspection was made of the rotary element and the conveniently accessible interior surfaces of the pump tank in the MSRE coolant salt pump as the rotary element was being removed from the coolant salt system. The observations are described below. The salt had drained well from all salt-exposed surfaces. The salt flow passages in the impeller, volute, and suction line were clean, and no evidence of wear of any kind was noticeable. A black carbon-like residue was seen on the exterior surfaces of the shield plug and its extension, both of which are parts of the rotary element, down to the salt-gas interface. There was a similar residue on the pump tank surface directly opposite the shield plug and on the cylinder which supports the pump volute down to the salt-gas interface. Apparently oil leakage from the lower shaft seal had drained from the catch basin, run down the outside of the shield plug, and decom- posed at the 1025°F temperature in the pump tank. These observations confirm the suspected leakage of a mechanical seal in the oil drainage system. This seal was replaced with a seal weld in the spare rotary elements for the fuel and coolant salt pumps. The rotating PHOTO 78332 SECTION K-L Fig. 5.8. Plugged drain line from Mark 2 pump test stand. Section I-J shows presence of zirconium oxide crystals in the salt; section K-L shows frozen normal salt. components, including the impeller, shaft, drive motor rotor, and all bearings and shaft seals, turned easily and smoothly to the hand, lending support to our belief that the pump would have operated satisfactorily at high temperature for many additional hours. Additional inspection of the pump will be performed while it is being prepared for use in one of the technology loops. 5.6.3 ALPHA Pump Water testing of the ALPHA pump,'” a centrifugal pump designed for capacities to 30 gpm, 250 ft head, and temperatures to 1400°F, was completed. Data indicate that the design hydraulic performance condi- tions can be met at a shaft speed of approximately 6750 rpm. The pump is to be used initially in molten-salt forced-convection loop’® MSR-FCL-2. The pump satis- fies the MSR-FCL-2 hydraulic requirements of 4 gpm at approximately 110 ft head at a shaft speed of approxi- mately 3900 rpm. Based on the water test results, the final design of the pump tank and internal parts was completed, and the parts were fabricated of Hastelloy N for use in the pump unit for MSR-FCL-2. Further water testing of the ALPHA pump is not planned at this time because the only available rotary element will soon be installed in MSR-FCL-2. However, additional testing may be desirable in the future to measure the fountain flow rate and to define in more detail the shaft deflection characteristics at higher pump capacities (speeds above 6000 rpm). 5.7 REMOTE WELDING P. P. Holz W. A. Bird C. M. Smith, Jr. W, R. Miller Most of the work previously supported under the MSRP remote welding program has been transferred into an automated welding program sponsored by the LMFBR program and will be reported under that program in the future. The new LMFBR program is directed toward developing suitable equipment, welding procedures, and programmed controls to produce butt welds with consistently high quality in pipe to meet LMFBR needs for reactor construction and loop main- tenance. It is planned to extend automated welding technology to pipe sizes as large as 30 in. diameter and to field test the equipment at construction sites. The 17. MSR Program Semiannu. Progr. Rep. Aug 31, 1970, ORNL-4622, p. 45. 18. Ibid., pp. 176—78. 59 LMFBR program makes use of the prototype machine- ry and controls developed in the MSR program.19-21 We are continuing 'to sponsor the work necessary to complete ‘and test the weld torch positioning mecha- nisms and "control circuitry; to define and document remote maintenance inspection, viewing, and alignment criteria; and to investigate pipe cleanliness requirements for maintenance welding in salt systems. 5.7.1 Automatic Controls We are striving to develop a welding system that can, without human intervention, handle all the perturba- tions, misalignments, etc., that are normally encoun- tered in welding. This effort should not be confused with programming the controls for automated welding. Programming simply provides preset values for specified variables at a prescribed time. A self-adaptive system must have the ability to correct for unexpected conditions at any time during the execution of the weld. Thus, the system mechanizes operator capabilities and reduces the level of skill required of welding personnel. Such a system is necessary for completely remote welding in radioactive areas. An automatic horizontal torch control and an auto- matic torch oscillator control are two new devices that are being added to our present equipment, which already includes an operational automatic vertical torch control. The automatic vertical torch control employs a servo-drive motor mechanism to control the torch-to- work spacing. The system steps the torch vertically within a narrow range to keep the arc voltage nearly constant and thus permits the wire feed rate control to operate as a fine control vernier. This combination of controls produces uniform weld metal deposits while maintaining proper arc voltage for uniform weld pene- tration. The automatic horizontal control will work in conjunction with automatic vertical control to position the torch at the optimum location between the beveled edges of the joint prior to making the weld pass. A second motor has to be added to drive the torch assembly horizontally across the weld groove area. The present equipment utilizes a dc motor with an adjust- able eccentric mechanism to oscillate the torch in a fixed pattern. Variations in the side wall of the groove 19. MSR Program Semiannu. Progr. Rep. Aug 31, 1969, ORNL-4449, pp. 79-82. 20. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp. 74-178. A 21. MSR Program Semiannu. Progr. Rep. Aug 31, 1970, ORNL-4622, pp. 45-50. sometimes produce insufficient tie-in or, at the opposite extreme, contact between the electrode and the side wall The improved system will continually sense the arc voltage and control the oscillator drive motor to keep the electrode within the proper distance from the wall of the groove. | 5.7.2 Pipe Cleaning Tests A -series of experiments to determine cleanliness 60 requirements for pipe joints to be welded and to check into possible contamination effects from salt plugs in the pipe near the weld zone produced promising results. - We were able to make butt welds manually to mieet nuclear code x-ray standards in ‘- and % -in. standard-- weight Inconel pipe that were plugged solid with LiF-BeF,-ZrF,;-ThF,-UF,; (68.4-24.6-5.0-1.1-0.9 mole %) by merely clearing a distance of 1 in. on both sides of the weld joint. Special provisions were made for purging the pipe interior. Pipe cleaning consisted in drilling out the plug and then wiping interior and exterior pipe surfaces with emery cloth to remove salt films and scale and to produce a reasonable surface finish. We were careful not to polish the surfaces because this would be difficult to do under working conditions in a reactor. Weld preparation and welding were done in a glove box.?? 22, P. P. Holz, Butr Welding Experiments with Salt Plugged Pipes, ORNL-TM report (to be issued). 6. MSBR Instrumentation and Controls 6.1 DEVELOPMENT OF A HYBRID COMPUTER SIMULATION MODEL OF THE MSBR SYSTEM O. W. Burke 6.1.1 Introduction As an extension of the work done by.Sides,' a hybrid computer simulation model of the MSBR has been under development for the past three months. The new model of the reactor, primary heat exchanger, etc., up to the steam generator, is essentially the same as that used by Sides. The model of the steam generator is similar to that reported in ref. 2. The bulk of the discussion here will pertain to the _steam generator model. 6.1.2 Steam Generator Model The steam generator is'a countercurrent, single-pass U-tube exchanger approximately 77 ft long and 18 in. in diameter. At the steady-state design point condition, water enters one end of the exchanger at a temperature of 700°F and at a pressure of 3752 psia. Salt enters the opposite end at a temperature of 1150°F. The salt flows through the shell side of the exchanger ‘and the water through the tube side. The mathematical model of the system consists in the conservation of mass, conservation of momentum, and conservation of energy equations of the salt and water. These equations are written in one space dimension, x (the direction of water flow), and time, ¢. In this initial model the variations in the density and velocity of the salt are neglected, and hence only the conservation of energy is considered for this part of the system. If it appears that a more detailed model of the salt side of the exchanger is warranted, it will be added later. The following equations were used: 1. W. H. Sides, I1., Control Studies of a 1000-Mw(e} MSBR, ORNL-TM-2927 (May 18, 1970). 2. C. K. Sanathanan and A. A. Sandberg, University of [llinois, Chicago, and F. H. Clark, O. W. Burke, and R. S. Stone, ORNL, “Dynamic Modeling of a Large Once-Through Steam Generator,” to be published in Nuclear Engineering and Design. Conservation of mass-(water), ap , 0 A 3 e PV =03 | o | (1) conservation of momentum (water), a(pv) or ox ax kop ' )= ey ) conservation of energ'y (wafer); ' ) ;) _ ' | B;(ph) +a(phv)—k1H(9 - T); : (3) conservation of energy (sait), (T—-06); 4) the equations of state for water, T="T(p, h); | ) p=pp, h). | (6) "The definitions of the variables used in the above equations are as follows: T = water temperature, p = water density, | v = water velocity, P = water pressure, ¢ = coefficient of friction, ~ k-= constant used to make units consistent, h = specific enthalpy of water, H = heat transfer coefficient, salt to water, k, =ratio of the surface area of a tube to the water volume in the tube, . : k,=r1atio of the surface area of a tube to the salt volume adjacent to the tube, p, = salt density (assumed constant), p = gpecific heat of salt at constant pressure, 8 = salt temperature, ' w = salt velocity. Boundary conditions on p and A are applied at the water entrance to the exchanger, on # at the salt entrance, and on v at the throttle. The critical flow at the throttle is expressed by the following nonlinear relationship among the system variables at a point just before the throttle: AT p 2 =M , o (Am) <1 + bT> where A, is the instantaneous value of the throttle opening, Ar o the initial steady state value, M the critical flow constant, and 5 an empirical constant. Ag o is taken as 1.0 and Ay is varied as a function of time during transients. It was determined in previous work? that a con- tinuous-space, discrete-time model is most satisfactory for this steam generator simulation. By simplification of Egs. (1), (2), (3), and (4), and using the backwards differencing scheme for the time derivative, the fol- lowing ordinary differential equations are generated: (7) 2 V—vV dp__pvdv_ev POV (M) dx k dx k k At dh _ 1 h—hy S ylkHE - D] -, (2M) dv__vdp PPk (3M) dx pdx pAt Hk,(T-6) 6-16 d@z_ 2( )+ k (4M) dx PsCp Vs v At where v, is salt velocity. In the above equations, the nonsubscripted variables are the ones being iterated for the values at the end of the (k + 1) time increment, while the variables with the k subscripts represent their values at the end of the kth time increment. The time increment is represented by At. By judicious choice of the direction of integration in space, of the various dependent variables, an initial value problem can be formed. Since the water enthalpy h and the water pressure p are known at the water entrance end of the exchanger (left end), these variables 62 will be integrated from left to right. For the same reason the water velocity (it can be calculated at the throttle) and the salt temperature will be integrated from right to left. . In the hybrid program the integrations are performed by the analog computer. The digital computer calcu- lates the terms of the derivatives of the differential equations, provides control for the calculation, and provides storage. The thermodynamic properties of water are stored in the digital computer as two- dimensional tables. An interpolation routine is used to get values from these tables. In the present model the derivative terms are updated at 1-ft intervals in the x dimension. The calculational procedure for a Ar step is as follows: With left boundary values of water enthalpy and water pressure as initial conditions, a left to right integration of these two variables is started. As the integration proceeds in x, the derivative terms of the differential equations are updated at 1-ft intervals. Values of p and & are stored at these 1-ft locations in x. The integrations are stopped when the x location corresponding to the water exit end of the steam generator is reached. In a procedure identical to that above, and with the current values of p and A, the salt temperature and the water velocity are integrated from right to left. The initial condition of the salt tempera- ture is that at the point where it enters the exchanger. The initial condition of the water velocity is the calculated velocity at the throttle. When the right to left integrations have proceeded to the left boundary, they are halted. The left to right integration of p and h is repeated, using the current values of p, 4, v, and 6. The right to left integration is repeated, etc., until the convergence is satisfactory. It appears that the A¢ can be of the order of 1 sec. The model of the reactor; primary heat exchanger, piping, etc., is a continuous-time model similar to those traditionally used on analog computers. It was felt that the detail required in these pélrts of the system was not as great as that required in the steam generator. The steam generator model will be coupled with the analog model of the remaining parts of the system in the following manner. The analog model of the system, exclusive of the steam generator, will be slowed down in time by a factor of 10 or 100, as required. At intervals of approximately 1 sec the water pressure and temperature at the left end of the steam generator and the salt temperature at the right end of the steam generator are read and stored in the digital computer. These values are used for a At step calculation on the hybrid computer as described above. At the end of the ~ At calculation, output values of p, A, 8, T, and v are stored and fed back to the analog simulation. This procedure is repeated at time intervals of Ar. If the analog simulation is slowed down by a factor of 100, for instance, the 1-sec At of the hybrid calculation will look like 1/100 sec to the analog model. By using a short time constant first order lag on the discrete outputs from the hybrid model, it is hoped that the outputs from the hybrid model of the steam generator to the analog model of the rest of the system will be fairly smooth in time. { 63 The analog model of the system, exclusive of the steam generator, has been developed and is ready to be patched on the analog patch boards. The patching and debugging of this model will require a few days. The hybrid program for the steam generator has been written, and practically all of the debugging has been accomplished. It is expected to be running within a very short time. The integration of the two models will require some time. The model should be complete before the target date of June 30, 1971. 7. Heat and Mass Transfer and Thermophysical Properties H. W. Hoffman 7.1 HEAT TRANSFER J. W.Cooke Heat transfer experiments employing a proposed MSBR fuel salt (LiF-BeF,-ThF,-UF,, 67.5-20-12-0.5 mole %) flowing in a horizontal tube have shown that the local heat transfer coefficient varies along the entire length of the tube in the Reynolds modulus range 2000 < Ng, < 4000 for heat fluxes from 0.7 X 10° to 3.0 X 10° Btu hr™' ft™2. It is hypothesized that a delay in transition to turbulent flow could result in such a variation in the transitional flow range.'™ To investigate the effect of flow development on heat transfer, a new test section, shown in Fig. 7.1, was installed in-the inert-gas-pressurized molten salt heat transfer system. This test section consists of a 48-in. length of 0.25-in.-OD by 0.035-in.-wall Hastelloy N tubing, with three electrodes welded to the test section so that the left half, the right half, or the entire length can be resistance heated. At each end are “disk- donut”-type mixing chambers. Since the flow alternates in direction, six different modes of operation of the heat transfer system are possible. For the initial studies, the left half was heated to provide a heat flux range from 0.5 X 10° to 4 X 10° Btu hr™' ft™, and the right half was heated just enough to make up for the radial heat loss to the surroundings (~0.05 X 10° Btu hr™! ft™?). Thus with flow to the left there is a 24-in. adiabatic-hydrodynamic entrance length, and with flow to the right there is a round-edge nozzle-type entrance. Over 200 calibration and data runs have been con- ducted with this new test section at two temperature levels (~1070 and ~1250°F) and three Reynolds 1. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 53-57. 2. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp. 87-88. 3. MSR Program Semiannu. Progr. Rep. Aug. 31, 1969, ORNL-4449, pp. 85-89. 64 J. J. Keyes, Jr. Table 7.1. Thermophysical prdperty data for molten salt mixture LiF-BeF,-ThF4-UF4 (67.5-20-12-0.5 mole %) used in the present calculation Uncertainty (%) u (b £t ™ hr1) = 0.187 exp {8000/T(°R)]? £25 k [Btuhr! ft7! CF)71] =0.69° 12 p (Ib/ft3) = 230.89 — 22.54 x 1073+ CF)? +3 Cp [Btulb™! CF)71) =0.324% +4 43. Cantor (ed.), Physical Properties of Molten-Salt Reactor Fuel, Coolants, and Flush Salts, ORNL-TM-2316 (August 1968). bMSR Program Semiannu. Progr. Rep. Aug. 31, 1969, ORNL-4449, p. 92. moduli (~3500, ~7000, and ~12,000) covering a heat flux range from 0.2 X 10° to 3.4 X 10° Btu hr ™! ft72. Data from 27 of these runs at the lower temperature level have been thoroughly examined thus far. The data have been analyzed as previously described® with the exception of two fundamental differences. First, the electrical heat generated in the tube wall (corrected for heat losses), rather than the enthalpy rise of the salt, was used to obtain the heat input to the salt. Second, the local bulk salt temperatures were calculated using the measured wall temperatures along the adia- batic entrance length rather than the mixing chamber temperatures. In addition, the more recent values of the thermophysical salt properties given in Table 7.1 were used, Preliminary results for runs 1 through 27 are given in Table 7.2. The odd-numbered runs were without (flow from left to right) and the even-numbered runs were with (flow from right to left) the 24-in. entrance length. Examining runs 1 through 12 (5000 < Ny, < 8000), one notes little variation in the heat transfer function, 4. MSR Program Semiannu. Progr. Rep. Aug. 31, 1967, ORNL-4344, pp. 96—100. 65 ORNL-DWG 71-6715 2-ft HEA]}D SECTICN 2-ft ENTRANCE LENGTH % s 7 N Lrél—ltingocoupLE //; . 3g-in. . RADIUS (TYP) ] Z//;‘“n ‘_“.‘i‘L“‘,‘L‘“I‘:‘_- fh / \ EEE'Egé. ot ' II | } 7 11 \1 1 - || “; % WQ »\‘ rg\l iy i A\*/,G-in. RADIUS (TYP) N £ 0.50-in. OD x 42-mil WALL 0.25-in. OD x 35-mi HASTELLOY N TUBING n X mil WALL HASTELLOY N TUBING "DISK AND DONUT" MIXERS 1-in OD x 60-mil WALL HASTELLOY N TUBING Fig. 7.1. Details of test section, entrance length, mixing chambers, and electrodes for pressurized flow system. " Table 7.2. Experimental data for heat transfer studies using the salt mixture LiF-BeF,-ThF4-UF,4 (67.5-20-12-0.5 mole %) Even-numbered runs were with and odd-numbered runs without an entrance length Modulus® h q/A ' Heat transfer i‘:}n (Zi:n) ?olg)t Aon (Btuhr™! ft ™2 bliea:;e — p— = Btu hr! ft=3 CF)"! function, ' 5 x 1075) alance N, Np, Ny, [P0M ] Ng.r 1 1085.7 1089.2 10.7 0.24 0.24 5208 15.1 48.6 2234 19.72 2 1090.1 1093.7 10.2 0.24 1.26 5197 149 51.2 2356 20.89 3 1090.7 11022 44 4 0.80 0.78 5483 14.7 50.5 2322 20.50 4 1099.5 11113 34.1 0.80 1.15 5486 14.2 50.8 2338 20.83 5 1106.1 11299 74.2 1.69 1.00 6051 136 49.7 2285 20.32 6 1118.3 11424 74.6 1.69 1.19 6211 13.1 494 2273 20.49 7 11324 1156.1 70.1 1.69 1.006 6512 12.5 525 2416 22.15 8 11395 1186.6 1209 3.38 1.09 7056 11.8- 60.8 2798 25.70 9 11684 12154 150.8 3.38 1.01 7696 10.8° 66.3 3049 28.96 10 1181.3 12286 1112 3.37 1.08 7941 104 66.1 3039 29.24 11 12034 1226.7 59.9 1.69 1.03 8230 10.2 61.4 2823 27.89 12 1203.2 1227.6 62.6 1.69 1.16 7883 10.2 587 2698 26.64 13 10646 1077.6 67.3 0.59 1.00 3313 16.0 19.3 887 7.50 14 1068.8 10814 58.6 0.59 1.07 3152 15.8- 219 1006 8.57 15 1077.2 10999 105.8 0.99 1.01 3370 15.0 204 937 7.96 16 1082.7 1106.6 85.8 0.99 1.08 3264 14.7 25.1 1153 9.94 17 10934 1129.5 1468 1.5 0.98 3467 139 224 1031 8.86 18 11044 11422 131.0 1.5 1.09 3352 134 242 1113 9.75 19 11185 11557 136.1 1.5 1.02 3554 12.8 235 1082 9.60 20 1059.6 1093.0 1409 1.5 1.05 3251 156 226 1038 8.58 21 10763 1109.3 1474 14 1.00 3447 148 215 988 8.30 22 1089.3 11355 170.7 2.0 1.13 3591 13.8 2511 1157 9.87 23 1114.2 11636 1898 2.0 1.09 3749 127 234 1076 9.40 24 1125.8 12043 2544 29 1.11 3659 116 246 1134 10.01 . 25¢ - 26 1080.3 11509 253.5 2.9 1.02 3524 13.6 26.0 1197 10.02 27 1113.8 11874 2728 29 1.03 3787 122 235 1081 9.33 %These are the Reynolds, Prandtl, and Nusselt moduli, respectively, calculated using the average of the local heat transfer coetg'lcient from the exi/t to within 5 in. of the test-section entrance. — A7 a7 1/3 0.14 CNS-T=NNu/NPr (.U/.U-s) . This run was deleted because flow was not constant. 66 ORNL-DWG 74-6716 16 | ] T | — ||/ 4N FROM W.M. McKAY, Npe =3260 Np =15.0 7 - CONVECTIVE HEAT _ 5 B \\ AND MASS TRANSFER (REF.5) GA=1X10" fgyiy =160 12 iy \ \sm_.__d Y T e \\// AN 1.0 :_“:E-— RUN 16 / RUN 17— 0.6 1.4 == G5 2 \ Npe=3525 Np, =13.6 Ol Q/A =3%105 hggiy=1300 [ Ak 8|3 A 2 40 oL - & Lg \‘--.__ RUN 26 | -l o8 N 1 Yi= ' RUN 27 =" Q|5 06 : O | 4l 14 4'::': \ S Npe = 5500 Np, =14.4 G/A=08%10° h,;4=2250 \ RUN 4 ——TRUN 3 08 | 14 o NRe=86500 Np,=12.8 "2 \ O/A=A7%40° Fexiy=2250 / P—— RUN 7 0.8 I 0 ' 20 40 60 80 100 120 DISTANCE FROM ENTRANCE TUBE INSIDE DIAMETER X/D, Fig. 7.2. Variation in ratio of local to exit heat transfer coefficient with distance from the entrance for the salt mixture LiF-BeF,-ThF4-UF, (67.5-20-12-0.5 mole %). Even-numbered runs with, odd-numbered runs without, an entrance length. NS-T_[NS-T = (NNU/NEI’P)(“/MS)O'“" where NNu and Np, are the average Nusselt and Prandtl moduli (based on all but the first 5 in. of the test section) respectively; u/u; is the viscosity ratio, bulk fluid to wall], beyond that which would be expected from the variation in Reynolds modulus. However, for runs 13 through 27 (3150 < Ng, < 3787), the heat transfer function averages 15% greater with a hydrodynamic entrance length than without. The influence of the entrance region on the heat transfer coefficients can be more easily seen from plots of the ratio of local to exit heat transfer coefficient as a function of axial position, as shown in Fig. 7.2. (The exit value obtained with the entrance length was used for each pair of plots.) Also shown in Fig. 7.2 for comparisons are the expected curves for similar en- trance and flow conditions extrapolated from published results for the simultaneous development of hydro- dynamic and thermal boundary layers.® Examining the lower-heat-flux, lower-Reynolds-modulus case first (upper curves), one can see that, with an entrance length (run 16) fully developed heat transfer coef- ficients are obtained within an L/D of 30, which is in agreement with expected behavior. However, with the round-edged nozzle-type entrance (run 17) full develop- ment was not obtained within the length of the tube. With a higher heat flux, fully developed heat transfer was not attained within the expected L/D of 40 even with an entrance region (run 26). At a Reynolds modulus of about 6000, there is less distinction between heat transfer development with (runs 4 and 6) or without (runs 3 and 7) the entrance region; and in both cases the the length required for constant heat transfer coefficients to be attained is in good agreement with the predictions of ref. 5. The undeveloped heat transfer behavior that has been observed in this study at low Reynolds moduli is difficult to explain in terms of commonly held theories regarding the combined development of the hydro- dynamic and thermal boundary layers in a circular tube with a constant heat flux. In fact, the thermal boundary layer should develop more rapidly for higher Prandtl modulus fluids, and the local heat transfer coefficient should be larger than the exit value until fully de- veloped conditions are reached. However, a theoretical study for incompressible flow has shown, in general, that the flow of a fluid whose viscosity has a large negative temperature dependence will be stabilized by 5. W. M. McKay, Convective Heat and Mass Transfer, pp. 186—-96, McGraw-Hill, New York, 1966. 67 neating and destabilized by cooling.® Specifically, computer results show that the critical Reynolds modulus (based on boundary layer thickness) for water at 60°F flowing over a heated flat plate can be increased by a factor of up to 22 above the expected isothermal value, depending on the heat flux, Thus the - distance required for the attainment of fully developed turbulent flow in the molten-salt system (du/d¢ for the salt at 1100°F is five times larger than for water at 60°F) could be greater than the length of the heated test section. Unfortunately, except for studies of gas flow over heated airfoils, no experimental results are available to quantitatively verify those theoretical studies. Our experimental system, however, with its capability of reversible flow, matched temperature measurements, uniform heat flux, and absence of pump-induced turbulence, is uniquely suited to measure precisely the effect of an adiabatic entrance region on the heat transfer of a moderately viscous fluid near its melting point. The local heat transfer coefficients and temperature profiles obtained from these measurements can then be used to check the stability theory. 7.2 THERMOPHYSICAL PROPERTIES J. W. Cooke 7.2.1 Wetting Studies The new technique' for determining the wetting characteristics of liquids has been used to study the wetting behavior of the molten salt LiF-BeF,-ZrF,- ThF,4-UF,; (70-23-5-1-1 mole %) on a Hastelloy N surface at 700°C (1292°F). The primary objective of this study was to determine the extent to which the wetting behavior of this salt could be controlled. The wetting behavior of several molten fluoride salt mix- tures has been found to be affected by addition of various metals to the melt.” The controlling mechanism is believed to be the oxidation state of the melt, which affects the liquid-solid surface energy level. In the present investigation, zirconium metal (in the form of a-0.25-in.-OD Zircaloy rod) was added to the melt to make it more reducing; nickel fluoride (in powder form) was added to make it more oxidizing. 6. A. R. Wazzan, “The Stability of Incompressibie Flat Plate Laminar Boundary Layer in Water with Temperature Depend- ent Viscosity,” pp. 184-202 in Proceedings of the Sixth Southeastern Seminar on Thermal Sciences, Raleigh, N. C., Apr. 13-14, 1970, 7. MSR Program Semiannu. Progr. Rep. Feb. 28, 1969, ORNL-4396, p. 205. Over a three-day period, 2000 observations of the contact angle were made. Since the initial intention of this study was to observe only the change from nonwetting to wetting conditions, no attempt was made to obtain precise values of the contact angle. By comparing the actual pressure traces in Fig. 7.3 with the predicted traces,' we infer that the molten salt mixture did not initially wet the Hastelloy N surface (contact angle in the range of 140 to 160°). However, a pressure trace, shown in Fig. 7.4, typical of wetting (contact angle ~20°) was obtained within 20 min after the introduction of the Zircaloy rod into the melt. Two hours later, 1 wt % of nickel fluoride powder was added ~ to the melt. Five hours after adding the powder, the melt was only partially wetting (contact angle ~90°), as shown in Fig. 7.5. The melt became nonwetting again 7 hr after adding the nickel fluoride, with the pressure traces again resuming the shape shown in Fig, 7.3. Based on these results, a new apparatus was con- structed to enable a more detailed study of the electrochemical processes involved in the wetting be- havior of metals in molten salts. To this end, a micromanometer has been added to this system to obtain more precise values of the contact angle. The system is currently being used to investigate the wetting behavior of the salt mixture LiF-BeF,-ThF,-UF, ORNL-DWG 74-6717 [FY] v > v wn L v a TIME Fig. 7.3. Pr&gsure trace with Hastelloy N unwetted by the salt mixture at 700 C. 68 (67.5-20-12-0.5 mole %) when beryllium metal is added to the mixture. - ORNL-DWG 74-6718 L/ PRESSURE TIME Fig. 7.4. Pressure trace with Hastelloy N wetted by the salt mixture at 700 C. ORNL-DWG 71-6722 —w PRESSURE TIME Fig. 7.5. Pressure trace with Hastelioy N partially wetted by the salt mixture at 700°C. ' 7.2.2 Thermal Conductivity Calibrations of the improved variable-gap thermal conductivity apparatus were completed using water at 12°C. The plot of overall thermal resistance as a function of gap thickness shown in Fig. 7.6 displays the precision of this apparatus. Qur measured value of 0.00622 for water at 12°C agreed with published values® to better than 5%, a factor of 3 less error than was associated with water tests using the previous apparatus. Since the apparatus is designed for operation with corrosive fluids at temperatures up to 850°C, such agreement with the specialized room temperature meas- urements published in the literature is outstanding. Soon after measurements were initiated with the molten salt LiF-BeF, (66-34 mole %), a leak developed in a weld in the concentrically guarded heat meter. Attempts to repair this weld were unsuccessful, and a new heat meter was fabricated. The eloxing process was utilized in the fabrication of the heat meter so that the welds could be eliminated. Conductivity measurements for the salt mixture LiF-BeF; (66-34 mole %) have not yet been resumed. 8. A. R. Challoner and R. W. Powell, “Thermal Conduc- tivities of Liquids: New Determinations for Seven Liquids and Appraisal of Existing Values,” Proc. Roy. Soc. A238, 90—106 (Dec. 4, 1956). ORNL-DWG 7{-6719 [ P . 30 T / z o § 25 o / - § 20 e a P [ L v o / w T : | 2 / (14 w40 // = L} - a € 5 wl > C ~ q% 0 0 002 004 006 008 010 02 014 Ax_ SPECIMEN THICKNESS (cm) Fig. 7.6. Data obtained in measurement of the thermal conductivity of water at 12°C using the improved variablegap apparatus, 69 7.3 MASS TRANSFER TO CIRCULATING BUBBLES T. S. Kress 7.3.1 Experiment A transient technique is being employed in the MSBR-related mass transfer studies to determine mass transfer rates of a gas dissolved in a turbulently flowing liquid to cocirculating helium bubbles.® Experiments have been performed using helium bubbles having mean diameters from 0.015 to 0.05 in. to extract oxygen from five mixtures of glycerin and water (Schmidt moduli of 419, 719, 1228, 2015, and 3446) over a Reynolds modulus range from 8.1 X 103 to 1.6 X 10°. Both horizontal and vertical flow in a 2-in.-diam conduit have been studied. It was previously reported'® that the bubble diameter distribution is described adequately by a relation that, when integrated, gives the interfacial area per unit volume, a,, as a function of the volume fraction, ®, and the number of bubbles per unit volume, V,: ay=%Gn/2 B N3 @23 (1) Note that it is necessary to evaluate a, in order to extract mass transfer coefficients from the data relating concentration to time. By making a small adjustment in the coefficient to allow for the velocity due to buoyancy, the volume fraction for flow in vertical conduits is approximated by: b= 0.73Qg/QL , (2) where Qg/QL is the volumetric flow ratio of gas to liquid. In horizontal flow, stratification of the bubbles affects the volume fraction to the extent that the vertical flow approximation applies only at high flows. It has been found experimentally that the horizontal- flow volume fraction can be correlated with the ratio of the mean axial velocity of the fluid to the bubble buoy- ant velocity in the radial direction, V,/V,. Assuming, following Peebles and Garber,'' that the terminal velocity, V,, is given by: V,= (4g‘7/3c)1 2 > 9. MSR -Program Semiannu. Progr. Rep. Aug. 31, 1968, ORNL-4344, pp. 74-75. 10. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 57-59. 11. F. N. Peebles and J. G. Garber, “Studies on the Motion of Gas Bubbles in Liquids,” Chem. Eng. Progr. 49(9), 95 (Feb- ruary 1953). where c=18.7/(V,dplw)°-** the desired velocity ratio becomes: ‘Va/Vr = 0.0188QL(;J/,0)0'5 1 5/31 273 The variation of measured volume fraction with this ratio is shown on Fig. 7.7 along with the corresponding least-squares curve: & =0.00179 + 0.021/(V,/V,) . (3) Using Eqgs. (1), (2), and (3) to establish the interfacial area, typical experimental mass transfer data for a 25% mixture of glycerin and water (Ng. = 1228) can be compared, as is seen in the upper portion of Fig. 7.8. Those results include mass transfer which occurs down- stream from the test section in the bubble separator. Certain aspects of those data are particularly revealing. ORNL-DWG 71-6720 . T T T T 0.515, _1.272 BASED ON I,/V,=0.0188 @ (u/p) J 0.008 F— u/ 3 H/p / ] WHERE ©=LIQUID FLOW,gpm p=LIQUID VISCOSITY, Ibm ft~' hr™! p= LIQUID DENSITY, Iby/ > 0.007 & = BUBBLE MEAN DIAMETER, in. __| 0.006 = o G 0.005 ) =] 0.003 0,d[2D)=0, a0 d . 3. —a—"-‘-:Oatr*= 55 (* 1/®)'/3 . . 