OCKHEED MARTIN ENERGY RESEARCH UBRARIES ORNL-5078 AT et 3 445k D445252 | MOLTEN-SALT REACTOR PROGRAM Semiannual Progress Repont Peniod Ending August 31,197 This document has been reviewed and is determined to be APPROVED FOR PUBLIC RELEASE. Name/Title: Leesa Laymance, ORNL TIO Date: 7/27/2017 OAK RIDGE NATIONAL LABORATORY CENTRAL RESEARCH LIBRARY DOCUMENT COLLECTION LIBRARY LOAN COPY DO NOT TRANSFER TO ANOTHER PERSON If you wish someone else to see this document, send in name with document and the library will arrange a loan OAK RIDGE NATIONAL LABORATORY OPERATED BY UNION CARBIDE CORPORATION FOR THE ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION Printed in the United States of America. Available from National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road, Springfield, Virginia 22161 Price: Printed Copy $8.50; Microfiche $2.25 This report was prepared as an account of work sponsored by the United States Government. Neither the United States nor the Energy Research and Development Administration, nor any of their employees, nor any of their contractors, subcontractors, or their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. ORNL-5078 UC-76 — Molten-Salt Reactor Technology Contract No. W-7405-eng-26 MOLTEN-SALT REACTOR PROGRAM SEMIANNUAL PROGRESS REPORT FOR PERIOD ENDING AUGUST 31, 1975 L. E. McNeese Program Director FEBRUARY 1976 OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37830 operated by UNION CARBIDE CORPORATION , for the ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION i 3 445k O4y5R52 b This report is one of a series of periodic reports that describe the progress of the program. Other reports issued in this series are listed below. 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 ORNL4254 ORNL-4344 ORNL-4396 ORNL4449 ORNL4548 ORNL-4622 ORNL4676 ORNL-4728 ORNL4782 ORNL-4832 ORNL-5011 ORNL-5047 Period Ending January 31, 1958 Period Ending October 31, 1958 Period Ending January 31, 1959 Period Ending April 30, 1959 Period Ending July 31, 1959 Period Ending October 31, 1959 -Periods Ending January 31 and April 30, 1960 Period Ending July 31, 1960 Period Ending February 28, 1961 Period Ending August 31, 1961 Period Ending February 28, 1962 Period Ending August 31, 1962 Period Ending January 31, 1963 Period Ending July 31, 1963 Period Ending January 31, 1964 Period Ending July 31, 1964 Period Ending February 28, 1965 Period Ending August 31, 1965 Period Ending February 28, 1966 Period Ending August 31, 1966 Period Ending February 28, 1967 Period Ending August 31, 1967 Period Ending February 29,1968 Period Ending August 31, 1968 Period Ending February 28, 1969 Period Ending August 31, 1969 Period Ending February 28, 1970 Period Ending August 31, 1970 Period Ending February 28, 1971 Period Ending August 31, 1971 Period Ending February 29, 1972 Period Ending August 31,1972 Period Ending August 31,1974 Period Ending February 28, 1975 Contents SUMM A RY . . e e PART 1 — MSBR DESIGN AND DEVELOPMENT 1. SYSTEMS AND ANALY SIS . ..o e e e e e 2 - 1.1 Tritium Behavior in Molten Salt-Systems . . ... . i e .. 2 1.1.1 MSBR Calculations . . ... ... e e 2 '1.1.2 Coolant Salt Technology Facility .. ....................... .. .. oo i 3 1.2 Xenon Behavior in MSBR . . ... . e 8 1.3 Neutronic ANalysis . ... ..o et e it e e e e et e e e e e e e e e 9 1.3.1 MSBR Studies .......... ..t e 9 1.32 TeGen Capsules .. ... ... e 12 1.4 High-Temperature Design Methods . .. ... ... .. . . 12 2. SYSTEMS AND COMPONENTS DEVELOPMENT ... ... ... . i 16 2.1 Gas-Systems Technology Facility . ........ .. i e 16 2.1.1 Cavitation and Salt-Pump Shaft Oscillations . . . . ... ... ... .. . 16 2.1.2 Salt-Pump Performance Data and Calibration of Variable-Flow Restrictors , ............... 18 2.1.3 Salt-Pump Fountain Flow .. ... . ... ... . ... . . . . . . . e 19 2.1.4 Densitometer Studies .............. ... ... .. ... .. P 22 2.2 Coolant-Salt Technology Facility (CSTE)} ... tvnn e et e et 22 2.2.1 LoopOperation ...................... e e e e 22 2.2.2 Salt Mist Test . ..ot e e 23 2.2.3 Tritium ExXperiments . .. .. ... e e 24 2.3 Forced Convection LOOPS ... ...t i e 26 2.3.1 Operation of MSR-FCL-2b ............ SR 26 2.3.2 Design and Construction of FCL-3 and FCL4 . ...... ... ... ... ... . . . .. ... .. . ..., 27 PART 2. CHEMISTRY 3. FUEL SALT CHEMISTRY ... ... e i [ ... 29 3.1 Compounds in the Lithium-Tellurium System .............. [ 29 3.2 Spectroscopy of Tellurium Species in Molten Salts . .......... ... .. ... ... . .. .. 30 3.3 The Uranium Tetrafluoride-Hydrogen Equilibrium in Molten Fluoride Solutions . ............ ... 31 3.4 Porous Electrode Studies in Molten Salts . ......... ... .. . . . 32 3.5 Fuel Salt-Coolant Salt Interaction Studies ...................... R L 3.6 Lattice and Formation Enthalpies of First-Row Transition Metal Fluorides . .................. 37 iii iv 4. COOLANT SALT CHEMISTRY . ... . e e e R 41 4.1 Chemistry of Sodium Fluoroborate ... ....... ... ... i i e 41 4.2 Corrosion of Structural Alloys by Fluoroborates .. ......... . .. .. . e 42 5. DEVELOPMENT AND EVALUATION OF ANALYTICALMETHODS . .......... .. ... .. ... .. 44 5.1 In-ine Analysis of Molten MSBR FUEL . . . .« .+ oo e oo e e e 44 5.2' Tritium Addition Experirfients in the Coolant-Salt Technology Facility ........................ 45 5.3 Electroanalytical Studies of Iron(II) in Molten LiF-BeF,-ThF, (72-16-12 MO0LE TB) . . o v oo e ettt e e e e e e e 47 5.4 Voltammetric Studies of Tellurium in Molten LiF-BeF, -ThF, (72-16-12 MO0LE ZB) v oo oottt e e e e e e e 48 PART 3. MATERIALS DEVELOPMENT 6. DEVELOPMENT OF MODIFIED HASTELLOY N .......... .. ... ... ... . ... e 52 6.1 Development of a Molten-Salt Test Facility ........... . .. . . . i 52 6.2 Procurement and Fabrication of Experimental Alloys .................. P e 65 6.2.1 Production Heats of 2% Ti-Modified Hastelloy N . . .. ... ... . . i e 65 6.2.2 Semiproduction Heats of 2% Ti-Modified Hastelloy N Containing Niobium .. ... .. ... ... . . e 69 6.3 Weldability of Commercial Alloys of Modified Hastelloy N . .................. e 69 6.4 Stability of Various Modified Hastelloy N Alloys in the Unirradiated Condition ............. ... .. . ... e h e aaee e iaaaeiaaan 74 6.5 Mechanical Properties of Titanium-Modified Hastelloy N Alloys in the Unirradiated Condition .. ........ ... ... ... . ... ..... e e 78 6.6 Postirradiation Creep Properties of Modified Hastelloy N .. ... ... . ... . .. . .. . . .. ... 82 6.7 Microstructural Analysis of Titanium-Modified Hastelloy N. . .. .. e T 84 6.7.1 Microstructural Analysis of Alloys 503 and 114 .. ... ... . ... ... ... .. P 85 6.7.2 Homogeneous Hastelloy N Alloys . ........ ... ... ... ... ... ........ e 88 6.8 Salt Corrosion Studies . .. . . P P P e R 91 6.8.1 Fuel Salt Thermal Convection Loops . .................... e 93 6.8.2 Fuel Salt Forced Circulation Loop ....... ... ... ........... IR T 94 6.8.3 Coolant Salt Thermal Convection Loops . ... ... ... .. . .. 94 6.9 Corrosion of Hastelloy N and Other AlloysinSteam .......... ... ... ... ... ... .. ... ... ..... 97 6.10 Observations of Reactions in Metal-Tellurium-Salt Systems . . . ... ..... e 100 6.11 Operation of Metal-Tellurium-Salt Systems ......... e T 101 6.11.1 Tellurium Experimental Pot Number 1 .. ... .. ... . . . . . 101 6.11.2 Chromium Telluride Solubility Experiment . ............ e 102 6.11.3 Tellurium Experimental Pot Number 2 .. ... ... .. .. . . . . . .. . . . . e 103 6.12 Grain Boundary Embrittlement of Hastelloy N by Tellurium .. .. ....... . ........ B 103 6.13 X-Ray Identification of Reaction Products of Hastelloy N o ' Exposed to Tellurium-Containing Environments . .................. P AP 107 6.14 Metallographic Examination of Samples Exposed to Tellurium-Containing Environments . . . ... .. ... e e e 108 6.15 Examination of TeGen-1 ... .. . . . . e e 119 6.15.1 Metallographic Observations ... . ... . ...t 123 6.15.2 Chemical Analyses for Tellurium . ... ... ... .. . . e 124 6.16 Salt Preparation and Fuel Pin Filling for TEGen-2and -3 .. ......... .. ... . ... .. oot 131 7. FUEL PROCESSING MATERIALS DEVELOPMENT . ... .. . . 132 7.1 Static Capsule Tests of Graphite with Blsrnuth and Bismuth-Lithium Solutions ...................... e e e e 132 7.2 Thermal Gradient Mass Transfer Test of Graphite inaMolybdenum Loop .. ... i e e 133 7.2.1 WeightChanges ........... ...t .. e e 133 7.2.2 Compositional Changes ......... ... ... ... ..l e 133 7.2.3 Microstructural Changes . . . ... ... . e e 137 7.2.4 Discussion of Results . . ... .. .. e 137 PART 4. FUEL PROCESSING FOR MOLTEN-SALT REACTORS 8. ENGINEERING DEVELOPMENT OF PROCESSING OPERATIONS . ......... ... . ... ... . ... 142 8.1 Metal Transfer Process Development ........................... e 142 8.1.1 Addition of Salt and Bismuth Phases to Metal Transfer ' Experiment MTE-3B ........... e 143 812 RunNd-1 ....... ... . ... .. . ... ... ... ... e e e 144 8.1.3 Run Nd-Z . . e e 145 8.1.4 Discussion of ReSUILS . .. ..t e e e e e e 145 8.2 Salt-Bismuth Contactor Development .......... ... ... . ... . i 147 8.2.1 Experiments with a Mechanically Agitated Nondlspersmg Contactor in the Salt-Bismuth Flowthrough Facility ........... ... .. ... . . .. . 148 8.2.2 Experiments with a Mechanically Agitated Nondispersing Contactor Using Water and Mercury ... ... ... i e ... 149 8.3 Continuous Fluorinator Development .. ... ... ... .. . ... . . . i 152 8.3.1 Installation and Initial Operation of Autoresistance Heating _ Test AHT -4 . .. e 152 8.3.2 Design of a Continuous Fluorinator Experiment Facility (CFEF) ....................... 155 8.3.3 Fluorine Disposal System for Bldg. 7503 . ... ..o\t tirtitit ettt e e 156 - 8.3.4 Frozen Wall Corrosion Protection Demonstration ........... e 156 8.4 Fuel Reconstitution Engineering Development .. ....... ... .. .. .. . . i 157 8.4.1 Instrumentation for Analyzing Reaction Vessel Off-Gases ... .................... ... ... 158 8.4.2 Design of the Second Fuel Reconstitution Engineering ' Experiment ........ .. ... e 160 8.5 Conceptual Design of a Molten-Salt Breeder Reactor Fuel Processing Engineering Center .. ...... ... o 161 PART 5. SALT PRODUCTION 9. PRODUCTION OF FLUORIDE SALT MIXTURES FOR MSR PROGRAM RESEARCH AND DEVELOPMENT e 163 9.1 Quantities of Salt Produced .. .. ... .. 163 9.2 Operating Experience in 12-in.-diam Reactor ......... .. .. ... . .. . il 163 9.2.1 Charging and Melting of Raw Materials . ......... .. ... .. ... ... . i, 164 9.2.2 Hydrofluorination and Hydrogen Reduction ......... ... ... ... ... .. ... 166 0.3 SUMMATY . oottt ettt e e e e e it ieee e e e 166 ORGANIZATION CHART . .ot e e e e e e e e e et 167 Summary PART 1. MSBR DESIGN AND DEVELOPMENT J. R. Engel 1. Systems and Analysis Calculations of the expected tritium behavior in the reference-design MSBR were continued with studies of An updated neutronics model of the 1000-MW(e) reference-design MSBR is being developed. Multi- dimensional, multigroup calculations will use the VEN- TURE code, with neutron cross-section data derived entirely from the ENDF-IV libraries. Processing of the cross-section data was completed for 38 of 39 nuclides ~at four temperatures of interest for the planned calcula- the possible effects of oxide films on heat exchange - surfaces in the steam system and on surfaces exposed to the containment atmosphere. The presence of oxide films with very low permeability on the heat transfer surfaces would significantly reduce the rate of tritium migration to the steam system because of the increasing importance of the oxide-film resistance at very low par- tial pressures of hydrogen and tritium. However, the reduction from this effect alone would be insufficient to limit the rate of tritium migration to the steam system to desired values. At high rates of tritium trans- port to the steam system, the presence of oxide-film resistances on loop walls tends to increase the rate of tritium flow into the steam. However, this effect is insignificant at the low migration rates required. Potential distributions of tritium in the Coolant-Salt Technology Facility ‘were estimated for the conditions of planned experiments. In the absence of tritium inter- action with the salt, other than simple dissolution, as much as 99% of the added tritium could be expected to escape through the loop walls. Removal of significant fractions in the loop off-gas could be expected only if the effective permeability of the loop walls were 10 to 100 times less than that of bare metal. Substantial chemical interaction of tritium with NaBF,-NaF was observed in the two tritium addition tests performed. Ratios of combined-to-elemental trit- tions. Cross-section data are also being examined for the two-step thermal reaction *®*Ni(n,y)*°Ni(n,a)®°Fe, . which is expected to be the principal source of helium ium in the salt, inferred from elemental concentrations in the off-gas and combined concentrations in the salt, ~ were 50 and 530 for the two tests. Approximately ' to Y. of the added tritium was removed in the off-gas stream, principally in a chemically combined, water- soluble form. vii in MSBR structural metals. _ A review of the data and calculations used to estimate tellurium inventories in the TeGen-1 experiment indi- cates an uncertainty of *20%. Work is continuing on the study of thermal ratch- etting and creep fatigue in reactor structural materials. Analytical methods are being developed which will be applied to the reference-design MSBR to evaluate the significance of these processes in Hastelloy N. 2. Systems and Components Development - The Gas-Systems Technology Facility was operated with water throughout the report period. Efforts to reduce the amplitude of the salt-pump shaft oscillations have been unsuccessful. The amplitude of these oscilla- tions is largely dependent upon shaft speed, so a larger- diameter impeller, which will give the design flow and head at lower speeds, is being fabricated. A method was developed for estimating the pump fountain flow. Since this flow was highier than desirable, back vanes will be used on the new impeller to limit the flow. Tests made at the loop indicate that the densitometer can be used to determine bubble-separator efficiencies if short-term tests are used. Routine operation of the Coolant-Salt Technology Facility was established with more than 2500 hr of salt circulation without plugging in the loop off-gas line. Measurements of the amount of salt mist in the off-gas stream showed 100 to 500 ng/cm?® (STP), depending on the salt temperature and the BF; flow rate into the loop gas space. The mist trap installed in the salt cold trap was effective in preventing the plugging that had been experienced earlier. Two tritium injection tests were conducted, in which 85 and 97 mCi, respectively, of tritiated hydrogen were added to the loop during two 10-hr periods. Frequent salt and off-gas samples were taken to monitor the tritium behavior in the loop. The forced-convection loop, MSR-FCL-2b, has accu- mulated 3000 hr of operation with MSBR reference fuel salt at design AT conditions with the expected low cor- rosion rates. Data obtained on the heat transfer charac- teristics of this salt are being analyzed. The design is essentially complete for forced-convection loops FCL-3 and FCL-4. Components are being fabricated and elec- trical installation is proceeding. PART 2. CHEMISTRY 3. Fuel-Salt Chemistry Relatively pure Li, Te (about 99% on a mole basis) was prepared by the controlled addition of tellurium to liquid lithium. The reaction was begun at 250°C, but ultimately temperatures greater than 500°C were re- quired to complete the reaction. LiTe; was prepared by reacting the stoichiometric amounts of Li, Te and tellu- rium for 2 hr at 550°C. Apparatus for the spectroscopic study of tellurium species in MSBR fuel salt has been assembled. Prelimi- nary work with lithium tellurides in chloride melts has shown that at least two light-absorbing species are present with compositions in the range Li, Te to LiTe,. Furthermore, studies with Te, in LiCl-KCl eutectic have shown that, in addition to Te,, a second species is present at high temperatures and/or high halide ion activity. : - ' Apparatus for the spectrophotometric study of the equilibrium UF4(d) + %4H,(g) = UF;3(d) + HF(g) has been assembled, and measurements using Li,BeF, as the solvent have begun. A preliminary value of about 107% was obtained for the equilibrium quotient at 650°C. This value is in good agreement with the value obtained previously by other workers. Development proceeded on porous and packed-bed electrode systems as continuous, on-line monitors of concentrations of electroactive species in molten salt solutions. The packed-bed electrode of glassy carbon spheres was calibrated using Cd** ions in LiCl-KCl eutectic before experiments were conducted with Bi** ions in solution. The results of the experiments demon- strated the capability of the electrode for monitoring these and other ions. viii Preliminary experiments were conducted to evaluate- some questions relating to the mixing of NaBF,-NaF coolant salt with MSBR fuel salt, LiF-BeF,-ThF, -UF, (72-16-11.7-0.3 mole %). The results showed that the rate of evolution of BF; gas on mixing was low. Mixing of small amounts of coolant salt with fuel salt did not result in the precipitation of uranium- or thorium- containing compounds. No data were obtained on the mixing of small amounts of fuel salt with coolant salt. Results of experiments in which a small amount of cool- ant salt containing oxide was mixed with fuel salt sug- gested that oxide species more stable than UQ,. were present, since no precipitation of UQ, was observed. A study of lattice enthalpies of first-row transition- metal fluorides was undertaken to provide a theoretical basis for evaluating thermochemical data for structural metal fluorides -being obtained from solid-electrolyte galvanic cells. Ligand-field corrections to plots of lattice enthalpy vs atomic number for the two series CaF, to ZnF, and ScF; to GaF; indicated that the standard enthalpies of formation (AH7 )of NiF, and VF; were satisfactory, but that more accurate experimental values of AHf for TiF;, VF,, CrF,, CrF;, FeF,, and FeF, would be desirable. 4. Coolant-Salt Chemistry Af;alyses of samples of condensate collected during operation of the Coolant-Salt. Technology Facility indicate that the vapor above the salt is not a single molecular compound but rather a mixture of simple gaseous species such as H, O, HF, and BF,. The conden- sate showed a tritium concentration ratio of about 10° relative to the salt. This result suggests a possible method for concentrating and collecting tritium in an MSBR. Related work showed that NaBF; OH dissolved in coolant salt undergoes a reaction that reduces the OH™ concentration in the salt, producing a volatile frac- tion. Physical and chemical observations were made on the system NaF-NaBF,-B,0; at 400 to 600°C. Work with- compositions typical of the usual coolant salt (oxide concentrations up to 1000 ppm) showed that at least two oxygen-containing species are present. One species is Nay;B3F¢O;; the other has not yet been identified. . Studies were continued to determine the extent to - which borides were formed in Hastelloy N and Inconel 600 by reaction with NaBF,-NaF at 640°C. Data ob- tained thus far indicate some formation of chromium and nickel borides; however, after four months of ex- posure of the alloy samples to salt the boride concentra- tion on the metal surfaces did not exceed 500 ppm in Hastelloy N and 1000 ppm in Inconel 600. The results also showed that chromium in these alloys was selec- tively oxidized by the salt. 5. Development and Evaluation of Analytical Methods During this period U*/U* ratios were monitored by voltammetric techniques in two thermal-convection loops and one forced-circulation loop. Stable redox con- ditions continue to exist in thermal-convection loops 21A and 23;the U*/U? ratio is approximately 7.6 X 10° and 5 respectively. In forced-convection l'oop FCL-2b, the U*/U* ratio is about 80. No attempts have yet been made to reoxidize the U*" in the melt by the addition of nickel fluoride or some other oxidant. The results from the first series of tritium addition experiments at the Coolant-Salt Technology Facility show that very little tritium exists in the off-gas in the elemental state; the bulk of the tritium occursin a com- bined or water-soluble form. It appears that about 50% of the injected tritium experienced significant holdup in the salt and was eventually removed in the system off- gas stream. It was observed that the Fe®* — Fe® electrode reac- tion in molten LiF-BeF,-ThF, (72-16-12 mole %) closely approximates the soluble product case at a gold electrode, the insoluble product case at pyrolytic graph- ite, and, depending on the temperature, both soluble and insoluble product cases at an iridium electrode. Voltammetric measurements were made in molten LiF-BeF,-ThF, following additions of Li,Te in an effort to identify soluble electroactive tellurium species in the melt. No voltammetric evidence of such com- pounds was obtained. These observations were in general agreement with chemical analysis that indicated <5 ppm Te in the salt. PART 3. MATERIALS DEVELOPMENT 6. Development of Modified Hastelioy N Work ‘is partially complete on the molten-salt test facility to be used mostly for mechanical property test- ing. Much of the test equipment is operational. All products except the seamless tubing of the 2% Ti—modified Hastelloy N were received. The first heat weighed 10,000 1b and had a fairly narrow working tem- perature range. The second heat weighed 8000 lb and had a wider working temperature range. Seamless tubing is being fabricated by two vendors. Weldability studies ix on these two heats showed that their welding charac- teristics were equivalent to those of standard Hastelloy N and that existing welding procedures for standard Hastelloy N could be used for the 2% Ti—modified alloy. The mechanical properties of Hastelloy N modified with titanium, niobium, and aluminum were evaluated in the irradiated and unirradiated conditions. These properties were used to estimate the individual and combined concentrations of titanium, niobium, and aluminum- required to produce brittle intermetallic phases. The .formation of brittle phases in the alloys containing niobium was enhanced by an applied stress. Specimens of modified Hastelloy N were exposed to tellurium from several different sources. The partial pressure of tellurium above Cr;Te, at 700°C seems reasonably close to that anticipated for MSBRs. Metal- lographic examination of the exposed specimens after straining revealed that alloys containing 0.5 to 1% Nb were resistant to intergranular cracking by tellurium. Further analysis of the data from TeGen-1 showed that most of the tellurium in each fuel pin was concen- trated on the tube wall. The concentration in the salt was 1 ppm or less. The salt has been prepared for filling the fuel pins in TeGen-2 and-3, and the pins for TeGen-2 have been assembled for filling. 7. Fuel Processing Materials Development Experiments were continued to evaluate graphite as a material for fuel processing applications. The penetra- tion of graphite by bismuth-ithium solutions was found to increase with increasing lithium concentration of the solution and pore diameter of the graphite. Decreasing the pore diameter of the graphite by pitch impregnation decreased the average depth of penetration. However, because the structure of the graphite was variable, greater-than-average penetration occurred in regions of low density. A thermal-convection loop constructed of molyb- denum contained ATJ graphite specimens in hot- and cold-leg regions and circulated Bi—2.4 wt % (42 at. %) Li for 3000 hr at 700°C maximum temperature, with a temperature differential of 100°C. Very large weight increases (30 to 67%) occurred in all of the graphite: samples, primarily as a result of bismuth intrusion into the open porosity of the graphite. Dissimilar-metal mass transfer between molybdenum and graphite was also noted. These results and previous capsule test results suggest that the presence of molybdenum enhances intrusion of bismuth-lithium solutions into graphite. Thin carbon layers were noted on the molybdenum. PART 4. FUEL PROCESSING FOR MOLTEN-SALT REACTORS 8. Engineering Development of Processing Operations Addition of the salt and bismuth solutions to the process vessels in metal transfer experiment MTE-3B was completed. Two experiments were performed to measure the removal rate and overall mass transfer coefficients of neodymium. In the first run about 13% of the neodymium originally added to the fuel salt (72-16-12 mole % LiF-BeF,-ThF,) in the fuel-salt reser- voir was removed during the 100 hr of continuous operation. Overall mass transfer coefficients for neo- dymium across the three salt-bismuth interfaces were lower than predicted by literature correlations, but were comparable to results seen in experiment MTE-3. For the first 60 hr of the second experiment, which was a repeat of the first experiment, the rate of removal of neodymium was similar. The second run was termi- nated because of unexpected entrainment of the fuel salt into the lithium chloride in the contactor, which resulted in depletion of the lithium from the Bi-Li solu- tion in the stripper and stopped further neodymium transfer. Future experiments in MTE-3B will depend on deter- mining the reason for the unexpected entrainment of fluoride salt into the lithium chloride, and it will be necessary to remove and replace the lithium chloride that is presently contaminated with fluoride salt. A hydrodynamic run intended to determine the effect of increased agitator speed on the extent of entrainment of one phase into the other in the salt-bismuth con-. tactor was performed. No visual evidence of gross en- trainment was found. Analytical results indicate that the bismuth concentration in the fluoride salt phase decreased with increasing agitator speed. This un- expected result is probably due to sample contamina- tion. Development work continued on an electrochemical technique for measuring electrolyte film mass transfer coefficients in a nondispersing mechanically agitated contactor, using an aqueous electrolyte solution and mercury to simulate molten salt and bismuth. During this report period experiments with Fe®-Fe®* were made with improved experimental apparatus. A stan- dard calomel electrode which enables measurement of the mercury surface potential was obtained. Electronic filters were attached to the inputs on the x,y plotter to damp out noise in the signal to the plotter. Near the end of the report period, a potentiostat was obtained which will automate the scan procedure now performed with the dc power supply. Copper, iron, and gold anodes have been tested. The gold anode is the most satisfac- tory choice, since it does not react with the electrolyte solution. By noting that the active anode area in the cell could be decreased with no resulting change in the dif- fusion current, it was determined that the mercury cathode rather than the gold anode is polarized. Results indicate that the ferric iron is being reduced by some contaminant in the system. Further tests with purified mercury and electrolytes in the absence of oxygen indi- cate that the contaminant was present in the mercury. Analytical results for Fe®" and Fe®* concentrations in the electrolyte phase are inconsistent with expected results. Qualitative results indicate that a buffered quinone-hydroquinone system may be useful as an alter- nate to the Fe**-Fe?* system. Installation of autoresistance heating test AHT-4, in which molten salt will be circulated through an autoresistance-heated test vessel in the presence of a frozen-salt film, was completed and operation was begun. A conceptual design was made of a continuous fluorinator experimental facility for the demonstration of fluorination in a vessel protected by a frozen-salt film. Design was completed and installation was begun on a fluorine disposal system in Building 7503, using a vertical scrubber with a circulating KOH solution. In- stallation was completed of equipment to demonstrate the effectiveness of a frozen-salt film as protection against fluorine corrosion in a molten-salt system. Off-gas streams from the reaction vessels in the fuel reconstitution engineering experiments will be con- tinuously analyzed with Gow-Mac gas density detectors. To determine whether hydrogen back-diffusion in the cell body will be a problem during the analysis of the HF-H, mixture from the hydrogenation column, the cell was calibrated with N,-H, mixtures. It was found that when the reference gas flow rate to the cell is suffi- ciently high, the effect of hydrogen back-diffusion is not seen. The second engineering experiment will be conducted in equipment which is either gold plated or gold lined to eliminate or minimize effects resulting from equipment corrosion. Several alternatives for gold lining or gold plating are discussed. The factors which must be considered in deciding between lining or plating are listed. A design is being prepared to define the scope, esti- mated design and construction costs, method of accom- plishment, and schedules for a proposed Molten-Salt Breeder Reactor Fuel Processing Engineering Center. The proposed building will provide space for prepara- tion and purification of salt mixtures, for engineering experiments up to the scale required for a 1000-MW(e) MSBR, and for laboratories, maintenance areas, and offices. The estimated cost of the facility is $15,000,000; authorization will be proposed for FY 1978. PART 5. SALT PRODUCTION 9. Production of Fluoride Salt Mixtures for Research and Development Activities during the report period fall in three categories: (1) salt production, (2) facility and equip- xi ment maintenance and modification, and (3) peripheral areas that include preparation of transfer vessels and assistance to others in equipment cleanup. Salt produced in this period, totaling about 600 kg, was delivered in more than 30 different containers. About one-half of the salt was produced in an. 8-in.- diam purification vessel and had acceptable purity levels. The remaining salt was produced in the 12-in.- diam purification vessel during five runs, each of which involved about 150 kg of salt. Part 1. MSBR Design and Development J. R. Engel The overall objective of MSBR design and develop- ment activities is to evolve a conceptual design for an MSBR with adequately demonstrated performance, safety, and economic characteristics that will make it attractive for commercial power generation and to de- velop the associated reactor and safety technology re- quired for the detailed design, construction, and opera- tion of such a system. Since it is likely that commercial systems will be preceded by one or more intermediate- scale test and demonstration reactors, these activities include the conceptual design and technology develop- ment associated with the intermediate systems. Although no system design work is in progress, the ORNL reference conceptual design' is being used as a basis to further evaluate the technical -characteristics and performance of large molten-salt systems. Calcula- tions are being made to characterize the behavior and distribution of tritium in a large system and to identify potential methods for limiting tritium release to the environment. These analytic studies are closely corre- lated with the experimental work in engineering-scale facilities. Studies were started, in this reporting period, to reexamine the expected behavior of xenon in an MSBR. This work will ultimately use information from experiments in the Gas-Systems Technology Facilitiy (GSTF) to further refine !'®° Xe-poisoning projections and to help define the requirements for MSBR core graphite. ' 1. Molten-Salt Reactor Program Stéff, Conceptual Design Study of a Single-Fluid Molten-Salt Breeder Reactor, ORNL- 4541 (June 1971). Additional core neutronics calculations are being made for the reference MSBR, using widely accepted, evaluated nuclear data and a two-dimensional computa- tional model. These calculations will provide updated estimates of the nuclear performance, as well as addi- tional information on core characteristics. Analogous methods and data are employed to provide support for in-reactor irradiation work. The GSTF is an engineering-scale loop to be used in the development of gas injection and gas stripping tech- nology for molten-salt systems and for the study of xenon and tritium behavior and heat transfer in MSBR fuel salt. The facilitiy is being operated with water to measure loop and pump characteristics that will be re- quired for the performance and analysis of develop- mental tests with fuel salt. The Coolant-Salt Technology Facility is being oper- ated routinely to study processes involving the MSBR reference-design coolant salt, NaBF,-NaF eutectic. Tests are in progress to evaluate the distribution and behavior of tritium in this system. Candidate MSBR structural materials are exposed to fuel salt at reference-design temperatures and tempera- ture differences (704°C maximum and 139°C AT) and representative salt velocities in forced-convection loops to evaluate corrosion effects under various chemical conditions. These operations, which are principally in support of the materials development effort, also pro- vide experience in the operation of molten-salt systems and data on the physical and chemical characteristics of the salt. One loop, MSR-FCL-2b, which is made of standard Hastelloy N, is in routine operation; two others, to be made of titanium-modified Hastelloy N, are under construction. L Systems and Analysis J. R. Engel 1.1 TRITIUM BEHAVIOR IN MOLTEN-SALT SYSTEMS Studies to elucidate the behavior of tritium in large molten-salt systems were continued in this reporting period. Additional calculations were made for the 1000-MW(e) reference-design MSBR to examine the effects that an oxide film on metal surfaces might have on the distribution of tritium. Analysis of the informa- tion being generated by the tritium addition experi- ments in the Coolant-Salt Technology Facility (CSTF) was begun. As additional data and results are developed, they will be incorporated into the MSBR studies. 1.1.1 MSBR Calculations G.T. Mays Calculations were performed to examine the potential effects on tritium transport to the steam system caused by the formation of oxide films on the steam side of the tubes in the steam raising equipment of an MSBR. The rate of diffusion of hydrogen (tritium) through metal oxides typically is proportional to the first power of the hydrogen partial pressure in the gas phase, as opposed to the % power for diffusion through metals (i.e., the diffusion process is molecular rather than atomic). In addition, at moderate hydrogen partial pressures, the permeability coefficients of the oxides may be as low or lower than those of pure metals. Thus, at the very low hydrogen partial pressures that would be expected in an MSBR, oxide films could offer substantial resistance to hydrogen (tritium) permeation. However, the efficiency of such films would be limited by the degree of metal surface coverage that could be established and main- tained during operation of the system. The computational model! for studying tritium behavior at steady state provides for variation of the metal permeability coefficients of the steam-system tubes, but assumes that diffusion through the tube walls varies only with the % power of hydrogen partial pres- sure. Variations in metal permeability were considered in previously reported results.> However, the model also includes the effect of a mass transfer coefficient for tritium transport through a salt film inside the tubes. Since transport through the salt film depends upon the first power of tritium concentration (or partial pres- sure), this value was used to estimate the effects of oxide films. Effective mass transfer coefficients were computed which included the resistances of the oxide films as well as those of the salt films.? Tritium distribution calculations were made for a variety of situations in which it was assumed that the effective permeabilities of the oxide coatings in the steam system were 1, 107!, 1072, and 1073 times those of the bare metal at a hydrogen partial pressure of 1 torr (130 Pa). These results were compared with cases without oxide coatings in which the permeabilities of the bare metal were reduced by factors of 1, 10, 102, and 10*. The comparisons were made at three values of the U** /U ratio (10%, 10%, and 10%) and, in all cases, sorption of hydrogen or HF on core graphite was as- sumed to be negligible. The results (Table 1.1) indicate that a low- permeability oxide coating would be more effective than a low permeability in the metal itself for limiting tritium transport to the steam system. When an oxide film resistance equal to that of the metal was added, the rate of tritium transport to the steam system was ap- proximately halved, as would be expected. (The total resistance to tritium transport was not doubled because of the contribution from the salt film.) The results with a factor of 10* reduction in a steam-tube permeability due to oxide formation indicate that tritium transport to the steam system could be limited to the design objective of 2 Ci/day. However, it may be unreasonable to expect to obtain and maintain oxide films of this quality in an operating system. 2 Additional calculations were performed to investigate the effect of reduced permeability of the primary and secondary loop walls through the formation of oxide coatings. These coatings can be expected to form in a manner similar to those expected on the steam equip- ment. For a given steam-tube permeability, reducing the permeabilities of the loop walls would be expected to increase the amount of tritium transported to the steam system. With the reduced loop-wall permeabilities, less 1. R. B. Briggs, A Method for Calculating the Steady-State Distribution of Tritium in g Molten-Salt Breeder Reactor Plant, ORNL-TM-4804 (April 1975). 2. G. T. Mays, in MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 31 2. 3. Although this calculational approach assumes that the oxide film is located inside the tubes rather than outside, it can be shown that, for given oxide and metal permeabilities, this arrangement slightly overestimates the rate of hydrogen permea- tion through the wall. Table 1.1. Effect of oxide films on tritium transport to the steam system of an MSBRZ Rate of * H migration to steam systemn (Ci/day) Ratio of oxide or metal permeability U**/U3* to neminal metal ratio Oxid~ film Reduced metal permeability‘b inside tubes® permeabilityd 1 10° 811 1425 1 10° 656 1169 1 10¢ 115 203 107! 10° 173 1351 107! 10° 138 1114 107! 10° 23 198 1072 10? 19 662 1072 10° 16 575 10_2 104 3 142 1073 107 2 93 1073 10° 1.5 84 1073 10¢ <1 31 INo sorption of H, or HF on core graphite. bAt a hydrogen partial pressure of 1 torr. CWith nominal metal permeability. 9No oxide film. tritium would permeate through the loop walls into the primary and secondary system containments, elimina- ting a potential sink for tritium. A higher tritium con- centration (or partial pressure) in the secondary system would result, creating an increased driving force for tritium transport to the steam system. The results of the calculations did indicate that, with- out the presence of a chemical getter in the secondary coolant, more tritium was transported to the steam system when the primary- and secondary-loop wall per- meabilities were reduced than in the same cases with reference permeabilities for loop walls. However, more importantly, for those cases where tritium transport to the steam system had been reduced to the design-limit objective of 2 Ci/day through chemical additions of H,, HF, or a chemical getter, results showed essentially no increase over the 2-Ci/day rate. Thus, it appears that reduced loop-wall permeability has little effect on trit- ium transport for cases where a tritium exchange mate- rial is present in the secondary coolant. 1.1.2 Coolant-Salt Technology Facility . J.R.Engel G.T.Mays A 1000-MW(e) MSBR is expected to generate about 2420 Ci of tritium per full-power day. Calculations showed* that, unless a major fraction of this tritium were converted to a chemical form less mobile than elemental HT, the rate of migration of tritium through the metal walls of heat exchange surfaces to the steam system could be unacceptably high. The purpose of the tritium addition experiments in the CSTF is to simulate the general conditions in the MSBR coolant-salt system to determine the extent to which tritium can be held up in the NaBF,-NaF salt. There is evidence that hydrogen- containing compounds in the salt may retain significant amounts of tritium. In the experiments, the first two in a planned series, tritiated hydrogen was diffused into the circulating salt through the walls of a hollow Hastelloy N tube. The tritium could accumulate in the salt, pass into the off- gas system, or permeate through the metal walls of the loop to the ventilated loop enclosure. Tritium concen- trations were monitored in the salt and in the loop off- gas. In the first two experiments, 85 and 97 mCi of tritium diffused through the Hastelloy N injection tube. For a detailed description of the experimental condi- tions, see Sect. 2.2. The computer program® for calculating the expected tritium distribution in a 1000-MW(e) MSBR was modi- fied to describe the CSTF and was used to calculate potential tritium distributions for the experiments under the following assumptions: 1. steady-state conditions, 2. only dissolution of elemental tritium (hydrogen) in the salt with no chemical reaction with the salt or any of its components, 3. all transport through metal walls varies as the % power of hydrogen partial pressure. Calculations were made for addition rates of tritiated hydrogen equivalent to those achieved in the CSTF ex- periments, using assumed loop-wall permeabilities rang- ing from the value expected for bare metal to 1072 of that value. The results (Table 1.2) show a significant effect of loop wall permeability on the fraction of the added material that could escape through the walls. This table also shows the calculated steady-state concentra- tions of elemental tritium in the salt (nCi/g) and in the off-gas (pCi/cm?®) in the same units that are being used in reporting experimentally observed concentrations. Also shown are the inventories of elemental tritium in the loop walls that would be associated with the calcu- lated transport rates through the walls. Since these cal- 4. G. T. Mays, in MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 3—-12. 5. R. B. Briggs and C. W. Nestor, A Method for Calculating the Steady-State Distribution of Tritium n a Molten-Salt Breeder Reactor Plant, ORNL-TM-4804 (April 1975). Table 1.2. Calculated steady-state tritium distributions in CSTF for experimental addition rates Addition rate Loop w.a‘ll Time to reach Fraction of addition Elementz.ll tr;tfl-um Eler.n.ental Tritium Tritiated permeability 85% of steady- te which removal in off-gas tritium . _ ) % of steady rate which permeates . inventory mixture (fraction of state conditions? loop walls Fraction of addition concentration . otal walls hydrogen Tritium bare-metal " (hr) %) rate removed Concentration in salt (i) (em®/hr) (mCi/hr) value) (%) (pCi/em?) {nCi/g) 3.18 7.9 1 0.3 99.4 0.6 400 1.3 0.1 107! 9.5 77.1 22.9 15000 50 0.8 1072 38 14.9 85.1 55000 200 1.6 1072 42 1.6 98.4 64000 220 1.7 3.3¢ 9.3 1 0.3 99.4 0.6 500 1.7 0.13 : 107! 10.3 75.7 243 19000 64 1.0 1072 37 14.3 85.7 67000 230 1.9 10-3 42 1.5 98.5 77000 260 2.0 2Salt and off-gas concentrations only, longer times required for steady-state permeation through toop walls, by tritium in hydrogen. ' €0.11% tritium in hydrogen. culations represent steady-state conditions, and the trit- ium addition experiments involve transients, it is useful to consider the time required to reach the steady state. At high loop-wall permeabilities, the concentrations of elemental tritium in the salt and off-gas are low and tend to reach steady values quickly for the assumed addition rates (see Table 1.2). Somewhat longer times are required to reach the higher concentrations associ- ated with lower assumed loop-wall permeabilities. In all cases substantially longer times are required to reach steady-state rates of tritium release through the loop walls. However, this has little effect on the ultimate steady-state levels in the salt and off-gas. Figures 1.1 and 1.2 show the results of tritium con- centration measurements in the salt and off-gas from the CSTF during the first and second tritium addition ORNL-DWG 75-42897 104 — 7 l ] I —] s é o SALT (nCi/g) _ - é ® OFF-GAS, WATER SOLUBLE (pCi/em3) | — B g A OFF-GAS, ELEMENTAL (pCi/cm3) _ 2 — / o . . —_ é 103 ? — . = n h— — § - TRITIUM ADDITION Y — c 2 ¢ — Q 5 7 o 102 é > [— @ h— S — A ° — 5 5 : / A 4 ® ° @ L : > [ ° o S 4l o : / NE - - 2 —. / o —_ > 7 o 3 ? N g 401 Y o O b - — (& - ® o) — v ‘ — 2 5 — = — _ & — ™ T A A ] 2 (o) A A — o o o o 10° :_‘% 4 = —4 | . = ST | % - 2 - | ® — 10-! 4 15 17 19 21 23 25 27 29 34| 4 JULY AUG DATE , 1975 Fig. 1.1. Observed tritium concentrations in CSTF, test 1. 10° TRITIUM CONCENTRATION (nCi/g or pCi/cm>) c-Sf\) 10 ot ORNL-DWG 75-12896 — |V = = T T T T T T 712 — o SALT (nCi/g) ] | ® OFF-GAS, WATER SOLUBLE (pCi/m®) — B A OFF-GAS, ELEMENTAL (pCi/cm?) — — | Z® — = . = — g . = A | _ \ / . o | o /) ® — ¢ — — . = — L—TRITIUM ADDITION ] | ® . — — . — e p - e o h— A O A S— o) — e (o] — — A _ — A A — — o] — N £ ° ] A A I o — / 0 0 = | ‘ ° = — ‘% A O— / |7 — % — - / — 7 Z 5 7 9 1 13 15 17 19 X DATE, AUGUST, 1975 Fig. 1.2. Observed tritium concentrations in CSTF, test 2. tests. The concentrations reported for the salt represent tritium in a chemically combined form, since any ele- mental HT trapped in the samples would have been re- leased in preparing them for scintillation counting. The tritium in the off-gas was present in two distinctly dif- ferent chemical forms. Part of the tritium activity was present in a water-soluble form, implying a chemical compound, since HT does not interact significantly with water at room temperature. The other form is presumed to be elemental HT, since it was trapped in water after passage of the sample stream through a bed of hot CuO. In all cases the results are presented in Figs. 1.1 and 1.2 as reported, with no corrections for apparent base- line concentrations. However, the results of samples taken before and after each test suggest that nonzero baseline concentrations were present. The apparent baseline concentrations for the two experiments were: _ Ist experiment 2d experiment In salt 1.7 nCi/g 1 nCilg In off-gas, elemental 1 pCi/cm3 1 pCi/cm3 In off-gas, water soluble 1 pCi/cm3 50 pCi/cm3 During the tritium addition period for the first experi- ment the tritium concentration in the salt (Fig. 1.1, open circles) increased, almost linearly, to a maximum of about 100 nCi/g, and then decreased approximately exponentially over the following 3 to 4 days to its pre- test baseline concentration. In the second experiment (Fig. 1.2, open circles) the tritium concentration in the salt reached a maximum of 70 nCi/g and returned to the baseline concentration about 6 to 7 days later. If base- line corrections are applied to the salt-sample data, the apparent half-lives for tritium removal from the salt for the two experiments are 9.2 and 12 hr respectively. ' If tritium removal from the salt is assumed to be a pure first-order process or combination of such proc- esses, and if it is assumed that the processes were also active during the addition period, the buildup of the tritium inventory in the salt (for a constant addition rate) should be described by N =Ly A where N(t) = tritium inventory in salt at any time ¢ during . the addition, - A = tritium addition rate, X =time constant for the removal process (or proc- esses). If several first-order processes were involved in :the trit- ium removal from the salt, the time constant A would be the sum of the several individual time constants, but the individual values would not be identifiable. Substi- tution of the actual addition rate into this equation gives -the expected tritium inventory in the salt at any time, if all .of the tritium were reacting with the salt. Conversely, substitution of observed inventory values permits evaluation of the effective addition rate (the rate at which tritium did react with the salt). In either case the results may be expressed as tritium trapping efficiencies with values of 85 and 50%, respectively, for the two experiments. These trapping efficiencies imply that significant quantities of the added material were reacting with and being trapped (at least temporarily) by the salt. Data from the second experiment suggest the - presence of -other mechanisms with significantly longer time constants for removal of tritium from the salt. Because of the apparent scatter in the data at longer times, the extraction of these time constants was not attempted. The water-soluble tritium in the off-gas during the first experiment (Fig. 1.1, closed circles) did not in- crease significantly until after the injection was com- pleted, and then rose to 1750 pCi/cm?. The level then dropped rapidly to about 50 pCi/cm?