(INTNRR © e = oo sesearcn o JNTRA RN DOCUMENT COLLECTION . 3 4456 0548275 0 S X ORNL-3812 UC-80 — Reactor Technology TID-4500 (41st ed.) MOLTEN-SALT REACTOR PROGRAM SEMIANNUAL PROGRESS REPORT FOR PERIOD ENDING FEBRUARY 28, 1965 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 U.S. ATOMIC ENERGY COMMISSION Printed in USA. Price $5.00. Available from the Clearinghouse for Federal Scientific and Technical Information, National Bureau of Standards, U.S. Department of Commerce, Springfield, Virginia LEGAL NOTICE This report was prepared as an account of Government sponsored work. Neither the United States, nor the Commission, nor any person acting on behalf of the Commission: A. Makes ony warranty or representation, expressed or implied, with respect fo the accuracy, this report, or that the use of any information, apporatus, method, or process disclosed in this report may not infringe privately owned right: B. Assumes any liab completeness, or usefulness of the information contained ies with respect fo the use of, or for domages resulting from the use of any information, apparatus, method, or process disclosed in fhis report. As used in the above, “person scting on behalf of the Commission'’ includes any employse or contractor of the Com: sion, o employee of such contractor, fo the extent that such employee or contractor of the Commission, or employee of such contractor prepares, disseminates, or provides access to, any information pursuant fo his employment or contract with the Commission, or his employment with such contractor. ORNL-3812 Contract No. W=7405-eng-26 MOLTEN-SALT REACTOR PROGRAM SEMTANNUATL PROGRESS REPORT For Period Ending February 28, 1965 R. B. Briggs, Program Director JUNE 1965 OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee operated by UNION CARBIDE CORPORATION for the U.S5. ATOMIC ENERGY COMMISSION I 3 445bL 05442750 SUMMARY Part 1, MSRE Operations and Construction, Engineering Analysis, and Component Development 1. MSRE Operations Operation of the plant on a 24-hr, 7-day basis began in September, after the initial operator training and the preoperational check-out of most of the system. The principal activities included leak testing, purging, and heating the salt systems, charging salt, completing startup checklists, and operating with salt circulating in the fuel and coolant loops. By the end of February the coclant lcocop had been full of salt for 1167 hr and the fuel loop for 914 hr, and the shakedown operation was nearing completion. About 90% of the precritical test program was accomplished., Operation disclosed the need for some modifications, the most 1m- portant of which are to the radiator doors, the freeze valve air supplies and controls, the thermal shield water piping, and several cooling air control valves. Generally, the performance of systems and components was very good. The problems which were encountered caused little delay in the testing, and none threaten the success of the MSRE. Among the more important experiments conducted during the pre- nuclear operation were the determination of entrained gas volume in the fuel loop and the measurement of time constants associated with removal of a noble gas from the salt and graphite. Design, procurement, fabrication, installation, and check-out of the MBRE instrumentation and controls systems are now essentially complete. All systems necessary for operation of the reactor during criticality and low-power experiments have been completed as originally designed and require only minor revisions and modifications to improve performance or to conform to recent changes in system design criteria. Except for a small amount of instrumentation on the vapor-condensing system, some additional safety instrumentation required to protect the contalnment system from excessively low pressures, and possibly some revision of the radiator door control system, the design of all systems required for high-power operation is complete. Installation and pre- liminary check-out of equipment and circuits for these systems is com- plete in all areas where the design 1is complete. Final check-out is proceeding as the systems become operational. Vendor fabrication of the data logger-computer is complete, and preliminary check-out and company acceptance tests are in progress at the vendor's plant. Design of signal interconnection and power wiring for the data logger is complete, and wiring installation is in progress. Installation and check-out of the data logger-computer is scheduled for April 1965. iil iv Three radiation-resistant closed-circuit television systems were specified and procured for use in remote maintenance of the MSRE. A review of instrument power system loads indicates that the present 25-kva generator will be grossly overloaded when all loads presently assigned to the reliable power bus are on line. Installation of additional capacity is planned. Components for the fuel sampling and enriching system and the coolant salt sampling system were completed. Fabrication of tanks for the wvapor- condensing system was started, and the water tank was completed. The gas tank was nearly completed. The fuel and coolant pumps were installed in the piping systems. After undergoing modification to eliminate tube vibration, the heat exchanger was installed. The installation of all component and pipe heaters, thermal insula- tion, and heater power circuits was completed, and the systems were checked. The high-bay containment was completed. All components, heaters, thermal insulation, and electrical circuits were installed in the drain tank cell. The charcoal beds and the vent house piping installations were completed. The control rod drives were installed and checked. The coolant sampling system was installed and the fuel sampling and enriching system installation was nearly completed. Installation of the vapor-condensing system