71 ORNL-DWG T7i1-672{ % ! —t 4 | MASS TRANSFER IN TEST ad SEGTION AND SEPARATOR [ [a]ie; Al A > e A P & 7 T n £ ._/’ N 7 MASS TRANSFER IN TEST SECTION 7 < - / /a ! (SEPARATOR MASS TRANSFER E o p==—=1= 7] kuunten ol s ] . P ¢ - ,.,/ X SUBTRAGCTED) S ESTIMATED VERTICAL® '@ fA f' /| b FLOW BEHAVIOR; \ ?3; ; § ! \""'IZ? 5/3 o ye) | T I 1] L A Qg/@_=0.003 ) [ dc g L_ . — z % GLYGERINE = 25 T o {SGHMIDT MODULUS ~1228) ] L, 0.5 CONDUIT DIAM =2.0in. - (%] g GRAVITATIONAL ORIENTATION MEAN BUBBLE DIAMETER < VERTICAL HORIZONTAL (in.) . o 0.045 n a] 0.02 A A 0.035 0.2 -_—— 0.4 ] { ‘ ‘ | 103 2 5 104 2 5 10° 2 5 108 REYNOLDS MODULUS Fig. 7.8. Measured mass transfer coefficient as a function of flow Reynolds modulus, bubble mean diameter, and gravitational orientation. Comparison of data before and after adjustments for the amount of mass transfer occurring in the separator. Using Eq. (4) as the basic analytical description, a theory of mass transfer from a turbulent liquid to circulating bubbles involves establishing a relationship between the relative eddy transport coefficient (u,/10), the fluid properties, and the turbulence characteristics of the field; that is, a function is desired such that U8 =frs, d, Ny, Nge) - ‘ Eddy diffusion coefficients for turbulent flow in pipes have been measured. The correlation of Groen- hof'? for the center-line region of the pipe is assumed to apply: u,=40Xx 10 J7, /oD, 12. H. C. Groenhof, “Eddy Diffusion in the Central Region of Turbulent Flows in Pipes and between Parallel Plates,” Chem. Eng. Sci. 25,1009 (1970). which can be converted to 1,/ =0.04Ng Ng,'® (5) While there have also been measurements of the variation in the coefficients with distance from a surface in shear fields with stationary no-slip interfaces, these results would not be expected to apply to a cocirculating bubble interface which may be either rigid (no slip) or mobile (a moving interface with internal circulation). Therefore an analytical expression had to be developed relating the eddy diffusivity to the concentration gradient and the individual velocity and frequency components of a Fourier analysis of the turbulence. Representing the turbulence field by super- position of sinusoidal idealized viscous “eddy cells” (a quasi turbulence), an integral expression has been obtained for the variation of eddy diffusivity with distance from an interface that may be either rigid or mobile. This variation, along with Eqgs. (4) and (5), is being programmed for solution on a digital computer. Nomenclature a4 = bubble interfacial area per unit volume C = drag ¢oefficient .d = bubble Sauter-mean diameter D = conduit diameter 1) = molecular diffusion coefficient | g = gravitational acceleration N, = number of bubbles per unit volume Qg = volumetric flow of gas bubbles @, = volumetric liquid flow r = spherical radial coordinate measured from center of a moving bubble V, = liquid mean axial velocity V., = bubble terminal velocity x = axial coordinate 72 Greek letters @ = concentration of a dissolved constituent 6o = radial average concentration of a dissolved con- stituent at x =0 M =liquid viscosity M, = turbulent eddy diffusivity p = liquid density 7,, = wall shear stress - ® = volume fraction occupied by the bubbles Dimensionless quantities 0.=0/6, re=rfD xXe =x/D Ny = V,Dolu Ng. =ulpl) Part 3. Chemistry W. R. Grimes The chemical research and development activities described below are conducted to establish the basic chemical information required for the development of advanced molten-salt reactor systems. A substantial fraction of these efforts continued to be devoted to the transport, distribution, and chemistry of fission products in the MSRE. Similar efforts seek to establish the nature and control of the interactions of tritium with molten salts, metal alloys, and graphite. Investigations of fission product behavior have been continued with specimens removed from the MSRE fuel circuit and by investigation of the chemistry of molyb- denum, niobium, and ruthenium in molten fluoride mixtures. Investigations into the chemistry of sodium fluoro- borate have continued and have been extended in efforts to identify the factors which will ultimately determine the applicability of fluoroborates as coolants. A broad program of fundamental investigations into the physical chemistry of molten-salt systems was maintained; from it are derived the basic data for reactor and chemical reprocessing design. Within the scope of these efforts are included research in solution thermodynamics and phase equilibria, crystal chem- . istry, electrochemistry, spectroscopy (both Raman and electronic absorption), transport processes, and theoret- ical aspects of molten-salt chemistry. Studies of the chemical aspects of separations meth- ods were continued. The results of these studies form the basis for evolving modifications of methods for reprocessing MSBR fuel salts. With adoption of a reference design for MSBR fuel reprocessing which effects transfer of rare earths from liquid bismuth to lithium chioride as an acceptor solvent, emphasis has been given to development of innovative means for separation of rubidium, cesium, and the rare earths from the fuel solvent and to the removal of solutes from the liquid metal extractant. 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 electroanalyti- cal 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. 8. Fission Product Behavior 8.1 DETERMINATION OF TRITIUM AND HYDROGEN CONCENTRATIONS IN MSRE PUMP BOWL GAS S.S. Kirslis F. F. Blankenship The dimensions of the problems caused by tritium diffusion through metal walls of a molten-salt reactor have been estimated using a particular mathematical model.! The validity of the model could be confirmed by measurements of the tritium concentration in .the 73 MSRE pump bowl gas. Therefore several different types of sample were taken while the MSRE was operating, and apparatus was constructed and tested for analyzing those samples. Since hydrogen partial pressure strongly influences the solubility and permeation rate of tritium from a gas mixture into a metal, it was also necessary to 1. R. B. Briggs and R, B. Korsmeyer, Distribution of Tritium in a 1000-MW(e)j MSBR, ORNL-CF-70-3-3 (Mar. 18, 1970) (internal memorandum). determine the hydrogen concentration in the pump - bowl gas. Several types of tritium sampling devices were used in the MSRE pump bowl. Advantage was taken of the unique properties of hydrogen and its isotopes in that they are highly soluble in and permeate rapidly through metals at red heat. The same properties would make it difficult to obtain a representative gas sample using an evacuated metal bulb with a resealable freeze valve. The simplest sampling device was a solid nickel cylinder, % in. in diameter and 6 in. long. The solubilities of hydrogen and tritium are such that at the expected partial pressures, the solid nickel bar at solubility equilibrium at 650°C would contain about ten times as much tritium per cubic centimeter as the surrounding gas. In the several samplings, exposure times in the pump bowl gas varied between 2 and 10 hr. The bar was pulled up into a cool section of the sampling line about 2 ft above the pump bowl and allowed to cool in essentially the same atmosphere. Each saturated nickel bar was taken to a hot cell, decontaminated in dilute nitric acid and 1 M H,0,—1 M oxalic acid, and stored in a hood in a Dewar flask under liquid nitrogen. A variation of this sampling method was a sealed thin-walled nickel capsule filled with nickel powder which had been deoxidized with hydrogen and gvacu- ated at red heat. The thought was that hydrogen and tritium should saturate the capsule wall and powder particles much more rapidly than in the case of the solid bar. A third sampling device was a thin-walled nickel capsule containing degassed CuO “wire.” At 650°C, ORNL—DWG 71-7244 TO GAS MANIFOLD BOURDON GAGE THERMOQ - QUARTZ COUPLE TUBE GAGE .| U POWDER UTs TUBE / FURNACE L4 IONIZATION / CHAMBER Fig. 8.1. Calibration apparatus. gaseous H, or HT permeating the capsule wall would react with the CuO to produce H, O and copper metal. The H,O or HTO could not diffuse back through the capsule wall and would thus be trapped in the capsule. Exposure times in the MSRE pump bowl gas phase varied between 2 and 10 hr for this type of capsule. A single capsule was prepared which was identical to the one just described except that one hemispherical end of the capsule was made of thin palladium sheet welded to the cylindrical nickel capsule body. Since palladium is much more permeable to hydrogen than nickel, a faster collection of H, O and HTO should have resulted. All of the different capsules were decontaminated and stored under liquid nitrogen in a Dewar flask, as described for the nickel bar samples. 8.1.1 Calibration Apparatus From the total amount of tritium and hydrogen found in the sample capsules or bars, it is theoretically possible to calculate how much tritium and hydrogen were in the MSRE pump bowl atmosphere, using capsule dimensions and literature values for the solu- bility of hydrogen in nickel and for the permeability of hydrogen through nickel at 650°C. However, in the case of the nickel bars, there is doubt whether the interior of each bar was fully saturated and doubt as to the fraction of dissolved gas lost during cooling. The latter doubt is of still greater concern for the nickel powder capsule. A more basic concern is whether Sieverts’s law (the solubility of hydrogen in a metal is proportional to the square root of the hydrogen pressure) is valid at the low pressures we are dealing with. It has been suggested ‘in the literature that the law fails at low pressure, since for many gases the extrapolations of plots of hydrogen solubility vs square root of pressure do not go through the origin. It is desirable to check this point since it has an important bearing on the effect of added hydrogen pressure on tritium permeation through metals. Finally, while no faults with the CuO method of sampling are obvious a priori, it is desirable to check new analytical tools and methods under conditions closely simulating those of actual use. For those reasons, a calibration apparatus (Fig. 8.1) was built in which the sampling devices could be tested in tritium and hydrogen concentrations similar to those expected in the MSRE pump bowl gas. The vertical quartz tube in which the sampling device was suspended was part of a thermal-convection loop constructed mainly of Pyrex tubing. Side loops through which the convective flow could be directed contained a tritium gas counter and a uranium powder—uranium hydride—uranium tritide trap. The counter was an ionization chamber operated at 500 V with the output current read with a vibrating reed electrometer. The uranium hydride—tritide trap could be heated to supply a pure dry mixture of 98% H, and 2% HT to the convecting gas. By cooling the trap the H, and HT could be resorbed on the powdered uranium. Provision was made for sampling the gas in the loop into a removable glass bulb. A side arm from the loop contained a Bourdon gage (vacuum to 15 psig) and a thermocouple gage. All glass valves in the loop system were of the greaseless O-ring variety in order to minimize exchange of tritium with the hydrogen in stopcock grease. Finally, the loop was connected to a gas manifold by means of which the loop could be evacuated or pressurized with dry helium, dry argon, or dry hydrogen. The complete calibration apparatus was installed in a hood in case of accidental tritium release. A tritium “sniffer”” was installed very close to the loop to warn of any tritium leakage. ' To date, the calibration procedure has been carried out only a few times, using the solid nickel bar type of sampling device. The procedure involved first evacuat- ing the nickel bar suspended in the heated quartz tube (650°C). Then dry argon was admitted to. nearly atmospheric pressure and allowed to circulate through the uranium tritide trap and the ionization chamber. The temperature of the trap was adjusted to give the desired H, and HT concentrations and the gas allowed to circulate past the nickel bar for several hours. The counter indicated the constancy of the HT pressure. A sample of the circulating gas was taken in the glass bulb, and the quartz tube and nickel bar were cooled rapidly to room temperature. The uranium tritide trap was then cooled to room temperature to resorb the circulating tritium, and the loop was evacuated and flushed with dry argon to remove the tritium. The ground joint at the top of the quartz tube was then opened (with dry argon flow blanketing the loop) and the nickel bar removed. The nickel bar was then placed in the quartz tube of the extraction apparatus in which the tritium was extracted from the nickel bar by heating it at red heat for several hours in a 99.9% He—0.1% H, flow. The gas then passed through a CuO trap at 500°C to convert HT to HTO, and the HTO was finally caught in a series of distilled-water bubblers. The water was analyzed for tritium with great sensitivity by scintillation counting. The glass bulb sample was analyzed for hydrogen by mass spectrometry and for tritium using the same extraction apparatus but omitting” the quartz heating tube. 75 The original rather simple design of the calibration apparatus had to be modified and refined repeatedly in order to perform as desired. To achieve the required leak rate and degassing rate (mainly H, O which would exchange with and dilute HT), the apparatus had to be leak-tight to the limit of a helium mass spectrometer leak detector, and ordinary glass stopcocks had to be replaced with greaseless glass valves. Before a gas was admitted to the loop, it had to contain less than 10 ppm of water as measured by a Meeco electrolytic water meter. It was necessary to pass the convective flow through the uranium tritide trap and through the ionization chamber to achieve controllable and steady tritium concentrations. The ionization chamber had an unfortunate memory effect. After once introducing a high tritium pressure (~100 p) for a special test, it was not possible to get rid of a high background due to tritium dissolved in the thick stainless steel wall of the ionization chamber. Weeks of evacuation and flushing while warming the chamber did not succeed in reducing the background to a satisfactory extent. A Pyrex ionization chamber was designed and built, with silvered interior surfaces for electrodes. The silver film is so thin that degassing of tritium should be rapid. The new ionization chamber has not yet been checked with tritium gas. ' The most successful calibration tests with a nickel bar sampling device achieved only a qualitative agreement with the glass bulb samples. It is thought that the failure to obtain exact agreement was due either to changing hydrogen and tritium concentrations during the exposures or to variations from Sieverts’s law for hydrogen solubility at low pressures. This will be checked in future tests by taking glass bulb samples at the beginning and end of each test and by working at different tritium and hydrogen partial pressures. 8.1.2 Analysis for Hydrogen Analysis of ‘the sampling devices for tritium is much simpler than analysis for hydrogen, since the nickel contains only about 0.02 g of H, per gram and it is very difficult to avoid picking up small amounts of water from the surfaces of gas handling equipment. For this reason an analytical method was devised which involves a minimum of gas handling. The standard vacuum fusion method is not sensitive enough. The nickel bar and nickel powder samples will be heated in a quartz tube containing hot CuQ to convert the extracted H, to H,0. At one end of the quartz tube there will be a simple dew-point measuring device by means of which the H, O concentration in the tube can be measured. The main difficulty with this method is expected to arise in desorbing water sufficiently from the system and sample surfaces in the early stages of heating under evacuation, before the valve to the pump is closed and the sample heated further to liberate its hydrogen. In the case of the CuO sample capsules, the hydrogen is already in the form of water, and it is only necessary to puncture the capsule with a special puncturing valve and let the inside gas into a dew-point meter. The dew-point method does not destroy the sample, so that the analyzed sample can be flushed through water bubblers for tritium analysis. For the nickel bar and nickel powder samples, a method for analyzing the extracted gas for H, by mass spectrometer has been devised in case the tramp water problem turns out to be too difficult. 8.1.3 Tritium Diffusion Studies An alternative reason for constructing the calibration apparatus with its convective circulation, its tritium counter, and its tritium-hydrogen supply is that it is an ideal system with which to study the effect of hydrogen pressure on tritium diffusion through Hastelloy N, using realistic concentrations of tritium and hydrogen. The system will also be useful in testing the effects of various metal surface treatments on the diffusion process. 8.2 EXAMINATION OF DEPOSITS FROM THE MIST SHIELD IN THE MSRE FUEL PUMP BOWL E.L.Compere E.G. Bohlmann In January 1971 the sampler cage and mist shield were excised from the MSRE fuel pump bowl by using a rotated cutting wheel to trepan the pump bowl top (see Sect. 1.3). The sample transfer tube was cut off just above the latch stop plug penetrating the pump bowl top; the adjacent ~3-ft segment of tube was inadvertently dropped to-the bottom of the reactor ceil and could not be recovered. The final ligament attaching the mist shield spiral to the pump bowl top was severed with a chisel. The assembly, of mist shield spiral sur- rounding the sampler cage attached to the latch stop plug, was transported to the HRLEL for cutup and examination. Removal of the assembly disclosed the copper bodies of two sample capsules that had been dropped in 1967 and 1968 lying on the bottom of the pump bowl. Also on the bottom of the bowl, in and around the sampler area, was a considerable amount of fairly coarse granular porous black particles (largely black flakes ~2 76 to S mm wide-and up to 1 mm thick). Contact of the heated quartz light source in the pump bowl with this material resulted in smoke evolution, and apparently some softening and smoothing of the surface of the accurmnulation. Periscopic examination of the interior of the pump bowl is described in Sect. 1.3. One of the sample capsule bodies was recovered and examined as described in Sect. 1.3. A few grams of the loose particles were also recovered and transferred in a jar to the hot cells; a week later the jar was darkened enough to prevent seeing the particles through the glass. An additional quantity of this material was placed loosely in the carrier can. Samples were submitted for analysis for carbon and for spectrographic and radio- chemical analyses. The results are discussed below. The sampler assembly as removed from the carrier can is shown in Fig. 8.2. All external surfaces were covered with a dark gray to black film, apparently 0.1 mm or more in thickness. Where the metal of the mist shield spiral at the top had been distorted by the chisel action, black eggshell-like film had scaled off, and the bright metal below it appeared unattacked. Where the metal had not been deformed, the film did not flake off. Scraping indicated a dense, fairly hard adherent black- ish deposit. On the cage ring a soft deposit was noted, and some was scraped off; the underlying metal appeared un- attacked. The heat of sun lamps used for in-cell photography caused a smoke to appear from deposits on bottom surfaces of the ring and shield. This could have been material, picked up during handling, similar to that seen on the bottom of the pump bowl. At this time samples were scraped from top, middle, and bottom regions of the exterior of the mist shield, from inside bottom, and from the ring. The mist shield spiral was then cut loose from the pump bowl segment, and cuts were made to lay it open using a cutoff wheel. A view of the two parts is shown in Fig. 8.3. In contrast to the outside, where the changes between gas (upper half) and liquid (lower) regions, though evident, were not pronounced, on the inside the lower and upper regions differed markedly in the appearance of the deposits. In the upper region the deposits were rather similar to those outside, though perhaps more irregular. The region of overlap appeared to have the heaviest deposit in the gas region, a dark film up to 1 mm thick, thickest at the top. The tendency of aerosols to deposit on cooler surfaces (thermophoresis) is called to mind. In the liquid region the deposits were considerably thicker and more irregular than elsewhere, as if formed from larger agglomerates, 77 R 53633 Fig. 8.2. Mist shield containing sampler cage from MSRE pump bowl. In the area of overlap in the liquid region this kind of deposit was not observed, the deposit resembling that on the outside. If we recall that flow into.the mist shield was nominally upward and then outward through the spiral, the surfaces within the mist shield are evidently subject to smaller liquid shear forces than those outside or in the overlap, and the liquid was surely more quiescent there than elsewhere. The condi- tions permit the accumulation and deposition of ag- glomerates. The sample cage deposits also were more even in the upper part, becoming thickest at and on the latch stop. The deposit on the latch stop was black and hard, between 1 and 2 mm thick. Deposits on the cage rods below the surface (see Figs. 8.4 and 8.5) were quite irregular and lumpy and in general had a brown-tan (copper or rust) color over darker material; some whitish material was also seen. Four of the rods were scraped to recover samples of the deposited material. After a gamma radiation survey of the cage at this time, the unscraped cage rod was cut out for metallographic examination; another rod was also cut out for more thorough scraping, segmenting, and possible leaching of the surfaces. The gamma radiation survey was conducted by lowering the cage in ‘-in. or smaller steps past a 0.020- by 1.0-in. horizontal collimating slit in 4 in. of lead. Both total radiation and gamma spectra were obtained using an Nal scintillation crystal. The radia- tion levels were greatest in the latch stop region at the top of the cage and next in magnitude at the bottom ring. Levels along the rods were irregular but were higher in the liquid region than in the gas area even though considerable material had been scraped from four of the five rods in that region. In all regions the spectrum was predominantly that of 367-day '°®Ru and 2.7-year '258b, and no striking differences in the spectral shapes were noted. . R 54220 Fig. 8.3. Interior of mist shield. Right part of right segment overlapped left part of segment on left 79 R 54219 Fig. 8.4. Sample cage and mist shield. R 54189 Fig. 8.5. Deposits on sampler cage. Ring already scraped. 80 Analyses of samples recovered from various regions inside and outside the mist shield and sampler cage are shown in Table 8.1. The samples generally weighed between 0.1 and 0.4 g. The radiation level of the samples was measured using an in-cell G-M probe at about 1 in. distance, and at the same distance with the sample surrounded by a %-in. copper shield (to absorb the 3.5-MeV beta of the 30-sec '°®™Rh daughter of 106Ru). Activities measured in this way ranged from 4 R/hr (2 R/hr shielded) to 180 R/hr (80 R/hr shielded). the latter being on a 0.4-g sample of the deposit on the latch stop at the top of the sample cage. Spectrographic and chemical analyses are available on three samples, (1) the black lumpy material picked up from the pump bowl bottom, (2) the deposit on the latch stop at the top of the sample cage, and (3) material scraped from the inside of the mist shield in the liquid region. The material recovered from the pump bowl bottom contained 7% carbon, 31% Hastel- loy N metals, 3.4% Be (18% BeF,), and 6% Li (22% LiF). Quite possibly this included some cutting debris. The carbon doubtless was a tar or soot resulting from thermal and radiolytic decomposition of lubricating oil leaking into the pump bowl. It is believed that this material was jarred loose from upper parts of the pump bowl or the sample transfer tube during the chise! work to detach the mist shield. The hard deposit on the latch stop contained 28% carbon, 2.0% Be (11% BeF,), 2.8% Li (10% LiF), and 12% metals in approximate Hastelloy N proportions, again possibly to some extent cutting debris. The sample taken from the inner liquid region of the mist shield contained 2.5% Be (13% BeF,), 3.0% Li (11% LiF), and 18% metals (with somewhat more Cr and Fe than Hastelloy N); a carbon analysis was not obtained. In each case, about 0.5 to 1% Zr (~1 to 2% ZrF, ) was found, a level lower than fuel salt in proportion to the lithium and beryllium. Uranium analyses have not yet been received, so we cannot clearly say whether the salt is fuel salt or flush salt. Since fission product data suggest the deposits built up over appreciable periods, we presume fuel salt. In all cases the dominant Hastelloy N constituent, nickel, was the major metallic ingredient of the deposit. Only in the deposit from the mist shield inside the liquid region did the proportions of Ni, Mo, Cr, and Fe depart appreciably from the metal proper. In this deposit a relative excess of Cr and Fe was found, which would not be attributable to incidental metal debris from cutting operations. It is also possible that the Table 8.1. Chemical and spectrographic énalysis of deposits from mist shield in the MSRE pump bowl - Radiation level Sample (R/hr @1 in.) Weight Percent . T_l ;. Percent Li Percent Be Percent Zr* Percent Ni’ Percent Mo? Percent Cr? Percent Fe? Percent Mn? Location , . (mg) C (dismin" g™ ") (shlelded) Pump bowl 10(5) 402 3.1 E10 6.00 3.42 0.5-1.0 20-30 2-4 1-2 0.5-1.0 0.5-1.0 bottom 25(12) 108 7.1 Latch stop 130(60) 291 1.85 El11 2.75 2.01 0.5-1.0 5-10 2-4 0.5-1.0 0.5-1.0 <0.5 180(80 365 28 _ Inside, 40(17) 179 4.7 E10 3.03 2,52 0.5-1.0 - 5-10 24 3-5 2—-4 <0.5 liquid region MSRE fuel - 11.1 6.7 11.1 (nominal) Hastelloy N 69 16 7 5 ~1 (nominal) aSemiquantitative spectrographic determination. I8 various Hastelloy N elements were all subject to mass transport by salt during operation, and little of that found resulted from cut-up operation. 8.2.1 Tritium Tritium determinations have been received on samples of material from the pump bowl bottom (14 mCi/g), latch seat (83 mCi/g), and inside liquid region of the mist shield (21 mCi/g). Tritium may well have ex- changed with the hydrogen of the oil entering the pump bowl and thereby been retained with any tar deposits. The loose material recovered from the pump bowl bot- tom contained 8 X 107° atom of T per atom of C, and the latch stop deposit contained 1.2 X 107 atom of T per atom of C. The agreement of the two values is useful. For comparison, assume oil (—CH,—) entered the pump bowl at w g/day, equivalent to 4.3 X 10*%w atoms of C per day and 8.6 X 10*?w atoms of H per day. Also at full power, 40 Ci/day of tritium was developed in the MSRE. If a fraction p does not pass through walls but remains available for exchange, we have 8.3 X 102%p atoms/day of T thus available. If a fraction X in fact does ex'change,. then there are 0.96 X 1072 (p/w)X atoms of tritium which are exchanged per atom of available hydrogen. If n hydrogen atoms are attached to each atom of carbon, then t T Xn oM —096x 1072222 atoms C w Because the observed value was ~107%, the ratio of observed to calculated indicates that pnX _ ~ 1072, w This does not seem unreasonable, since p is less than 1, w somewhat greater than 1, and » is probably 1 or less. We conclude thereby that X is of the order of at least a few percent. Probably more of this tar was carried out of the pump bowl by off-gas than remained. In the two determina- tions reported above we found 0.2 and 0.3 Ci of tritium per gram of carbon. If a (low) few grams of lubricating oil passed through the pump bowl and was associated similarly with tritium by exchange reactions, then up to a curie or so of tritium per day could enter the off-gas in this way. Presumably much would deposit or condense and not emerge from the off-gas system. Tritium as HT or CH3T, etc., would of course pass through. | 82 In an MSBR, utilization of exchange processes to affect tritium behavior would depend on additional factors which will not be considered here. We now come to consideration of fission product isotope data. These data are shown in Table 8.2 for deposits scraped from a number of regions. The activity per gram of sample is shown as a fraction of MSRE inventory activity per gram of MSRE fuel salt, to eliminate the effects of yield and power history; materials concentrated in the same proportion should have similar values. We first note that the major part of these deposits does not appear to be fuel salt, as evidenced by low values of ®°Zr and '**Ce. The values 0.13 and 0.11 for 144 Ce average 12%, and this is to be compared with the combined 24% for LiF + BeF, determined spectro- graphically as noted above. These would agree well if fuel salt had been occluded steadily as 24% of a growing deposit throughout the operating history. For '?7Cs we note that samples below liquid level inside generally are below salt inventory and could be occluded fuel salt as considered above. For samples above the liquid level inside, or any external sample, values are two to nine times inventory for fuel salt. Enrichment from the gas phase is indicated. Houtzeel? has noted that off-gas appears to be returned to the main loop during draining, as gas from the drain tanks is displaced into a downstream region of the off-gas system. However, our c'ieposit must have originated from something more than the gas residual in the pump bowl or off-gas lines at shutdown. An estimate sub- stantiating this is as follows. With full stripping, 3.3 X 10'7 atoms '37chain per minute enter the pump bowl gas. About half actually goes to off-gas, and most of the rest is reabsorbed into salt. If we, however, assume a fraction f is deposited evenly on the boundaries (gas boundary area ~16,000 cm?), the deposition rate would be ~2 X 10'3f atoms '37chain per square centimeter per minute. Now if in our samples the activity is / relative to inventory salt (1.4 X 10'7 atoms '37Cs per gram) and density is ~2, then the time ¢ in minutes required to deposit a thickness of X centimeters would be .o L4X10V X IX2X X 2X 10'3f A I = 14X 1047X,min. In obtaining our samples we generally scraped at least 0.1 g from perhaps 5 cm?, indicating a thickness of at 2. A. Houtzeel, private communication. Table 8.2. Gamma spectrographic (Ge-diode) analysis of deposits from mist shield in the MSRE pump bowl 99TC 95Nb 103Ru 106Ru lZSSb 127mTe 137CS 9521.0 144Cea Half-life 2.1 X 10° years 35 days 39.6 days 367 days 2.7 years 105 days 30 years 65 days 284 days (after °5Zr) Inventory, dis min ! g~! 24 uglg 8.3 E10 3.3 E10 3.3 E9 3.7E8 2.0E9 6.2 E9 9.9 E10 59E10 Sample activity per gram expressed as fraction of MSRE inventory activity/grams fuel salt? Pump bowl bottom, 0.23 36 52 20 17 2.1 0.13 + 0.03 0.10 loose particles ~ Latch stop 265 364 1000 5.7 69 31 0 0 Top . Qutside 122 167 328 104 98 9.5 0 0 Inside 42+ 14 237 £ 163 273 1000 69 4.3 0 0 Middle outside ’ 277 531 646 563 54 8.0 0 Below liquid surface - Inside No. 1 354 164 224 271 143 0.6 0 0.13 +0.02 Inside No. 2 463 148 292 310 72 98 1.1 0 0 ' Cage rod 305 221 692 198 187 0.2 0 0 Bottom . ' Outside (0, <60) (0, <60) 1210 167 232 - 3.2 0 0 Inside 16 £ 9 159 189 51 27 0.32 0 0.11 ?Background values (limit of detection) were as follows: °°Zr, 2-9 E10; '*%Ce, 1-2 E10; ' 3% Cs, 2-9 E8; ! '®Ag, 1-3 E9; ! 5*Eu, 1 E8—2 E9. bUncertainty stated (as * value) only when anappreciable fraction (>10%) of observed. €8 least ~0.01 c¢m, and [/ values were ~4, whence ¢ = 600/f. Thus, even if all (f ~1) the !'*7chain entering the pump bowl entered our deposits, 600 min flow would be required to develop their ! *?Cs content — too much for the 7-min holdup of the pump bowl, or even the rest of the off-gas system excluding the charcoal beds. It appears more likely that ! *7Cs atoms, from ' 37 Xe atoms decaying in the pump bowl, were steadily incorporated to a slight extent in a slowly growing deposit. The noble metal fission products, 35-day ?°Nb, 39.6-day '°3Ru, 367-day '°°Ru, 2.7-year '258Sb, and 105-day ‘2?7 Te, were strongly present in essentially all samples. In all cases 35-day °3Nb was present in quantities appreciably more than could have resulted from decay of ®° Zr in the sample. Antimony-125 appears to be strongly deposited in all regions, possibly more strongly in the upper (gas) region deposits. Clearly '*5Sb must be considered a noble metal fission product. '?7™Te was also found, in strong concentration, frequently in similar proportion to the 23 Sb of the sample. The precursor of 27" Te is 3.9-day '?7Sb. It may be that earlier observations - about fission product tellurium are in fact observations of precursor antimony isotope behavior, with tellurium remaining relatively fixed.' The ruthenium isotopes were present in quantities comparable to those of *>Nb, '25Sb, and '?7" Te, If the two ruthenium isotopes had been incorporated in the deposit soon after formation in the salt, then they should be found in the same proportion to inventory. But if a delay or holdup occurred, then the shorter-lived 193 Ru would be relatively richer in the holdup phase®: the activity ratio '°3Ru/!'°®Ru would exceed the inventory ratio, and material deposited after an appreci- able holdup would have an activity ratio ! °3 Ru/!°6Ru which would be less than the inventory value. Examina- tion of Table 8.2 shows that in all samples, relatively less '°?Ru was present, which indicates the deposits were accumulated after a holdup period. This appears to be equally true for regions above and below the liquid surface. Thus we conclude that the deposits do not anywhere represent residues of the material held up at the time of shutdown, but rather were deposited over an extended period on the various surfaces from a common holdup source. Specifically this appears true for the lumpy deposits on mist shield interior and cage rods below the liquid surface. 3. E. L. Compere and E. G. Bohlmann, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 60—66. 84 Data for 2.1 X 10%-year ?