®, rose again.to about 300 pCi/cm® 6.5 days after the addition, and then decreased to lower values. In the second experi- ment the watersoluble tritium in the off-gas rose rap- idly during the addition period and reached a maximum value at the end of the addition of 13,100 pCi/cm?3. Also, the ratio of the concentration of water-soluble tritium in the off-gas to that of the elemental form was substantially greater in the second experiment than in the first. . ‘ Owing to the apparent scatter in the data involving the water-soluble tritium in the off-gas for the first ex- periment, no ‘quantitative evaluation was attempted. However, in the second test, the initial decrease in con- centration has an apparent half-life of 18 hr, but the data again suggest the presence of other time constants. An attempt was made to separate the time constants by assuming that the decay curve was made up of two simple, first-order exponentials. This led to apparent half-lives of 9.3 and 37 hr for the two processes. Numer- ical integration of the water-soluble tritium data for the second experiment yieided a total flow of 58 mCi through the off-gas line during the removal period and 7.5 mCi during the addition period. Thus, a total of 65 mCi or about 65% of the tritium added is accounted for as combined tritium in the off-gas stream during this test. Since the concentration of elemental tritium was always less than 0.01 of the combined tritium concen- tration, the presence of any elemental tritium does not significantly affect this observation. The concentration of elemental tritium in the off-gas samples rose during the addition phase of each experi- ment and apparently began to decrease as soon as the addition was stopped. The maximum concentration in the first test was about 800 pCifcm?® and only 40 pCi/cm? in the second. In both cases the decrease in concentration with time after the addition was too irregular to justify any quantitative evaluation. Although no measure of elemental tritium concentra- tion in the salt is available, a value can be inferred from the concentration in the off-gas by (1) assuming that the elemental tritium in the off-gas samples represents release from the salt and only from the salt, and (2) assigning reasonable values to gas stripping parameters in the CSTF pump tank. Concentrations of elemental tritium calculated in this way indicate that the ratios of combined/elemental tritium in the salt were about 50 and 530 in the first and second experiments respec- tively. It appears that chemical interactions between the tritium-containing compound in the off-gas and the new metal of the sample line may have been responsible for the high concentrations of elemental tritium in the off- gas samples from the first test and that the actual ratio of combined/elemental tritium in the salt may have been higher than 50. The inferred maximum concentration of elemental tritium in the salt during the second experiment is about 0.13 nCi/g. Extension of the calculated tritium distribution with nominal metal-wall permeability (Table 1.2) to lower concentrations indicates that, at 0.13 nCi/g, tritium permeation through the loop walls could account for no more than about one-third of the trittum added to the system. Since this is close to the amount not accounted for in the off-gas samples, it appears that the effective permeability of the loop walls is near that of bare metal. 1.2 XENON BEHAVIOR IN THE MSBR G. T. Mays The computer program MSRXEP (Molten-Salt Reac- tor Xenon Poisoning) describing the !?*Xe behavior in the reference-design MSBR was used to perform calcula- tions to study the effects of the Knudsen diffusion coef- ficient for xenon in the bulk graphite and graphite coat- ing of the reactor core on the !3%Xe poison fraction. The program has been described previously .®»’ Following the fission of the fuel, the decay of tfie mass-135 fission fragments is assumed to follow the decay chain shown below: g 135, B (18.2 sec) 135 Sb (1.7 sec) 1351 1 BSmXe *(15.29 min) 135 (g & 135, (3.0 X 10° yr) (stable) 1351 (6.59 hr) 5 o &9 ) 135Xe (9.17 hr) This diagram illustrates the half-life of each isotope and the branching ratio of the 351 decaying to '*3MXe and '3%Xe assumed for this study. Along with this decay chain the following input data were used: Bubble concentration: 44 bubbles per cubic centimeter of salt Total heljum dissolved in salt and present in gas bub- bles: 1.0 X 107 mole/cm? Bubble separator efficiency: 90% For these conditions the mass transfer correlation in the program gives a bubble mass transfer coefficient of 0.0166 cm/sec, which leads to a loop-averaged void frac- tion of 0.55% with an average bubble diameter of 0.65 mm. The calculated ! *® Xe poison fraction is 0.0046. The reference Knudsen diffusion coefficients for the bulk graphite and graphite coating associated with the 0.0046 poison fraction are 2.58 X 107° and 2.58 X 107® cm?/sec respectively.* The bulk graphite values were varied from'2.58 X 107° to0 2.58 X 107 c¢m?/sec, assuming no graphite coating was present (Table 1.3, cases 1, 3, 5, 7), to observe the effect on the poison fraction. The low-permeability graphite coating — 0.28 mm thick — was assigned bulk-graphite values for the Knudsen diffusion coefficient and porosity, making the coating part of the bulk graphite for calculation pur- poses. Under these conditions the porosity of the bulk graphite was held constant at a value about 31 times *The complete units for diffusion coefficient are (cm?® gas)/ (sec + cm graphite). . 6. H. A. McLain et al., in MSR Program Semiannu. Progr. Rep. Aug. 31, 1972, ORNL-4832, pp. 11, 13. 7. H. A. McLain et al., in MSR Program Semiannu. Progr. Rep. Feb. 29, 1972, ORNL4782, pp. 13, 16—-17. Table 1.3. '**Xe poison fraction as a function of the Knudsen diffusion coefficient for the graphite coating and bulk graphite of the reactor core Knudsen diffusion coefficient (cm® gasfsec +cm graphite) ¢ : poison fraction Bulk graphite? Graphite coating 1. 258x10°¢b No coating 0.0153 2. 2.58% 107 2.58 X 10°° 0.0152 3. 258x 1077 No coating 0.0140 4, 258X 10°¢ 2.58 X 1077 0.0145 5. 2.58x10°® No coating 0.0113 6. 2.58X 10 258X 10°* 0.0107 7. 258X 10°° No coating 0.0077 8. 258X 10°¢ 2.58 X107% ¢ 0.00469 ZBulk graphite porosity value remains constant in all cases, V31 times greater than the value for the graphite coating. bReference value for bulk graphite. CReference value for graphite coating. dPoison fraction for reference case. greater than that of the graphite coating.* In addition, the Knudsen diffusion coefficient for the graphite coat- ing was varied within the same, aforementioned range while the diffusion coefficient of the bulk graphite was held constant at its reference value of 2.58 X 107° cm? /sec to observe the effects on the poison fraction (cases 2, 4, 6, 8). The previously stated values involving bubble characteristics and mass transfer were held con- stant throughout this series of calculations. The results (Table 1.3) indicate that Knudsen diffu- sion coefficients for the bulk graphite and graphite coat- ing at least as low as the reference values (2.58 X 107° and 2.58 X 107° cm?/sec, i.e., case 8) would be re- quired to meet the 0.005 target value for the '’*Xe poison fraction. A diffusion coefficient of less than 2.58 X 107% c¢m? /sec would be required for the bulk graph- ite with no coating, because of its higher porosity. If the permeability of the graphite coating did not yield a dif- fusion coefficient equal to that of the reference value, such a coating would have little effect on xenon poison- ing. The penalty for not coating the graphite is about 0.01 in xenon poison fraction, or 0.01 in breeding ratio, if no attempt is made to decrease the permeability or porosity of the base material. It may be noted in case 3 that a slight reduction in the Knudsen diffusion coefficient for the bulk graphite is *In practice, substantial reductions in the Knudsen diffusion coefficient probably would be accompanied by reduced po- rosity. Calculated '?*%Xe more effective in reducing the '?°Xe poison fraction than a similar reduction in the Knudsen diffusion coeffi- cient for the graphite coating in case 4. In cases 6 and 8, where the permeability of the coating is very low, re- ducing the Knudsen diffusion coefficient for the graph- ite coating affects the '?°Xe poison fraction much more strongly. 1.3 NEUTRONIC ANALYSIS H.T.Kerr D.L.Reed E.J. Allen The neutronic analysis work during this reporting period has involved several tasks aimed at additional description of the neutronic characteristics of an MSBR and the provision of neutronics information for the fueled in-reactor irradiation experiments. 1.3.1 MSBR Studies Neutronic analysis studies for the reference-design MSBR are in progress in three areas: 1. development of a two-dimensional neutronic com- putational model of the MSBR, using the computer code VENTURE and reestablishing the operability of a reactor optimization code (ROD), 2. updating the neutron cross-section data base used by various computer programs, 3. calculation of the rate at which helium will be pro- duced in the reactor vessel of the MSBR. Computational models. A neutronic computational model of the MSBR, using the computer code VEN- TURE,® is being developed. VENTURE is a multi- dimensional, multigroup, neutron diffusion computer code. The MSBR model will have nine neutron energy groups and the (r-z) geometry shown in Fig. 1.3. The various zones in the model allow for different core com- positions and cross-section sets. In addition to providing a check of the design studies made with the ROD? code using one-dimensional calculations, this model will per- mit explicit evaluation of the nuclear reactivity effects associated with localized core perturbations, such as limited core voiding. Previously such effects were con- servatively estimated from calculations for an infinite medium of salt and graphite. 8. T. B. Fowler, D. R. Vondy, and G. W. Cunningham III, VENTURE: A Code Block for Solving Multigroup Neutronic Problems Applying the Finite-Difference Diffusion-Theory Approximation to Neutron Transport, ORNL-5062 (October 1975). 9. H. F. Baumann et al,, ROD: A Nuclear and Fuel Cycle Analysis Code for Circulating-Fuel Reactors, ORNL-TM-3359 (September 1971). 10 ORNL-DWG 75-13175 I | I[ ll _ $ 1 | ' Tl T T T T ! T L R | | | | I | TR o | | | | I | I | | | | | | | | ; l -9 | | ' | | | 9 L NN o t 7 |8] 5 4 P32 v 12| 3 4 56| 7 l I | I S | : Co o A Lo E o I A o S B |- N Lo I | I N L | ‘1 1 | L 11— T 1T 1 i 1 | | PERCENT THICKNESS (cm) ZONE NO. _ FUEL SALT RADIAL AXIAL C . CONTROL RODS ' 17.2 219.5 2 CORE 1A - 13.2 ©32.8 . 219.5 3 CORE 1A 13.2 50.0 219.5 4 CORE 1B 132 119.5 219.5 ‘5 CORE Il A AND B 37.0 g4 - © 25.4 6 SALT ANNULUS 100.0 ' 5.1 5.4 7 GRAPHITE REFL. 1.0 76.2 © 61.0 (MAX.) 8 SALT ANNULUS 100.0 0.6 0.6 9 REACTOR VESSEL ' 5.1 5.1 Fig. 1.3. Two-dimensional computational model of MSBR. ROD is a computer program for nuclear and fuel-cycle analyses of circulating-fuel reactors. It consists essen- tially of a neutronics subprogram, an equilibrium- concentration subprogram, and an optimization subpro- gram. Variables such as breeding ratio, fuel composition, etc., can be optimized with respect to cost. o The operational status of the ROD code has been re- established by running a test case for the reference- design 1000-MW(e) MSBR, using the old cross-section data previously generated for the MSR program. The test case will be rerun using the ENDF/B-IV cross sec- tions, and any significant differences will be evaluated and reported. Generation of updated cross-section data. The neces- sary descriptive information for the neutronic model for use in the computer code VENTURE has been col- lected, and the most recent ENDF'? cross-section data are needed. (The neutron cross-section data used for MSBR analysis were originally derived from the GAM-II and ENDEF/B-I libraries with some ORNL modifica- tions,!! and no recent updates have been made.) The new cross-section data are being obtained exclusively from the ENDF/B-IV data files, using the AMPX!? processing system. This effort will provide evaluated cross-section data and neutron energy spectra for 11 4. Perform a one-dimensional neutron transport calcu- lation of the MSBR core to determine 123-group spectra and collapse the 123-group cross-section set to nine groups for each of the various zones in the model. ' . Reorder the nine-group set from nuclide ordering to group ordering; the cross sections are then ready for use in VENTURE and ROD. The initial processing step is capable of treating nuclides in groups of from 1 to 3, depending upon the amount of data in the ENDF/B-IV file for each nuclide. This step is now complete for all the nuclides of interest, except 232Th, - Helium production in reactor vessel. The helium pro- duction in the reactor vessel for the present reference ~ design and possible alternate designs will be estimated in typical regions of an MSBR and will serve as the data base for subsequent MSBR nuclear analyses. The steps involved in this process are: 1. Calculate 123-neutron-energy-group cross sections from the ENDF/B-IV library. The ENDF point data for 39 nuclides are weighted over an assumed energy spectrum to derive multigroup cross sections. Thermal scattering cross sections are treated at 300, 600, 900, and 1200 K for each nuclide. . Determine contributions to the multigroup cross sec- tions from resolved resonances; resonance self- shielding is treated for the various fuel configura- tions at 900 K. Perform fuel-moderator cell calculations for four geometries to adjust the cross sections for the flux depressions in regions having a high concentration of fuel or moderator (cell homogenization calcula- tions). 10. ENDF/B-IV is the Evaluated Nuclear Data File-Version IV and is the national reference set of evaluated cross-section data. 11. O. L. Smith, Preparation of 123-Group Master Cross Sec- tion Library for MSR Calculation, ORNL-TM4066 (March 1973). 12. N. M. Greene et al.,, AMPX: A Modular Code System for Generating Coupled Multtgroup Neutron-Gamma Libraries from ENDF/B, ORNL-TM-3706 (1974). conjunction with the neutronic modeling of the MSBR. (Neutron energy spectra and flux magnitudes in the reactor vessel as obtained from the. neutronic model provide the basis for calculating helium productlon rates.) ‘ Helium is produced in nickel- base alloys primarily from these reactions: ENi M5 Ni M "®Ni 5 sSFe +4He, 56Fe + “He, SONi 7, $7Fe+“He. The ®Ni(n,«) and the ®°Ni(n,a) reactions are induced only by high-energy neutrons, whereas the *®Ni(n,y) and 5°Ni(n,a) reactions are induced primarily by low- energy neutrons. In highly thermalized neutron energy spectra, as in the MSBR vessel, the two-step reaction 58Ni(n,y)*?Ni(n,a) is the principal source of helium. The *?Ni cross sections are not well known, but differ- ential measurements are being made by ENDF partici- pants.}® Also, helium analyses are available from several irradiated nickel specimens, and effective integral cross “sections will be derived from these data for comparisons with the measured cross sections. : : At present, some cross-section information is avallable for the *?Ni(n,a) reaction. Values for.the 2200-m/sec (i.e., 0.0253-eV) cross section have been reported as 13.7 barns!® and 18 barns.!® It has also been re- ported!? that a large resonance occurs at 203.9 eV with a total width, T, of 13.9 eV. From this information a preliminary estimate of the shape and magnitude of the 13. F. G. Perey, Report to the U.S. Nuclear Data Commzttee ORNL-TM-4885 (April 1975). 14. H. M. Eiland et al., Nucl. Sci. Eng., 53, 1 {(January 1974). cross section can be deduced and 123-group cross sec- tions generated. From the Breit-Wigner one-level formula, o) = (KIE) {1/1E-E,)" + 1714 where K = constant, E = neutron energy, E, = resonance energy (203.9 eV), ' = total width (13.9 eV). The constant K can be determined from the value of the cross section at 0.0253 eV, which for this study is assumed to be either 13.7 or 18 barns. Energy- dependent cross sections can be generated, and the helium production can then be estimated with the fol- lowing equation: Nye(t) = [03 0,/(02 — 0,)] N° x {11 —exp (- 0, 9010, ~ [l —exp (0, 01 fos }, where g, = (n,y) cross section of 3®Ni, o, = absorption cross section of * ?Ni, 03 =(n,a) cross section of 3 °Ni, N° ¢ = neutron flux, initial >®Ni concentration, t =time, Ny (1) = helium concentration at time ¢. 1.3.2 Analysis of TeGen Experiments Fission rates and tellurium production rates for the fuel pins in the TeGen-1 irradiation experiment were reported in the preceding MSR' semiannual report.'® The fission rates were estimated by a flux mapping experiment, direct flux monitoring of the TeGen-1 cap- sule, and computational analyses. The tellurium concen- trations in the fuel pins were calculated from these fis- sion rates, but no estimates for the accuracy of the calculated tellurium concentrations were given in the report. The accuracy of the ?33U fission product yield data'® leads to an estimated uncertainty for the yield of tellurium in the TeGen-1 capsule of about 13.5%. Assuming that the uncertainty in the estimated fission rates 1s *15%, the uncertainty in the reported tellurium concentrations is about 20%. 12 The TeGen-2 experimental capsule is scheduled to be inserted into the ORR for irradiation in October 1975. Flux monitors will be loaded into the capsule prior to the capsule’s insertion into the ORR. After the TeGen-2 capsule is removed from the reactor, the monitors will be recovered and their induced activities measured to develop estimates of the tellurium production rates for TeGen-2. 1.4 HIGH-TEMPERATURE DESIGN METHODS G.T. Yahr Thermal ratchetting and creep-fatigue damage are important considerations in the structural design of high-temperature reactor systems. Simplified analytical methods in ASME Code Case 1592 (ref. 17) and RDT Standard F94T (ref. 18) permit the assessment of ratchetting and creep-fatigue damage on the basis of elastic-analysis results, provided a number of restrictive conditions are met. Otherwise, detailed inelastic analy- ses, which are usually quite expensive for the conditions where they are currently necessary, are required to show that code requirements are met. Analytical investi- gations to extend the range over which simplified ratch- etting and creep-fatigue rules may be used to show compliance with code requirements are being performed under the ORNL High-Temperature Structural Design Program, which is supported in part by the MSRP. Modeling procedures for applying the simplified ratchet- ting rules to geometries and loadings prototypic of those encountered in LMFBR component designs are to be identified. Then the conservative applicability of these ratchetting rules and procedures and of elastic creep-fatigue rules will be demonstrated and placed on a reasonably sound and defensible engineering basis. Finally, an assessment will be made of the applicability of the simplified design methods to Hastelloy N under MSBR design conditions, and the importance of thermal ratchetting in an MSBR will be determined. 15. H. T. Kerr and E. J. Allen, in MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 14—15. 16. M. E. Meek and B. F. Rider, Compilation of Fission Product Yields Vallecitos Nuclear Center, 1 974, General Elec- tric Company, NEDO-12154-1, (January 26, 1974). 17. Code Case 1592, Interpretations of ASME Boiler and Pressure Vessel Code, American Society of Mechanical Engi- neers, New York, 1974. 18. RDT Standard F9-4T, Requirements for Construction of Nuclear System Components at Elevated Temperatures (Supple- ment to ASME Code Cases 1592, 1593, 1594, 1595, and 1596), September 1974. The detailed plans for achieving the stated objectives were given in a previous progress report.'® The basic approach is to perform a relatively small number of carefully planned and coordinated rigorous elastic- plastic-creep ratchetting-type analyses of the geometries illustrated in Fig. 1.4. Each geometry is subjected to the axial, bending, thermal transient, and pressure loadings described in Table 1.3.1 of ref. 19. Structural problems 1 and 2 are being analyzed at ORNL, using the PLACRE computer program,*® while problems 3 and 4 are being analyzed by Atomics International and Com- bustion Engineering, respectively, using the MARC com- puter program.”! Each inelastic analysis will include a complete code evaluation for accumulated strains and creep-fatigue damage. Also associated with each in- TYPE I: NOTCHED CYLINDRICAL SHELLS TYPE 2' CYLINDRICAL SHELLS Q=o—!—- ——]—-Q=0 AT, {b) UNIFORM WALL WITH DIFFERENTIAL RATCHETTING (a) STEPPED WALL THICKNESS r 'Tz NNNNARNARNNNY _1(‘: [ (¢) UNIFORM WALL WITH {d} BUILT-IN CYLINDER AXaL TEMPERATURE VARIATION 13 elastic analysis are a number of elastic analyses to pro- vide the input parameters required to apply the various simplified ratchetting rules and procedures and elastic creep-fatigue rules. The progress to date on these studies is discussed below. Both Al and CE have encountered difficulties in their three-dimensional inelastic analyses. Although consider- 19, J. M. Corum and G. T. Yahr in MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 15-22. 20. W. K. Sartory, “Finite Element Program Documenta- tion,” High-Temperature Structural Design Methods for LMFBR Components Quart. Progr. Rep. Dec. 31, 1971, ORNL-TM-3736, p. 66. 21. MARC-CDC, developed by MARC Analysis Research Corporation, Providence, RI. ORNL-DWG 757768 TYPE 3: NOZZLE-TO-SPHERICAL SHELL |_—16"0D. X | 0375" WALL Z e S\ JUNCTION DETAIL {TYPES 384! 76"1D. X 11875" WALL TYPE 4 NOZZLE-TO-CYLINDRICAL SHELL (IHX INLET NOZZLE) 18"0D. X 0.375" WALL 76" 1D X 11875 WALL Fig. 1.4. Structural configurations used in the analytical investigation of the applicability of simplified ratchetting and creep- fatigue rules. able effort has gone into developing finite-element models that are of a size that can be accommodated on present-day computers and into improving the MARC computer program, the large 3-D inelastic analyses are proving considerably more expensive to. run than had been expected. The experience at Al and CE indicates the importance of developing simplified methods of analysis. Three- dimensional inelastic analysis of many realistic com- ponent geometries is too expensive and time consuming at present to be used routinely. Although developments in computers and stress analysis programs may bring the cost down in the future, it is desirable, meanwhile, to minimize the number of inelastic analyses that must be done. 1.4.1 Circular Cylindrical Shells* Nine cases of circular cylindrical shells have been pro- posed for the present study.!® Two of the cases involve notched shells. The other seven cases involve axial varia- tions in temperature, pressure, and/or wall thickness, or a built-in wall. All nine cases were to be analyzed using the ORNL in-house finite-element program PLACRE. A ten-cycle inelastic analysis and a one-cycle elastic analysis have now been completed for all nine cases. Both the inelastic and the elastic results for all nine cases have been completely postprocessed. Because of modifications to the creep-fatigue damage rules presently under study by the ASME Boiler and Pressure Vessel Code Working Group on Creep/Fatigue, it may be necessary to modify the ORNL postprocessor and repeat some of the postprocessing to keep the present study up to date. 1.4.2 Nozzle—to—Spherical Shell® After some difficulties, the MARC computer code is operational on the IBM computer at the Rockwell Inter- national Western Computing Center, and check cases have demonstrated that this code will perform satisfac- torily. Considerable effort has gone into developing the finite-element model of the nozzle—to—spherical shell. An isoparametric three-dimensional 20-node brick ele- ment will be used to model the entire geometry. Be- cause of symmetry about the plane of the applied moment, only half of the nozzle—to—spherical shell has to be modeled. There are six 30°-wide elements around *Work at ORNL, by W_K. Sartory. TWork at Atomics International, by Y. S. Pan. 14 the half-model. There are three elements through the wall at the root section of the nozzle and only one element through the wall in both the nozzle and the sphere away from the intersection region. A series of elastic analyses must be done, since this is a thermal stress problem in which temperature varies with time. Since the moment applied to the nozzle is the only nonaxisymmetric load, the principle of super- position will be used to reduce the cost of the elastic analyses. A series of axisymmetric analyses were done to determine the stresses due to the internal pressure and temperature, and one three-dimensional analysis was done to determine the stresses due to the moment applied to the nozzle. The stresses from the three- dimensional analysis will be added to the stresses from the axisymmetric analyses to obtain the total elastic stresses. The axisymmetric model in the elastic analyses was used to determine what maximum thermal load incre- ment may be employed without having to do an exces- sive number of iterations during each increment. On this basis, the first cycle of the three-dimensional inelastic analysis was divided into 32 increments. The first three increments of the three-dimensional inelastic analysis have been completed. The computer cost for these three increments was higher than anticipated. Efforts will be made to find some way to reduce the cost to an accept- able level. 1.4.3 Nozzle-to-Cylinder Intersection® The original concept for the inelastic ratchetting-type analysis of the nozzle-to-cylinder intersection was to perform two separate analyses: (1) a thin-shell analysis of the whole structure, and (2) a detailed three- dimensional solid analysis of :the intersection only. Dis- placements and forces to be applied at the boundaries of the three-dimensional solid model of the intersection were to be determined from the shell model at the end of each loading increment. The total computer time of the two analyses would be less than that required for the solution of the problem, using one model of the complete nozzle-to-cylinder intersection with suffi- ciently small elements in the intersection region. How- ever, the transfer of the forces and deflections from the shell analysis to the three-dimensional solid analysis was found to be more difficult than anticipated. Because the shell element and solid element have different displace- ment functions, a special constraint must be imposed on the shell elements at the boundaries of the three- dimensional solid model to assure compatibility. This FWork at Combustion Engineering, by R. S. Barsoum. stiffens the intersection in the shell model. When run- ning the initial elastic analyses, it was found that small 15 changes in the displacement boundary conditions applied to the solid model would produce large changes in the results of the analysis. From a pragmatic view- point, the biggest difficulty with the two-model method is assuring that the correct data are transferred from the shell analysis to the solid analysis at every increment in loading. ' ' Due to the above considerations, it was decided to do the "analysis by using only one model, made up of a combination of a reduced integration shell element and a 20-node solid element which are fully compatible with each other. - ‘ , It was necessary to restructure a large portion of the MARC program to perform the inelastic analysis for the 3-D model of.the nozzle-to-cylinder intersection. This restructuring made a larger core available for the analy- sis. The restructuring involved stripping unneeded por- tions of the program, putting common space on low- cost storage, and eliminating mesh optimization and its correspondence table. The inelastic analysis of the nozzle-to-cylinder inter- section was - started. The full pressure and nozzle- moment loadings were imposed on the structure, which resulted in stresses less than 0.936 of the yield stress at 870 K (1100°F). When the first increment of thermal load was applied, convergence was not obtained because of an error in the computer program, which is being corrected. 2. Systems and Components Development R. H. Guymon 2.1 GAS-SYSTEMS TECHNOLOGY FACILITY R. H. Guymon G.T. Mays After a brief shutdown at the beginning of this report- ing period to modify running clearances in the pump, water operation of the Gas-Systems Technology Facility (GSTF) was resumed on March 11, 1975, with the bypass loop blanked (Fig. 2.1). Considerably larger salt- pump shaft oscillations were encountered than before the labyrinth clearances were increased.! After obtain- ing calibration data for the main-loop variable-flow re- strictor and for the salt pump at low flows, the loop was shut down to install the bypass loop variable-flow re- strictor. Water testing was then resumed on April 14 and continued throughout the period. Data for calibration on the bypass loop variable-flow restrictor and for the salt pump were obtained. At normal pump speed the head-capacity performance of the installed impeller was ~3% below the nominal loop design conditions. At the nominal liquid flow rate and pressure drop in the main loop, the flow rates from the gas outlets of the bubble separator were satisfactory. Although loop cavitation (as indicated by noise level) was reduced by replacing the variable-flow restrictors with orifices, the amplitude of the salt-pump shaft oscil- lations was not reduced appreciably. Preliminary infor- 1. R. H. Guymon and W. R. Huntley, MSR Program Semi- annu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 23-25. SALT PUMP PURGE GAS SALT PUMP OFF-GAS BUBBLE GENERATOR BULK SALT SEPARATOR G | mation shows that leakage past the salt-pump shaft labyrinth is higher than desirable, and attempts will be made to reduce this. Tests under actual operating conditions with water in the loop indicated that the densitometer will be satisfac- tory for salt operation. Preliminary information ob- tained from saturating the loop water with air and then stripping the air by injecting helium at the bubble gener- ator indicated the need for monitoring the oxygen con- centration in the off-gas from the bulk salt separator, in the off-gas from the salt pump, in the loop water, and perhaps in the water in the pump tank. Difficulties were also encountered with the response time of the oxygen monitors and with the reproducibility of their readings. 2.1.1 Cavitation and Salt-Pump Shaft Oscillations Data on the salt-pump shaft deflections and oscilla- tions obtained during the previous period! indicated that the running clearances at the labyrinth (fountain flow area) and at the impeller hub should be increased to prevent contact of the metal surfaces during opera- tion with salt (Fig. 2.2). After increasing the clearances, water operation was restarted with the bypass loop blanked off. The shaft oscillations were much larger than they had been previously under similar conditions. Turbulence or cavitation as indicated by noise was the apparent cause. For more flexibility in operating condi- tions, the bypass-loop variable-flow restrictor was in- stalled. Loop parameters were then recorded at many ORNL-DWG 75- 12614 BULK SALT SEPARATOR OFF-GAS FLOW REST ...... Fig. 2.1. Gas-Systems Technology Facility. ORNL -DWG 75-12642 /;’ SHAFT \ \ 7 . BOWL N \ \ ? N N N ! |~ SLINGER | FOUNTAIN FLOW . el L—— FOUNTAIN FLOW N el ==_"-F UPPER AND LOWER N / - =- = = LABYRINTH \ p— —_— —_ T x e— — f . \:__., - ’___—..____ ' N——— - voLuTe — = Q———-- - - —— -NIMPELLER"— — T \‘_ —_— = N —— IMPELLER HUB LABYRINTH Fig. 2.2. GSTF salt pump. combinations of salt-pump speed and settings of the main-loop and bypass-loop variable-flow restrictors. Log-log plots were made of pressure drops across various sections of the loop as functions of the flow rates through the segments. Since the head loss for a fixed resistance is proportional to a fixed power of the fluid velocity, the curves should be straight lines unless the character of the resistance changes due to cavita- tion. The plots indicated that cavitation was occurring in the main loop between the inlet to the main-loop variable-flow restrictor (FE-102A) and the throat of the bubble generator at flow rates above 320 gpm (1200 liters/min) with the variable-flow restrictor set at 1 in. (25 mm), above 470 gpm with the variable-flow restric- tor at 2 in. (1800 liters/min at 51 mm), above 600 gpm with the variable-flow restrictor at 3 in. (2300 liters/min at 76 mm), and above 630 gpm with the variable-flow restrictor at 4 in. (2400 liters/min at 102 mm). The data were not sufficiently precise to determine whether cavi- tation was also occurring in the bypass loop; however, noise indicated that it was. Since the loop turbulence and/or cavitation as indi- cated by noise and the salt-pump shaft oscillations were unacceptable at conditions required by the bubble separator design, changes were made in the main-loop and bypass-loop flow restrictions. By replacing each variable-flow restrictor with two or more orifices in series, the loop noise level was decreased, but there was 17 little or no decrease in the amplitude of the shaft oscil- lations. The amplitude of the shaft oscillations was plotted as a function of salt-pump speed at various operating con- ditions (Fig. 2.3). At salt-pump speeds less than about 1600 rpm, the oscillations were reasonably small and at any given speed appeared to be unaffected by: (1) flow rates between 450 and 1050 gpm (1700 to 4000 liters/ min), (2) salt-pump overpressures between 5 and 15 psig (1.3 X 10° to 2.0 X 10° Pa), (3) type of restriction (variable-flow restrictors, orifices, or a combination of these), or (4) flow route (through the main loop, bypass loop, or both). At higher speeds the oscillation ampli- tude increased rapidly with increases in speed, and there was more scatter in the data, making it difficult to eval- uate improvement in cavitation and effects of other variables. However, at any given speed above about 1700 rpm, increasing the flow rate (between 450 and 1050 gpm) caused larger oscillations. One possible explanation for the increased amplitude of the oscillations at higher speeds is that the shaft is approaching its critical vibration frequency and is there- fore more sensitive to disturbances, such asloop turbu- lence or cavitation. The critical speed of this impeller assemnbly is 2280 rpm in air,' which would indicate a maximum normal operating speed of 1710 rpm, using the normal industrial practice of operating pumps at less than 75% of critical speed. If the pump shaft oscillations were, in fact, a con- sequence of operation near the critical speed of the rotating assembly, two obvious alternatives were avail- able to reduce the amplitude of the oscillations: 1. further reduction of the loop disturbances to mini- mize the driving forces that cause oscillation, 2. operation at lower speeds to reduce the oscillatory response to disturbances. The first alternative was rejected because it would have required extensive modification of the loop, and it was difficult to guarantee that all sources of such disturb- ances could be reduced to satisfactory levels. Design calculations showed that the desired flow and head (3800 liters/min at 30.5 m, or 1000 gpm at 100 ft) could be obtained by replacing the present 113%-in.-diam (290-mm) impeller with a 13-in.-diam (330-mm) unit and operating it at 1500 rpm. A larger impeller is being machined from an available Hastelloy N rough casting. Since the larger impeller will be somewhat heavier than the original one, it will cause a reduction in the critical speed of the rotating assembly. The estimated critical speed with the new impeller is 2000 rpm, which makes the operating speed 75% of the critical speed. ORNL-DWG 75-12613 140 o 120 . ®* 450-649 gpm TOTAL FLOW 4 100 o 650-849 gpm TOTAL FLOW Ao a B850 -1049 gpm TOTAL FLOW ‘a ] g E ;_:; 8o a ~J a o - @ o 60 e - w . ® 40 & o . & o ¢ ¢ mjta O [ ] o.o ° [ %] ® 20 S o000 " Y [ ] [ ] [+] b L ] ® oP [ ] © S R Y SN SR o 1000 1400 1200 1300 1400 1500 1600 1700 1800 SALT PUMP SPEED (rpm) Fig. 2.3. GSTF pump shaft oscillations. At a few off-design conditions during some of the later runs, the pump shaft deflection records showed random spikes in one direction superimposed on the relatively uniform oscillations described earlier. These occurred with higher than normal flow rates in the main loop or at reduced system overpressure. Since either increasing the overpressure or injecting gas at the bubble generator reduced or eliminated these random oscilla- tions, it was concluded that they were a consequence of cavitation at the bubble generator. Such cavitation and the attendant oscillations are not expected to occur at normal operating conditions. 2.1.2 Salt-Pump Performance Data and Calibration _of the Variable-Flow Restrictors The original design of the GSTF provided for varying the salt-pump speed and/or changing the variable-flow restrictor settings to obtain different flow rates or pres- sures needed for future experiments. However, instru- mentation will not be provided for measuring the bypass-loop flow rate during salt operation, and only two salt pressure measuring devices will be installed (at the salt-pump discharge and at the bubble-separator dis- charge). Also, since the salt pump was modified and has a mismatched impeller-volute combination, no perfor- mance data were available. Therefore, extra pressure indicators were installed for the water tests, and loop pressure profiles were obtained at various pump speeds, flow rates, and variable-flow restrictor settings to evalu- ate the pump performance. The calibration of the main-loop variable-flow restric- tor and of the salt pump at low flow rates was straight- forward, since, with the bypass loop blanked off, the total pump flow was measured directly by the main- loop venturi. However, once the bypass-loop variable- flow restrictor was installed, the calibration of it and the pump was complicated. The main-loop variable-flow restrictor was closed, and the bypass-loop variable-flow restrictor was calibrated at low flow rates, using the pump calibration curves established before it was in- stalled. The main-loop variable-flow restrictor was then opened to various settings, and the pump calibration curves were extended by adding the measured flow through the main loop to the flow through the bypass loop taken from the bypass-loop variable-flow restrictor calibration curves. Then using these extended head- capacity curves for the pump, it was possible to extend the calibration curves to higher flows. The pump calibration curves (Fig. 2.4) indicate that at 1770 rpm the pump flow rate will be 970 gpm (3700 liters/min) at 100 ft (30.5 m) of head. The original design called for 500 gpm (1900 liters/min) through each loop; however, the bypass flow rate can be reduced to 470 gpm (1800 liters/min) without compromising any of the objectives. To determine the main-loop variable-flow restrictor setting for normal operation with a flow rate of 500 gpm in the main loop, plots were made of the pump head vs flow for several settings of the flow restrictor. From these, a curve was made of pump head at 500 gpm (1900 liters/min) vs settings (Fig. 2.5). A 1.85-in. (47-mm) setting will give the desired head of 100 ft (30.5 m) at 500 gpm. ' ORNL —DWG 7512614 T 140 — | i —— . 120 1770 rpm ——- S — ' l ; | { ! N 100 | : : —_ : W §- \1500 rpm : - 80 ™~ : (=} H E I Q w 60 I a = 2 a . 40 —_ ———— .- ,XQ,, — —— 20 T~ 0 . 0 200 400 600 800 1000 1200 FLOW (gpm) Fig. 2.4. Head capacity curves for the GSTF pump. . . ORNL-DWG 75-12615 5 T l 1 ~_§ MAIN VFR (FE-102A) SETTING FOR 500 gpm 2 BY-PASS VFR (FE-104A) " SETTING FOR 500 gpm 2\ 3 BY-PASS VFR (FE-1044) SETTING FOR 470 gpm i ~N 5 | | | | _ /] rd PUMP HEAD {feet of liquid} / 7 2 2 N o' 0 ! 2 3 4 5 6 VARIABLE FLOW RESTRICTOR SETTING (in) Fig. 2.5. Variable-flow restrictor calibrations. The bypass variable-flow restrictor settings were deter- mined similarly and found to be 1.85 in. (47 mm) at 500 gpm (1900 liters/min) or 1.70 in. (43 mm) at 470 gpm (1800 liters/min). 2.1.3 Salt-Pump Fountain Flow The GSTF salt pump is a centrifugal sump pump having an impeller which rotates in a volute section which in turn is located in a pump bowl. The clearance between the impeller and the volute assembly at the pump inlet allows leakage from the discharge directly to the pump suction (see Fig. 2.2). A second bypass flow, called the fountain flow, escapes through the clearances between the impeller shaft and the volute assembly. This bypass stream flows into the pump bowl, circulates downward, and reenters the main stream at the pump suction. Due to the large liquid holdup and large surface area in the pump bowl, significant gas-liquid mass trans- fer can occur in the fountain flow stream, and therefore its flow rate is important in analyzing mass transfer processes in the loop. Since the fountain flow is not measured directly, a method using mass balances on measured gas flows was developed to determine this flow rate. A lumped-parameter model of the GSTF was used to develop equations from which an expression for the fountain flow was derived. The system model contains two major regions: the pump bowl and the primary loop (main and bypass segments), consisting of a gas section and a liquid section. Each section was assumed to be perfectly mixed. The three general time- dependent equations for a specific gas in a gas mixture are given in Table 2.1, representing gas mass balances for the pump-bowl gas section, the pump-bowl liquid section, and the primary-loop gas section (circulating voids). These were simplified by applying the following assumptions: 1. There is no gas carry-under in the pump bowl, which implies that (2) the efficiency for separation of bub- bles from the fountain flow (ef) is unity and Fj, = Fr(1 — ¥, )+ Fpg,(b) the bubble surface area in the pump bowl (4, ), the void fraction in the pump bowl (¥ppg), and the concentration of gas in the pump bowl (Cs) are nonapplicable or zero. . Mass transfer equilibrium exists in the primary loop, which implies that the mass transfer term 4343 (Cs- KRTC,) is zero. . Steady-state conditions exist, making all time deriva- tives zero. . Fp =0, since there was no gas purge flow during the experiments. Therefore, Eq. (3), Table 2.1, reduces to Fp—F¥; —Q¥,e,=0. (4) Solving for ¥ , v, =F3/(Ff+Q€s)- (5) By adding Eqgs. (1) and (2), Table 2.1, and simplifying, Fp¥; Cs + FAl — ¥, )Cs + FggCa — FC —F{1 = ¥,)C; +FpgCy =0. (6) By substituting Eq. (5) into Eq. (6), Fr may be ex- pressed in terms of a quadratic equation: (Ca — Cg)Ff2 t [(Qe, — Fg + Fpg)Cs — Cy) +FpCy — FCi ] Fp Qe [Fge(Cy —C2) —F,C,] =0. (7) Equation (7) is a general solution for the fountain flow, which depends upon the gas concentrations in 20 each of the three sections of the model. If it is assumed that mass transfer equilibrium exists at the gas-liquid interface in the pump bowl, the gas concentration in the pump bowl liquid (C,) is related to the corresponding concentration in the pump bowl gas section (C;) by Henry’s law. If only one gas is involved (e.g., helium), C, follows directly from the pump bowl overpressure. Further, since mass transfer equilibrium was assumed for the primary loop, the gas concentration in the loop liquid (for a single gas) follows from the loop average pressure and Henry’s law. Thus, the fountain flow may be evaluated from Eq. (7), using other known liquid flow rates and measurable gas flow rates into and out of the system. If no mass transfer is assumed to occur at the gas-liquid interface in the pump bowl, the gas con- centration in the liquid leaving the pump bowl is the same as that in the entering liquid (i.e., C; = C,), and Eq. (7) reduces to Qe F.= ] T (FRCyIF Cy) =1 @) Since the rate of mass transfer in the pump bowl is neither infinite nor zero, Eq. (7) will give a low indica- tion and Eq. (8) will give a high indication of the foun- tain flow rate. The deviation from the actual fountain flow rate will depend on how much mass transfer actu- ally occurs in any experiment. If the loop void fraction is increased (by increasing the gas input rate), the con- tribution of mass transfer across the gas-liquid interface to the flow rate of gas out of the pump bowl will be reduced relative to the bubble contribution. Therefore, a plot of the calculated fountain flow vs the reciprocal of the gas input rate at several different conditions should give the actual fountain flow rate when extra- polated to zero (infinite gas flow rate), using either Eq. (7) or (8). The preceding equations and approach were used to calculate the fountain flow for the GSTF pump. Results from the plot indicated that the curves generated were not defined well enough to provide accurately the re- quired extrapolations. The range of fountain flows at the highest gas input rate at which data were obtained was 100 to 200 gpm (380 to 760 liters/min). Even the lower estimated value for the fountain flow may excessively complicate future mass transfer experi- ments; so efforts will be made to reduce this flow. Since the labyrinth clearances cannot be reduced without incurring metal-to-metal contact between the pump shaft and the volute, back vanes will be installed on the top of the impeller to minimize the differential pressure which drives the fountain flow. Table 2.1. Gas mass balance equations for computation model of GSTF Rate of change of gas purge ' bubbles separated : mass transfer of flow of oft- gas inventory in = flow in + from fountain + dissolved gas across - gas from _ pump bowl in gas space flow ‘ liquid-gas interface ‘ pump bowl Equation V, ci'cr‘l = FpCp + Ff\lrLefCJ + hA (C, —KRTC)) - F.C 1 . Rate of change dissolved gas dissolved gas mass transfer mass transfer dissoived gas of gas inventory present in ' present in of dissolved of dissolved present in flow dissolved in = incoming fountain + flow from — £as across — gas to bubbles _— from pump pump bowl liquid flow bubble genera- liquid interface in pump bowl bowl to loop tor via bulk salt dcC separator Equation V, a’tz = Ff(l - )HC, + FpoC, - hoA, (C, —KRT(C,) - h,A, (C, —KRTC) - Fp (1 -¥pp)C, 2 Ratio of change flow of flow of bubbles mass transfer bubbles in dissolved gas bubble removed « of gasinventory gas to from pump bowl of dissolved fountain . in fountain by bubble in loop voids = bubble + toloop + gas to bubbles — flow — - flow — separator dv generator in loop ' Equation V, L = Fp + Fi¥pp + hid, (C, — KRTC,) - Fe¥per R A &) - OV jcg 3 dt 12 2.1.4 Densitometer Studies The void fraction of the liquid after it leaves the bub- ble separator must be known in order to evaluate the bubble-separator efficiency. Densitometer instrumenta- tion (using a digital voltmeter for readout) was installed at the loop, and tests were made using the 30-Ci (1.1 X 10'% dis/sec) '37Cs source. The effects of changing void fraction were simulated by inserting plastic sheets 3 and 6 mils (0.076 and 0.152 mm) thick between the source and detector and by using the metallic shim slide containing stainless steel calibration plates 10 to 250 mils (0.254 to 6.35 mm) thick which were designed for this purpose. The drift encountered during development testing was still present. The hourly drift would be equivalent to a change in bubble-separator efficiency of about 10% 22 during salt operation (assuming a void fraction of 0.3% at the inlet to the bubble separator). Thus, short-term tests will be required to determine the bubble-separator ‘efficiencies. Based on densitometer readings, the bubble-separator efficiency was greater than 98% at various operating conditions with water, which is slightly higher than predicted. : Nonmenclature A, = area of liquid-gas interface in pump bowl A, = bubble surface area in pump bowl A5 = bubble surface area in loop C, = gas concentration in pump bowl gas space C, = gas concentration in pump bowl liquid C; = gas concentration in the loop bubbles C, = gas concentration in loop liquid C; = gas concentration in pump bowl bubbles Cp = gas concentration in gas purge entering the pump bowl gas space F, = flow of total off-gas from pump bowl Fp = gas flow rate to bubble generator Fgs = liquid flow from bubble separator via the bulk salt separator to the pump bowl (assume no bub- bles) F¢ = fountain flow (liquid and bubbles) F, = flow of gas purge into pump bowl gas space F; =flow of liquid and bubbles from pump bowl to loop h, =mass transfer coefficient for gas dissolved in pump bowl liquid to gas space in pump bowl h, = mass transfer coefficient for gas dissolved in pump bowl liquid to bubbles in pump bowl h, = mass transfer coefficient for dissolved gas in loop liquid to bubbles in loop liquid K = Henry’s law (solubility) coefficient Q = flow of liquid and bubbles to bubble separator R = universal gas constant T = temperature ¥, = total gas volume in pump bowl V', = volume of liquid and bubbles in pump bowl ¥y = volume of circulating liquid and bubbles in main loop €s = bubble separator efficiency er = efficiency for separation of bubbles from foun- tain flow WV, =void fraction in loop fluid W pp = void fraction in pump bowl fluid 2.2 COOLANT-SALT TECHNOLOGY FACILITY (CSTF) A. N. Smith Modifications to the salt cold trap (SCT) were com- pleted,” and the loop was started up on March 14, 1975, and operated for 1279 hr to check the effective- ness of the salt mist filter, to obtain data on salt mist generation rates under different operating conditions, and to obtain salt and off-gas sample data in preparation for the tritium tests. Work was completed on design, fabrication, installation, and checkout of the tritium addition system, and the tritium test program was started. At the end of the report period, two tritium additions had been completed, and plans were being made for additional tritium addition tests as well as tests designed to examine how the tritium behavior is affected by the injection of steam into the salt. 2.2.1 Loop Operation The SCT flow lines were disconnected from the sys- tem, and the loop was started up on March 14, 1975. The loop operated continuously until May 6, 1975, when it was shut down to permit installation of equip- ment in the containment enclosure for the tritium tests. The loop was started again on June 27, 1975, and it was still in operation at the end of the report period, when more than 2500 hr of operating time had been logged without plugging in the off-gas line. This is convincing evidence that the salt mist filter has been effective, since the off-gas line had plugged after only 240 hr of opera- tion before the mist filter was installed.? The loop is operating at a pump speed of 1790 rpm ~ (estimated salt flow rate, 54 liters/sec) and a pump bowl 2. A. N. Smith, MSR Program Semiannu. Progr Rep. Feb. 28, 1975, ORNL-5047, pp. 25-29. 