was started, Excavation work was completed, and the water tank was delivered to the site. 2. Component Development Prototypes of the removable heater for 5-in. pipe and the drain tank heater completed over 8000 hr of satisfactory test operation. The drain tank cooler test was shut down due to failure of one of the 1/2-in. water tubes after a total of 2551 thermal cycles from 1200 to 200°F., The tube had remained intact through 1632 cycles, and this life is believed adequate for service in the MSRE. Testing is continu- ing to determine the life of other parts of the cooler assembly. Thermal cycling of a prototype freeze valve was started to supple- ment a previous test in which a valve had been subjected to over 200 complete freeze-thaw cycles without a detectable change. The five freeze flanges in the 5-in. pipe in the MSRE were success- fully assembled. The temperature distributions on those flanges were essentially the same as that found in the test of the freeze-flange prototype. A method was devised and demonstrated for repairing the seal- ing surfaces of the cone seal disconnect used in the leak detector lines to the freeze flanges. The freeze valves in the drain tank cell, coolant cell, and reactor cell were tested as part of the prenuclear operation of the reactor. It was found to be necessary to increase the cooling air flow to all but the reactor drain valve, and revisions to the freeze valves in the coolant system are being made to provide for an automatic drain on a power failure. Fabrication and run-in of the control rods were completed, and the rods were shipped to the reactor. The materials of construction in the highest temperature zones were changed from stainless steel to Inconel and INOR-8 to improve the oxidation resistance. Tests were started to evaluate the effects of the changes. The prototype control rod drive completed 124,400 cycles of 102 in. travel per cycle in 150°F ambient temperature. Gears of several different materials were tested and a fully hardened stainless steel ASTM 4276 type 440C was found acceptable for use in the rod drives. The MSRE control rod drives were received from the manufacturer. A1l the units were accepted with a variance in the finish specified for the worm and worm gear, and these gears will be replaced before nuclear operation of the reactor. The units were run in at the test stand and then installed at the reactor. The prototype unit was reworked to make it acceptable as a spare for use at the reactor. A study was made of several different makes of pressure regulators to determine the relative susceptibility of diffusive inleakage of moisture through the regulator diaphragm. The results indicate that the regulator presently in the system permitted an inleakage which resulted in 1 ppm of moisture in the helium stream. Another regulator was chosen and will be installed for evaluation in the system. Temporary samplers were designed and installed on the fuel drain tank and fuel pump bowl for use during the prepower operations. The coolant system sampler installation was completed and is being operated routinely by the Reactor Operations Group. The Engineering Test Loop was shut down after 15,400 hr of trouble- free operation. A test which used a cold zone in the fuel pump bowl indicated « high accumulative rate of zirconium oxide on the cold zone. This cold trapping effect may be useful in the control of similar oxides in large systems. The program to demonstrate remote maintenance tools and techniques was continued in conjunction with the installation of the reactor com- ponents. The operations necessary to disengage the large components were tried and cataloged. Practice with the portable maintenance shield was obtained through use during operations of the freeze flanges and handling of the pump bowl and motor and other small components. The tooling for operating the freeze flanges was revised. The graphite sample assembly and the control rod drives were installed using remote means. Design and fabrication of several small tools and viewing devices were completed., vi Assistance was provided in the design, fabrication, and testing of an ultrasonic molten-salt level probe being developed on an AEC contract. Design and fabrication were completed, and testing is in progress. Test results are encouraging. A similar ultrasonic probe will be installed in the MSRE fuel storage tank. Design and fabrication of this probe are in progress. Testing of the prototype float-type molten-salt level transmitters has been terminated. The two systems installed on the level test facility operated satisfactorily for 29 months. Examination of one transmitter after the test was terminated showed little damage or deterioration. The other transmitter was left in service and is being used in other test operations on the level test facility. Design, development, and testing of a high-temperature transformer for use with a float-type molten-salt level transmitter on the Mark IT fuel circulating pump have been completed. Installation of four conductivity-type single-point level indicator probes in MSRE drain tanks was completed. Three of these probes have operated satisfactorily since installation. A fourth probe failed in operation because of oxidation and embrittlement of a copper-clad, mineral-insulated copper-wire excitation cable. This cable is being replaced with cables designed for high-temperature operation. Except for some minor troubles with purge flow control and an un- expectedly high purge line pressure drop, the bubbler-type molten-salt level indicators installed in the MSRE performed satisfactorily during startup and precritical operation of the MSRE systems. Drift testing of thermocouples fabricated from MSRE stock is continuing. Performance of MSRE prototype thermocouples installed on the engineering and prototype pump test loops continues to be