°Tc are available for one sample taken from the inner mist shield surface below the liquid level. The value 1.11 X 10* ug/g, vs inventory 24 ug/g, shows enhanced concentration ratio similar to our other noble metal isotopes and clearly substantiates the view that this element is to be regarded as a noble metal fission product. The consistency of the ratios to inventory suggests that the noble metals represent ~5% of the deposits. The quantity of noble metal fission products held up in this pump bowl film may not be negligible. If we take a median value of ~300 times inventory per gram for the deposited material, take pump bowl area in the gas region as 10,000 cm?® (minimum), and assume deposits 0.1 mm thick (~0.02 g/cm?; higher values were noted), the deposit thus would have the equivalent of the content of >60 kg of inventory salt. There was ~4300 kg of fuel salt, so on this basis deposits containing about ~1.4% or more of the noble metals were in the gas space. At least a similar amount is estimated to be on walls, etc., below liquid level; and no account was taken for internal structure surfaces (shed roof, deflector plates, etc., or overflow pipe and tank). Since pump bowl surfaces appear to have more (~10 times) noble metal fission products deposited on them per unit area than the surfaces of the heat exchanger, graphite, piping, surveillance specimens, etc., we believe that some peculiarities of the pump bowl environment must have led to the enhanced deposition there. We first note that the pump bowl was the site of leakage and cracking of a few grams of lubricating oil each day. Purge gas flow also entered here, and hydro- dynamic conditions were different from the main loop. The pump.bowl had a relatively high gas-liquid surface with higher agitation relative to such surface than was the case for gas retained as bubbles in the main Joop. The liquid shear against walls was rather less, and deposition appeared thickest where the system was quietest (cage rods). The same material appears to have deposited in both gas and liquid regions, suggesting a common source. Such a source would appear to be the gas-liquid interfaces: bubbles in the liquid phase and droplets in the gas phase. It is known that surface- seeking species tend to be concentrated on droplet surfaces. The fact that gas and liquid samples obtained in capsules during operation had '®3Ru/'°®Ru activity ratios higher than inventory® and deposits discussed here had '°?Ru/'°®Ru activity ratios below inventory suggests that the activity in the capsule samples was from a held-up phase that in time was deposited on the surfaces which we examined here. The tendency to agglomerate and deposit in the 85 less-agitated regions suggests that the overflow tank . may have been a site of heavier deposition. The pump bowl liquid which entered the overflow pipe doubtless was associated with a high proportion of surface, due to rising bubbles; this would likely serve to enhance transport to the overflow tank. The binder material for the deposits has not been established. Possibilities include tar material and per- haps structural or noble metal colloids. Unlikely, though not entirely excludable, contributors are oxides formed by moisture or oxygen introduced with purge gases or in maintenance operations. The fact that the mist shield and cage were wetted by salt nevertheless suggests such a possibility. 8.3 SYNTHESIS OF NIOBIUM FLUORIDES C.F. Weaver J.S.Gill Various methods for synthesis of NbF; and NbF, have been reported previously.*—8 In attempts to synthesize the fluorides of lower oxidation numbers, we have utilized the disproportionation of NbF,. Experi- ments conducted by heating NbF, in closed evacuated quartz tubes with one end at room temperature have suggested that the NbF, disproportionates in the temperature range 250 to 350°C under conditions which maintain an NbF; pressure of a few hundred microns. The disproportionation of NbF, at 350°C was found to produce a compound identified by x-ray diffraction as NbF3; (ASTM-9-168). This material is extremely sensitive to air, reacting to form a vapor and NbO, F, as identified by x-ray diffraction. The NbO, F decomposed above 300°C in a vacuum to form NbOF, vapor, identified with a mass spectrometer, and a residue of niobium oxides. In this regard the dispropor- tionation of the niobium compounds is similar to that of the molybdenum fluorides. 4. L. M. Toth, H. A. Friedman, and C. F. Weaver, MSR Program Semiannu. Progr. Rep. Feb. 29, 1968, ORNL-4254, p. 137. 5. L. M. Toth and G. P. Smith, Reactor Chem. Div. Annu. Progr. Rep. Dec. 31, 1967, ORNL-4229, p. 64. 6. F. P. Gortsema and R. Didchenko, Inorg. Chem. 4, 182-86 (1965). 7. C. F. Weaver et al., MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4548, pp. 124-29. 8. C. F. Weaver et al., MSR Program Semiannu. Progr. Rep. Aug 31, 1970, ORNL-4622, pp. 71-74. 8.4 REACTION KINETICS OF MOLYBDENUM AND NIOBIUM FLUORIDE IN MOLTEN Li, BeF, SOLUTIONS C.F.Weaver J.S.Gill The behavior of dilute solutions of trivalent molyb- denum fluoride in molten Li,BeF, at 500°C was summarized in ref. 8. The molybdenum left the system by a half-order process which was unaffected by helium flow rate, surface area of the copper container, presence of UF,, presence of graphite, or quantity of molyb- denum metal produced. The behavior of similar solutions at higher tempera- ture-9— 12 suggests that the mechanism of removal of Mo>* from molten Li,BeF, is different in the range 600 to 700°C than near S500°C. A pair of experiments conducted at 600°C emphasize this behavior. Both solutions were prepared at 500°C. On increasing the temperature to 600°C, molybdenum vanished from the solution. These experiments were performed with iden- tical procedures except that in one the solution was held at 500°C for only one day (Fig. 8.6), while in the second the solution was held for 1000 hr (Fig. 8.7) in a copper container with ten times the surface area present in the first case. Clearly, either the age of the solution or the surface area of the container had a pronounced effect on the rate of loss of molybdenum, although no such effects were observed at 500°C. Previous descriptions® of the behavior of niobium fluoride solutions have indicated the existence of an intermediate oxidation state, probably (IV), stable in molten Li, BeF, at 500° for periods as long as a month; this species was tound to ve reauciole by hydrogen only with difficulty. A value of Py /Py, 1/2 %107 atm!/2 was observed for the reaction NbF,(d) + 2H, = 4HF + Nb° with a concentration of 1000 to 1600 ppm of niobium in solution. 9. C. F. Weaver, H. A. Friedman, and D. N. Hess, Reactor Chem. Div. Annu. Progr. Rep. Dec. 31, 1967, ORNL-4229, pp. 36—37. ' 10. C. _F. Weaver, H. A. Friedman, and D. N. Hess, MSR Program Semiannu. Progr. Rep. Feb. 29, 1968, ORNL-4254, pp. 132-34, 11. C. F. Weaver, H. A. Friedman, and D. N. Hess, MSR Program Semiannu. Progr. Rep. Aug. 31, 1968, ORNL-4344, pp. 15455, : 12. C. F. Weaver et al.,, Reactor Chem. Div. Annu. Progr. Rep. Dec. 31, 1968, ORNL-4400, pp. 34-39. ORNL-DWG 71-7245 25 ‘\ oo \ d LN ) 'y 20 .\.\ [ ] E N, a ~ 2 CNG _ p=d 15 TN O = \ . e e L L J O ] § 12 liters /hr He FLOW \ R4 600°C \ 5 oy L 1Y . @ Q i) 500 1000 1500 2000 2500 3000 TIME (hr) Fig. 8.6. Removal of Mo™ from molten Li,BeF,. ORNL-~DWG 70-13508 35 T T T : . 500°C | 600°C —pr————— | ——— I.""'--..._.__ 30 e~ --..__.\ — . \\ T 2 \ a ® ‘-o; - \ & 20 |— : \ = o S 15 \‘ Q =z 8 \ ? {0 12 liters/hr He FLOW \ 5 \ L ] 0 \ ¢} 200 400 600 BOO 1000 1200 1400 TIME (hr) Fig. 8.7. Removal of Mo from molten Li, BeF,. We have now exposed these solutions at higher temperatures and noted instability. The niobium con- centration in molten Li, BeF, decreased at 700°C from 1200 to 950 ppm in 1190 hr (50 days) and at 900°C from 950 to 55 ppm in 847 hr (35 days) (Fig. 8.8). Since there was no evidence of corrosion of the copper container, it is assumed that the loss was a result of disproportionation of the niobium compound. The expected increase in nobility of the niobium with temperature was manifested by a value of =~1072 atm1/2 for Py F/PH2 1/2 3t 900°C and 3000 ppm of ORNL-DWG 71-7246 2 4°<.\ 700°C ! 900°C 10} —ve— e oy i\ \. ] \e CONCENTRATION {ppm) | \ 102 \ 0 400 800 1200 1600 2000 2400 TIME (hr) Fig. 8.8. Removal of Nb* from molten Li;BeF,. niobium in solution. The stability of these solutions contrasts sharply with the ease of decomposition of pure NbF,, as described in Sects. 8.3 and 8.5. The observation that NbF,, when dissolved in molten Li;BeF,;, seems to disproportionate in the operating temperature range for molten-salt reactors indicates that whenever oxidized niobium exists in the reactor fuel some NbFs must exist in the gas phase even though the species in solution has a lower valence. 8.5 MASS SPECTROSCOPY OF NIOBIUM FLUORIDES C.F.Weaver J.D. Redman Earlier work® 12-14 on the mass spectroscopy of niobium fluorides consisted in observing the pentafluo- ride polymers, the associated oxyfluorides, the fluorina- tion of niobium metal, and the disproportionation of the lower fluorides. Both the disproportionation of NbF, and the decomposition of NbO, F mentioned in Sect. 8.3 were followed by mass spectroscopy. The 13. C. F. Weaver et al., MSR Program Semignnu. Progr. Rep. Feb. 28, 1969, ORNL-4396, pp. 157—-62. ‘ 14. C. F. Weaver et al., MSR Program Semiannu. Progr. Rep. Aug. 31, 1969, ORNL-4449, pp. 113-21. 87 cracking pattern- for NbOF; was somewhat different The. tentative cracking patterns previously reported from that previously observed at 800°C, with the for Nb, F,, and NbF; have been confirmed with minor principal change being in the relative intensity of the refinement. The cracking pattern for NbF;, including NbOF; ion. These small changes, shown in Table 8.3, the intensities for the doubly charged ions, is shown in are probably temperature effects. Table 8.4. Table 8.3. Cracking pattern for NbOF4 Table 8.4. Cracking Pattern for NbF; Monomer Relative intensity Ion Relative intensity Ion = = " 800°C - 450-650°C NbF, 100 . NbF4 " 8 NbOF; ™" 50 68 NbF, " 13 NbOF, " 100 100 NbF* o NbOF* 13 8 Nb* 5 NbO™ 8 NbF,2* 0.1 No* 14 1 NbF, 2" 10 NbFy 1 5 NbF, 2+ . 5 NbF, " 13 12 NbE2* . 9. Coolant Salt Chemistry and Tritium Control The eutectic mixture formed from NaBF, and NaF (92-8 mole %), melting point 383°C, is the proposed MSBR coolant salt. In projected pump-loop experi- ments (see this report, Sect. 5), engineering experience with this salt will be gained over a period of several years. Although its cost and most of its chemical and physical properties are favorable for its adoption as the coolant, the fluoroborate salt suffers potential disad- vantages, as compared with the ? LiF-BeF, coolant used in the MSRE; the chief disadvantage is that the fluoroborate coolant exhibits a significant vapor pres- sure at operating temperatures and would exert an even higher vapor pressure if there was accidental mixing of the coolant and fuel salts. The strong possibility that the coolant salt will have to be the sink for the tritium produced in an MSBR places an additional criterion on this or any other coolant to be used in an MSBR; the coolant should be able to contain a sufficient concentration of hydrog- enous species to exchange isotopically with the tritium passing into the coolant circuit. At the same time the coolant with the hydrogenous additive should be compatible with its alloy containment system. Success in meeting the performance criteria requires very carefully conducted research efforts. The investi- gations described below are motivated by the desire to meet these criteria. At the current stage of develop- ment, none of these efforts has advanced to the point where it provides a clear solution to the problems associated with the application of fluoroborates as coolants, although some results seem encouraging. (R. E. Thoma) 9.1 STUDIES OF HYDROGEN EVOLUTION AND TRITIUM EXCHANGE IN FLUOROBORATE COOLANT S. Cantor R. M. Waller We seek to determine how much chemically bound hydrogen can be retained in molten fluoroborate coolant. To the extent that this hydrogen does not corrode metals in contact with the coolant, it is available for isotopic exchange with any tritium enter- ing the coolant circuit of an MSBR, 88 In these experiments evacuated nickel capsules con- taining NaBF, -NaF (92-8 mole %) and metal coupons are heated within a silica vessel which is connected to- gas-handling and pressure-measuring apparatus. Hydro- gen (or tritium) gas diffusing through the nickel capsule can be readily determined since vessels of fused silica are virtually impermeable to hydrogen. 'In a previously reported’ experiment, hydrogen (in concentrations equivalent to 60 ppm H,O) in the salt reacted with chromium coupons at 500°C and was completely converted to H,(g). The evolved hydrogen was initially established by gas-chromatographic analy- ses; the loss of hydrogen from the salt sample was confirmed by a subsequent tritium tracer analysis. Two experimental runs (see Table 9.1) performed with nickel as the only metal in contact with the salt indicated that low, but encouraging, levels of chemi- cally bound hydrogen can be retained in the salt. In both runs the salt contained hydrogenous impurities incorporated during the recrystallization of NaBF, from aqueous media. Although the postexperimental examination of the salt after both runs confirmed isotopic exchange, it should be noted that the salt samples could not be stirred in these experiments and perhaps had not achieved equilibrium. Further tests will be necessary to verify isotopic exchange. Should iso- topic exchange be confirmed, experiments in a circu- lating salt system will be required to establish isotopic exchange in the coolant as a practical method for tritium control. Five experiments were carried out with nickel cou- pons in which NaOH and H;BO; were encapsulated with the coolant salt. The amounts of hydrogen added with the NaOH and H3;BO; were many times greater than the hydrogen believed to be bound in the fluoroborate itself. Two capsules containing 0.4 mole % NaOH were maintained at 600°C, one for 192 hr and the other for 387 hr. In both cases the evolved hydrogen gas, determined chromatographically, was equivalent to the chemically bound hydrogen loaded into the capsules. : 1. S. Cantor and R. M. Waller, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, pp. 79-80. Table 9.1. Details of two experiments involving isotopic exchange in NaBF,-NaF Run No. Hydrogenous impurity H,(g) Conditions and procedure accounting (ueq H/g salt) Analytical examination of salt sample after opening capsule 6, nickel capsule and coupons 16, nickel capsule and coupons 80 hr at 520°C; cool down over 102 hr at 600°C; during most 5.0 £ 0.5, all collected prior to weekend; 10 hr at 600°C; the 22-hr evacuation. during and between heating, gas samples were collected and ‘ analyzed chromatographically. Capsule and silica vessel then evacuated for 22 hr at 600°C; Ta(g), 0.1 Ci, introduced into system, which was maintained at 600°C for 140 hr. 14 £ 1.4, collected prior to of this time the silica vessel introduction of D,(g). was evacuated by means of a Toepler pump from which exiting gas was analyzed for H;(g); then D,(g) at about 22 torrs was introduced and maintained at 600°C for 90 hr. By counting tritium in the salt and by prior mass-spectrographic analysis of H,(g) and HT(g), it was deduced that 0.5 + 0.1 ueq of H per gram of salt had been retained in the salt. Infrared absorption spectrum of 15-mg pellet of salt showed an —OD peak whose absorbance was 31/2 times greater than the —OH peak; however, neither peak provides a quantitative analysis for hydrogen in the salt. 4Assuming a 10% relative standard error in the gas-chromatographic analysis. 68 Two capsules containing 0.5 mole % boric acid were heated at 600°C, one for 435 hr and the other for 644 hr. In these time intervals the evolved hydrogen gas, when analyzed, corresponded to about 60% of the hydrogen introduced into the capsule. However, in-a third capsule, with an initial concentration of 0.66 mole % H3BO; and heated at 600°C for 300 hr, virtually all the chemically bound hydrogen in the capsule escaped as H,(g). The disparity between the first two H3BOj; experiments and the third can probably be attributed to inadvertent pumping away of H, in the first two experiments. Post experimental infrared spectral examination (by John Bates) of the salts in these five latter experiments revealed low OH concentrations, hardly different from that found in fluoroborates, without added NaOH or H;BO;, that had been subjected to similar treatment. These infrared results, togethier with the quantities of H,(g) collected during the experimental runs, suggest that the NaOH and H;BO; did not augment the capacity of the coolant salt to retain hydrogen. Further- more, the higher level of protons increased the corro- sion of nickel. After the experiments the salt and capsule walls were examined and found to contain an -easily observable yellowish skin, identified as NaNiF;. With the fluoroborate salt alone, such corrosion prod- ucts, though possibly formed, have not been detected. 9.2 REACTION OF SODIUM FLUORIDE—-SODIUM TETRAFLUOROBORATE WITH WATER H. W. Kohn Attempts to refine our previous measurements of the equilibrium constants for the reactions NaBF, + H,0 = NaBF, OH + HF (1) and - NaBF;OH - NaBF,0 + HF (2) by improving analytical techniques for the transpiration method have so far met with failure. These have included installation of thermal conductivity apparatus to measure gas composition, sodium fluoride traps for absorption of HF and BF; from the gas stream, and azeotropic distillation from pyridine solutions of sam- ples of the effluent gas collected from the transpiration experiment. During the course of these expe}iments, we have observed some HF present at all times (nickel containers and copper lines and samplers are used). Consequently, our approach has been to determine 90 more accurately the role of HF in stabilizing OH™ and in producing H, O. We have improved on some earlier experimental results by using more refined analytical techniques to measure the extent of reactions (1) and (2). By collecting the off-gas containing BF; and HF in aqueous sodium fluoride followed by CaCl, precipi- tation: BF;(g) + NaF(aq) » NaBF, , (3) NaBF, + 2CaCl, + 3H,0 - H3;BO; +2CaF,{ + NaCl + 3HCl, (4) 2HF + CaCl, — CaF,{ + 2HCI | (5) followed by a double-end-point titration of the result- ant HC1 and H; BO; -mannitol complex, we ascertained that in fluoroborate to which H, O was added slowly as vapor over a period of several hours, there was always a small pressure of HF which would persist for two or three days. The HF pressure measurements so obtained are still erratic since, for example, any error in H3 BO, determination is reflected threefold in the determina- tion of HF, but the implication is that under our experimental conditions (nickel containers, copper lines, 400°C) the equilibrium content of HF in the argon sweep gas is approximately 0.1 meq/liter. Part of the difficulty stems from the apparent high solubility of HF in fluoroborate at 400°C. We believe that the HF is due to reaction (2) above. Using the following numbers, oxide = 400 ppm ob- tained by inventory of the H,0O, and HF = 0.1 meq/liter, we can calculate an equilibrium quotient = [HF] [0*]/[OHT] = 10™* at 400°C (when HF is expressed in atmospheres and the other quantities are expressed in moles per liter). The OH is calculated from the vapor pressure of water and the flow rate of argon through a side stream. We assume all the water reacts, since virtually none is found, under these conditions, by Karl Fischer titration of the effluent gas. Using the temperature variation of the equilibrium determined previously,? this means that (at the same oxide and hydroxide levels) the equilibrium pressure . of HF required to stabilize hydroxide at 650°C would be about 50 mm. We have obtained preliminary values for the solubility of HF in the NaF-NaBF, eutectic mixture with HF at 2. H. W. Kohn, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, p. 81. 600 550 91 ORNL- DWG 71- 2732 TEMPERATURE (°C) A 500 450 400 0. | | | A P B /“ e A 0.2 / SHAFFER - WATSON E A ¢’ e / a} _;:__-" / Ny rY ?:‘ 0.1 2k = A C 0.08 e - & 5 I 4 0.06 o (/3] " MEASURED BY ] 0.04 |—o O SATURATION AND STRIPPING | ® PRESSURE RISE AND TITRATION A PRESSURE RISE CORRECTED 0.02 . 110 1.15 1.20 1,25 1.30 1.35 1.40 1.45 1.50 f000/7 (oK) Fig. 9.1. Henry’s law solubility of HF in NaBF, —9% NaF eutectic. 0.29 atm in a nickel vessel. The HF was supplied in a gas mixture composed of Ar, BF;, H,, and HF. Solubility was determined by sparging the HF from the saturated melt; the amounts of HF removed were measured by alkalimetric titration. The removal of HF was first order as expected, but the curve of concentra- tion in the gas vs sparging time showed a long tail; that is, the last traces of HF were difficult to remove. The value of the Henry’s law constant at 400°C is 0.33 0.03 equiv liter™! atm™!. During the three measure- ments the oxide concentration ranged from 500 to 300 ppm, determined by KBrF,, but the change in oxide content did not seem to affect the solubility markedly in this range. Solubilities at other temperatures were measured in three different ways: (1) by saturation and stripping, (2) by saturation at 400°C followed by raising the temperature and titrating the HF in the cover gas, and (3) by saturation at 400°C, followed by raising the temperature and measuring the corresponding pressure rise and subsequently correcting it for expansion of the cover gas and for the contribution from BF; overpres- sure. In Fig. 9.1 the results of these three methods are presented, along with a correlation developed by Shaffer and Watson® which related HF solubility in NaF-BeF, and in NaF-ZrF; with the free fluoride content of the melt. The Shaffer-Watson formula is given below. K = ex A+BRC_D+ERC P\ RT R where A, B, D, and E are empirical constants, Cis thg fraction of free fluoride in the melt, R is the gas constant, T is the absolute temperature, K is the Henry’s law constant. One may extend this correlation to the NaF-BF; system by using the vapor pressure of BF; over the eutectic from Cantor’s data® to calculate the free 3. J, H, Shaffer and G. M. Watson, Reactor Chem, Div. Annu. Progr. Rep. Jan, 31, 1960, ORNL-2931, p. 32. 4. S. Cantor (ed.), Physical Properties of Molten Salt Reactor Fuel, Coolant, and Flush Salts, ORNL-TM-2316 (August 1968). fluoride in the mixture as a function of temperature and use these numbers in the Shaffer-Watson formula to calculate HF solubility. The results of this calculation are shown on the graph. A more refined apparatus for measuring HF solubility has been constructed and is presently being tested. We have also used infrared spectroscopy to determine the stability constant referred to before. This is reported separately. 9.3 IDENTIFICATION OF CORROSION PRODUCTS IN THE BUBBLER TUBE OF THE FLUOROBORATE TEST LOOP S. Cantor For approximately 1.3 years, an Inconel pump loop (operated by the Reactor Division under the supervision of A. N. Smith) circulated molten NaBF,-NaF (92-8 mole %). Welded into the pump bowl was a % -in.-ID Inconel tube through which helium and BF; entered the circulating salt. After shutdown of the loop, the lower part of this tube was cut off, sectioned, and split axially. One half of the -axially split portion was submitted to the Reactor Chemistry Division for chemical examination of the corrosion and mass trans- fer products deposited on the inner surfaces of the tube. A full description of the appearance of the bubbler tube is given elsewhere in this report.’ Five corrosion and mass transfer products were identified by x-ray diffraction and electron microprobe analyses: 1. agreen salt, NajCrFg; 2. ayellow salt, NaNiF;; 3. a black powdery magnetic material with the stoichi- ometry Ni; Fe; . a porous, gray metallic mass, which plugged the mouth of the tube, composed of nickel metal and NaBF, in approximately equal parts; 5. a millimeter-thick layer, running through most of the length of the tube, composed of nickel and nickel oxide. A plausible inference which may account in part for 92 the morphology of the deposit assumes that Na;CrFy, - NaNiF;, and Ni; Fe were present in a surface scum on the fluoroborate melt prior to migrating up the bubbler 5. A. N. Smith, “MSBR Design and Development,” this report, Part 2. tube; the migration possibly occurred when liquid droplets were formed by bubbles breaking at the mouth of the bubbler. A second transfer mechanism might have been backflow up the tube; however, such events were infrequent and of short duration. The presence of the salts Na;CrFg and NaNiF; has been noted previously as corrosion products from molten NaBF,. Both are only slightly soluble in the melt. They arise initially through oxidation of the metal. In the case of Na;CrF,, the oxidation of chromium in the loop probably arose from three sources: 1. H,O entered as an impurity in the helium and possibly in the fluoroborate salt charge (and also on one occasion deliberately introduced®); the reactions may be written: H, O(g) + 2F “(in the salt) = 2HF(g) + O* (in the salt) , Cr + 3HF(g) + 3NaF(l) > Na;CrF + %, H,(g) . 2. Air trapped in lines when tanks of gas were changed; besides the moisture thereby introduced, 0, (g) would also lead to oxidation: Cr+ %,0,(g) + 3NaF(l) + BF3(g) - Na3 CI'F6 + 1/2 B2 03(1) . 3. Metallic ions already present in the melt oxidize chromium; for instance, the fluoroborate charged into the loop contained about 200 ppm iron; the reaction with chromium may be written: Cr+ % Fe™() > Cr¥*() + %, Fe . The trivalent chromium subsequently precipitated as Na; CrFg. ' In accounting for the formation of NaNiF;, reactions similar to those given under 1 and 2 above seem likely: Ni + 2HF(g) + NaF(l) > NaNiF, + H, , Ni+ %4, 0,(g) + NaF(1) + % BF;(g) > NaNiF; + 4B, 05(1) . The reasons for the formation of Ni; Fe are not easily ascertained. One can speculate that ionic iron and nickel in the salt oxidize metallic chromium and 6. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 41-44. coprecipitate as an alloy: %,Cr + 3Ni?*(1) + Fe?*(l) > % Cr**(1) +[Nis Fe . Most of the Cr** would subsequently precipitate as Na;CrF¢. The dissolved ionic nickel .and iron could have originated by means of the oxidation by H, O(g) and O,(g). . The origin of the “extra” nickel in the tube (noted above under products 4 and 5) is uncertain. The nickel plug formed in a position where the salt was exposed to lower temperatures; perhaps the reaction NiF, (dissolved in melt) + H,(g) = 2HF(g) + Ni permitted the deposition of nickel at the mouth of the tube. , In summary, the corrosion products identified in the bubbler tube originated with the introduction of oxidants, the most important being water vapor. The water vapor led to the formation of HF, which corroded the Inconel. This corrosion, which implies the dissolution of chromium, iron, and nickel ions, eventu- ally led to the precipitation of Na;CrF¢ and NaNiF;, some of which were carried on the surface of the molten fluoroborate. Dissolved iron and nickel also reacted further with metallic chromium and were apparently precipitated out as the alloy Niz Fe. Some of this alloy was also present on the surface of the molten salt. 9.4 MASS SPECTROSCOPY OF FLUOROBORATE MSR COOLANTS C.F.Weaver J.D.Redman With respect to tritium control in a molten-salt reactor, F. F. Blankenship has recently suggested that hydrogenous material might be added to and main- tained in the proposed coolant salt, NaBF,-NaF (92-8 mole %), by incorporating 1 or 2% NaOH into the liquid if the behavior of this system as such or its corrosive attack on structural materials is acceptable. Consequently, we have initiated studies of the vapors over such materials using a time-of-flight mass spec- trometer to detect the species effusing from a Knudsen cell. 93 behaved in a more complex way. In addition to the previously reported monomer and dimer, both the trimer and tetramer of NaOH were observed. Water was ~ also evolved from material which had been dried and The first experiments were made with the separate - components BF;, NaBF,, and NaOH in nickel cells to determine cracking patterns and to confirm that the boron compounds behaved as commonly described. The BF, yielded a cracking pattern essentially like that in the literature, and no polymers of BFj3 were observed. The NaBF, decomposed, yielding BF; mon- omer as the only vapor species. However, the NaOH outgassed at 200°C. This was attributed to the decom- -position 2NaOH — Na,O + H,01. In addition a small amount of H, was evolved, probably by Ni + 2NaOH —~ Na, O + NiO + H,, although the condensed products of this reaction are not well established. By far the most intense of the reduced species was sodium vapor, attributed to the reaction Na, O + Ni - NiOJ + 2Na?t. The Na vapor was detected by mass spectrometry and the NiO precipitate by x-ray diffraction. Severe attack of the nickel cell was also evident by direct observation. The standard free energy change for the reaction as written at 1000°K is +29 kcal, which implies that P, = 7 X 107? torr, well within the range of detection of the mass spectrometer. The possibility of adding hydrogenous material to the fluoroborate coolant as water either intentionally or accidentally as a steam leak suggested that the reaction of moisture with NaBF, should also be investigated. Two approaches were used to obtain reaction prod- ucts of the NaBF,-H,O system. One was to add H,0 (0.1 g) to NaBF, (0.2 g) and to reflux at 300°C under an atmosphere of helium for an hour. After refluxing, the cell was cooled, evacuated, and studied in the normal manner. Vapor over the temperature range of 25 to 750°C was quite similar to that over pure NaBF,, even to the amount of water vapor evolved. No reaction products had been formed, at least to the 300° refluxing temperature limit; moreover, the NaBF, displayed a very impressive, nondeliquescent character. In the second approach to obtain reaction products from the NaBF,-H, O system, steam was admitted over the temperature range 25 to 750° at an approximate leak rate of 1077 torr liter sec™!. The complexity of the reaction was apparent. For convenience, an attempt was made to assign fragments to possible molecular precursors, almost arbitrarily (Table 9.2). The only Table 9.2. Partial pressure of ‘assumed molecules in vapor at 300° from the NaBF ,-H, O reaction Molecules Pressure (torrs) x 1072 H,0 1.0 BF4 2.5 HF 1.0 NaBOF, 0.008 BF3 'Hzo 0.2 NaBF, 0.003 ? {mass 19) 0.01 criterion was to group fragments displaying the same coefficient of temperature dependence with the same precursor. Some fragments certainly had more than one precursor. Nevertheless, at 300°, reaction products of NaBF, and steam accounted for at least 30% of the total vapor pressure, most of which was HF.. 9.5 SPECTROSCOPIC INVESTIGATIONS OF HYDROGEN- AND DEUTERIUM-CONTAINING IMPURITIES IN NaBF, AND NaF-NaBF, EUTECTICS John B. Bates Harold W.Kohn Jack P. Young Marvin M. Murray George E. Boyd Evidence for the existence of hydrogen-containing impurities in NaBF,; was first obtained from near- infrared spectra of the molten salt and in mid-infrared spectra of pressed pellets of the “pure” crystalline material. Bands observed in the region between 1.35 and 2.55 p (Fig. 9.2) are believed to correspond to overtones and combinations of fundamental vibrations of an OH species (either free OH™ or BF;OH ") in the melt, Although the vibrational spectrum of crystalline NaBF, has been thoroughly investigated,” ® no infrared measurements have been reported in the region above 3000 ¢cm ™' with thick samples for transmission meas- urements. We have recently repeated infrared transmission meas- urements on a single crystal of NaBF, supplied by L. O. Gilpatrick. This crystal was polished to a thickness of about Y% to Y% mm before examination with a Perkin-Elmer model 621 spectrophotometer. A single sharp band at 3641 cm ™! was observed which exhibited about a 2:1 dichroic ratio as the infrared polarizer was rotated through an angle of 90°. This band was assigned tentatively to an OH-containing molecular species (most likely BF3;OHT) which is effectively isolated in an NaBF, matrix. Subsequent spectroscopic studies'® on relatively pure samples of crystalline NaBF3;OH appear to confirm this assignment. Spectroscopic observation of isotopic exchange of hydrogen with deuterium was investigated in solid-state and in molten-salt exchange reactions. In the first experiments, D, O vapor was bubbled slowly into a 7. H. A. Bonadeo and E. Silberman, Spectrochim, Acta 26A, 2337 (1970). 8. J. B. Bates, A. S. Quist, and G. E. Boyd, J. Chem. Phys. 54,128 (1971). 9. J. B. Bates, ibid., 54 (1971) (in press). . 10. J. B. Bates and A. S. Quist, work in progress (1971). 