3. Ibld p. 26. gas overpressure of 2.67 X 10° Pa (2000 mm Hg abs). The pump bowl off-gas flow; which consists of helium containing a few percent of BF3, plus trace quantities of condensible material, is about 2 liters/min (STP). The BF; concentration of the off-gas is a function of the BF; partial pressure in the salt, which in turn is a strong function of the salt temperature. Except for short periods of time when special tests required a different setting, the salt circulating temperature has been main- tained at 535 to 540°C, at which point the BF, concen- tration in the off-gas stream is about 2.5% by volume. The loop off-gas stream, except for a 100-cm?/min sample stream, is passed through a —72°C cold trap (dry ice—alcohol bath). Material which is a:dirty white solid at trap temperature and a dirty brown fluid upon warm- ing to room temperature, and which is rich in tritium (about 10°® nCi/g), continues to collect in the cold trap at a rate of 1 to 10 ng/cm?® (STP) of off-gas. This mate- rial is believed to be a variable mixture whose composi- tion depends on the relative partial pressures of BFj; HF, and H, O over the salt (see Sect. 4.1). As of 0800 on August 31, 1975, the loop had accu- mulated 3073 hr of salt circulating time since being reactivated in December 1974. - 2.2.2 Salt Mist Test Between March 25, 1975, and April 24, 1975, a series of tests was run to determine the concentration of salt 23 mist? in the off-gas stream as a function of salt tempera- ture and BF; flow. For each test, the -loop operating conditions were set at the desired values, and the off-gas stream was shunted through a metallic 5- to 9-um filter which was inserted into the-salt-sample access nozzle on the pump bowl. The salt mist concentration was calcu- lated using the gain in weight of the filter and the total flow of gas. A total of ten tests were carried out. The test time was normally about 12 to 15 hr, but in two cases it was shortened to about 3 hr because of the buildup of a high-pressure drop across. the filter. Pump bowl pressure was 2.67 X 10° Pa, and total off-gas flow was 2.1 liters/min (STP). When BF; was added to the helium entering the pump bowl, the BF; flow was ad- justed so that the BF, partial pressure in the incoming gas was the same as the calculated partial pressure of the BF; over the salt, assuming the eutectic mixture of NaBF, and NaF. The salt circulating temperature was controlled at either 535 or 620°C. The observed con- centrations of mist in: the off-gas (Table 2.2) ranged from ~100 ng/cm?®-at the lower temperature to as high as 500 ng/cm?® at the higher temperature. At the lower temperature; where the expected partial pressure of BF, in the salt was low, the addition of BF; with the cover gas was ineffective in reducing the amount of mist in the off-gas. At the higher temperature (higher BF, 4. Ibid, p. 27. Table 2.2. CSTF salt mist test, March 1975 Calculated Salt BF, BF, - . temperature vapbr flow Salt mist concentration - : ‘ 3 °C o . Q) pressure? (cm?®/min, STP) L(ng 5alt/(.m- STP off-gas) (mm Hg) 620 250 250 198 336 160 Average 230 620 250 0 308 ‘ 532 532 Average 455 535 50 50 80 100 Average 90 - 535 50 0 111 88 Average 95 2 Assuming the eutectic composition. partial pressure in the salt) a significant reduction was observed in the mist concentration when BF3; was added with the cover gas. However, it was not reduced to as low a value as at the lower temperature. These results, while not completely definitive, suggest that BF; evolution from the salt may not be the only mist- producing mechanism in the pump tank; that is, simple mechanical agitation of the salt in the pump bowl may also produce some mist. In addition, no data were ob- tained with excess BF5 concentrations in the cover gas. Since the installation of the salt-mist filter in the off-gas line was effective in eliminating the operational prob- lems in the CSTF caused by the mist, further investiga- tions of methods to limit or control the mist have been deferred in favor of experiments to study tritium be- havior in the system. 2.2.3 Tritium Experiments The decision to use tritium rather than deuterium as a test gas® in the CSTF necessitated additional design effort and a somewhat more elaborate test setup in order to satisfy applicable radiation safety require- ments. A conceptual design was prepared for the tritium addition system, and a preliminary radiation safety analysis was performed for the proposed test. Engineer- TRITIUM TRANSFER CYLINDER . 400 °«C — ] PURIFIER SAMPLE i - HYDROGEN 24 ing design, procurement, fabrication, and installation of the tritium addition system were completed by the third week in June 1975. The addition tube and the addition procedure for tritium are essentially the same as those devised for the addition of deuterium. The inner tube of the addition assembly is pressurized with hydrogen containing a small amount of tritium, and the gas is allowed to diffuse through the Hastelloy N tube which forms the lower end of the addition tube, and which is immersed in the {lowing salt stream (see Fig. 2.6). The Hastelloy N tube is 120 mm long by 12.7 mm in OD by 10.6 mm in ID, and provision is made to fasten metallurgical specimens to the upstream face. The portion of the addition tube immediately adjacent to the Hastelloy N section is surrounded by an evacu- ated annulus monitored to check for extraneous tritium leakage. The hydrogen-tritium mixture is passed through a purifier (Pd-Ag tube) to remove impurities such as O,, N,, and H, O which might interfere with the permeation process. The probe volume, V), (injec- tion tube plus adjacent tubing), and a calibrated refer- ence volume, V,, are interconnected and pressurized with the H,-T, mixture at the start of the test. The two volumes are then isolated from each other while the addition is in progress. At the end of the addition ORNL-DWG 75-12616 7 gz VACUUM 1 m_ 35°C VACUUM ANNULUS~_|| 100 °C INJECTION TUBE SALT o FLOW °32 C& Fig. 2.6. Tritium addition system for GSTF. period, the difference in pressure between Vy, and V, is recorded; then the two volumes are equilibrated and the final equilibrium pressure is recorded. The initial and final pressures in Vp, py and p, respectively, the final equilibrium pressure, p3, and the known volume and temperature of ¥, are then used to calculate the amount of gas which permeated the addition probe according to the equation (1 —p2)(P, —Pa)x V, (p3 — P2) RT, r n= where 7 is the number of moles of gas transferred, R is the molar gas constant, 7, is the"reference volume tem- perature, and the other symbols are as previously de- fined. ' During the addition, the amount of extraneous leak- age is calculated from pressure rise measurementsin the evacuated annulus, and this quantity is subtracted from n to obtain the net amount of gas transferred into the salt. The tritium content of the hydrogen-tritium mix- ture is determined by mass spectrometer analysis, and the net amount of added tritium is then calculated. Tritium (and hydrogen) which enters the salt stream is 25 assumed either to remain in the salt or to leave the salt - by one of two paths: either by permeation through the walls of the loop piping or by transfer to the gas phase in the pump bowl or in the salt monitoring vessel (SMV) and leaving the loop with the off-gas stream. During and after an addition, the tritium content of the salt is mon- itored by taking samples of salt from the salt pool in the pump bowl or in the SMV, and the tritium content of the off-gas stream is monitored by taking samples from the off-gas line at a point about 1 m downstream of the pump bowl. The off-gas sample stream is passed first through a water trap to collect chemically combined (water-soluble) tritium, and then through an oxidizing atmosphere to convert elemental tritium to tritiated water, which is collected in a second trap. The tritium contents of the salt samples and of both the off-gas samples are determined by a scintillation counting tech- nique. During the initial tritium addition experiments, no provision was made for measuring loop wall permea- tion, so that the tritium lost by this mechanism is assumed to be the difference between the amount of tritium added and the sum of the quantities which leave in the off-gas stream and which remain in the salt. During the March 14, 1975, to May 6, 1975, oper- ating period, a number of salt and off-gas samples were taken to obtain baseline values for tritium concentra- tion and to shake down and evaluate the sampling tech- niques. During shutdown of the CSTF in May and June 1975, the tritium addition probe was installed in the surveillance-specimen access tube, and the final installa- tion work was done on the tritium addition system. A stainless steel valve (HV-253A) and some stainless steel tubing, which were part of the original off-gas sample- line installation, were removed and replaced with a Monel valve and Hastelloy N tubing, because it was felt that the Monel and Hastelloy N would be less likely to react with the off-gas sample stream. Two 2.5-cm-diam X 45-cm-long Hastelloy N tubes were filled with salt from the drain tank and set aside as representative samples of the salt as it existed prior to the start of the tritium tests. On June 27, 1975, the loop was filled dnd salt circula- tion was resumed. Several additions of hydrogen were made to check out the operation of the addition system and to obtain data on permeation rates. A total of 313 cm® (STP) of hydrogen was added in these tests, and the last addition (31 ¢cm® STP) was made on July 16, 1975. With the addition tube pressurized to 1.37 X 10° Pa, the measured permeation rate was about 3 em? /hr, compared with a predicted value of 1.3 cm?/hr. The first addition of tritium was made on July 17, 1975, and a second addition, with conditions essentially the same as for the first addition, was made August 5, 1975. In each case, salt and off-gas samples were taken during the tritium addition (about 10 hr) and for about 2 weeks afterward, until sample results indicated that the tritium levels had returned to their pretest values or had stabilized. Data for calculation of the amount of added gas are shown in Table 2.3. A mathematical analysis and discussion of the sample results are presented in Sect. 1.1.2. Table 2.3. Tritium addition data for CSTF tests Test number T1 T2 Date T-17-75 8-5-75 Addition started 1002 0846 Addition ended 2050 1912 Addition time (hr) 10.8 10.4 Initiat pressure, p | (Pa) 1.38 x 10° 1.37 X 108 Final pressure, p2 (Pa) 1.06 x 10° 1.07 x 108 Equilibrium pressure, p3 (Pa) 1.24 X 10° 1.23 X 10° V, volume (cm?) 155 155 T, temperature ("K) 303 306 Gross permeation rate (cm?® /hr) 3.10 3.33 Average leak rate (cm?/hr) 0.07 0.04 Net permeation rate (cm?/hr) 3.03 3.29 Tritium concentration in mixed 1010 1100 gas (ppm) ‘ Total tritium added {cm?) 0.033 0.038 Total tritium added {(mCi) 85 97 - 2.3 FORCED-CONVECTION LOOPS W.R. Huntley M. D. Sllverman H. E. Robertson 'The Forced Convection Corrosion Loop Program is part of the effort to develop a satisfactory structural ‘alloy for molten-salt reactors. Corrosion loop MSR-FCL-2b is operating with reference fuel salt at 'typic‘:al MSBR velocities and temperature gradients to evaluate the corrosion and mass transfer of standard ‘Hastelloy N. Addition of tellurium to the salt in MSR-FCL-2b is planned after baseline corrosion data are obtained in the absence of tellurium. At this time, the loop has operated approximately 3000 hr at design AT conditions with the expected low corrosion rates. Two additional corrosion loop facilities, designated MSR-FCL-3 and MSR-FCL4, are being constructed. They are being fabricated of 2% titanium—modified Hastelloy N alloy which is expected to be more repre- sentative of the final material of construction for an MSBR than standard Hastelloy N. 2.3.1 Operation of MSR-FCL-2b Loop ‘FCL-2b was operated continuously for about 3000 hr from February to June 1975, under design AT (565°C minimum, 705°C maximum) conditions. During this period, standard Hastelloy N corrosion specimens installed in the loop in January 1975 were exposed to circulating fuel salt at three different temperatures, (565, 635, and 705°C). As expected, corrosion rates were low; the highest value was 0.1 mil/year (3um/year) at the highest temperature station. Salt samples taken at intervals have been analyzed for major constituents, metallic impurities, and oxygen 26 (Table 2.4). Except for an occasional high value for oxygen or iron, the analyses are relatively consistent and indicate that the observed corrosion processes have had very little effect on the concentrations of the various species present in the fuel salt. Analytical probe readings for the U*/U3 ratio, indicative of the redox condition of the salt, have been taken on a weekly basis. This ratio, which was about 7 X 10? at the beginning of the corrosion run, rapidly dropped to about 1 X.103 after the first 24 hr of operation. The ratio then gradu- ally fell to ~v1 X 10% by the end of March (1500 hr elapsed time), and it has remained at that level during the latter part of the operation. After the corrosion specimens were removed for the 3000-hr weight-change measurements, preparations were made for obtaining heat transfer data on the Li-Be-Th-U fuel salt (71.7-16-12-0.3 mole %). At this time' a Calrod electric tubular heater failure was dis- covered on the pipe line (12.7-mm-OD X 1.1-mm-wall) which runs from metallurgical station No. 3 to the inlet of cooler No. 1. After removing the thermal insulation, about 10 to 20 cm® of salt was found on the loop piping and the burned-out heater. Grainy material was present on the heater sheath at three locations directly opposite peeled-off sections of oxide layer on the Hastelloy N piping. A small crack ("v5 mm long) was found on a tubing bend directly under the failed heater. Whether the heater arced, causing the piping to fail, or whether the salt leak from the loop caused the heater burnout is uncertain at this time. Examination of speci- mens from these regions is continuing. The fuel salt was drained from the loop into the fill- and-drain tank after the leak was discovered. Analytical results on a sample taken from the tank indicated that no obvious contamination of the fuel salt had occurred. A new section of piping was installed (approximately 2.4 m, from metallurgical station No. 3 to the inlet to cooler No. 1)..During the shutdown, several defective thermocouples and two defective - clam-shell electric heaters were replaced. Ball valves were refurbished, numerous small repairs were made, and instruments were recalibrated. After the thermal insulation had been replaced, baseline heat loss measurements were made, with no salt in the loop, in preparation for taking heat transfer data. The loop was ready for refilling at the end of July, approximately four weeks after ‘the salt leak was discovered. After filling the loop, heat -transfer measurements were -obtained with flowing salt.” The ALPHA pump speed was varied from 1000 to 4600 rpm, resulting in salt flows of approximately 2.7 to 16 liters/min, which correspond to Reynolds numbers that vary from 1600 to 14,000. The lower limit for salt-flow was set to prevent freezing, and the upper limit was dictated by the power reqt{lred for driving the pump. At the lowest flow rate, unusual wall temperature profiles were noted, which probably were caused by entrance conditions and transitional flow effects. The heat trans- fer measurements were completed near the end of this reporting period, and analysis of the data'is in progress. The stringers containing the Hastelloy N -corrosion specimens were reinserted in the loop, and AT opera- tion (565°C minimum, 705°C maximum) was resumed in order to complete the-originally planned 4000-hr cor- rosion run. If no unusual corrosion behavior is encoun- tered in the next 1000 hr of operation, nickel fluoride (NiF,) additions will be made to the loop in order to raise ‘the oxidation potential of the salt to a level corre- sponding to a U*/U% ratio of about 10%, and a new set of corrosion specimens will be exposed. 27 Table 2.4. Salt sample analysis during MSR-FCL-2b operation with LiF-BeF, -ThF, -UF,“ Total hours Trace materials Date Major components Sa;lnple mpled , of sa.l.t (wt %) - (ppm) Notes s (1975) circulaticn L Be , - when sampled i e Th 9] F Fe Cr Ni O C S 1b 1-17 : 0 78 2.21 43.2 1.11 46.7 101 40 23 <50 11 9.9 Flush salt 2b 1-23 48 . 60 42 5.2 3b 1-28 0 7.99 1.74 43.0 098 46.3 75 70 15 125 29 43 New salt 4b 2-11 177 ' i 137 63 60 15 : 5b 2-18 355 154 64 68 45 6b 2-24 498 98 63 28 48 7b 33 676 7.68 2.36 428 1.05 46.3 147 67 35 45 23 17 8b 3-25 1146 7.01 2.55 43.0 1.04 45.8 256 59 57 <125 14 9 9b 4-16 1647 8.16 229 432 097 45.2 45 70 30 20 78 15 10b 5-12 2197 8.29 2.64 43.0 1.03 45.5 62 85 . 30 60 11ib 6-9 2838 8.23 2.25 42.7 1.00- 445 30 70 25 140 12b 6-23 3173 8.20 2.08 43.3 1.04 45.1 35 15 25 152 13b 7-3 3177 8.30 2.18 430 1.04 452 70 .80 40 30 Fill-and-drain ‘ ) ’ . ) tank 3246 7.28 2.03 1.00 450 45 85 70 58 14b 8-7 45.0 471.7-16-12-0.3 mole %. 2.3.2 Design and Construction of FCL-3 and FCL4 RN The design work for FCL-3 and FCL-4 was essentially completed; any changes or revisions which occur during construction of FCL-3 will also be made on FCL4. The piping support frame for FCL-3 was installed, and installation of electrical equipment is proceeding. Con- - duit lines have been run from the variable-speed motor- generator set on the ground floor up to the electrical rack installed on the experiment floor, and a sizable number of transformers, starters, switches, etc., have been installed. The instrument panel cabinets have been positioned, and cable trays are now being installed. Fabrication of two ALPHA-pump rotary elements and two pump bowls is 90% complete. A large number of completed items for both loops (e.g., dump tanks, auxil- iary pump tanks, cooler housings, blower-duct assem- blies, electric drive motors, purge gas cabinets, etc.) are on hand, awaiting installation. Fabrication of the titanium-modified Hastelloy N tubing for the salt piping of the loop is in progress. Part 2. Chemistry L. M. Ferris Chemical research and development related to the design and ultimate operation of MSBRs are still con- centrated on fuel- and coolant-salt chemistry and the development of analytical methods for use in these systems. Studies of the chemistry of tellurium in fuel salt have continued to aid in elucidating the role of this element in the intergranular cracking of Hastelloy N and related alloys. An important initial phase of this work involves the preparation of the pure tellurides Li, Te and LiTe, for use in solubility measurements, loop experiments, electroanalytical studies, and studies of tellurium redox behavior in molten salts. Techniques for preparing these tellurides have been developed, and experimental quan- tities have been prepared. Spectroscopic studies of tellu- rium chemistry in molten salts and of the equilibrium %H,(g) + UF4(d) = UF;(d) + HF(g) have also been initiated. In work using molten chloride solvents, at least two light-absorbing tellurium species have been shown to be present. These species are as yet unidenti- fied, but have compositions in the range Li,Te to LiTe, . Preliminary values of the quotients for the above equilibrium have been obtained, using Li;BeF, as the solvent. These values are in reasonable agreement with those obtained previously by other workers. " A packed-bed electrode of glassy carbon spheres was constructed, calibrated with Cd?* ions, and used in experiments with Bi®" ions in LiCI-KCl eutectic. It was concluded that this electrode was prototypic of one that could be used for the electroanalysis or electrolytic removal of bismuth, oxide, and other species in MSBR fuel salt. Preliminary experiments were also conducted to evaluate some questions relating to the mixing of fuel 28 and coolant salts. The results suggest that, on mixing small amounts of coolant salt with large amounts of fuel salt, the rate of evolution of BF; gas will not be intoler- ably high and that some oxide can be present in the coolant salt without effecting precipitation of UQ, or ThO,. Lattice enthalpies of first-row transition metal fluorides were calculated to provide a theoretical basis for evaluating thermochemical data for structural-metal fluorides. ' Work on several aspects of coolant-salt chemistry has continued. Analyses of condensates from the Coolant- Salt Technology Facility (CSTF) indicate that the vapor above the salt is a mixture of simple gases such as BF, HF, and H, O rather than a single molecular compound. Tritium concentrates in the condensates by about a factor of 10° relative to the salt. Studies of the system NaF-NaBF4-B,0; at 400 to 600°C show that at least two oxygen-containing species are present in typical coolant salt. One species is Na; B; F, O5, while the other has not yet been identified. The development of analytical methods for both fuel and coolant salt was also continued. An in-line voltam- metric method was used to monitor U*/U3* ratios in two thermal-convection and one forced-circulation loops. Two additions of tritium were made at the CSTF. The salt in the loop did significantly retain tritium, and the tritium ultimately appeared in the off-gas. Work was begun on using various electrodes for determining iron in MSBR fuel salt. Previous work had been conducted with solvents that did not contain thorium. Preliminary voltammetric experiments were conducted to identify soluble electroactive tellurium species in MSBR fuel salt. 3. Fuel-Salt Chemistry A.D. Kelmers 3.1 COMPOUNDS IN THE LITHIUM-TELLURIUM SYSTEM D.Y. Valentine A.D. Kelmers It has been demonstrated that tellurium vapor can induce shallow grain-boundary attack in Hastelloy N similar to that observed on the surfaces of the fuel-salt circuit of the MSRE.! However, the actual oxidation state or states in which tellurium is present in MSBR fuel salt, an LiF-BeF,-ThF4-UF; mixture, and the chemical reactions with the Hastelloy N surfaces remain to be determined. The lithium-tellurium system is being investigated to determine which Li-Te species can be present and to synthesize samples of all possible lithium tellurides. The solubility of these compounds in the fuel salt will then be determined. In addition, they will be used in spectrophotometric and electrochemical investi- gations of tellurium species in melts. During this report period, samples of Li; Te and LiTes were prepared. The preparations were made in an argon-atmosphere vacuum box equipped with an en- closed evacuated heater which held a molybdenum crucible. All handling of Li-Te compounds was done in inert-atmosphere boxes; sometimes the compounds were sealed under vacuum to minimize oxygen, nitro- gen, or H,O contamination. Lithium having an oxygen content of <100 ppm was supplied by the Materials Compatibility Laboratory, Metals and Ceramics Divi- sion. Tellurium metal of 99.999+ wt % purity was obtained from Alpha Ventron Products. The Li, Te was first prepared by dropping small pieces of lithium into molten tellurium contained in a molyb- denum crucible at 550°C. The reaction was extremely exothermic, emitting fumes and light flashes after each lithium addition. Solid formation occurred at lower lithium concentrations than expected from the reported phase diagram.? Further lithium additions continued to be absorbed after first melting on the surface of the solid phase. An amount of lithium necessary to satisfy the Li, Te stoichiometry was taken up in this manner. However, because of the loss of vapor and of some solid material which splashed out of the crucible during the early additions of lithium, it is doubtful that the stoichi- ometry was in fact preserved. The x-ray diffraction pattern showed a single phase, identified as Li, Te, having a face-centered cubic struc- 29 ture with a lattice parameter of 6.5119 + 0.0002 &.° The oxygen contamination in the product totaled about 375 ppm. Spectrographic analysis reported 0.5 wt % molybdenum present. Since the oxygen level and mo- lybdenum impurities were fairly low, a larger-scale prep- aration was attempted as well as a direct preparation of LiTe; by the same method. In both cases the product was contaminated unacceptably with molybdenum, and these preparations were discarded. Apparently the first preparation had affected the surface of the crucible such that the reaction with molybdenum was acceler- ated in these subsequent experiments. The molybdenum crucible was used for one further preparation, after cleaning and polishing the inside sur- face. The Li, Te was prepared from the lithium-rich side of the phase diagram by dropping tellurium into molten lithium. Since molybdenum is relatively inert toward lithium,* less reaction with the crucible was expected. In addition, this preparation could be made at a much lower temperature. The tellurium was added to the lithium in small increments with the temperature held at 250°C. Each of the first additions resulted in a smooth, quiet reaction with a solid phase forming on the bottom of the crucible. However, since completion of the reaction was not visibly apparent, the tempera- ture of the system was increased above the tellurium melting point to about 550°C to ensure that unreacted tellurium was not on the bottom of the crucible. More additions of tellurium were then made. Above 500°C a popping noise was heard after each addition of tellu- rium. After about three-fourths of the tellurium had been added, the system was mostly solid. As more tellu- rium was added, the amount of solid in the system became so great that further additions of tellurium were 1. A..D. Kelmers and D. Y. Valentine, MSR Program Semi- annu. Progr. Rep. Feb, 28, 1975, ORNL-5047, p. 40. 2. P. T. Cunningham, S. A, Johnson, and E. J. Cairns, J. FElectrochem, Soc.: Electrochem. Sci. Tech, 120, 328 (1973). 3. X-ray lattice parameters were measured by O. B. Cavin of the Metals and Ceramics Division. The value 6.5119 + 0.0002 A measured for Li, Te is in agreement with the value 6.517 A reported by E. Zintl, A. Harden, and B. Dauth, Z. Elektrochem, 40, 588 (1934). The value 6.1620 + 0.0002 A measured for LiTe, is in agreement with the value 6.162 A reported in ref. 2. 4. H. W, Leavenworth and R. E. Cleary, Acta Met. 9, 519 (1961). not covered by the liquid. Subsequent additions pro- duced light flashes and popping associated with the highly exothermic reaction as encountered in the pre- vious preparations of Li, Te. Finally, enough additional tellurium was added to the system to satisfy the Li; Te stoichiometry, and the system was allowed to cool to room temperature. ' Upon crushing the cooled product, four differently colored ‘substances were distinguishable: gray opaque material, wine-red to pink opaque material, colorless translucent crystals, and metallic tellurium. Analyses were performed separately on each type of material: 1. Gray opaque material. The x-ray diffraction pattern ~revealed Li,Te and LiTe3‘; no other lines were present. The oxygen level was about 218 ppm. Spec- trographic analysis indicated the presence of about 0.1 wt % molybdenum. . Red-hue material. Only a few crystals of all-red material could be isolated. The remainder of the red-hue material was ground together with some sur- rounding gray material. The x-ray diffraction pattern corresponded mainly to Li;Te. A small amount of LiTe; was also present: The oxygen content was reported to be about 275 ppm. Spectrographic analysis reported <0.01 wt % molybdenum. Colorless translucent and isolated red crystals. Both these products gave an x-ray diffraction pattern corresponding to pure Li, Te with no indication of a second phase. To ensure a uniform product, all the 'various colored materials were recombined and thoroughly mixed. The Li, Te mixture was then placed in a 2-in.-diam tungsten boat, which had previously been enclosed in a quartz bottle. The quartz bottle was then evacuated, sealed, and heated to 550°C for about 16 hr. The product obtained after cooling was almost completely cream- white. However, when the bottle was broken open, the product began to turn beige upon exposure to the envi- ronment of the inert-atmosphere box. The product was then crushed roughly and placed in sample bottles. On standing in the bottles, the product gradually reverted to the red-gray color it had been before the heat treat- ment, with the exception that the product in one bottle remained beige. The reason for the lack of uniform behavior is as yet unknown. Some of the darkened product was returned to the tungsten boat in another quartz bottle and the heat treatment repeated. It again turned the cream-white color. The products, both the light beige and the red-gray color forms, gave x-ray dif- fraction patterns for a single phase, Li, Te. Analysis of this Li, Te is given in Table 3.1. 30 Table 3.1. Analysis of Li, Te and LiTe, Li, Te LiTe, Li (wt %) 9.5=+0.1 1.7+ 0.1 Te (wt %) 89.9+0.5 984+ 0.5 Li, Te (mole %) . 98.7+ 1.6 LiTe, (mole %) 14+05 829: 5.0 Te {mole %) ) : 17.1 £ 15.0 X-ray diffraction ‘Single phase 740 (£ 10%) 0.05 <0.01 Single phase 275 (£10%) <0.01 " <0.01 Oxygen (ppm) Molybdenum (wt %) Tungsten (wt %) Red-gray Li, Te was mixed with the amount of tellu- rium required to satisfy the LiTey stoichiometry. The mixture - was then sealed in a quartz bottle under vacuum and heated to 550°C for 2 hr. The hot liquid was dark metallic gray: On cooling, the solid appeared bright silver-gray. The x-ray diffraction pattern con- firmed the presence of a single phase, LiTe;, having a near body-centered cubic structure with a pseudocell lattice parameter of 6.1620 + 0.0002 A.> The well- exposed Debye-Scherrer diffraction patterns suggest that the structure of this compound is more complex than previously reported.> Work will continue in an effort to describe this structure. The oxygen content was reported to be 275 ppm. Spectrographic analysis reported no molybdenum or tungsten contamination. Analysis of this LiTes is also given in Table 3.1. 3.2 SPECTROSCOPY OF TELLURIUM SPECIES IN MOLTEN SALTS B. F.Hitch L.M.Toth A spectroscopic investigation of tellurium behavior in molten salts has been initiated to identify the species present in solution and to obtain thermodynamic data which will permit the determination of the species’ redox behavior in MSBR fuel salt. A previous investiga- tion> had indicated that Te; is present in LiF-BeF, (66-34 mole %) on the basis of an absorption band occurring at 478 nm when LiTe; was the added solute; however, the work was terminated before these observa- tions had been fully substantiated. The current work is an extension of those earlier measurements which 5. C. E. Bamberger, J. P. Young, and R. G. Ross, J. [norg. Nucl. Chem. 36, 1158 (1974). should lead ultimately to a measurement of redox equi- libria such as LiTes + %H, + SLiF = 3Li, Te + SHF, (1) % Te, + LiF + 4 H, = LiTe; + HF . (2) These data should then permit the prediction of tellu- rium redox chemistry as a function of UF;/UF, ratio. During the past several months, most of the effort was devoted to assembly of the apparatus necessary for the fluoride measurements. This involved fabrication and assembly of the following: a furnace for the fluoride studies, diamond-windowed spectrophotometric cells, a vacuum and inert gas system, and a KHF, saturator through which H, is passed to generate HF-H, mixtures of known proportions. Also during the period of preparatlon some attention was given to a supporting study in chloride melts.® The advantages of working in chlorides are: 1. previous ground-work mvest1gat10ns have already been reported,’ : : 2. chlorides are easier to hold in silica cells without container corrosion, 3. the greater solubility of the tellurides in chlorides may reveal greater detail because of more intense spectra. Absorption spectra have been measured for Li,Te, LiTe5, and tellurium solutes as- well as during titrations of Li, Te with Te, in the LiCI-KCl eutectic at 450 to 700°C. These data indicate that at least two light- absorbing species are present in molten chlorides con- taining lithium tellurides with compositions in the range 31 Li, Te to LiTé,~ Furthermore, an exarhination of Te, in - the LiCl-KCl eutectic has indicated that there is a second species present besides Te, which is formed at high temperatures and/or high halide ion activity. More detailed experiments are anticipated using purer lithium telluride solutes in the diamond-windowed cell to demonstrate that the additional species are not related to impurities from the reagent or silica corrosion. 6. This work has done in cooperation with J. Brynestad of the Metals and Ceramics Division.’ 7. D. M. Gruen, R. L. McBeth, M. S. Foster, and C. E. Crouthamel, J. Phys. Chem. 70(2),472 (1966). 3.3 URANIUM TETRAFLUORIDE-HYDROGEN EQUILIBRIUM IN MOLTEN FLUORIDE SOLUTIONS* L. O. Gilpatrick L. M. Toth The equ111br1um ' UF4(d) + %Ha (g) = UF3(d) + HF(g) (1) is under investigation, using improved methods of analy- ses and control. The effects of temperature and solvent composition changes on the equilibrium quotient [UF3] Pyr . [UFa](p )/2 Q = (2) are thq immediate objectives of this work and are sought to resolve previous discrepancies noted in fuel- salt redox behavior.® : The procedure involves sparging a small (approx1- mately 1 g) sample of salt solution (UF, concentration of 0.038 to 0.13 mole/liter or 0.065 to 0.22 mole %) with H, gas at 550 to 850°C until partial reduction of UF, to UF'3 is observed; HF is added to oxidize the desired amount of UF; back to UF,. When an equilib- rium between the HF/H, gas mixture and the UF;-UF, in solution is reached, a spectrophotometric determina- tion of the UF; and UF,; concentrations is made. These data are combined with the analytically determined® HF/(Hg)l/z ratio to obtain the equilibrium quotient at a given set of conditions. The assembly -of the system for this experiment has been completed, and measurements of equilibrium quotients, using LiF-BeF, (66-34 mole %) as the sol- vent, have been initiated. Some "delay has occurred because of trace water in the HF/H, sparge gas which was responsible for the hydrolysis of uranium tetra- fluoride and the subsequent precipitation of UO,. The problem has been partially alleviated by treatment of the KHF, saturator, gas supply lines, and spectrophoto- metric furnace with fluorine at room temperature. How- ever, back diffusion of water vapor into the furnace from the exit gas line has also caused substantial solute losses and has been reduced by using higher HF-H, flow *This research in support of the MSBR Program was funded by the ERDA Division of Physical Research 8. L. O. Gilpatrick and L. M. Toth, “The Uranium Tetrafluoride—Hydrogen Equilibrium in Molten Fluoride Solu- tions,” MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047,p.43. rates. Together, these modifications have reduced the solute losses to an acceptable level (2% per day). Equilibrium has been achieved at 650°C for measured UF, /UF, ratios of approximately 0.2 X 1072 t0 3.1 X 1072, Although the UF; /UF, values are reproducible at fixed KHF, saturator temperatures, the analytically determined HF values are not as yet. Consequently, the standard error (approximately 50%) in the equilibrium quotients is still rather high. So far, a value of 0= 107° has been determined at 650°C, which compares favor- ably with the previous® value of 9.16 X 1077, Most of the immediate effort is being devoted to improving the precision of the HF determination. Tritium control in an MSBR would be favored by higher equilibrium quotients. In an MSBR, the UF;/UF, ratio will probably be fixed by equilibria involving the structural metals. The tritium inventory will be established by the tritium production rate and the various tritium removal processes. If @ is larger than previously anticipated, the partial pressure of HF would be higher and the partial pressure of H, would be lower than previously estimated. Thus, TH would be available at a lower concentration for permeation through the heat exchanger to contaminate the coolant loop (and ultimately the steam system), and a larger proportion of the tritium would be present as TF, which would be removed in the helium gas stream. 3.4 POROUS ELECTRODE STUDIES IN MOLTEN SALTS H. R. Bronstein F. A. Posey Work continued on development of porous and packed-bed electrode systems as continuous on-line monitors of the concentrations of electroactive sub- stances, especially dissolved bismuth, in MSBR fuel salt. In previous work,'®'! a prototype packed-bed elec- trode of glassy carbon spheres (100 microns in di- ameter) was tested in the LiClI-KCl eutectic system. Linear-sweep voltammetric measurements, carried out in the presence of small amounts of iron and cadmium salts, showed that the cell, instrumentation, and auxil- iary systems functioned successfully and demonstrated 9. G. Long and F. F. Blankenship, The Stability of Uranium Trifluoride, ORNL-TM-2065, Part II (November 1969), p. 16; Eq. 6 with xyg, = 0.002. 10. H. R. Bronstein and F. A. Posey, MSR Program Semi- annu. Progr. Rep. Aug. 31, 1974, ORNL-5011, pp. 49-51. 11. H. R. Bronstein and F. A. Posey, MSR Program Semi- annu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 44. 32 the sensitivity of this method of analysis. However, these measurements showed the need for redesign of the experimental assembly to permit removal and replace- ment of the cell and addition of substances to the melt. During this report period the redesigned packed-bed electrode of glassy carbon spheres was tested again in LiCI-KCl (58.8-41.2 mole %) eutectic, since the be- havior of a number of electroactive substances has al- ready been established in this medium. The packed bed of glassy carbon spheres was supported on a porous quartz frit and contained in a quartz sheath. Another porous quartz frit pressed on the bed from above. A glassy carbon rod penetrated the upper quartz frit to provide compaction of the bed and electrical contact with a long stainless steel rod which was insulated from the surrounding tantalum support tube. The electrode assembly was dipped into the melt so that the molten salt flowed up through the interior of the bed and out an overflow slot. By this means it was possible to obtain a reproducible volume of melt inside the packed-bed electrode. Voltammetric and coulometric scans of the pure melt at 395°C showed that the background current was small. A typical set of current-potential and charge- potential background curves is shown in Fig. 3.1. A 2-V ORNL-DWG. 75-11288 11]Tlrlllllillllllllllll =400 TEMPERATURE : ~395°C REFERENCE ELEGTRODE: Ag/AgCI{~0.1 M ) in —-300 LiCI-KCI {eutectic}/ Pyrex glass CADMIUM GONTENT : 0.064 coulombs /ml ~3.3 x10™*¢ ~37ppm | PAGKED-BED ELECTRODE VOLUME : ~1.75mi _ GLASSY CARBON SPHERE DIAMETER: ~10C microns w -200 2 [=] 5 3 E —-100 , (] o < I BACKGROUND © > 511 — o - o START 5 o a | O E £ 2 —E = - S START E o = I_’.._ - L & } E § BACKGROUND N S | A - cd®-» ca?t ER [&] +5 1.1 1 |[ 1 1.1 1 ll |- | | Lol .1 II [ S 1.0 0.5 0.0 -0.5 -1.0 -1.5 ELECTRODE POTENTIAL (w#s Ag/AgCl) (volis) Fig. 3.1. Linear-sweep voltammetry and coulometry of cad- mium in LiCI-KCl (58.8-41.2 mole %) eutectic with a packed- bed electrode of glassy carbon spheres. Curve A: current back- ground (sweep rate = 10 mV/sec); curve B: current with Cd** present (sweep rate = 5 mV/sec); curve C: background charging curve {(sweep rate = 10 mV/sec); curve D: charging curve with Cd?* present (sweep rate = 5 mV/sec). range of electrode potential could be swept without evidence of significant amounts of oxidizable or reduc- ible impurities in the melt. For calibration purposes, a known quantity of cadmium ions (Cd**) was added to the melt by anodization of molten cadmium metal con- tained in a specially designed graphite cup which could be lowered into the melt. The amount of cadmium added (Fig. 3.1) was monitored by use of an electronic autoranging coulometer. Following addition of cadmium, the voltammetric and coulometric scans indicated that only a small fraction of the known cadmium content inside the void space of the packed-bed electrode was being measured. After removal of the cell assembly, examination showed that the glassy carbon contact rod had somehow fractured, possibly due to excessive pressure from the mating stainless steel contact rod, and resulted in loss of elec- trical contact with the packed-bed electrode. The cell assembly was then redesigned and rebuilt to permit electrical contact to be maintained without undue pressure and to allow accurate measurement of the working volume of the packed-bed electrode. The new design was similar to that of the previous cell, except that the upper fritted quartz disk was perma- nently sealed to the surrounding quartz sheath. A small hole in the center of the disk permitted loading of the glassy carbon spheres into the electrode assembly and provided accurate positioning of the glassy carbon con- tact rod into the bed. Prior to loading of the spheres, the volume contained between the porous quartz disks was measured with mercury. Some voltammetric and coulometric scans in the pres- ence of cadmium ions are shown in Fig. 3.1. Asin pre- vious studies in aqueous media with the packed-bed electrode,'? more accurate analytical results were obtained on the anodic half cycle (stripping) than on the cathodic half cycle (deposition). Approximately 40 mC of cadmium was estimated to be within the packed- bed electrode. The coulometric results shown in Fig. 3.1 are quite consistent with this value. Thus it is possible, knowing the geometry of a packed-bed electrode, to estimate the response and sensitivity within reasonable limits (the accuracy of estimation depends upon void fraction, the accuracy of the volume measurement, and other factors). Repeated scans over a period of many days showed good reproducibility and also established that diffusion through the quartz frits during the time of measurement (only a few minutes) has very little effect on the results. 12. H. R. Bronstein and F. A. Posey, Chem. Div. Annu. Progr. Rep. May 20, 1974, ORNL4976, pp. 109-11. 33 Another cell was packed with 200-u-diam glassy carbon spheres and used to obtain the results shown in Fig. 3.2. In this case a quantity of Bi*" ions had been anodized into the melt in a manner similar to that used for cadmium. At the time of these measurements the same melt had been in use for many weeks. Fig. 3.2 shows voltammetric and coulometric anodic stripping curves in the vicinity of the anodic peak for stripping of bismuth which had previously been deposited on the internal surfaces of the electrode during the cathodic half cycle. In agreement with observations of others, we found that volatility of BiCl; precluded close corre- spondence between added and observed quantities of bismuth, and that the bismuth peak decreased steadily with time. The appearance of the bismuth peak suggests that possibly some alloying of bismuth with the cad- mium took place. Other experiments on bismuth reduction and strip- ping will be carried out in the future in which cadmium, used for calibration of the cell system, is absent. In addition, the present apparatus will be used to study the electrochemistry of lithium telluride in the LiCl-KCl eutectic. Observations on the tellurium system in the chloride melt may be useful in interpretation of tellu- rium behavior in later studies with MSBR fuel salt. The ORNL-DWG, 75-11289 T T T T T T T T T TEMPERATURE : 392°C REFERENCE ELECTRODE : Ag/AgCH(~OiM) in LiCI-KCI (eutectic)/Pyrex glass i GLASSY CARBON SPHERE DIAMETER: ~ 200 microns 00 — O - 3 A — - g —— — ¢ < - P4 g F l ,/'—’ ”” ] (E) g T ~2mC LT 13 = 051 — 2 E = - 1E Bio——D Bi3+ — =N Tw 0O 2+ W i 4@ x | Possibly Fe2+-+ Fe3+ Cd"—Cd 1% 3 1ok from Iron ;mpurity T 5 5mC A 1.5 | | 1 1 | | | | l 1 04 03 02 01 00 -0 -0.2 -03 -0.4 -05 -06 ELECTRODE POTENTIAL (vs Ag/AgCl) {volts) Fig. 3.2. Linear-sweep anodic stripping voltammetry and coulometry of bismuth electrodeposited onto a packed-bed electrode of glassy carbon spheres. Solid lines: experimental current-potential and charge-potential curves in the region of the bismuth stripping peak; dashed lines: estimated background charging curves. capability of the packed-bed . electrode of glassy carbon spheres for monitoring electroactive species in molten salts has been shown to be satisfactory. Consequently, plans are now under way for design and fabrication of cells and apparatus for testing the electrode system in molten fluoride media including MSBR fuel salt. In bismuth-containing fluoride melts, whether bismuth is present as Bi® or Li; Bi, or both, it should be possible to identify and determine the quantities of each species. The packed-bed. electrode offers hope of removing, as well as monitoring, dissolved bismuth in the fuel salt which may be present as a result of the reductive extrac- tion process for removal of fission products. 3.5 FUEL SALT—COOLANT SALT INTERACTION STUDIES A.D.Kelmers D.E. Heatherly In the alternate coolant evaluation,!® several areas of potential concern were defined with regard to the appli- cability of the conceptual design coolant salt [NaBF, - NaF (92-8 mole %)] for MSBRs. These centered primar- ily around events associated with off-design transient conditions, particularly primary heat exchanger leaks, which would allow intermixing of fuel salt and coolant salt. If coolant salt leaked into the fuel salt, the quan- tity and rate of evolution of BF; gas from reaction (1), NaBF4(d) coolant - BF3(g) + NaF(d)fuel salt » (1) would determine the transient pressure surges to be en- countered in the heat exchanger and reactor. Also, pre- vious work!'* indicated a substantial redistribution of the ions Li*, Na*, Be**, F~, and BF,” between the result- ing immiscible two-phase system formed on mixing Li,BeF4 and NaBF,. The solubilities of UF, and ThF, have not been measured in such systems; thus the dis- tribution of uranium and thorium between such phases and the resulting concentrations are unknown: In addi- tion, if oxide species were present in the coolant salt, either deliberately added to aid in tritium trapping or inadvertently present due to steam leaks in the steam- raising system, the precipitation of UO, following mix- ing of an oxide-containing coolant salt with fuel salt has not been investigated. Therefore, a series of experiments were carried out to investigate these areas. 13. A. D. Kelmers et al., Committee Report: Evaluation of Alternate Secondary (and Tertiary) Coolants for the. Molten- Salt Breeder Reactor (in preparation). 14. C. E. Bamberger, C. F. Baes, Jr., J. P. Young, and C. S. Sherer, MSR Program Semiannu. Progr. Rep. Feb. 20, 1968, ORNL-4254, pp. 171-73. 34 The experimental apparatus consisted of a quartz vessel heated by a quartz furnace so that the volume of the resulting phases could be observed at temperature and measured with a cathetometer. The quartz vessel extended up out of the furnace and was closed with an O-ring fitting and end plate. A nickel stirring shaft, driven by a constant-speed dc motor, penetrated the end plate and during the tests was driven at a speed adequate to stir the two phases without appreciable visible dispersion. Access for sample filter sticks was provided through the end plate as was done also for the argon inlet and exit lines. A very low argon flow main- tained an inert atmosphere over the melt during the experiment. Predetermined weights of fuel salt [nominal composi- tion LiF-BeF, ThF,-UF, (72-16-11.7-0.3 mole %)] and coolant salt [nominal composition NaBF,-NaF (92-8 mole %)] were placed in the quartz vessel and rapidly heated to 550°C. Bubbles of gas could be observed due to BF; generation via reaction (1) as soon as the coolant salt melted during the heatup period. When the temperature reached 550°C, counted as time zero, stir- ring was initiated. The volume of the phases was period- ically determined, and filter-stick samples were taken at 30- or 60-min intervals. ‘ The reaction between the fuel salt and coolant salt proceeded slowly: approximately 30 to 60 min was required to complete the visible evolution of BF, gas at 550°C. When the initial coolant salt content was 20 wt % or less of the total material, no coolant salt phase remained after approximately 1 hr. All the NaF dis- solved in the fuel salt phase and all the BF; gas left the reaction vessel. With larger initial weights of coolant salt, up to 50 wt %, a small residual volume of coolant salt phase could be observed after 1 to 3 hr. Severe corrosion of the quartz reaction vessel occurred at the interface between the coolant salt phase and the argon cover gas in the experiments with the larger initial weights of coolant salt, presumably due to attack of the quartz by BF; via a reaction such as 2BF;(g) + % Si02(c) =% SiF,4(g) + B, 05(d). ‘ (2) In experiments 6 and 7, holes were corroded completely through the vessel wall, and the surface of the stirred molten coolant salt phase was exposed to air for I to 2 hr at 550°C. ' : In most of the experiments, samples of the fuel-salt phase were withdrawn at intervals of 1, 2, 3, and 4 hr. Samples were also taken after the conclusion of the experiment; after the melt had cooled to room tempera- ture the quartz vessel was broken away from the solid salt. All these samples gave essentially identical analyti- Table 3.2. Composition of fuel-salt phase and coolant-salt phase after contact at 550°C Experiment Initial mixture (wt %) Fuel-salt phase (mole %) Coolant-salt phase (mole %) No. Fuel salt? Coolant salt? LiF NaF BeF, ThF, UF, LiF NaF BeF, ThF, UF, Na, SiF, B,0, NaBF, 2 100 0 68.8 2.3 17.2 11.5 0.24 3 90 10 633 93 167 105 0.24 } . 