satisfactory. Routine observation and logging of data on these couples have been discontinued., Radiation damage testing of a typical extension cable, disconnect, and thermocouple assembly was terminated after eight months exposure to a ®9Co gamma source. Gas was generated to the end of the test, but the resistivity of the insulation remained high. A ceramic—vitreous-enamel material shows promise for use as an end sealant on mineral-insulated copper wires sheathed in a stainless steel tube. Tests were performed to determine the effect of mismatch between thermocouple and extension lead-wire materisls on the accuracy of a differential temperature measurement. The effects were found to be serious enough to require careful design to minimize junction effects and careful matching of materials to obtain the desired accuracy at the MSRE. vii Installation of the MSRE temperature scanner systems was completed, and the systems were used during initial heat-up of MSRE piping and components and during subsequent operations. Difficulties were experi- enced with electrical noise pickup, calibration drift, and signal identifi- cation. These difficulties have been corrected, and the scanner systems are performing satisfactorily. The life of the mercury switches used to scan the signals has been much longer than was expected. A stable, adjustable millivolt reference supply equipped with auto- matic cold-junction compensation was developed. Some difficulty was experienced in obtaining reliable operation of the single-point temperature alarm switches used in the MSRE. Modifica- tions made on the switch modules offer promise of correcting the trouble. Four resistance thermometers were operated at 1350°F for periods up to 1850 hr. Three of the four thermometers failed before the conclusion of the tests. Calibration drift was experienced in one of the two NaK-filled dif- ferential pressure transmitters installed at the MSRE. A spare trans- mitter appears stable, so the trouble 1s believed to be a result of faulty fabrication. Four helium control valves failed in service at the MSRE due to galling between the close-fitting 17-4 PH plug and Stellite No. 6 seat. Replacement trim fabricated for use in repair of these valves also failed in the same manner. Other trim material combinations are being investi- gated. A motion-multiplylng device was developed to obtain a 1l-in. stroke from a valve actuator with a 1/2-in. stroke. Assistance was given during installation and initial hot operation of the fuel and coolant salt pumps in the MSRE. Two spare rotary assem- blies for the reactor pumps in the MSRE were assembled and subjected to shakedown tests. The spare for the coolant pump was prepared for delivery to the MSRE. The spare for the fuel pump was refurbished after a rubbing incident in which the axial running clearance between impeller and volute was lost during cooling tests of the upper pump tank shell. A new design of radiation densitometer for measuring the concentration of undissolved helium in circulating salt was fabricated and installed on the prototype pump test facility. Failure of the electrical insulation in the pump motors installed in the MSRE lubrication systems was traced to the in- trusion of moisture; moisture-resistant coatings were applied to four pump motors. Delivery of the last of four drive motors for the fuel and coolant salt pumps was accepted. The water mockup tests for the MK-2 fuel tank, as well as the initiation of tests with the pump having a molten-salt bearing and the PKP molten-salt pump, were delayed by the emphasis on delivering pumps, lubrication systems, and spare equipment to the MSRE. viii 3. MSRE Reactor Analysis An analysis of the stability of the MSRE was completed, The study included latest values of the system parameters and the effects of un- certainties in these parameters and in the theoretical dynamics model. The system was found to be inherently stable, not only at the design point but for any combination of parameters within the predicted range of un- certainty. The effectiveness of borosilicate Raschig rings in suppressing criti- cality at the bottom of the reactor cell in event of rupture of the MSRE primary circulating system was evaluated. Use of commercially available rings containing 4% by weight natural boron should ensure that a consider- able margin of subcriticality is maintained for any mixture of fuel salt and water that might be dumped into the bottom of the cell. The use of an unmoderated radial blanket of molten salt for improving the breeding capability of a single-fluid, graphite-moderated molten-salt breeder reactor was analyzed., The reactor considered was a 2500-Mw (ther- mal) system with an average power density of 400 w per cm® of core salt. The optimum carbon-to-233U ratio, which maximizes the production of ex- cess neutrons available for absorption in thorium, was found to be in the range of 2500 to 4500, To a close approximation, the breeding potential of the core is insensitive to the C/?33U ratio in this range. Some gain in reactor breeding ratio was obtained by use of unmoderated fuel salt blankets of thicknesses between 1.0 and 1.5 ft, but for fuel salt thick- nesses greater than 1.5 ft the gain was very small. When fuel inventories were taken into consideration, even for blanket thicknesses less than 1.5 ft the gain in breeding ratio was not sufficient to compensate for the cost of the required additional uranium inventory in the radial blanket. Part 2. Materials Studies 4. Metallurgy INOR-8 was found to be compatible with a nitrogen atmosphere con- taining 0.03 to 5.6% O, at 1300 and 1400°F. Reaction rate curves show an increase in reaction with increased oxXygen content. The maximum attack measured was equivalent to an oxidation depth of 0.05 mil in 700 hr. Alterations on the MSRE heat-exchanger tube bundle were successfully completed