94 ORNL-DWG. 71-3460 ! & 5 W—’\:— > = | 2690 FREQUENCY (cm™h Fig. 9.4. Band due to absorption by OD-containing species in the infrared spectrum of molten NaBF,4 at 425°C. The above experimental results demonstrated that samples of presumed high-purity NaBF, contained OH impurities and that exchange with deuterium atoms occurs in the molten state. We thus desired to obtain a direct measurement of the OH absorption in molten NaBF, as a function of time. It was observed that the bands due to overtones of OH disappear from the near-infrared spectrum of the melt in a matter of hours. However, pellets pressed from the frozen salt always showed the sharp absorption at 3641 cm™". These two observations can be reconciled if it is assumed that a small, but unknown, concentration of OH or BF;OH™ is in equilibrium with NaBF,. Indeed, the rate of decrease of the near-infrared OH peak does not appear to follow simple first-order kinetics. Therefore, com- peting, consecutive, or higher-order processes must be involved. It is interesting also to note that the same near-infrared spectrum is observed if either NaOH or H, BO; is added to molten NaBF,, . Near-infrared spectra of molten NaBF,; were meas- ured with SiO, cells and a high-temperature furnace described previously.!! For measurements in the mid- 11. J. P. Young, Inorg. Chem. 6, 1486 (1967), infrared region, samples of NaBF, were spiked with D;BO; because it was possible to study only the absorption by OD rather than that by OH with the available equipment. The infrared spectrum of a molten NaBF,-D;BO; solution in the 2700-cm™' region at about 425°C is shown in Fig. 9.4. The broad OD band is centered at about 2690.cm !, in excellent agreement with the frequency of the sharp band observed at 2688 cm ™! in the pellet spectra (Fig. 9.3). A slight decrease in the absorption of the 2690-cm ™ band with time was observed. However, the band was still observable after 48 hr in a 2-mm-path-length SiO, cell. Under our experimental conditions, the OD ab- sorption in the melt was stable in the presence of added INOR-8 but unstable in the presence of pure chromium metal. It is not known how much OH or OD was actually observed in either the melt or pellet spectra. Since it was demonstrated that OD species can be seen spectrally in the melt, an apparatus which does not use Si0, as a window material is being designed and fabricated for use in further studies on molten NaBF,. The results of these spectroscopic studies indicate that an OH-containing impurity occurs in “pure” samples of NaBF, , that exchange with deuterium atoms takes place in the molten state, and that the OH species persists at melt temperatures of 425°C over a period of days. Additional experiments are under way to deter- mine quantitatively the amount of OH-containing impurity in NaBF, and the extent of isotopic exchange of H for D or T in the molten state at temperatures of about 600°C. ' 1 9.6 RAMAN SPECTRA OF THE HIGH-TEMPERATURE PHASE OF POLYCRYSTALLINE NaBF, Arvin S. Quist John B. Bates George E. Boyd Raman spectra were measured with polycrystalline NaBF, at temperatures just above and below the dimorphic crystal transition (245°C). The results of this study rationalize anomalies in x-ray diffraction data for NaBF, and related structures. The spectrum of the high-temperature solid phase indicates the presence of a highly symmetric effective field about the BF,” ion similar to that previously observed in melt spectra. A complete disordering of the fluoride ion positions must be assumed for the spectral results to be consistent with recent x-ray data on crystalline NaBF,; powders at 265°C, which have been indexed in the hexagonal system. 96 The room-temperature crystalline form of NaBF, is orthorhombic, space group Cmcm, and is isostructural only with the room-temperature forms of CaSO, and NaClO,4. All three compounds undergo phase transitions at elevated temperatures. Sodium perchlorate has a cubic structure above 308°C which is said to be " isostructural with the high-temperature phases of many perchlorates, tetrafluoroborates, and sulfates. The high- temperature phase of CaSO, (above 1210°C) was reported to be hexagonal. At 245 + 1°C, NaBF, undergoes a change in crystal structure to a form first reported as monoclinic. Dworkin and Bredig,!? how- ever, have expressed the view that a lowering of the symmetry of the high-temperature form without a lowering of the number of molecules per unit cell is quite unlikely. They. have indexed the powder pattern reported by Pistorius, Boeyens, and Clark'?® on the basis that a mixture of the orthorhombic low- temperature phase with a high-temperature hexagonal phase was measured. Recently, Bredig'? obtained x-ray powder patterns of NaBF, at 265°C with a Buerger precession camera. The powder photographs were indexed in the hexagonal system with parameters ¢, = 500 £ 002 and ¢ = 7.75 * 0.03 A. Acceptable agreement between observed and calculated diffraction intensities, however, could not be obtained for ordered hexagonal space groups. In view of the uncertainties in the structure of the high-temperature phase of crystalline NaBF,, we have measured the Raman spectrum of this material in an effort to provide additional information regarding the lattice symmetry. Static and dynamic field effects on the vibrational modes of BF,” in the orthorhombic (room temperature) NaBF, structure have been estab- lished by previous studies.!5:'® The two BF, ions in the primitive unit cell of this crystal occupy C,, sites, so that all the components of the doubly degenerate v,(e) mode and of the triply degenerate v;(f,) and v4(f,) modes are observed in the Raman spectrum. An increase in the site symmetry of the BF,” ion in the high-temperature form may cause a collapse of the 12, A, S. Dworkin and M. A. Bredig, J. Chem. Eng. Data 15, 505 (1970). 13. C. W, F. T. Pistorius, J. C. A. Boeyens, and J, B. Clark, High Temp.—High Pressures 1,41 (1969). 14, M. A. Bredig, ORNL MSRP Monthly Progr. Rep. (December 1970—-January 1971), MSR-71-13, p. 21 (internal memorandum). 15. J. B. Bates, A. S. Quist, and G. E. Boyd, J. Chem. Phys. 54,124 (1971). 16. J. B. Bates, ibid., 54 (1971) (in press). splitting, and hence only one frequency for each of the vibrational modes would be observed. The presence or absence of such effects in the spectra of the high- temperature phase would establish the symmetry of the effective field about the BF,;” ion in the high- temperature lattice and perhaps also resolve the existing conflict in the x-ray data. In the current Raman studies of the high-temperature phase of NaBF, only the »,(a,), v;(e), and v,(f;) modes were examined in detail;! 7 the intensities of the v3(f,) modes are quite weak under these conditions. A detailed comparison of the v, and v, bands at 240 and 254°C is given in Fig. 9.5, which clearly shows the abrupt change in the Raman spectrum of crystalline NaBF, as it undergoes the phase transition. The experimental observations indicate that the effec- tive environment about the BF,” ion in the high- temperature crystalline phase of NaBF, has tetrahedral (or higher) symmetry and that there is no measurable (or allowed) dynamic coupling between neighboring BF,” ions. This observation is consistent with the recent. x-ray results,!* which led to a hexagonal structure only if a complete disordering of the fluoride positions is - assumed. With this complete disordering, the effective field experienced by a BF,; ion would be highly symmetric, and the lack of splitting of the », and v, bands in the Raman spectra could be explained. This structure also would be consistent with recent infrared spectra obtained to 270°C in which the »; mode disappeared in the high-temperature phase and the three components of v, observed at room temperature were reported to merge into a single component above the transition point.!® Furthermore, the assumption of a disordered lattice structure explains the similarity be- tween the Raman spectrum of the high-temperature crystalline phase and that of ihe melt.!® In the high-temperature crystalline phase the BF,” ions are undergoing large-amplitude random reorientations (librations), which gives rise to the disorder, and a given BF, ion “sees” an average symmetric field similar to that experienced in the melt. Because the neighboring BF, .ions are closer together in the high-temperature solid than they are in the melt,?® the F—F nonbonded repulsion is larger in the solid and a higher »; frequency is observed. 17. A more extensive discussion of the results of the present study is contained in a paper by the authors which is to be published in J. Chem. Phys. 18. H. A. Bonadeo and E. Silberman, Spectrochim. Acta 26A, 2337 (1970). 19. A. S. Quist, J. B. Bates, and ‘G. E. Boyd, J. Chem. Phys. 54,4896 (1971). ‘ 97 ORNL- DWG. 71-962 240°C 4.0 cm! slit 5x10%¢c/s > 1 l ] | 1 | 1 l I | I I 5 550 530 510 370 350 330 = L 2| me, 4.0 cm ' slit 5x102¢/s 1 ] ) | ) | ! | | I | 550 530 510 370 350 330 FREQUENCY (cm™!) Fig. 9.5. Raman spectra of the v, and v, regions of polycrystalline NaBF, at 240 and 254°C. A phase transition occurs at 245°C. A disordered hexagonal structure also can explain the observation of optical isotropy of NaBF, in the high-temperature phase.”" The fluorine atoms make the largest contribution to the polarizability of the crystal, and their orientation will determine the symmetry of the crystal indicatrix. Complete disordering of the F atoms would give a nearly spherical indicatrix, and the crystal would appear to be optically isotropic. 20. The density of NaBF, is higher in the high-temperature solid than in the melt, and the anion-anion separation is generally known to increase in going from the solid to the molten state (J. Braunstein, “Statistical Thermodynamics of Molten Salts and Concentrated Aqueous Electrolytes,” in Jonic Interactions. Dilute Solutions to Molten Salts, S. Petrucci, ed., Academic, in press). 21. George Brunton, ORNL, Reactor Chemistry Division, private communication (1970). 9.7 ANEW METHOD FOR SYNTHESIS OF NaBF, OH L. O. Gilpatrick C. J. Barton Laboratory studies are continuing in attempts to synthesize hydrogen-containing species for the reten- tion of tritium in the MSR coolant salt. One of these species is the first hydrolysis product of NaBF, in aqueous solution, which has been reported to be NaBF;O0H.?2>23 Ryss and Slutskaya® have reported a successful synthesis of this compound in which boric acid is reacted at 0°C with a saturated solution of NaHF, as follows: INaHF, + H, BO, % NaBF,OH + NaF . 2 This synthesis has the disadvantage that it produces a mixture of NaBF;OH and NaF and in addition also becomes contaminated with some NaBF,; produced by a side reaction when attempts are made to recrystallize it from water solutions. A new synthesis has been devised for NaBF;OH which employs the following reactions in a saturated solution of NaHCO;: BF, + H,0 "~ H(BF,OH) , H, O H(BF, OH) + NaHCO, 3—2%» NaBF;OH + CO, + H,0 . One mole of solid NaHCO; and BF; gas are added progressively to 5 moles of water at 0°C, which produces a clear solution. Four volumes of cold 95% ethanol are then added. Storage at 0°C gives a good yield of product which has a better purity than that produced by the method of Ryss. The identification of this material is relatively certain and direct based on its properties. Sodium hydroxytri- fluoroborate is very water soluble and shows rapid hydrolysis to NaF, H;BO;, and NaBF, when stored as an aqueous solution. It is birefringent to polarized light and exhibits refractive indexes from a maximum of 1.350 to a minimum of about 1.343 at 25°C. The crystals belong to the hexagonal system and are polar, 22. A. Travers and L. Malprade, Bull. Soc. Chim. 47, 788 (1930). 23. C. A. Wamser, J. Amer. Chem, Soc. 70,1209 (1948). 24. L. G. Ryss and M., M. Slutskaya, J. Gen, Chem, USSR 22, 45 (1952). 98 as has been shown by a recent x-ray structure study?®® which makes positive identification convenient. ’ 9.8 SOLUBILITY OF BF, GAS IN FLUORIDE MELTS S.Cantor W.T.Ward The purposes of these measurements are to relate BF, solubility to changes in the thermodynamic properties of molten fluorides and to provide data relevant to the use of BF; as a burnable neutron poison for purposes ~of reactor control. The single-vessel apparatus and the procedures of measurement have been described previ- ously.?® Solubilities of BF; have been determined in five molten-salt solvents composed of LiF and BeF,. The chief results obtained thus far are: 1. In each solvent, the higher the temperature, the lower the BF; solubility; that is, the enthalpy of solution is exothermic. . At constant temperature and pressure, the higher the concentration of LiF, the higher the solubility of BF;; indeed, the Henry’s law constant appears to be linear with the thermodynamic activity of LiF. 3. Henry’s law is obeyed in almost all cases; however, where -the concentration of BF; in the melt approaches 1 mole % or greater, there are discernible positive deviations from Henry’s law, ' The data and derived thermodynamic information are summarized in Table 9.3. The magnitude and negative sign of the enthalpy of solution suggest a strong interaction of BF;(g) with the melt. There does not appear to be a correlation of enthalpy of solution (of BF3;) with melt composition. The approximate con- stancy of AH in Table 9.3 signifies that plots of log K (Henry’s law constant) vs reciprocal temperature (in °K) are approXimately parallel. As may be noted in column 3 in the Table, the higher the mole fraction of LiF, the higher the Henry’s law constant. In Fig. 9.6, the Henry’s law constant at 600°C is plotted vs activity of LiF reported®”’ for these melts; the excellent linearity of these two properties also holds at other temperatures. An interpretation of this linear 25. M. T. R. Clark and H. Lyton, Can. J. Chem, 48, 405 (1970). 26. S. Cantor and W. T. Ward, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, pp. 78-179. ' 27. B. F, Hitch and C. F. Baes, Jr., Inorg. Chem. 8, 201 (1969). 99 Table 9.3. Solubility? of BF3 in molten LiF-BeF, solvents; enthalpy and entropy of solution Solvent Temperature Henry’s law b b composition range measured K (mole fraction AH AS_, -1 (mole % LiF) ©o) BF 3/atm) at 600°C (keal/mole) (cal mole = deg ") 6364 470-653 0.00207 T —15.1 -29.6 66 520-725 0.00242 -15.6 -29.9 70 548-732 0.00354 -15.5 : -29.0 7SC 641857 0.00500 —15.9 -28.7 80 714866 0.00613 -15.6 -28.0 2Pressure range, 1.3—3.0 atm, AS AH bCalculated from the equation In K = — — R RT’ ¢Measurements in this solvent composition not completed. dIn the pressure range 1,3-3.0 atm, at temperatures below 520 °C. there are positive deviations from Henry’s law. ORNL-DWG 71-7250 0.006 {mole % LiF) 80’ / / 0.005 ~J [ ] 0.004 70,0 HENRY'S LAW CONSTANT {mole fraction BF3/aim) 0003 0.002 63 .V 04 0.2 03 04 05 ACTIVITY OF LiF 06 or Fig. 9.6. Henry’s law solubility of BF3 vs activity of LiF in molten LiF-BeF, at 600°C. behavior is that the ionic reaction equilibrium F~(1) + BF;(g) = BF, (1) occurs, from which it follows that aBF‘;_ - KfPBF3 ag -, (1) mole fraction BF3 dissolved /atm ORNL-DWG T{-7254 00059 | ‘ ? 00057 |— ' / * MEASURED { o EXTRAPOLATED FROM » HIGHER TEMPERATURE 0.0055 , S 00053 / — / | *_IFHENRY'S LAW HELD __ 00051 \ = | EFFECT ON SOLUBILITY ’OF AN ERF\‘OR OF {°C 0.0049 f 0 1 2 3 PRESSURE OF BFz (atm} Fig. 9.7. Solubility of BF; in 63-37 mole % LiF-BeF, at 517.6 + 0.5°C. where ag . - and ag - are, respectively, the activities of fluoroborate and fluoride jons; K; is the equilibrium constant of the reaction (like any equilibrium constant, K; varies with temperature); and Py is the equilib- rium pressure of BF;. At constant temperature and pressure, Eq. (1) explicitly states that app,- varies linearly with ap-. In this interpretation of the phe- nomena depicted in Fig. 9.6, the concentration of dissolved BF; is identified with the activity of BF,4 in the melt, while the activity of LiF is, of course, identified with the activity of fluoride ion in the melt. In the pressure range under study (1.3—3 atm) the solubility of BF; is sufficiently low that the dissolved molecules do not interact with one another. This - noninteraction between solute molecules is the basis of Henry’s law. In this investigation virtually all the solubilities of BF; obey Henry’s law. However, Henry’s law does not seem to hold at .relatively low tempera- tures in the melt of composition LiF-BeF, (63-37 mole 100 %). At temperatures below 520°C the solubilities of BF; become 1 mole % or greater. Figure 9.7 shows data which do not obey Henry’s law; although the data exhibit some scatter, the solubility of BF; per unit pressure almost certainly increases with pressure; that is, the deviations from Henry’s law are positive. Further measurement will be made to confirm these deviations. 9.9 EQUILIBRIUM PHASE RELATIONSHIPS IN THE SYSTEM RbF-RbBF, L. O. Gilpatrick C. J. Barton An investigation of the equilibrium phase diagram for the RbF-RbBF, system was completed. Final revisions to the system diagram, as reported previously,*® resulted from measurements with purer rubidium fluoride than was previously employed. New RbF has been obtained and purified by treat- ment with dry H, and HF at temperatures up to and above the melting point to remove water and hy- droxide. Rubidium fluoride was produced which has a melting point of 793°C, which is in good agreement with the literature value of 795°C. Binary mixtures of RbF and RbBF, were prepared at 5 mole % intervals ranging up to 20 mole % RbBF,. 28. L. O. Gilpatrick and C. J. Barton, MSR Program Semiannu. Progr, Rep, Feb, 28, 1970, ORNL-4548, p. 133. Differential thermal analysis (DTA) was used to examine these compositions, which gave the expected higher temperatures in each case, lending support to the belief that the irregularities in the original part of the study were due to RbF of insufficient quality. The phase diagram in its final form is shown in Fig. 9.8. 9.10 ACTIVITIES IN ALKALI FLUORIDE — FLUOROBORATE MIXTURES D. M. Moulton J. Braunstein Thermodynamics of mixing in alkali fluoride — alkali fluoroborate mixtures are of interest in connection with proposed coolants for the MSBR. Deviations from ideality, although small, are less well understood in such - common cation binary mixtures (and have been less thoroughly studied) than in common anion binary mixtures. Endothermic mixing is frequently observed rather than the exothermic effect expected from electrostatic considerations, and may be associated with changes in packing of large anions of differing size as well as with the dispersion energy of the mixed anions.?® Here we report some activity coefficient 29. J. L. Holm and O. J. Kleppa, J. Chem. Phys. 49, 2425 (1968); O. J. Kleppa, Annu. Rev. Phys. Chem. 16, 187 (1965); J. Brdunstein in lonic Interactions, S. Petrucci, ed., p. 179, Academic, New York, 1971. ‘ ORNL-DWG 70-30234A 800. 1 \l\ | l I R 4 LIQUIDUS \n\ o SOLIDUS 700 ~] e POLYMORPHIC TRANSITION - \ IN RbBF, 600 \\ LIQUID " b LIQUID + RbF /n/. T 500 Q = LIQUID + & RbBF, g ———o— d42°C ‘;\‘5/.* o] & 400 14 a = a RbBF, + RbF E 300 | - 247°C . 200 B RbBF, + RbF 100 0 RbF 10 20 30 40 50 60 70 80 90 RbBF, RbBF,; (mole %} Fig. 9.8. The system RbBF4-RbF. calculations from available data in fluoride—fluoro- borate mixtures. We have used the phase diagrams of the alkali fluoride—fluoroborate systems (Na, K, Rb) to calculate the component activities along the liquidus lines. The method was the usual one of integrating the Schréder— van Laar equation, to give the ideal liquidus and then comparing actual and ideal mole fractions. A power series was used for the heat capacity difference. Each activity refers to the hypothetical pure liquid at the same temperature. The sources of the data are Barton et al.,>° Kubaschewski et al.,>! and Dworkin and Bredig.?? In all three cases the behavior of the activity coefficients is qualitatively the same (see Table 9.4). The fluoride shows positive deviation at first but then becomes negative before it reaches the eutectic. The fluoroborates always show negative deviation (with sodium it is only very slight). In both cases the sodium salt shows the smallest and rubidium the largest tendency toward negative deviation. A comparison with the activity coefficients of Cantor,>® which were derived from decomposition pressure measurements, shows that for NaF we agree fairly well up to about 50 mole % but then deviate increasingly. For KF the agreement is worse, and for Table 9.4. Activity coefficients in alkali fluoride— fluoroborate mixtures at 1000°K for Na and 1100°K for K and Rb Xpp, 02 04 06 07 08 09 095 YNg 110 120 123 119 112 0.89 J[ 1.00(-) yna® 115 126 131 132 133 133 H 1.00(+) Tk 1.05 1.08 1.02 0.97 H 0.98 099 1.00(-) y® 114 124 132 13514101 1.00 1.00 Yrp 102 100 093W 094 097 099 099 dCantor, [[ Eutectic: fluoride to left, fluoroborate to right. 101 the fluoroborates the direction of deviation is the opposite. The activity coefficients are shown; they have been put on the same standard temperature T, as Cantor’s by saying that Yr, = 'yT(T/TO) (i.e., ideal entropy of mixing), which does not change them much and in particular does not affect the shape of the curve or the direction of deviation. The differences in the calculated activity coefficients may indicate a significant temperature dependence of the activity coefficients since the eutectic temperatures lie some 500 to 700° below the temperatures of the fluoroborate decomposition pressure measurements. Deviations from ideality calculated from the NaF- NaBF, and KF-KBF, phase diagrams®® are compared with those calculated from the NaF-Nal and KF-KI phase diagrams®** and are shown as the interaction parameters for excess free energy at the eutectic in Table 9.5. In both the BF,-F~ and 1°-F~ systems, deviations from ideality become less negative with decreasing size of the common cation, as would be expected if changes in anion packing contribute positive deviations. 9.11 HIGH-TEMPERATURE CRYSTAL STRUCTURE AND VOLUME OF SODIUM TETRAFLUOROBORATE AND RELATED COMPOUNDS M. A, Bredig - High-temperature x-ray diffraction patterns of NaBF, and CaSQ, (isostructural) reported by several authors 30. C. J. Barton et al.,, MSR Program Semiannu. Progr. Rep. Dec, 31, 1967, ORNL-4191, p. 158. 31. O. Kubaschewski, E. L. Evans, and C. B. Alcock, Metallurgical Thermochemistry, Pergamon, Oxford, 1967. 32. A. S. Dworkin and M. A, Bredig, J. Phys. Chem. 64, 269 (1960); A. S. Dworkin and M. A. Bredig, J. Chem. Eng. Data 15, 505 (1970). 33. 8. Cantor, MSR Program Semiannu. Progr. Rept. Aug. 31, 1968, ORNL-4254, p. 170; MSR Program Semiannu. Progr. Rep. Feb. 28, 1869, ORNL-4344, p. 159. 34, E. M. Levin, C. R. Robbins, and H. F. McMurdie, Phase Diagrams for Ceramists, 1969 Supplement, The American ~ Ceramic Society, Columbus, Ohjo, 1969. Table 9.5. Interaction parameters for excess free energy at the eutectic NaF-NaBF, —0.53 Binary system GE/x x5 (kcal/mole) KF-KBF, -0.73 NaF-Nal +0.62 KF-KI —-0.70 Table 9.6. Powder x-ray diffraction pattern of high-temperature form of NaBF, at 265°C hkl Intensity d, obs. (A) d, calc. (A)? 100 ww 434 433 002 3.90 to 3.88 101 v 3.77 3.78 102 m 2.89 2.89 110 w 2.50 2.50 200 2.165 112 m 2.097 2.100 202 w 1.884 1.890 104 w 1.770 1.770 203 1.659 210 w 1.645 1637 114 1.531 212 vs 1525 1.508 302 1.353 9Hexagonal: 2 = 5,00 + 0.02 &;¢c = 7.75 £ 0.03 A. were critically reviewed and reinterpreted in earlier ~ reports.>®> A hexagonal crystal structure was found to take the place of a monoclinic one suggested by Pistorius et al.>® for NaBF, and of a cubic one given by Floerke®” for CaSO,. The hexagonal structure seems to be a novel one among compounds of the stoichiometry M(ZX,). Somewhat like (NH,4)F, it appears to be related to the hexagonal structure of ZnS, wurtzite, or, much less likely, that of nickel arsenide (NiAs), in a manner similar to the relationship between the cubic high-temperature form of the other alkali fluoroborates and the structure of rock salt.>® After constructing a simple heating device, auxiliary to the Buerger x-ray diffraction precession camera, the hexagonal, bimolecular unit cell with 4 = 5.00 £0.02 and ¢ = 7.75 + 0.03 A has now been directly confirmed on powder samples of NaBF, at 265°C, that is, about 20° above the transition point (Table 9.6). From this, one calculates the molar volume as 50.6 * 0.6 cm?®/ mole and the specific gravity as 2.13 + 0.02 g/cm?. When, on considering atomic positions, by analogy with (NH4)F the space group Cg, (C6,mc) was used, 35. M. A. Bredig, Chem. Div. Annu. Progr. Rep. May 20, 1968, ORNL-4306, pp. 129—-30; Chem. Div. Annu. Progr. Rep. May 20, 1970, ORNL-4581, pp. 116—18. ' 36. C. W. F. T: Pistorius, J, C, A, Boeyens, and J. B. Clark, High Temp.—High Pressures 1,41-52 (1969). 37. O. W. Floerke, Naturwissenschaften 39,478 (1952). 38. Ralph W. G. Wyckoff, Crystal Structures, 2d ed., vol, III, Interscience, New York, 1963. 102 distances in the unit cell were too short to accommo- date an F (or an O) as a bridge between B and Na (or S and Ca), corresponding to the hydrogen bridging in (NH,)F. For ordered orientation of the BF, (or S0,%") ions within the cation tetrahedra, no agreement between diffraction intensities observed and calculated (with the help of G. D. Brunton) was obtained thus far. Attempts to reconcile the discrepancies by a highly disordered model, or an even more modified structure, are being continued. A high degree of (rotational) disorder of the tetrafluoroborate ions would be in agreement not only with the relatively large entropy of transition,>® but might also reconcile the optical isotropy, observed by G. D. Brunton, and apparent degeneracies in the Raman spectrum, measured by Quist, Bates, and Boyd (Sect. 9.6), with the hexagonal symmetry based on the powder x-ray pattern. The volume change, AV/V,, from room temperature (Vo) to slightly above the transition point of NaBF, is 15.5%, which is slightly smaller than a previous, rough estimate of 18%.*° If for comparison one takes the more accurate high-temperature data for NaClO, of Braekken and Harang®' instead of those given by Wyckoff>® based on the results of Herrmann and Iige,*? AV/V, for NaBF, is not similar to but is almost twice as large as that for NaClO, . Figure 9.9 shows the corresponding volume changes for alkali tetrafluoroborates (partly based on Brunton’s work®? and on my own recent high-temperature measurements), for alkali perchlorates,®®>*! and the sulfates of Ca, Sr, and Ba.’® The value of AV/V, for the tetrafluoroborates is consistently larger (~30%) than that for the corresponding perchlorates and alkaline-earth sulfates. The earlier speculative estimate of similarity in AV/V, (ref. 40) was based on a value for the volume of the high-temperature form of KBF, given by Finbak and Hassel,** which I found to be too small by 6%. , Figure 9.9 demonstrates the large departure of the two sodium salts from the trends found with the other 39. A. S. Dworkin and M. A. Bredig, J. Chem. Eng. Data 15, 505-7 (1970). ~40. Stanley Cantor, Dana P. McDermott, and L. O. Gil- patrick, J. Chem. Phys. 52,4600—-4604 (1970). 41. H. Braekken and L. Harang, Z, Kristallogr. 75, 53849 (1930). 42, K. Herrmann and W, Ilge, ibid., 41-66 (1930). 43. George Brunton, Acta Crystallogr. B24, 1703-4 (1968); ibid., B25, 2161-62 (1969). 44, Chr. Finbak and O. Hassel, Z, Phys. Chem. B32,433-38 (1936). ORNL- DWG 71-7252 AV/Vyy, RELATIVE VOLUME INCREASE (%) 25 \ f N ’ KBF>\ 20 / / [ srs0, 8™\ RbBF, f — AN d ,/ KCIO:“.‘quo4 15 |NaBF, — . / RbCIO, // \\CSBF4 10 A % J*® by means of the Clausius-Clapeyron relationship. These calculations indicate that actually the relative volume change on transition, AV, /V, for NaBF, is quite similar to that for KBF, and RbBF,, namely, 10%. Thus the apparent deviation for NaBF, in Fig. 9.9 must be caused by the thermal expansion coefficient in the low-temperature orthorhombic phase being considerably smaller than in those of KBF, and RbBF, of a different structure. This is reminiscent of the difference in the thermal expansion of NaNO; and KNO,. Additional x-ray diffraction measurements are planned to confirm these observations. 45, C, W, F. T. Pistorius and J. B, Clark, High Temp.—High Pressures 1,561-70 (1969). 46. Carl W. F. T. Pistorius, J. Phys. Chem, Solids 31, 38589 (1970). 103 9.12 HYDROGEN PERMEATION THROUGH OXIDE-COATED METALS R. A. Strehlow The control of tritium flow in a molten-salt reactor system involves both knowledge of permeation rates through steam generator heat exchanger material and the extent to which-permeation is decreased due to the oxide coating which is expected to exist in contact with the steam. It has been long recognized that oxide coatings on metals impede hydrogen permeation.®”-*® Occasionally large impedance to flow of hydrogen has been observed and attributed or related to an oxide film,*®+*? The impedance to hydrogen flow is often, however, observed as an artifact of an experimental determination of hydrogen permeability. It is reflected as an aging or conditioning time often of subsidiary interest. - A An experimental program was begun to study hydro- gen permeation under conditions closely analogous to those which might be expected in a molten-salt reactor steam generator heat exchanger. The apparatus is shown schematically in Fig. 9.10. Since several modes of operation are possible, note need be made only that the diffusion membrane (a piece of tubing of normal wall thickness) separates two regions, A and B. Region A is analogous to the steam side of a heat exchange tubing section in that the gas which can be admitted and analyzed may be admitted wet or dry. Region B is analogous to the inside of a heat exchange tubing section only to the extent that a gas may be flowed through the tube and analyzed. (A suitable mode of operation would be with this side evacuated.) Salt has not been used in permeation studies conducted so far. A quadrupole mass analyzer was used variously to assess leak-tightness of the apparatus, gas purity, and approximate analysis, as well as to follow the changes in gas composition in both regions A and B for different experiments. The mass analyzer was equipped with a source which permitted fairly high pressures to be used (in the range of 0.1 to 0.3 torr). Under these conditions an enhanced sensitivity for several species is observed due to the occurrence of ion-molecule reactions. Two stainless steel samples were selected for two initial experiments in order to correlate this work with earlier work on hydrogen permeation. Estimates of flow 47. C. J. Smithells and L. E. Ransley, Proc. Roy. Soc. A150, 172-97 (1935). 48. P.S. Flint, KAPL-659 (Dec. 14, 1951). 49, Tennyson Smith, J. Nucl. Marer. 18,323-35 (1966). ORNL-DWG 71-7253 TO PUMP, ARGON-HYDROGEN —= /DIFFUSION SAMPLE MASS-ANALYZER M GAS INLET FCR [ === GAS INLET (WET OR DRY) —==TO VENT I TO PUMP AND MASS-ANALY ZER Fig. 9.10. Schematic diagram of permeation measurement apparatus. were made using ordinary techniques of pumping speed and volume measurement. Thermocouple gages were used for measurement of pressures. The two experi- ments studied (1) permeation of protium from a wet or dry argon-hydrogen mixture in region A through a stainless steel (304) tube into a deuterium atmosphere in region B and (2) a similar measurement of the permeation of protium through a stainless steel type 347 tube into evacuated region B, with monitoring of hydrogen concentration in region A. The first experiment was performed as an attempt to follow the development of an oxide layer by watching the fall of protium partial pressure in region B after adding steam to the gas in region A. It was observed that the protium partial pressure increased when water vapor was added to the argon-hydrogen mixture. The second experiment was conducted to clarify the ques- tion of the hydrogen partial pressure in region A when the gas was saturated with water vapor. Analysis of the data indicated that the increase was less than 2 or 3%. The first experiment was conducted at a temperature of 579 +-2°C. The results are shown in Fig. 9.11. Deuterium flowed through a Y, 4-in. tube 32 in. long with a wall thickness of 0.010 in. Evacuation of region A vyielded a log P vs time plot with a slope corre- sponding to 0.018 cm?®/sec, a value of 1 atm for the deuterium pressure, and 45 u for the steady-state (corrected) pressure of deuterium in region A measured by a thermocouple gage; a value of 1/2 3 0‘09cm mm atm hr cm? was obtained for the 304 stainless steel tube used here. This compares quite well with the value obtained by Webb,3® although it was not intended to be a precise 50. R, W. Webb, Permeation of Hydrogen through Metal, NAA-SR-10462 (July 25, 1965). ORNL-DWG 71- 7254 SCAN NUMBER o 100 200 300 400 500 600 100 rp—\wmi? 3 (DHM) O Y | O S o} 50 ’é; S~ I MAaSS 2 (DY) 20 o R e : MASS 5 (DpH™) I* {arbitrary units) ) T 5 A Tl ] I—- Ar, Hp, Hp0 2 — DRY Ar, H2 ON T DRY Al’ HZJ ] EVACUATE REGION A J D, OFF (REGION E!) 1 o 700 1400 2100 2800 3500 4200 TIME (sec) Fig. 9.11. Protium permeation into deuterium from steam- hydrogen-argon. measurement of the permeability. Since mass analysis was conducted at high pressure (150 u in the ion source) numerous ion-molecule reactions occurred. From deuterium, masses 2, 4, and 6 are obtained with mass 4 (D,") dominant, mass 6 (D3") to the extent of about 10% of mass 4, and mass 2 (D) to the extent of about 1.5% of mass 4. The amount of protium measured was about 0.6% of the deuterium peaks and was therefore reflected almost entirely in changes of the mass 3 and 5 peaks. These changes with alteration of the experimental conditions are shown in Fig. 9.11. On starting a flow of dry argon — 4% hydrogen, a rapid parallel increase in the protium-containing peaks (3 and 5) occurred with no significant change in the mass 2 peak. After a steady state was reached (at about 700" sec) the gas mixture was valved through a water bubbler, which should have produced about a 2% water content. However, instead of a decrease in protium in region B, a slow increase was observed, which is attributed to oxidation of the metal by the water. This increase lasted about 12 min, after which a low decrease was observed with a slope reflected both in the mass 3 and 5 curves of dlog Py, ~ . —r 03hr . Since a slight decrease in mass 2 also occurred (with a slope of 0.15 hr™"), the decrease of hydrogen permea- tion into the steel tube (presumed due to the oxide buildup) was probably about 0.2 for the value of the slope. After 2000 sec the argon-hydrogen mixture was valved so that the water bubbler was bypassed. A level for protium was reached which was about the same as the initial rate with dry gas. (The scatter of points for mass 3 between scans 210 and 400 was due to a recorder scale change.) The chamber was evacuated to permit measurement of the permeability of the sample as described above. : The question of whether there was an alteration of gas composition in region A during this study was explored in a second experiment, A sample of stainless steel (type 347, Y, in. in diameter, 11 in. long, 0.030 in. wall thickness) was evacuated inside (region B) and exposed to the same perturbations of the outside as before (region A). The permeability measured (at 620°C) was 1/2 _, cm® mm atm 43 X 10 hr cm? To determine whether there was an appreciable change in the H,/Ar ratio when the gas was admitted wet, use was made of the fact that although the reactions 105 H, +H,0* >H,0*+H —44 AH kcal/mole and H,*+H,0-H,;0"+H —202 AH kcal/mole are exothermic and will occur, the ion-molecule reac- tions involving the production of ArH* from Ar* do not involve water or water ions, Hence the ratio of mass 41 (ArH*) to mass 40 (Ar*) should reflect the hydrogen gas concentration.