4 80 20 604 123 16.5 10.6 0.25 5 70 30 544 190 15.5 10.9 0.24 21.2 19.1 4.4 0.18 0.013 d 55.1 0 -6 60 40 508 263 13.1 9.6 0.22 128 36.6 3.8 0.50 0.017 3.1 38.6 4.7 7 50 50 44.0 36.6 8.9 10.4 0.20 94 439 3.9 0.74 0.028 1.9 23.3 16.9 2Nominal composition LiF-BeF, -ThF,-UF, (72-16-11.7-0.3 mole %). bNominul composition NaBF, -NaF (92-8 mole %). “No coolant-salt phase remained. dNot analyzed. S€ cal values; therefore, the fuel-salt phase analyses (Table 3.2) represent an average of 3 to 5 values. Further sup- port for the contention that reactions involving the fuel-salt phase were complete in 60 min or less is shown by the plots of volume vs time in Fig. 3.3. The coolant- salt phase volume decreased rapidly for about 30 min due to reaction (1); thereafter, the volume change was slower, presumably due to reaction (2). It was impossible to obtain coolant-salt phase samples with the filter sticks, both because the phase volume was small and the salt tended to drain out of the filter sticks. Therefore, all coolant-salt phase analyses (Table 3.3) were from samples obtained after completion of the experiment and represent only single values. The analyses (Table 3.3) show substantial redistribu- tion of the ions Li*, Na', and Be®* between the two phases. Thorium and uranium exhibited low solubility in the coolant-salt phase. Neither NaBF, nor the oxy- genated fluoroborate compound (represented as B,0, in the table) was soluble in the fuel-salt phase. Fuel salt stirred in contact with a coolant-salt phase containing up to about 50 mole % B, O; showed no precipitation of UO,. The coolant phase compositions were ex- ORNL-DWG. 75-13748 | [ I | [ 60 W o {ml) H O u!t = S 301 O = 20— Coolant Salt Phase 10— — o l | ] | ] 0 20 40 60 80 100 120 TIME (min) Fig. 3.3. Volume of coolant-salt phase and fuel-salt phase vs time in mixing experiment No. 6. Initial mixture was 60 wt % fuel salt and 40 wt % coolant salt. After heating to 550°C, stirring was begun and the depth of the two phases was periodi- cally measured. 36 Table 3.3. Composition of coolant-salt phase expressed as components of the ternary system MF-B, O, -NaBF, [nitial mixture Coolant-salt phase (wt %) (mole %) Fuel salt Coolant salt MF4 B,O, NaBF, 70 30 47.5 525 0 60 40 64.2 31.9 3.9 50 50 65.3 20.1 14.6 MF = Na — NaBF, + Li+ 2Be + 4Th + 4U + 48i. pressed (Table 3.3) in terms of the ternary system MF-B,0;-NaBF,;. The compositions are close to the glass-forming regions of the ternary phase diagram'3 for the system NaF-B,0;-NaBF,, where discrete com- pounds have not been established. The following pertinent observations can be made: 1. The rate of evolution of BF; gas on mixing was low; presumably the rate-limiting step is the transfer of NaF across the salt-salt interface. Thus, in a reactor system with turbulent flow, the release would be more rapid; however, these results are encouraging relative to MSBRs in that very rapid gas release re- sulting in significant pressure surges was not experi- enced. . No tendency was observed for the fuel salt constit- uents thorium or uranium to redistribute or to form more concentrated solutions, or to precipitate fol- lowing mixing of coolant salt into fuel salt. These experiments do not yield information relative to mixing fuel salt into coolant salt, since it was impossible to contain predominantly coolant-salt phase mixtures in quartz at 550°C. Thus the ques- tion of uranium (and/or thorium) precipitation as UF,-NaF complexes, as observed!® in an engineer- ing loop, remains unresolved. . Apparently an oxide species forms in the coolant-salt phase which is more stable than UQO,, since no UO, precipitation was observed. Thus, large amounts of oxygenated compounds could be added to the flu- oroborate coolant salt for the purpose of seques- tering tritium, since leakage of such a coolant salt into the fuel salt would not lead to uranium or thorium precipitation. 15. L. Maya, Sect. 4.1, this report. 16. H. F. McDuffie et al., Assessment of Molten Salts as Intermediate Coolants for LMFBR s, ORNL-TM-2696 (Sept. 3, 1969), p. 20. 3.6 LATTICE AND FORMATION ENTHALPIES OF FIRST-ROW TRANSITION-METAL FLUORIDES* S. Cantor The primary purpose of this investigation is to provide a theoretical basis for critically evaluating the thermo- dynariic data that will be obtained in an experimental program recently initiated with Division of Physical Research . funding. In the experiments, free energies of formation will be deduced from emf measurements of solid-electrolyte galvanic cells. The first-row transltion metals include common structural metals (Fe, Ni, Cr) and other metals (Ti, V)-which may be-used in fission or fusion reactors. When' these metals are corroded or otherwise oxidized in fluoride media used in these reac- tors, metal fluorides are formed; reliable thermo- dynamic information for these compounds is valuable in predicting their chemical behavior in the reactor system. For a metallic fluoride, MF,, (where n is the valence of the metallic ion), the relationship between lattice enthalpy AH; and enthalpy of formatron AHy is given by the equation’ AHy = AHy - AHy+ —nAHp. (1) The lattice enthalpy is the heat of the reaction M™(g) + nF(g) =MF,(c) (2) at 298.15°K. The lattice enthalpy is very similar to the lattice “‘energy,” the latter being somethat more diffi- cult to obtain from experimental information: AHy is the standard heat of formation of MEF, (c) AH,,+ is the standard enthalpy of formatron at 298.15°K, of the gaseous cation and electrons (e) formed from the crystalline metal: M(c)—>Mn+(g)+.ne'(g). _ o - | (3) AHF‘ is the standard enthalpy at 298.15°K of a mole of gaseous fluoride ions formed from the ideal gases, electrons, and d1atom1c fluorine: B+~ Flo)- @) The enthalp1es of formation for reactrons (3) and (4) are deduced mostly from atomic or molecular data. AH,+ is obtained by summrng the first n ionization potentials of M and its enthalpy: of sublimation and coverting these quantities, where necessary, to 298.15°K; values of AH,+ are given in Tables 3.4 and 3.5. The enthalpy of reaction (4) at 298.15°K is —61.24 *This research'in support of the MSBR Program was funded by the ERDA Division of Physical Research. Table 3.4. Standard enthalpies of formation of 3d divalent fluorides (AHf) and their cations (AH2+), lattice enthalpies (AHL) . _AH _AH,. —AH Fluorlde "(kcal/mole) (kcal/mole)? (kcal/m{;le) CaF,” . 291.5% 460.3 629.3 ScF, (239)° 540 (657 TiF, (217)¢ 586.6 (681)¢ . VF, (208)¢ . 620.0 (706)¢ CrF, 1869 + § 6353 699 + 5 MnF, 205.4 + 1€ 602.2 685.1 + 1 FeF, 170€ £ 658.3 706 + 5 CoF, 160.5 + 18 682.3 7203 %1 NiF, 157.2 = 0.47 703.5 7382+ 0.4 CuF, 131.2 £ 0.8¢ 732.3 741.0 £ 0.8 ZnF, 182.7¢ 665.1 725.3 Zlonization potentials trom C. E. Moore, NSRDS-NBS 34 (1970); enthalpies of sublimation from ref, 19; valence state preparation energy corrections from ref 20. PNBS Technical Note 270-6 (1971). CEstimated in this investigation. dNBS Technical Note 270-4 (1969). eT._N. .R_ezul(hina et al., J. Chem. Then‘nodyn.‘ 6, 890 (1974). fDerived from Asolid galvanic-cell emf data given in W. H. Skelton and J. W Patterson J. Less- Common Metals 3l 47 (1973). EJANAF: Thermochemical Ta bles, (1971). ME. Rudzitis et al. ,J. Chem. Eng. Data 12,133 (1967) 'NBS Technical Note 270-3 (1968). 2d ed., NSRDS-NBS 37 kcal per gram-ion; it is based on Popp’s value!” (3.400 eV) for the electron affinity of fluorine, the dissociation energy (1.58 eV) measured by Chupka and Berkowitz,'® and the enthalpy difference of F, (ideal gas), H, 95 — Hy, listed by Hultgren et al.!? : The lattice -enthalpies of the divalent fluorides are listed in Table.3.4 and are plotted against atomic num- ber in Fig. 3.4. The curious double hump has been inter- preted??.in terms of ligand-field.theory. By this theory, differences between actual values of AH; and those lying on a smooth curve-drawn to fit the data of CaF,, MnF,, and ZnF, are primarily due to ligand-field stabi- lization energy (LFSE). For this series of compounds, 17. H. P. Popp, Z. Naturforsch. 22a, 254 (1967). 18. W A, Chupka and J. Berkowitz, J. Chem. Phys. 54 5426 (1971) "'19. R. Hultgren et rrl Selected Values of the Thermo- dynamic Properties of the Elements, p. 177, American Society of Metals; Metals.Park, Ohio, 1973. - 20 P, George and D. S. McClure, p. 381 in Progress in Inorganic Chemistry, vol. 1, F. A. Cotton, ed., Interscience, New York, 1959, 38 . ORNL-DWG. 75-i37'49 5] S I o e ~ 700 Q © E ~ o QL X -~ I ¥ 650 ¥ S i T O N T I d® d' d¢ d3 d* d°> d® d7 d8& qd° d'" Fig. 3.4. Lattice enthalpies of 3d divalent fluorides. Solid circles (e) are experimental AH; calculated from Eq..(1); error . Table 3.5. Lattice enthalpy (a#/,) and standard enthalpy of formation (AH ;) of 3d trivalent fluorides; standard enthalpy ol"}ormatlon of 3d cations (AH,.) : . —AH —AH3+ —AH : A_Fluonde (kcal/mole) (kcal/mole)? (kca]/mldle) " ScF, 394.1 1 20 1112 13224+ 2 - TiF, 336.1 + 3.5¢ 1222 1374 + 3.5 “VE, 2959+ 5 1298 . 1409 . CrF, 277¢ 1351.6 1445 " MnF, 238+ 7€ 1381 1436 £ 7 " FeF, 249 + 3¢ " 13654 14313 CoF, 193.8¢ 1455.6 1465.7 NiF, (165)% 1515.5 (1497)8 -CuF, (120¥ 1584 (152088 GaF, 278M 1389 1483 é[oniiation potentials from C. E. Moore, NSRDS-NBS 34 " (1970); enthalpies of sublimation from ref. 19; valence state _ preparatlon energy correctlons from ref 20. €Q. Kubaschewskl et al., bT N. Rezukhma et al., J. Chem. Thermodyn 6, 890 (1974) pp. 334, 354 in Metallurgical Thermo- chemistry, 4th ed., Pergamon, Oxford, England, 1967. dDer;ved from solid galvanic-cell emf data given in W. H. " Skelton and J. W. Patterson, J.. Less-Common Metals 31, 47 bars are uncertainties in AH . Open circles (o) are AH; minus - . ligand-field stabilization eneIgy. Triangles (A) are. AH; minus - both LFSE and Jahn-Teller energy. Solid. squares (w) were esti- mated by adding LFSE plus 3 kcal/mole as an empirical correc- tion to the smooth curve. ; the ligands are fluoride ions octahedrally coordinated (1973). -ENBS Tebhméai Note 270~ 4 (1969). fJANAF Thermochemical Tables, 2d ed., NSRDS-NBS 37 (1971). BEstimated in this investigation. hNBS Technical Note 270-3 (1968). ‘open circles) above the smooth curve by 5 and 7 _ kcal/mole respectively. Both CrF, and CuF, (and to a (except for CaF,) to the cation; the octahedral “field” - - of fluoride ions acts to “stabilize” "the. splitting of d-electron energy. levels of the metal ions. In a spheri- cally symmetrical field, the d-electron levels would be degenerate, that is, be at the same energy. In Fig.. 3.4, the smooth curve drawn through, for examp]e NiF, represents. the AH; NiF, would have if the field of lesser extent, FeF,) are known from crystal structure data to exhibit major tetragonal departures from octa- “hedral coordination geometry: This is attributed to the * Jahn-Teller effect,? which confers an additional stabili- " zation energy. (Jahn-Teller energy is abbreviated herein . as JTE.) For CuF, and FeF,, the JTE can be derived optically (Table 3.6). When LFSE and JTE are addi- - tively applied to CuF, and CrF;, the lattice enthalpy is . overcorrected as is shown by the triangles in Fig. 3.4. fluoride ions around the nickel cation were spherically symmetrical. Octahedrally coordinated cations with unfilled, half-filled, or fully filled 3d orbitals will not have a ligand-field stabilization energy; hence a smooth The methods outlined above can be applied to predict AH; for VF, and TiF,. Neither compound would be - expected to have a significant JTE. The formula used is curve is drawn through Can(dO) Man(d ), and' ZnF,(d'%). - ‘spectra. For Fer, NiF,, and CoF,, subtraction of optically derived LFSE (Table 3.6) from AH[ yields values (open circles in Fig. 3.4) which are above the smooth curve by 2 to 3 kcal/mole. Similar LFSE sub- AH; (kcal/mole) = AH; "+ LFSE+ 3, (5) ~ where AH; ' is the value for the compound lying on the Values of the LFSE can be deduced from optical tractions fo; CrF, and CuF, yi_él_d values (dedptéd by ° 1972. smooth curve in Fig. 3.4. The 3 kcal/mole on the right- “hand side of Eq. (5) is an empirical correction reflecting pp. 590-93 in Ad- , Interscience, New York, 21. F. Al Cotton and'G. Wilkinson, vanced Inorganic Chemistry, 3rd ed. 39 Table 3.6. Ligand-field parameter (10Dq}), stabiliiation energy (LFSE), ' Jahn-Teller splitting (5) and energy (JTE) for the di- and trifluorides of 3d metals Electron Fluoride 10Dq LFSE 5 JTE levels » . (cm ™' )2 (kcal/mole) (cm ™' )? (kcal/mole) d! " ScF, ' ' (—12)P Lo TiF, - 16,000¢ —-18.2 910 - 1.7 d* TIF, (10,7004 (-24) : VF, 15,900¢ -36.4 d? VE, 11,000 (—38) - CIF, 14,600 ~50.1 d* CrF, " 11,000 —18.9 (-11)8 MnF, . 17,400 ~29.8 - 9000 -12.9 d* FeF, 6,900 ~7.9 . 1400 — 2.7 CoF, 11,400 ~13.0 : d? . CoF, .- 7,200 ~16.5 NiF, . . 16,200/ -37.0 das NiF, . 7,400 —25.4 CuF, 14,100 _48.4 o d® CuF, - 7,400 ~15.3 7500 - - —10.7 2Values given in-D. Oelkrug, Struct. Bonding (Berlin) 9, 1-26 (1971) unless otherwise indicated. ' » bLFSE estimated very roughly as % of TiF, . €Based on K, NaTiF, . , dEstimated by assuming 10Dq is %, that of TiF, . €Based on (NH, ), VF,. TEstimated by Jorgensen’s method: 10Dgq=g X k;g=12,300 cm ', k=0.9. ERough estimate from JTE of CuF, . N Based on K, CoF,. IG. C. Allen and K. O. Warren, Struct. Bonding (Berlin) 9, 107 (1971). ORNL- DWG. 75-13750 R 1550 | I 1500 1450 B Q Q —AHL {Kcal/mole) r— s by E o I PO | ScFy TiFy VF3 Crfy MnF3 FeF3 CoFy NiFy CuF3 (ZnFz) GaFy dO dl dZ d3 d4 d5 dG d7 d8 d9 d10 Fig. 3.5. Lattice enthalpies of 3d trivalent fluorides. Solid circles (®) are experimental AHL calculated from Eq. (1); error bars are published uncertainties in AHp . Open circles (0) are AHL minus ligand-field stabilization energy. Solid squares (=) are estimated by adding LFSE to the smooth curve. the difference between theoretical and thermochemical lattice enthalpies for NiF, and CoF,. The standard enthalpy of formation (AHy) for TiF, and VF, is then obtained from Eq. (1) and is listed in Table 3 .4. 40 Analogous considerations were applied to study AH| for trivalent fluorides. The data and results are pre- sented in Table 3.5 and in Fig. 3.5. The double-hump pattern of the data is evident in Fig. 3.5. Subtraction of LFSE (given in Table 2.6) yields very satisfactory agree- ment between theoretical and experimental lattice enthalpies of VF; and CrF;; the agreement for TiF; (and for CoF,) is less satisfactory. As may be seen by the open circle below the curve in Fig. 3.5, subtraction of LFSE from AH; overcorrects MnF;. This is some- what surprising, since MnF,, with its 3d* electronic configuration for Mn**, also has a sizable JTE (Table 2.6). If the JTE were also subtracted, the discrepancy from the smooth curve would be much greater. In short, the thermochemical data for MnF; are questionable. In estimating AH; and AHy for NiF; and CuF, (Table 3.5), only the LFSE was added to the spherically symmetrical values (i.e., smooth curve values) of AHy. In other words, Eq. (5) was applied without the empiri- cal correction of 3 kcal/mole. With regard to the AHy of the structural-metal fluo- rides, the theory, as applied above, suggests that there is little need to determine AHf for NiF,. Moreover, from the value of AHy of TiF, obtained in this study, it is understandable why TiF, has never been prepared as a pure solid; it can be easily shown that TiF, would readily disproportionate to TiF; and Ti. However, a more accurate experimental determination of AHf for TiF; would be desirable for both practical as well as theoretical reasons. The same may be said for VF,, VF;, CrF;, CrF;, FeF,, and FeF;. 4. Coolant-Salt Chemistry A.D. Kelmers 4.1 CHEMISTRY OF SODIUM FLUOROBORATE L.Maya W.R. Cahill* The composition of the condensable fraction of the vapor phase in equilibrium with molten fluoroborate can be defined by the system H; OBF,-HBO, ‘H, 0, as described in the previous report.! The work done ‘dur- ing this report period was aimed at spectroscopic identi- fication of the molecular species present. The 1B NMR as well as IR and Raman spectra of BF3;2H20, ' HBF,(OH),, and of other intermediaté compositions was obtained. Dihydroxyfluoroboric acid (DHFBA) participates in éxchange processes which could be dés- cribed by the following equilibria: 2HBF2(OH)2 = BF3 'VHQO + H3 BO3 , BF3'H2 O+ H3BO3 = BF3 '2H2.0 + HB()2 The presence of H3BO3; and HBO, was detected by IR and Raman spectra, and the pronounced broadening of the '°F and '!B NMR Signals is an indication of ex- change processes. The Raman spectrum of DHFBA indicates that this compound is a tetrahedral molecule. Exchange processes were not detected for BF3-2H, 0. This compound appears to be stable at room tempera- ture. The structural information derived from the Raman spectrum, which identified BF3-2H,0 as a © *QRAU summer -participant. 1. L. Maya, MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047,p.47. tetrahedral molecule, agrees with the x-ray structural determination? of this compound. Additional samples of condensate collected during the operation of the Coolant Salt Test Facility (CSTF) were analyzed (Table 4.1). Silicon is present because of attack on the glass trap used to collect the condensate. Variations in the chemical composition of the samples can be interpreted as an indication that the condensed - material is not a single molecular compound but, rather, a mixture formed by combination of the simpler gas- eous. species present in the system, that is, H, O, HF, ~and BFj;. The relatively high tritium content of these fractions should be noted. Tritium is present in the system, since some of the Hastelloy N in the loop was originally used in the MSRE. The condensates show a tritium concentration factor of about 10° relative to the salt, suggesting that fluoroborate coolant salt [NaBF;-NaF (92-8 mole %)] may be an effective means ~of concentrating and conveying tritium out of the system. Attempts were made to generate a condensable frac- tion in laboratory-scale experiments by heating coolant salt containing up to 200 ppm H as NaBF; OH to 400°C in a closed system equipped with a cold finger. The OH™ concentration in the salt decreased to 50 ppm, and the composition of the condensate in a typical run was 53.2% H, OBF,, 14.8% (H,0), SiF¢ and 32% free water found by difference. The boron concentration in the condensed material did not reach as high a level as in 2. W. B. Bang and G. B. Carpenter, 4cta. Cryst. 17, 742 (1964). . . Table 4.1. Analyses of CSTF trap condensates Chemical comnposition b Tritium content Sample Operation period Amount? | (Wt %) (mCi/g) "H,OBF, HBO, SiF, 2 1 1972 Not avail. 60.4 15.7 Not det. 0.8 to 3.0¢ 2 1/14/75 — 1/24/75 100 mg 92.3 0 2.1 5.7 3 3/14/75 — 4/15/759 25¢g - 84.1 12.4 4.0 34 4 4/15/75 — 5/6/75 800 mg 83.0 12.1 0.1 0.6 aApproximate amount, Some of the material remained in the trap. bDifference from 100% is H, O. CGiven as a range. Apparently moré than one sample was anatyzed for tritium content. Data from A. S. Meyer and J. M. Dale, Anal. Chem. Div. Annu. Progr. Rep. Jan. 1974, ORNL4930, p. 28. dThe loop was not in operation between 1/24)75 and 3/14/75. 41 the CSTF samples, and there was considerable cor- rosion. Nevertheless, these experiments showed a pos- sible mechanism for the conversion of dissolved NaBF, OH into a volatile fraction. o An apparatus was assembled to measure the vapor density of BF;:2H,0 and related compounds at ele- vated tempérgtures to determine the degree of dissocia- tion of these materials. This work tested the hypothesis that the condensable materials collected in the opera- tion of the CSTF are completely dissociated at oper- ating temperatures (400—600°C) and only combine to form more complex molecules in the colder parts of the system. The procedure consists in measuring the pres- sure developed in a closed system containing a known amount of BF;+2H, 0 or DHFBA in order to establish the degree of dissociation according to the equilibrium described below: BF3'2H20 =BF3 + 2H20 At this time, volumes in the apparatus have been deter- mined, and pressure determinations have been made using argon as a test gas. Initial runs with BF,+2H,0 indicate that this compound may be completely dis- sociated at 400°C. _ Work on determining the oxide species present in molten fluoroborate is being continued, and the survey' of the system NaF-NaBF,-B,0; at 400 to 600°C has been extended to include IR and x-ray diffraction analyses in addition to physical and chemical observa- tions of the behavior of selected compositions. The observations indicate that there are three main areas in the system: 1. A region of compositions in which BF; is evolved. This occurs with compositions having a deficiency, in terms of equ1molar ratios, of NaF relative to the B, O3 present. . A region of compositions in which stable glasses are formed on cooling. This corresponds to mixtures containing more than 33 mole % B, 0,. . A region in which crystalline phases and glasses co- exist. decreases with decreasing B, 04 content, Usually coolant salt [NaBF,-NaF (92-8 mole %)] contains relatively small amounts of oxide, up to 1000 ppm, and its composition lies within area 3; thus, work has been directed toward characterizing the oxide species in this area. At least two species were present; one formed at the boundary of the glass area (high oxide content), and the other was Na; B3 F¢ O3, .which formed in compositions having NaF:NaBF,:B,03 mole ratios of 2:2:1 and 2:4:1 and was possibly present in 42 The tendency to form glasses on cooling compositions containing as little as 3 mole % B,0;. Experiments at the 1.5 to 4.0 mole % B,0; level, approaching the coolant composition, have been im- peded by the relatively low sensitivity of IR and x-ray powder diffraction. The difficulty with IR, using the KBr pellet method, arises from the fact that, at these oxide levels, the only band not covered by BF,~ absorptions is the one at 810 cm™!. This band has a relatively low absorptivity, and it is common to NaBF;0H, Na,B,F;0, Na;B;F,0,, and possibly other BOF compounds, although the intensity and line shape are different for each compound. A more certain IR identification can be made only when at least two typical bands can be idéntified (presently observable only at higher concentrations), as was the case in the identification of NayB3;F40, at an oxide level corre- sponding to 14 mole % B,0j;. Difficulties with x-ray diffraction arise from the low sensitivity of this tech- nique coupled with the fact that the species have a ten- dency to form glasses. Raman work on melts is being planned as the next step in this study. 4.2 CORROSION OF STRUCTURAL ALLOYS BY FLUOROBORATES S. Cantor D.E. Heatherly ~ B.F. Hitch - Alloys containing chromium in contact with molten NaBF,-NaF would be expected to form a borlde be- cause the reaction (1 +x)Cr(c) + NaBF,(d) + 2NaF(d) = Na; CrFg(c) + CryB(c) (1) has a negative standard free-energy change (AG®). At a temperature of 800°K, AG2y0 = —10 kcal. This value is based on an estimated standard free energy of forma- tion (AGJ?) of Nay CrFg of —600 kcal/mole. In reaction (1), the exact value of x is unknown; however, AG? of the more stable chromium borides (Cr, B, Crs B; ) is esti- mated to be —22 kcal per gram-atom of boron.? In nickel-base alloys, reaction (1) may proceed more readily because of the probable exothermic nature of the reaction Cry B(c) t yNi(alloy) = Cr(alloy) + Nij, B(c). (2) 3. O. H. Krikorian, Estimation of Heat Capacities and other Thermodynamic Properties of Refractory Borides, UCRL- 51043 (1971). 4. O. S. Gorelkin, A. S. Dubrovin, O. D. Kolesnikova, andN A. Cherkov, Russ. J. Phys. Chem. 46,431 (1972). 43 Assuming that AG? of N1 B equals ts enthalpy of = formation,* AG; for reactron (2) is about - gram-atom of boron. An experiment to determine the extent of borrde —3 keal pe r. formation in the nickel-base alloys ‘Hastelloy N (7% Cr)' : and Inconel 600 (15% Cr) has been in progress for sev- eral months. In this experiment, metal speeim'ens are equilibrated with NaBF,-NaF (92-8 mole %) at 640°C The 'onnly plausible explanation seems. to be that some 'NaBF, remains on (or in).the metal surface despite the washing (5—10 "min in boiling water) intended to ‘remove adhering traces of salt. Some of the scatter in ‘the boron analyses. by SSMS is probably due to salt “contamination of the metal surface. Inconel 600 speci- - mens scanned by IMMA showed a similar pattern of under an argon atmosphere and are periodically re- . moved, washed free of salt, using water, and analyzed- by spark-sourcé imass spectrometry - (SSMS):and; -less ' ~ routinely, by ion microprobe mass analysis (IMMA). 5 Analysis for boron on specimen surfaces by SSMS sug- '. gests some boride formation. Hastelloy N specimens surface inclusions containing B, Na, and F. Unfortu- nately, IMMA does not provide quantitative analyses for ~ these elements. As yet, the extent of boride formation cannot be quantified in either Hastelloy N or in Inconel ' 600 by a combination of SSMS and IMMA. Probably, that had equilibrated for up.to 129 days were found'to - contain 30 to 1000 ppm B; Inconel 600 specimens con- . ; _'tamed 80. to 2000 ppm B. Control spécimens that had . _ not been in contact with the molten salt showed 5 to 20 ppm when analyzed by SSMS. Boron in Inconél 600 increased with equilibration time; but, with Hastelloy * however, reactions (1) and (2) occur to a small extent; boride is deposited at levels not greater than 500 ppm on Hastelloy N and not exceeding 1000 ppm on Inconel 600 "after four months of contact with molten NaBF, NaF - IMMA was also used to obtam depth profiles of alloy - ."-constltue.nts through about 5000 layers. In control N, the data were much more scattered and showed vir- ' ‘tually no time dependence. - Several specimens analyzed by SSMS were also 1nvest1 , gated by IMMA. Boron was present w1thm the first few-' " hundred monolayers of metal, in inclusions also con- taining sodium and .fluorine in specimens of 2% Ti— modified Hastelloy N that had equilibrated for 72 days. “These contamed 150 ppm B as determmed by SSMS. 3 épecimens, elemental concentrations were uniform with -depth. In equilibrated Hastelloy N, molybdenum was uniformly distributed throughout the depth explored, but chromium and titanium concentrations increased linearly from the surface inward; the iron concentration appeéared to.decrease slowly with depth. Equ111brated Inconel 600 showed V1rtually no chromium in the first ; A.SOO Tayers, but chromium increased linearly in the next 5. Spark source mass spectrometry and ion. mlcroprobe mass . analysis performed by the Analyttcal Chemrstry D1vrsron T 4500 layers; iron and nickel were uniform through the depth studied. Thus, IMMA indicatés that chromium is ‘selectively oxidized by NaBF,-NaF (92-8 mole %) or by oxidants contained in this molten mixture. 5. Development and Evaluation of Analytical Methods - A.S. Meyer 5.1 IN-LINE ANALYSIS OF MOLTEN MSBR FUEL B. R. Clark A.S. Meyer R.F.Apple D. L. Manning Corrosion test loops described previously! have con- tinued operation with circulating reference fuel carrier salt, LiF-BeF, ThF, (72-16-12 mole %). No additional loops have been placed in operation during this report- ing period, although several are expected-to begin opera- tion within the next few months, - Measurements of the U* /U3 ratio in the forced con- vection loop (FCL-2b) indicate a “‘steady-state’ value of about 100 (Fig. 5.1). This is somewhat lower? than the 1. H. E. McCoy et al., MSR Program Semmnnu Progr Rep. Aug. 31, 1974; ORNL- 5011 ,p. 76. 2. A.S. Meyer et al., MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 52. . ORNL- DWG, 75-12056 | U4+ log 637 . -, « ——Loop Down 2 Days e —=—_oop Down { Day * ——|oop Down 8 Days 1 I | I 50 t00 150 ELAPSED TIME ({days) 200 Fig. 5.1. U /U ratios in forced convection loop FCL-2b. apparent steady-state value obtained with the fluoride mixture, LiF-BeF,-ThF, (68-20-12 mole_%), indicating a less oxidizing melt. The melt, which started at a ratio of around 1000, reached this level via a redox process which' presumably involves reaction with the chromium in the walls of the véssel or in the specimens. No at- tempts have yet been made to reoxidize the U3 in the melt by suifable additions of NiF, or some other oxi- dant. It is mterestmg that the decrease to a steady-state value occurred after about 75 days, with a rapid de- crease in the first 30 days. Previous data from the exper- imental fuel showed?® a rather stable value near 10* for about 60 days, until beryllium additions were made to forcé reduction of the U%*, 'Some of oscillations in the data probably result from air contammatlon with subsequent oxidation when the loop was down. This was most prominént with the experimental melt (68-20-12 mole %) when the U3 ratio was substantially greater than the steady state value reached at a later date. Ratios of U**/U3* measured in the two thermal con- vection loops, NCL 21A and NCL 23, are summarized in Figs. 5.2 and 5.3’ respectively. No unusual trend is apparent in the oxidation-state history of the fuel melt in NCL 21A. This loop was operated for about 240 days with Hastelloy N corrosion specimens: The curve shows a rather dramatic rise in the ratio whenever new spéci- mens are added. This effect is attributed to additions of moisture and air which partially oxidize U**. A recovery to lower ratios follows each increase in repetitive fash- ion. ORNL-DWG. 75-12057 —1 *8 T i I =3 § 3 2% 3 R4 g < . @ sh8E E g ~ ,Ig g E “ & & % . Zz DU.D of 2 i e | s 7 8 v, l £ - l." . ‘ 41— .'.-.. “., "ete s — | 1 | 1 | 50 100 150 200 250 300 ELAPSED TIME (days) Fig. 5.2. U**/U* ratios in thermal convection loop NCL-21A. ORNL-DWG. 75-12055 » | | | 1 o E. g - v - 2r Eg g g o] — 33 » S 2o c o =< L < R o) + £ . B [& c 8 + |+ ® o . 8. GEJ %/hr. One heat (425) with the reverse combination, low in niobium and 4, H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb. 29, 1972, ORNL4782, pp. 167-69. ORNL-DWG 75-13743 3.0 -_.. P - N 4 . \\ 25 |— h 430 ® 431 428 < % 427" 429 ® 432 i 74-533 @ ® 74-557 74-534 @ . 20 ——971-114 N 704 °C T ~ 72-503 | | Ty | & ® 74-901 " 650 °C - o Woys < o) Q i ' NO 1.0 4 0.5 0 0 0.2 0.4 06 0.8 10 Al CONTENT (%) Fig. 6.24. Proposed boundaries separating stable from un- stable alloys of Ni—12% Mo—7% Cr + Al and Ti with respect to gamma prime precipitation at 650 and 704°C. Alloys above the lines will form gamma prime and those below will not (see Table 6.4 for the compositions of the various alloys). Table 6.7. Stress-rupture data for several heats of titanium-aluminum modified Hastelloy N? at 704°C and 35.0 X 10° psi Composition (wt %) Rupture TOtfll Age hardens Heat . life strain o b Ti Al C (hr) (%) at 704°C 474-901 1.8 0.10 0.06 193.2 394 No 474-533 2.17 048 0.05 196.0 42.0 No 427 2.4 0.18 0.014 82.0 234 No 428 2.47 0.16 0.064 201.8 61.0 No 429 2.4 0.35 0.017 200.6 22.7 No 430 2.5 0.34 0.073 212.4 55.8 No 431 2.5 0.74 0.016 2938.3 6.8 Yes 432 2.35 0.69 0.057 3611.5 13.7 Yes Base: Ni—12% Mo—7% Cr. b Results of hardness measurements taken on aged, unstressed specimens. 78 ORNL-DWG 75-43714 20 0433 A Rg HARDNESS 4| — o425 3.5 4.0 4.5 5.0 at. % (Nb + Ti+Al} Fig. 6.25. Change in hardness of unstressed specimens of various heats of Nb—Ti—Al-modified Hastelloy N after aging 1000 hr at 650°C (see Table 6.4 for the compositions of the various alloys). high in titanium, does not seem to show much aging. Alloys containing 3 to S at. % (Nb + Ti + Al) age appre- ciably with creep rates of about 1 X 1073%/hr or less. The above results indicate that alloys containing approximately 2.5 at. % or less (Nb + Ti + Al) will be satisfactory from the aging standpoint. Such an alloy would be represented by the composition on a weight percent basis of 0.5% Nb—1.5% Ti—0.1% Al. Alloys having concentrations of Nb + Ti + Al above the 2.5 at. % range will be more susceptible to aging. Future work will include evaluation of microstructure for a number of these specimens to confirm present conclusions, evaluation of additional alloys to test the indicated boundaries for stable compositions, and an extension of data analysis to determine whether a quan- titative relationship can be derived that separates the relative effects of the individual elements, Nb, Ti, and Al, on the stability of alloys of this type. 6.5 MECHANICAL PROPERTIES OF TITANIUM-MODIFIED HASTELLOY N ALLOYS IN THE UNIRRADIATED CONDITION T. K. Roche J.C. Feltner B. McNabb Several tests were completed or are in progress to determine the mechanical properties of recently re- ceived heats of 2% Ti—modified Hastelloy N in the unir- radiated condition. These alloys include two production heats (74-901 and 75-421) and six semiproduction heats (74-533, 74-534, 74-535, 74-539, 74-557, and 74-558). Table 6.8. Comparison of hardness changes? and creep behavior? of several heats of Hastelloy N¢ modified with Nb, Ti, and Al Composition Hardness, Rp Minimum creep rate Heat wt % at. % Annealed? 1000 l‘;lr Change (%/hr) Nb Ti Al C Nb + Ti + Al at650°C 237 1.03 0.04 <0.05 0.84 1.5x 1072 63 2.5 <0.01 0.13 1.66 1.1 x10-? 181 1.85 0.50 <0.01 0.045 1.86 3.1 % 1072 69-648 1.95 0.92 0.05 0.043 2.55 7.0x 107¢ 69-344 1.7 0.77 0.24 0.10 2.63 26X 1073 70-835 2.6 0.71 - 0.10 0.053 2.82 6.0x 1074 425 048 1.9 0.08 0.037 2.90 80.1 84.1 4.0 7.8% 1073 421 1.04 1.9 0.07 0.048 3.24 87.3 86.5 —0.8 1.3x 1073 424 1.34 1.8 0.1 0.063 3.38 88.6 90.2 1.6 7.1x107* 418 1.92 2.0 0.05 0.058 3.91 89.1 90.4 1.3 22x 107 420 1.90 1.8 0.15 0.055 3.87 88.7 101.5 12.8 1 x10°° 433 1.89 22 0.33 0.024 4.75 84.8 104.3 19.5 2 X10°° 434 1.86 2.2 0.32 0.061 4.70 93.3 107.0 13.7 3 x10°% 9 Alloys aged for 1000 hr at 650°C and hardness measured in unstressed condition. bCreep tested at 650°C and 47.0 X 103 psi. CBase: Ni—12% Mo—7% Cr. 41 hrat 1177°C. 79 ORNL-DWG 75-13742 6 > 0433 ¢ 434 ~d 4 RS S €420 N~ [* 48 + o~ = S+ 4 e424 + T'|'--°\421 L A 70-835 ¢ RN ARG 3 ' L' 69-344 TN _ g . 69-648 oY © \ 2 \ p 181 ® 63\ ' \ 4 \\ 4 IR 237 \ 0 100 2 5 1074 2 5 10° 2 5 {072 2 5 40! MINIMUM CREEP RATE (% /hr) Fig. 6.26. Minimum creep rate of various heats of Nb—Ti—Al-modified Hasfelloy N tested at 650°C and 47.0 X 10° psi (see Table 6.4 for the alloy composition). Four of the six semiproduction heats contain small additions of rare earths, lanthanum, cerium, and misch metal. The compositions of these alloys were chosen to study the effectiveness of rare-earth additions for mini- mizing the extent of shallow intergranular cracking. Each of the six semiproduction alloys was the prodxct of a 120-1b double-melted (vacuum induction plus elec- troslag remelt) heat produced by an outside vendor. The chemical analysis of the alloys was reported pre- viously.® The mechanical-property studies include the determination of room- and elevated-temperature tensile properties and creep-rupture properties in air at 650, 704, and 760°C. These data serve as a reference for comparison with the properties of standard and other modified Hastelloy N alloys both in the unirradiated and irradiated conditions. A The principal effort during this report period was directed toward completing the creep-rupture data on the above heats. Tests are being run at three stress levels for each of the three test temperatures. Most of the tests were completed, with the major exception being heat 75-421, the 8000-b production heat, for which specimens are being prepared. Specimens of the other 5. T. K. Roche, B. McNabb, and J. C. Feltner, MSR Pro- gram Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 61 and 65. heats were obtained from swaged rod and were annealed for 1 hrat 1177°C prior to test. Figures 6.27 through 6.29 are plots of rupture time as a function of stress at 650, 704, and 760°C, respec- tively, for the 2% Ti-modified Hastelloy N heats and are compared with plots for a previous heat (471-114) of the same alloy and standard Hastelloy N. Minimum creep rates measured from these tests at the three tem- peratures are shown in Fig. 6.30 as a function of stress. As concluded previously, the more recent heats of 2% Ti—modified Hastelloy N are essentially equivalent in strength to the earlier heat, and there is no significant effect resulting from the addition of rare earths to the 2% Ti—modified alloy. As determined from past work and confirmed by the recent tests, the modified alloy exhibits longer rupture lives than standard Hastelloy N at the three temperatures. Additionally, the first of eight creep machines capable of tests in molten fluoride fuel salt was put into opera- tion. A specimen of heat 474-533 has been in test at 650°C and 30.0 X 10® psi for slightly over 1300 hr. Data are not available, as yet, on this same heat in air, but a comparison is made in Table 6.9 with an air test of an earlier heat (471-114) of the same nominal com- position. The data appear to be falling within a normal scatter band for alloys with the same nominal composition that 80 ORNL—DOWG 75-437M1 © M P TTHI 60 N.[2% Ti-MODIFIED HASTELLOY N (HEAT 471-114) b | F 8 -51. \\ \A (17.8 —51.0) ~ 50 SN~ || @ | F | Hrowme]| 4 a (19.8~42.9) Y 5 STANDARD HASTELLOY N T™NL |1 Z 40 roQ mmal (23.3 - 42.8) | b I ' | nt ~d w \ o o ¢ 474-533 \\\ = | @ 474-534 A © 30 I 4 474-535 N A 474-539 - o 474-557 20 |— ® 474- 558 . ° o 474- 901 TEST TEMPERATURE : 650 °C S L] L []] 10 2 5 102 2 5 10% 2 5 104 RUPTURE TIME (hr) Fig. 6.27. Stress-rupture properties of several heats of 2% Ti—modified Hastelloy N and standard Hastelloy N at 650°C. (Ranges of rupture strain indicated in parentheses.) ORNL-DWG 75-13710 60 50 ."\- iy ~ 40 ~+.| 2% Ti-MODIFIED HASTELLOY N (HEAT 471-114) @ “T-h__ [ T TTTT " S {39.4-62.8) g \L\ e Lo 30 = = — (37.2-47.6) L @ T T~ ™ 0 (36.8-47.6) = ¢ 474-533 ] \“‘*-I l ‘ l 20 ® 474-534 "‘-..._‘ Tt T —T1 1 5 474-535 ~{q=-STANDARD HASTELLOY N A 474-539 0 474-557 10 |—w® 474-558 o 474-901 I ] ] TEST TEMPERATURE : 704 °C . T Y 10! 2 5 10° 2 5 103 2 5 104 RUPTURE TIME (hr) Fig. 6.28. Stress-rupture properties of several heats of 2% Ti—modified Hastelloy N and standard Hastetloy N at 704°C. (Ranges of rupture strains indicated in parentheses.) 81 ORNL- DWG 75-13709 35 ‘ \\ 30 \{,2% Ti-MODIFIED HASTELLOY N (HEAT 471-114) N 25 ~ o (47.2-57.5) N \\ Z ~L L “t 20 < eo‘mto-(m.g-ss.g) Q h N — '\‘ - \ N ;) n w 15 STANDARD HASTELLOY N-ZP o a—0-(257-54.2) & \\ N Z R N ~ 0 0 474-533 N ® 474-534 3 a 474-535 N & 474-539 5 o 474-557 o 474-901 TEST TEMPERATURE 76C°C ] TN N 10! 2 5 102 2 5 10° 2 5 104 RUPTURE TIME (hr) Fig. 6.29. Stress-rupture properties of several heats of 2% Ti—modified Hastelloy N and standard Hastelloy N at 760°C. (Ranges of rupture strains indicated in parentheses.) ORNL-DWG 75-13708 80 70 — o 474-533 ® 474-534 - -~ HEAT s 474 -535 1 LA 47114 60 |- & 474-539 T 0 474 —557 PP i ® 474 —558 I~ ~ 50 | © 474-901 @ =} Lt p o P le 40 — g 471114 ] 1 - & T 704°C — L] n P - 4 30 = — 1" Jar-na b 1 6’_’ 20 . £7600C - 10 = 2% Ti—MODIFIED HASTELLOY N (47t-114) O L L 1073 2 5 1072 2 5 107! 2 5 10° MINIMUM CREEP RATE (%/hr) Fig. 6.30. Creep properties of several heats of 2% Ti—modified Hastelloy N at 650, 704, and 760°C. Solid lines are for heat 71-114, and the dashed lines indicate bands which contain the data for the other modified alloys. Table 6.9. Creep data for specimens of two heats of 2% Ti—modified Hastelloy N in salt and air environments? Accumulated strain (%) Time (hr) 474-533 471-114 (fluoride fuel) (air) 500 1.3 2.4 1000 2.8 4.4 1300 3.5 5.5 9Tests run at 650°C and 30.0 X 102 psi. are tested under similar conditions, and there is no indication that corrosion by the molten fluoride fuel salt represents a significant factor. 6.6 POSTIRRADIATION CREEP PROPERTIES OF MODIFIED HASTELLOY N H.E.McCoy T.K.Roche Postirradiation creep tests are in progress on speci- mens from five experiments that were irradiated in the 82 poolside of the ORR. Each experiment contains 102 miniature creep specimens in an instrumented facility in which temperatures can be measured and controlled by supplying heat from auxiliary heaters. Only 12 in-cell creep machines are available for postirradiation creep _testing; hence the testing proceeds rather slowly. The most recent tests have concentrated on (1) the prop- erties of six 125-lb semiproduction heats that contain 2% Ti and low concentrations of rare earths, and (2) the properties of several alloys containing both niobium and titanium. The results of tests completed to date on the six heats that contain titanium and rare-earth additions and the 10,000-lb commercial heat that contains titanium are summarized in Table 6.10. Previous tests at tempera- tures of 650 and 704°C showed that the creep prop- erties of these heats are about equivalent.® The rupture life at 650°C and 40.0 X 10 psi varied from 1200 to 1800 hr, and the rupture life at 704°C and 35.0 X 103 psi varied from 170 to 200 hr. Final conclusions con- 6. T. K. Roche, ). C. Feltner, and B. McNabb, MSR Pro- gram Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 78. Table 6.10. Postirradiation creep properties of titanium-modified Hastelloy N at 650°C after irradiation at the indicated temperature? Irradiation Minimum Rupture Total fracture Alloy Test temperature Stress. creep life strain Composition? 3 positi number CO) (10° psi) rate (hr) %) (%/hr) 474-533 R-1912 650 40.0 0.025 231.1 7.2 2.17% Ti, 0.48% Al R-1908 650 47.0 0.050 111.8 7.8 704 40.0 58.2 6.6 R-1929 760 35.0 0.035 202.2 7.8 474-534 R-1913 650 40.0 0.021 572.2 16.2 2.09% Ti, 0.53% Al, 0.013% La R-1909 650 47.0 0.087 66.0 6.9 704 40.0 58.8 3.5 R-1930 760 35.0 0.0085 73.9 2.8 474-535 R-1915 650 40.0 0.016 767.1 13.6 2.13% Ti, 0.55% Al, 0.04% rare earth R-191] 650 47.0 0.099 93.0 7.5 R-1922 704 40.0 0.023 467.7 12.3 R-1926 771 35.0 0.019 6454 13.9 ) 474-539 R-1914 650 40.0 0.020 601.8 10.8 1.93% Ti, 0.20% Al, 0.03% Ce R-1910 650 47.0 0.11 73.9 9.7 R-1921 713 40.0 0.031 429.5 16.5 R-1925 774 35.0 0.0080 1220.0 11.4 474-557 R-1920 671 47.0 0.045 217.4 10.6 2.14% Ti, 0.02% Al R-1923 713 40.0 0.044 186.2 9.5 . R-1927 771 35.0 0.019 434.5 8.3 474-558 R-1916 650 47.0 0.092 74.8 79 2.05% Ti, 0.02% Al, 0.02% La R-1924 716 40.0 0.021 179.3 8.6 R-1928 795 35.0 0.012 42.5 1.1 474901 R-1936 650 47.0 0.069 171.6 13.6 1.80% Tt, 0.10% Al R-1937 704 47.0 0.15 33.2 5.2 R-1907 732 47.0 0.14 52.5 74 “All specimens annealed 1 hr at 1177°C prior to irradiation for ~1100 hr to a thermal fluence of N3 X 107° neut.rons/cm2 . Ib'Alloy nominal base composition of Ni—12% Mo—-7% Cr—0.05% C. cerning the postirradiation properties (Table 6.10) are not possible, because the test matrix has not been com- pleted. Specimens irradiated at 650°C and tested at 650°C have rupture lives that are about half those of the unirradiated specimens, but there are no differences in the properties of the various heats that are considered significant in view of the limited data. The properties of all heats are considered good after irradiation at 650°C. After irradiation at 704°C and testing at 650°C and 40.0 X 10* psi, the rupture lives of heats 474-533 and 474-534 appear to be lower than those for the other heats by a factor of 3. In all cases the rupture life and the fracture strain were lower after irradiation at 704°C than at 650°C. After irradiation at "760°C and testing at 35.0 X 10° psi at 650°C, the rupture life varied from 43 to 1220 hr and the fracture strain from 1.1 to 11.4% 83 respectively. Thus, differences in creep behavior of these alloys likely become progressively more important as the irradiation temperature is increased. The fracture strains of the various heats appear to show significant trends with increasing irradiation tem- perature. Heats 474-533 and 474-557 have good frac- ture strains (6 to 10%) which do not decrease appreci- ably with increasing irradiation temperature. The frac- ture strains of heats 474-535 and 474-539 are in the range of 10 to 16% and do not change appreciably with irradiation temperature. Alloys 474-534 and 474-558 show decreasing fracture strains with increasing irradia- tion temperature. The behavior of alloy 474-901 appears to be unique in that it shows a marked drop in fracture strain as the irradiation temperature is in- creased from 650° to 704°C. However, this effect may Table 6.11. Postirradiation creep properties of several modified Hastelloy N alloys at 650°C? Minimum Total Rupture Alloy? Test Stiess' creep life fractgre number (10° psi) rate (hr) strain (%/hr) (%) 428 R-1948 47.0 0.043 222.1 11.9 474-533 R-1908 47.0 0.050 111.8 7.8 R-1912 40.0 0.025 231.1 7.2 430 R-1947 47.0 <0.0049 972.0¢ 4.8¢ 432 R-1946 47.0 <0.00051 972.0¢ 0.5¢ 431 R-1945 47.0 <0.0024 972.0¢ 2.3¢ 424 R-1919 35.0 ~0 316.0¢ 40.0 0.00024 140.64 47.0 0.00033 452.64 55.0 0.0066 . 949 49 7.84 424 R-1944 63.0 0.0096 3554 10.4 420 R-1918 35.0 ~0 532.24 40.0 ~0 140.59 47.0 0.00021 450.99 55.0 0.00037 695.99 63.0 0.00080 33434 70.0 0.0062 277.6¢ 4.3d 420 R-1943 63.0 <0.0017 1068.0¢ 1.8¢ 418 R-1917 35.0 0.00002 65244 40.0 0.00007 140.64 47.0 0.00014 452.34 55.0 0.0040 794 .84 5.4d 418 R-1942 63.0 0.0022 559.2 7.2 434 63.0 <0.085 12.9 1.1 433 R-1949 63.0 <0.0011 660.0¢ 0.75¢ 2All specimens annealed 1 hr at 1177°C prior to irradiation. Irradiation carried out at 650°C for approximately 1100 hr to a thermal fluence of ~3 X 102° neutrons/cm?. bSee Table 6.4 for detailed chemical analyses. CTest still in progress. d3tress increased on the same specimen in the increments shown. 84 Table 6.12. Summary of information relative to metallurgical stability of several compositions of modified Hastelloy N Age hardens at 650°C based on indicated parameter Hardness, N Postirradiation Composition (wt % osition (at. % Heat? unstressed Unirradiated creep behavior ? ) Wi ) Compos ; ar 7) . creep . . ’ Nb Ti Al Nb + Ti + Al specimen, 5 behavior’ irradiated at 1000-hr anneal 650°C for ~1100 he? 428 No No No 2.47 0.16 3.57 474-533 No No No 2.17 0.48 392 430 Yes Yes Yes 2.5 0.34 4.02 432 Yes Yes Yes 2.35 0.69 4.63 431 Yes Yes Yes 2.5 0.74 4.94 424 No Yes Yes 1.34 1.8 0.10 3.38 420 Yes Yes Yes 1.90 1.8 0.15 3.87 418 No Yes Yes 1.92 2.0 0.05 3.91 434 Yes Yes Yes 1.86 2.2 0.32 4.70 433 Yes Yes Yes 1.89 2.2 0.33 4.75 2See Table 6.4 for detailed chemical analyses. bFrom Table 6.4. CFrom Table 6.6 based on considerations of data on rupture life and total strain. dFrom Table 6.11. be related to the strain rate. In summary, these sparse data suggest that the fracture strains of alloys con- taining only titanium and those containing titanium plus cerium remain at adequate levels as the irradiation temperature is increased, while the fracture strains of the two alloys (474-534 and 474-558) that contain lan- thanum do not. Additional specimens were irradiated and tested to check this important point. Section 6.4 of this report deals in detail with the metallurgical stability of alloys containing Nb, Ti, and Al in the unirradiated condition. Some of these alloys have been irradiated, and limited test results are avail- able (Table 6.11). The alloys were annealed for 1 hr at 1177°C prior to irradiation for about 1100 hr at 650°C. The anneal at 1177°C should have dissolved most of the alloying elements, and the subsequent period at 650°C may have resulted in the formation of gamma prime. Precipitation of this embrittling phase also strengthens an alloy; hence the postirradiation creep tests should show whether significant quantities of gamma prime were formed. As discussed in Sect. 6.4, precipitation of this phase may be strain induced, and a detailed analysis of the creep data will be required to determine whether the gamma prime formed in the specimens during irradi- ation or whether it formed as the specimens were stressed initially. _ The data from Table 6.11 and information from Sect. 6.4 are summarized in Table 6.12, which shows that the same conclusions are reached with regard to aging of creep specimens in the unirradiated and irradiated con- ditions. However, hardness measurements on unstressed, unirradiated specimens fail to be a good indication of aging in alloys containing niobium. Alloys having a com- bined titanium and aluminum content as high as 3.57 at. % had excellent postirradiation properties. Alloys with higher combined concentrations are quite strong, but no conclusion can be made about their fracture strains. All of the alloys containing Nb, Ti, and Al are quite strong, and can take considerable strain before fracturing. Alloy 434 has a low fracture strain, while no conclusion can be drawn relative to alloy 433. The alloys containing Nb, Ti, and Al which have been evaluated thus far are likely too highly alloyed even though some of the fracture strains are acceptable. Less highly alloyed materials are being irradiated. 