in which four tubes were removed and the stub ends were plugged. Welding conditions are reported. Creep-rupture and elevated-temperature tensile properties of INOR-8 weld metal were found to compare favorably with the properties of wrought INOR-8, and stress relieving in an argon or hydrogen atmosphere appeared to result in improved mechanical properties of weld metal. ix The morphology of INOR-8 weld metal was studied, and a phase asso- ciated with weld cracking was found to contain more aluminum and silicon than exists in the INOR-8 composition. Brazing studies were begun on combinations of materials expected to be useful for making graphite-to-INOR-8 joints., It was observed that in metal-to-metal combinations with relatively wide differences in coefficient of thermal expansion, ductile braze metal 1s required for crack-free joints. In graphite-to-metal joints, the limiting factor for making sound joints is the difference in coefficient of expansion. Palladium- nickel braze alloys are being investigated for metal-to-graphite Jjoining. Oxygen contamination was found to meet specifications in the graphite core bar and lattice bar specimens. The oxygen concentration did not vary appreciably with the size of specimen or the section of bar from which it was obtained. Accessible void measurements using two wetting agents, xylene and liquid sulfur, indicated that penetration is limited to 1/4 in. of the outer surfaces in CGB graphite. A creep-test experiment was designed and built to study in-reactor creep of INOR-8 as part of an expanded program to study the effects of irradiation on the elevated-temperature properties of INOR-8. Surveil- lance specimens were fabricated for insertion in the MSRE core and for use in the control test rig that was designed to simulate the MSRE temperature profile and major temperature fluctuation. 5., 1In-Pile Tests of MSR Materials A series of in-pile tests of the compatibility of Molten-Salt Reactor materials has been completed. Farlier tests in the series furnished evi- dence, such as CF, and F, in the gas phase, that raised searching ques- tions about the stability of the fuel under irradiation. Favorable an- swers to these questions have been confirmed by the most recent test. The key factor was the use of heaters to maintain the temperature of the fuel during periods when the pile was inoperative. Under these cilrcumstances there was no evidence of F, release from the fuel, and virtually none of the untoward effects encountered earlier were manifested. To a consider- able extent, this relieved doubts about whether the crystal demage, and consequent release of F,, at room temperature could account for all the previously observed behavior of an unfavorable nature. Off-gas from in-pile capsules was analyzed for CF,, but none could be detected. The maximum sensitivity of the measurements was such that CF, would have been detected if its rate of production was 0.1% of that of xenon. This is lower by a factor of 1000 than the rate at which CF,; pro- duction in the MSRE would be of practical significance. The amount of uranium deposited from the fuel on graphite proved to be negligibly small, again in contrast to the behavior in earlier tests carried out without heaters. No evidence of radiation-induced incompatibility could be found. Fission product iodine and tellurium were partially removed from capsules that were swept with helium during the in-pile exposure. 6. Chemistry Equilibrium phase behavior was examined in systems of relevance to molten-salt reactor technology. A three-dimensional model of the LiF- BeF,-ZrF, phase diagram was constructed to afford a simple graphic dis- play of the crystallization behavior of the MSRE fuel and coolant salt. Reexamination was made of the LiF-BeF, system using very pure mixtures of LiF and BeF,. Significant refinement in the liquidus wvalues was achleved. The phase diagram of the system UF;3-UF, was constructed as a part of a study of UF3 = UF, high-temperature equilibria. The system was found to be characterized by a substantial solid solubility of UF, in UF3. Fractionation experiments were conducted with the MSRE four- component fuel mixture, LiF-BeF,-72rF,-UF,, at cooling rates approximating those in the reactor drain tanks; little compositional variation was ob- served. Zone melting experiments revealed that the rare-earth trifluo- rides CeF3, GdF;, and LuF; were usefully removed from an ingot of LiF in from 1 to 12 passes of the molten zone, Tests of MSRE fuel doped with rare earths failed to show an effective separation of rare earths. Transpiration measurements of the reactions between H, O-HF mixtures and molten-salt mixtures have been extended to mixtures of LiF-BeF,-ZrF, in order to learn more about the behavior of oxides as contaminants in molten-salt fuel systems. The results permitted the calculation of sparg- ing efficiency in the removal of oxide from melts as part of the produc- tion process; calculated values were in reasonable agreement with the pro- duction data., The transpiration results were also used to calculate the oxide tolerance of the MSRE fuel and coolant salts; at 600°C the oxide tolerance of the flush salt is indicated to be 0.011 mole/kg, much lower than a previous estimate of 0.06 mole/kg, but the tolerance of the fuel salt is now estimated as being considerably higher than previously esti- mated, perhaps as high as 0.045 mole/kg. The potential advantages of the use of HF-Hy, mixtures for on-stream or side-stream sparging of molten-salt reactor fuels are being explored. They include continuous removal of oxide, control of the oxidation state of the fuel to compensate for the oxidizing nature of the fission process, control of the corrosion of a nickel-based container alloy, and the pos- sibility of removing continuously the 1351, which is the 6.7-hr principal precursor of 135Xe, the