® ! The results summarized in Fig. 9.12 show an initial condition with an ArH*/Ar* ratio of 2.29 * 0.05 initially and a precipitous drop on diverting the flow through the bubbler due to the presence of air in the bubbler volume. (N, has a proton affinity of 131 kcal/mole. Thus the ArH*/Ar” ratio dropped to 0.4 for 2 min.) The level of the ratio was reestablished after .3 min and maintained at 225 * 0.02. Some erratic behavior was observed during reestablishment of dry conditions. The pressure level in region B showed a 25% increase of pressure as compared with 45% observed in the first experiment. Aside from the point of no appreciable hydrogen being made by the admission of water is the sensitivity for water shown in Fig. 9.12. Both by charge exchange from the dominant gas, argon, and proton transfer (principally from H,"), the water produces the domi- nant ions in the spectra. While this is expected, it is interesting to tabulate the spectra to show the enhance- ment of sensitivity which is obtained by using ion- molecule reactions. Table 9.7 shows the ionization produced by these chemical ionization reactions and compares them with the data from electron impact ionization performed by increasing the mass spectrom- etry repeller potential and decreasing the ion source pressure to a few microns. In summary, after establishment of steady permeation of protium from an argon—4% hydrogen atmosphere through types 304 and 347 stainless steel, switching to wet conditions (about 2% H,O) yielded an increase in protium permeation which is attributable to a corrosion reaction. As the oxide coating increased in thickness, the permeation rate slowly decreased. The protium 51. This approximation would not be valid at very high water concentrations since the proton affinity of H,O is 169 kcal/mole and that of argon is but 54: fewer average coilisions would be needed to be reflected in an ArH™ decrease. For the ion source used here approximately 20 collisions are calculated at 100 u. 106 Table 9.7. Comparison of chemical and electron impact ionization for Ar—4% H, —2% H, O as percent of total Argon H, ' H,O (40 + 20 amu peaks) (2 amu) (17 + 18 amu) Electron impact 92 5 3 ionization (40 amu) (41 amu) (18 + 19 amu) Chemical (ion-molecule) 9 21 70 ionization ' ORNL-DWG 71-7255 increase is not due to an increase in hydrogen partial pressure but to a direct charging of the hot metal with hydrogen during the oxidation process. No diminution of deuterium permeation was observed. The questions on permeation processes still to be l l resolved are: acat, 19, H30* 1. the extent to which the slow oxide growth in the K 3 presence of steam at any pressure up to normal steam generator conditions will ultimately impede 18, "sz0+ ~ permeation, o” 2. the extent to which independently applied coatings A : - sddage—p—a | Y can be used to decrease permeation rate under +_ T Ad AedbAAA A 41 41, ArH 2 steady-state conditions, 3. the functional relationship between permeation and partial pressure at very low partial pressure of N ‘k hydrogen, a trtrirenies 2 b : 4, whether the assumption of general equivalence of \ protium permeation and deuterium permeation is \ \ valid, that is, whether at any protium pressure \ \ competition for desorption sites might impede out- \ \ flow of deuterium under conditions like the first experiment reported here, ION CURRENT (arbitrary units} 19 | \ 5. the relations between alloy composition, structure of the stable oxide coating, and effectiveness in \ reducing hydrogen permeation, 6. the possible role of molten salts in the permeation 29(x ¥5) NoH* process. \ 28(x Y5} Nyt w1 P 0 200 400 600 80C " {000 TIME (sec) . Fig. 9.12. Hydrogen-argon ratios as monitored by masses 40, 41, 19 amu. 10. Physical Chemistry of Molten Salts 10.1 THERMODYNAMICS OF LiF-BeF, MIXTURES FROM EMF MEASUREMENTS OF CONCENTRATION CELLS D.D.Sood! K.A.Romberger? J. Braunstein Electromotive force measurements of the molten-salt concentration cell with transference LiF ! LiF Be I Be (A) BeF, | BeF, I II have been reported previously along with their applica- tion to refinement of the liquidus of the LiF-BeF, phase diagram®»* and determination of the ionic transference numbers,> when combined with pre- viously reported emf measurements of concentration cells without transference (The transference number of an ionic constituent is the fraction of the electric current carried by that constituent relative to a refer- ence constituent.) Here we report some excess chemical potentials and partial molar enthalpies of mixing derived from such measurements as part of a study of the thermodynamics of mixing in molten salts of reactor interest. ' The electromotive force £ of the cell (A) may be written® 1. Representative of the Indian Department of Atomic Energy, on assignment at QORNL from Bhabha Atomic Re- search Centre, Bombay, India. 2. Present address: Kawecki-Berylco Corporation, Hazelton, Pa. . 3. K. A. Romberger and J. Braunstein, MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, p. 161; J. Braunstein, K. A. Romberger, and R. Ezell, J. Phys. Chem. 74, 4383 (1970). 4. R.E. Thoma et al., J. Nucl. Mater. 27(2), 176 (1968). 5. K. A. Romberger and J. Braunstein, fnorg. Chem. 9, 1273 (1970). 6. B. F. Hitch and C. F. Baes, Jr., Inorg. Chem. 8, 201 (1969). : 107 (1 X duy;r » where F is the faraday, ¢ ; is the transference number of lithium ion relative to fluoride, x is the mole fraction of BeF, in a BeF,-LiF mixture, and HpeF, and up;p are the chemical potentials of BeF, and LiF. In order to derive chemical potentials from isothermal measure- ments of a cell with transference, it is necessary to know the concentration dependence of the transference numbers. In this system, it has been shown that t; =1 at temperatures between 500 and 700°C and composi- tions between 0.3 and 0.7 mole fraction BeF,. (The transference number probably is also equal to unity in the concentration range between 0 and 0.3 mole fraction BeF, since the transference number of a constituent in a molten salt generally approaches unity as its mole fraction approaches unity.) The isothermal composition dependence of the excess chemical potential may be obtained from emf measure- ments of the cell (A) at closely spaced compositions in the half-cell (II) (compartment I is a reference half-cell). From Eq. (1) and the definition of the chemical potential, the composition dependence of the excess (a#fm> ) 3E - x /o ox/) p Measured values of AE/Ax at 500°C are shown in Fig. 10.1; the smooth curve is (0£/0x). Calculated values of (aufiF/ 0x) are plotted against composition in Fig. 10.2. The excess chemical potential of LiF along the liquidus curve, calculated from the liquidus data,® can be represented by the equation, fitted by least squares, chemical potential is X 1 +x 108 ORNL-DWG 74-3550 1600 I jLIQUIDUS l . ! SYSTEM LiF-BeF, H o | et T =498 +0.5°C 1400 Z._’“ ..qm.; +—— SILICA CELL hi H O —— METAL CELL H s LH H S Tl H A DL [Hg) 1200 : . S H A,A‘H 2Y iR ‘ ™, T 1000 = z \T — ( ] /—l- b_“‘ wlx |0 = 800 “r\_q e \H H HH 600 '_'t_u—&_k‘ W ~ 400 -——\«\ = — l\ 200 0.30 0.34 0.38 0.42 0.46 0.50 0.54 0.58 0.62 X (mole fraction Bef,) Fig. 10.1. Isothermal composition dependence of (AE/Ax)T for the concentration cell (A) with transference employing beryllium electrodes. ORNL-DWG 713551 - 1 10 I} o — £ O —==3 0 S £ . B \t E = — 2 -0 \-\ pd -0 8 3 N s =E Yo Th mg: § -20 -20 w3y S —— -30 -30 0 0.1 02 03 04 05 08 X (mole fraction BeF,) E kcal/mole HLiF S = —4.7141 X liguidus —10.5916x — 61.282x* . (3) The derivative [dufiF/dx] liquidus 2lso is shown in Fig. 10.2; it may be written E E [d'uLiF] _ (a‘uLiF> dx 1 jiquidus ox /r E OML iF dT + _ . (4) oT X dx liquidus Fig. 10.2. Composition dependence of the excess chemical potential of LiF in BeF,-LiF mixtures at constant temperature and along the LiF liquidus. Since (a,ufiF/aT) = _SEEF, the partial molal excess entropy and enthalpy may be calculated as E E (af-"L iF/ax)T - [d”L iF/-dT] liquidus [dT/dx] iquidus (5) 3 (6) ~E _ E oE Hi g =uLir t TSLiF - These equations, based on isothermal emf measure- ments and accurate liquidus data, lead to the estimates: - x=03,T=530°C SE.p=+025eu fifiF = —1013 cal/mole E.r = —1215 cal/mole) — — o cE _ l x=033,T=450°C §F..=086eu IjE HLiF E - LiF = _-]1051 cal/mole —1670 cal/mole) Data at other temperatures and compositions are being analyzed, and experimental work is in progress on the development of an Li-Bi alloy electrode to obtain additional thermodynamic data in this and other systems. 10.2 EQUILIBRIUM PHASE RELATIONSHIPS - IN THE SYSTEM LiF-BeF, -CeF; L. O. Gilpatrick C. J. Barton A tentative phase diagram was presented earlier’ for the system LiF-BeF,-CeF;. Additional data have been obtained using the differential thermal analysis (DTA) technique. No major changes in the proposed diagram have resulted from this new information. Liquidus temperatures were measured for six compositions rang- ing from 5 to 20 mole % CeF;, which as a group indicate that the liquidus contours within this composi- tion region are some 50°C higher than those proposed earlier on an a priori basis. Further revision of the equilibrium phase diagram will be delayed until addi- tional data become available. 10.3 ELECTRICAL CONDUCTIVITY OF MOLTEN AND SUPERCOOLED NaF-BeF, (40-60 MOLE %) G. D. Robbins J. Braunstein Preliminary electrical conductance measurements of supercooled molten NaF-BeF, (40-60 mole %) have been obtained with an all-metal conductance cell, described previously.® These extend from 418° to 310°C, 95° below the liquidus.®s'® The purpose of these measurements is to relate the temperature de- pendence of conductance in glass-forming beryllium fluoride systems to their glass transition temperatures. After loading the component starting materials (222. g), always handled in a protective atmosphere, resist- ance was monitored as a function of temperature and measuring frequency for more than two weeks, during which period the frequency dependence gradually decreased until the change of resistance at constant temperature and constant frequency remained ~0.01 Q/day for several days. Data obtained 17 days after 7. L. Q. Gilpatrick, H, Insley, and C. J. Barton, MSR Program Semiannu, Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 89, 8. G. D. Robbins and J. Braunstein, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p, 98, 9. R. E. Thoma, ed., Phase Diagrams of Nuclear Reactor Materials, ORNL-2548, pp. 34135, 10. D. M. Roy, R. Roy, and E. F. Osborn, J. Amer. Ceram. Soc. 36, 185 (1953). 109 ORNL— DWG 74-7519 12 I T TEMPERATURE (°C) /O 3|O°M-‘ \ \ b 326° \ | pzase 0 363° |0 380° Lo | — 393° -_______—-O 404° "] Lo—0—"] 0 HO° 2 W?—T" TP aig>_| §g}@o——°w —-—"'——- 1 0 - 0 0.004 0.008 0.012 0.016 Fig. 10.3. Measured resistance of NaF-BeF, (40-60 mole %) as a function of the square root of the reciprocal frequency IN/Y initiation of the experiment are shown in Fig. 10.3, where the measured resistance is plotted against the reciprocal square root of frequency and extrapolated to infinite frequency. These data were taken over a two-day period in the order indicated on the Arrhenius- type plot (Fig. 10.4) of the logarithm of the extrap- olated resistance (R_ ) vs reciprocal temperature. Calibrations of the all-metal conductance cell have been carried out at 25°C in 0.01 and 0.001 N aqueous KC1 solutions and 0.001 N dioxane (79 wt %)—H,O solutions (each corrected for solvent conductance) to determine the variations of cell constant with electrode depth (the major variable), solution volume, dielectric constant, and measured resistance. From the calibration data, computer-fitted interpolation functions will be used for final calculation of the specific conductance. Based on a glass transition temperature of 117°C (see following section), the data presented here encompass the interval 7(°K)/T, = 1.77—1.49. From these prelimi- nary results, apparent activation energies for specific ORNL— DWG 71 - 7520 7(°C) 425 400 375 350 325 300 2.4 . : | i | ) I [ — 10 [ ] 2.2 : yi 9__ 9 w | / x — o | / 2.0 g1 - a | | 1) 1.8 S / — 6 w | / - | / | 1.6 B4 — 5 2 | / 5 7o ta|— 5 / . 4+ — 4 . / 8 ! / SUPERCOOLED c 12 ' /£ uaub . — £ I 24 'S | / -~ 3 | 1.0 | 7 , 1/ 0.8 : ¥ 4 6 s 1, ( 04 0.6 7 ./ A 0.4 ;’fl 34 : 0.2 : | | 0 L { .40 1.45 150 155 160 165 170 {75 1000/7 (oK) Fig. 10.4. Logarithm of extrapolated resistance vs reciprocal temperature CK). conductance vary from 13 to 17 kcal/mole over this 108° interval. For the glass-forming pyridinium chlo- ride—zinc chloride system'' at 58.4 mole % ZnCl, (T, = 1°C), activation energies vary from 6.5 to 10.2 kcal/mole at T(°K)/Tg = 1.77-1.49, increasing to 20.9 kcal/mole at 1.26. The molten fluoride data presented here contain a large (relative) frequency extrapolation at the lower resistances and should be viewed with appropriate uncertainty (perhaps £20%). Studies are in progress at higher BeF, contents, where considerably larger meas- ured resistances are expected. However, in order to obtain satisfactory data in the composition region of the larger glass transition thermal effects, it will be necessary to modify the cell design (electrode size) so that a higher resistance is measured, as the low measured resistance amplifies apparent frequency (elec- trode surface) effects.!2+13 10.4 GLASS TRANSITION TEMPERATURES IN THE NaF-BeF, SYSTEM G. D. Robbins J. Braunstein The relation of glass transition temperatures to models of transport in molten salts has been discussed in an earlier report.'® Experimental techniques and results were reported for glass transition temperatures, Tg, obtained by differential thermal analysis (DTA),' 51 for NaF-BeF, mixtures of composition 50, 55, and 60 mole % BeF,. These studies have been expanded to encompass what appear to be the limits of the glass-forming region in this system. Evaluation of these glass transition temperatures will permit compari- son with “theoretical glass transition temperatures” (believed to be near, but slightly below Tg”), as indicated by the temperature dependence of activation energies of electrical conductance, currently being obtained. Glass transition temperatures were measured in DTA heating curve experiments with 3-g samples of NaF- BeF, mixtures which had previously been heated for ~20 hr at 850°C in evacuated, sealed nickel tubes'® and quenched in ice water. The samples were heated slowly (2°/min); the transition temperatures observed in the glasses are listed in Table 10.1, together with data for the three previously reported mixtures. Column 2 of Table 10.1 lists the temperatures at which a transition occurs in the plots of temperature difference between an Al,O; reference and a fluoride glass sample vs the temperature of the fluoride sample. These thermal effects are interpreted as resulting from glass transitions — the relaxation on heating of the frozen glass into a supercooled liquid. ' The new transition temperatures reported here were measured with a calibrated Pt vs Pt—10% Rh thermo- couple. Two thermal transitions which occur in this 11. A. J, Easteal and C. A. Angell, J. Phys. Chem. 74, 3987 (1970). 12. G. J. Hills and S. Djordjevié, Electrochim. Acta 13, 1721 (1968). : 13. G. Jones and S. M. Christian, J. Amer. Chem. Soc. 57, 272 (1935). 14. G. D. Robbins and J. Braunstein, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 99. 15. Obtained with the apparatus of L. O. Gilpatrick, whose technical advice is gratefully acknowledged. 16. L. O. Gilpatrick, S, Cantor, and C. J. Barton, “D.T.A. Apparatus with Agitation and Sealed Specimens for Molten Salt Phase Studies,” pp. 85--96 in Thermal Analysis, vol. 1, ed. by R. F. Schwenker, Jr1., and P. D. Garn, Academic, New York, 1969. Table 10.1. Glass transitions of NaF-BeF, mixtures Mole % BeF, Tg Q) AT (°C) 30 38 130, 127 .5 0.5,0.5 43 123.5, 124 1.1,1.2 46 120.5, 120.5, 120.5, 120.5, 0.7,1.1,1.1,1.1,0.5 118.5 50! 117.5,117 0.8,0.8 55! 116.5,118, 115,115 0.5,0.7,0.7,0.5 60 11’/'.5,1 118,' 117, 116.5,117 0.8,0.4,0.7,0.9,0.8 80 118,119.5 0.3,0.3 90 114.5,111, 113, 113 <0.1 1.00 No effect No effect temperature range and which have been found useful for DTA temperature calibration!”? (the crystalline inversions of potassium nitrate at 128°C and ammo- nium nitrate at 125°C) appeared 1.5° too low when measured with this thermocouple. Hence a 1.5° correc- tion was applied to all data obtained with the thermo- couple. Redetermination of the previously measured value of T, for Xp,p, = 0.60 (and adding 1.5°) resulted in the last three values listed for that composi- tion in column 2. These values agree well with the previously reported values obtained with calibrated Chromel-Alumel thermocouples. Although tempera- tures in column 2 are reported to the nearest 0.5°C to avoid round-off error in computation, resolution of the transition temperatures more closely than 1°C is con- sidered uncertain. Column 3 contains an indication of the relative magnitudes of the glass transitions, reported as the maximum temperature shift at the transition from the basedine DTA trace. On increasing the beryllium fluoride content from the eutectic at 43 mole % BeF,,'%:'® the thermal effect decreases until at 90 mole % BeF, the break is at the limit of detection. No such thermal effect could be found on samples of pure beryllium fluoride. However, since pure BeF, is known to form a glass easily,?® it appears that the glass- forming region extends to this end member. A quenched sample containing 30 mole % BeF, did not 17. 1. Barshad, Amer. Mineral. 37, 667 (1952). . 18. R. E. Thoma, ed., Phase Diagrams of Nuclear Reactor Materials, ORNL-2548, pp. 34—35 (November 1959). 19. D. M. Roy, R. Roy, and E. F. Osborn, J. Amer. Ceram. Soc. 36, 185 (1953). 20. H. Rawson, Inorganic Glass-Forming Systems, pp. 235-48, Academic, New York, 1967. 111 ORNL- DWG 71- 7524 ~ 122 2 ¥ . -~ - : ~ \ - 1 - ¥ ~124 DR ?L { |‘ ~ . \\‘| Ty i \ \ II o \ 120.5 # — - - N S J 120.5R:j7' Y N Y (SO N E H N 80 90 100 MO 420 - 130 140 150 460 #(°C) 100 150 200 250 300 350 400 7 (°C) Fig. 10.5. Differential thermal analysis traces of an NaF-BeF, (5446 mole %) mixture. exhibit a glass transition; instead only the expected solid-state transitions’®:!® were observed. It thus ap- pears that the region of glass formation, under the conditions described above, extends from the mid-30% beryllium fluoride region to pure BeF,. This may be compared with the variously reported glass-forming regions in this system?® of 60—100,2' 40—72,2%, and 45-70%3 mole % BeF,. In one case we were able to observe the reverse glass transition on cooling. Several of these DTA traces are shown in the upper portion of Fig. 10.5 for XBBF2 = 046. Observation of T, while cooling is especially difficult due to the slow rate of heat loss from the furnace smearing out the transition. The data shown in Fig. 10.5 were all obtained in experiments where the heating rate was 2°/min, but with an average cooling rate of 0.3°/min. 21. D. M. Roy, R. Roy, and E. F, Osborn, J. Amer. Ceram. Soc. 33, 85 (1950). . 22. W. Vogel and K. Gerth, Glastech. Ber. 31, 15 (1958). 23. M. Imaoka and S. Mizusawa, J. Ceram. Ass. Jap. 61, 13 (1953), as reported in ref. 9. ORNL-DWG 71-7522 NO GLASS 140 _f ' i GLASS 130 3 — g ¥ B i 120 | %fl $ § 4 & | 1 | 1o : 100 03 04 05 06 Q.7 08 09 1.0 XBer Fig. 10.6. Glass transition temperatures of NaF-BeF, mix- tures vs mole fraction BeF,. The lower portion of Fig 10.5 demonstrates the effect of heating a quenched sample beyond T, until the supercooled liquid crystallizes, followed by the ex- pected crystalline inversion in 2NaF-BeF, (320°), then by the solidus (340°) and liquidus (~365°) transi- tions.’®>!? Marked undercooling is observed on the cooling cycle. Such extended heatings were used to confirm the interpretation of the glass transitions. Figure 10.6 shows a plot of the observed values of T, as a function of mole fraction beryllium fluoride. The open circles indicate experimental data, with duplicate values represented as ears on the circles. The average value of 7, at a given composition is shown as a solid circle. The vertical bars represent +2° of the average T, value, the estimated accuracy of these results being no greater than this. It appears from the data presented here and the reported behavior of pure BeF,2° that glasses can be formed from NaF-BeF, mixtures of composition 38 to 100 mole % beryllium fluoride. This region includes the primary phase fields of BeF,, NaF-.BeF,, and 2NaF-BeF,.'%:'® For compositions less than 31 mole % BeF, the precipitating phase is NaF,'®:'® and no glass formation could be detected in the sample of 30 'mole % BeF,. The central portion of the glass-forming region exhibits glass transitions at 117 + 2°C from 50 to 80 mole % BeF,, rising with higher NaF content and, perhaps, decreasing near pure BeF,. 10.5 RAMAN SPECTRA OF BeF, 2 IN MOLTEN LiF AND NaF TO 686°C A.S.Quist J.B.Bates G.E. Boyd Raman spectral studies are particularly well suited for investigations on BeF,?”, because all of the four normal 112 modes of vibration of the tetrahedral anion are Raman active, whereas only two modes are infrared active, Molten Li,BeF, and Na, BeF, were studied to obtain information on the effect of the nature of the cation on the vibrational spectrum of BeF,?". In molten Na, BeF; and Li, BeF; only the minimum fluoride ion concentration necessary for complete 4-coordination of beryllium is present; hence, it was of interest to measure the spectrum of BeF,? in an excess of F™. Accordingly, a solution of 17 mole % BeF, in NaF-LiF also was studied. Additional measurements were made "on aqueous BeF,* solutions at 25°C to obtain spectra under conditions of reduced interionic interactions. Raman spectra of molten Li,BeF, were obtained from 487 to 640°C.2*A typical spectrum taken at 533°C is shown in Fig. 10.7. Three of the four bands are clearly visible at frequencies near 390, 550, and 800 cm!. The expected fourth band is located near 260 cm~! but is somewhat obscured in Fig. 10.7 by the steeply rising background; this band is more apparent in other spectra, such as in Fig. 10.8 for molten Na, BeF, at 616°C. The vibrational frequencies observed with molten Li; BeF, at several temperatures, with Na, BeF, at 616°C, with the NaF-LiF-BeF, mixture at 645 and 686°C (no difference), with aqueous BeF,?™ containing excess F~ ion, and with solid Li,BeF,; at 25°C are listed in Table 10.2. The frequencies of the BeF,? vibrations observed from Raman spectral measurements on the molten salts and in aqueous solution (Table 10.2) are relatively insensitive (within the experimental uncertainty) to the cation, the temperature, or the composition. The polarization (Fig. 10.8) and intensity of the band at 550 + 3 cm™! in the Raman spectra make it logical to assign it to the totally symmetric stretching mode, v1(ay), of the tetrahedrally coordinated BeF4% anion. This frequency is essentially the same in aqueous solution as in the melts; only in solid Li, BeF, does it occur at a different (higher) frequency. The observed frequency and assignment are consistent with previ- ously reported values for the Raman spectra of solid alkali metal tetrafluoroberyllates?> and for aqueous solutions.?® Although the position of the », band in the melt does not vary greatly with temperature or cation, its half-bandwidth decreases considerably when 24. A more extensive discussion of the results of the present study is contained in a paper by the authors which is to be published in J. Chem. Phys. (1971). 25. A. . Grigorev et al., Dokl, Akad. Nauk USSR 152, 762 (1963) (English transl.). 26. R. E. Mesmer and C. F. Baes, Jr., Inorg. Chem. 8, 618 (1969). : 113 Table 10.2. Vibrational frequencies observed in the Raman spectra of BeF42 in melts in solid Li, BeF,, and in aqueous solutions (frequencies in cm ™) Mol NaF-LiF-BeF, 2.5M Li, BeF 4 melts olten (53-30-17 (NH,),BeF, Li, BeF,(s), 25°C ) " - Na,BeF,4, . Assignment 487°C 533°C 640°C 616°C mole %), in 2 M NH4E Raman Infrared 645, 686°C 25°C 255 240 260 265 255 257 va(e) 295 348 360 385 390 " 390 385 380 . 380 377 372 v4(f2) . 402 405 440 435 475 463 a . 500 547 550 545 550 552 548 563 vi(@y) 775 775 800 800 800 . 800 800 795 795 805 v3(f3) ' 850 860 2Vibrations assigned to the Li* sublattice moving against the F ~ sublattice. ORNL-DWG. 70-5696R ! I ] [ | | [ LiyBeFy (y) 533°C : 90cm i SLIT , 5 x 102 cps i : I i i ,::"; -fé T 3 +::_$- H > : SR ; E g : : 2 g, : Ll HEC HIEHT H i = + = ‘ B Y, B v T JIHHJ ; H . SR SEhE o At 1 i e e N M NS HERERE I B HRE R R S S e it e 1 I I | | I | | I 1000 900 800 700 600 500 400 300 200 FREQUENCY (cm™) Fig, 10.7. Raman spectrum of molten LiyBeF, at 533°C. 114 ORNL-DWG. 71-2575R Hay NozBeF4 ei6°C B 10 e SLIT 1x10° cps [=] P T 4880 A jifitig . Iy i o E[F'. RhByak] - > T = iR 8 ;i %) et Pl i Al TR MNENEY =z W AN ] L LV —— ; H _:]f!i LY. A “ |‘ :;: | i [|; ' R e R N PR ER e SR [ o R SR AT YT | TN AL S R i g i P AR i 53 T H R | 1 ] 500 FREQUENGY (cm™) 300 100 Fig. 10.8. Raman spectrum of molten Na,BeF,; at 616°C. Incident light polarized (a) perpendicular, (b) parallel to plane containing the slit and laser beam. lithium is replaced by sodium. Thus, in molten Li, BeF, at 582°C the v, half-bandwidth is about 100 ¢cm™, whereas in molten Na, BeF, at 616°C this value is near 50 cm™". The origin of this effect appears to be cation dependent since any significant change in anion-anion interaction would probably have been reflected in a frequency shift of v, . The triply degenerate va(f,) and v4(f;) vibrational modes of tetrahedral ions are both Raman and infrared active. Infrared studies of solid alkali metal fluoro- beryllates have led to the assignment of bands near 800 em™! to the »32%:2772% and of the bands near 380 cm™! to v,25°2%2° vyibrations. The Raman spectrum of solid Li, BeF, exhibits three bands in each of these regions, consistent with a lowering of the T; symmetry of the BeF;% ion when it is in a hexagonal unit cell lattice.>® In molten and aqueous tetrafluoroberyllates, only single peaks at 800 + 10 and 385 £ 5 cm ™' are observed for the »; and v, vibrations, respectively, indicating a symmetric environment for BeF,% (no complexing) in these solutions. Half-bandwidths for v, 27. R. D. Peacock and D. W. A. Sharp, J. Chem. Soc., 2762 (1959). 28. 1. LeComte, C. Duval, and C. Wadier, Compt. Rend. 149, 1991 (1959). 29. E. Funck, Ber. Bunsenges. Phys. Chem. 68, 617 (1964). 30. J. H. Burns and E. K., Gordon, Acta Cryst. 20, 135 (1966). and v, in Na, BeF, melts are approximately 110 and 75 cm ™! respectively. The double degenerate v,(e) vibration has not been reported previously. In solid Li, BeF,, peaks at 257 and 295 cm™! are assigned to this vibrational mode. In melts this band occurs at 255 * 10 cm™!; it was not detected in aqueous BeF,? because of the high background in the spectra. The half-bandwidth was approximately 90 cm ™' in molten Na, BeF, . No other bands were observed in the Raman spectra of the melts. However, the spectrum of solid Li, BeF, contains additional bands near 440 and 475 cm ™! and possibly other very weak bands near 125, 165, and 210 cm™'. The latter may be caused by external lattice modes. The strong infrared bands at 500, 463, and 435 ecm™! correspond to the weak bands observed in the Raman spectrum at 475 and 440 cm™. We have assigned these bands to vibrations of the Li* sublattice against the sublattice of F atoms. Similar bands observed at 400 to 550 cm ™ in solid Li, CO53! were found to shift to higher frequencies when °Li was substituted for 7Li. These bands were assigned to stretching vibrations of LiO, tetrahedra. The Li* ions in Li; BeF, also are tetrahedrally coordinated;?° hence, the assignment of the bands between 430 and 500 .em™ (Table 10.2) to motions of “LiF, tetrahedra” seems reasonable. 31. P. Tarte, Spectrochim. Acta 20, 238 (1964). The results of the study reported here indicate that the BeF,% anion retains tetrahedral symmetry in its aqueous solution and in Li; BeF; and Na, BeF,; melts. Both the frequencies and force constants of “free” BeF,?” were well characterized from these studies. A comparison of the frequencies and half-bandwidths for v, in molten Li,BeF,; and Na,BeF, showed that cation-anion interactions are important, while effects of anion-anion interactions were not apparent. 10.6 BUBBLE FORMATION BY IMPINGEMENT OF A JET STREAM ON A FLUID SURFACE H. W. Kohn J. R. Tallackson During the operation of the MSRE, the void fraction was observed to be somewhat dependent on flow rate of the primary salt and on the salt level in the pump bowl. A qualitative investigation of this phenomenon was made using two spray jets of water and flash photography. Two mechanisms were postulated for bubble formation, depending on whether the stream was continuous or had broken into drops. An air sheath which surrounds the liquid jet is apparently dragged beneath the surface of the water, where it breaks up into bubbles. When individual droplets hit the surface, holes may be punched which are then closed over, forming bubbles, a mechanism established by Worthing- ton®? over 70 years ago. A proportionality of total void fraction to stream velocity was noted. A more quantitative study of the phenomenon was made by an MIT Practice School team®® and will be issued as a Practice School report. A preliminary abstract follows. Bubble formation by impingement of a liquid jet on a fluid surface was studied. During the Molten-Salt Reactor Experi- ment helium bubbles were formed in the pump bowl in this fashion. However, variations in the void fraction were observed with changes in salt flow rate and salt level in the pump bowl. The purpose of this study was to determine the dependence of bubble surface area, mean diameter, and average residence time on surface tension, viscosity, angle of impingement, flow rate, tube diameter, and stream length. ’ A two-level half-factorial experimental design was performed using an air—aqueous-solution system to determine the effect of each of the six independent variables on the three responses. Additional data were obtained using an air-water system and varying only flow rate, tube diameter, and angle of impinge- ment. The instantaneous bubble surface area was found to be proportional to a Weber number based on stream velocity and tube diameter to the 1.2 power. Bubble mean diameter was 32. A. M. Worthington and R. S. Cole, Phil. Trans. Roy. Soc. London A189, 137 (1897); A194, 175 (1900). 33. J. T. Boepple and J. B. Cabellon, Bubble Formation by Impingement of a Liquid Jet on a Fluid Surface, CEPS-X-122. 115 found to be proportional to a Reynolds number based on stream velocity and tube diameter to the 0.6 power. The average residence time was found to be proportional to the product of the stream velocity, tube diameter squared, and the square of the sine of the angle of impingement to the 1.3 power, Bubbles were observed forming via the sheath mechanism proposed by Tallackson. A sheath of air was dragged by the liquid jet stream beneath the fluid surface and entrained as bubbles when the flow patterns covered the air with fluid from the tank. Examination of high-speed photographs taken at 1000 frames per second shows that entrainment of an extended air sheath and its subsequent breakup into bubbles is a compara- tively rare occurrence. 10.7 THE SOLUBILITY OF HYDROGEN IN MOLTEN SALT J. E. Savolainen A.P. Malinauskas In .order to determine, and then to control, the transport of tritium which is formed in the operation of a molten-salt reactor, it is desirable to ascertain the extent to which this species interacts with the molten salts in the various loops of the reactor. To achieve this end, we have initiated a program of study of the solubility of hydrogen and its isotopes in molten salt. Because no adequate theoretical treatment of gas solubility -in salt miXtures exists, our experimental program was formulated on the assumption that the hydrogen—molten-salt interaction was similar to that between neon and salt, within a factor of 10 or so. This assumption necessitated that we obviate or minimize the effects of two properties of the system to be investigated. The first .of these, as is apparent from a consideration of Table 10.3, is the significantly higher solubility of hydrogen in candidate apparatus materials compared with the expected solubility in molten salt. Table 10.3. Hydrogen solubility values at 700°C and 1 atm saturation pressure 108 X solubility Solvent (moles H, /cm3 solvent) Reference Fe 651 1a Ni 2579 14 Cu 196 _ le Mo 98 1d Pt 87 o le Au 39 2 Li,BeF, (10) 3 1. S. Dushman, Scientific Foundations of Vacuum Tech- nique, Wiley, New York, 1962, 2d ed. a, p. 522; b, p. 525;¢, p. 528;d,p. 533;e,p. 535. 2. W. Eichenauer and D, Liebscher, Z. Naturforsch, 17a, 355 (1962). 3. Estimate. The second complicating feature, which is illustrated by the data of Table 10.4, concerns the transparency of most metals to hydrogen at elevated temperatures. (Note in particular the rate of transport through gold as opposed to the rate of transport through quartz. Until recently, gold was thought to be impermeable to hydrogen.) - Because it appeared unlikely for us to employ a technique and construction material which would com- pletely eliminate solubility and transport effects, we decided to consider what at first seemed to be a totally illogical approach, viz., to employ Hastelloy N, which, though compatible with the molten salts of interest in this work, is both quite soluble and transparent to hydrogen. This approach, however, led to the apparatus design which is sketched in Fig. 10.9. The apparatus itself is essentially that first employed by Grimes, Smith, and Watson,>* but appropriately modified for use with hydrogen. In brief, the technique involves saturating a molten-salt supply in one chamber of the apparatus and then transferring part of the saturated liquid to another chamber where the gas is stripped from the salt and collected for measurement. Hydrogen losses from the salt in the saturator are eliminated by doubly containing the salt in this region 34. W. R. Grimes, N. V. Smith, and G. M. Watson, J. Phys. Chem. 62, 862 (1958). 