6.7 MICROSTRUCTURAL ANALYSIS OF TITANIUM-MODIFIED HASTELLOY N D. N. Braski J. M. Leitnaker G. A. Potter The first part of this section presents the results of microstructural studies of two titanium-modified Hastelloy N alloys, 472-503 (designated as 503) and 471-114 (designated as 114). Previous analyses of these same two alloys dealt with their microstructures after aging” and after postirradiation creep tests.® In the present investigation the microstructures of both alloys were analyzed in an attempt to explain some unusual postirradiation creep results in specimens that were given a slightly higher solution annealing treatment before irradiation. The analysis showed that many of the test specimens were quite inhomogeneous and that the poor creep properties could in some cases be related to the inhomogeneities. This finding prompted a study aimed at producing more homogeneous Hastelloy N alloys. The problem is being approached by reducing the carbon content of the alloy and by giving careful attention to the fabrication parameters. The results of initial experiments to fabricate homogeneous alloys are presented in the second part of this section, 6.7.1 Microstructural Analysis of Alloys 503 and 114 Postirradiation creep tests. The results of creep tests on specimens of alloys 503 and 114, which had been previously irradiated in the ORR at 760°C, are given in Fig. 6.31. The creep tests were conducted at 650°C at a stress level of 35.0 X 107 psi. This particular series of specimens was designed to show the effect of solution annealing temperature on the postirradiation creep rup- ture life of the materials. The solution anneal was a 1-hr heat treatment and was given to all specimens before they were irradiated. As seen in Fig. 6.31, the 503 speci- men given the standard 1 hr at 1177°C solution anneal demonstrated good creep rupture life, while the 114 specimen given the same treatment had a comparably short lifetime. However, with an increase of only ~30°C in annealing temperature, alloy 503 had a greatly reduced lifetime, while alloy 114 showed marked improvement. It was considered unlikely that these results could be caused by changes in solution annealing temperature alone, and other possible expla- nations were sought. It is important to note that despite the apparent instability in creep behavior, the properties of the 2% Ti—modified alloys are generally good. The problem is thus to determine why some specimens have poor properties. A solution to this problem was sought by carefully analyzing the microstructures of the two alloy 503 and 114 specimens described above. 7. D. N. Braski, J. M. Leitnaker, and G. A. Potter, MSR Program Semiannu. Progr. Rep. Aug. 31, 1974, ORNL-5011, pp. 62—-68. 8. D. N. Braski, J. M. Leitnaker, and G. A. Potter, MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 83-90. 85 ORNL-DWG 75-3454 1600 T : STRESS LEVEL = 35,000 psi IRRADIATION TEMPERATURE = 760°C 1400 1200 R N | w 1000 - - 503 ‘ « 3—’1 2 800 ~ u_ - z / - | \ . - . < R o | x T / « N \ 400 « f \ g / \11a 2] \ 200 | } \.\ / \ P U 0 — B 1000 1100 1200 1300 SOLUTION ANNEALING TEMPERATURE (°C) Fig. 6.31. Creep-rupture life of 2% Ti—modified Hastelloy N alloys 503 and 114 at 650°C after irradiation in ORR for 1200 hr. Transmission electron microscopy. Samples were pre- pared for transmission electron microscopy (TEM) by electropolishing small transverse sections of the tested creep specimens in perchloric acid solutions. Figures 6.32 and 6.33 show electron micrographs representative of 503 and 114 specimens respectively. Figure 6.324 shows an area near a grain boundary in the 503 speci- men annealed at 1177°C. The microstructure was ob- served to contain MC-type carbides — both in the grain boundary and in the form of small platelets. Disloca- tions were nearly always found to be associated with the MC platelets. The 503 specimen annealed at 1204°C had similar features (Fig. 6.32b). In both specimens the MC platelets were concentrated near the grain bound- aries. This suggests that the element or elements (prob- ably titanium) making up the MC-type carbide in both specimens were not uniformly distributed throughout the matrix. Figure 6.33 shows electron micrographs of the 114 specimens annealed at 1177°C (Fig. 6.33a) and 1204°C (Fig. 6.33h). These specimens also contained fine MC-type carbides, but instead of forming platelets they precipitated out on stacking faults. The stacking fault - precipitates initiate from dislocations associated with preexisting or primary MC carbides and grow along (111) planes.® The primary MC carbides (the dark 9. J. M. Silcock and W. J. Tunstall, “Partial Dislocations Associated with NbC Precipitation in Austenitic Stainless Steel,” Phil. Mag. 10, 360—89 (1964). spherical particles in Fig. 6.33) were not dissolved dur- ing the solution anneal. Although not shown in these micrographs, a nonuniform distribution of MC precipi- tates was also observed near many grain boundaries in both 114 specimens. However, in this case, a denuded zone was observed rather than a higher concentration of MC at the grain boundaries. This implies that at least Fig. 1177°C. (b) Solution anneal of 1 hr at 1204°C. 86 one of the elements (probably titanium) needed to form MC-type carbides remained in the volume surrounding the primary carbides. Since most of these primary car- bides were located within grain interiors, the areas near grain boundaries were free of carbides. Transmission electron microscopy has revealed several interesting dif- ferences in carbide morphology between specimens of 6.32. Specimens of alloy 503 after 1200 hr in ORR at 760°C and creep testing at 650°C. (a) Solution anneal of 1 hr at YE-11275 =R Fig. 6.33. Specimens of alloy 114 after 1200 hr in ORR at 760°C and creep testing at 650°C. (a) Solution anneal of 1 hr at 1177°C. (b) Solution anneal of 1 hr at 1204°C. the two titanium-modified Hastelloy N alloys. The investigation also showed two different types of non- uniform MC-type carbide distributions near grain boundaries in both alloys. Unfortunately, it was not possible to identify differences in microstructure by TEM that could be directly related to the wide variation in creep properties with changes in solution annealing temperature. Metallography. The same two pairs of 503 and 114 specimens studied by TEM were also examined in the Radiation Metallography Laboratory at ORNL. Polished longitudinal sections of 503 and 114 specimens were examined near the fracture surfaces. Micrographs of these areas are shown for 503 and 114 in Figs. 6.34 and 6.35 respectively. The 503 specimen annealed at 1177°C (Fig. 6.34a), which had a long creep rupture life, had many intergranular cracks along its outer sur- face. Several cracks or evidence of grain boundary sepa- ration was also observed within the specimen interior. Carbide stringers were found to be distributed uni- formly across the entire section of the sample. The car- bide stringers are composed of numerous MC-type par- ticles which lie in lines parallel to the primary working direction in fabrication. Some carbide stringers are more clearly illustrated in Fig. 6.36 by the higher magnifica- tion micrographs taken of the 503 specimen. The 503 specimen annealed at 1204°C had only a few cracks (Fig. 6.34b), but one of them was apparently able to 87 propagate quite rapidly, leading to early fracture. Note also the striking differences in microstructure of the surface region, which extends to a depth of ~0.012 in. This region was free of carbide stringers and had a grain size of at least twice that seen in more central areas. Quite similar microstructural features were also ob- served for the 114 specimens having poor creep prop- erties. The 114 specimen annealed at 1177°C, which had a short creep rupture life, also had a ~0.0124n.-thick carbide-free layer (Fig. 6.352). How- ever, a large grain size was not observed in the layer as seen in the 503 specimen (Fig. 6.34b). It was also found that surface cracks never penetrated beyond the carbide-free layer. The 114 specimens annealed at 1204°C (Fig. 6.35h) demonstrated better creep proper- ties and had a microstructure much like that shown for the better SO3 samples (Fig. 6.35¢). The carbide-free layer found in the 503 and 114 specimens with poor creep properties may have been caused by decarburiza- tion during the solution anneal at 1177°C in argon. Decarburization layers of ~1 mm (0.039 in.) were observed in Inconel 617 after creep tests at 1000°C for 127 hr in helium containing 500 ppm oxygen.'® The lack of carbides in the surface layer may have, in turn, 10. Y. Hosoi and S. Abe, “The Effect of Helium Environ- ment on the Creep Ripture Properties of Inconel 617 at 1000°C." Met. Trans. 6A, 117178 (June 1975). | R-70167 0.25mm Fig. 6.34. Microstructure of spec nens of alloy 503 near fracture surfac after 1200 hr in ORR at 760°C and creep testing at 650°C. (a). Solution anneal of 1 hr at 1177°C. (h) Solution anneal of 1 hr at 1204”C 88 oot on_| 025mm Fig. 6.35. Microstructure of specimens of alloy near fracture surfaces after 1200 hr in ORR at 760°C and creep testing at 650°C. (@) Solution anneal of 1 hr at 1177°C. (b) Solution anneal of 1 hr at 1204°C. encouraged grain growth in the 503 specimen (Fig. 6.35b) during the solution anneal. It is unclear as to why only certain specimens have the carbide-free layers when all specimens were supposedly fabricated in the same way. The results of the metallographic and TEM examina- tions cannot be used to fully explain the mechanisms which led to the early creep failure of two of the speci- mens studied. However, we have shown that a number of microstructural inhomogenieties exist in the 2% Ti modified Hastelloy N alloys, including carbide-free sur- face layers, large-grain-size surface layers, generalized carbide stringers, and nonuniform distributions of MC-type carbide near grain boundaries. Some of these inhomogenieties appeared to affect the results of mechanical tests and may also influence other impor- tant tests, such as those relating to tellurium attack. Consequently, a study was initiated to produce Hastel- loy N alloys with more homogenous microstructures. 6.7.2 Homogeneous Hastelloy N Alloys The problem of producing Hastelloy N alloys with homogeneous microstructures is being approached in two ways. The first is to reduce the carbon content in the alloy to ensure that all of the MC-type carbides are dissolved during the solution annealing treatment. If all the carbides could be held in solution during fabrica- tion, the formation of carbide stringers might be elimi- nated. The second approach is a detailed evaluation of the fabrication process. This latter effort is primarily aimed at identifying the steps at which the different inhomogenieties are introduced and finding suitable alternate processing methods to remove the inhomo- genieties. A definite concern throughout the entire study is that any successful fabrication changes also be compatible with commercial practices. Carbon content. The first series of experiments was designed to eliminate carbide stringers by reducing the carbon content of the alloy. Thermodynamic calcula- tions using data from previous experiments indicated that all of the carbides should dissolve at 1177°C in alloys with carbon contents of less than 0.045 wt %. Therefore, two alloys, 451 and 453, both with a nomi- nal Hastelloy N composition (13 wt % Mo, 7 wt % Cr, bal Ni) and 1.94 wt % Ti, were cast into l-in-diam ingots having carbon contents of 0.017 and 0.035 wt % respectively. The fabrication schedule called for the cast ingots to be hot swaged at 1177°C from a l-in. to a 0.430-in. diameter and then to be annealed at 1177°C for 1 hr. The rods were further reduced to a 0.337-in. diameter by cold swaging, annealed at 1177°C for 1 hr, and cold swaged to a final diameter of 0.250 in. One- inch-long samples were then cut from each alloy rod, 89 R-69236 Fig. 6.36. Carbide stringers in specimen of alloy 503 after irradiation at 760°C and creep testing at 650°C. encapsulated in quartz under an argon atmosphere, and aged at 760°C for 116.5 hr to precipitate the carbides. After aging, the carbides in alloy 453 (0.035% C) were extracted electrochemically in a methanol—10% HCl solution. Consecutive extractions produced the profile shown in Fig. 6.37 of wt % carbide precipitate through the thickness of the sample. The profile for alloy 453 is considerably more uniform than those observed for alloys 503 and 114 specimens aged at 750°C for 1000 hr. The difference may not be entirely due to a reduc- tion in carbon content, because the 503 and 114 speci- mens were swaged from bars cut from %-in.-thick plate, not from drop-cast ingots. (Carbides are fairly uni- formly distributed in the grain boundaries of the 2-Ib laboratory ingots, while they appear as stringers in the J-in. plate.) Metallographic examination of the aged 451 and 453 samples (Fig. 6.38) showed that the reduc- tion in carbon content did not eliminate the carbide stringers. However, the stringers were finer and more evenly distributed than those observed previously (Fig. 6.34b). Carbide-free surface layers were observed in both specimens; a typical surface layer in a heavily etched 453 sample is shown in Fig. 6.39. The depth of the carbide-free surface layer was ~0.003 in. Fabrication. One of the most critical steps in fabri- cating Hastelloy N alloys with respect to its effect on microstructure is the solution anneal. Electrochemical extractions on an as-swaged alloy 453 (0.035% C) sam- ple showed that a moderate number of carbide particles ("0.2%) was present in the microstructure after proc- essing. It is suspected that the sample was not ade- quately annealed at 1177°C prior to the final cold swag- ing operation. That is, the annealing time was too short or the annealing temperature was actually less than ORNL-DWG 75-3455R 0 T ESTIMATED ERROR w 08 o w ! \ ‘\Qs. H : "CI \\ o L os ,‘ C & S / w 8 2N / ~. | o e — : ‘; 453w T~d | 02 E 3 0 O 0025 0050 0075 0100 0425 DISTANCE FROM CENTER OF SAMPLE (in.) Fig. 6.37. Distribution of carbides through the specimen thickness of titanium-modified Hastelloy N alloys. Alloys 503 and 114 solution annealed at 1177°C and aged at 650°C for 1000 hr. Alloy 453 swaged and aged at 760°C for 116.5 hr. 90 1177°C. Therefore, the response of titanium-modified Hastelloy to solution annealing at 1177°C was studied as a function of time at temperature. Samples of alloy 451 (0.017% C) were annealed in argon at 1177°C for 15 min to 8 hr. The samples were cleaned electrochemically for 6 hr to remove any sur- face effects, and the carbides were electrochemically extracted, separated, and weighed. The results of this experiment are plotted in Fig. 6.40. Only extremely small amounts of precipitates were present in samples annealed for 2 hr or more. At times less than 2 hr, there was some scatter in the data, but, in general, slightly more precipitate was extracted. These results indicate that 30 to 60 min are needed in addition to the stan- dard 1-hr solution anneal at 1177°C to dissolve the car- bides completely. Micrographs of sections from each of the samples of alloy 451 from the first annealing series are shown in Fig. 6.41. Little grain growth was observed between the 15-min and 1-hr anneals, while slight grain growth was evident after 2 hr at 1177°C. As expected, rather extensive growth occurred at the longer annealing times of 4 and 8 hr. Y-133206 ®) 0.010in. 0.25mm Fig. 6.38. Microstructure of titanium-modified Hastelloy N alloys 451 (0.017% C) and 453 (0.035% C) after cold swaging and aging at 760°C for 116.5 hr. (a) Alloy 451. (b) Alloy 453. Although most of the effort in this study has been directed toward elimination of carbide stringers in the alloys, experiments are also under way to determine the cause of the carbide-free layers. In one experiment, sam- Y-133201 Fig. 6.39. Carbide-free surface layer in sample of alloy 453 (0.035% C). 91 ples will be examined metallographically before and after a solution anneal at 1177°C. This is a reasonable starting point, because carbide-free layers are found along the reduced section of machined tensile speci- mens; that is, any effects of hot or cold swaging would have been removed by machining in that area. Finally, there is the consideration of fabrication practice, which may, in fact, be the key to producing a homogeneous alloy. A number of relatively minor changes in the way the material is handled may have dramatic effects on the resultant microstructure. Duplicate ingots of alloys 451 and 453 are available and will be fabricated to 0.250-in.-diam rod, with special attention paid to the fabrication parameters. Metallographic samples will be cut from the work piece throughout the processing so that the associated microstructural features may be observed. 6.8 SALT CORROSION STUDIES J.R.Keiser J.R.DiStefano E.J. Lawrence The corrosion of both nickel- and iron-base alloys by molten fluoride salts has been the subject of extensive research for many years. Results show that impurities such as FeF,, NiF,, and HF in the salt react with con- stituents of the alloys, but corrosion from these sources is limited by the supply of reactants. The strongest oxidant of the normal constituents of fuel salt is UF4, and of the major constituents of most iron- and nickel- base alloys, chromium forms the most stable fluoride. Consequently, the major corrosion reaction between ORNL-DWG 75-142243 0.2 .t R o 4st ANNEALING RUN o 5 2nd ANNEALING RUN = o0 3rd ANNEALING RUN = 0 4th ANNEALING RUN & o o * ) © 3 A\ [} T ole 2 1 T =4 o] ! 2 3 4 5 6 7 8 TIME AT 1477 °C (hr) Fig. 6.40. Amount of carbides extracted from alloy 451 as a function of time at solution annealing temperature of 1177°C. 92 0.0101n. it @ (e) Fig. 6.41. Micrographs of samples of alloy 451 after solution annealing at 1177° 1 hr;(d) 2 hr; (e) 4 hriand (f) 8 hr. nickel- or iron-base alloys and molten-salt reactor fuel salt has been found to be 2UF4(d) +Cr(c) 2 2UF5 (d) + CrF, (d) . Because the equilibrium constant for this reaction has a small temperature dependence, temperature gradient mass transfer can occur and results in continuous re- moval of chromium from the hotter sections of a system and a continuous deposition of chromium in the cooler sections. The experiments described in this section are being conducted to determine the corrosion rate of various (f) C for different times. () 15 min; (b) 30 min; (c) salt-alloy systems under controlled test conditions. The variables include composition of the alloy, oxidation potential of the salt, temperature, and exposure time. All loops incorporate electrochemical probes to measure the concentrations of uranium and transition-metal flu- orides. The systems used to conduct these experiments include one forced circulation loop operated by person- nel in the Reactor Division and three thermal convec- tion loops. Five additional thermal convection loops have been constructed and are being prepared for opera- tion. The status of these eight thermal convection loops is summarized in Table 6.13. 93 Table 6.13. Status of thermal convection loop tests on August 31, 1975 ' Loop Insert Loop : numbet’ material specimens Salt type Status Purpose 21A Hastelloy N Hastelloy N MSBR fuel salt Operating 1. Analytical method development? 6969 hr 2. Baseline corrosion data 3. Tellurium mass transfer studies 23 Inconel 601 Inconel 601 ‘ MSBR fuel salt Qperating 1. Baseline corrosion data for 6035 hr high-chromium atloy under MSBR conditions 2. UF, -graphite reaction 31 Type 316 Type 316 Li, BeF, Operating 1. Baseline corrosion data SS SS 248 hr 2. Effect of oxidation potential . of salt on corrosion rate 18B Hastelloy N Modified MSBR fuel salt In preparation 1. Screening test loop for Hastelloy N modified alloys 24 Hastelloy N Nb-Ti modified MSBR fuel salt In preparation 1. Baseline corrosion data for Hastelloy N modified alloys 2. Tellurium mass transfer 25 Hastelloy N Nb-Ti modified MSBR fuel salt In preparation 1. Baseline corrosion data for Hastelloy N modified alloys 2. Tellurium mass transfer 27 Type 316 Type 316 MSBR. fuel salt. In preparation 1. Oxidation potential studies S8 SS 2.. Tellurium mass transfer and other iron- 3. Effect of tellurium on base alloys mechanical properties of specimens 29 Hastelloy N Standard and MSBR coolant salt In preparation 1. Analytical method development modified 2. Baseline corrosion data Hastelloy N 3. Tritium transport data 2 All eight thermal convection loops are equipped with electrochemical probes. 6.8.1 Fuel Salt Thermal Convection Loops Two thermal convection loops, NCL 21A and NCL 23, have been operating with MSBR fuel salt (LiF- BeF,-ThF,;-UF,, 72-16-11.7-0.3 mole %) to obtain baseline corrosion data. NCL 21A is a Hastelloy N loop with specimens of the same material. As with all ther- mal convection loops, eight specimens are inserted in the hot and the cold legs. The 16 specimens are re- moved periodically for visual examination and weighing. The results of the weight change measurements are shown in Fig. 6.42. The corrosion rate of the hottest specimen in this loop is somewhat higher than has been observed in other Hastelloy N systems (see Sect. 6.8.2 discussion of FCL-2b). The higher corrosion rate of loop 21A relates to the relatively high oxidation poten- tial of the salt in this loop (U**/U" about 10*). How- ever, assuming uniform removal of material, the cor- rosion rate of the hottest specimen was 0.24 mil/year, which is within acceptable limits. This loop will con- tinue to be used to obtain corrosion data for Hastelloy N in salt with a relatively high oxidation potential. Loop NCL 23 is constructed of Inconel 601 and has specimens of the same material. A loop was built of Inconel 601 because of this alloy’s resistance to grain boundary penetration by tellurium. Since the alloy con- tains 23% Cr, there was concern about its ability to resist attack by molten fluoride salt. The corrosion rate of Inconel 601 in fuel salt was determined from weight measurements of the 16 specimens of loop 23, and the results are shown in Fig. 6.43. All specimens lost weight, and the loss shown by the hottest specimen was very large. The material lost by the hottest specimens did not result in uniform removal of the surface, but resulted in the formation of the porous surface struc- ture shown in Fig. 6.44. As shown in Fig. 6.45, electron microprobe examination of this specimen showed high thorium concentration in the pores. The only known source of thorium was the salt which contained ThF,, so it is very likely that the salt penetrated the pores. Continuous line scans with the microprobe indicated a depletion of chromium near the surface. Figure 6.46 shows the results of analysis for Ni, Cr, and Th. This figure clearly shows the chromium concentration gra- 94 "ORNL-DWG 75-12246 { e o - 566 °C N o . o O » - E 635°C ] ) z - A <1 I . %) \ E -2 A \\k g_l \ 704°C _3 \ 0 1000 2000 3000 4000 5000 SPECIMEN EXPOSURE TIME (hr) Fig. 6.42. Weight changes of Hastelloy N specimens from loop NCL-2IA exposed to MSBR fuel salt at the indicated - temperature. ORNL-DWG 75-12245 0 -4 P, e —— w \ 635°C g -2 ™ o \ E N W 3 N prd 3 ~. o4 N - \ I . . 5 g < [¥1] S \ -6 704°C -7 o - - 500 1000 1500 2000 SPECIMEN EXPOSURE TIME (hr) Fig.- 6.43. Weight changes of Inconel 601 -specimens from loop NCL-23 exposed to MSBR fuel salt at the indicated tem- perature. dient and ‘provides further evidence of the presence of thorium in the pores. Deposits such as those shown in Fig. 6.47 formed on the specimens in the cold leg, and the deposits’ were identified by microprobe analysis as chromium. - This compatibility test of Inconel 601 in MSBR fuel salt shows a relatively high corrosion rate, and it is doubtful that this alloy would be suitable for use in an MSBR under the conditions of this test. The lower limit for the U**/U3" ratio in an MSBR will likely be determined by the conditions under which the reaction 4UF, +2C23UF, + UG, proceeds to the right. Because the salt in loop NCL 23 is strongly reducing with a U**/U3" ratio of less than 6, it was decided to try to reproduce the results of Toth and Gilpatrick,!! which predicted that at temperatures below 550°C and U*/U*" ratios below 6 the UC, would be stable. However, graphite specimens exposed to the salt for 500 hr did not show any evidence of UC,. The specimens used were made of pyrolytic graph- ite, and it is likely that the high density of the material limited contact of the salt and graphite. The experiment is being repeated with a less dense graphite. 6.8.2 Fuel Salt Forced Circulation Loop Hastelloy N forced circulation loop FCL-2b has been operated during this reporting period to gather baseline corrosion data under conditions where the Uy ratio was relatively low (see Sect. 2.3). Eighteen Hastel- loy N specimens were exposed to MSBR fuel salt with a U# /U ratio of about 100. The specimens were re- moved at predetermined intervals for visual examination and weighing, and the weight changes are shown in Fig. 6.48. Six specimens were held at each of three tempera- tures: 704, 635, and 566°C. Of the six specimens at each temperature, three were exposed to salt having a velocity of 0.49 m/sec and three to salt having a ve- locity of 0.24 m/sec. No effect of salt velocity on the - corrosion rate was found, so each data point represents the average weight loss of the six specimens. The weight loss of the specimens at the highest temperature corre- sponds to a uniform corrosion rate of 0.11 mil/year. Uniform corrosion at this rate is acceptable and well within the limits which can be tolerated in an MSBR. Following termination of the ~3200-hr corrosion experiment, FCL-2b was to be used to make heat trans- fer measurements. This operation has been delayed, because a salt leak developed and a section of the %-in.- diam Hastelloy N tubing had to be replaced (see Sect. 2.3). Examination of the tubing in the vicinity of the leak is under way. : ‘ Further corrosion measurements will be made in ‘this loop with the U* /U ratio at about 103. Additions of NiF, to the salt will be made to raise the U**/U3* ratio to the desired level. » 6.8.3 Coolant Salt Thermal Convection Loops Thermal convection loop NCL 31 is constructed of type 316 stainless steel and contains LiF-BeF, (66-34 mole %) coolant salt. The 16 removable corrosion speci- mens are also made of type 316 stainless steel. The maximum temperature of the loop is 639°C, and the minimum temperature is 482°C. The initial objective of 11. L. M. Toth and L. O. Gilpatrick, The Equilibrium of Dilute UF, Solutions Contained in Graphite, ORNL-TM-4056 (December 1972). 95 Y —129632 60 MICRONS 500X INCHES \ < O o O )¢ a Fig. 6.44. Microstructure of Inconel 601 exposed to MSBR fuel salt at 704°C for 720 hr. As polished 20 1 Y-131294 ’ Backscattered Electrons ThMa X-Rays Fig. 6.45. Electron beam scanning images of Inconel 601 exposed to MSBR fuel salt for 720 hr at 704°C. Y-131219 96 Ni-~3000 COUNTS FULL SCALE CR--3000 COUNTS FULL SCALE. TH--1000 COUNTS FULL SCALE i v POy PRI, Y L {10 MICRONS © ;| T & Fig. 6.46. Microprobe continuous line scan across corroded area in Inconel 601 exposed to MSBR fuel salt for 720 hr at 704°C. 97 Y-131514 2 [ gxu gOT 5go =S o @0’ o ¥ g Q10 Fig. 6.47. Microstructure of Inconel 601 exposed to MSBR fuel salt at 566°C for 720 hr. As polished. ORNL-DWG 75-12244 "¢ os < 3 566°C oo 2 5 o z [635°c i :7051 s 3 704°C $ -0 == 0 500 4000 4500 2000 2500 SPECIMEN EXPOSURE TIME (hr) 3000 3500 Fig. 6.48 Weight changes of Hastelloy N from loop FCL-2b exposed to MSBR fuel salt at the indicated temperature. this loop is to provide baseline corrosion data on a com- mercial iron-base alloy. The loop has been in operation for 248 hr. 6.9 CORROSION OF HASTELLOY N AND OTHER ALLOYS IN STEAM B.McNabb H. E. McCoy The corrosion resistance of several heats of standard and modified Hastelloy N and other iron-, nickel-, and cobalt-base alloys is being evaluated in the unstressed condition in the TVA Bull Run Steam Plant. Two heats of standard Hastelloy N tubing (N15095 and N15101) are being evaluated in the stressed condition from 28.0 X 10® to 77.0 X 10° psi. The method whereby the specimens are stressed is shown in Fig. 6.49. The wall thickness of the gage sec- tion of the specimens was varied from 0.010 in. (77.0 X 10° psi) to 0.030 in. (28.0 X 10 psi) to produce the desired stress range. The %-in.-OD capillary tube con- nects the annulus between the two tubes to the con- denser. When the inner tube ruptures, steam passes through the capillary, and a rise in temperature of a thermocouple attached to the capillary indicates rup- ture. Time to rupture can be taken directly from the multipoint recorder and plotted vs stress for design pur- poses. Data of this type for periods as long as 11,000 hr were reported previously.'? A photograph of the specimen holder (Fig. 6.50) shows the ten instrumented stressed specimens, the four uninstrumented stressed specimens in the filter basket, and the unstressed sheet specimens bolted to the speci- 12. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 94-101 98 ’ — — STEAM SUPPLY 3500 psig, 1000 °F, 16-17 1bs/min ORNL-DWG 68-3995R2 . PRESSURE TRANSDUCER™ SAMPLE HOLDER SAMPLE VESSEL SUPPORT AND /1 THERMOCOUPLE 9 weLL =S FILTER/ ale \-MULTIPOINT RECORDER SAMPLE HOLDER | FLOW RESTRICTER REDUCED WALL THICKNESS ] | EVENT CONDENSER }: WATER QUT RETURN TO CONDENSATE STORAGE < — CAPILLARY TUBE TUBE BURST SPECIMEN { TYP. 10) B bz WATER IN Fig. 6.49. Schematic of double-walled tube-burst specimen. men holder. The filter basket bolts to the small flanges on each side of the sheet specimens (shown exposed), so that the specimens are covered and the flow of steam is directed over the specimens rather than around them. The steam enters the specimen chamber near the middle of the stressed specimens in front of the unstressed specimen holder and is directed lengthwise over the two stacks of 2-in.-long X %-in.-wide X 0.035-in.-thick sheet specimens. The steam passing over the specimens flows through the Neva-Clog filter to prevent scale from enter- ing the flow restricter orifice or the remainder of the steam system. The steam is condensed and returned to the condensate storage vessel. No specimen has lost any scale so far, but some of the Croloy-type alloys are beginning to develop blisters, a prelude to scaling. The oxide on all Hastelloy N specimens is thin and ad- herent, with no evidence of scaling. Some of the un- stressed Hastelloy N specimens have been exposed to steam for 19,000 hr at 538°C and 3500 psig. Several alloys were included in this study, and, as reported pre- viously,'* they displayed a wide range of oxidation rates. Several obeyed the parabolic rate law, Aw - Kt where Aw is the weight change in mg/ecm?, ¢ is the time in hours, and K is a constant. Figure 6.51 is a log-log plot of weight change in mg/cm? as a function of time in hours. Note the sudden increase in the rate of weight change, with each alloy gaining approximately 0.5 mg/cm® over the last 4000 hr. This probably indi- cates deposition of some substance on the specimens at a rate that was equal for all specimens. We noted pre- viously that fine particles of iron oxide that was en- trained in the steam had deposited on the specimens, but this deposition occurred at a much lower and con- stant rate. The increased rate of weight gain for the specimens was discussed with Bull .Run engineers. The Bull Run facility has had several instances of condenser tube leaks in the last year of operation, whereas in previous years, few if any condenser leaks occurred. The cooling water in the condensers is at higher pressure than the condens- ing steam to prevent back pressure on the turbines, and . when a leak occurs, untreated cooling water is intro- duced into the steam system hot well. Continuous monitoring of silicon in the four hot wells (condensed 13. H. E. McCoy and B. McNabb, Corrosion of Several fron- and Nickel-Base Alloys in Supercritical Steam at 1000°F, ORNL-TM-4552 (August 1974). 99 Photo 224475 Fig. 6.50. Photograph of the steam corrosion chamber after 19,000 hr of exposure. Features to note are the stressed but uninstrumented specimens in the filter (foregound), the two groups of unstressed specimens, and the ten instrumented stressed specimens. The stressed specimens have an outside diameter of 1 in. and a length of 3 steam wells) indicates a condenser leak when the silicon level increases, and the leaking condenser can be iso- lated and repaired. The condensed steam (and any cool- ing water introduced by condenser leakage) passes through demineralizers and is monitored again, with sili- con and other impurities being held below acceptable limits before the condensate is returned to the steam system. Even though care is taken to prevent excessive amounts of impurities in the steam system, the facility is evidently operating with a different level of impurities than had been experienced before condenser problems developed. Some evidence of sodium silicate, as a black- ish gray deposit, has been observed on some safety-valve seats, and this is possibly the material that has deposited on the specimens. The oxide on most of the specimens is black or gray, and no changes in its appearance were noticed during routine examination and weighing of the specimens. When the specimen holder is removed for the next scheduled examination, an effort will be made to determine the composition and nature of the deposit. A scheduled replacement of condenser tubes is planned by Bull Run engineers in the near future to eliminate the problem of condenser leaks. Some of the alloys represented in Fig. 6.51 lost weight initially before gaining at an accelearated rate during the last 4000 hr. These alloys were Hastelloy X, Haynes alloy 188, and Inconel 718, and they contain approximately 20% Cr. Other investigators have re- ported weight losses due to loss of chromium in steam at high temperatures. It is probable that these alloys would have continued to lose weight if the steam con- ditions had not changed; new specimens of some of the alloys will be inserted in the test facility when steam conditions improve. ORNL-DWG 75-15519 10 N \\\ AN R CROLOY 1-9% Cr 2 3 - s NE 1 )LJ j \ ol ~— O o o N I £ —— /1 05 1 + HASTELLOY N E / porser = sl T q% ] 3 [ . 4 E 0.2 / IAW K| o5 5 IAW Kt o AW =Kt ul ; 0.1 l : ol 7% 1 > 1t o INcoLoY 800 003 X1/ 1 J 1T & HASTELLOY X \ / Py o H-188 ® INCONEL 600 0.02 |— \ a4 347 STAINLESS STEEL N / < INCONEL 718 001 5 S II4 L L 11 . 10° 2 5 10 2 5 10 TIME (hr) Fig. 6.51. Corrosion of several alloys in steam at 1000°F (538°C) and 3500 psi. 6.10 OBSERVATIONS OF REACTIONS IN METAL-TELLURIUM-SALT SYSTEMS J. Brynestad - Several criteria must be met for a good screening test system for the tellurium corrosion of Hastelloy N: 1. The tellurium activity must be appropriate, repro- ducible, and known. . The tellurium must be delivered uniformly over the sample surfaces and at a rate sufficient to prevent excessive testing times. ‘ . Preferably, the system should operate under invari- ant conditions during the test run. . The system must be relatively cheap, simple, and easy to operate. In the MSBR the productlon of tellurium per time unit will quickly reach a constant value, and in due time a steady state will be reached where tellurium is re- moved from the melt at a rate that equals the rate at which tellurium is produced. 100 In addition to reacting with the material of which the primary circuit is constructed, tellurium could be re- moved by several means which include the following: 1. The processing system: Since the MSBR is to be equipped with a processing system to remove fission products, tellurium might be effectively removed from the salt by appropriate measures. . The gas phase: If the gas phase is contacted with a getter such as chromium wool, the tellurium activity in the melt might be kept close to that defined by the equilibrium YaCryTeg(s) @ Tey(g) + 7 Cr(s) This activity is sufficiently low that Hastelloy N would not be attacked. . A “‘getter” immersed in the salt melt: Obvious dis- advantages of this arrangement would be the prob- lems of mass transport in temperature gradients and the lack of a candidate material. Until the steady-state condition in an MSBR is more clearly defined, it is impossible to state the likely tellu- rium activity. It is only known that in the MSRE, stan- dard Hastelloy N was embrittled (probably by tellu- rium). In the MSRE the steady-state tellurium activity — if ever reached — probably was defined by gas phase removal and was likely rather high. Until the steady-state situation in the MSBR is de- fined, it must be assumed that one must deal with the MSRE condition, under which standard Hastelloy N is embrittled. In order to define this condition, we have tested several systems with defined tellurium activities with regard to their behavior toward Hastelloy N: 1. equilibrium mixture of Cr; Te; (s) + Cr; Tes(s), 2. equilibrium mixture of Ni3Te,(B,, 41 at. % Te) + NiTeq 775(y, v 43.7 at. % Te), equilibrium mixture of Cry Teq(s) + Crs Teg(s), equilibrium mixture of Ni3 Te, (s) + Ni(s). 3. 4. The systems are arranged in sequence of decreasing Te, activity, as determined by isopiestic experiments. Typical corrosion experiments were conducted at 700°C for 250 to 1000 hr. The arrangements were by isothermal gas phase transport of Te, in previously evacuated, sealed-off quartz ampuls; by embedding the specimens in the mixtures; and, in the Cr, Te;-Cr; Te, and Cr3Te,s-CrsTeq cases, by transport in molten salt. The most pertinent results are as follows: 1. Hastelloy N samples exposed to NizTe,(s) + Ni(s) (system 4) did not show intergranular cracking. This is promising, because if one can establish a steady- state condition in which the tellurium activity is lower than that defined by this system, standard Hastelloy N will not be embrittled. . Systems 1 and 2 have tellurium activities that are too high. These systems corrode Hastelloy N severely under all the experimental arrangements used. . System 3 [Cr3Tes(s) + Crs Tes(s)] shows promise as a tellurium-delivery method in molten salt, since it is sufficiently corrosive to cause intergranular cracking of Hastelloy N but does not form reaction layers. o It is of value to note that the system Cr,Teg(s) + Cr(s) has a tellurium activity that is much lower than the system NizTe,(s) + Ni(s). This system also is prom- ising, since high-surface chromium might be used as a tellurium getter in the gas phase. Experiments are under way to measure the tellurium activities of the above systems. 6.11 OPERATION OF METAL-TELLURIUM-SALT SYSTEMS J. R. Keiser J. R. DiStefano 1. Brynestad E.J. Lawrence The discovery of shallow intergranular cracking of Hastelloy N parts of the Molten-Salt Reactor Experi- ment which were exposed to fuel salt led to a research effort which identified the fission product tellurium as the probable cause of the cracking. Experiments showed that Hastelloy N specimens which had been electro- plated with tellurium or exposed to tellurium vapor exhibited shallow intergranular cracking like that of specimens exposed in the MSRE. Subsequently, a pro- gram was initiated to find an alloying modification for Hastelloy N which would enhance its resistance to tellu- rium. The resistance of these modified alloys to crack- ing is measured by exposing specimens to tellurium vapor, deforming them, and then evaluating their sur- faces by metallographic and Auger methods. However. the chemical activity of tellurium in these experiments is significantly higher than it was in the MSRE. In order to simultaneously expose specimens to the combined corrosive action of molten fluoride salt and tellurium at a more realistic chemical activity, a method is being sought for adding tellurium to molten salt in a manner that would simulate the appearance of tellurium as a fission product. Experiments have been started that will permit evaluation of several methods to determine whether they will produce the desired conditions. 6.11.1 Tellurium Experimental Pot 1'% Tellurium experimental pot 1 was built to evaluate the use of lithium telluride as a means for adding tellu- 101 rium to salt. This pot (Fig. 6.52) allows tellurium to be added periodically in the form of salt pellets containing a measured amount of lithium telluride. Three viewing ports permit observation of the pellets after their addi- tion to the salt. Electrochemical probes are inserted through Teflon seals and are used to detect and measure the concentration of certain species in the salt. The salt 14. The lithium telluride for this experiment was prepared by Valentine and Heatherly. The electrochemical measurements were made by Meyer and Manning Y—129592 Access for Additions, Sampling, and Electro- chemical Probes Viewing Port Reaction Vessel Fill or Dump Vessel Fig. 6.52. Lithium telluride experimental pot No. 1. used is LiF-BeF,-ThF, (72-16-12 mole %), and its tem- perature is maintained at 650°C. In the first experiment, tellurium was added as the lithium telluride Li, Te, which was prepared by the Chemistry Division (Sect. 3.1). Initially, two pellets containing a total of 0.070 g of Li, Te were added to 600 m] of salt. Electrochemical examination of the salt by members of the Analytical Chemistry Division gave no indication of the presence of tellurium (Sect. 5.3). Subsequently, three more Li, Te pellets were added, and a sample of the salt was taken for chemical analysis. Three additions of CrF, totalling 1.82 g were then made, followed by the addition of three more Li, Te pellets. A final addition consisting of 0.2 g of BeO was made. Results can be summarized as follows: . The Li, Te pellets did not melt and disappear imme- diately; some evidence of the pellets remained on the salt surface for the duration of the experiment. . No electrochemical evidence of a soluble tellurium species was detected. . Following the Li, Te additions, visibility through the view ports was limited by a bluish-grey deposit that was subsequently identified as being predominately tellurium. . Chemical analysis of the salt sample taken after the addition of five Li, Te pellets showed that the tellu- rium content was less than 5 ppm. The conclusion is that use of the compound Li,Te does not provide an adequate means for adding tellu- rium to MSBR fuel salt. However, it is thought that LiTe; is much more soluble in MSBR fuel salt than is Li, Te. Considerable difficulty was encountered by the Chemistry Division in synthesizing LiTe; (Sect. 3.1), but material is now available and the experiment will be repeated with LiTe; as the source of tellurium. 6.11.2 Chromium Telluride Solubility Experiment The addition of a soluble chromium telluride, either Cr,Te; or CryTes, represents another method for adding tellurium to molten MSBR fuel salt. If an excess of chromium telluride is maintained, the activity of tellurium in solution in the salt will be constant pro- vided the temperature is not changed and no changes are made in the salt composition. To determine whether there is a temperature at which either of these chro- mium tellurides can provide a reasonable amount of tellurium in solution, an attempt was made to deter- ‘mine the solubility of the chromium tellurides as a func- tion of temperature. ' 102 A Hastelloy N pot was filled with the salt LiFF-BeF, - ThF, (72-16-12 mole %) and the temperature con- trolled at 700°C. After a sample of the salt had been taken, Cr; Te, was added and a small Hastelloy N sheet specimen was inserted into the salt. After 170 hr the specimen was removed and after 250 hr a salt sample was taken. The temperature was then lowered to 650°C, and a day later a salt sample was again taken. This sequence was then repeated at 600°C. Next, the salt temperature was raised to 700°C, Cr,Te; was added, and another Hastelloy N specimen inserted. The speci- men was removed and salt samples were taken under the same time-temperature conditions as discussed above. The two Hastelloy N specimens were weighed and submitted for Auger examination. No weight changes were detected, but evidence of tellurium in the grain boundaries was found (Sect. 6.12). The results of the chemical analysis of the salt sample are shown in Table 6.14. Tellurium concentrations at 700°C were not as high as was expected, but the fact that some tellurium was in solution is demonstrated by the tellurium found on the specimens. Additional Cr,Te; was added to the solution, and two salt samples were taken. Preliminary results indicate that the solution may not have been saturated when the first series of salt samples was taken, Following the solubility measurements, two tensile specimens were exposed to the salt-Cr, Te; solution. Both specimens, one of regular Hastelloy N and one of 2.6% Nb—0.7% Ti—modified Hastelloy N, showed a weight increase after 500 hr exposure at 700°C. After room-temperature tensile testing, the regular Hastelloy N specimen was observed to have significantly more and deeper cracks than did the modified Hastelloy N speci- men (Sect. 6.14). Table 6.14. Results of chromium telluride solubility experiment Tellurium and chromium content of salt samples (ppm) i f Sampling Background After After temperature (no Te) Cr,Te, Cr,Te, C0) © addition addition 700 Te 5 Te <5 Te <5 Crd4 Cr75 Cr90 650 Te 15.1 Te 7.5 Cr 105 Cr 120 600 Te <5 Te <5 Cr? Cr 88 2]nsufficient sample. 103 The experimental assembly is being used to expose standard Hastelloy N specimens to salt containing Cr, Te; to obtain data on the extent of attack at 700°C as a function of time. 6.11.3 Tellurium Experimental Pot 2 When a technique for introducing tellurium into salt at an acceptable chemical activity has been developed, a method will be needed for exposing a large number of specimens to salt-tellurium solutions. A large experi- mental pot has been constructed for this purpose. The pot has a stirring mechanism, facilities for introduction of electrochemical probes, and sufficient accesses to allow simultaneous exposure of a large number of speci- mens. Operation of the system will begin when a satis- ~ factory tellurium addition technique is available. 6.12 GRAIN BOUNDARY EMBRITTLEMENT OF HASTELLOY N BY TELLURIUM R. E. Clausing L. Heatherly Auger electron spectroscopy (AES) is a powerful technique for studying grain boundary embrittlement of Hastelloy N by tellurium. The recent development of the technique to permit AES analysis using a small- diameter ("v5-u) electron beam to excite the Auger elec- trons of a specimen surface has made truly microscopic analysis possible.!> The development of techniques for scanning the beam and the development of electronic data processing equipment have continued to be a cen- tral part of our efforts. As the techniques improve, our ability to see the details of the tellurium embrittlement process improves dramatically. We can now not only provide a qualitative image of the elemental distribution on intergranular fracture surfaces at a magnification of several hundred times, but we can also provide a semi- quantitative elemental analysis as the beam is scanned along a line across the sample. However, it is not pres- ently practical to provide a quantitative analysis along a line across an intergranular fracture surface, since Auger intensities at each point on a rough surface vary accord- ing to topography. This effect can be corrected in prin- ciple by a normalization technique, but data for each point must be normalized individually, and the present equipment cannot handle the volume of data required. The data presented below are typical of several samples of tellurium-embrittled Hastelloy N that were examined recently. These samples are being studied in various 15. R. E. Clausing and L. Heatherly, MSR Program Semi- annu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 104. parts of our previously outlined efforts to understand the tellurium embrittlement of nickel-based alloys. The sample chosen for the present discussion demonstrates our state-of-the-art capabilities and limitations and at the same time provides some new insights into the nature of the tellurium embrittlement of Hastelloy N. A sample of Hastelloy N that had been exposed to tellurium vapor at low partial pressure for 500 hr at 700°C was fractured in the AES system, and the result- ing fracture surface was analyzed using Auger electron spectroscopy. The fracture surface is shown in Fig. 6.53. The scanning electron micrographs, made by Crouse, reveal that intergranular fracture occurred along the edges of the sample and that the central region failed in a ductile manner. One fairly large area of ductile shear can be seen. Three types of Auger data presentations are used below: imaging, line scans, and selected area analyses. The first is qualitative, while the second and third are progressively more quantitative. Figure 6.54a is an image obtained using the scanning beam in the AES system and the absorbed sample cur- rent to produce the image contrast. It is similar to the scanning electron micrograph (1), but, because of the larger electron beam and the different method for pro- ducing the image contrast, the resolution in Fig. 6.54a is poorer, and some distortion is evident. Nevertheless, it is relatively easy to correlate the features shown in Fig. 6.54a with those in Fig. 6.53a. Figure 6.54b 1s an image of the same area shown in Fig. 6.54z but with image contrast produced by the tellurium Auger signal. A care- ful comparison of the areas of high tellurium concentra- tion with the areas of intergranular fracture shows that a good correlation exists between the two. No tellurium can be detected in the regions of ductile or shear frac- ture. Figure 6.55 is a series of line scans showing the peak-to-peak intensity of the Auger signals for nickel, molybdenum, chromium, and tellurium as the electron beam was scanned along the path shown by the bright line in Fig. 6.544. Some of the observations that can be made are: (1) The intensities of the Auger signals are influenced considerably by topography; that is, some features, such as the shear region between feature Y and the tellurium-embrittled region below it, show lower Auger emission for all elements (this dependence on topography accounts for much of the jagged nature of the line scan). (2) The tellurium concentration is quite high in the region of intergranular fracture near each original surface. (3) There is a definite tendency for the concentration of molybdenum to be higher in the regions of intergranular fracture. (4) The nickel and chromium concentrations are in approximately the same ratio throughout the scan. = 0t 350 300 200X INCHES 0.010 MICRONS 0.005 0003 15 MICRONS 25 2000X ° 3 3 3 z ] 8 8 S 10 Fig. 6.53. Scanning electron micrographs of fracture surface of Hastelloy N sample exposed to tellurium vapor at 700°C for 500 hr showing the region examined by Auger electron spectroscopy. (@) 200X, (h) 500X, (¢) 1000X, (d) 1000X, (¢) 2000X. The lower half of (c) shows an area of ductile shear. 