primary neutron absorber produced in fission. Laboratory experiments have shown effective removal of iodine from LiF- BeF, melts by the use of HF-H, gas mixtures at 480°C. Preliminary calcu- lations indicate that only a modest side stream from the reactor {(a very small fraction of the total flow through the system) would have %o be stripped of its iodine in order to provide an attractive improvement in neutron economy. xi The stability of UF3, both as a solid in the presence of solid UF, and as a dissolved component of molten fluoride mixtures containing LiF and BeF,, has been studied through measurements of the equilibrium pres- sures of HF and H, associated with the equation H, + UF, = UF3 + HF . Equilibrium quotients were obtained, and thermodynamic values were derived. These indicate that the disproportionation of U(IV) toc U(O) + U(III) in molten~-salt reactor fuels has a much smaller tendency than was previously predicted; melts containing 0.5 mole % each of UF3 and UF, at 1000°K are indicated to be in equilibrium with uranium metal at the very low activity of 1.5 x 1077, The activity coefficients for UF, in MSRE fuel, as esti- mated from the UF; stability studies, were found toc be in good agreement with those derived from other chemical studies. Further study of the viscosity of LiF-BeF; mixtures over the tempera- ture range 376-1112°C and the composition range from 36 to 100 mole % BeF, has yielded values of A and B for the equation log n (cp) = A/T (°K) — B which vary smoothly with composition. The activation energy for viscous flow decreases sharply from 58.5 kcal/mole for pure BeF, to 9.5 kcal/mole for 36 mole % BeF,, while the viscosity at 600°C drops from 63,800,000 to 11.3 centipoises over the same composition range. The production of coolant- and flushing-salt mixtures for the MSRE was completed, and these mixtures were transferred to the reactor tanks for use in prenuclear operation. Approximately 16,000 1b of the binary mixture, “LiF-BeF, (66-34 mole %), was required to make the coolant and flushing salts. The production of three different fluoride mixtures for use in preparing the MSRE fuel was essentially completed. These mixtures were a barren fuel solvent, LiF-BeF,-ZrF, (64.7-30.1-5.2 mole %), a de- pleted uranium concentrate, LiF-UF, (73-27 mole %), and an enriched ura- nium concentrate of the same chemical composition. Some 10,000 1b of barren fuel solvent and 600 1b of depleted fuel concentrate are being made, and some 350 1b of enriched fuel concentrate, containing 90 kg of highly enriched 235U, has been made (in six batches, each containing 15 kg of enriched uranium)., The enriched fuel concentrate is to be sub- divided into smaller containers for use in the approach to criticality when the MSRE fuel is finally constituted. Chemical support to the MSRE during prenuclear operations has in- cluded arrangements for and interpretation of chemical analyses of the fluoride mixtures added to the reactor and arrangements for following the changes in chemical composition of the composited flushing and coolant salt during scme 1000 hr of prenuclear cperation. The chemical analysis of as-received flush and coolant salts revealed an Li:Be ratio which was significantly different from that intended; an as-yet-unex- plained systematic bias in the chemical analysis was inferred from these xii results when various other methods of analysis indicated conclusively that the Li:Be ratio was that which was intended. During prenuclear op- eration, the concentrations of dissolved plus suspended oxide, iron, nickel, and chromium were followed by chemical analysis. The nickel re- mained low, the iron fell smoothly, the chromium rose slightly, and the oxide generally decreased. The overall results were not compatible with explanations based on oxidation-reduction reactions in the system but seemed more likely to reflect the slow settling out of small traces of metallic iron and perhaps oxide which had been passed through the 0,0015- in.-diam pores of the sintered nickel filters used in the final transfer of the material to the reactor. The overall results suggest that no measurable corrosion of the container metal occurred during approximately 1000 hr of prenuclear operation. Development and evaluation of equipment for use in Analytical Chem- istry Division hot cells for analyzing MSRE fuel samples were continued for improvement in the design and efficiency of cell operation. The initial training program was completed, with additional training sched- wled after final equipment modification. The equipment was installed and tested in Cells 5 and 6 of the High-Radiation-Level Analytical Lab- oratory. Development studies were continued on methods for determining reduc- ing power and oxides in MSRE fuel. Satisfactory precision limits were established for reducing power under bench-top conditions. Studies of the application of electrochemical methods for possible direct analyses in the MSRE fuel and coolant salts were continued. Eval- uations of new reference-electrode systems and indicator-electrode designs are being made. Preliminary voltammetric measurements indicate that chromium(II) in the MSRE fuel solvent undergoes a reversible reduction to the metal at the pyrolytic graphite electrode. Investigations on the coolant salt are, at present, concerned with a cathodic wave which may be due to the reduction of hydroxide. 7. Fuel Processing The design, procurement, and construction of the MSRE fuel processing system were essentially completed except for the salt sampler and the ura- nium absorption equipment. An electrolytic hygrometer is being tested for in-line monitoring of the removal of oxide from molten salt by treatment with hydrogen and hy- drogen fluoride. Initial results are encouraging, but they indicate that HF will have to be completely removed from the gas that is bypassed to the analyzer. Study of methods for the removal of volatilized chromium fluoride from the off-gas stream during fluorination of molten salt has begun. Some data have been obtained for the sorption of CrF; on NaF pellets at 400°C., xiii Part 1. MSRE OPERATIONS AND CONSTRUCTION, ENGINEERING ANALYSTS, AND COMPONENT DEVELOPMENT l. MSRE OPERATIONS. ............ ® & 8 2 & & 5 0 80 00 e & 8 &0 0 8B O PR B PR PSS PPE & ® 0 B0 Chronological AcCOUnt.cisesesssssasccscansane e Component and System Performance...... cresesecsseseasessesven e Heaters and Insulation..... covssecuns secsssssanenns cecernans Freeze FlangeS.eeessescens sevesessssseaas ceessccas sesesscass 5 5 6 6 7 Freeze ValveS..e... ceesan cerecens chriasesananesas Certesesseaans 8 Reactor Access Nozzle..... P 9 Fuel and Coolant PumpS.cesceccccss Cesenevasenane cecssnasenns 9 Radiatoresce.. ceesessessareans ceeeesiesatssessessasssnsenans 10 Weighing Systems teesceseeresttrresesbeanasnossn e ciscaana 10 Drain Tank Heat Removal Systems............. ............ sese 11 Thermal Shield Cooling Water..ceeeeesse ceessesessensessnaaes 11 Helium and Off-Gas Systems.................................. 11 Operations Analysis ® o % o9 & % & & 0 O S S S OB S B S S s e e BB o 5 2 0 0 0 09 " 0 12 Behavior of Noble Gas in the Fuel SystemMececerieecessccserene 12 Gas in Circflating LOOPS & 5 5 & s 80 8 ® 8 & 8 8 0 80 ® 0 9 0 @ ® S0 0 PSS O N O DS DS 14 Overflow from Fuel PUlDeecsscscssscossesscooossossacacsseas cees 14 Fuel and Coolant Pump Cool-Down RateS.sseess cessesssceseses . 15 Drying Out the Salt SystemS.eeceescssecssscacsann vessessevse i5 Instrumentation and Controls Design and Installation...ceeeeese 17 GENEra8lecescsassccsssssssnsscsssnss csesscessserennas cecssanns 17 Reactor Process InstrumentationNeeeceececces cesesscsseesecsas . 17 Reactor Nuclear Instrumentation..eeciececcese cesssssscsssanans 18 Electrical Control CircultSeecescseccsses teeesecseseansesesnsens 18 Control Panels and CabinetSececeseessssee Ceeesssssnsssassannae 19 Data SyStelMleececssscsccsssssssonseasssrcssersossassnsssssasnsas . 22 Fuel Sampler-Enricher and Chemical Processing System SEAMPD e e seseacsccvssssssansscsnsasesenosnesses cecscarsssss soee 22 Coolant Salt Sampler cecesenesnanas ceesessena ciceene cesacnnse 22 Personnel and Stack MonltorsS.ieeceeseecsss cecasaae cecesascnrens 23 Reliable Instrument Power SystelM...scececeesces cesssensansas 23 Remote-Maintenance Closed-Circuit TelevisioON.ieseescsecescsce 23 MSRE Component FabricatioNesceececssessssoasssssescassoscsssnss 24 Sampling and Enriching SystemS.csseessesessnssesssavsaessenes 24 Vapor-Condensing SysteM.ceseesasscscessesecsccccsscscscscccns 25 MSRE Construction and Installationeissccccccesscscsasscssscsscses 25 Pump Installation.eeseeersscsvacesssecesaccsscscsccnsscse oo 25 Heat Exchanger Modification and Installation..ecesececsecacs 25 Heater and Electrical InstallatioN.sesccscesssescscsssasccee 26 High-Bay Containment...ceeeecee crvassanns cesesesssesessnsssese 26 Drain Tank SysteMesececsceses cessevsennen cesessssesenssensns 26 2. Charcoal BeASeseescocscsssccsasecsassccascasacssanssscnssenss Control ROAd DriveS.ceceseeccscsssssccsccsenscsscnssasnanasns sese Sampling and Enriching SystemS..cecceccocececcssncococcsscosse Vapor-Condensing SysteMseecesccecscacaes cessssssescesccscsnse COI@ONENTDEVEIJOPD{EN‘I'lo-o.-ocno-oo-..oo.- ........ o & &8 & 0 * 0 8 88 0 2 s Life TeStS'...........I.................................l..l..l Pipe Heaters.l.....l.......................'....l...'....... Drain Tank Healerleeeseeeesssosescsccasossasoscsnsssesssososonas Drain Tank COOlereuiscecses cecreessseesecesessenes st st et Freeze Valve.ceseeeeosscsosesoessoscsssascssssssnssssosssasnses Check-0Out and Startup of ComponentS.ceseseecceasse cesesessenssese Freeze FlangeSeeceesescsrsroersscccscsccsoccss ceenue ceesssncse Freeze Valves...... T T Y T T T Y rarareran Control ROASeeecsescccscacesaonosccssoscsnsccssossssscncsanas Prototype Control Rod Drive Test.eesessecsoscossssscacssnnas Control Rod Drive UnitSeecececesossocscssssssscscsasscssccnss Diaphragm 1.eaKage.vessesoessreasssssessssscsssssssscenscscocs S AMD ] e Seeecacsssssssassssssassssssssssssssoccssssssnsasscscses Temporary Drain Tank Sampler.ccsecscescesscscessssssssscsoss Temporary Fuel System Sampler..ccsceeseescscsssosssascsossas Coolant Salt System Sampler.scecesesosocccsseasscscssesscssssoes Engineering Test LOOPiecscsessscscscassssssasssasesssosssosnsnnas Zirconium Oxide COLA TroPesescsscossscosonss teecnans teecvenes Maintenance..eeeecescsoses T T Instrument Development..seeeeseesoecssssssecssssssssnsnscssnans Ultrasonic Single-Point Molten-Salt Level Probee.icecerieescss Float-Type Molten-Salt Level TransmititersSeescsscccssesseosscs Conductivity-Type Single-Point Molten-Salt Level Probe...... MSRE Bubbler-Type Molten-Salt Level Indicatoriesecsesescseses Thermocouple Development and Testingeeeceeecesacsaoss sevesenes Temperature SCaNNercsccscsssscsssssssccsssosnsscnssscsososss Single-Point Temperature Alarm SWitCheS..eseescessossscssses High-Temperature Resistance ThermometerSeeeessecsesscocsanas High-Temperature NaK-Filled Differential Pressure PransSmitleraeeeseesccscsssasscossoseassssnsossssssssssnssnsss Helium Control Valve Trim Replacement...cceceececessessoscss Control Valve Actuator Motion Multiplier.cecescescesscesseas Pump Development .. ceecesssceesccecscscssscessscssscssscsccccncssce MSRE PUmDS.seesessessescesssessssssnsnssssnsasesssssscsssasnse Other Molten-Salt PUlDSeseescssscessscsccsscssrsscssossssscses MSRE REACTORANA-I‘YSIS...........l...................I.. ..... [ ] MSRE Stability AnalySiS.eesseeesscesscssssssssssscsssssssnsnasse Methods Us€d.essesesssessescsscsssssesosnsssrssssssasossosnsss Resuwlts..... e Suppression of Criticality in MSRE Cell in Event of the Maximum Credible Accident.cceesccecsccscscosossansssossransssns Effectiveness of a Radial Molten-Salt Blanket on the Breeding Potential of a One-Fluid Graphite-Moderated Reactor.cececesces Res.