116 and by filling the annulus between the containers with hydrogen. Thus, in the saturation procedure, the hydrogen is bubbled through the salt in the saturator and then made to flow around the salt container before being vented. Once saturation has been achieved, a freeze valve, located at F in Fig. 10.9,is opened, and part of the salt is transferred into the stripper chamber. The stripper, like the saturator section, is doubly contained also, but in this case the annulus between the containers is Table 10.4. Rates of transport of hydrogen at 700°C and 1 atm pressure through a 1-mm-thick plate of 1 cm? cross section ] Material (mi)(l)es :157;;“) Reference Fe 83 la Ni 112 la Cu 3.16 la Mo 1.98 la Pt 7.85 la Au. 0.71 2 Quartz?3 0.04 " 1b 1. S. Dushman, Scientific Foundations of Vactuum Tech- nique, Wiley, New York, 1962, 2d ed. a, p. 573; b, p. 497. 2. W, Eichenauer and D. Liebscher, Z. Narurforsch. 17a, 355 (1962). 3. Probably similar to Li,BeF,4 relative to hydrogen trans- port, ORNL-DWG 71-1632 —><}— VACUUM =P <1 R GAS GAS —rd i BURET AND I VACUUM™] ¥ vent | STRIPPER SATURATOR \ F | ( ] 2! Fig. 10.9. Sketch of the hydrogen solubility apparatus. continuously pumped and the gas collected for meas- urement. The hydrogen which does not permeate the Hastelloy N container during the transfer and subse- quent strip operations is stripped directly from the salt through the use of a xenon purge. Separation of the hydrogen and xenon is readily accomplished in a trap cooled with liquid nitrogen. : Several experiments had been conducted to demon- strate our ability to recover virtually all of the hydrogen which is delivered to the stripper compartment. The data from one of these experiments are presented in Table 10.5. The sample of hydrogen which was col- lected at the conclusion was found to contain 29.2% impurity as determined by mass spectrometric analysis. Moreover, xenon appeared to be virtually the only significant impurity (27.4%). On the basis of this analysis, the experiment yields an effective recovery of about 95% after 98 min of operation. Since the amounts recovered from the annulus are entirely free of xenon, we estimate that 3.57 cm® (STP), or about 65% of the original amount of hydrogen admitted, was actually recovered through the Hastelloy N! Several preliminary experiments have been performed to determine the solubilities of both helium and hydrogen in Li;BeF,. To date, the reproducibility of the data is poor (within a factor of 2). Faulty temperature control and measurement have been dis- covered as a possible cause, and this condition is Table 10.5. Hydrogen recovery experiment Hydrogen admitted into strip section at 500°C and at ¢ = 0 min: 5.42 cm?® (STP) Time Re 'or’1 Amount recovered Rate (min) & [cm3 (STP)] [cm3 (STP)/min] 28 Annulus 1.41 0.050 41 Stripper 3.27 . 78 Annulus 1.64 0.033 83 Stripper 0.41 98 Annulus 0.52 0.026 Total 7.25 % 0.708 = 5.13 Table 10.6. Solubility of helium and hydrogen in Li, BeF, G Temperature? Solubility e O (moles gas/cm? salt-atm) X 1077 He 630 1.89 H, 620 248 H, 617 - 3.29 2These values may be too large by about 30°C. 117 currently being' corrected. Typical results of the pre- liminary experimentation are listed in Table 10.6. It must. be noted that these results may be in significant error (also within the factor indicated above). 10.8 ENTHALPY OF UF, FROM 298 TO 1400°K A. S. Dworkin The unusually low literature values for the enthalpy and entropy of fusion of UF,; (102 kcal and 7.7 eu respectively) are suspect because the calorimetric meas- urements®> from which they are derived were prema- turely terminated when leaks developed in the con- tainer. A recent paper>® reports a polymorphic solid transition in UF,; involving a large amount of heat to explain the low fusion value. However, neither the calorimetric study®® nor the many investigations of UF, at thislaboratory, including a recent DTA study by Gilpatrick,®” have shown any indication of such a tran- sition. We have, therefore, remeasured the enthalpy of UF, from room temperature up into the liquid phase in an attempt to clarify this matter. Our results give no indication of a transition. What was taken to be a transition was possibly caused by an impurity present in the material used in ref. 36. The following equations represent our measured enthalpy data for UF, in cal/mole: Hyp — Hyg =—9650 +29.53T+1.15X 107°T? +2.21 X 10% 77! (298--1309°K), AHg, o = 11,230 cal/mole AStusion = 8-6 eu (1309°K) , Hp— Hygg =—9420 + 39.57T (1309—1400°K) . Our entropy of fusion is somewhat but not signifi- cantly larger than that in the previous measurement.?3 It is still quite low for a compound with five atoms. Table 10.7 shows a comparison of the entropies for the isostructural (monoclinic) compounds ZrF,, UF,, and ThF,. The entropy of melting for ZrF, of 12.7 eu is 35. E. G. King and A. U. Christensen, U.S. Bur. Mines, Rep. Invest. 5909 (1961). 36. L. A. Khripin, Y. V. Gogarinski, and L. A. Lukyanova, Izv. Sib. Otd. Akad. Nauk SSSR, Ser. Khim. Nauk 1(3), 14 (1965). 37. L. O. Gilpatrick, MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, p. 148. Table 10.7. Comparison of entropies of ZrF,;, UF,, and ThF, Tyr ASy 8% 98 S°1400—5°208 51400 (°K) (eu/mole) (eu/mole) (eu/mole) {eu/mole) ZiF, 1205 12.72 25.0% 59.7¢ 84.7 UF, 1309 8.6 36.39 56.0 92.3 ThF, 1383 ? 34.0¢ ? ? aGee ref. 37. bSee ref. 38. CSee ref, 39, dgee ref. 40. €See ref, 41, considerably larger than the unusually low value of 8.6 eu for UF,. However, the absolute entropy of UF, at 1400°K is larger than that for ZrF, by about 7 eu. This is approximately what would be expected considering the differences in size, mass, and magnetic character- istics of the cations. An attempt was also made to measure the enthalpy of fusion of ThF,. However, experimental difficulties were encountered due to the high temperatures in- volved, and the measurements were terminated to prevent irreparable damage to the calorimetric equip- ment. Enough qualitative data were collected to in- dicate that, as expected, the fusion behavior of ThF, approximated that of UF, rather than that of ZrF, . 10.9 ABSORPTION SPECTROSCOPY OF MOLTEN FLUORIDES: THE DISPROPORTIONATION EQUILIBRIUM OF UF; SOLUTIONS L. M. Toth An investigation of chemical factors which affect the stability of uranium trifluoride as a dilute species in molten fluoride mixtures was initiated sometime ago utilizing absorption spectroscopic techniques. In this research, molten fluoride solutions of dilute UFj3, contained in graphite, are examined to determine the effects caused by changes in the concentration of fluoride ions in the solvent. That such changes affect the coordination number of fluoride ions around U(IV) was demonstrated previously®® by results which showed that the fluorine coordination number de- creases from 8 to 7 as the fluoride ion concentration in the solvent decreases. Such results allow measurements of the spectrum of U(IV) in a molten solution to be used to estimate the F~ concentration of that solution 38. L. M. Toth, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, pp. 103—5. 118 relative to others and to facilitate studies of equilibria which are affected by these concentrations. Although the stability of UF; in molten fluorides has been the subject of previous investigations,>® none of the experiments employed graphite as a container material. From thermodynamic data it was inferred that solutions in which up to 60% of an initial 1 mole % of UF, is converted to UF; should be stable in the presence of graphite. It was not the purpose of the current work to reinvestigate the stability of UF;, but rather to develop laboratory procedures for spectro- photometric studies of UF; molten fluoride solutions which employ graphite spectrophotometric cells. Anomalous behavior of UF; was observed, however, when it became apparent that the stability of UF; in ~the molten fluoride solvent mixtures was less than predicted originally and, in addition, when results showed that the stability of UF; was more sensitive to the acid-base properties of the solvents than was anticipated. These observations motivated the investiga- tion described below, which is still in progress. The equilibrium 2C(graph1te) + 4UF3 = 3UF4 + UC2 (1) has been identified as the disproportionation reaction of dilute UF; solutions in graphite by two approaches: (1) the forward reaction in which UF; is allowed to - disproportionate and reach equilibrium and (2) the back reaction in which UF, and UC, are equilibrated, forming the same approximate (see discussion follow- ing) U(LII)/U(IV) ratio achieved by the forward reac- tion. These measurements were performed in the diamond-windowed spectrophotometric cell*® at tem- peratures up to 750°C in a helium atmosphere with samples of <1 cc volume. The LiF-BeF, solvent system has been used to effect changes in F~ activity by varying the concentrations of solvent components from 48-52 mole % LiF-BeF, to 66-34 mole %. The temperature effect on the equilibrium has also been studied. Uranium(lIT} and uranium(IV) concentrations were measured spectrophotometrically, taking account of absorption coefficient changes due to both solvent composition and temperature. Uranium dicarbide, as a product of reaction (1), was identified by Debye- 39. (@) G. Long and F. F. Blankenship, The Stability of UFj;. Part 11, Stability in Molten Fluoride Solution, ORNL-TM-2065 (November 1969). (b) F. F. Blankenship et al., unpublished data on “The Stability of UF; in Molten Fluoride.” 40. L. M. Toth, J. P. Young, and G. P. Smith, Anal. Chem. 41, 683 (1969), Scherrer patterns of scrapings®! taken from the walls of the graphite cells. 119 To study the back reaction, pyrolytic-graphite-coated . UC, microspheres®? were crushed and used as the reductant for UF, analysis of the microspheres revealed the presence of graphite, UC,, and approximately 10% UC. The effect of this UC phase is estimated to produce UF; at a concentration which represents an upper limit for the disproportionation equilibrium.*® The UF; concentra- tion produced by this mixed reductant was found to be somewhat greater than that achieved by the forward reaction. Although reaction (1) is identified as a dispropor- tionation reaction, the overall reaction rate which is measured experimentally need not necessarily be sec- ond order as is found for elementary homogeneous disproportionation mechanisms. The rate for reaction (1) is expected to be typically first order, being con- trolled by diffusion to the surface of the graphite. This reaction is expected also to be surface-to-volume depend- ent so that caution should be exercised in making comparison of rates between systems of different sizes. The only meaningful data in support of the dispropor- tionation mechanism are then theresults of the tests of equilibrium performed in the two experimental ap- proaches, together with the overall material balance, which showed that in solution 4 moles of UF; produce 3 moles of UF,. It is impossible to account quantita- tively for the insoluble components of the equilibrium under the conditions of these experiments. The immediate results of this work show that not 60% (approximately) of the total 1 mole % uranium is stable in solution but instead approximately 6%. This number is, however, very dependent upon temperature and fluoride ion concentration, approximately doubling for F~-deficient solutions at high temperatures (700 to 750°C) and decreasing to much less than 6% for F “-rich solutions at lower temperatures. Data which will more clearly describe this stability are currently being evaluated and will be given later in detail. It is interesting to note that the MSRE solution composition, LiF-BeF, -ZrF, (64-29-5 mole %), behaves 41. Grateful acknowledgment is made to H. L. Yakel, Metals and Ceramics Division, and H. W. Dunn, Analytical Chemlstry Division, for these determinations, 42, Grateful acknowledgment is made to J. L. Scott and W, M. Proaps for supplying this material. 43. W. R. Grimes, Chemical Research and Development for Molten-Salt Breeder Reactors, ORNL-TM-1853, pp. 37-39 (June 1963). ' in solution. A Debye-Scherrer as an F-deficient solution,**® whereas the proposed MSBR composition, LiF-BeF,-ThF, (72-16-12 mole %), appears to be quite rich in F~ as determined by UF, coordination number determination — much more than would be expected for a solution with 12 mole % ThF,. ' 10.10 THE OXIDE CHEMISTRY OF Pa** IN MSBR FUEL SOLVENT SALT C. E. Bamberger R.G.Ross C.F. Baes, JIr. While we have demonstrated that the addition of 0% to a fluoride melt containing Pa** at 563°C precipitates Pa, 05 quantitatively,*s it was also of interest to study the behavior of Pa** in the presence of variable amounts of 0. _ The experimental procedure consisted in stepwise additions of ThQ, microspheres to 200 g of LiF-BeF,- ThF, (72-16-12 mole %) containing 1800 ppm protac- tinium, maintained in the tetravalent state by a hydro- gen atmosphere. The protactinium solution was prepared by extensively hydrofluorinating, at 700°C, a mixture of the salt with Pa,Os and ThOQ,. This treatment was followed by a lowering of the tempera- ture to 560°C, and since this produced no detectable precipitation of Pa, Qs, it was concluded that the oxide removal was essentially complete. The protactinium- containing melt was then extensively reduced by sparging with hydrogen at two temperatures; because of the negative temperature coefficients for the reduction by hydrogen of metallic oxidizing impurities*® (i.e., Ni**, Fe?*, Cr**), they were removed by the hydrogen treatment at the higher temperature, 720°C. The subsequent hydrogen reduction was carried out at a lower temperature, 560°C, where the protactinium was presumed to be more easily reduced to Pa** (see Fig. 10.12). - The equilibration with omde was conducted by stirring both phases under a flowing hydrogen atmo- sphere, while monitoring the content of HF in the effluent gas. This kept the system sufficiently reducing, while the rate of HF evolved provided an indication of whether the system was accidentally contaminated with oxidizing impurities (i.e., air, moisture). 44. Private communication with D. L. Manning, Analytical Chemistry, is gratefully acknowledged. 45. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr., MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 92. 46. C. F. Baes, Jr., Symposium on Reprocessing of Nuclear Fuels, Nuclear Metallurgy, vol. 15, CONF-690801 (1969). ORNL-DWG 71— 7523 663 °C 5 S £ ~ - —— ~ 'I-.- o ~A oy patt AT 569°C ™~ - S \‘~‘§ 5 T 3 < & L Pa”* AT 668°C \ S \ 2 AN g b, £ o0 N = \ g - \o \ PeS* AT 563°C N \1 L Y \\ \ . 0 0 10 20 30 millimoles of 0%~ added /kg of melt Fig. 10.10. Precipitation of Pa** and Pa® oxides from LiF-BeF;-ThF4 (72-16-12 mole %) by addition of ThO,. The data for Pa”" are from ref. 45; the solid line corresponds to the quantitative precipitation of PayOs. ' This experiment is still in progress. The data presently availablé are shown in Fig. 10.10 as millimoles of Pa** in solution per kilogram of melt vs millimoles of 0% added per kilogram of melt. For comparison we have also plotted the behavior of Pa®* reported earlier®® at nearly the same temperature. It can be seen that the behavior of the two oxidation states of protactinium is quite different; the Pa*'-containing phase appears con- siderably more soluble than Pa, Og. Although we have not yet isolated and analyzed the solid Pa**-containing phase, preliminary material balance calculations would indicate that it is a (Pa-Th)O, solid solution. A less likely possibility is that it is an oxyfluoride, PaOF,. Work continues in order to characterize the solid phase and to study the effect of temperature and of the composition of the solid phase on the distribution of Pa** between the fluoride melt and the oxide phase. 10.11 THE REDOX POTENTIAL OF PROTACTINIUM IN MSBR FUEL SOLVENT SALT R.G.Ross C.E.Bamberger C.F. Baes, Jr. It has been previously reported*®” that the precipita- tion of protactinium oxide from MSBR fuel salt holds 120 promise for a very efficient and simple way of removing protactinium from the fuel stream of an MSBR. Our previous work indicated that under oxidizing conditions pure or nearly pure Pa,O;s precipitates by addition of O*" to the molten fluoride mixture and that Pa,Os has a considerably lower solubility than UQ,. In order to select the best conditions for removing protactinium from MSBR fuel salt by the oxidative precipitation of protactinium as Pa,Os, values for the redox potential of the couple Pa**/Pa®* are required. We have previously reported limits estimated from direct measurements of the equilibrium PaFs(d) + 1/2H,(g) = PaF,(d) + HF(g) , (1) Ql :Xpa4+.PHF/(XPas+.PH21/2) , where d, g, and X; denote, respectively, dissolved and gaseous species and mole fractions. In these measure- ments the amounts of Pa** and Pa*" present during the hydrogen reduction were estimated by a material balance based on the amount of HF evolved. The reduction of impurities such as Ni?* and Fe®* (which also produced HF) occurred simultaneously, however, necessitating corrections which could be estimated only roughly. Hence only limits for the values of Q, could be obtained. In order to overcome the above-mentioned difficulties we chose to measure the equilibrium quotient of 1/2Pa, O5(c) + 5/4ThF,4(d) + 1/2H, (g) = PaF4(d) + 5/4ThO,(¢) + HF(g) , (2) Q2 =Xp,¥, 'PHF/XThF45/4 Py, v and to combine Q. with @5, measured previously, 1/2Pa,; Os(c) + 5/4ThF,(d) = PaF;5(d) + 5/4ThO,(c), (3) Qs = XPaFS/XThF4S/4 - Reaction (2) has the following advantages: 1. the concentration of PaFs is held constant by the presence of the saturating phases Pa, O and ThO,, or more likely, by Pa, O5 and a dilute solid solution of Pa** in ThO, ; 47. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr., MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 92. 2. the concentration of PaF; is known from Q5 3. the concentration of PaF, can be easily estimated by measuring the concentration of total protactin- ium dissolved in the melt and subtracting the small concentration of PaF calculated from Q5 4. as long as the partial pressure of HF is measured under equilibrium conditions, it is irrelevant whether the HF is generated by the reduction of PaFg, by reducible metallic ions, or by both. ' The experimental procedure consisted in equilibrating under a hydrogen atmosphere the solid phases, Pa; Os and ThO,, with the melt, LiF-BeF,-ThF,; (72-16-12 mole %), in a stirred vessel assembled inside a glove box suited for work with alpha-active elements. Since the reaction studied (2) seemed to involve slow kinetics, the hydrogen was not flowed through the system, as in a transpiration experiment; rather its pressure was held constant, and only the hydrogen lost by diffusion through the walls of the container was replenished. At various intervals, approximately one-fourth of the gas was withdrawn into an evacuated copper vessel and then flushed from it into a small titration vessel with argon. Since the HF pressures measured were in the order of 1072 to 10™* atm, the titrations were carried out with care in order to avoid CO, contamination and to obtain sharp end points. The melt was also sampled at the time of the gas sampling. Since hydrogen was present, we used a special shroud to surround the Teflon seal of the salt sample tube with a flowing atmosphere of argon. This prevented either air from leaking into the equilibration container or pure H, from leaking out into the glove box. All the results obtained are presented in Fig. 10.11. On the left side of this figure the PaF, concentration of the melt is shown as a function of time in each of two runs at each temperature. The amount of PaF, was increased between runs by protonged hydrogen reduc- tion. At the two higher temperatures the product Xpar, Pup. plotted on the right, remained constant within the scatter of the data in spite of the changes in the PaF, concentration; that is, the partial pressure of HF was found to be inversely proportional to the PaF, concentration in accord with equilibrium (2). At the lowest temperature (584°C), however, an obvious change in behavior had occurred in that the PaF, concentrations in the two runs tend to approach a. single value while giving an Xp,p, Py product much lower -than would be expected by extrapolation of the otherresults in Fig. 10.11. It is presently believed that at 584°C a major fraction of the Pa** was in the form of a (Pa-Th)O, solid solution, which should have been 121 3 ORNL-DWG 74-7524 -6 10° = = = —= 10 = 732°C = O~ = = —] l '-_._-‘-o — = — Lo Lo 102 — — :d:' ——-&)—__.—E = 10—7 L PR NS = R _] 10’ 1078 10> = = = 107® = 659°C = - = — w : H-.\‘A\_d_“— = = a ] o 2 10% = Bk o S ) NS b s e i N [T, — — - o | o E = = = f— o. o s oy L = - —_ x =Y — e —daed b — — —] — — — 103' 1078 10° = o = — 107~ = 5_84E = 10% = s = L S T — = Ea+% = Ehfl_‘ p— E p— l— "u--..___. - = - 10‘ - -8- 10— T [|]" 10—9 0 60 120 180 O - 60 120 180 TIME {hr} TIME {hr) Fig. 10.11. Variation of Pa* concentration and of Xpaa+* Py 2s afunction of equilibration time for two different initial concentrations of Pa® at each temperature. appreciably richer in Pa*" than the solid solutions formed at the higher temperatures. With most of the Pa** present in a (Pa-Th)O, solid solution, even a dilute one, the composition of this phase would have re- mained relatively fixed during the last two runs at the lowest temperature. This, in turn, would tend to fix the value of Xp,p, by the equilibrium PaF, (d) + ThO,(ss) = PaO, (ss) + ThF,(d) , Xpao, XThF, ) XTh02 Q4 XPaF4 - The relatively constant level of Xp g, thus produced apparently was lower than that required by equilibrium (2), and hence it appears that Pa, O5 was dissolving, but at too slow a rate, owing to its low solubility, to permit equilibrium with respect to reaction (2). At 659 and 732°C, however, it appears that equilib- rium with respect to reaction (2) was maintained, presumably because the Pa*®* present in ThO, solid solution was but a small fraction of the amount of PaF, in the fluoride phase. The composition of the solid- solution phase therefore could be controlled by the Pa** in the melt and, hence, by reaction (2). ORNL-DWG 71- 7525 0.1 /l' 0.05 L7 Q < L 0.02 X e (=] [+ N Ny / LL¢ . 2 0.0f / x 7 / H] o // 0.005 // / 0.002 : 0.95 1.00 1.05 1.10 145 1000/ (o) Fig. 10.12. Variation of 0 = (XpaF,/XpaFs)PHE/PH, /) as a function of 1/7(C K). Figure 10.12 shows a plot of log Q, as a function of 1/T(°K) estimated for the two higher temperatures only. This plot indicates that molten fluoride solutions of PakF, are oxidized more easily by HF at the higher temperatures and hence that solutions of PaF are reduced more easily by hydrogen at the lower tempera- tures. Due to the importance that the redox potential of Pa*/Pa®* has in an efficient protactinium-removal process as well as in the control of MSBR fuel chemistry, the magnitude of Q, and its temperature dependence will be investigated by means of other reactions. 122 10.12 THE CRYSTAL STRUCTURES | OF COMPLEX FLUORIDES G. D. Brunton The development of automated techniques for rapid collection of diffraction data from single crystals has greatly improved the precision of crystallographic meas- urements. As a result, studies of the crystal structures of the complex compounds encountered in molten salt reactor technology have advanced to the point that the stoichiometry of such crystal phases can often be determined independently and unequivocally by single crystal diffraction methods. Application of this capa- bility to MSR R&D programs provides a powerful means to advance understanding of chemical phe- nomena which involve liquid-solid interactions. A brief review of the results obtained in recent studies of the structures of complex fluorides is included here to illustrate some details of the coordination chemistry which typifies many of the complex compounds which are formed by the heavy metal fluorides. 10.12.1 The Crystal Structure of CsUF, s*® The complex fluoride CsUgF,s crystallizes with space group P63/mmc; a, = 8.2424(4) A, c, 16.412(2) A, Z =2, and the calculated density = 7.0013 . g/cc. The U* ion is coordinated by 9 F~ ions at distances of 2.267(8) to 2.54(4) A and the Cs* ion by 12 F~ at distances of 3.12(2) and 3.45(4) A. 10.12.2 The Crystal Structure of a-KThF, 43 Crystals of a-KThgF,s are hexagonal-rhombohedral R3m with hexagonal axes ao = 8.313(2) and co = 25.262 A, Z = 3, and the calculated density = 6.281 g/cc. The structural units of a-KTheF,s (Fig. 10.13) are identical to those of CsUgF,s; two rings of Th-F polyhedra surrounding a central K ion. The structural units are stacked ABCABC along ¢, in @-KThgF, 5 and ABAB in CsUgF, 5 (Figs. 10.14 and 10.15). 10.12.3 The Crystal Structure of Li, MoF*? Crystals of Li,MoF, are tetragonal P4,2,2 with ay = 4.6863(7) and c, 9.191(2) A, Z = 2, and the calculated density = 3.687 g/cc. The Li* and Mo** ions are octahedrally coordinated. The Li-F distances range from 2.017(2) to 2.102(7) A and the Mo-F distances range from 1.927(2) to 1.945(2) A. The MoF4? ion is coordinated by ten Li* (Fig. 10.16). 48. Acta Crystallographica (in press, 1971). 49. J. Crystal Growth (in press, 1971). 123 ORNL-DWG 71-1795 Fig. 10.13. KThgF, 5 structural unit. ORNL-DWG 71-2330 Fig. 10.14. KTh¢F, s structural stacking. ORNL-DWG 71-2331 Fig. 10.15. CsUgF, 5 structural stacking. 124 ORNL-DWG 71-1686 Fig. 10.16. Li,MoFs. ORNL—DWG 70-13164 Fig. 10.17. RbTh;F ;. 10.12.4 The crystal Structure of RbTh, F, ;%8 of these ions has 9 nearest-neighbor fluorine jons at the : : . corners of capped trigonal prisms. The Th-F distances The complex fluoride RbThFy; crystallizes with 00 trom 2 32(3) to 2.48(2) A. The rubidium jon has space group P2,ma; aq =8.6490(5), by = 8‘17§(2)’ and nearest-neighbor fluorine jons in the range 2.79(3) Co = 74453(4) A. There are two formula weights per t0 3.37(2) A (Fig. 10.17). The final discrepancy index is unit cell, and the calculated density is 6.488 g/cc. There 0.0710 for 2134 Ag Ka reflections >a. are two thorium ions in the asymmetrical unit, and each Fig. 10.18. 10.12.5 The Crystal Structure of RbsZr, F, *® The. complex fluoride RbsZr,F,, crystallizes with space group P2,; 2o = 11.520(5), by = 11.222(5), and co = 7.868(2) A; cos § = —0.1445(3). The calculated density = 3.930 gfcc, and Z = 2. The structure was solved by a tangent formula procedure and refined by Fourier and least-squares methods to a final discrepancy index (R) of 0.0508 for 1376 observed Cu Ka; reflections. The F~ coordination polyhedra are dif- ferent for each of the four crystallographically inde- pendent Zr ions. Seven F~ ions are nearest neighbors to Zr(1) at the corners of a pentagonal bipyramid, and the interatomic distances are 1.99(2) to 2.20(2) A. The second Zr ion has eight nearest-neighbor F~ [1.98(2) to 2.24(2) A] at the corners of an irregular antiprism. The third Zr ion_is octahedrally coordinated by six F~ at distances of 1.90(2) to 2.10(2) A. The fourth Zr ion has seven F~ nearest neighbors at the corners of an irregular antiprism with one corner missing. The interatomic distances Zr(4)-F are 1.95(2) to 2.19(2) A. The Rb-F distances range from 2.68(2) to 3.21(3) A. The struc- ture is composed of cross-linked chains of Zr-F poly- hedra. The chains are connected by Zr(2)-Zr(4) edge- sharing polyhedra, and the space between the polyhedra is filled with Rb ions (Fig. 10.18). 10.12.6 Discussion With the elucidation of these structures it becomes increasingly evident that the complex fluorides have a number of common features from which some generali- zations can be made. A few examples are noted, as 125 ORNL-DWG 70-14373 s | D follows. The heavy-metal-to-fluorine bond distances are generally found to range from 1.9 to 2.5 A, dependent approximately on the charge/size values for the cation; the fluorine coordination numbers for the heavy metals are consistently large, one almost never found to be less than 6, and may be as large as 9. Coordination numbers are not infrequently variable within a single structure. Although such a phenomenon does not necessarily imply the existence of variable quasi-anionic character among structurally distinguishable cation-anion clusters, it does suggest ‘that success in the development of theoretical models of liquid-state behavior may be dependent upon cognizance of some of the charac- teristic details of such structures as are cited here. ' 10.13 NONCRYSTALLINE BeF, AT 25°C: STRUCTURE AND VIBRATIONAL MOTION A. H. Narten J. B. Bates A program of investigation of molten salts by x-ray and neutron diffraction was resumed in support of spectrographic researches with these materials. Prelimi- nary to the study of complex systems, initial studies . were devoted to glasses of BeF, and SiO,. These compounds form highly viscous melts near the fusion point, and the structure of the melts, though of considerable interest, is poorly understood. Upon quenching, the melts solidify into glasses, and the structure of these is thought to be similar to that of the liquid phases. A study of noncrystalline BeF, and SiO, at room temperature by x-ray diffraction has yielded new insight into the average configuration and motion of atoms in these phases. The polymorphism of crystalline SiO, and, to a lesser - extent, of crystalline BeF, is well known. All crystal- line modifications of these compounds contain SiO, and BeF, tetrahedra joined at the vertices so that an oxygen or fluorine atom is common to two tetrahedra. The crystalline modifications differ in the packing of these tetrahedra. The diffraction pattern of noncrystalline SiQ, and BeF, is characterized by the absence of Bragg peaks, and hence no long-range order is present in these glasses. Atom pair correlation functions derived from the diffraction data show very sharp peaks charac- teristic of near-neighbor interactions, but all positional correlation is lost about 8 A away from any starting point. The average distances between near-neighbor atoms in vitreous Si0, and BeF, are equal to those found in the high-temperature crystalline modifications of both compounds (8-quartz). As in the crystalline forms, deviations from ideal tetrahedral geometry are very small in the noncrystalline phases at room temper- ature. In contrast, the average coordination numbers are slightly lower in the glasses. Intensity and correlation functions calculated for a model based on the B-quartz structure are in quantitative agreement with those derived from the diffraction data. The slightly lower density and coordination number found for the non- crystalline phases are described by the model in terms of random vacancies in the tetrahedral networks (12% for SiO, and 6% for BeF,), and these defects result in the progressive loss of all positional correlation with separation. With a structural model thus available, it was possible to calculate the vibrational modes to be expected. Based on the f-quartz structure, the k = o (optical) crystal vibrations of SiO, and BeF, were calculated using an eight-parameter modified valence force field. Both the optical selection rules and calculated frequen- cies were in good agreement with the vibrational spectra of Si0, and BeF, measured in this laboratory and with spectra reported in the open literature. In particular, the theory predicts that the only totally symmetric vibration of the -quartz structure is a bending mode of the 0-8i-O (in SiO, ) or F-Be-F (in BeF,) valence angles. The strongest bands observed in the Raman spectra of Si0, and BeF, were observed at 437 and about 280 cm ™!, respectively, and were therefore assigned as due to the totally symmetric bending modes in each case. The vibrational spectra measured on the glasses agreed with the calculations both in frequency and in the selection rules. This agreement strengthens the con- fidence which can be placed in the structure deduced from diffraction. Neither method can give an unambig- 126 uous structure, and it is therefore useful to employ them together. 10.14 RELATIONSHIP BETWEEN ENTROPY AND SONIC VELOCITY IN MOLTEN SALTS 'S. Cantor A correlation of the entropies of crystalline and liquid ionic compounds with molar weight and with molar volume was reported previously.