105 ¥-133510 Crack Te Embrittled Feature X Feature Y Te Embrittled Y—-133512 eature X eature Y Te Embrittled 125 MICRONS 375 150X 0.005 INCHES ~ 0.015 Fig. 6.54. (a) Scanning electron micrograph obtained in the AES apparatus using absorbed sample current to produce image contrast. The bright vertical line shows the path of the line scan and identifies regions in Fig. 6.55. (b) Image of the same region at the same magnification using the tellurium Auger signal to produce contrast. The embrittled grain boundary regions next to the original sample surfaces are obvious. (c) Same image as (a) with the regions analyzed and reported in Table 6.15 identified. 106 ORNL-DWG 75-15427 \_ Te (483 eV) Mo (22 eV) WAV Cr (529 eV) — AREA NO. \ Ni (848 eV) 3 v ORIGINAL b | EMBRT'FTLED X Te EMBRITTLED - Fig. 6.55. Auger signal intensities for scans along the path indicated in Fig. 6.54a. The vertical axis is displaced and the vertical scales arbitrarily varied to permit a qualitative comparison of the variations of Ni, Cr, Te, and Mo as a function of distance along the scan line. The zones and features identified along the horizontal axis are also identified in Fig. 6.54a and c. The AES analysis of the regions labeled area 1, area 2, and area 3 are given in Table. 6.15. Another observation based on the detailed examina- tion of this and other samples is that the tellurium con- centration in the grain boundary is not a monotonically decreasing function as one proceeds inward from the original surface. On nearly all of the embrittled samples examined thus far, the tellurium concentration is uni- formally high throughout the embrittled area as, for example, is shown on the right in Fig. 6.55. (The signal intensity on the left is strongly influenced by topo- graphy. If this effect were removed by a normalization process, this area would have a more nearly uniform composition similar to that on the right.) The high, rela- tively uniform tellurium concentration in the embrittled regions suggests that either a particular grain boundary phase of fixed composition may exist or that the tellu- rium atoms fill all of the appropriate grain boundary sites in the embrittled region. Sputtering this fracture surface (and those of similar samples) to a depth of a few-atomic layers (3 to 10) reduced the tellurium con- centration to below 1 at. %, showing that the tellurium is concentrated very sharply in the grain boundary. It is therefore unlikely that the tellurium present in the grain boundary exhibits the properties of a bulk telluride. The molybdenum concentration remained high during sputtering operation, indicating that the concentration of molybdenum is high in the bulk phase, perhaps in a phase that has precipitated in the grain boundary. Table 6.15 shows quantitative selected-area analyses made in the three regions of the sample indicated in Fig. 6.54c¢. The compositions have been normalized to equal 100 at. % in each row. The three rows for each area are obtained from one Auger spectra, but some elements were ignored in the first two rows to make changes in the relative amounts of the other elements more obvious. These results confirm the above conclusions and show (1) that tellurium is present in relatively large amounts in the embrittled regions and (2) that area 3, which is near the extreme of the depth to which the tellurium penetrated, contains about as much tellurium as area 2, which is located near the center of the upper embrittled region. Regions 2 and 3 are both enriched in molybdenum and carbon as indicated in the line scans. 107 Table 6.15. Composition of regions on the fracture surface of a Hastelloy N sample exposed to tellurium for 500 hr at 700°C ) Composition (at. %)b Region? _ Ni Mo Cr Te C g S S Composition in the lower region 70 19 11 area 3 (intergranular fracture) 64 17 10 9 40 11 6 6 33 2 1 1 Composition in central region . 75 16 9 area 1 (ductile fracture) 75 16 9 61 13 8 13 1 1 3 Composition in the upper - 64 25 12 region area 2 (intergranular 58 23 11 8 fracture) 42 17 8 6 23 1 1 2 2 Areas identified in Fig. 6.54c¢. bThe composition in each row is normalized to equal 100 at. %. The three rows for each region are from the same data, but are normalized so as to make changes in relative amounts of the elements more obvious. For convenience and consistency in reporting data we assume the AES spectra (Paul Palmberg et al., Handbook of Auger Electron Spectroscopy, Physical Electronics Industries, Inc., Edina, Minn., 1972) are accurate and directly applicable to our data. Elemental sensitivities are taken directly from the spectra presented in the handbook with no attempt to correct for chemical effects, line shape, matrix effects, escape depth, or distribution of elements as a function of depth in the sample. The analyzer used is Varian model 981-2707, operated with an 8000-eV electron beam energy. The above results suggest the need for a detailed examination of the causes and effects of the high mo- lybdenum and carbon contents in the grain boundary region and also an examination of the implications that the presence of a two-dimensional tellurium-rich grain boundary phase may have on the time dependence of tellurium penetration into the alloy. 6.13 X-RAY IDENTIFICATION OF REACTION PRODUCTS OF HASTELLOY N EXPOSED TO TELLURIUM-CONTAINING ENVIRONMENTS D. N. Braski " Hastelloy N and several modifications of the alloy have been exposed to tellurium to determine their rela- tive susceptibilities to intergranular cracking. Different methods for exposing samples to tellurium have also been studied in an attempt to develop a suitable screen- ing test for the alloy development program. Some speci- mens were exposed directly to tellurium vapor at 700°C, while others were subjected to attack by nickel or chromium tellurides at 700 and 750°C respectively. This section presents the results of x-ray diffraction analyses of reaction products produced during the tests. Knowledge of the reaction products aids in evaluating a given method of tellurium exposure and may provide information relating to the mechanisms of intergranular cracking. ‘ A number of Hastelloy N tensile specimens and flat x-ray samples were exposed to tellurium vapor at 700°C for 1000 hr in an experiment conducted by Kelmers and Valentine.!® The specimens were positioned in the top portion of a long quartz tube having'a small amount of tellurium at the bottom. The tube was evacuated, backfilled with argon, and placed in a gradient furnace with the specimens at 700°C and the tellurium source at 440°C. With this arrangement, tellurium vapor diffused upward through the tube at a rate dependent on the temperature difference between the specimens and the tellurium (~0.05 mg Tefhr).!® At the end of 1000 hr exposure, the specimens were covered with a very fine, hairlike, deposit similar to that observed previously in creep tests at 650°C.!7 The results of x-ray diffraction analyses on these deposits are given in Table 6.16. The first alloy listed is standard Hastelloy N, while the other three have titanium and niobium additions. The main 16. A. D. Kelmers and D. Y. Valentine, MSR Program Semi- annu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 40—-41. 17. R. E. Gehlbach and H. Henson, MSR Program Semiannu. Progr. Rep. Aug. 31, 1972, ORNL4832, pp. 79-86. 108 Tabie 6.16. X-ray diffraction results for specimens exposed for 1000 hr at 700°C Wt % alloying Method of Heat number additions to nominal tellurium Surface reaction products Hastelloy N composition exposure 405065 None Kelmers-Valentine Ni,Te, + Cr,Te, experiment? 472-503 2.16% Ti Kelmers-Valentine Ni, Te, experiment? 470-835 0.71% Ti + 2.6% Nb Kelmers-Valentine Ni, Te, + Cr,Te, experiment? 180 1.84% Nb Kelmers-Valentine Ni, Te, experiment? 474-533 2.0% Ti Brynestad? — low Ni, Te, + unidentified substance Te activity exposure Ni, Te, (s) + Ni(s) 405065 None Brynestad: Ni, Te, + NiTe, 4, LiCl + Cr, Te, 2A. D. Kelmers and D. Y. Valentine, MSR Program Semiannu. Progr. Rép. Feb. 28, 1975, ORNL-5047, pp. 40—-41. by, Brynestad, MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 102. reaction product was Ni;Te,, which was detected on the surfaces of all four. alloys. (Ni; Te, was found on earlier samples exposed for shorter times in the same apparatus.'®) X-ray lines which could be indexed as Cr3 Tey were also found on standard Hastelloy N and on the alloy modified with 0.71% Ti plus 2.6% Nb. The Cr3Tes interplanar spacings and relative intensities were calculated by H. L. Yakel, Metals and Ceramics Divi- sion, from the crystallographic data in ref. 18. The pres- ence of Cr3Te, in the reaction layer is reasonable, be- cause both chromium and tellurium were detected pre- - viously on Hastelloy N exposed to nickel telluriddes by electron microprobe analysis.'”> In addition, chromium tellurides were previously identified by x-ray diffraction on Hastelloy N exposed to tellurium vapor.!? Brynestad®® exposed 2% Ti—modified Hastelloy N specimens to a low tellurium activity (Ni; Te, + Ni mix- ture) at elevated temperatures. The specimens were first placed in a quartz tube and the Ni;Te, + Ni powder mixture packed around the specimens. The tube was then sealed off under vacuum and placed in a furnace at 700°C for 1000 hr. The reaction products obtained in this test also contained Ni; Te,, but the remaining four lines could not be satisfactorily indexed to any of the 18. A. F. Bertaul, G. Roult, R. Aleonard, R. Pauthenet, M. Chevretou, and R. Jansen, “Structures Magnetiques de Cr, X, (X =8, Se, Te),” J. Phys. Radium 2(5),582-95 (1964). 19. D. N. Braski, O. B. Cavin, and R. S. Crouse, MSR Pro- gram Semiagnnu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 105--09. 20. J. Brynestad, MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 102. Ni, Cr, or Mo tellurides. The unusually broadened x-ray diffraction peaks suggest that a complicated telluride, such as Ni-Cr-Te, may have been formed. In another tellurium experiment, Brynestad exposed a standard Hastelloy N tensile specimen to a melt of LiCl contain- ing Cr; Te; (solid) at 750°C. Some Cr, Te; dissolved in the LiCl melt and reacted with the Hastelloy N. After 146 hr the tensile specimen was removed, and the flat surface on one end was analyzed by x-ray diffraction. The results (Table 6.16) showed that Ni3Te, and NiTey o were produced. In summary, these tests have shown that the primary reaction product between Hastelloy N and tellurium near 700°C is Ni;Te,. X-ray lines corrsponding to Cr3 Teq were also present in patterns from the surfaces of several Hastelloy N alloys exposed to tellurium vapor at 700°C. Exposure of Hastelloy N to tellurium at low activities (Ni;Te, + Ni mixture) may have produced some complicated Ni-Cr-Te compounds in addition to Ni3 Te, as evidenced by the unusually broadened X-ray lines. '6.14 METALLOGRAPHIC EXAMINATION OF SAMPLES EXPOSED TO TELLURIUM-CONTAINING ENVIRONMENTS H.E.McCoy B.McNabb I.C. Feltner Several samples of modified Hastelloy N were exposed to tellurium-containing environments. They were de- formed to failure at 25°C, a procedure which forms surface cracks if the grain boundaries are brittle; a 109 metallographic section of each was prepared to deter- mine the extent of cracking. These tests have two objec- tives. The first is to develop a method for exposing samples to tellurium to produce a reaction rate com- parable to those anticipated for an MSBR. This rate is thought to be a flux of tellurium of about 10'° atoms cm™? sec™'. The second is to compare the cracking tendencies of various alloys of modified Hastelloy N. A new technique developed for measuring the extent of cracking is more nearly quantitative than that used previously, In the new technique a mounted and polished longitudinal section of a deformed specimen is viewed on a standard metallurgical microscope. The eyepiece has a filar which can be moved to various locations in the field being viewed. The filar is attached to a trans- ducer which produces an output voltage that is a func- tion of the location. The output signal is interfaced with a small computer which will, on command, compute crack lengths and several statistical parameters. The information is displayed on a teletypewriter. The cracked edge of the mounted specimen is scribed every 0.1 in., and the operator measures all cracks in succes- sive 0.1-in. intervals until at least 30 cracks have been measured. The computer then calculates and displays the average crack length, the maximum crack length, the standard deviation, and the 95% confidence interval. A typical scan requires about 10 min and is consider- ably faster than other methods used thus far. The experimental conditions associated with the ten experiments to be discussed in this report are summa- rized in Table 6.17. The chemical compositions of the alloys studied are given in Table 6.18. In all cases the Table 6.17. General description of Te—Hastelloy N exposures Experiment E . ¢ Exposure Alloys General Designation Xpenmenters conditions included® comments 75-1 Brynestad LiCi + Cr, Te, 405065 Heavy reaction layers for 146 hr at ~750°C 75-2 Kelmers, Te vapor for Whisker growth, evidence Valentine 1000 hr at 700°CP 405065, of inhomogenous reaction 470-835, with Te 472-503, - 180 753 Brynestad 250 hr at 650°C 405065, Heavy reaction layers packed in Cr, Te, 474-534, 474-535 754 Brynestad 200 hr at 700°C 405065 Heavy reaction layers . packed in Cr;Te, 75-5 Brynestad, 504 hr at 700°C 405065 Reaction layers Keiser in salt + Cr, Te, 470-835 75-6 Brynestad 1000 hr at 700°C 405065 No visible reaction layers with vapor above Cr,Te, 75-7 Brynestad 1000 hr at 700°C 405065 Shallow reaction layers ' with vapor above Cr, Te, 75-8 Brynestad 1000 hr at 700°C with 405065 No visible reaction layers vapor above 8, + 1, nickel tellurides 75-9 McNabb, 250 hr at 700°C in 405065,471-114,474-534, No visible reaction layers McCoy vapor above Te at 474-535, 600600, 62, 63, 300°C 181, 237,295, 297, 298, 303, 305, 306, 345, 346, 347,348,21543,469-344, 469-648,469-714, 470-786, 470-835 75-10 McNabb, 250 hr at 700°C in 405065, 21543, 345, 348, No visible reaction layers McCoy vapor above Te at 411,413,421,424,425 300°C ZSee Table 6.18 for chemical compositions. bA. D. Kelmersand D. Y. Valentine, MSR Program Semiannu. Progr. Rep. Feb: 28, 1975, ORNL-5047, pp. 40-41. 110 Table 6.18. Chemical analyses of nickel-base alloys used in tellurium cracking studies (weight percent) Heat number Mo Cr Fe Mn C Si Ti Nb Al QOther 62 11.34 752 a 020 0.042 001 a 1.9 a 63 11.45 733 4 020 0.135 0.01 a 2.5 a 180 11.2 70 0 0.040 022 0.046 001 <0.02 184 a 181 11.5 684 0.054 023 0.45 0.0l 050 185 & 0.03 W 237 12.0 6.7 43 049 0032 4 0.04 1.03 <0.05 295 114 8.06 4.02 028 0.057 <002 <002 085 a 0.05 W 296 11.5 8.09 3.96 0.28 0.059 <0.02 <002 1.2 a 02 W 297 12.0% 700 a0b 020 0.06 0.02 024 057 a 298 12.0° 7.00 agb 0.2 0.06 002 <001 20 a 303 12.00 700 aob 020 0062 002 049 084 4 305 11.2 825 4.16 022 0.072 0.09 088 13 a 306 10.6 8.04 3.11 0.18 0065 027 001 055 4 345 11.0 7.1 3.8 026 0052 022 002 045 4 346 11.0 6.7 3.7 0.18 0.05° 0.48 002 049 4 347 120 76 4.3 025 005® 047 <002 088 4 348 12.0 7.2 0.07 0.19 0.05° 047 <0.02 062 a4 411 12.0¢ 700 4 020 0050 4 a 115 a 413 12.00 700 4 020 0052 o 100 113 a4 421 12.00 700 4 028 0052 219 1.04 007 424 12.00 7.0 4 026 0052 4 1.8 134 0.10 425 12.00 700 4 026 o0 ¢ 1.9 048 0.08 405065 16.0 7.1 4.0 055 0.06 057 <001 & <0.03 472-503 12.9 679 0089 <00l 0066 0089 216 005 0.09 471-114 12.5 7.4 0.062 002 0058 0.026 175 a 0.07 474-534 11.66 712 006 <001 0.08 0.03 209 a 053 0.14 W, 0.013 La 474-535 11.79 730 005 <001 0.08 0.03 213 @ 055 0.10W, 0.010 La, 0.03 Ce 600600 1600 8.0 0.19 0.27 (Inconel 600) 469-648 12.8 6.9 0.30 034 0043 @ 092 195 4 469-714 13.0 8.5 0.10 035 0013 a 080 160 a . 470-835 12.5 7.9 0.68 060 0052 « 071 260 a 0.031% Hf 470-786 12.2 7.6 0.41 043 0.044 a 0.82 0.62 a 0.024 Zr 469-344 13.0 74 40 056 0.11 2 077 1.1 a 0.019 Zr 421543 0.04 0.08 0050 0019 0.7 0.02 12.4 7.3 9Not analyzed, but no intentjonal addition made of this element. bNot analyzed, but nominal concentration indicated. sample was a small tensile specimen ¥% in. in diameter X 1% in. long having a reduced section ' in. in diameter X 1% in. long. All specimens were annealed 1 hr at 1177°C in argon prior to exposure to tellurium. The results of crack measurements and data resulting from the tensile tests at 25°C that were used to open the embrittled grain boundaries are shown in Table 6.19. Experiment 75-1 was run by Brynestad and involved a sample of standard Hastelloy N that was immersed in LiCl saturated with Cr,Te; for 146 hr at 750°C. The specimen formed a heavy reaction layer (Table 6.17) but lost weight (Table 6.19). Figure 6.56 shows that the reaction was rather extensive, with some obvious grain boundary penetration which resulted in extensive crack formation in the deformed section. The extent of reac- tion in this experiment was higher than anticipated for an MSBR, and therefore it is not believed that the experimental conditions employed constitute a good screening method. Experiment 75-2 was run by Kelmers and Valentine, and the detailed results were described previously.?! All samples lost weight in this experiment (Table 6.19). Although the samples had more reaction product on 21. A. D. Kelmers and D. Y. Valentine, MSR Program Semi- annu. Progr. Rep. Feb, 28, 1975, ORNL-5047, pp. 40-41. Table 6.19. Intergranular crack formation and tensile 'pr_Operties” of samples exposed to tellurium and strained to failure at 25°C s X Experiment Cracks/unit length Depth (u) Stapdfard cor?fifidzbnce Weightb Yield U‘l::':;:e Fracture l_Jnifor‘m Fract‘ure Rtaduc!ion number Heat number Crackeli Crack Mo deviation interval change streSS' stream suess‘ elongau?n strain in area racksfin. racks/fem Average aximum () o (mg) (10° psi) (10 pi) 10° psi) (10° psi) (%) %) 75-1 405065 310 122 101.1 187.2 46.3 16.6 -1.7 48.9 [14.3 111.8 384 395 75-2 405065 320 126 46.3 75.7 12.8 4.5 -1.0 51.7 124.7 117.8 40.4 41.8 42.1 470-835 167 66 26.7 70.9 15.6 44 -2.1 571.7 142.0 136.0 42.5 44.0 353 472-503 300 118 533 97.0 264 9.6 -1.8 56.9 133.8 123.0 38.9 40.4 41.0 180 63 25 243 60.2 £3.2 6.1 -4.7 54.5 124.2 115.8 3422 39.¢6 23.7 75-3 405065 240 95 27.2 429 8.0 2.7 —287 46.7 116.0 109.0 40.2 424 48.4 474-534 220 87 14.5 276 4.5 1.4 —245 56.6 114.0 109.0 41.1 442 52.7 474-535 230 91 20.3 344 54 1.6 -231 459 110.6 99.7 47.4 50.6 543 754 405065 410,570 161,224 69.3,59.1 118.1,123.5 283,25.1 8.9,6.7 —788, —631 44.8,44.0 103.7,100.9 99.3,96.8 32.2,31.0 33.2,324 48.6,46.0 75-5 4035065 370 146 42.5 ©69.0 159 5.2 -60.1 50.9 121.2 117.3 36.5 37.2 28.2 470-835 180 71 331 547 10.8 36 -58.8 55.6 131.2 126.7 37.0 37.9 384 75-6 405065 215 85 85.2 148.3 294 9.0 +0.02 51.6 121.3 117.2 404 42.0 339 15-7 405065 520 205 40.0 106.6 249 6.9 -25.5 524 124.9 118.5 38.7 394 333 75-8 405065 360 142 59.5 92.0 219 7.3 -36.3 52.2 123.8 117.6 40.0 41.4 329 75-9 405065 360 142 41.6 75.3 15.1 5.0 -7.6 529 127.5 122.0 39.5 41.1 394 471-114 185 73 374 639 13.3 4.4 +8.4 479 113.4 105.0 49.6 54.1 46.6 474-534 360 142 225 435 10.7 36 =27 56.9 126.9 1124 43.3 45.5 488 474-535 240 95 20.1 371 9.1 . 37 -0.9 525 1224 113.3 46.3 48.3 479 600600 265 104 17.8 325 6.6 1.8 +3.0 39.9 101.9 79.4 321 37.5 57.2 345 100 39 256 64.1 11.3 36 +0.7 61.7 119.9 107.7 © 391 41.8 46.3 346 160 63 20.7 44 8 10.0 35 +0.4 55.1 1289 111.7 42.1 44.7 47.1 348 9 4 9.6 i5.9 3.1 2.1 -0.3 54.8 123 .4 107.5 42.5 459 503 347 270 106 18.8 373 6.6 1.8 +2.9 58.8 131.2 116.2 40.1 431 46.3 306 450 177 21.0 379 7.3 2.2 +0.8 60.0 130.0 119.2 376 40.1 42.8 303 430 169 14.6 26.8 6.1 1.9 +1.3 56.2 122.3 1100 471 495 49.0 297 _ 310 122 18.0 41.0 6.2 2.2 +2.3 61.1 129.2 117.2 353 378 48.5 295 25 10 10.1 17.9 36 23 -2.5 53.5 122.1 1145 43.3 46.2 44 9 231 85 33 14.8 385 8.5 2.9 -0.9 58.4 122.6 108.5 43.5 46.5 494 305 360 142 16.9 32.7 7.3 24 +0.5 53.6 126.5 117.3 46.8 489 47.7 298 330 130 17.3 30.7 6.4 2.2 +0.4 72.3 123.6 1104 42.7 499 49.7 181 220 87 16.7 38.6 8.7 37 -7.4 549 127.1 118.8 454 45.6 395 62 30 12 29.3 146.0 44.5 25.7 -5.9 47.9 117.5 110.1 49.7 50.7 42.8 63 440 173 12.7 26.7 53 . 1.6 -89 58.1 133.4 121.7 333 38.0 423 469-648 330 130 14.8 433 7.4 26 +0.2 58.9 135.7 127.3 43.7 459 443 469-714 83 33 124 504 12.9 4.5 +0.08 49.9 121.7 116.3 53.7 53.3 43.6 470-835 93 37 8.6 2477 5.0 19 -0.03 55.9 137.5 1304 46 .4 47.7 43.9 470-786 32 13 13.8 578 14.5 7.3 -03 49.5 116.7 107.3 50.6 52.7 40.8 469-344 360 142 9.1 17.1 38 1.3 +0.07 58.7 138.8 129.0 38.0 39.7 414 421543 85 33 8.8 296 48 1.7 —1.2 454 110.6 99.1 54.7 579 53.3 75-10 405065 340 133 17.7 48.5 7.8 1.9 +0.04 53.1 1326 122.0 40.9 364 40.9 - 425 300 118 27.1 548 11.2 4.1 +0.5 51.8 126.0 117.9 46.7 48.5 42.1 425 240 95 227 51.0 124 36 +0.02 51.8 126.3 1163 46.3 42.0 47.9 421 340 133 244 60.1 10.8 2.6 -0.03 53.0 129.7 122.3 43.7 45.6 42.9 424 430 169 26.7 50.8 84 26 -0.03 57.4 1393 132.7 44.2 45 .8 383 298 208 82 8.7 206 34 0.74 +0.1 53.6 125.8 1120 47.5 51.2 52.7 295 26 10 10.6 29.5 6.9 38 +0.2 534 122.1 115.7 42.4 44 6 47.5 348 34 13 16.1 36.2 7.2 3.5 +54 49.8 1219 107.1 44.8 476 54.7 345 13 5 89 14.1 33 3.0 +0.01 59.1 119.8 107.9 44.3 474 454 413 80 32 10.5 25.7 53 1.5 +0.08 52.0 1269 115.5 46.9 494 44.0 411 2 9 14.8 339 9.2 5.1 +14 42.1 1136 97.9 51.9 54.2 536 421543 17 7 109 237 6.1 39 +2.9 47.3 1139 101.4 523 55.7 53.7 9dTensile test run at 25°C at a strain rate of 0.044 min~' bTotal weight of specimen 6.1 t0 6.2 g. It 112 Y—133193 (a) (b) ¥ Y—-133195 Fig. 6.56. Standard Hastelloy N (heat 5065) exposed to LiCl saturated with Cr, Te, at 750°C for 146 hr. (a) Edge of unstressed portion of specimen, () edge of stressed portion of specimen. As polished. 100X. one end than the other, the extent of cracking seemed reasonably uniform. Typical photomicrographs of the four materials are shown in Fig. 6.57. Alloys 405065 (standard) and 472-503 (2.16% Ti) formed extensive cracks, but alloys 470-835 (0.71% Ti, 2.60% Nb) and 180 (1.847% Nb) were considerably more resistant to cracking. In experiment 75-3, three samples were packed in CryTey granules for 250 hr at 650°C. The samples formed heavy nonadherent reaction products and lost weight (Table 6.19). All three materials formed exten- sive cracks (Table 6.19, Fig. 6.58), with the depth of cracking being slightly less in the two modified alloys (474-534 and 474-535) than in standard Hastelloy N (heat 5065). However, the extent of reaction is too high under these conditions for the results to be meaningful. In experiment 754, duplicate samples of standard Hastelloy N (405065) were packed in granules of CryTes and heated 200 hr at 700°C. The samples formed nonadherent reaction films and lost weight dur- ing the test (Table 6.19). The reaction layer and the intergranular cracking produced during stressing are shown in Fig. 6.59. Again, the reaction rate was un- reasonably high for use of the exposure conditions as a screening test. In experiment 75-5, Brynestad and Keiser exposed two specimens to MSBR fuel carrier salt (containing no uranium) that was saturated with Cr,Te;. The ex- posure was for 504 hr at 700°C. These samples formed reaction layers but lost weight (Table 6.19). As shown in Fig. 6.60 both materials formed reaction layers, but in heat 470-835 (0.71% Ti, 2.60% Nb) there appeared to be less penetration of the reactants along the grain boundaries. The standard Hastelloy N had regions where layers of grains dropped out during the exposure. The number and depth of cracks in the stressed portion of the samples were less for heat 470-835 than for stan- dard Hastelloy N, but both materials formed extensive intergranular cracks. Since the samples packed in the various tellurides reacted extensively, several experiments were run in which the samples and the telluride were separated in 113 Y—-134367 Y—134365 (b) Y —134362 (c) Y —134369 0.010in 0.25mm Fig. 6.57. Specimens from experiment 75-2 which were exposed to a low partial pressure of tellurium for 1000 hr at 700°C. (a) Heat 405065, () heat 472-503 (2.16% Ti), (¢) heat 470-835 (0.71% Ti, 2.6% Nb), (d) heat 180 (1.84% Nb). As polished. 100X Y-134623 - ] v-134361 (e) v 134360 0.010in 025 mm Fig. 6.58. Specimens from experiment 75-3. Packed in Cr, Te, granules for 250 hr at 650°C and deformed to fracture at 25°C. (@) Heat 405065, unstr , (b) heat 405065, stressed, (c) heat 474-534 (2.09% Ti, 0.013% La), unstressed, (d) heat 474- stressed, (¢) heat 474-535 (2.13% Ti, 0.01% La, 0.03% Ce), unstressed, (f) heat 474-535, stressed. As polished. 100X. 115 (® Y 134626 Y—134625 Fig. 6.59. Specimen of standard Hastelloy N (heat 405065) from experiment 754. Packed in Cr, Te, granules for 200 hr at 700°C and deformed to fracture. (a) Edge of unstressed shoulder, (b) edge of stressed gage length. As polished. 100X the reaction capsule. In experiment 75-6, standard Hastelloy N was reacted with the vapor above Cr3Te, at 700°C for 1000 hr. The specimen gained a small amount of weight (Table 6.19), did not form an obvious reaction layer (Fig. 6.61), but did form extensive inter- granular cracks (Fig. 6.61, Table 6.19). Experiment 75-7 was run in the same way, but CryTe; was used. The sample lost weight, formed a surface reaction prod- uct, and formed intergranular cracks when strained (Table 6.19, Fig. 6.62). In experiment 75-8 the source of tellurium was two nickel tellurides, 8, and v, . The specimen lost weight, did not form a visible surface reaction product, and did form intergranular cracks (Table 6.19, Fig. 6.63). From these experiments it was concluded that the tellurium activity produced by Cr3 Tes was likely that best suited for screening studies. Experiment 759 included 25 alloys which were exposed to tellurium vapor at 700°C for 250 hr. The weight changes covered a range of +8.4 to —7.4 mg, with no obvious correlation between weight change and crack depth or number (Table 6.19). These specimens were sealed in four different capsules for exposure to tellurium, and there were differences in the extent of discoloration of the samples. These differences are likely associated with slight differences in the extent of reaction due to variation of the temperature of the tellurium metal in the various capsules. Thus, it is ques- tionable as to how far one should carry the analysis of the data from this experiment. One further problem concerning data analysis which applies equally well to all data sets is the basis that should be used for comparison. The number of cracks and their average depth are two very important param- eters. However, it is possible that a specimen can have a large number of shallow cracks (e.g., heat 63, Table 6.19) or a few rather deep cracks (e.g., heat 62, Table 6.19). The formation of intergranular cracks of any depth is important, because this may indicate a ten- dency for embrittlement. The depth of the cracks is important, because this is a measure of the rate of pene- tration of tellurium along the grain boundaries. How- ever, for a relatively short test time [test 759 (250 hr)], the formation of numerous shallow cracks may be indicative of a near-surface reaction which will not lead to rapid penetration with time. Obviously, longer-term tests are needed to determine the rate of penetration of tellurium into the metal. On the basis of number of cracks formed, the alloys in experiment 75-9 which formed less than 40 cracks per centimeter were 345, 470-835, 421543, 237, 469-714, 470-786, 62, 295, and 348. The alloys forming cracks with an average depth of 7 u were 63, 469-714, 295, 348, 469-344, 421543, and 470-835. Several of the alloys appear good on the basis of both criteria. These alloys all contain niobium and several contain niobium and titanium. Another parameter used for comparison was the product of the number of cracks and the average crack depth. The alloys from experi- 116 Y—132811 (a) Y-132812 Y-132813 (c) Y_132814 (@ 0.010in 0.25mm Fig. 6.60. Samples from experiment 75-5. Samples exposed to fuel salt saturated with Cr, Te, for 504 hrat 700°C and strained to fracture at ". (a) Standard Hastelloy N, unstressed shoulder, (b) standard Hastelloy N, stressed gage length showing region where grains were lost during salt exposure, (c) heat 470-835 (0.71% Ti, 2.60% Nb), unstressed shoulder, (d) heat 470-835, stressed portion. As polished. 100x. 117 (a) (b) = Y—132881 Y—132882 0.25mm Fig. 6.61. Standard Hastelloy N (heat 405065) from experiment 75-6. Samples exposed to the vapor above Cr, Te, at 700°C for 1000 hr and strained to failure. (¢) Edge of unstressed portion, (b) edge of stressed portion. As polished. 100X, ment 75-9 are ranked on this basis (Table 6.20). Stan- dard Hastelloy N is significantly different from all other heats on this basis. There are large variations among the other heats, but it is difficult to pick out general trends on the basis of niobium and titanium concentrations. Several typical photomicrographs of samples from experiment 75-9 are shown in Fig. 6.64. No reaction films were visible on any of these specimens. The pic- tures show clearly the wide range of cracking experi- enced by the various heats. The mechanical property data show small, but signifi- cant, variations in the yield and ultimate tensile stresses of the various heats (Table 6.19). The higher stresses are generally associated with the alloys containing the higher amounts of niobium and titanium. However, the high fracture strain and reduction in area for all heats indicate that only very small (if any) amounts of gamma prime formed during the 250 hr at 700°C. In experiment 75-10, steps were taken to ensure that the specimens were at a uniform 700°C and that the tellurium was at 300°C. The weight changes were very erratic and show no correlation with the number of cracks or the depth of crack formation (Table 6.19). A sample of heat 425 was included in each of the two capsules used in this experiment to obtain some idea of reproducibility. The reproducibility was reasonably good. Samples of alloys 405065, 298, 295, 348, and 345 were included in experiments 759 and 75-10. Heats 405065, 298, and 345 in experiment 75-9 cracked more severely than in experiment 75-10. Alloy (a) Y-132816 ® - de i : ; 0.010 in. 0.25mm Fig. 6.62. Standard Hastelloy N (heat 405065) from experiment 75-7. Samples exposed to the vapor above Cr, Te, at 700°C for 1000 hr and strained to failure. (¢) Edge of unstressed portion, (b) edge of stressed portion. As polished. 100X . Y-132818 Fig. 6.63. Standard Hastelloy N (heat 405065) from experiment 75-8. Samples exposed to the vapor above g, + v, nickel tellurides at 700°C for 1000 hr and strained to failure. (a) Edge of unstressed portion. (b) edge of stressed portion. As polished 100% 119 Table 6.20. Ranking of materials from experiment 75-9 Product of number of cracks and average depth Heat Concentration (%)? ? A typical fuel pin is shown schematically in Fig. 6.67. The segments marked “A” were subjected to tensile tests, using the fixture shown in Fig. 6.68. The mechanical property data obtained from the rings and the results of limited metallographic examination were reported previously.>> More detailed metallographic studies have been completed during this report period. The segments marked “B” were used for chemical studies. The salt from each segment was analyzed, and the fission product distributions on the tube surface and a short distance into the tube were determined from two successive leach solutions. The first leach used a “verbocit” solution (sodium versenate, boric acid, and sodium citrate) which should have dissolved only resid- ual salt from the metal surface. The second solution was aqua regia, and the time was sufficient to remove about 22. B. McNabb and H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 123-36. TYPE DESCRIPTION AND USE A Y6 in. RING FOR MECHANICAL PROPERTIES 8 Y4 in. FOR LEACH (2 STEP) c SECTION TO BE RETAINED Fig. 6.67. 1 mil of the tube. Both solutions were subjected to various chemical procedures to analyze for various nuclides and elements. These results are partially ana- lyzed, and the results for tellurium will be discussed. The tube segments marked “C” were retained for pos- sible future studies. 6.15.1 Metallographic Observations Photomicrographs of the three materials in the un- deformed condition are shown in Fig. 6.69. Numerous voids were present near the surface of the Inconel 601 specimen to a depth of about 0.2 mil. Voids were likely caused by the removal of chromium from the alloy via reaction with UF,; in the salt. The Hastelloy N section shows no evidence of chemical reaction with the salt. The type 304 stainless steel shows some grain boundary attack to a depth of about 0.5 mil. This was likely caused by selective removal of chromium along the grain boundaries. The features in the type 304 stainless steel appear much like shallow cracks and may have influenced the number of cracks that were observed in stressed samples of this material. Composite photomicrographs of the Inconel 601 rings after straining to failure are shown in Fig. 6.70. Rings 2 and 4 from near the salt-vapor interface exhibit some evidence of attack, but the other samples are almost entirely free of indications of chemical reaction. Photomicrographs of the deformed rings from the Hastelloy N capsule are shown in Fig. 6.71. The count ORNL-DWG 75-9062 END CAP DN AND SALT LEVEL DI Schematic diagram of individual fuel pin showing the locations of test specimens. 124 ORNL- DWG - 75+ 17506 metallographic sample includes the fracture and an adjacent segment. Since the fracture occurred at dif- ferent locations, the metallographic specimen contains varying amounts of inhomogeneously deformed mate- rial. For example, Fig. 6.71¢ includes a very small seg- ment of homogenously deformed material, whereas Fig. 6.71f includes a relatively long segment. As shown by the photomicrographsin Fig. 6.71 and the data in Table 6.22, specimens from the vapor region (2-A-1), the salt- vapor interface (2-A-2), and the bottom of the salt (2-A-16) cracked most severely. Three samples from other locations formed shallower cracks. It is not known whether these differences are significant. Typical photomicrographs of deformed rings from the type 304 stainless steel capsule are shown in Fig. 6.72. These specimens, located on the inside surface, had shal- low cracks with an average depth of about 0.4 mil (Table 6.22). These cracks were rather uniformly distri- buted in the samples from all four locations. As noted in Fig. 6.69, the unstressed specimen also contained cracklike features having a maximum depth of about 0.5 mil. Hence, the cracks in the stressed specimens may simply be the result of further opening of features that are likely related to corrosion. 6.15.2 Chemical Analyses for Tellurium The tube segments designated B-1, B-2, and B-3 in Fig. 6.67 were subjected to several types of chemical analyses, but only the results for tellurium have been analyzed in sufficient detail to report at this time. The results for the three pins are shown in Table 6.23. The Table 6.22. Summary of crack frequency and depth information for rings from TeGen-1 fuels after straining to failure at 25°C G Crack . Fig. 6.68. Fixture for tensile testing rings from TeGen fuel =~ Specimen frequency Crack depth (mils) pins. number (cracks/in.) Average Maximum of crack frequency shown in Table 6.22 was made in an Hastelloy N effort to detect significant differences in cracking — 2-A-1 480 0.80 2.0 among the various specimens. These counts are subject g:fij :‘158 (l) 'éo f; to numerous ?roplems, the main one -bemg the inhomo- 2-A-5 480 0.58 12 geneous distribution of strain within the sample. In 2-A-8 330 046 1.0 deforming the ring specimens in the fixture shown in 2-A-16 380 1.4 2.5 Fig. 6.68, the small portions of the ring located between the two parts of the fixture likely deformed very uni- Type 304 stainless steel formly, but this length is very short relative to the total ~ 3-A-2 160 0.49 1.2 length. The part of the ring that contacted the fixture g:fig’ gég 8§Z ig likely deformed in some areas but was restrained in 3.A-16 202 0.37 1.0 other areas by surface friction from the fixture.' The 125 (a) R-70188 : % = . » (b) - - () e i 20 4‘0 60 MICRONS 100 120 140 — 500X ———L 0.00t INCHES 0.005 Fig. 6.69. Undeformed rings (sample No. 9) from each TeGen fuel pin near the middle of the fuel salt. (a) Inconel 601, (b) Hastelloy N, (c) type 304 stainless steel. As polished. 500X 126 R—70185 R—70182 | (b) Fig. 6.70. Samples from Inconel 601 fuel pin from TeGen-1. Rings taken from the locations shown in Fig. 6.67 and deformed to failure at 25°C. Portion of specimen exposed to fuel salt is on the lower side of each figure. (¢) Location A-1, (b) location A-2, (c) location A-4, (d) location A-5, (e) location A-8, (f) location A-16. 127 R—70180 Fig. 6.71. Samples from Hastelloy N fuel pin from TeGen-1. Rings taken from the locations shown in Fig. 6.67 and deformed to failure at 25°C. Portion of specimen exposed to fuel salt is on the lower side of each figure. (¢) Location A-1, (b) location A-2, (c) location A4, (d) location A-5, (e) location A-8, (f) location A-16. 128 R—70177 Fig. 6.72. Samples from type 304 stainless steel fuel pin from TeGen-1. Rings taken from the locations shown in Fig. 6.67 and deformed to failure at 25°C. Portion of specimen exposed to fuel salt is on the lower side of each figure. (@) Location 2A, (b) location 4A, (¢) location 8A, (d) location 16A. 129 Table 6.23. Chemical analyses for '? 7™Te and ' ?® MTe nuclides in the three fuel pins from TeGen-1 Sample Type Concentration of 2 7MTe Concentration of ' **MTe Pin identification i - chemistry B . z " ocation sample? dpm total or dpm/g g/lem? or gfg dpm total or dpm/gb g/cm? or gfg€ No. 1 — Inconel 601 1B1 A <22 X107 d <9.5 % 108 d 1B1 B 4.65x 10* 257%x 107 237 x 10° 446X 10°° 1B2 A <2.5 % 10’ d <2.37 x 107 d 1B2 B 1.59 X 10° 8.75x 10" 6.00 x 10 1.13x 1077 1B2 C 7.85 x 10° 1.14 x 107° 7.44 X 10°¢ 4.5%x 1077 1B3 A <53 x 108 d <1.7 X 10°¢ d 1B3 B 6.27 x 10® 345x 10°°® 3.06 x 10® 5.76 x 10°* 1B3 C 247 x 107 3.58x10°° 1.63 x 107 9.9x 10°° No. 2 — Hastelloy N 2B1 A <6.0 X 10¢ d <8 64 X 10¢ d 2B1 B 7.49 x 10° 38 %107 4.97 x 10* 094 X 1077 2B2 A <8.5 x 107 d <54x107 d 2B2 B 295 x 10* 1.52x 107 2.02 x 10* 3.76 x 10°® 2B2 C 6.78 X 107 6.58x 10°° 5.94 x 107 24x10°° 2B3 A <54 x 10° d <5.8x 10°¢ d 2B3 B 8.19 x 10° 4.22x10°® 8.05 x 10° 1.51x 1077 7 2B3 C 3.48 x 107 269X 10°° 3.93 x 107 1.61x 1078 No. 3 — type 304 - 3B1 A 5.52x 10° 3.04x10°® 9.59 x 10° 1.80x 1077 stainless steel iB1 B 1.31 x 10® 0.72x 1078 1.83 x 10® 3.44 x 10°® iB2 A 9.54 x 107 6.29 X 10°° 3.04 X 107 8.51x 10°* iB2 B 2.50 x 10° 1.66 X 107* 3.60 x 107 10.1 x 10°° 3B2 C 9.00 X 107 1.31 x10°® 10.62 x 107 6.48x 107 3B3 A 1.66 X 107 L.I0X 107 5.42 X 10° 3.3% 107 3B3 B 1.27 x-10°® 0.84 x 107 3.15 x 107 1.9x 1078 3B3 C 5.38 x 107 7.8x 107° 3.23 x 107 197 x 1073 4A denotes 100 cm® solution obtained by leaching the metal sample in “‘verbocit” (sodium versenate, boric acid, and sodium citrate}. B denotes 100 cm? solution obtained by leaching the metal sample in aqua regia to remove about 1 mil of metal. C denotes 100 cm® solution obtained by dissolving about 1 g of salt in nitric acid (8 M) saturated with boric aci d. bCounts for individual nuclides given in disintegrations per minute (dpm) total for chemistry sample types A and B and dpm per gram of salt for type C chemistry sample. These counts are laboratory numbers and subject to several corrections which have not been made. “These concentrations are expressed as grams of the particular nuclide per cm?® of metal surface for chemistry sample types A and B and as grams of nuclide per gram of salt for chemistry sample type C. The values have been corrected back to the conclusion of the irradiation. dConcentration throught sufficiently low to be ignored. sample numbers énding with 1 (i.e., 1B1,2B1,and 3B1) designate the material that came from the fuel pin wall exposed to the gas space above the salt. The sample numbers ending with 2 designate material that came from the fuel pin exposed to the fuel salt just below the salt-gas interface, and the sample numbers ending with 3 designate material that came from the portion of the fuel pin exposed to fuel salt near the bottom of the capsule.. Solutions were prepared for analysis by leach- ing metal samples of each tube in “verbocit™ to remove residual salt (type A solution in Table 6.23), leaching the rings in aqua regia (type B solution in Table 6.23), and dissolving about 1 g of salt removed from the metal rings in nitric acid (type C solution in Table 6.23). These solutions were counted to determine the amounts of 1*"MTe and '?°™MTe present. The direct results of- these analyses are presented in Table 6.23, but cannot be interpreted directly because a number of corrections have not been made. The data have been corrected as well as possible to reflect the concentration of each nuclide at the end of irradiation. The concentrations for the leaches from the metal specimens are expressed as grams per square centimeter of tube wall exposed to the fuel salt, and the concentrations for the salt samples are expressed as grams per gram of salt. The ORIGEN code was used by Kerr and Allen to predict the concentrations of tellurium isotopes that should have been present. These calculations have been used extensively in the subsequent analysis of the data. Table 6.24 compares the quantities of '*7"Te and 129mTe found in the three fuel pins with those pre- dicted to be present by the ORIGEN calculations. For 130 each fuel pin the one sample taken of the tube in the gas space was assumed to be typical of that region, and the two samples from the salt-covered parts were aver- aged to obtain a typical value for the salt-covered region. As shown in Table 6.24, generally about 20% of the 127MTe and '*°"Te was found. The percent of tellurium found in the Inconel 601 capsule was appreci- ably higher due to the higher amount found on the salt-covered metal surfaces. There are several possible explanations why the con- centrations of '>7MTe and '2°"Te found are only about 20% of those produced. One possibility is that the amounts calculated are too high. This appears not to be the case, but the calculations will be checked further. The most likely explanation is that the acid leach was not sufficient to remove all of the tellurium from the wall. The tube segments were suspended in the acid with the inside and outside surfaces of the tube wall exposed, as well as the cut surfaces on each side of the %-in. tube segment. Based on the weight changes ob- served and the assumption of uniform metal removal, the thickness of metal removed appears to be about 0.8 mil. Since the cracks extended deeper than 0.8 mil in the Hastelloy N, the tellurium likely penetrated deeper than did the leaching solution. However, the cracks in the other two materials were very shallow, and the 0.8-mil dissolution should have recovered a higher frac- tion of the tellurium if one can equate the depth of cracking to the depth of tellurium penetration. The results in Table 6.24 show no evidence of a systematic variation in the percent recovered from the three tubes. - Several possible explanations for the apparent discrep- ancy in the quantities of tellurium generated and that actually found are being investigated, but none appears reasonable at this time. The concentrations of 27" Te and '?° " Te found in the salt can be used to predict upper limits for the solubility of tellurium in fuel salt under these condi- tions. The '27™MTe nuclide concentration in the salt ranges from 1,14 X 107° to 1.31 X 1078 g per gram of sait (Table 6.23). The ORIGEN calculations were used to estimate the ratio of 127" Te to total tellurium, and this ratio was used to convert the above concentrations of 127MTe to total tellurium concentrations of 0.07 to 0.83 ppm. Similarly, the concentration of '??™Te ranged from 4.5 X 107° to 6.48 X 107® g per gram of salt, and these correspond to total tellurium concentra- tions of 0.08 to 1.13 ppm. The low values in both cases were noted in the Inconel 601 pin, and the higher values were observed in the type 304 stainless steel pin. The concentrations in the Hastelloy N pin were only slightly less than noted for the type 304 stainless steel pin. The Table 6.24. Amount of Tellurium in various locations of fuel pins from TeGen-1 (g) Inconel 601 Hastelloy N Type 304 Location stainless steel 127}nTe l‘z9mTe lZTmTe IZBmTe lITmTe 129mTe Salt 4,1 x10°°® 1.3x 1077 1.0x 1077 3.5 x 1077 1.8 1077 7.4 % 1077 Metal-vapor space 1.4X 1077 24 x 1077 21x1077 50x 1077 20X1077 1.2x10°¢ Metal-salt covered 1.7x 10°¢ 2.3 x10°° 78 x 1077 26 X 10°¢ 4.4 %1077 39x 1077 Total found 1.9x 10°¢ 277 %x10°¢ 1.1 x 10°¢ 345%X 10°¢ 8.2x 1077 2.3x 10°¢ Total formed 3.62X 107¢ 1.34 X 10°% 40x10°¢ 1.48x 1075 3.6x10°¢ 1.34 X 10°°¢ " Percent found 52 20 28 23 23 17 Table 6.25. Concentration of tellurium in various locations of fuel pins from TeGen-1 (108 g/cm?) Inconel 601 Hastelloy N Type 304 Location stainless steel 12‘1Te 129Te 127Te 129Te lz‘?Te l‘nge Metal-vapor space, Bl 2.57 446 3.86 94 3.76 21.4 Metal-salt location, B2 8.75 11.3 1.52 3.76 2.29 1.86 Metal-salt location, B3 3.45 5.76 422 15.1 0.95 1.03 Average if total 11.2 41.3 12,3 45.6 11.2 41.3 yield evenly distributed 131 higher chromium concentration of the Inconel 601 may have caused the lower tellurium concentration in the fuel salt. : The concentrations of '*7”Te and '*°MTe are expressed in Table 6.25.in terms of grams per unit sur- face .area. There appear to be significant variations within each capsule, but there is no consistency be- tween the various pins. The high value for '**”Te in the vapor space of the type 304 stainless steel pin is likely anomalous, since the '?7”Te is not as high. Thus, at this time we conclude that the tellurium is distributed uniformly over the entire surface area of the pin. 6.16 SALT PREPARATION AND FUEL PIN " FILLING FOR TeGen-2 AND -3 M. R. Bennett = A. D. Kelmers The purpose of this portion of the TeGen activity is to prepare purified MSRE-type fuel salt containing 233U and to then transfer a known quantity of this salt into fuel pins for subsequent irradiation in the ORR. One batch of purified salt will be prepared and used, in two filling operations, to fill two sets of six fuel pins each, identified as TeGen-2 and TeGen-3. Similar activi- ties in 1972 to fill the fuel pins used in experiment TeGen-1 have been previously described.?® To MSRE- type fuel carrier salt containing LiF-BeF,-ZrF, (64.7-30.1-5.2 mole %), sufficient 233 UQ, and 238 UF, were added to produce a final composition of LiF-BeF, - ZrF,-*33UF,-?38UF, (63.08-29.35-5.07-1.00-1.50 mole %) after.hydrofluorination to reduce the oxide content. The uranium will be reduced by hydrogen, or by beryllium if necessary, to a U* content of 1.0 to 1.8%, and a measured volume of salt will be transferred into the fuel pins. The design permits obtaining a pre- determined volume in the pins by flushing through an excess salt volume and then blowing back the salt in the upper portion of the pins to leave a predetermined vol- ume. . The equipment in Bfiilding 4508 used previously for this work was reactivated and modified where appropri-. ate. A safety summary and step-by-step operating proce- dure have been prepared and approved. During the lat- ter part of this report period the salt components were charged to the salt purification vessel, and a 36-hr hydrofluorination at 600°C was. completed. Both fil- tered and unfiltered samples were obtained after hydro- fluorination .in- copper filter sticks. After analytical results indicating satisfactory removal of oxide have been received, hydrogen reduction of about 1% of the UF, will be carried out. . 23. R. L. Senn, J. H. Shaffer; H. E. McCoy, and P. N. Haubenreich, MSR Program Semiannu. Progr. Rep. Aug. 31, 1972, ORNL-4832, pp. 90-94. 