‘.fl-ts.................'..l...............................I 4. Part 2., MATERIALS STUDIES mTAHJ[.]RGY...'Q..'...‘.........Il.......'....l..'...'l......'.. Reaction of INOR-8 with Impure NitrogeNiceisseveescsesssssananas Alteration of MSRE Heat Exchanger Tube Bundleé..sccosecssscnsses INOR-8 Welding StudieS.eeceesseesesocasscnaass ceeeee ceeesencense Mechanical Properties of INOR 8 Weld Metal.................. INOR-8 Welding Microstructure Study.. creesesetsassasncanae . Graphite-to-Metal Joining Development... ..... Pesssssasasssanssas Transition JointS.eceeeesscacensss cuscenssseresseas ceeevanes Brazing Alloy Development.seeeescssesassaes D Evaluation of MSRE Graphite.seseesscocsecsccsccccssessncscsansas Oxygen Contamination of the MSRE Core Bars and Lattice BarSeeececssescsssrsscccosscssssssosscscssssssesconcns Accessible Voids Content of MSRE Core Graphitesicesescecesses Metallographic Examination of Bayonet Tube in Drain Tank Cooler TeSTeeeeseesesssasovssssesasssssassassansannnace crasnrnes Mechanical Properties of Irradiated INOR-8..eceevesesssscccssss MSRE Materials Surveillance Testingeescsseccoesas tesessecareease Control Test for Survelllance SpeCimMeN..secsevsvovossscessacss RADIATION CHEMISTRY ¢ evevceccscssssncsnosvanressases essenenseses IntrodUuCtioNecessseesescsssorsssossnsssocssscaanssscosssasnascnss . Experiment MIR-47-5.c0evecsses ceseeescanes Ceeessrsessaaane cesies GraphitC.icessscssccesoenns cesscsnsns tesescscassss cecesaseasen INOR-84vevensns cesesesssssenscaannanrs cessacras cessecsssesaas Fuel Salt.seceese. S cessenssvene Summary of MIR-47-5 Postirradiation ExaminationsS..ceeeceececsseces Experiment MIR-47-6.cceecccnsss cecessases cecesananse ceceacerene ObJeCtiveSeseeessecosssossccnns ceveesenne D Experimental..eceeeces cesssessssvsenss *ssasacsanss cesessanses Results and DiscuUSSiONeseecesne cesssnas cressersasecenesssssas Summary of Experiment MIR-47-6.ccsveectccsccsenscanns ceessenras CHEMISTRY e v eceveceese cesesseases ceesccscensscnacse ¢cesccansanssa High-Temperature Fluoride Phase Equilibrium Studi€S.csesecesse . Fuel System for the Molten-Salt Reactor Experiment.......... Reactions in Molten Salt SystemS.ceeecscssceesscscee cessesresnne . HF-H,0 Equilibrium with Molten Fluorides...... crescense ceeoe Advantages of On-Stream H¥-II, Sparglng of MSR Fuels. ........ The Stability Of UF3ececvecsscaacas cecescesesneans cecescssraans Viscosity in the LiF-BeF, System..... tetssecsestetsetsecesssans Fuel, Coolant, and Flush Salts for the MoRE...cveeceeceensnnees The Production ProcesS.cescssscescss aesesenonn tessceessesses Coolant and Flush Salt MixtureS..seceeecececssee cercscsnsssensue Component Mixtures for the MSRE Fuel.iiieeesressncesnccasane Chemistry of Prenuclear Use of Fuel and Flushing Salts in tHe MORE.veeereossssasesasnecnsssssossssossssssssossassosnssosscs Compositional Analysis of MSRE S@ltSeessscccccsosccssrsasssas Chemical Analyses of Fuel and Circult Salts During Prenuclear Tests of the MSRE..eseeeeseansens cessessssecsssse xvi Development and Evaluation of Equipment for Analyzing Radioactive MSRE Fuel SampleS.ccesssccccsesssscssssscssssssses 155 Sample PreparationN.ieeseescssscscsccccsccesossascrssssssccavanses 155 Sample ANBlySES.ecessssssvsesnsosssassnsscssassssasssssssssasss 100 Development and Evaluation of Methods for the Analysis of the MSRE FU€leeecescascssscssscncsasssssscssssasansasssssssaanss 160 0Xid€eesevsceacseescsennenencnne ccescssassnssnsaccassnsessass 160 Reducing POWeI'esessesesssessscssssonscsonsssnssoscvossscsossss 103 Electrochemical AnalySEeS..sceeeccescsssssssossscassssssceasess 104 mPRO@SSING......................".'l...........'......... 169 MSRE Fuel Processing System StatuS.eeseeevrersesescssscneecsescss 169 WaterMonitorl......Il...............l........l................ 170 Chromium Fluoride TrapPiNgececeesescsssscsersosscerssssnonsnasas L71 INTRODUCTION The Molten-Salt Reactor Program is concerned with research and de- velopment for nuclear reactors that use mobile fuels, which are solutions of fissile and fertile materials in suitable carrier salts. The program is an outgrowth of the ANP efforts to make a molten~salt reactor power plant for aircraft and is extending the technology originated there to the development of reactors for producing low-cost power for civilian uses. The major goal of the program is to develop a thermal breeder reac- tor. Fuel for this type of reactor would be 23 UFy or 235UF4 dissolved in a salt of composition near 2LiF-BelF,. The blanket would be ThF, dis- solved in a carrier of similar composition. The technology being devel- oped for the breeder is applicable to, and could be exploited sooner in, advanced converter reactors or in burners of fissionable uranium and plutonium that also use fluoride fuels. Solutions of UCls and PuClis in mixtures of NaCl and KC1 offer attractive possibilities for mobile fuels for fast breeder reactors. The fast reactors are of interest too but are not a significant part of the program. Our major effort is being applied to the development, construction, and operation of a Molten-Salt Reactor Experiment. The purpose of this Experiment is to test the types of fuels and materials that would be used in the thermal breeder and the converter reactors and to obtain several years of experience with the operation and maintenance of a small molten- salt power reactor. A successful experiment will demonstrate on a small scale the attractive features and the technical feasibility of these sys- tems for large civilian power reactors. The MSRE will operate at 1200°F and atmospheric pressure and will generate 10 Mw of heat. Initially, the fuel will contain 0.9 mole % UF,, 5 mole % ZrF,;, 29.1 mole % BeF,, and 65 mole % LiF, and the uranium will contain about 30% ?3°U. The melting point will be 840°F. 1In later operation, highly enriched uranium will be used in lower concentration, and a fuel containing Th¥, will also be tested. In each case the composition of the solvent can be adjusted to retain about the same liguidus temperature. The fuel will circulate through a reactor vessel and an external pump and heat exchange system. All this equipment is constructed of INOR-8,* a new nickel-molybdenum-chromium alloy with exceptional resist- ance to corrosion by molten fluorides and with high strength at high tem- perature. The reactor core contains an assembly of graphite moderator bars that are in direct contact with the fuel. The graphite is a new material® of high density and small pore size. The fuel salt does not wet the graphite and therefore should not enter the pores, even at pres- sures well above the operating pressure. 