®® The correlating equation is essentially a method for estimating a quantity € which is related to the distribution of acoustical vibrations in a crystal. For crystals, 8 is usually determined from calorimetric data, but it can also be obtained from elastic-constant data; @ is related to the elastic constants through the mean velocity of sound: (1) where C is the velocity of sound; V is the molar volume; n is the number of atoms in the chemical formula of the compound; and k, h, and N are, respectively, Boltz- mann’s, Planck’s, and Avogadro’s constants. For a molten salt, we may use entropy, either predicted or measured, to provide values of § by means of an equation derived from the Debye theory of lattice vibrations: 7] T > where Si(liq) is the entropy of the molten salt at absolute temperature 7 and R is the gas constant [1.987 cal mole™ (°K™')]. The 6 calculated from Eq. (2), when substituted into Eq. (1), permits the calcula- tion of the sonic velocity of the molten salt. The sonic velocity, thus calculated, is compared with experi- mental values of several binary salts in Table 10.8. The agreement between calculated and experimental sonic velocity is quite good. Thus it would seem that a “liquid” @ derived from entropy is an effective method for predicting sonic wvelocities of binary molten salts. The deeper significance of the good agreement in Table 10.7 may be that “solid-like” models of molten salts lead to useful relationships between physical properties; when these models succeed they ought to be refined and applied to other physical properties: Sr(liq) = 3nR (g — 1n (2) 50. Stanley Cantor, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 95-98. Table 10.8. Velocities of sound in molten salts (km/sec) 1000°K 1200°K Calculated Experimental? Calculated Experimental? LiF 2.57 (2.72) 2.48 2.45 LiCl 1.88 1.93 1.83 1.77 NaF _ 1.78% 2.05% NaCl 1.45 (1.81) 1.38 1.63 NaBr 1.15 (1.34) 1.12 1.21 KF 1.59 (1.94) 1.53 1.76 KI 0.802 1.09 0.787 1.02 CsCl 0.908 1.10 0.844 0.970 CsBr 0.731 0.955 0.680 0.826 CaCl, 1.71 (2.10) SrBr, 1.19 1.54 BaCl, 1.44¢ 1,72¢ 2The experimental data were reported by J. O’M. Bockris et al. and M. Blanc et al. and are summarized in G. J. Janz, Molten Salts Handbook, p. 251, Academic, New York, 1967. Paren- theses indicate experlmental data extrapolated into super- cooled-liquid region, bat 1300°K. CAt 1235°K. What value of @ predicts the sonic velocity of more complex molten compounds such as nitrates? When molar entropies are substituted into eq. (2), the calculated sonic velocities are too low by about a factor of 10. However, in these compounds the total molar entropy includes contributions from internal degrees of freedom — interatomic vibrations and internal rota- tional modes of the complex ion. When we subtract these entropy contributions, representing them as ‘“‘gas- like” behavior (i.e., unhindered rotational and internal vibrational entropies of NO;~ calculated from the standard partition functions of an ideal gas), we obtain a residual entropy associated only with “lattice” contri- butions. Correspondingly, the complex ion is countéd as one particle; for instance in the case of KNO;,n =2 in Egs. (1) and (2). This “lattice” entropy, when substituted into Eq. (1), yields sonic velocities in excellent agreement with experimental values: Velocity of sound at 700°K (km/sec) . Calculated Experimental NaNOs(liq) 1.76 1.67 KNO;(lig) 1.62 1.76 AgNO;(liq) 1.37 1.42 These results do not prové that nitrate ions execute unhindered rotation in the melt. What has been shown is that the propagation velocity of sound waves in molten salts containing nitrate ions appears to depend 127 only on “lattice” vibrations. The entropy associated with the “lattice’ vibrations is approximated by sub- tracting entropy contributions assumed to be caused by free rotation and internal vibrations of the complex ions. 10.15 A REFERENCE ELECTRODE SYSTEM FOR USE IN FLUORIDE MELTS H. R. Bronstein Further study of the galvanic cell comprised of the half-cell Be/LiF, BeF, (67-33 mole %) and the Ni-NiF, reference electrode assembly, as illustrated in Fig. 10.19,51733 led to findings believed to be of consider- able interest and significance. Deterioration of the initial cell voltage with time was found to be caused by the depletion of the small NiF, content of the reference half-cell, as evidenced by its conversion into an equivalent amount of finely dis- persed nickel metal. Since a static atmosphere of highly purified argon or helium gas was maintained in the quite limited volume of the apparatus, an impurity in this small amount of cover gas could not be held _responsible for the reduction of the NiF, . This conclu- sion was verified by subsequent experiments which established the actual mechanism involved. ) This phenomenon continued to occur with all ex- ternal measuring circuits disconnected, as demonstrated by measuring the potential at periodic intervals. Also, beryllium metal has no measurable vapor pressure at 600°C, the temperature of the experiments. The con- clusion had to be drawn that an internal “short” in the cell supplied the electrons for NiF, +2¢~ = Ni® + 2F . Obviously, the beryllium metal, the only material present capable of causing the reduction, must have reached the NiF, in some manner. Therefore, one must not only invoke the transport of -beryllium metal to the reference electrode cell through the salt melt, either by dissolution or by the.so-called “chunking effect,”4:%% but also an electronic component in the conductance of the single-crystal LaF;. The beryllium arriving at or in 51. H. R. Bronstein and D. L. Manning, MSR Program Monthly Report for December 1969 and January 1970, MSR-70-10, p. 28 (internal memorandum). 52. D, L. Manning and H. R. Bronstein, MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, p. 184. 53. H. R. Bronstein, Chem. Div. Annu. Progr. Rep. May 20, 1970, ORNL4581, p. 119. 54. The Encyclopedia of Electrochemistry, ed. by C. A. Hampel, pp. 44—51, Reinhold, New York, 1964. 55. H. Aida, J. Epelboin, and M, Garreau, J, Electrochem, Soc. 118,243 (1971). ORNL-DWG. 71-2238 |<—3/4in.—)1 : — % BORON NEAN Ni NITRIDE —= R || - . \ __'.:.'-? |...|F SINGLE 2\ i 13 in. N \ . CRYSTAL—Hh M\ ,'::.'Ez‘d’ \\\\\ ! Z(C) Ni FRIT Fig. 10.19. Reference electrode assembly. the nickel frit would act as a beryllium atom, if the “chunking effect” were valid, or perhaps as a subvalent Be™ or Be,?* ion, if the solution assumption is valid. Nevertheless, the mechanism would be as follows: Be® - Be?* + 2¢” if as a “chunk of beryllium,” and Be* - Be** +e”, Be,** = 2Be?" + 2e” if as a subvalent species dissolved in the melt, with transport of the electrons through the LaF; crystal and reaction with the Ni** ions to form nickel metal. The excess fluoride ions are also transported (in the op- posite direction) through the LaF; crystal®*!»®? into the main bulk of the melt to maintain electrical neutrality. The overall reaction is then simply Be + NiF, (from the electrodes) to give nickel metal in the LaF; cup and beryllium fluoride in the main melt. A series of experiments was conducted to verify this mechanism. First, a beryllium electrode was mounted through an inverted BeO crucible. Since solid beryllium metal is less dense than the melt, any solid metal detached from the electrode by the so-called ‘“‘chunking 128 effect” would rise to, and be trapped at, the melt surface within the crucible. Dissolved metal, on the other hand, would still reach the reference electrode. The observation was that under these conditions the NiF, in the LaF; cup was still transformed into Ni metal. _ Second, with the same crucible-encapsulated beryl- lium electrode, the reference electrode assembly was modified by. cutting the LaF; crystal below the cup portion and placing a thin disk of CaF, Y4 in. thick between the two portions of the LaF;. It was hoped that the CaF, crystal would perhaps have little or no electronic conduction and so act as a filter of electrons. This cell maintained the theoretical voltage for a period five times longer than the single-crystal LaF; before onset of deterioration. The CaF, apparently had some small electronic conductance of its own. A third, and definitive, experiment was made. The reference electrode assembly was remade without the CaF, disk. To the main melt, NiF, was added as a “getter” for the dissolved Be. The beryllium electrode was now encased in a somewhat porous BeO crucible of Y% in. wall thickness by % in. ID containing the pure melt LiF-BeF, (67-33 mole %). The crucible prevents the mixing of the two solutions, its porosity allowing ion migration. Any Ni** ions entering the crucible are immediately reduced to Ni metal by the saturated beryllium solution. The same result occurs if any dissolved beryllium diffuses into the main melt sat- urated with NiF,. By this means the theoretical cell potential was maintained for a period of three days. Termination of the experiment apparently was caused by the loss of ion migration through the BeO crucible walls plugged up by the deposition of nickel metal. The voltage recorded in all these cells before onset of deterioration was 2.045 + 0.003 V. According to the compilation of Baes®® for the free energy of formation ~of NiF, and of BeF, in BeF,-LiF (33-67 mole %), this cell should yield 2.041 + 0.025 V at 600°C. These experiments have established that beryllium does dissolve in the LiF-BeF, (67-33 mole %) melt to some small extent, yet to be determined, and that, somewhat surprisingly, the conductance in the LaF, crystal has a small electronic component. ‘With this knowledge, the technique described can and will be used to obtain reliable thermodynamic data on a variety of systems of interest. 56. C. F, Baes, J1., “Nuclear Metallurgy,” vol. 15, AIME Symposium on Reprocessing of Nuclear Fuels, pp. 615—-44, 1969, 11. 11.1 EXTRACTION OF RUBIDIUM AND CESIUM FROM MSBR FUEL SOLVENT INTO BISMUTH BY REDUCTION WITH LITHIUM AT 650°C D. M. Richardson J. H. Shaffer The experimental determination of the distributions of the alkali metals in the metal transfer process was initiated over a year ago to assure that the behavior of these metals as fission products, contaminants, or intentional diluents could be predicted.. The distri- butions of sodium and potassium were previously reported for the complete metal transfer process.' Distributions in the chloride/bismuth extraction were reported for cesium,? and the distributions of rubidium and cesium in the fluoride/bismuth extraction at 650°C are reported here. The reductive extractions were performed indi- vidually in 4-in. IPS 304L stainless steel vessels with mild steel liners. Two %, -in. steel pipes extended to within % in. of the bottom of the vessels and provided separately for adding lithium directly to the metal phase and for withdrawing bismuth samples in graphite ladles. Additional ports were provided for material additions, inserting beryllium electrodes, salt sampling with hydrogen-fired copper filter sticks, gas exhaust, and thermocouples. The vessels were each charged with approximately 3000 g of bismuth and were hydrogen sparged at 600°C for 8 hr. The fluoride salts consisted of LiF-BeF,-ThF, (72-16-12 mole %) in batches of approximately 3000 g that were individually treated with carrier salt and radiotracer: 1.48 X 1072 mole % RbF (with *Rb)and 0.95 X 1072 mole % CsF (with '27CsF). These salts were treated in nickel preparation vessels at 650°C according to standard hydrofluorination procedures, followed by hydrogen sparging. Approximately 3000 g of each salt was transferred at 650°C under flowing argon to the extraction vessels. 1. D. M. Richardson and J. H. Shaffer, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 107. 2. D. M. Richardson and J. H. Shaffer, MSR Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, p. 172. 129 Chemistry of Molten-Salt Reactor Fuel Technology Reductions were performed by successive additions of 0.25 or 0.5 g of clean lithium metal, and samples of salt and metal phases were taken after a minimum of 5 hr of argon sparging at 1 liter/min. Analyses of lithium, thorium, and the corrosion metals were performed by the Spectrochemical Labo- ratory and by the General Analysis Laboratory. Anal- yses of rubidium and cesium were made by gamma- spectrometric counting of the 1.077-MeV gamma of 86Rb and the 0.661-MeV gamma of '27Cs. Over a week was allowed for decay of thorium daughter activities in the bismuth samples. Thorium activities in the salts were corrected by means of a salt blank without radiotracer. The equilibrium quotient obtained for rubidium was K, = (Dry/Dpp) = 8.62 + 1.26, the mean of three samples at the 0.95 confidence level. The corresponding distribution equation for this composition of salt was where X . is the mole fraction of lithium in bismuth. The cesium salt preparation was used twice for a reductive extraction into bismuth. In the first instance there was large scatter of the sample data obtained, and the spare samples were expended in the course of measurements of phase segregation of solute in frozen bismuth pellets.? Subsequently, this salt was transferred to a second extraction vessel containing freshly pre- pared bismuth, and the reductions were repeated. The equilibrium quotient obtained in this experiment was Ko =(DcgDLj)=15.05 £081, the mean of 15 samples at the 0.95 confidence level. The corresponding distribution equation for this com- position of salt was log D¢ = log X ; +1.320. The relative extractabilities of the alkali metals from fluoride salt into bismuth containing lithium at 650°C areCs :Na:Rb:K:1:0.80:0.57:0.53. 3. D. M. Richardson and J, H. Shaffer, sect. 11.4, this report. 11.2 DISTRIBUTION OF THORIUM BETWEEN MSBR FUEL SOLVENT AND BISMUTH SATURATED WITH NICKEL AND THORIUM AT 650°C D. M. Richardson J. H. Shaffer Occasionally, in reductive-extraction experiments where various metals were reduced from LiF- BeF,-ThF, (72-16-12 mole %) into bismuth at 650°C, there were instances where the solubility product constant of ThNiBi, in bismuth was reached during the later stages of reduction. It was of interest to examine the experimental data obtained under these conditions to determine whether or not the normal distribution of thorium prevailed. The other metals whose distributions were indi- vidually measured in each of these experiments were cesium,® potassium,® rubidium,® and sodium.® The thorium and nickel solubility data of concern here are plotted in Fig. 11.1, together with the derived line (with theoretical slope): Xt X Xy; = 2.36 X 1076, where X is the mole fraction in bismuth. The experi- mental data shown in Fig. 11.1 were obtained from spectrographic analyses and provide confirmation of the reported solubility product constant® at 650° of 1.5 X 107°. The thorium and lithium data for the same samples are plotted in Fig. 11.2, together with the derived line {(with theoretical slope): XTh/(XLi)4 =47 X 10—7 s where X is the mole fraction in bismuth. The corre- sponding log,, distribution equation for this compo- sition of fluoride salt is log Dt =4 logDyp; +8.022. The distributions of thorium and lithium for this fluoride salt at 600°C and 700°C have been reported’ and are described by the derived equation log [Dyp/(D;)*] = 8987(1/T°K) — 1.4844 . 4. D. M. Richardson and J. H. Shaffer, sect. 11.1, this réport. 5. D. M. Richardson and J. H. Shaffer, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 107. 6. F. J. Smith and L. M. Ferris, J. Inorg. Nucl. Chem. 32, 2863 (1970). 7. L. M. Ferris and J. }, Lawrance, MSR Prograrh Semiannu, Progr. Rep. Feb. 28, 1969, ORNL-4396, p. 284. 130 _2 ORNL— DWG 7{— 7526 10 > N 8 Tl ¢ PN PR N 2 [ ] - 9 z \ b -3 . 210 /A b = _ — G 5 Xpp* Xy =2.36 X140 @ L 4 Z . = 5 ™ ] z O I - 2 104 10"4 2 5 1073 2 5 1072 NICKEL IN BISMUTH (mole fraction) Fig. 11.1. LiF-BeF,-ThF, (72-16-12 mole %) and bismuth at 650°C. Five experiments at nickel and thorium saturation. 102 ORNL—DWG 74— 7527 S U - L4 LJ / . 2 /’ L L 4 £~ . 1070 -fi'.l. - L] 5 . Il X =47 x40 (XLI) 2 / XLi Fig. 11.2. LiF-BeF,-ThF4 (72-16-12 mole %) and bismuth at 650°C. Five experiments at nickel and thorium saturation. The resulting estimated distribution equation for 650°C is log Dy, =4 log Dy ; +8.251 . Considering the scatter of the present data and the inaccuracy of simple interpolation of log K, there is close agreement of these distributions at 650°C. Within the limitations mentioned, the present data show that the distribution of soluble thorium is unaffected by the presence of ThNiBi, in the bismuth phase. 11.3 BISMUTH-MANGANESE ALLOYS AS EXTRACTANTS FOR RARE EARTHS FROM MSBR FUEL SOLVENT D. M. Moulton J. H. Shaffer Bismuth-manganese alloys have been used to extract cerium from LiF-BeF,-ThF,; (72-16-12 mole %). Man- ganese forms reasonably low-melting eutectics and no strong compounds with thorium and rare earths, and it seemed that adding it to the bismuth might raise the thorium solubility without any undesirable effects. An extraction was carried out at 600° with 24 mole % Mn in the Bi, using Th metal as the reductant. Then more Mn was added — enough to increase the concentration to 36.5 mole %, but the solubility is reported to be only 32% — and the extraction was continued. Finally the temperature was raised to 700°, where all the Mn should have dissolved, and more additions were made. Analysis of the samples was by gamma counting and spectrographic techniques. Thorium solubility increased substantially over that in bismuth alone, to 1.36 and 2.38 wt % at 600 and 700°, and it is not certain that saturation was reached even then. The maximum distributions (0.075 and 0.13) were not correspondingly high because of the lower average atomic weight of the metal. Although thorium distribution continued to rise, that of cerium did not; it leveled off as if saturated, but its concentration was so low (0.1 wt % in the salt) that it does not seem likely that it formed a primary solid phase. The distributions are shown in Fig. 11.3, and the equilibrium constants are given in Table 11.1; these latter values are not very different from the figures in pure bismuth. The two manganese concentrations gave the same results and are not distinguished. If the tailing off of the cerium, so far unexplained, turns out not to be real, then it should be possible with this solvent to reduce the amount of liquid metal circulating in the reductive-extraction process without much penalty in the separation. | 131 ORNL-DWG 71-7528 i R T T | N W T T T [ o 700° 05 [-—® 600:] MAXIMUM POINTS IN PURE Bi —— ® 700° ) (MSRP SEMIANNUAL Pz | FEB 1969, p. 191) — / jel 0.2 5] -g/ // S 0.4 Y ) Q . S oe” v 0.05 1 & a SGASLoPE = V3 0.02 0.01 ' 0.001 0.002 0.005 001 002 005 04 0.2 Dy, Fig. 11.3. D, vs Dy, into Bi-Mn solutions from LiF- BeF;-ThF, (72-16-12 mole %). Table 11.1. Logs of equilibrium constants in Bi-Mn solutions 0 Dy/Dc, Dy*/Dyy, D¢ /Dy 600 —7.77+0.18 ~10.17+0.19 +0.81 + 0.09 700 —6.74+0.15 —8.90 + 0.29 +0.26 + 0.39 11.4 REMOVAL OF SOLUTES FROM BISMUTH BY FRACTIONAL CRYSTALLIZATION D. M. Richardson J. H. Shaffer The occurrence of sometimes large variations in solute concentrations was noted in reductive-extraction exper- iments where metal samples were taken in graphite ladles directly from the bismuth phase. Efforts to determine the causes for this have resulted in finding evidence that solutes are strongly expelled to the surface of small bismuth pellets during the process of freezing. This effect could be utilized in a process for concentrating solutes and, conversely, for the purifi- cation of bismuth. The graphite “ladle” is a 7 4-in.!'® For the radioisotope solutes the specific activity of single bismuth pellets was measured before and after each of several scrubbings to remove material from the surface. The change in weight of the pellet was then used to compute the average specific activity of the material removed, which was then normalized as a ratio by dividing by the average specific activity of the original unscrubbed pellet. These ratios are plotted as the ordinate in Fig. 11.4 against the fraction of original weight removed by scrubbing as the abscissa. In the case of thorium, chromium, nianganese, nickel, and lithium, the ratios were deduced by com- parison of the two pellets within a sample pair and consequently are less precise. For all of the metals studied there was a significant segregation of solute in the outer layers of the pellets. For cerium, chromium, and thorium the average con- centration in the outer 1% by weight was 60 or 70 times greater than the average concentration of the original pellet. Expressed differently, this means that two-thirds of all of these solutes was located in this outer shell. Although it has not been proved that high-concentration pellets were not the result of in- homogeneities in the liquid bismuth phase, it is ap- parent that solutes were strongly expelled to the surface 8. D. M. Richardson and J. H, Shaffer, sect. 11.1, this report. 9. D. M. Richardson and J. H. Shaffer, MSR-70-62, p. 11 (Sept. 2, 1970) (internal memorandum). 10. D. M. Richardson and J. H. Shaffer, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 107. and, as a consequence, that low-concentration pellets might be expected. In the second fluoride/bismuth experiment involving 137Cs,® it was found that the top position pellets were consistently lower in thorium and cesium concentration than the corresponding twins from the bottom position. The median difference was 20% for thorium and 5% for cesium. This result can be rationalized by the argument that the pellet in the bottom position is usually the first 133 to freeze since it is farther from the hot mass of the metal sampling rod and since it is upstream in the convected cover gas that sweeps up the hot sampler when it is raised into the standpipe. In this case the slower. cooling of the top position pellet would allow more complete expulsion of solute and a greater likelihood that solute would be lost to the graphite ladle. 12. Development and Evaluation of Analytical Methods for Molten-Salt Reactors A.S. Meyer 12.1 ELECTROANALYTICAL STUDIES OF TITANIUM(IV) IN MOLTEN LiF-NaF-KF (46.5-11.5-42.0 MOLE %) F.R.Clayton' D.L.Manning G.Mamantov?® We are investigating the electroanalytical behavior of titanium(IV) in molten LiF-NaF-KF, NaBF,, and LiF- BeF,-ZrF, . Titanium is present in the structural metal (Hastelloy N) of the molten-salt reactor and, from free energy considerations, should be somewhat more re- active than chromium,. ' Voltammetric studies of the reduction of Ti(IV) in molten LiF-NaF-KF (46.5-11.5-42.0 mole %) at ~500°C were carried out using a graphite cell enclosed in a Pyrex jacket under a helium atmosphere. Platinum (A = 0.1 cm?) and pyrolytic graphite (PG) working electrodes, a large Pt wire counterelectrode, and an Ni wire quasi-reference electrode (QRE) were employed in the experiments. Titanium(IV) was added as K, TiF, (obtained from Harshaw Chemical Company or D. E. LaValle, ORNL). The cyclic voltammograms (scan rate = 0.1 V/sec) obtained prior to the addition of K, TiFq showed two small reduction waves at —0.20 and —0.48 V (vs Ni QRE) at both the Pt and PG electrodes {(probably due to iron and chromium impurities in the melt). In addition, a larger impurity wave (possibly the reduction. of OH™) was observed at the Pt electrode at —0.95 V (PG electrode cannot be used at negative potentials in LiF-NaF-KF due to predeposition of potassium). Addition of K, TiF¢ [Ti(IV) concentrations ~0.1 M] resulted in a large well-defined reduction wave (£t 2 =+0.05 V vs QRE) at both Pt and PG electrodes. The peak current (ip) for this reduction wave reached a maximum after 1 to 2 hr and then decreased to ~0.1 of maximum i, after ~24 hr. Further Ti(IV) additions restored the wave, although it decreased with time. At 1. Student participant, University of Tennessee, Knoxville. 2. Consultant, Department of Chemistry, University of Ten- nessee, Knoxville. the Pt electrode a second, larger and steeper, wave was . observed at —1.45 V. This wave was less reproducible 134 than the wave at +0.05 V. A stripping-type reoxidation peak clearly related to the wave at —1.45 V was observed at —0.80 V. The reduction wave at —1.45 V was reasonably stable over a four-day period. Estimates of the number of electrons, »n, involved in the two waves (from logarithmic plots as well as cathodic to anodic peak potential separation) gave n = | for the +0.05-V wave and n = 3 for the —1.45-V wave respectively. Parameter ip/v”2 values for the wave at +0.05 V were independent of the scan rate (up to 50 V/sec) and Ti(IV) concentrations <0.1 M, indicating no kinetic complications. The above experimental observations suggest that Ti(IV) is reduced reversibly to Ti(III), which in turn is reduced to the metal at a potential ~1.5 V more cathodic than that of reduction of Ti(IV). The disap- pearance of Ti(IV), but not Ti(III), with time suggests that Ti(IV) is reduced by graphite and/or the nickel QRE. The examination of the Ni QRE at the end of the experiment showed noticeable attack. A second experi- ment in a graphite container using a Pt QRE gave similar results. Again, Ti(IV) could not be kept in solution for more than one to two days. In this experiment all potentials were more cathodic by ~350 mV due to the shift in the QRE. Half-wave potentials for polarograms constructed from current-time curves for the two Ti waves were —0.30 V [Ti(1V) — Ti(III)] and —1.75 V [Ti(Ilf) — Ti(0)]. Additional evidence for n = 1 for the first wave was obtained from the peak current ratio for derivative vs normal voltammetry. The experimentally observed ratio of 0.500 is in very good agreement with the theoretically predicted value of 0.508 for n =1 and v= 0.1 V/sec at S00°C. ' In view of the difficulty encountered in maintaining a stable solution of Ti(IV) in a graphite cell, we con- ducted further experiments in a platinum cell using only platinum electrodes. In this cell, Ti(IV) was stable and could be kept in solution for several weeks. The Ti(IV) to Ti(IIT) voltammogram was well defined and reproducible. The peak current was found to be proportional to concentration of Ti(IV) and to the square root of the scan rate (up to ~10 V/sec). The Ti(IIT) to Ti(0) wave was preceded by what appears to be a predeposition wave, probably due to alloying of Ti with Pt. Presence of strong alloy formation is supported by the unusually large separation between anodic and cathodic peak potentials for this wave. The determina- tion of #, from the ratios of voltammetric peak currents and limiting currents for polarograms constructed from current-time curves, gave n,/n; of 3. The half-wave potentials for the two.waves (measured at a potential corresponding to 0.85ip) with respect to an Ni/NiF (sat) electrode contained in boron nitride were found to be +0.15 and —1.45 V respectively. Chronopotentiograms for the reduction of Ti(IV) to Ti(IIT) were reasonably well defined. The ir'/? product (r = transition time) was essentially constant in the current range 1 to 10 mA. The diffusion coefficient, calculated from Sand’s. equation, was 5.3 X 107 cm?/sec (at 480°C). Excellent verification of n = 1 for the first wave was achieved from the ratio of voltam- metric ip/v” 2 to chronopotentiometric ir'/* (for this determination of #, concentration, diffusion coef- ficient, and electrode area do not need to be known). Placement of a titanium electrode in the melt resulted in the expected conversion of Ti(IV) to Ti(Ill), as . evidenced by the disappearance of the first Ti reduction wave and a gross shift of the Pt QRE with respect to the Ni(IT)/Ni BN electrode (the QRE potential before the immersion of the Ti electrode was +585 mV; after the immersion it eventually became —515 mV). Voltammo- grams run with the Ti electrode as the reference [presumably poised to Ti(III)/Ti] showed that the oxidation of Ti(IIT) was occurring at about +1.5 V and the Ti(III) reduction at 0.0 V. We are now studying the behavior of titanium in Li, BeF, and NaBF, . 12.2 REFERENCE ELECTRODE STUDIES IN MOLTEN NaBF, D. L. Manning A stable and rugged reference electrode is needed for the in-line analysis of the MSR coolant salt. Providing a fixed potential reference for voltammetric scans (re- placing the quasi-reference electrode) will simplify the 135 measurement of corrosion products. Also for in-line measurement on the coolant salt technology loop, it can be used to obtain a “‘redox” potential of the coolant which can perhaps be correlated with corrosion rates or proton concentration in the coolant. Accord- ingly, a model of the Ni/NiF, (LaF,) electrode,® which has been found to serve as a practical reference for electroanalytical measurements-in fuel melts, was tested in molten NaBF, at ~400°C. _ Upon immersing the reference electrode in the melt and monitoring the emf vs a platinum electrode (assumed to be poised at the equilibrium potential of the melt), the platinum electrode measured approxi- mately —150 mV vs the Ni/NiF, (LaF,) reference. The -drift and unsteadiness of the emf, however, was much more pronounced in molten NaBF, than was experi- enced in LiF-BeF,-ZrF, melts.* A nickel electrode immersed in the melt measured ~ —175 mV vs the reference. Nickel fluoride (450 mg) was next added to the melt (80 g) to establish the Ni/NiF, redox potential. ~ Assuming the Ni/NiF, couple is the poised system in the melt inside the LaF; crystal, the magnitude of the emf between the reference electrode and the Ni electrode in the molten NaBF, should decrease as the NiF, dissolves. This was not observed. Within the scatter of the measurements, the emf of the nickel electrode and the platinum electrode did not signifi- cantly change. Voltammograms recorded at a pyrolytic graphite electrode did not reveal a reduction wave that could be attributed to the NiZ* + 2e — Ni® electrode reaction, and only a very small stripping wave for Ni°® - Ni** + 2e could be developed. This -suggests that the solubility of NiF, in NaBF, is very limited (<25 ppm). The apparent limited solubility of NiF, may in part explain the erratic-behavior of the reference electrode. - The Ni/NiF, couple was probably not the emf- determining system; consequently, the electrode mostly resembled a quasj-reference. At this time, an Ni/NiF, (LaF;) reference- with a saturated solution of NiF, inside the LaF; crystal for use in NaBF, melts does not appear promising. Further tests are planned, however, with an alternate model of the electrode® which makes use of a solid Ni/NiF, pressed pellet as the potential- determining couple. Other reference couples will also be tested in the LaF, electrode. ' 3. D. L. Manning and H. R. Bronstein, MSR Program Semiannu, Progr, Rep, Feb, 28, 1970, ORNL-4548, p. 184. 4, D. L, Manning and F. R, Clayton, MSR Program Semi- annu. Progr, Rep, Aug, 31, 1970, ORNL4622, p. 115. 12.3 ELECTROCHEMICAL-SPECTRAL STUDIES OF MOLTEN FLUORIDE SALT SOLUTIONS J.P.Young F.L.Whiting' Gleb Mamantov? The present phase of the electrochemical generation and spectral characterization of unusual oxidation states of solute species in molten fluoride salts has been concluded. This work has been summarized in the Ph.D. thesis of F. L. Whiting. The solvent system has been mainly, but not exclusively, LiF-NaF-KF. The study was carried out to survey the utility of this technique as a preparative and investigational tool. Our original assignment of the absorption spectrum of superoxide, 0,7, in molten LiF-NaF-KF has proved to be in error.’ Superoxide does exist in this melt; it exhibits an ultraviolet absorption peak at 234 nm and a Raman spectral peak at 1107 cm™'. The yellow solution obtained by the addition of NaQO, to the melt, however, is due to the oxidation of chromium impurities (<10 ppm) to CrO, %", It has been reported that O, dissolved in liquid NH, also forms a yellow solution with an absorption peak near 360 nm.® [t was not possible to confirm this; rather, the addition of NaQO, to liquid NH; yielded a colorless solution with an absorption peak at 252 nm. Other investigations have also indi- cated the error of the original liquid ammonia study.” In other results of this work, we found it possible to generate electrolytically Mn(III), Cu(Il), U(III), and CrO4* and to obtain their spectra in the melt. Although Co(II) could be oxidized electrolytically in the melts studied, the product, probably Co(IIl), was not stable under the experimental conditions of the study, and no spectrum could be obtained. A method was developed for removing dissolved chromium and other reducible metal ions to less than ppm amounts by treatment of molten LiF-NaF-KF with elemental sili- con. Futher, the generation and spectral measurement of CrO,?*", by the oxidation of Cr(Il) or (III) in the presence of excess oxide ion, can be used for ‘the spectral determination of chromium to less than 1 ppm in LiF-NaF-KF. Likewise, it would seem that the generation and spectral measurement of CrO,%” by the oxidation of excess chromium in the presence of oxide ion could be applied to a colorimetric oxide analysis in amenable fluoride melts. 5. J. P. Young, F. L. Whiting, and Gleb Mamantov, MSR Program Semiannu, Progr. Rep. Aug. 31, 1969, ORNL-4449, p. 159. 6. J. K. Thompson and J. Kleinberg, J. Amer. Chem, Soc. 73, 1243 (1951). 7. G. Czapski and B. Halperin, Israel J, Chem, 5, 185 (1967). 136 12.4- SPECTRAL STUDIES OF MOLTEN FLUORIDE SALTS J.P. Young As a result of the graduate research of F. L. Whiting (reported above) it was apparent that oxygenated solute species can indeed exist in fluoride melts, and the kinds of such species are perhaps much more numerous than had previously been expected. During this period, various spectral studies of these oxygenated species in molten fluoride media were carried out, One of the end results of the study of oxyanions culminated in a cooperative spectral study of the reaction of OH™ and OD™ with molten NaBF,. The results of this study are reported in another section of this report.