7. Fuel Processing Materials Development J. R. DiStefano The processes that are being developed for isolation of protactinium and removal of fission products from molten-salt breeder reactors require materials that are corrosion resistant to bismuth-lithium and molten fluo- ride solutions. Past experience has indicated that, al- though their solubilities in bismuth are low, iron-base alloys mass transfer rapidly in bismuth at 500 to 700°C. "The most promising materials for salt processing are molybdenum, Ta—10% W, and graphite. Molybdenum has been tested in a wide range of bismuth-lithium solu- tions for up to 10,000 hr and has shown excellent com- patibility. Thermodynamic-data and literature reports indicate that molybdenum will also be compatible with molten fluoride mixtures. Ta—10% W also has excellent compatibility with bismuth-lithium solutions, but tests are required to measure its compatibility with molten fluoride salts. A thermal convection loop has been constructed of Ta—10% W, and a test with LiF-BeF,-ThF,-UF, (72-16-11.7-0.3 mole %) will be started during the next reporting period. Graphite has shown excellent compatibility with both bismuth-lithium solutions and molten salts. Although no chemical interaction between bismuth-lithium solu- tions and graphite has been found, the liquid-metal solu- tion tends to penetrate the open porosity of graphite. Recent tests have evaluated the extent of penetration as a function of structure of the graphite and the lithium concentration of the bismuth-lithium solution. Dynamic tests of graphite with bismuth-lithium have thus far been limited to quartz loop tests circulating Bi—0.01 wt % (0.3 at. %) Li. During the report period a test was H. E. McCoy completed in which graphite samples were exposed to Bi—2.4 wt % (42 at. %) Li in a molybdenum thermal convection loop for 3000 hr at 600 to 700°C. 7.1 STATIC CAPSULE TESTS OF GRAPHITE WITH BISMUTH AND BISMUTH-LITHIUM SOLUTIONS J. R. DiStefano Samples of graphite with varying densities and pore diameters were exposed to Bi—0.17 wt % (4.8 at. %) Li and Bi—3 wt % (48 at. %) Li in capsule tests for 3000 hr at 650°C. Two of the graphites (Table 7.1) were pitch impregnated to increase their densities and reduce their pore sizes.! The relatively high densities of these graph- ites indicate that impregnation was effective, but the pore size distribution in the samples shows that some of the larger pores were unfilled or only partially filled. Specimens were graphite rods 6 mm (0.24 in.) X 38.1 mm (1.5 in.) long that were threaded into an ATIJ graphite holder. The specimens and holder fit into a graphite capsule which contained the bismuth-ithium solution (Fig. 7.1). The entire assembly was sealed in a stainless steel outer capsule by welding in argon. Sam- ples exposed to Bi—0.17 wt % (4.8 at. %) Li showed little evidence of penetration except in low-density areas (Fig. 7.2). Samples exposed to Bi—3 wt % (48 at. %) Li were penetrated more uniformly, and the depth 1. All graphites were fabricated by C. R. Kennedy of the Carbon and Graphite Group, Metals and Ceramics Division, ORNL. Table 7.1. Penetration of graphite by bismuth-lithium solutions in capsule tests for 3000 hr at 650°C Maximum pore Graphite Bul}( Range.of diameter that Penetration (mils) . e s density pore diam contributes 10% - - - identification (g/em?) (1) to total porosity Bi-0.17% Li ~ Bi-3% Li () 33-6K 1.84 0.1-1 1 0 5 44-25K49 1.84 0.1-2 1.2 0-172 8 33-38K 1.80 0.1-2 1.5 05> 5 44-26K9 1.80 0.1-3.5 1.5 0-2b . 8 44-23K 1.59 0.1-4.5 4.5 0—2b 15 ZImpregnated. bNonuniform, penetration in one or two areas only. 132 133 ORNL-DWG 75-14509 rt—— STAINLESS STEEL OUTER CONTAINER ATJ GRAPHITE CAPSULE GRAPHITE ROD SPECIMEN (0.24-in-0CIAM x 1.5-in. LONG) Bi—-Li INCHLS | Fig. 7.1. Graphite (bismuth-lithium) capsule test assembly. of penetration increased with increasing pore size and decreasing density. Results from previous tests have been inconclusive as to the effect of lithium concentra- tion in bismuth on penetration of graphite. In the cur- rent series all graphites were penetrated to a greater extent by Bi—3 wt % (48 at. %) Li than by Bi—0.17 wt % (4.8 at. %) Li. Tests of 10,000 hr duration with these graphites are continuing. - 7.2 THERMAL GRADIENT MASS TRANSFER TEST OF GRAPHITE IN A MOLYBDENUM LOOP J. R. DiStefano Although graphite has low solubility in pure bismuth (less than 1 ppm at 600°C), capsule test results have shown that higher carbon concentrations are present in Bi—2 wt % (38 at. %) Li and Bi—3 wt % (48 at. %) Li solutions after contact with graphite. To avoid the joining problems associated with fabrication of a graph- ite loop, a molybdenum loop was constructed, and interlocking, tabular graphite specimens were suspended in the vertical hot- and cold-leg sections.? In addition to mass transfer of graphite from hot- to cold-leg areas, penetration of graphite by bismuth-lithium and mass transfer between graphite and molybdenum were evalu- ated. 7.2.1 Weight Changes The loop (CPML4) circulated Bi—2.4 wt % (42 at. %) Li for 3000 hr at 700°C (approximately) maximum temperature and 600°C minimum temperature. Weight changes in the graphite samples are given in Tables 7.2 and 7.3. After the bismuth-lithium solution was drained from the loop, the samples were removed and weighed (“after-test” column in Tables 7.2 and 7.3). Sub- sequently, they were cleaned at room temperature in ethyl alcohol and in an H, O—HNO, (100 ml H,0-30 ml 90% HNO,) solution to remove bismuth-lithium adhering to the surfaces of some samples. Samples from the cold leg were weighed and then kept in air for two days prior to the alcohol treatment. After soaking in alcohol, these samples showed larger weight gains than the after-test weight gains, and this is attributed to reac- tion of lithium in the sample with moisture in the air during the two-day period. All samples showed large weight gains (33—67%), and gains in hot-leg samples were, on the average, larger than those in the cold-leg samples. 7.2.2 Compositional Changes Graphite samples were analyzed before and after treatment with H, O—HNO,, and the results are shown in Table 7.4. These results indicate that bismuth was primarily responsible for the large weight increases and that samples picked up molybdenum, but treating them with H,O-HNO; completely removed the molyb- denum. An electron-beam microprobe analysis of a graphite sample before acid cleaning showed that molybdenum was present on the outer surface of the specimen (Fig.-7.3). Chemical analyses of other graphite samples after acid cleaning are shown in Table 7.5. Analyses of bismuth-lithium samples from the loop are shown in Table 7.6. For sampling, the hot leg was sectioned so that one sample came from the surface that was in contact with the molybdenum tube wall while the other sample was taken from the interior of the section, away from the wall. The concentration of carbon in the melt was highest in the sample from the hot leg, and both molybdenum and carbon concentra- 2. 1. R. DiStefano, MSR Program Semiannu. Progr. Rep. Feb, 28, 1975, ORNL-5047, pp. 140—-41. o =1.8g/cm’ 1.8g/cm ~75% of pores = 1 diam ~25% of pores = 1-4y diam 8i-0.17 Li Fig. 7.2. Penetration of graphite as a function of structure of graphite and lithium in bismuth. Conditions: 1000 hr at 650 135 Table 7.2. Weight increases in ATJ graphite hot-leg samples from CPML-4 Weight (g) Weight :Sfiglei Befaore After A’fte‘r A'fte.r increase test test .cleanmg . cleaning @ % in alcohol in H, O-HNO, 59 0.4563 0.8244 0.8122 0.7006 0.2443 54 7 0.5220 0.9706 0.9656 0.8298 0.3078 59 9 0.4994 0.9551 0.9336 0.8298 0.3304 66 10 0.4800 0.9596 09819 0.7944 0.3144 66 11 0.4326 0.8339 0.8303 0.7102 0.2776 64 12 0.4594 0.9302 0.9263 0.7669 0.3075 67 13 0.4753 0.9796 0.9764 0.7686 . 0.2933 62 16 0.4632 09175 09152 0.7658 0.3026 65 17 04742 0.9099 0.9058 0.7694 0.2952 62 18 4.5070 0.9005 0.8956 0.7975 0.2905 57 19 0.4709 0.8189 0.8142 0.7168 (.2459 52 20 0.5369 -0.9184 0.9149 0.8107 0.2738 51 21 0.5163 0.8652 0.8629 0.7626 0.2453 48 22 0.5184 0.8499 0.8459 0.7591 0.2407 46 23 0.5389 0.9103 0.9066 0.7475 0.2067 38 240 0.5405 0.8545 0.8497 (.7235 0.1830 33 4Tgp of hot leg; temperature: 680—700°C. bBottom of hot leg; temperature: 600—620°C. Table 7.3. Weight changes in ATJ graphite cold-leg samples from CPML-4 Weight (g) After Weight Sample Before After ’ standing in . Af.ter , increase number test test air for two days cleaning in © %) and then cleaning H, O-HNO, in alcohol S 274 3.4614 0.7469. 0.7622 0.6408 0.1794 39 28 0.4821 0.8284 0.8412 0.6260 0.1439 30 29 0.4717 0.7793 0.7913 0.6433 0.1716 36 30 04776 - 0.7136 0.7239 0.6291 0.1515 32 31 0.4848 0.7209 0.7315 0.6396 0.1548 32 32 0.4827 0.7629 0.7744 0.6647 0.1820 38 33 0.4788 0.7193 0.7300 0.6331 0.1543 32 34 0.4708 0.7566 0.7672 0.6536 0.1828 39 35 0.4624 0.7382 0.7485 0.6290 0.1666 36 36 0.4823 0.7496 0.7598 0.6343 0.1520 32 37 0.4667 0.7331 0.7432 .0.6239 0.1572 34 38 04713 0.7513 0.7619 0.6418 0.1705 36 39 0.4745 0.7498 0.7603 0.6506 0.1761 37 40 0.4628 0.7285 0.7390 0.6233 0.1605 35 41 0.4725 0.7458 0.7553 0.6419 0.1694 36 42 0.4800 0.7427 0.7525 0.6526 0.1726 36 43b 04766 0.7280 0.7401 0.6478 01712 36 2Top of cold leg; temperature: 660—680°C. bRottom of cold leg; temperature: 620—630°C. 136 Table 7.4. Chemical analyses of graphite tions were higher than were found previously in quartz exposed to bismuth-lithium solution loop tests circulating Bi-0.01 wt % (03 at. %) Li.? ina mlybdeaim lodp Quartz loop test 11 contained molybdenum samples, and analysis of the bismuth-lithium solution after test Sample number Condition ;‘v% showed that it contained 2.5 ppm molybdenum. Quartz : : loop 8 contained samples of three different grades of 4 (hot leg) Uncleaned 46 0.6 0.1 graphite, and the bismuth-ithium solution contained 10 4 (hot h‘;» Acid cleaned 03 <0.01 to 15 ppm carbon after the test. 26 (cold leg) Uncleaned 43 1.2 O e e e 26 (cold leg) Acid cleaned 40 04 <0.01 3. 0. B. Cavin and L. R. Trotter, MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL4676, p. 228. Y-133056 40 pm BACKSCATTERED ELECTRONS Bi Mg X-RAYS Mo LG X-RAYS Fig. 7.3. Electron beam scanning images of graphite exposed to bismuth. View (4) is backscattered electron picture of sample surface; dark material is graphite and bright material is bismuth and molybdenum as indicated by (8) and (C) Table 7.5. Chemical analyses of graphite samples? from CPML-4 Sample ] Concentration (wt %) Location number Bi Li Mo 1 Hot leg 37 0.35 <0.01 9 Hot leg 42 0.44 <0.01 15 Hot leg 46 0.47 <0.01 19 Hot leg 36 0.35 <0.01 24 Hot leg 23 0.25 <0.01 27 Cold leg 36 0.35 <0.01 31 Cold leg 21 0.25 <0.01 36 Cold leg 19 0.26 <0.01 42 Cold leg 23 0.26 <0.01 ?Samples acid cleaned in H, O-HNO, prior to analysis. Table 7.6. Analysis of bismuth-lithium solution from CPML-4 Concentration? Sample location | C Mo Li (ppm) (ppm) (%) Hot leg (core)? 43 17 2.3 Hot leg (surface)® 106 102 3.0 Cold leg (core)? 24 14 1.6 20n a weight basis. binterior sample. €Surface sample in contact with molybdenum tube wall. 7.2.3 Microstructural Changes Selected graphite samples from hot- and cold-leg regions are shown in Fig. 7.4. The white phase distrib- uted throughout the samples is bismuth. These samples were acid cleaned and it is evident that bismuth was dissolved from the area near the surface. Molybdenum samples from hot- and cold-leg regions are shown in Fig. 7.5. Surface layers measuring 0.015 to 0.025 mm (0.6—1 mil) thick were found on the hot-leg sample. In some areas there was a single layer, while a double layer was found in other areas. Electron-beam microprobe analysis indicated the single layer and/or outer layer to be primarily molybdenum. This layer was much harder than the base metal (1000-1200 DPH compared with about 200 DPH), indicating that it is probably Mo, C. Where there is a double layer the outer layer appears to be Mo, C, but the inner layer is primarily bismuth. One explanation is that the Mo,C layer cracked and/or spalled, allowing the entry of bismuth which did not drain when the test was terminated. The molybdenum sample from the cold leg also exhibited a surface layer less than 0.1 mil thick that was similar in composition to that found in the hot leg. Samples of molybdenum from hot and cold legs have been submitted for chemi- cal analysis. 7.2.4 Discussion of Results The principal objective of this experiment was to eval- uate temperature-gradient mass transfer of graphite in bismuth containing a relatively high concentration of lithium. However, mass transfer data were obscured by the gross pickup of bismuth by the graphite samples. Previous capsule and quartz loop tests with ATJ graph- ite had indicated much less intrusion of the graphite by bismuth than occurred in the molybdenum loop test. This suggests that the permeability of ATJ graphite to bismuth-lithium does not depend simply on the po- rosity of the graphite. It is generally accepted that some fraction of the pores in graphite is effectively sealed off and contributes nothing to flow. Therefore, the con- nected pore system controls the permeability, The shape of the connected pores influences the type of flow and the length of the path the fluid takes through the sample. For a nonwetting liquid, the external pres- sure forcing the liquid into the pores, AP, must over- come the surface tension of the liquid. This defines a critical pore radius, 7., and until the pressure exceeds the value given by AP =2y cos 8/r, (1) where 7 is the surface tension and 8 the wetting angle, the pore cannot support flow. Thus, for a given AP, 7. is the minimum pore size that will be penetrated. In both the metal and quartz thermal convection loops AP is determined by the argon overpressure (<1 atm) and the height of bismuth-lithium solution above the sample, and these were essentially the same in both types of tests. Temperature affects both ¢ and @0, but all of the tests were operated under similar time-temperature-AT conditions. Graphite samples used in the quartz loop tests had almost four times the surface area of the tabu- lar specimens used in the current test, but they were almost three times as thick. The larger surface area of the quartz loop specimens should have increased the relative amount of bismuth-lithium intrusion, but the greater thickness of these samples would reduce the per- centage increase. ATJ graphite samples from the quartz loop tests increased in weight by 0.1 to 0.6 wt %, far less than the 30 to 67 wt % increases noted in samples from the current loop test. Thus, specimen geometry alone does not seem to explain the differences noted. However, the surface tension o and wetting angle § were Hot Lea No. 10 Hot Leg No. 13 Fig. 7.4. Graphite samples after exposure to Bi—2.4% Li for 3000 hr in CPML-4. Y-132704 Hot Leg No, 23 600 700 i -8 B il " z S & o H 100 200 8€T Y-133405 60 MICRONS ——————1—500x—L 40 S == P e o _— ' = N = — B3 - M S = 9 - et e > 3 3 S = i —= HE : L ~ e —— S AA = = - - Top of Hot Leg: 700°C g Top of Hot Leg: 700°C 5 g 0.001 Bottom of Hot Leg: 600°C specimens. probably different because the lithium concentrations of the bismuth-ithium solutions were different, and molybdenum was present in the current test. It is pos- sible that the presence of molybdenum on the surfz of the graphite had a marked effect on the contact angle 0. In an earlier series of tests, the bismuth content of e graphite specimens was much higher when they were tested in molybdenum capsules instead of graphite cap- Bottom of Cold Leg: 620°C . Molybdenum tube wall from thermal convection loop CPML-4 that circulated Bi-2.4% Li and contained graphite sules.* Accordingly, data on the wetting of graphite by bismuth containing lithium and other constituents of processing solutions would be useful for predicting the resistance of graphite to penetration. 4. J. R. DiStefano and O. B. Cavin, MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-S047, pp. 137 -39, Part. 4. Fuel Processing for Molten-Salt Reactors J. R. Hightower, Jr. The activities described in this section deal with the development of processes for the isolation of protac- tinium and for the removal of fission products from moltensalt breeder reactors. Continuous removal of these materials is necessary for molten-salt reactors to operate as high-performance breeders. During this report period, engineering development progressed on continuous fluorinators for uranium removal, the metal transfer process for rare-earth removal, the fuel recon- stitution step, and molten salt—bismuth contactors to be used in reductive extraction processes. Work on chemistry of fluorination and fuel reconstitution was deferred to provide experienced personnel for the prep- aration of salt for the TeGen-2 and -3 experiments (Sect. 6.17). The metal transfer experiment MTE-3B was started. In this experiment all parts of the metal transfer process for rare-earth removal are demonstrated using salt flow rates which are about 1% of those required to process the fuel salt in a 1000-MW(e) MSBR. This experiment repeats a previous one (MTE-3) to determine the reasons for the unexpectedly low mass transfer coeffi- cients seen in MTE-3. During this report period the salt and bismuth phases were transferred to the experi- mental vessels, and two runs with agitator speeds of 5 rps were made to measure the rate of transfer of neo- dymium from the fluoride salt to the Bi-Li stripper solu- tion. However, in these runs the fluoride salt was en- trained at low rates into the LiCl, which resulted in depletion of the lithium from the Bi-Li solution in the stripper. Fuel-salt entrainment was unexpected, since no entrainment was seen in experiment MTE-3 under (as far as can be determined) identical conditions. The measurement of mass transfer coefficient in these first two runs was not compromised by the entrainment. The measured mass transfer coefficients were lower than predicted by literature correlations, but the values are comparable to those obtained from experiment MTE-3. Mechanically agitated nondispersing salt-metal con- tactors of the type used in experiment MTE-3B are of interest, because entrainment of bismuth into the fuel salt can be minimized, because very high ratios of bis- muth flow rate to salt flow rate can be more easily handled than in column-type contactors, and because these contactors appear to be more easily fabricated from molybdenum and graphite components than are column-type contactors. Attempts were made to meas- ure entrainment rates of fluoride salt in bismuth and entrainment rates of bismuth in fluoride salt under con- ditions where the phases were not dispersed and under conditions where some phase dispersal was expected. These measurements were made in the 6-in.-diam (0.15-m) contactor installed in the Salt-Bismuth Flow- through Facility. The results indicate that mild phase dispersal with its concomitant high mass transfer coeffi- cients might be allowable in the reductive extraction processes. We are continuing development of methods for measuring mass transfer coefficients in mercury- water systems to learn how to scale up contactors which would be used with salt and bismuth. A nonradioactive demonstration of frozen salt cor- rosion protection in a continuous fluorinator requires a heat source that is not subject to attack by fluorine in the fluorinator. To provide such a heat source for future fluorinator experiments we have continued our studies of autoresistance heating of molten salt. During the report period we have completed new equipment for studying autoresistance heating of molten salt in a flow * system similar to a planned continuous fluorinator ex- 140 periment; three preliminary runs have been made with the equipment. The design was started for a facility for developing continuous fluorinators, and equipment is being installed for an experiment to demonstrate the effectiveness of frozen salt for protection against fluo- rine corrosion. The uranium removed from the fuel salt by fluorina- tion must be returned to the processed salt in the fuel reconstitution step before the fuel salt is returned to the reactor. An engineering experiment to demonstrate the fuel reconstitution step is being installed. In this experi- ment gold-lined equipment will be used to avoid intro- ducing products of corrosion by UF¢ and UF;. Alterna- tive methods for providing the gold lining include elec- troplating and mechanical fabrication. The choice be- tween the two depends on availability of gold from ERDA precious-metal accounts and the price of gold from the open market. Instrumentation for the analysis 141 of the vessel off-gas streams has been installed and is being calibrated. Future development of the fuel processing operations will require a large facility for engineering experiments. A design report is being prepared to define the scope, estimated design and construction costs, method of accomplishment, and schedules for a proposed MSBR Fuel Processing Engineering Center. The building will provide space for preparation and purification of salt mixtures, for engineering experiments up to the scale required for a 1000-MW(e) MSBR, and for laboratories, maintenance areas, and offices. The estimated cost of this facility is $15,000,000; and authorization is pro- posed for FY 1978. 8. Engineering Development of Processing Operations * J. R. Hightower, Jr. 8.1 METAL TRANSFER PROCESS DEVELOPMENT H. C. Savage During this report period the salt and bismuth solu- tions were charged to the process vessels of the metal transfer experiment MTE-3B.! Two experiments were completed in which the rate of removal of neodymium from molten-salt breeder reactor fuel salt (72-16-12 mole % LiF-BeF,-ThF,) was measured. The MTE-3B process equipment (Fig. 8.1} consisted of three interconnected vessels: a 14-in.-diam (0.36-m) fuel salt reservoir, a 10-in.-diam (0.25-m) salt-metal con- tactor, and a 6-in.-diam (0.15-m) rare-earth stripper. The salt-metal contactor is divided into two compart- ments interconnected through two 0.5-in.-high (13-mm) 1. H. C. Savage, Engineering Development Studies for Molten-Salt Breeder Reactor Processing No. 20, ORNL- TM-4870 (in preparation). by 3-in.-wide (76-mm) slots in the bottom of the divider. Bismuth containing thorium and lithium is cir- culated through the slots. Thus fluoride fuel salt was in contact with the Bi-Th in one compartment, and LiCl was in contact with the Bi-Th in the other compart- ment. The stripper contains lithium-bismuth solution (595 at. %) in contact with the LiCl. Mechanical agita- tors having separate blades in each phase in the con- tactor and stripper were used to promote mass transfer across the three salt-metal interfaces. The fluoride fuel salt was circulated between the reservoir and contactor by means of a gas-operated pump with bismuth check valves. The LiCl was circulated between the stripper and contactor by alternately pressurizing and venting the stripper vessel. The bismuth-thorium phase was circulated between the two compartments of the contactor by the action of the agitators, and no direct measurement of this flow rate was made during the experiment; however, meas- urements made in a mockup using a mercury-water system indicated that the Bi-Th circulation rate between ORNL-DWG-71-147-RI ‘ AGITATORS VENT LEVEL ELECTRODES LEVEL ELECTRODES FLUORIDE " SALT PUMP VENT —— E —— [ — ARGON SUPPLY ARGON SUPPLY 125 V2 -7 AV 77 Loin | - s - . 4 b - < / \‘\—ucr~/’4 72-16-12 mole % Bi-Th \\ Li-Bi LiF'Ber‘ThF4 FL%%?PE SALT - METAL RARE EARTH RESERVOIR CONTACTOR STRIPPER Fig. 8.1. Flow diagram for metal transfer experiment MTE-3. 142 the two compartments should be high enough to keep the concentration of rare earths in both compartments essentially the same.> This was found to be the case in the two experiments in MTE-3B. In this experiment, neodymium is extracted from the fuel carrier salt into the thorium-bismuth solution. Next, the neodymium is extracted from the thorium- bismuth into molten LiCl, and finally, the neodymium is stripped from the LiCl into bismuth-lithium alloy. Operating variables in the experiment are: . the flow rate of the fluoride fuel salt between the fuel salt reservoir and the contactor, . the flow rate of the lithium chloride salt between the contactor and the stripper vessel, . the degree of agitation of the salt and bismuth phases in the contactor and stripper, . the amount of reductant (lithium) in the bismuth phase in the contactor. The operating temperature of the system is Vv650°C. Overall mass transfer rates for representative rare-earth fission products are determined by adding the rare earth to the fluoride fuel salt in the reservoir and observing the rate of transfer of the rare earth across the three salt-bismuth interfaces as a function of time by periodic sampling of all phases. 2. H. O. Weeren and L. E. McNeese, Engineering Develop- ment Studies for Molten-Salt Breeder Reactor Processing No. 10, ORNL-TM-3352 (September 1974) pp. 57-59. 143 During the course of the experiments the concentra- tions of neodymium in each phase were determined by counting the 0.53-MeV gamma radiation emitted by 147Nd tracer added to the neodymium originally in the fuel salt. This provided a rapid method for following the transfer rate. More accurate data necessary for calculat- ing the overall mass transfer coefficients at each of the three salt-metal interfaces were obtained by analyzing samples of the salt and bismuth phases for total neo- dymium via an isotopic dilution mass spectrometry technique. Use of this technique allows measurement of neodymium concentrations as low as 0.01 ppm (wt). 8.1.1 Addition of Salt and Bismuth Phases to Metal Transfer Experiment MTE-3B The quantities of salts and bismuth charged to the process vessels of experiment MTE-3B are listed in Table 8.1. All internal surfaces of the carbon-steel ves- sels were hydrogen treated at 650°C for 7 hr to re- move.any oxides prior to the addition of the salt and bismuth solutions. The auxiliary charging vessels used in the additions were also hydrogen treated. Subsequently a purified argon atmosphere was maintained in all the vessels to prevent oxide contamination (via ingress of air or moisture) of the vessels and process solutions. The charging vessels were 10-in.-diam (0.25-m) carbon-steel vessels of about 22 liters (0.022 m®) in volume equipped with .electric heaters for melting the salts and bismuth. Nozzles and access ports were pro- Table 8.1. Quantities of salts and bismuth for experiment MTE-3B Volume? at Weicht Material Vessel 650°C (kg) g-moles " (liters) E Fluoride fuel salt? Reservoir - 29.4 97.0 1535 (72-16-12 mole % LiF-Bel', -ThF,) ‘ ' Fluoride fuel salt Contactor 3.1 10.2 161 (72-16-12 mote % LiF-BeF,-ThF,) Bismuth-thorium [~1500 ppm (wt) Th, Fluoride salt .29 27.6 132 ~50 ppm Li] side of contactor ' Bismuth-thorium [~1500 ppm (wt) Th, LiCl side of 3.5 33.8 161 ~50 ppm Li] contactor Lithium chloride Contactor 2.9 4.3 101 Lithium chloride Stripper 3.8 5.6 132 Bismuth-5 at. % lithium in " Stripper 4.3 41.8 200 stripper 2Densities at 650°C: fluoride fuel salt = 3.30 g/cc; LiCl = 1.48 g/ce; Bi = 9.66 g/cc. bmole weight = 63.2 g. 144 vided for the addition of the salts and bismuth, argon and hydrogen purge gas lines, and lines required to transfer the salt and bismuth phases into the process vessels. Bismuth, hydrogen treated in the charging vessel to remove oxides, was the first material to be added to the contactor. The fluoride fuel salt was then contacted in the charging vessel (using argon sparging) with a bismuth—0.15 wt % thorium solution (50% of Th satu- ration) for several days prior to transfer into the fuel- salt reservoir and the fluoride salt compartment of the contactor. Thorium metal (0.1197 kg) was then added to the 61.4 kg of bismuth in the contactor. This quan- tity of thorium is about 50% of the amount that would be soluble and was calculated to produce a lithium con- centration of 40 ppm (wt) in the thorium-bismuth phase in the contactor based on previously reported data® on the distribution of thorium and lithium be- tween molten bismuth and fluoride fuel salt. Following the additions of bismuth to the contactor and the fluoride fuel salt (72-16-12 mole % LiF-BeF,- ThF,;) to the contactor and fuel-salt reservoir, a new charging vessel was installed for makeup and charging of the bismuth—-5 at. % lithium to the stripper and the LiCl to the contactor and stripper. First, bismuth was added to the charging vessel and was hydrogen treated to remove oxides by sparging with hydrogen at ~600°C (873°K) for ~7 hr. The charging vessel contained 67.87 kg of bismuth to which was added 0.120 kg of lithium metal to produce the bismuth—5 at. % lithium for the stripper. Part of the bismuth--5 at. % lithium solution (41.8 kg) was then transferred into the stripper vessel. Thorium metal (0.109 kg) was added to the 26 kg of bismuth-lithium solution remaining in the charge vessel, and 15.88 kg of LiCl that had been oven dried at 200°C (473°K) was added to the charge vessel. The bismuth- lithium-thorium and LiCl phases were sparged with argon, using a gas-lift sparge tube, for four days. The" LiCl was then transferred into the LiCl side of the con- tactor and the stripper vessel. The salt and bismuth solutions were filtered through molybdenum filters [~30 u (3.0 X 107> m) in pore diameter| installed in the transfer lines during transfer from the charging vessels into the MTE-3B process vessels. 8.1.2 Run Nd-1 For the first run in MTE-3B, 3300 mg of NdF; (2360 mg of Nd) was added to the 97 kg of fluoride fuel salt (72-16-12 mole % LiF-BeF,-ThF,) in the fuel salt reser- voir on June 6, 1975, The neodymium contained 72.2 mCi of '*7Nd tracer (r,, = 11 days) at the time of addition. The neodymium concentration in the fuel salt in the reservoir was calculated to be 24 ppm (wt), which approximates that expected in the fuel salt of a single- region 1000-MW(e) MSBR. Neodymium was chosen as the representative rare-earth fission product for the first series of experiments in MTE-3B for several reasons: 1. results could be compared with those obtained using neodymium in the previous experiment,* MTE-3, 2. "7Nd tracer used for following the rate of transfer of neodymium has a relatively short half-life (11 days), which would prevent excessive levels of radio- activity in the experimental equipment as additional neodymium, containing '*7Nd, was added to the fuel salt during the experiment. 3. neodymium is one of the more important trivalent rare-earth fission products to be removed from MSBR fuel salt. An attempt was made to start the first run (Nd-1) on June 9, 1975. However, a malfunction in the electronics of the speed control unit for the stripper-vessel agitator prevented startup. After this unit was repaired, run Nd-1 was started on June 15, 1975, and the scheduled period of operation (100 hr) was completed on June 20, 1975. Operating conditions of run Nd-1 were: 650 to 660°C (923 to 933°K), 5 rps agitator speeds in both contactor and stripper, fluoride salt flow rate of 35 cc/min (5.8 X 1077 m3/sec), and LiCl flow rate of 1.2 liters/min (2.0 X 107° m?3/sec). After 100 hr of fluoride salt and LiCl salt circulation, the fluoride salt circulation was stopped and the run was continued for 16 hr. This was done to observe the expected larger decrease in the concentration of neo- dymium in the smaller amount of fluoride salt in the contactor (10.2 kg), as compared with the 107.2 kg contained in both the contactor and reservoir. These data would provide a more accurate measure of the rate of transfer of neodymium across the fluoride salt— bismuth—thorium interface. Finally, the circulation of LiCl was also stopped. The agitators in the contactor and stripper vessels were then operated for 24 hr over a three-day period (8 hr each day) to allow the salt and bismuth phases to equilibrate in an attempt to determine neodymium distribution coefficients between the phases. 3. L. M. Ferris, “Equilibrium Distribution of Actinide and Lanthanide Elements Between Molten Fluoride Salts and Liquid Bismuth Solutions,” J. Inorg. Nucl. Chem. 32, 2019-35 (1970). 4, Chem. Technol. Div. Annu. Progr. Rep. March 31, 1973, ORNL-4883, p. 25. The experimental equipment operated satisfactorily throughout run Nd-1. All operating variables were main- tained at desired conditions. Results obtained during run Nd-1 are discussed in Sect. 8.1.4. 8.1.3 Run Nd-2 Run Nd-2 was done with the same operating condi- tions as run Nd-1 except for run duration (139 hr in- stead of 100 hr). Prior to run Nd-2, 3590 mg of NdF, (2580 mg of Nd) containing 101 mCi of '*7Nd tracer was added to the 97 kg of fuel salt in the reservoir. Including the neodymium remaining in the fuel salt at the end of run Nd-1, estimated to be 18 ppm (wt), the neodymium concentration in the fuel salt in the reser- voir at the start of run Nd-2 is estimated to be 45 ppm. The neodymium concentration in the fuel salt in the contactor is estimated to be 9 ppm at the start of run Nd-2. We are uncertain of the amounts of neodymium in the other phases at the beginning of run Nd-2, as discussed in Sect. 8.1.4. Run Nd-2 was started on July 13, 1975, and was ter- minated on July 19, 1975, after 139 hr of operation. During the first 50 hr of operation the rate of transfer of neodymium into the lithium-bismuth phase in the stripper appeared to be about the same as observed during run Nd-1, based on counting of the !*7Nd tracer in samples taken at regular intervals. After about 60 hr of operation, the transfer of neodymium into the bismuth-lithium phase in the stripper suddenly stopped, and it was observed that neodymium was being ex- tracted from the bismuth-lithium phase in the stripper into the LiCl in the stripper and contactor. During the run a significant decrease in the emf between the strip- per vessel and the contactor occurred (from 160 mV to ~25 mV over a 30-hr period), indicating loss of lith- ium reductant from the bismuth-lithium phase. The run was terminated after 139 hr of operation, when it be- came clear that useful information could no longer be obtained and it appeared that fluoride salt was being entrained into the LiCl in the contactor. 8.1.4 Discussion of Results Subsequent investigation and results of chemical anal- yses of samples of the salt and bismuth phases indicate “that fluoride fuel salt was being entrained into the LiCl in the contactor throughout both runs Nd-1 and Nd-2. Estimates of the amount entrained are shown below: Estimated amount of fluoride salt transferred into LiCl Nd-1 (104 hr) Nd-1 and -2 (301 hr) Basis of estimate 0.292 kg Fluoride in LiCl phase Thorium in Bi-Li phase Increase in LiCl level in stripper 0.607 kg 0.400 kg 145 Based on fluoride analyses of LiCl samples taken during run Nd-1, the entrainment of fluoride salt appears to have occurred at a relatively constant rate throughout the run. The total amount of neodymium which trans- ferred into the Li-Bi phase in the stripper during run -Nd-1 is estimated to be 300 mg. The amount of neo- dymium contained in the entrained fuel salt is estimated to be 6 mg. Thus, most of the neodymium which trans- ferred into the Li-Bi in the stripper vessel was by mass transfer rather than as a result of entrainment. The reason for the observed entrainment is not clear at present. One explanation is that the 5.0-rps agitator speed is sufficient to cause entrainment (entrainment of fluoride salt into the chloride salt occurred in the pre- vious experiment MTE-3 at 6.7 rps, but not at 5.0 rps). Experiments are in progress for determining whether this explanation is correct, and results to date indicate that entrainment does not occur at 3.3 rps. Further experiments in MTE-3B will depend on determining the reason for the unexpected entrainment of fluoride salt into the LiCl. However, it appears feasible to continue rare-earth mass transfer experiments in MTE-3B by removing the LiCl (contaminated with fluoride salt) and the Li-Bi solution from the stripper vessel, after which purified LiCl and Li-Bi solution will be added to the system. The main purpose of the metal transfer experiment is to measure mass transfer coefficients for the rare earths at the various salt-metal interfaces in the system and to determine whether a literature correlation® (based on studies with aqueous-organic systems) which relates many transfer coefficients to the agitator speed and other physical properties of the system is applicable to molten salt—bismuth systems. Data obtained from run Nd-1 have been analyzed, and estimates have been made of overall mass transfer coefficients for neodymium at the three salt-metal interfaces. Even though entrainment of fluoride salt into LiCl occurred during run Nd-1, it is believed that the mass transfer rate for neodymium was not significantly affected. The concentration of fluoride in the LiCl at the end of run Nd-1 was ~1.3 wt % or 0.03 mole fraction. Based on previous studies,’® the dis- tribution coefficient, D;,, for neodymium between the molten bismuth-thorium solution and LiCl (mole frac- tion Nd in bismuth/mole fraction Nd in LiCl), would be decreased by ~20%, while the distribution coefficient for thorium would be decreased by a factor of V150. This would result in a decrease in the separation factor 5. 1. B. Lewis, Chem. Eng. Sci. 3,248-59 (1954). 6. L. M. Ferris et al., “Distribution of Lanthanide and Actinide FElements Between Liquid Bismuth and Molten LiCl-LilF and LiBr-LiF Solutions,” J. lnorg. Nucl. Chem. 34, 313-20 (1972). 146 between neodymium-thorium from ~10* to ~102%, with an increase in the amount of thorium transferred into the LiCl. The rate of transfer of neodymium across the three salt-metal interfaces was determined by analyses of sam- ples taken throughout the run. Two analytical methods were used: (1) counting of the 0.53-MeV gamma radia- tion emitted by the !*7Nd tracer and (2) isotopic dilu- tion mass spectrometry. Based on counting of the 147Nd tracer, a material balance of the neodymium of >95% obtained at the end of run Nd-1 indicated that about 13% of the neodymium originally added to the fuel salt reservoir had been transferred into the Li-Bi solution in the stripper. The counting technique is a rapid method and pro- vided information on the rate of transfer while the runs were in progress. However, it does not provide the accuracy required for calculation of the overall mass transfer coefficients, particularly in the Bi-Th and LiCl phases in the contactor and stripper vessels in which the neodymium concentrations are usually less than 1 ppm (wt). The isotopic dilution analysis is capable of accu- rately determining neodymium concentration down to ~0.01 ppm (wt), and results obtained from the isotopic dilution technique for the Bi-Th and LiCl phases were used for calculations of the overall mass transfer coeffi- cients. Values for distribution coefficients for neodymium at the three salt-metal interfaces were measured at the end of run Nd-1 for comparison with those calculated from the data of Ferris®>” (Table 8.2). The experimental values are in reasonable agreement with the calculated values in the absence of fluoride contamination;indica- 7. L. M. Ferris et al., “Distribution of Lanthanide and Actinide Elements Between Molten Lithium Halide Salts and Liquid Bismuth Solutions,” J. lnorg. Nucl. Chem. 34,2921-33 (1972). Table 8.2. Distribution coefficients? for neodymium in experiment MTE-3B, run Nd-1 Salt-metal interface Calculated? Experimental Fluoride salt—Bi—Th 0.006 0.013 LiCl-Bi-Th 0.94 0.64 LiCl-Li-Bi 3.5x 107 >1x 10° ?Distribution coefficient = (m.f. Nd in bismuth)/(m.f. Nd in salt). bConditions: 650°C (923°K), Li concentration in Bi—Th = 40 ppm, Li concentration in Li—Bi = 5 at. %, no fluoride in LiCl phase. ting that the distribution coefficients for neodymium were not seriously affected by the entrainment of the fluoride salt into the chloride. Data obtained during run Nd-1 were analyzed by simultaneous solution of seven time-dependent differen- tial material-balance equations (Fig. 8.2) that relate the ORNL DWG. 7I-13962R2 FLUORIDE LiCl SALT SAl Li-BISMUTH dx i _ _ —_ dXz _ _& _ Vza— z - Killi()(2 DA)+ F1 (X1 Xz) dxs _ X3 _ - Vagp = + KA (%58 - Fy (X3, X4 = — KyA, (X5DgXe) * Fp (X3~ Xg) ng:—i =+ KzAz (X4" DBX5)-F3(X5-XG) dx X Vegi~ =~ KaAs (Xe"fiz)*':flxs'xe) dx Xz Vgl T+ K3A3(X5—55) V = MOLAR VOLUME OF EACH PHASE F = FLOW RATE, MOLES /SEC. D, ,Dg,D. = RARE-EARTH DISTRIBUTION COEFFICIENTS, MOLES/MOLE A = AREA AT EACH INTERFACE, CM° ,,K.= RARE - EARTH OVERALL MASS TRANSFER COEFFICIENT, CM/SEC X,., = RARE- EARTH CONCENTRATION IN EACH PHASE, MOLES/CM?® EQUATIONS USED TO CALCULATE MASS TRANSFER COEFFICIENTS FOR METAL TRANSFER EXPERIMENT MTE-38B Fig. 8.2. Equations used to calculate mass transfer coeffi- cients for the metal transfer process experiment. V = volume of each phase; x = rare-earth concentration; ¢+ = time; 4 = mass transfer area; F = flow rate; D = rare-earth distribution coeffi- cient; K = overall mass transfer coefficient. rate at which the rare earths are transferred through the several stages to the distribution coefficients of the rare earth, the mass flow rates, and the mass transfer coeffi- cients at each salt-metal interface.® The set of equations was solved, using a computer program, by selecting values for the mass transfer coefficients which.resulted in the best agreement between the experimental data on rate of change of neodymium concentration in all phases in the system and the calculated values. Several trial-and-error iterations were required, using adjusted . values of mass transfer coefficients, until a “best-fit” solution was obtained. The final calculated results for run Nd- 1 are shown in Table 8.3, where the values for the overall mass transfer coefficients are given and are compared with values cal- culated by the correlation of Lewis.® The coefficients are lower than predicted and are similar to results obtained in the previous experiment, MTE-3.* Final analytical results for run Nd-2 are not yet available. However, for the first 50 hr of operation, the rate of accumulation of neodymium in the Li-Bi solution in the stripper appeared to be similar to that observed in run Nd-1. The significance of these absolute values of mass transfer coefficient cannot be assessed until the scaling laws in this type of contactor are known. 8. L. E. McNeese, Engincering Development Studies for Molten-Salt Breeder Reactor Processing No. 11, ORNL-TM- 3774 (in preparation). ~ 147 8.2 SALT-BISMUTH CONTACTOR DEVELOPMENT C. H. Brown, Jr. Mechanically agitated nondispersing salt-bismuth con- tactors are -being . considered for the protactinium removal step and the rare-earth removal step in the reference MSBR processing plant flowsheet. These con-. tactors have several advantages over packed column salt-bismuth contactors: 1. they can be operated under conditions that minimize entrainment of bismuth to the fuel salt returning to the reactor, 2. they can be fabricated’ more economically from graphite and molybdenum components, 3. they can handle more easily large flow-rate ratios of bismuth and molten salt. Experimental development of stirred interface contac- tors is being carried out in two different systems, a facility in which molten fluoride salt is contacted with bismuth containing a dissolved reductant and a system in which mercury and an aqueous electrolyte phase are used to simulate b1smuth and molten salt. These two systems and the development work performed during this report period are-described in Sects. 8.2.1 and 8.2.2. Table 8.3. Overall mass transfer coefficients? for neodymium. in metal transfer experiment MTE-3B, run Nd-1 K, (mm/sec)b K, (mm/sec)b K, (mm/sec)b Measured® Predicted value Measured¢ Predicted value Measured¢ Predicted value value (%) value (%) value (%) 0.0035 39 0.25 20 0.13 : 2.5 2Based on the neodymium in the salt phase. 1 1 1 . b—— =— +-————at fluoride salt—Bi~Th interface. ! 1+DB t LiCl-Bi—Th interf: = Ty T d -Bi1-—- erface. K. k. &, 1t Ly i—Th interta 1 1 1 ‘ o — =t , at LiCl- Li—Bi interface, K, k, kDo : , , = individual mass transfer coefficient, fluoride salt to bismuth, k k, = individual mass transter coefficient, bismuth to fluoride salt, k, = individual mass transfer coefficient, bismuth to lithium chloride, k, = individual mass transfer coefticient, lithium chloride to bismuth, where DA = distribution coefficient between fluoride salt and bismuth, DB = distribution coefficient between chloride salt and bismuth, D = distribution coefficient between chloride salt and lithium-bismuth, CAgitator speed is 5.0 rps. 8.2.1 Experiments with a Mechanically Agitated Nondispersing Contactor in the Salt-Bismuth Flowthrough Facility Operation of a facility has continued in which mass transfer rates are being measured between molten LiF- BeF,-ThF, (72-16-12 mole %) and molten bismuth in a mechanically agitated nondispersing contactor. The equipment consists of a graphite-lined stainless steel vessel, salt and bismuth feed and receiver vessels, and the contactor vessel. In the first of these the salt and bismuth phases are stored between runs. The other vessels allow for treatment of the phases with HF and H,. The contactor consists of a 6-in.-diam carbon-steel vessel containing four l-in.-wide vertical baffles. The agitator consists of two 3-in.-diam stirrers having four noncanted blades. A %-in.-diam overflow at the inter- face allows removal of interfacial films, if present, with the salt and metal effluent streams. During a run the salt and bismuth phases are fed to the contactor by con- trolled pressurization of the respective feed tanks; the phases return to the receiver vessels by gravity flow. A detailed description of the facility and operating pro- cedures has been previously reported.” A total of nine mass transfer runs have been completed to date along with one hydrodynamic run intended to determine the amount of entrainment of one phase into the other at a series of different agitator speeds. Results from the nine mass transfer runs have been previously reported.’ ~!2 The experimental procedure for, and résults obtained from, the hydrodynamic run and treatment of the salt and bismuth with HF and H, are discussed in the re- mainder of this section. Experimental operation during the hydrodynamic run. The hydrodynamic run was performed with salt and bismuth flow rates of 150 and 140 cc/min respectively. The agitator was operated at three dif- ferent speeds during the run, 250, 310, and 386 rpm. At 250 and 310 rpm, three sets of unfiltered salt and bis- muth samples from the contactor effluent streams were taken at 4-min intervals. Three sets of unfiltered ef- fluent samples were also taken with the agitator operat- ing at 386 rpm, but the samples were taken at 2-min intervals. To avoid contamination of the sample contents with extraneous material, the sample capsules were cleaned of foreign matter by the following procedure: Gross amounts of salt or bismuth were first removed with a file; then the sample capsule was polished with emery cloth, and finally the capsule was washed with acetone. The sample capsules were then cut open with a tubing cutter, and the contents of each sample were drilled out 148 and visually inspected for the presence of one phase in the other. No such evidence of gross entrainment was found. In some of the salt samples, small flecks of metal were noticed which were probably small pieces of the sample capsule produced during the drilling operation. The contents of each sample were then sent to the Ana- lytical Chemistry Division for determination of bismuth present in the salt samples and beryllium present in the bismuth samples. It is assumed that any beryllium present in the bismuth is indicative of entrained fluoride salt. The results of these analyses are given in Table 8.4. The bismuth concentration in the salt samples shows a general decrease with increasing stirrer speed, with very low values occurring at the highest stirrer speed. It also seems evident that the bismuth concentration in the salt phase may have been a function of the run time, since - after the fourth sample the bismuth concentration re- mained at a relatively constant value of 50 %11 ppm, which is quite different from the values reported for the first four samples, which ranged from 1800 to 155 ppm. These results are significantly higher than those of Lindauer,'®> who saw less than 10 ppm of bismuth in 9. J. A. Klein et al., Engineering Development Studies for Molten-Salt Breeder Reactor Processing No. 19, ORNL-TM- 4863 (July 1975) pp. 21-38. 10. C. H. Brown, Jr., Engineering Development Studies for Molten-Salt Breeder Reagctor Processing No. 21, ORNL-TM- 4894 (in preparation). 11. J. A_ Klein, Engineering Development Studies of Moiten- Salt Breeder Reactor Processing No. 18, ORNL-TM-4698 (September 1974) pp. 1-22. ‘ 12. C. H. Brown, Ir., Engineering Development Studies for Molten-Salt Breeder Reactor Processing No. 20, ORNL-TM- 4810 (in preparation). 13. R. B. Lindauer, Engineering Development Studies of Molten-Salt Breeder Reactor Processing No. 17, ORNL-TM- 4178 (in preparation). Table 8.4. Analyses of salt and bismuth samples taken during the hydrodynamic run Agitator speed Bisample Bein Bi Salt sample Biin salt (rpm) number (ppm) number (ppm) 250 428 215 437 1800 250 429 125 438 205 250 430 215 439 155 310 431 85 440 270 310 432 910 441 53 310 433 442 34 386 434 110 443 64 386 435 175 444 54 386 436 50 445 43 fluoride salt in contact with bismuth in several different contacting devices. It is likely that sample contamina- tion is a contributing factor to the high bismuth concen- trations measured. Three possible sources of sample contamination have been reported:!? 1. contamination by withdrawing samples through a sample port which has been in contact with bismuth, 2. contamination during sample handling and in the analytical laboratory by the use of equipment rou- tinely used for bismuth analyses, 3. contamination from a low-density bismuth-containing material which may be floating on the salt surface. Since no maximum permissible rate of bismuth entrain- ment in the fuel salt going to the bismuth removal step or in the salt returning to the reactor from the fuel processing plant has been set, it is difficult to assess the significance of these results. However, the bismuth con- centrations in the salt do not seem to be inordinately high at the highest stirrer speed, and it seems likely that some degree of phase dispersal might be tolerated in order to achieve higher mass transfer rates. The beryllium concentrations in the bismuth samples at each agitator speed show both high and low values with no discernable dependence on agitator speed. These results agree well with previously reported data’® for beryllium concentration in the bismuth phase during mass transfer runs in this system at agitator speeds of 124, 180, and 244 rpm. Previous experiments with water-mercury and organic-mercury systems suggest entrainment of the light phase into the heavy phase at an agitator speed of about 170 rpm. The concentration of beryllium in the bismuth phase is not significantly different from previous results observed at lower agita- tor speeds. The effect of entrained fluoride salt in the bismuth would be most detrimental in the metal trans- fer process, where fluoride salt in the chloride salt phase decreases the separation factors between thorium and the rare-earth fission products. H,-HF treatment of salt and bismuth. The mass trans- fer runs completed to date in the salt-bismuth contactor have all been performed under conditions where the controlling resistance to mass transfer is in the inter- facial salt film. One final mass transfer run will be per- formed in which the bismuth-film mass transfer coeffi- cient is measured. In preparation for this run, the salt and bismuth in the graphite-lined treatment vessel were treated with HF diluted with H, to oxidize the reduc- tants present in the bismuth phase. The procedure used was essentially that reported previously.!® The salt and bismuth at "600°C were sparged with 25 scfh of 30% (mole) HF for 9 hr. The HF utilization decreased from 149 75% at the beginning of treatment to 35% during the ~ final 2 hr of treatment. Analysis of the salt and bismuth phases before and after treatment with HF and H, in- dicated that essentially all of the reductant in the bis- muth phase was oxidized by hydrofluorination. The uranium distribution ratio decreased from 740 moles/ mole prior to the treatment to 0.03 mole/mole after the HF-H, treatment. 8.2.2 Experiments with a Mechanically Agitated Nondispersing Contactor Using Water and Mercury We have continued development of a mechanically agitated nondispersing two-phase contactor, using an aqueous electrolyte and mercury to simulate molten salts and bismuth. As previously reported,'® we have investigated the feasibility of using a polarographic technique for measuring electrolyte-film mass transfer coefficients in this type of contactor. During this report period we have 1. tested three different anode materials, 2. produced cathodic polarization waves corresponding to the reduction of Fe®*, complexed with excess oxalate ions, at the mercury surface, 3. obtained and calibrated a slow-scan controlled- potential cyclic voltameter, 4. examined the quinone-hydroquinone redox couple as a possible alternate to the Fe®" -Fe** couple now being used. Modifications to experimental equipment. With the ex- ception of the tests made with the quinone-hydroquinone redox couple, all tests made during this period were performed with the equipment previously described.!® The equipment consists of the 5 X 7 in. Plexiglas con- tactor used in previous work with the water-mercury system. The mercury surface in the contactor acts as the cathode in the electrochemical cell. The cathode is elec- trically connected to the rest of the circuit by a %-in.- diam stainless steel rod electrically insulated from the electrolyte phase by a Teflon sheath. The anode of the cell is suspended in the aqueous electrolyte phase and consists of a metallic sheet formed to fit the inner perimeter of the Plexigas cell. The current through the cell is inferred from the voltage drop across a 0.1-§2 * 0.5%, 10-W precision resistor. The signal produced 14. B. A. Hannaford et al., Engineering Development Studies for Molten-Salt Breeder Reactor Processing No. 3, ORNL-TM- 3138 (May 1971) p. 30. 15. C. H. Brown, Jr., Engineering Development Studies for Molten-Salt Breeder Reactor Processing No. 22, ORNL-TM- 4041 (in preparation). across the resistor is recorded as the y coordinate on a Hewlett-Packard x,y plotter. The x coordinate on the plotter is produced by the potential difference between the mercury surface and a standard calomel electrode (SCE) suspended in the electrolyte phase. Prior to studies made on the quinone-hydroquinone system, a slow-scan controlled-potential cyclic volta- meter (potentiostat) was obtained from the Analytical Chemistry Division to replace the Hewlett-Packard dc power supply previously used. The cyclic voltameter is a three-electrode instrument which controls the potential between the mercury surface and a standard calomel reference electrode while passing a current between the auxiliary electrode and the mercury surface. Voltages can be scanned between +2 V vs SCE and —2 V vs SCE at a scan rate up to 1 V/min. The potentiostat can carry a current of up to 2.5 A between the auxiliary and mercury electrodes. Experiments with the Fe®* -Fe* system. The electro- lyte used for all the experiments performed during this report period was nominally 0.001 M Fe?* obtained from ferrous sulfate, 0.00025 M Fe3* obtained from ferric sulfate, and 0.8 M potassium oxalate. The oxalate ions form a stable complex with both the Fe3* and Fe?* 150 which facilitates measurements of the Fe®' reduction wave directly. Three anode materials have been tested:. copper, iron, and gold; and satisfactory polarization waves were pro- duced with all three materials. However, the copper and iron reacted with the electrolyte solution. This addi- tional side reaction caused poor reproducibility in the data and could also possibly alter the properties of the solution. To avoid this problem, an anode was fabri- cated by plating gold on a 0.0625-in.-thick sheet of nickel, which was formed to fit the inner perimeter of the electrochemical cell. ' Shown in Fig. 8.3 is a polarogram measured with the electrolyte described above in the 5 X 7 in. Plexiglas contactor using the gold anode with phase volumes of about 1.8 liters each and no agitation. The cell current is plotted as a function of the mercury surface potential vs the SCE. The current increases from zero at zero applied potential to a relatively constant value at an applied potential of about —0.35 V vs SCE. In this region continuous electrolysis is taking place in the cell, corresponding to reduction of Fe(C,0,);* ~ at the mer- cury cathode. In the region of applied potential from —0.35 V vs SCE to —0.80 V vs SCE, the cell current ORNL DWG 75- 8425 | | | I I ] | 3 — (] ‘ 2 | _ []!’ 1 - | — In i.l“ [ z w @ « = o 1= EVZ = 0.245 Y 0 L | I | l | I -0. -0.2 -03 -0.4 -05 -0.6 -07 -0.8 VOLTAGE vs SCE Fig. 8.3. Cathodic polarization wave for Fe(C,0,),> ~measured in the 5 X 7 in. rectan- gular contactor with no agitation. increases only a small amount; here, the current is limited by the rate of diffusion of the Fe(C;04)3>~ to the mercury surface, where this ion is reduced. The dif- fusion current can be related to the mass transfer . coefficient through the electrolyte film as previously discussed.*® The half-wave potential is defined as the potential at which the current is equal to one-half the limiting value. Figure 8.3 shows the measured half-wave potential for the ferric oxalate complex. The half-wave potential of —0.245 V measured in the contactor agrees well with the value reported in the literature of —0.24 V vs SCE for the reduction of ferric oxalate.'® Under ideal conditions, the diffusion current is directly proportional to the polarized electrode surface area and the bulk concentration of the limiting ion. To determine that the mercury surface was actually being polarized, two tests were performed. First, the anode surface area was decreased by about 48%. This had no effect on the magnitude of the diffusion current, indi- cating that the mercury surface (cathode) was polarized rather than the anode surface. In the second test, the ‘concentration of the ferric ion was doubled, but no concomitant increase in diffusion current was seen. Since the diffusion current is directly proportional to the concentration of the limiting ion (Fe®"), the current should have doubled. The only explanation for this behavior is that the Fe’ had been reduced by some contaminant in the system, possibly present in the mer- cury. This would have caused ferric ions to be present at only a very low concentration during cell operation, due to electrolytic oxidation of the ferrous iron. To eliminate the possibility of reductant being present in the mercury, a supply of purified mercury was ob- tained from the Analytical Chemistry Division. A test was performed using the purified meré:ury and an elec- trolyte having the same nominal Fe®" and Fe* concen- trations given above. Preparation of the electrolyte was completed in the absence of oxygen to preclude pos- sible oxidation of Fe?* to Fe*. Again, the anode sur- face area was decreased with no discernable decrease in the diffusion current, indicating -that the mercury sur- face was polarized. An increase of the Fe* concentra- tion from ~0.25 mM to ~0.5 mM resulted in an in- crease in the diffusion current by a factor of 2, indi- cating that the wave being measured was the ferric ion reduction wave. However, the half-wave potential was measured to be —-0.75 V vs SCE, which is about three times the reported value. To calculate the aqueous-film mass transfer coeffi- cient from polarographic data, the bulk concentration of the oxidized species must be accurately known. The 151 method for this measurement is a polarographic tech- nique using the dropping-mercury electrode. Samples of the electrolyte used in the second of the two tests men- tioned above were analyzed for Fe¥ and Fe?®' by this method. Results indicated that the Fe®* and Fe®* con- centrations were 1.7 and 0.28 mM respectively, which is in poor agreement with the expected values of 0.50 mM Fe* and 1.0 mM Fe**. One possible cause for the poor agreement is that the Fe?* was oxidized to Fe** during the period when the solution was held in the sample bottles. However, this was not expected, since the elec- trolyte had been sparged with argon to remove dissolved oxygen, and the sample bottles were purged with argon to remove air. To aid in determining if the reported analytical results were in error due to analytical technique or to method " of solution preparation, two standard solutions were prepared and sampled for analysis. One solution was . prepared to contain 56 ug/ml Fe®", and the other solu- tion was prepared to contain 56 ug/ml Fe**. Both solu- tions were 1 M in K,C, 0,4 -H; 0. Subsequent analytical results indicated that both solutions had essentially the same concentrations of Fe® and Fe?", 50 and 27 ug/ml respectively. Further investigation will be necessary to determine the correct method for preparing and/or analyzing iron oxalate solutions. Experiments with the quinone-hydroquinone system. A possible alternate to the Fe* -Fe?* system for meas- uring electrolyte-phase mass transfer coefficients is the reversible reduction of quinone to hydroquinone at the mercury cathode. The reaction under consideration is C6H4O2 +2H++2€—2C6H4(0H)2 . (1) Since hydrogen ion as well as quinone is a reacting material, a strong buffer must be present to serve as a supporting electrolyte. The buffer causes the H™ con- centration to be essentially constant across the inter- facial electrolyte film, because the rate at which the buffer equilibrium is established is relatively rapid com- pared with the quinone diffusion rate.!” A qualitative test was made with the quinone system to determine whether acceptable polarization waves could be measured and to determine whether the qui- none electrolyte is inert to mercury. The electrolyte was 0.01 M hydroquinone and 0.005 M quinone with a 0.05 M phosphate buffer at a pH of 7.0. Satisfactory polari- 16. [. M. Kolthoff and 1. J. Lingane, p. 484 in Polorography, Interscience, New York, 1946. 17. C. A. Lin et al., “Diffusion-Controlled Electrode Reac- tions,” Ind. Eng. Chem. 43, 2136—43 (1951). 152 zation waves were obtained in a small cell with a large copper anode and a mercury pool cathode. The electro- lyte was chemically inert to mercury during the tests. The color of the quinone electrolyte changed from a light yellow to deep brown within several hours. This phenomenon is due to the decomposition of quinone by ultraviolet light. Further studies in the 5 X 7 in. contac- tor will be done to determine whether this system is suitable for mass transfer measurements. 8.3 CONTINUOUS-FLUORINATOR DEVELOPMENT R. B. Lindauer Continuous fluorinators are used at two points in the reference flowsheet for MSBR processing. The first of ‘these is the primary fluorinator, where 99% of the uranium is removed from the fuel salt prior to the re- moval of *23Pa by reductive extraction. The second point is where uranium produced by decay of 232 Pais removed from the secondary fluoride salt in the protac- tinium decay tank circuit. These fluorinators will be protected from fluorine corrosion by frozen-salt layers formed on the internal surfaces of the fluorinator which are exposed to both fluorine and molten salt. To keep frozen material on the walls while maintaining a molten-salt core in the fluorinator, an internal heat source is necessary to support the temperature gradient, Heat from decay of the fission products in the salt will be used in the processing plant. However, to test frozen-wall fluorinators in nonradioactive systems, another internal heat source which is not attacked by fluorine is needed. Since electrolytic or autoresistance heating of molten salt has proven to be a feasible means for providing this heat source, studies of autoresistance heating of molten salts are continuing. A conceptual design was made for a continuous fluorinator experi- mental facility (CFEF) to demonstrate fluorination in a vessel protected by a frozen-salt film. Design was com- pleted and installation was begun of a fluorine disposal system in Building 7503 which uses a vertical spray tower and a recirculating KOH solution. Installation was completed of equipment to demonstrate the effective- ness of a frozen-salt film as protection against fluorine corrosion in a molten salt system. 8.3.1 Installation and Initial Operation of Autoresistance Heating Test AHT4 Equipment for autoresistance heating test AHT4 was installed in cell 3 of Building 4505. In this system (Fig. 8.4) molten LiF-BeF,-ThF, (72-16-12 mole %) is cir- culated by means of an argon gas lift from a surge tank to a gas-liquid separator from which the salt flows by gravity through the autoresistance electrode, through the test vessel, and returns from the bottom of the test vessel to the surge tank. The test vessel (Fig. 8.5) used in experiment AHT-3 was decontaminated, equipped with new cooling coils, heaters, and thermocouples, and reinstalled for experiment AHT 4. The test vessel is made of 6-in. sched-40 nickel pipe with a 44-in.-long (1.1-m) cooled section from the elec- trode to below the gas inlet side arm. The cooled sec- tion is divided into five separate zones, each with two ORNL DWG 75-498% GAS-LIQUID T SEPARATOR _— FREEZE VALVE HEAT FLOWMETER » OFF-GAS ARGON AUTORESISTANCE : HEATING POWER SUPPLY TEST VESSEL Fig. 8.4. Flowsheet for autoresistance heating test, AHT-4. 154 parallel coils through which an air-water mixture flows. - The gas outlet section above. the salt level has an in- creased diameter for gas-salt disengagement and is made of 8-in. sched40 pipe. The surge tank has a 46-in.-long (1.2-m), 6-in:-diam (0.15-m) section to provide submer- gence for the gas lift. The upper section of the surge tank is 24 in. (0.61 m) in diameter and provides suffi- cient capacity to contain the salt inventory for the en- tire system. The gasliquid separator is an 8-in.-diam (0.20-m) conical-bottom vessel with baffles and York mesh in the upper part for gas-liquid disengagement. In the heat flowmeter the salt is heated by an internal cartridge heater, and the flow rate is calculated from the ‘heat input and the temperature rise of the salt stream. The system is started up by heating the equipment and lines to 600°C (873°K). The argon gas lift is started, and initially the salt flow rate is determined by the decrease in surge-tank liquid level. After the salt levels in the tank, separator, and test vessel are constant, cooling of the test vessel is started. The resistance be- tween the high-voltage electrode and the test vessel walls is checked periodically by applying a low voltage to the electrode and measuring the current. As cooling progresses, this resistance will increase until the point is reached where heat can be produced in the salt at a significant rate (several hundred watts) without causing a reduction (shorting) of the resistance. The 80-iter salt batch was charged to the surge tank; and after minor modifications to the heating system, operation was started. Four preliminary runs were made, lasting from 4 to 12 hr (from the time the gas lift was started until plugging occurred). In the first run, plugging apparently occurred in the electrode when the liquid level.in the separator fell too low to provide suf- ficient head for flow to the test vessel. Salt flow in the second run was much smoother, and circulation continued for 11 hr without adjustment of the gas lift. During this time the test vessel was being cooled,-and the salt flow rate slowly decreased by 7%, from 450 to 425 cm? /min. This was probably caused by an increase in salt viscosity, a buildup of frozen salt in the test vessel, or a combination of the two. The steady salt flow rate and higher salt temperature (>873°K and 20—30°K higher than in run No. 1) kept the electrode from freezing, but the heat supply at the bottom of the test vessel was insufficient to keep the salt outlet from freezing, which terminated run 2. The resistance be- tween the high-voltage electrode and the vessel wall increased from 0.01 to 0.08 £, but autoresistance heat- ing was not attempted. The vertical portion of the test section had been cooled to 639°K (solidus temperature, 623°K). Before the third run, the output of the powerstat con- trolling the test vessel bottom heaters was increased by 44% to keep the salt outlet above the freezing point. The run was terminated by salt freezing in the elec- trode. This resulted from too low a salt flow rate (the heat flowmeter was inoperative because of a burned out heater) and too low an initial temperature (723°K vs 823°K in the second run) in the vertical section of the side arm through which the electrode passes. The fourth run was started with some heat on the vertical section of the side arm. This section was un- heated previously. As cooling progressed, the bottom heaters on the test vessel were inadequate at the salt flow rate being used. Increasing the salt flow rate pre- ‘vented freezing at the bottom of the test vessel. After 7% hr of operation, the liquid levels in the separator and test vessel started to increase, indicating salt flow prob- lems both at the inlet and exit of the test vessel. Al- though the salt resistance had only increased from 0.01 to 0.03 € and the average test vessel wall temperature (in the cooled zone) was 658°K, autoresistance heating was started. This freed the plug in the electrode, allow- ing salt flow from the separator to the test vessel, and the increased flow raised the test vessel bottom temper- ature, and flow resumed from the test vessel. However, salt flow rates were erratic for the next 2 hr, and 9% hr after the start of the run the test vessel level started to rise, indicating a frozen salt restriction in the vessel. It was decided to try to transfer the molten salt from the test vessel to the surge tank before complete plugging occurred. This was done successfully, and 5.6 liters of salt was transferred to the surge tank. After cooling, radiographs were taken of the test vessel by the Inspec- tion Engineering Department, using a 35-Ci '®2Ir source. In the test section of the test vessel, radiation penetration was insufficient to permit measurement of the film thickness. The bottom of the vessel between the salt outlet and the gas inlet was free of salt as ex- pected, and the radiograph of the top of the vessel showed a 25-mm-thick ring of salt above the normal liquid level. This is salt deposited on the colder pipe wall by the action of the gas bubbling through the salt. Calculations from the volume of salt transferred indi- cated an average film thickness of 45 mm (a 65-mm- diam molten core). The salt resistance at the end of the run was 0.18 £2, and the maximum autoresistance heat- ing used was 450 W. The main problem seems to be the forming of a uni- form salt film. Near the electrode where the hot molten salt enters, cooling is much slower than in the vertical section above the gas inlet. It is probably in the vertical section where the salt film becomes too thick and re- stricts the salt flow. 8.3.2 Designof a Continuous-Fluorinator Experimental Facility (CFEF) The purpose of the CFEF is to measure the perfor- mance of a continuous fluorinator which has frozen- wall corrosion protection in terms of uranium removal. The uranium which is not volatilized, but is oxidized to UF,, will be reduced back to UF, in a hydrogen reduc- tion column. The facility will be used to obtain operat- ing experience and process data, including fluorine utili- zation, reaction rate, and flow-rate effects; and to demonstrate protection against corrosion, using a frozen salt film. : . o The facility will be installed in a cell in Building 7503 to provide beryllium containment. The system will con- tain about 8 ft* (0.23 m?®) of MSBR fuel carrier salt (72-16-12 mole % LiF-BeF,-ThF,) containing 0.35 mole % uranium initially. The salt will be circulated “ through the system at rates up to 50% of MSBR flow’ rate (6.7 X 107> m?>/sec). Because of the short fluori- nator height (1 to 2 m) the amount of uranium volatil- ized will be between 80 and 95% per pass. The variables 155 of salt flow rate, fluorine flow rate, and fluorine con- centration will be studied by measuring the UF, con- centration in the fluorinator off-gas stream and by sam- pling the salt stream after reduction of UFs to UF,. The fluorinator will have two fluorine inlets to provide data for determining the column end effects. Reduction of UFs will be carried out in a gas lift in which hydro- gen will be used as the driving gas and also as the reduc- tant. If additional reduction is required, this can be done in the salt surge tank. The surge tank is designed " to provide sufficient salt inventory for about 10 hr of fluorination with 95% uranium volatilization per pass. About 99% of the uranium should have been removed from the salt batch after this period of time. The facility flowsheet is shown in Fig. 8.6. Salt will enter the fluorinator through the electrode in a side arm out of the fluorine path. The electrode flange will be insulated from the rest of the fluorinator, and the auto- resistance power will be connected to a lug on the flange. The salt will leave at the bottom of the fluori- nator below the fluorine inlet side arm. The fluorinator wall will be cooled by external air-water coils to form the frozen salt film which will serve the dual purpose of preventing nickel corrosion and of providing an electri- cally insulating film for the autoresistance current. Below the fluorine inlet the fluorinator wall will not be cooled, and the molten salt will complete the electrical ORNL-DWG 73-8426 : . TO OFF-GAS TO FLUORINE DISPOSAL SYSTEM : GAs-Liquio| || |- P HF SAMPLER SEPARATOR TRAP HEAT i UFO TRAPS F‘LOWMETER E ] l A I FLUORINE o ] AUTO sunslr; REDUCTION N MN (DRESISTANCE TANK coLvS POWER GAS-LIFT SUPPLY —_ PUMP FREEZE VALVE ‘| FLUORINATOR HYDROGEN = | | FREEZE vALvE | DRAIN [ TANK : i ! Fig. 8.6. Continuous fluorinator experimental facility flowsheet. FLUORINE - CONTAINING GAS 69 FE) (FE 156 ORNL DWG 75-15057 TO HOT OFF-GAS SYSTEM PHOTOMETRIC ANALYZER KOH SURGE TANK Fig. 8.7. Fluorine disposal system. circuit to the vessel wall. Since all of the uranium will not be volatilized from the salt, there will be some UFg in the salt at the bottom of the fluorinator. The fluori- nator bottom, exit line, and reduction column will be protected from the highly corrosive UF; by gold lining or plating. The molten salt containing UF5 will enter the bottom of the column where the salt will be con- tacted with hydrogen. The hydrogen will enter through a palladium tube, which will result in the formation of atomic hydrogen and greatly increase the reduction rate to UF,. The hydrogen reduction column will also act as a gas lift to raise the salt to a gasdiquid separator. The salt will then flow by gravity to the fluorinator through a salt sampler, surge tank, heat flowmeter, and electrical circuit-breaking pot. Off-gas from the separator which contains HF and excess hydrogen will pass through an NaF bed for removal of the HF. Uranium hexafluoride from the fluorinator will also be removed by NaF. Mass flowmeters before and after the NaF beds will be used to continuously measure the UF, flow rate. 8.3.3 Fluorine Disposal System for Building 7503 The CFEF (Sect. 8.3.2) will be the first test of the frozen-wall fluorinator using fluorine. For the disposal of the excess fluorine, a vertical scrubber is being in- stalled in Building 7503. A flow diagram of the system is shown in Fig. 8.7. The scrubber is a 6-in.-diam. 8-ft-high (0.15- by 2.4-m) Monel pipe with three spray nozzles in the upper half of the vessel. The surge tank contains 200 gal (0.95 m*) of an aqueous solution con- taining 15 wt % KOH and 5 wt % KI. This equipment is designed to be able to dispose of one trailer of fluorine (18 std m*) at a flow rate of 1.2 scfm (9 X 107 std m?/sec). The KOH solution will be circulated through the spray nozzles at a total flow rate of 15 gpm (0.001 m?*/sec). The fluorinator off-gas stream will flow cocur- rently with this stream. The scrubber exit stream passes through a photometric analyzer for monitoring the efficiency of the scrubber. 8.3.4 Frozen-Wall Corrosion Protection Demonstration Equipment has been installed for demonstrating that a frozen salt film will protect a nickel vessel against fluo- rine corrosion by preventing the NiF; corrosion prod- uct film from being dissolved in the molten salt. A small vessel containing 6 X 107 m? of molten LiF-BeF,- ThF, (72-16-12 mole %) will be used for the demon- stration (Fig. 8.8). The fluorine inlet consists of three concentric tubes which provide a path for an air coolant ORNL DWG. 75-8949 FLUORINE IN —1 L -~ ARGON OUT L -—— _ARGON IN ————=———— ARGON PURGE [————h FLUOQRINE OUT i SALT LEVEL—» Fig. 8.8. Frozen salt protection demonstration test vessel. stream that will be used for freezing a salt film on the outside of the outer tube. The wall of the inner tube through which the fluorine will flow is 31 mils (0.79 mm) thick. The inner tube of the fluorine inlet will not be protected from corrosion. The vessel wall is also unprotected but is 280 mils (7.11 mm) thick. Fluorine will be passed at a low flow rate (830 mm?/sec) through the salt until failure occurs, which is expected in less than 100 hr at the tip of the probe near the gas-liquid-solid interface. Wall thickness measurements before and after the demonstration will show to what extent the salt film afforded protection. A flow diagram for the system is shown in Fig. 8.9. The argon back pressure will be recorded to provide an indication of corrosive failure. Failure of the tube below the salt film will allow some salt to leak into the argon cooling annulus. The salt will be entrained up into the cool portion of annular space, causing a restriction to 157 the argon flow. The system will be designed such that the fluorine flow is terminated automatically when either a low argon pressure is detected in the annulus or when a high argon back pressure occurs. 8.4 FUEL RECONSTITUTION ENGINEERING DEVELOPMENT R. M. Counce The reference flowsheet for processing the fuel salt from an MSBR is based upon removal of uranium by fluorination to UFg as the first processing step.'® The uranium removed in this step must subsequently be re- turned to the fuel carrier salt before its return to the reactor. The method for recombining the uranium with the fuel carrier salt (reconstituting the fuel salt) consists in absorbing gaseous UFy into a recycled fuel salt stream containing dissolved UF,; according to the reac- tion UFg(g) + UF4(d) = 2UF5(d) . 2) The resultant UF; would be reduced to UF, with hydrogen in a separate vessel according to the reaction UFs(d) + %2H, (g) = UF4(d) + HF(g) . () Engineering studies of the fuel reconstitution step are being started to provide the technology necessary for ‘the design of larger equipment for recombining UF, generated in fluorinators in the processing plant with the processed fuel carrier salt returning to the reactor. During this report period, equipment previously de- scribed'? was fabricated and has been installed in the high-bay area of Building 7503. This report describes instrumentation for off-gas analysis, including a prelimi- nary calibration curve, and two alternatives for pro- viding corrosion-resistant gold linings for equipment to be installed later. The nickel reaction vessels presently installed will be used to test the salt metering devices and gas supply systems. After the initial shakedown work is completed, the UFg absorption vessel, H; reduction column, flowing-stream samplers, and associated transfer lines will be replaced with gold or gold-lined equipment. Gold is being used because of its resistance to corrosion by UF¢ gas and UF; dissolved in the salt. 18. Chem. Technol. Div. Annu. Progr. Rept. Mar. 31, 1972, ORNL-4794,p. 1. 19. R. M. Counce, Engineering Development Studies for Molten-Salt Breeder Reactor Processing No. 19, ORNL-TM- 4863 (July 1975) pp. 38—42. ORNL OWG 75-8948 FCV-3 ' £, ! 100 cc/MIN HASTINGS MASS FLOWMETER 2 .&]_(FI) 5 PSIG | ‘ i V_g _____________________________ ! . o VALVE CLOSES ON 1+ CELL 3B y : HIGH Ar BACK } FIRST FLOOR 8,";2 ! PRESSURE [ g ettt tdied ke 0s150 " | (OUTSIDE 2l IN. WATER | | : - BLDG) w I _——_=Jd ! V-5 /@ a8 ! I 1! w i Qv 0-30 ! 1 - q 1 ' _ - b s} = : PSIG e i V-2 His v-4 g 8 1 I-CFM i ) - I @ @ 1 xe © ® I 0-15 L___! | V-3 4 PSIG | J (s) v-ejg V-7 : /® ARGON \ o I /@ 20 PSIG L AR | ' ] l ./@ > . - © IE—(9 | DELIVERY | | VESSEL | I ACTIVATED ALUMINA TRAP TEST VESSEL Fig. 8.9. Frozen salt protection demonstration flowsheet, 8.4.1 Instrumentation for Analyzing Reaction Vessel Off-Gases The equipment for the second phase of the experi- ment will consist of a feed tank, a UF, absorption ves- sel, an H, reduction column, flowing-stream.samplers, a receiver tank, NaF traps for collecting excess UFg and for disposing of HF, gas supplies for argon, hydrogen, nitrogen, and UF,, and means for analyzing the gas streams from the reaction vessels (Fig. 8.10). The equip- ment will be operated by pressurizing the feed tank ~with argon in order to displace salt from the feed tank to the UF, absorption vessel. From the UF¢ absorption vessel, the salt flows by gravity through a flowing- stream sampler into the H, reduction column. From the H; reduction column the salt {lows by gravity through a flowing-stream sampler to the receiver tank. Absorption of gaseous UF¢ by reaction with dissolved UF, will occur in the UFg absorption vessel, and the resultant UFs will be reduced by hydrogen in the H, reduction column. The effluent salt is collected in the receiver tank for return to the feed tank at the end of the run. The off-gas from the absorption vessel and the reduc- tion column will be analyzed for UF4 and for HF re- spectively. The respective off-gas streams will be continuously analyzed with the use of the Gow-Mac gas density balance. A sample stream is taken from the main off-gas stream and passed through the balance for analysis (Fig. 8.11). These analyses will be used in determining the efficiencies of UF4 absorption and H, utilization. The efficiency of UF¢ absorption will be determined by metering UF¢ and Ar to the UF, reaction vessel and determining the UF¢ content in the vessel off-gas, using a model 11-373 Gow-Mac gas density cell.2® The H, utilization will be determined similarly. Hydrogen will be metered to the H, reduction column, and the column off-gas will be analyzed for H, content, also using a model 11-373 Gow-Mac gas density cell. The Gow-Mac cell, commonly used as a gas chromatograph 20. Gow-Mac Instruments Company, 100 Kings Road, Madison, New Jersey. . : 159 ORNL DWG 74- {1{666R3 ] | — UFg HF ANALYS!S| IANALYSIS HF UFg TRAP TRAP 1 — UFg —@—— | SAMPLE =— 55— PORT ; \e—————p BUILDING = OFF - GAS # SYSTEM SAMPLE PORT UFg ABSORPTION VESSEL H2 REDUCTION COLUMN g —— v ME ST wTw- FEED R RECEIVER Hp Fig. 8.10. Flow diagram of equipment used in the second fuel reconstitution engineer- ing experiment. ORNL DWG 75-8309 OFF- GAS EAIECARS® B0 ’ v BUBBLER REACTION | VESSE : __GAS SSEL oECaTY VENT . CELL BUBBLER ROTAMETER REFERENCE GAS Fig. 8.11. Schematic diagram of fuel reconstitution -engineering experiment off-gas system. detector, provides a continuous signal which varies directly with the density of the sample gas, allowing continuous analysis of the sample gas stream with accu- racies of 3 to 4%.2! Because the detector elements are not exposed to the sample stream, the gas density cell is useful in analyzing corrosive gas mixtures. Nitrogen and argon will be reference gases for the gas density cells used for analyzing the off-gas from the UF¢ absorption vessel and the H, reduction column respectively. The response of the gas density cell is fairly insensitive to changes in the sample gas flow rate when nitrogen or argon is used as a reference gas.*? To measure varying Ar-UFg and H,-HF ratios with the gas density detectors, it is necessary to control the refer- ‘ence gas flow rate precisely. However, high precision is not required for controlling the sample gas flow rate. The reference gas flow rates are controlled sufficiently by rotameter and separate gas supply systems. A satis- factory means for providing reproducible sample flow rates has been developed. The sample stream is taken from the main off-gas stream (Fig. 8.11) and flows through a capillary tube, the gas density detector, an NaF trap to remove the corrosive constituent (UF, or HF), and a bubbler to provide a constant downstream pressure. The pressure upstream from the capillary is maintained at a higher constant value by means of a similar bubbler in the off-gas line downstream from the NaF trap. The NaF traps provide sufficient volume in the lines so that small pressure fluctuations from bub- bles in the process vessels and in the bubblers are effec- tively damped out. The flow rate is not constant (although it is reproducible), because, as the concentra- tion of the sample gas changes, its viscosity changes, producing changes in sample flow rate under the prevail- ing conditions. These flow rate changes superimposed upon concentration changes in the sample stream to the gas density detector result in a nonlinear response of the gas density detector to changes in concentration. The effects are reproducible, however, and a reproducible calibration can be obtained. Such a calibration was obtained with mixtures of hydrogen and nitrogen (Fig. 8.12). For sample gases containing hydrogen and at refer- ence gas flow rates below a certain critical flow rate, hydrogen will diffuse countercurrently into the refer- ence gas stream to the area of the detector elements. 21. J. T. Walsh and D. M. Rosie, J. Gas Chromatogr. 5(5), 23240 (May 1967). 22, C. L. Guillemin and M. F. Auricourt, J. Gas Chromatogr, 1, 24 -29 (October 1963). 160 ORNL DWG 75-8310 100 | E— | — T T T 1 " GOW MAC GAS DENSITY CELL 90 MODEL I1- 373 A DETECTOR CURRENT 70mA w SENSITMVITY: 32mV Z 80r- DETECTOR TEMR 28°C 7 Sw REFERENCE GAS : ARGON e S 4o REFERENCE GAS FLOW RATE | 66 “Srin ¥ SAMPLE GAS FLOW RATE :APPROX. 33°C/min ] 2y® =Z3 50~ ] x0T s =l 40| - L Wl oo 0 30 — <« (L) 20F _ 1o - | | 1 | i | 1 1 1 0 50 00 %H, ;00 T T Ll L 5’0 Li L] L L) 8 % N Fig. 8.12. Calibration curve of Gow-Mac gas density cell model in fuel reconstitution engineering equipment for H, and N,. Due to the high thermal conductivity of H,, the back diffusion of H, can greatly affect the sensitivity of the gas density cell. However, if sufficiently high reference flow rates are maintained, this problem can be over- come. 8.4.2 Design of the Second Fuel Reconstitution Engineering Experiment The design of equipment for the second fuel reconsti- tution engineering experiment (FREE-2) is continuing. The equipment for FREE-2 will be similar in design to the equipment for experiment FREE-1 except for the addition of an intermediate liquid-phase sample port be- tween the UFg absorption vessel and the H, reduction column (Fig. 8.10). In addition, all vessels and transfer lines exposed to dissolved UF, with the possible excep- tion of the receiver tank, will be gold or gold lined. Gold sheet 0.010 in. (0.25 mm) thick is on hand for the fabricated liner of the UF, absorption vessel. Two alter- natives exist for lining the H, reduction column and the receiver vessel: interior gold plating or a fabricated gold liner, The minimum plating thickness that would probably provide a pinhole-free lining is approximately 0.005 in. (0.13 mm). The minimum thickness for a fabricated gold liner in vessels of this size is approximately 0.010 in. (0.25 mm). Fabricated gold liners are economically competitive with gold plating in the thicknesses men- tioned, because gold sheet is available at ERDA precious-metal account prices, approximately $34.99/troy oz ($1.13/g), and gold in commercial gold- plating solutions is. available only at market prices of about $164/troy oz ($5.27/g) as of June 18, 1975. Some comparisons important in the choice between interior gold plating or fabrication of a gold liner are: 1. the technology involved in fabricating a welded gold vessel is available, while some technology would need to be developed for interior plating of vessels having a- high length/diameter ratio such as the H; reduction column, . the time involved in both approaches is approxi- mately the same, . the plating will be difficult to inspect, and there will be no guarantee of pinhole-free coverage, while dye penetrant examination of welded joints is available for a fabricated liner. . Because it is unclear whether there is sufficient gold in the ERDA precious-metals account for lining the re- ceiver tank liner, gold plating is favored. There is the 161 additional alternative of not lining the receiver tank, . since corrosion of the receiver vessel by UF; in the salt could be tolerated, and corrosion products could be re- moved by hydrogen reduction and filtration between runs. ' : 8.5 CONCEPTUAL DESIGN OF A MO-LTEN-SALT BREEDER REACTOR FUEL PROCESSING ENGINEERING CENTER D. L. Gray* J.R. Hightower, Jr. A conceptual design is being prepared to define the scope, estimated final design and construction costs, method of accomplishment, and schedules for a pro- posed MSBR Fuel Processing Engineering Center (FPEC). The proposed building will provide space for the preparation and purification of fluoride salt mix- tures required by the Molten-Salt Reactor Program, for intermediate- and large-scale engineering experiments associated with the development of components re- quired for the continuous processing capability for an MSBR, and for laboratories, maintenance work areas, and offices for the research and development personnel assigned to the FPEC. *ORNL Engineering Division. The project will consist of a new three-story engineer- ing development center approximately 156 ft (47.5 m) wide by 172 ft (52.4 m) long. The building will have a gross floor area and volume of 54,900 ft> (5100 m?) and 1,218,000 ft® (34,500 m?), respectively, and will be constructed of reinforced concrete, structural steel, concrete block masonry, and insulated metal paneling. The building will be sealed and will be operated at nega- tive pressures of up to 0.3 in. of H, O.(75 Pa) to provide containment of toxic materials. The FPEC will be located in the 7900 area approximately 300 ft (91 m) west-southwest of the High Flux Isotope Reactor. The engineering center will contain: 1. Seven multipurpose laboratories built on a 24 X 24 ft (7.3 X 7.3 m) module, for laboratory-scale experi- ments requiring glove boxes and walk-in hoods; A high-bay area, 84 X 126 ft (25.6 X 38.4 m), equipped with a 10-ton (9000-kg) crane, for large- scale development of processes and equipment for fuel processing at the pilot-plant level; . A facility for preparing and purifying 16,000 kg per year of fluoride salt mixtures needed for the Molten-Salt Reactor Program; . Support facilities, including counting room, process control rooms, change rooms, lunch and conference room, and data processing room,; Fabrication and repair shop, decontammatlon room, and clean storage areas; _ A truck air lock to prevent excessive ingress of out- side air during movement of large equipment items into and out of the high-bay area; Two S-ton (4500-kg) service elevators, one inside the building to service the regulated areas and one out- side to service the clean areas and to move filters to. filter housings on the third floor and roof; General service and building auxiliaries, including special gas distribution systems, liquid and solid waste collection and disposal, and filtered air- handling and off-gas scrubbing facilities. The experimental program planned for the building involves large engineering experiments that use 238U, 232Th, Be, hazardous gases (F,, H,, and HF), molten bismuth, and various fluoride and chloride salts. Ini- tially, radioactivity will be limited to that necessary for low-level beta-gamma tracer experiments. The labora- tory area can later be upgraded, if desired, for use with alpha-emitting materials at levels up to 1 kg of **°Pu. The laboratory area will consist of seven 24 X 24 ft (7.3 X 7.3 m) modular-type laboratories and a general- purpose room. Bench-scale experiments of the type now performed in Buildings 4505, 3592, and 3541 will be carried out in these laboratories. Problems encountered in the large-scale experiments can be studied via small subsystems. Inert-atmosphere glove boxes will provide space for examination of samples removed from both the large and the small experiments. The laboratory area will be maintained at a negative pressure of 0.3 in. of H, 0 (75 Pa). The high-bay area will be the main experimental area where large engineering experiments will be performed. Experiments will involve circulating molten mixtures of LiF-BeF,-ThF,, lithium chloride, and molten Bi-Li alloys. The experiments will also use elemental fluorine, hydrogen fluoride, hydrogen chloride, and hydrogen gases as reactants and will use purified argon for purg- ing. Excess fluorine, hydrogen fluoride, and hydrogen chloride will be neutralized in a caustic scrubber using KOH solutions, and the cleaned and filtered off-gas will be ducted to a building exhaust system. The experi- mental equipment and components will be housed in steel cubicles with floor pans which can contain any salt spill. The cubicles will be maintained at a negative pres- sure with respect to the high-bay ambient. The high-bay area can be supplied with up to 45,000 cfm (21.2 m?/sec) of air. This air can be from recirculated inside 162 air or fresh air from the outside. The high-bay exhaust system will be designed for 30,000 cfm (14.2 m?/sec) at floor level and 50,000 cfm (23.6 m3/sec) at the roof framing level. All exhaust ducts will contain fire barriers upstream from the double HEPA filter banks. The salt preparation and purification area will consist of a 25-ft-wide by 35-ft-long by 14-ft-high (7.6 X 10.7 X 43 m) raw materials storage room, a 22 X 22 X 28-ft-high (6.7 X 6.7 X 8.5 m) room for weighing and blending the salt constituents, and a 40-ft.-wide by 45-ft-long by 28-ft-high (12.2 X 13.7 X 8.5 m) room for melting, H,-HF treating, and filtering the fluoride salt mixtures. This facility should be capable of producing 16,000 kg per year of fluoride salt mixtures, using the batch processing method in use at the facility at Y-12. The estimated cost for the FPEC is $15,000,000, of which $5,200,000 provides for inflation during the three years required for design and construction of the building. The design is essentially complete, and the conceptual design report is scheduled to be issued in September 1975. Authorization for this project will be proposed for FY 1978, Part 5. Salt Production 9. Production of Fluoride Salt Mixtures for MSR Program Research and Development F. L. Daley A salt production facility is operated by the Fluoride Salt Production Group for preparation of salt mixtures required by experimenters in the MSR Program. The group is responsible for blending, purifying, and pack- aging salt of the required compositions. Much of the salt produced is used in studies on Hastel- loy N development in which the concentrations of metal fluorides, particularly nickel, iron, and chromium, are important study parameters. It is thus desirable to use salt in which the concentrations of these metal fluorides are low and also reproducible from one salt batch to the next. Oxides are undesirable salt contami- nants primarily because of the adverse effect of uranium precipitation, and also because of the effect of oxides on corrosion behavior of the salt. Sulfur is another con- taminant present in the raw materials used for preparing salt mixtures. Sulfur is quite destructive to nickel-based alloys at temperatures above 350°C, because a nickel— nickel sulfide eutectic which melts at about 645°C penetrates the grain boundaries and leads to intergranu- lar attack of the metal. The maximum desired levels for these contaminants in the fluoride salt mixtures are iron, 50 ppm; chromium, 25 ppm; nickel, 20 ppm; sul- fur, <5 ppm; oxygen, <30 ppm. Other duties of the group include procurement of raw materials, construc- tion and installation of processing equipment, and re- finement of process operating methods based on results from operation of the production facility. When the facility was reactivated during 1974, initial production was carried out in existing small-scale (8-in .- *Consultant. R. W. Horton* diam) reactors while new large-scale (12-in.-diam) reac- tors were being installed. Experience with both the small and large units is summarized in the remainder of this chapter. 9.1 QUANTITIES OF SALT PRODUCED The 8-in.-diam reactor was used for production from startup of the program in early 1974 through the first three months of 1975. During this period, a total of nine full-scale batches (315 kg total) were processed and made available to investigators. Salt from the nine batches was shipped in a total of 23 containers of ap- propriate sizes. In general, operation of the 8-in.-diam reactor proceeded smoothly, and the resulting salt was of acceptable composition and purity. Production in the 12-in.-diam reactor was started in March 1975. Five production runs, each involving about 150 kg of salt, have been carried out. Of the five salt batches processed, four were suitable for use; most of the salt from these four runs was used for fuel proc- essing experiments. In contrast to the earlier runsin the 8-in.-diam reactor, difficulty has been observed in the 12-in.-diam reactor with corrosion of dip lines in the meltdown vessel and with increasing concentrations of metallic impurities in the product salt. 9.2 OPERATING EXPERIENCE IN 12-in.o~0w0€wmez“wwwéw€>EFFOOH < T o = = C. L. Matthews L. Maya r 103. H. F. McDuffie : 127. G.P. Smith 104. C.J. McHargue - 128. L. Spiewak 105. H. A. McLain 129. J. O. Stiegler 106. B. McNabb 130. R.E. Thoma 107. A.S.Meyer 131. A.J. Thompson 108. R. L.Moore ' 132. L. M. Toth 109. F. H. Neill 133. D. B. Trauger 110. P. Patriarca 134. D.Y. Valentine 111. T. W. Pickel ‘ 135. T.N. Washburn 112. C. B. Pollock 136. A. M. Weinberg 113. F. A. Posey . 137. J. R. Weir 114. H. Postma 138. J. C. White 115—116. H.P. Raaen 139. M. K. Wilkinson 117. D. L. Reed 140. W. R. Winsbro 118. T.K. Roche 141. J. W, Woods 119. M. W. Rosenthal 142. R. G. Wymer 120. H.C. Savage 143. G.T. Yahr 121. C. D. Scott 144. J.P. Young 122. W. D. Shults 145. E. L. Youngblood 123. M. D. Silverman 146—147. Central Research Library 124. M. J. Skinner 148. Document Reference Section 125. A.N. Smith 149—151. Laboratory Records 126. F.J. Smith 152. Laboratory Records, ORNL R.C. EXTERNAL DISTRIBUTION 153. Research and Technical Support Division, ERDA, Oak Ridge Operations Office, Post Office Box E, ~ Oak Ridge, TN 37830 154. Director, Reactor Division, ERDA, Oak Ridge Operations Office, Post Office Box E, Oak Ridge, TN 37830 155—156. Director, Division of Reactor Research and Development, ERDA, Washington, DC 20545 157—258. For distribution as shown in TID-4500 under UC-76, Molten Salt Reactor Technology category (25 copies — NTIS) v U.S. GOVERNMENT PRINTING OFFICE: 1976-748-189/236