® Because of the chromate (CrO,27) formation in LiF-NaF-KF, the generation and spectral studies of chromate in other melts were undertaken. It was found that CrO,*” dissolved in LiF-BeF, melts. The spectrum is altered from that seen in LiF-NaF-KF, and it is believed that the dissolved species is dichromate (Cry0,%). Again, reduced chromium can be oxidized, with MnF;, in the presence of oxide ion to Cr,0,*". Chromate dissolves in molten NaF-NaBF,, and the spectrum indicates Cr,0,%” is the resultant species. It has not been possible, however, to generate Cr,0,° in NaBF, by the oxidation of chromium in the presence of added oxides. The nonreactivity could result from several causes, such as insolubility of chromium fluoride or the absence of free oxide ion via the formation of BF,0". The ultimate sensitivity of any possible spectral determination of O*” in molten salts is limited by the molar absorptivity of the resultant oxidized oxyanion. The molar absorptivity of the most intense peak of CrO,; %", at 372 nm, in molten LiF-NaF-KF is approxi- mately 4000 M~ at 500°C. The molar absorptivity of the isoelectronic VO, 3™ at its peak maximum, 272 nm, should be near 8000 M~ ¢cm™; VO, 3 is soluble in molten NaF-NaBF, or LiF-NaF-KF and exhibits the expected spectrum. The oxidation of lower-valent vanadium to VO, has yet to be studied. The spectrum of nitrite (NO,") has been observed in molten NaBF,. Nitrite exhibits a peak at 350 nm; the sensitivity of this absorption is not accurately known, but the peak is sufficiently sensitive to detect NO,™ at concentrations less than 1000 ppm. If further consider- ation is given to the use of a nitrate-nitrite secondary 8. John B. Bates, J. P. Young, H. W, Kohn, M, M, Murray, and G. E. Boyd, sect. 9.5. coolant for the MSBR, the results of this spectral study of NO, can be applied to a possible determination of the secondary coolant in molten NaF-NaBF,. It was also observed that NaNO, dissolves in LiF-NaF-KF; the spectrum of NO, in this melt is quite similar to that seen in molten fluoroborate melts. It was observed in earlier studies of NaBF, melts that many -salts, such as some transition-metal fluorides, rare-earth fluorides, and actinide fluorides, did not possess a sufficiently high solubility to be detected spectrally. For this reason it was thought that analyt- ically, spectral measurements might not be of much value for determinations of solute species in this proposed coolant salt, Earlier spectral data, further- more, on what turned out to be impure NaF-NaBF, indicated that the melt absorbed strongly in the ultraviolet below 300 nm. If NaBF, is purified, depending on the method of purification, the salt is quite transparent even at wavelengths of 200 nm or less. Based on chemical analyses, this increase in trans- parency is caused by the removal of Fe(II) and/or Fe(II) from the salt. Obviously, then, Fe(II) or Fe(IIT) in ppm amounts could be spectrally determined by the ultraviolet absorbance of molten NaBF,. Other solute species could interfere; in this region of the spectrum one observes allowed transitions of many ions. The extent of such interference is being investi- gated. Presently, the extent of melt purification can be followed by absorbance measurements in the ultraviolet region of the spectrum below 300 nm, 12.5 ANALYTICAL STUDIES OF THE NaBF, COOLANT SALT J.M.Dale R.F.Apple A.S.Meyer In the previous report® it was noted that S. Cantor had raised questions concerning the meaning of the Karl Fischer (KF) determinations of “water” in samples of NaBF, coolant salt. We cited earlier analyses in which comparable results were obtained by the direct KF titration and by titration of the water separated by azeotropic distillation from coolant samples as evidence that these analyses represented some protonated species in the salt, It was found, however, that azeotropic distillations of current samples yielded titrations much lower than those obtained by direct titrations, and therefore a critical evaluation of the azeotropic method was made. 9. R. F. Apple et al., MSR Program Semiannu. Progr. Rep. Aug, 31, 1970, ORNL-4622, p. 116, 137 In the azeotropic procedure, about 450 ml of pyridine (1000 ppm maximum H, O) is placed in a glass still fitted with a Vigreux column. After a period of total reflux, about 50 ml of the distillate is discarded, and successive 3-ml test portions of the distillate are titrated with coulometrically generated KF reagent until a constant titration establishes a blank for the apparatus. The sample of salt is then added to the stiil pot through a standard taper joint, and titration of the distillate is continued until the titration drops to the original blank value, usually three to four test portions of distillate. : | It was noted that the blanks could contribute as much as several hundred micrograms to a sample titration (depending in part on the relative humidity in the laboratory), a value sometimes in excess of the net titration for .smaller samples. Also, the addition tech- nique was subject to some uncertainties. Although no apparent introduction of water was observed during simulated sample additions, the pulverized samples were briefly exposed to the atmosphere during an addition. We have modified the technique by eliminating all unnecessary glass joints, minimizing the length of rubber connectors, substituting a molecular sieve column for the Ascarite—magnesium perchlorate drier on the still vent, using manual rather than automatic termination of the titration current, and using pyridine reagent only from selected manufacturers’ lots. With these modifications the blank titrations were reduced to about 25 to 50 ug, at least a fivefold improvement. Also, a solid sample injector was developed to eliminate contamination during sample additions. This injector is basically a plunger with a recess that can be charged with approximately 1 g of salt in a dry box and sealed in a Teflon sleeve., The loaded assembly after weighing is in turn sealed in a metal standard taper joint which couples to the still pot. After the system is blanked the plunger can be depressed to inject the sample without exposure to the atmosphere. With this injector, about 95% of the water from a sodium tartrate dihydrate standard is recovered. With the improved procedure, negligible titrations have been obtained on current coolant samples. Samples taken from the FCL-1 thermal-convection loop are conservatively reported as <50 ppm vs 1000 ppm by direct titration. It should be noted that the blank titrations reported cannot explain the much higher azeotropic values (up to 2500 ppm) that were obtained on earlier samples, nor do the present low titrations completely eliminate the possible existence of protons in these samples. It will be necessary to perform azeotropic distillation on salt of known proton concen- tration (e.g., NaBF3;OH when prepared) to evaluate the meaning of these negative results. We are proposing to use the azeotropic technique for the determination of “water’ that is present as hydroly- sis products in NaBF, cover-gas streams, The sensitivity of the direct KF titration of these gases is limited by the high blank resulting from the reaction of BF; with the constituents of the reagent. In the proposed method the water and BF; would first be stripped from the “cover gas by sparging it through cold dried pyridine; subsequently, the water would be separated for titra- tion by azeotropic distillation, In a preliminary test of this approach, H. W. Kohn!® found that - negligible blanks. were obtained for gas streams that contained only BF,; but when HF (required for his reprocessing studies) was absorbed in the pyridine, an unacceptably high blank was titrated in the distillate. This is apparently caused by the reaction of pyridinium hydrofluoride with the glass apparatus. In KF titrations at room temperature, excess pyridine eliminates the interfering reactions of HF, but these reactions are obviously accelerated at the temperature of boiling pyridine. A metal azeotropic still has been designed and will be tested for this application. 12.6 IN-LINE CHEMICAL ANALYSIS J. M. Dale T. R. Mueller The project for making in-line measurements of U(IV)/U(III) ratios and corrosion products in MSR fuel salt is in the stage of testing of the individual components involved in the experiment. The thermal- convection loop, designed by members of the Metals and Ceramics Division, has been delivered to the X-10 site, and heaters are being installed prior to a salt flushing operation to ensure that the internal surfaces of the loop are clean. Five electrode ports have been provided in a test chamber of the loop. Three of these will be used for the electrodes for the voltammetric measurement of the U(IV)/U(III) ratios and corrosion product concentrations in the melt, and the other two 10. Reactor Chemistry Division, unpublished work. 138 will be used for a reference electrode system for following the redox potential of the melt. These measurements will be made during the studies to correlate the effect of melt potential on the corrosion of test specimens to be installed in the loop. ' A newly constructed cyclic voltammeter, which pro- vides several new capabilities for electrochemical studies on molten-salt systems, will be used in this work. A new plug-in module was developed to differentiate the current-potential voltammeter output used for measure- ment of U(IV)/U(III) ratios. The differentiator has 11 ranges for input signals of different rise time which are designed to achieve optimum signal-to-noise ratio with minimum wave-shape distortion. The new instrument also includes an integrating circuit, provisions for compensating that portion of the circuit resistance (mostly in the cell) not compensated by the potentio- stat, a preselect cyclic counter, a circuit layout that minimizes cross talk existing in the prototype instru- ment, a chopper-stabilized sweep generator for “drift- free”” operation in the “cyclic” or “hold” modes, and a potentiostat capable of delivering 500 mA of cell current (100 mA in prototype). The voltammeter, which can be directly operated by the PDP-8 family of computers, has a terminal that permits ground isolation between the computer and the experiment if this is needed. A PDP-8/1 digital computer purchased for application to in-line analytical methods will be used ‘in this study to allow unattended measurement of the U(IV)/U(III) ratio in the melt. A computer program’! developed by M. T. Kelley and R. W, Stelzner will be used in a modified form for this purpose. Previous work on this measurement was made in an electrically ungrounded cell. Before installation on the thermal loop, it is planned to check out the compatibility of the com- puter-voltammeter system on a melt in a grounded system. The experience derived from this study will be most valuable for application to computer-operated in-line measurements involving other types of chemical transducers. 11. M. T, Kelley, R. W. Stelzner, and D. L. Manning, MSR Program Semiannu, Progr. Rep. Aug. 31, 1969, ORNL-4449, p. 157. Part 4. Materials Development 13. Examination of MSRE Components " H. E. McCoy Operation of the MSRE was terminated after success- fully completing its mission. Further information can be gained by examining selected parts of the system. The examination of the coolant circuit was reported previously,! and work on the primary circuit has been partially completed during this reporting period. The components examined include a graphite moderator element from the core, a Hastelloy N control rod thimble, the sample assembly from the pump bowl, a lost copper sample tube, the shell and several tubes from the primary heat exchanger, and a freeze valve that failed as operation was terminated. All of these parts have been examined visually, and metallographic samples of most parts have been examined. Work is still in progress to define the concentrations of fission and corrosion products on many of the graphite and metal surfaces. This work is partially completed and will be reported at a later date. Concentration gradients near the metal surfaces that are indications of corrosion behavior are being evaluated by the electron probe microanalyzer and will be reported later. 13.1 EXAMINATION OF A GRAPHITE MODERATOR ELEMENT B.McNabb H. E. McCoy One of the five removable graphite core moderator blocks was examined in HRLEL after completion of the operation of the MSRE. Visual examination revealed machining marks still plainly visible in the fuel channels and on surfaces not exposed directly to flowing fuel salt, indicating little if any change during operation. Figure 13.1 is a photograph of the fuel channel facing the center of the reactor showing the excellent condi- tion of the core block. The small thumbnail flaw in the fuel channel about 39 in. from the top is believed to be 1. H. E. McCoy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 119. a small crack that was in the bar during fabrication, since tool marks are on the top of the flaw and skip over an area at the end of the flaw. The few scratches on the surface are believed to be caused by handling and are not very deep. There was a crack in the graphite around the Hastelloy N lifting stud at the top of the bar that extended about halfway around the bar, as shown by the short arrows in Fig. 13.2. The long arrow indicates the same channel in the different views. The small punch mark visible in the top view was made to indicate the fuel channel toward the center of the reactor, with a control rod on one side and the surveillance sample basket on the other side. The crack in the graphite would not affect the integrity of the bar, since it was on the end and several threads remained to secure the lifting stud. The crack was probably caused by the approximate factor of 2 difference in thermal expansion of graphite and Hastel- loy N. ' Measurements were made of length, width, and bowing. There were no significant changes since all measurements were within the original tolerance for fabrication of the core block. The measured length of the approximately rectangular portion of the bar was 64Y, + Y, in. Thus, there was no measurable change in length. Width measurements across the flat parallel surfaces were made using 1-to-2-in. micrometers and read through the Kollmorgen. The original dimensional tolerances were 1.998%0-99% i Measurements shown in Table 13.1 were made at ~1-ft intervals from top to bottom of the side facing the center of the reactor. Measurements were made near the top, middle, and bottom at 90° to this on the side that had 0.075 in. relief for flow between the channels starting 2 in. from either end, top = 1.9226 in., middle = 1.9231 in., and bottom = 1.923 in. All measurements were within tolerance. No actual measurements of the individual bars were recorded before operation, but assuming the 139 PHOTO 0370-71 Fig. 13.1. Photograph of one of the removable graphite moderator elements after operation in the MSRE, showing the fuel channel facing the center of the reactor. Machining marks are plainly visible in the fuel channel, showing the excellent condition of the core block. (U2} 141 PHOTO 0371~ 74 Fig. 13.2. Photograph of the top of the graphite moderator element, showing the crack originating at the INOR-8 lifting stud and extending partway around the graphite block shown by the short arrows. The long arrow indicates the same channel in the different views. The crack was probably caused by the approximate factor of 2 difference in thermal expansion between the INOR-8 and the graphite during thermal cycling of the reactor. maximum and minimum of the fabrication tolerances there were no significant changes or observable trends (Table 13.1). Measurement of bowing was made by stretching a string tightly between two C-clamp-type fixtures at either end of the bar and measuring the distance between the bar and the string with a cathetometer, then rotating the bar 90° and sighting on another bar surface and measuring again. The maximum observed bowing was 0.012 in. with the string being closer to the bar in the center, possibly indicating slight sagging of the string. It was concluded that there was no appreci- able bowing. No tolerances for bowing were in the specifications, and the bar was not supported on a flat Table 13.1. Dimensional measurements of graphite moderator element? Change from original Measured width dimensions, in inches, Locaiea (in.) assuming — Maximum size Minimum size Top 1.9947 ~0.0033 0.0017 1t 1.9954 ~0.0025 0.0024 2t 1.9944 ~0.0036 0.0014 36t 1.9948 ~0.0032 0.0018 4ft 1.9942 ~0.0038 0.0012 Bottom 1.9971 ~0.0009 0.0041 +0.000 @Dimensional tolerances were 1,998 in. '~0.005 PHOTO Fig. 13.3. Comparison of the bottom positioning stud on the graphite moderator element when initially installed in the MSRE and after operation. The large picture was made before the reactor was operational, and the small picture in the lower right corner was made in the hot cells after the moderator element had been at temperature for over three years. (44" 143 PHOTO 0372-71 ETCHED 5008 Fig. 13.4. Hastelloy N thread that was in contact with graphite for about three yearsat 650°C. The modified layer is likely due to working effects from machining the thread. Reduced 70.5%. surface; so absolute measurements were not made or considered necessary since the bar removed easily. Photographs were taken of the assembled graphite core blocks before inserting in the reactor core. Figure 13.3 shows the bottom of core block 1184 C19 in the assembled position, and the inset shows the bottom of the same core block as it appeared in HRLEL after operation for the full life of MSRE. There is no apparent change, as the small cracks and the small chip missing from the bottom are apparent in both photo- graphs and the numbers are just as legible as before operation. Core-drilled samples were taken from the fuel flow channel near the midplane and used as electrodes in the mass spectrograph, but the results have not been analyzed. Core-drilled samples were taken at the top, middle, and bottom of the same fuel channel for Auger analysis of the surface layers for fission product deposition, but the results have not been received yet. A cross section of the Hastelloy N lifting stud in contact with the graphite is shown in Fig. 13.4. There appears to be little if any reaction with the graphite. There is a surface modified layer about 1 or 2 mils deep similar to that observed on surveillance samples and attributed to cold working of the surface and subse- quent heat treatment during service. There appears to be very little difference in the side of the thread in contact with the graphite and the root of the thread, which probably was not in intimate contact. 13.2 AUGER ANALYSIS OF THE SURFACE LAYER ON GRAPHITE REMOVED FROM THE CORE OF THE MSRE R. E. Clausing The detection and quantitative analysis of thin layers of fission products deposited on the graphite and metal surfaces of the MSRE is necessary to predict the requirements for afterheat removal. Auger electron spectroscopy is a new technique which offers unique capabilities in this respect. It provides, under suitable conditions, excellent sensitivity (down to 0.1% of a monolayer) and resolution down to one to three atom layers.”> The technique is applicable to all elements 2. L. A. Harris, /. Appl. Phys. 39, 1419 (1968). 3. L. A. Harris, J. Appl. Phys. 39, 1428 (1968). except hydrogen and helium and, when combined with sputter thinning, can provide profiles of concentration as a function of depth below the original surface. 13.2.1 Auger Electron Spectroscopy If a vacancy is created in an inner electron shell of an atom by electron bombardment or any other method, the excited atom will revert to the ground state by emitting characteristic x radiation or alternatively by emitting an electron as the result of a radiationless transition called an Auger transition. This Auger elec- tron has an energy which is determined by the energy levels in the parent atom. Figure 13.5 shows schemat- ically the Auger process for the creation of the KL; L4 (2144 eV) and L,VV (~149 eV) Auger electrons characteristic of sulfur. The energy of the Auger electron is obtained by subtracting the energy of the two final vacancies from the energy of the initial vacancy. The most useful Auger electrons have energies between 10 and 3000 eV, and when they originate more than a few atom layers below the surface they cannot escape from electrically conducting solids with- out energy losses. This effect produces the excellent depth resolution. ORNL-DWG 71-6723 h KVV AUGER KL3 ELECTRON AUGER ELECTRON —] } VALENCE ZZ@ZZ%ZZ ELECTRON VIS My ! ‘\ ; LEVELS L3 > = 164 Lo e oA~ // K [ VACANCY CREATION SULPHUR (Z=16) 2472 Fig. 13.5. Energy level diagram showing schematically the emission of KL 3L 3 and L, V'V Auger electrons from sulfur. 144 Special techniques are required to detect Auger electrons stimulated by electron bombardment of solids because the Auger electrons represent only a small signal superimposed on a large continuous back- ground.?’? Figure 13.6 shows how this small signal can be recovered from an energy distribution curve through differentiation. The differentiated curve is the one normally reported. The differentiated curve is usually obtained using the instrumentation shown in Fig. 13.7. Electrons from the target are energy analyzed using a retarding potential applied to the second and third grids of a set of three grids spaced between the target and a hemispherical electron collector. The retarding potential is modulated by superimposing an ac¢ voltage (K sin wt) on the retarding voltage E£,. The amplitude of the ac compo- nent (K''sin cwr) in the collector current is proportional to the number of electrons in the energy range between Ey + %K and £, — LK. A plot of K’ as a function of Ey then gives an energy distribution curve. The amplitude of the ac component (K" sin 2wt) is ORNL-DWG 7t-6724 I CURVE A I I = Lo REGION 251! REGION G| - = | | REGION = f ‘ .3 | ! | | | T 1 J1J 1 I | i I i I 'V\’ ] A ! | i P! | 3 s | AUGER PEAKS APPEAR N THESE REGIONS | ------------------------------------------------------ | [ ] [ o I : | CURVE B I p-d i b O N } q - | -5 | 8 z @ ¥ A= ! w2 | w - c% N o ! u JY ! <= W Q l [ }v\v | [ | 0 50 100 1500 V (eV) ELECTRON ENERGY SPECTRA 2000 2500 Fig. 13.6. Method of enhancing the visibility of Auger spectra. 145 ? TARGET ORNL- DWG 71 - 6725 RETARDING GRIDS = 210V ¥ I(E) . Kk sin w? 6 LOCK—IN-AMP ISOLATION %’0 Q — TRANSFORMER il = = (1:1 RATIO) - Y INPUT X-Y RECORDER 0-1500 V dec VOLTAGE SUPPLY | X INPUT (AUTOMATICALLY SCANNED) Fig. 13.7. Instrumentation for Auger electron energy analysis using the retarding-potential techniquel proportional to the derivative of the energy distribution curve.® The lock-in amplifier output, which is propor- tional to these amplitudes, can thus be used together with a signal proportional to £, to plot directly either an energy distribution curve or its differential by tuning it to the fundamental of the modulation frequency or its second harmonic respectively. 13.2.2 Results and Discussion An apparatus has been built specifically for the Auger analysis of radioactive samples from the MSRE, and preliminary results are reported below. Figure 13.8 shows the apparatus. In addition to the basic equipment for Auger analysis, an ion gun is provided for sputter thinning the sample, and provision is made to heat the sample by electron bombardment. A movable shield is 4. P. W. Palmberg and T. N. Rhodin, J. Appl. Phys. 39,2425 (1968). _provided to minimize the contamination of the appara- tus. : Figure 139 shows a curve obtained from MSRE sample 6. This sample is from a core surveillance specimen removed in April 1968, after run 14. A number of Auger peaks are present. They have been tentatively identified as indicating the presence of carbon, oxygen, sulfur, molybdenum, niobium, tellur- ium, technetium, uranium, lithium, and beryllium. Contributions of these elements to the spectra are - indicated in the figure caption. There is some overlap of spectra which complicates the interpretation; however, the use of better resolution for intense peaks and increased sensitivity for less intense peaks will minimize this problem. Table 13.2 shows how several of the peaks changed during sputter thinning. These peak heights are propor- tional to the concentration of the parent atoms at the surface of the sample. Since no suitable standards were available, no absolute concentrations can be given yet. Additional samples and standards are now available, and quantitative information will be available soon. f 146 ORNL~ DWG 70-6794 RETARDING POTENTIAL GRIDS S COLLECTOR «& ‘_ § ELECTROSTATIC SHIELD SAMPLE MOVABLE SPUTTER SHIELD FILAMENT FOR STATIONARY ELECTRON BEAM SPUTTER SHIELD HEATING {ON GUN FOR SPUTTER CLEANING Fig. 13.8. Apparatus for Auger analysis of low-level radioactive samples. ORNL-DWG 70-6801R ] | .y Vad \“*MET;M 510 300 : Wjim I r/ [ 430 "‘-‘\_‘\ "‘-u-h_ QEGV\/ L37 240\ i e N | 222 E.B. CLEANED 3 / / /HOA/|<200 2 min, 7mA, 3.4 kY N . 800°C © / { /( I/\182 MODULATION 10 VPP 120 | ATTENUATION AS MARKED . | 270 160 I / fM4B 0 100 200 300 400 500 600 700 ELECTRON ENERGY (eV) " Fig. 13.9. Auger electron spectra from MSRE graphite sample 6 after heating to 800°C for 5 min by electron bombardment. The following elements have been tentatively identified as making contributions to the peaks: § 148; 0 510, 490; C 270, 253, 240; Mo 222, 185, 160, 149, 137, 118; Nb 200, 160, 92; Tc 250, 220, 180, 145; Te 490, 410, 278; U 92 (86), (72}, (60); Li (39); Be 92. Values in parentheses are from a low-energy scan not shown. ' 147 Table 13.2. Auger electron intensity as a function of depth below the original surface of sample MSRE 6 (film side) Normalized to carbon (270) equal 100 Accumulated Estimated Technetium, argon ion depth - sulfur, ;Zi;g;g:;?; Niobium bombardment (atom molybdenum (200 eV) (X 10'5/ecm?) layers) (148 eV) (182 ¢V) 0 29 9 .2 5 1 9 712 4 13 6 16 3 23 5 7 19 3 = 58 10 5 23 4 136 25 4 16 3 280 55 4 12 3 540 110 3 12 2 1240 250 2 10 1886 380 2 9 3464 700 nd? 9 3570 710 2 14 3700 740 2 9 6764 1350 8 11,907 2400 nd? 9hd = not detected. 13.3 EXAMINATION OF HASTELLOY N CONTROL ROD THIMBLE B.McNabb H. E. McCoy Control rod thimble 3 was cut off just above the midplane of the reactor core. Figure 13.10 shows the electric arc cut made at the left and enlarged views of the spacer sleeves. Some areas show the possibility of some wetting of the Hastelloy N by the fuel salt near the center of the reactor, where the témperature and flux were higher, as shown around the left and center spacer sleeves, but not at the bottom, as shown on the right sleeve. Some of the variations in light and dark areas are due to not obtaining even lighting for the pan photographs made in the hot cells. The control rod thimble was fabricated from Hastelloy N heat Y8487 with chemical composition of Ni—70.80%, Mo— 16.78%, Cr—7.32%, Fe—4.1%, Mn—0.3%, C—-0.05%, Si—0.17%, S—0.0075%, P—-0.004%, Cu—0.03%, Co— 0.1%, Al-0.16%, Ti—0.25%, B—0.007%; the melt was made and fabricated in 2-in.-OD tubing by the Inter- national Nickel Company. Tensile tests were conducted on some Y-in.-wide rings cut from the 2-in. control rod thimble, using the fixture shown in Fig. 13.11. The circular ring is placed in the fixture when it is closed, and the two halves are pulled apart in the tensile test. Most of the strain occurs in the two sides between the grips, but the gage length changes during the test and makes it impossible to define elongation as a percent. Therefore crosshead travel and the reduction in area at the fracture are reported. Table 13.3 compares the properties of the unirradiated and irradiated samples tested in the same manner. It shows that there is little change in the yield and ultimate strengths. The room-temperature reduc- tion in area is reduced from about 45% in the as-received, unirradiated condition to 28% for the irradiated rings. Some of the ductility at 25°C can be recovered by annealing at 871°C for 8 hr but is not improved further by annealing 141 hr at this tempera- “ture. The elevated-temperature reduction in area was reduced more severely but still was 9.7% at a fairly slow deformation rate of 0.002 in./min. Similar changes were noted in the properties of the surveillance samples reported previously. The control rod thimbles were exposed to fuel salt and fission product deposition only on the exterior of the tube, and the irradiated rings only exhibited fissuring or cracking on the exterior of the ring, as shown in Fig. 13.12. The ID of the tube was slightly 148 R-54116 Fig. 13.10. 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. Table 13.3. Tensile data on rings from control rod thimble 3 (heat Y-8487) Specien : Bosticradiatios Test Crosshead Yield Ultimate Crosshead Reduction in No. Condition ot temperature speed stress stress travel area “©0) (in./min) (psi) (psi) (in.) (%) 5 Unirradiated 650 0.05 39,100 76,700 0.37 28.7 6 Unirradiated 25 0.05 52,000 114,400 1.20 445 T Unirradiated 25 0.05 56,900 117,500 L17 46.0 8 Unirradiated 25 0.05 58,000 124,300 118 43.0 9 Unirradiated 650 0.002 40,700 62,300 0.20 20.6 2 Irradiated None 25 0.05 54400 105,100 0.54 23.2 3 Irradiated None 25 0.05 53,300 102,500 0.51 29.7 4 Irradiated None 25 0.05 60,500 110,800 0.42 285 6 Irradiated None 650 0.05 38,300 51,200 0.099 121 7 Irradiated None 650 0.002 34200 38200 0.061 9.7 5 Irradiated 8 hr at 871°C 25 0.05 49,300 100,500 0.36 34.9 8 Irradiated 141 hr at 871°C 25 0.05 48,700 104,900 0.39 344 “Based on 0.002 in. offset of crosshead travel. 149 oxidized by the reactor cell atmosphere, and the oxide cracked during testing, but the cracks did not propagate - into the metal wall. Figure 13.13 shows a cross section of the control rod thimble and spacer sleeve near the midplane of the core. The cross section was made through the button weld that kept the spacer sleeve from sliding up or down the thimble. The OD of the sleeve was exposed to flowing salt and the ID to salt in a 0.005-in. clearance or crevice between the sleeve and the thimble. The OD of the thimble at this point was exposed to the same conditions as the ID of the sleeve, and the ID of the thimble was exposed to the control rod thimble cooling atmosphere. The spacer sleeve was made of heat 5060, and the weld bead was made with heat 5090 weld wire, both air-melted heats of standard Hastelloy N. There appears to be very little attack on any of the materials, with only the 2- to 4-mil surface modification similar to that noted previously. Surface layers of the thimble material are presently being dissolved for fission product analysis at the midplane and bottom of the core, but results have not been received. Fig. 13.11. Fixture used for pulling Hastelloy N rings cut from the control rod thimble. R-54166 0.07 INCHES T 50x o 5 Fig. 13.12. Hastelloy N ring from control rod thimble stressed at 25°C. The cracked surface was exposed to the fuel salt, and the oxidized surface was exposed to the cell environment of 2 to 5% O, and AS POLISHED AS POLISHED 100X ETCHED 100X 150 FHOTO 0373-7 [FrT— ETCHED QD OF SLEEVE 500X ETCHED D OF SLEEVE 500X E ETCHED. 0D OF THIMBLE 500X ETCHED 1D OF THIMBLE 500X Fig. 13.13. Cross section of Hastelloy N control rod, spacing sleeve, and weld bead. Reduced 72%. Figure 13.14 is a photomicrograph of the cross section of the tube wall, near the midplane of the core showing the OD in the as-polished condition, showing a grain boundary network by reliéf polishing, 4 or 5 mils deep. Figure 13.15 is the same area in the etched condition showing the modified etching characteristics of the surface layers. 13.4 EXAMINATION OF THE SAMPLER ASSEMBLY B.McNabb H. E. McCoy One of the Y-in.-diameter rods from the sampler enricher cage reported elsewhere® was sectioned for metallography and fission product deposition analysis. 5. E. L. Compere and S. S. Kirslis, MSR Program Monthly Progress Report December 1970 and January 1971, MSR-71-13, p. 16 (internal memorandum). A piece was cut from the bottom of the rod in the liquid or fuel salt region of the pump bowl and one at the top in the vapor or gas region above the liquid level Figure 13.16 shows a photomicrograph of the rod above the liquid level or in the vapor region of the pump bowl. There is very little if any attack in this area, either as polished or etched. Figure 13.17 is a photomicrograph of the bottom of the rod in the liquid region. There was quite a bit of deposited material on the lower end of the rod in the liquid region, but there was little attack on the rod as shown in the micrograph, unless it was even removal of all constituents of the surface. Electron microprobe examination and electro- Iytic dissolution of the surface layers are planned for one of the rods, and the electrolytic dissolution of the specimen from the mist shield in the liquid and vapor regions is planned for determination of fission products and the deposited material on the rods. 151 Shaas o 0.007 INCHES 500X o ™ Fig. 13.14. Photomicrograph of the cross section of the Hastelloy N control rod thimble showing the outside diameter in the as-polished condition. \ 0.007 INCHES T 500% © ~ - a - \ -~ ¥ ls ! A i Kf e i) S , Sl | fi 1 g4 ; °s - \\ { { F I xPe co@ D> O - \\ o il y . e \. Fig. 13.15. Photomicrograph of the cross section of the Hastelloy N control rod thimble showing the outside diameter in the etched condition. Etchant: aqua regia. 0.035 INCHES T> 100X (a) o010 ™. 0.035 INCHES T 100X foo30m. Fig. 13.16. Photomicrographs of a Hastelloy N rod from the sampler assembly located above the normal salt level. (a) As polished: (b) etched with aqua regia 153 (a) too0 . 0.035 INCHES 100X T to030 . Fig. 13.17. Photomicrographs of a Hastelloy N rod from the sampler assembly. Normally located below the salt level. (a) As polished: (b) etched with aqua regia. 13.5 EXAMINATION OF A COPPER SAMPLE CAPSULE B. McNabb H. E. McCoy A copper sample bucket that was probably in the sampler cage of the pump bowl from August 1967 to January 1971 was examined. It had been flattened, probably during retrieval attempts in April 1968, when heavy magnets were used with no cushioning effects of salt and with the pump bowl at high temperature. Figure 13.18 is a photomacrograph of a cracked area of the sample capsule showing some deposited material on the outside of the capsule. Figure 13.19 is a photo- micrograph of the cracked area polished down below the surface, showing the intergranular nature of the cracks and the extent of them. The capsule was made of OFHC copper and hydrogen fired 2 hr at 1200°F before use. Figure 13.20 is a photomicrograph of a cross section of the copper capsule, showing the thick deposit of material on both the OD and ID of the capsule. An electron probe microanal grain boundaries and x-ray scans for the distribution of nickel, iron, chromium, and copper in the surface layer sis of the material in the 154 Fig. 13.18. Photograph of the crack formed in the copper sampler that was retrieved from the pump bowl. The outer surface was coated with deposited material. AW U R-54647 < g N )| Fig. 13.19. Section of copper capsule showing extensive intergranular cracking. 155 PHOTO 1183-71 in the area around the large metallic particle on the OD were made. Figure 13.21 shows the distribution of these elements, and it appears that the surface layers are made up mostly of copper and nickel with some areas containing high concentrations of chromium and mo- lybdenum. The ID appears similar to the OD of the capsule. The grain boundary precipitates were only 28% copper, and no other element could be detected. This observation indicates that the remainder is a light element that could not be detected by the instrument. Beryllium and lithium are possibilities for the unidenti- fied element. The surface deposits were mechanically ground away e 3 2 S from two small pieces about %4 in. square by % in. ' : / == long for electrodes for the mass spectrograph. We do - ; -~ not know the exact depth of sampling, but suspect that it was near the surface. The analysis of these is shown in _ Table 13.4. Several important points can be made from S5 this analysis. The spectrograph was not adjusted to detect lithium, but the presence of uranium, zirconium, e beryllium, and fluorine indicates that the fuel salt was present. More specifically, the isotopic concentrations of U correspond to those of the first fuel charge rather than the later fuel charge, where the uranium was predominantly 2**U. The weight percents of various elements present in the initial fuel charge were approxi- 4 mately 10.9 Zr, 6.3 Be, 67 F, 5.1 U,and 10.8 Li. Using : U as a base, Be should be higher by a factorof 2.1,and F higher by a factor of 13. The analytical values in Table 13.4 show the factors based on U to be 0.7 for Be, 6.6 for Zr, and 0.7 for F. Thus, using U as a base, the material in the Cu capsule is enriched in Zr and Table 13.4. Composition of copper capsule Bement Coneentration || L Concentration (ppm) (ppm) As 1 Si 70 Be 1000 sn ~20 Ca ~5 5t - Cr 30 Te