1501d commercially as Hastelloy N and Inco No. 806. 2Grade CGB, produced by the Carbon Products Division of Union Carbide Corp. Heat produced in the reactor will be transferred to a coolant fuel in the heat exchanger, and the coolant salt will be pumped through a ra- diator to dissipate the heat to the atmosphere. A small facility is being installed in the MSRE building for occasionally processing the fuel by treatment with gaseous HF and Fj. Design of the MSRE was begun early in the summer of 1960. Orders for special materials were placed in the spring of 1961. Major modifi- cations to Building 7503 at ORNL, in which the reactor is installed, were started in the fall of 1961 and were completed by January 1963. Fabrication of the reactor equipment was begun early in 1962. Some difficulties were experienced in obtaining materials and in making and installing the equipment, but the essential installations were completed so that prenuclear testing could begin in August of 1964. The prenuclear testing was essentially completed without major difficulties at the end of February 1965. The critical experiments are expected to begin late in April. They should be completed early in the summer of 1965 and will be followed by several months of operation at intermediate levels in ralsing the reactor to full power. Because the MSRE is of a new and advanced type, substantial research and development effort 1s provided in support of the design and construc- tion. Included are engineering development and testing of reactor com- ponents and systems, metallurgical development of materials, and studies of the chemistry of the salts and their compatibility with graphite and metals both in and out of pile. Work is also being done on methods for purifying the fuel salts and in preparing purified mixtures for the re- actor and for the research and development studies. This report is one of a series of periodic reports in which we de- scribe briefly the progress of the program. ORNL-3708 is an especially useful report because it gives a thorough review of the design and con- struction and supporting development work for the Molten-Salt Reactor Experiment. It also describes much of the general technology for molten- salt reactor systems. ORNL-~-2474 Period Ending Januvary 31, 1958 ORNL-2626 Period Ending October 31, 1958 ORNL-2684% Period Ending January 31, 1959 ORNL-2723 Period Ending April 30, 1959 ORNL-2799 Period Ending July 31, 1959 ORNL-2890 Period Ending October 31, 1959 ORNL-2973 Periods Ending January 31 and April 30, 1960 ORNL-3014 Period Ending July 31, 1960 ORNL-3122 Period Ending February 28, 1961 ORNL-3215 Period Ending August 31, 1961 ORNL-3282 Period Ending February 28, 1962 ORNL-3369 Period Ending August 31, 1962 ORNL-3419 Period Ending January 31, 1963 ORNL-3529 Period Ending July 31, 1963 ORNL- 3626 Period Ending January 31, 1964 ORNL-3708 Period Ending July 31, 1964 Part 1. MSRE OPERATIONS AND CONSTRUCTION, ENGINEERING ANALYSIS, AND COMPONENT DEVELOPMENT 1. MSRE OPERATIONS Chronological Account From August 1964 through February 1965, the operators were trained, all nonnuclear systems were put into operation, and about 90% of the pre- critical testing program was completed. Figure 1.1 is an outline of the principal activities during this period. Instruction of the operators, which began in July, was continued through August and into September, with emphasis on flowsheets, control circuits, and integrated operation of the plant. As construction was completed on more parts of the plant, the effort spent on checking, calibrating, and testing increased. In late September operations were placed on a 24-hr, 7-day basls. On-the-job training in all nonnuclear operations continued throughout the remalnder of the period. By early October, preoperational tests of components were complete, the auwxiliary systems were in operation, and the salt systems had been closed and proved leak-tight by testing with helium at 40 psig. The next two months were occupled in purging the salt systems of moisture, heating them to 1200°F, and charging salt into the coolant drain tank and a fuel drain tank. Salt of the same composition was used in both systems: 66 LiF—34 BeFs, 5756 1b in the coolant system and 9230 1b in the fuel system. The first operations with salt were transfers among the tanks in the drain tank cell. These served to calibrate the weighing devices, check elevations and volumes, and establish the operating requirements of the freeze valves. Meanwhile, we completed the extensive Startup Check List, which checks all instruments and controls and places all auxiliary sys- tems in service 1n preparation for heating and filling the salt circulat- ing loops. ORNL-DWG 65-4468 AUGUST SEPTEMBER OCTOBER NOVEMSER DECEMBER JANUARY FEBRUARY CHARGE CIRCULATE FLUSH SALT CALIBRATE, START, AND TEST AUX, SYSTEM CHARGE COOLANT SALT FLUSH SALT | | [ CIRCULATE COOLANT SALT INITIAL OPERATOR TRAINING HEAT {C) CELL MEAT (D} CELL TEST TRANSFER (FVs) FILL AND DRAIN TESTS ('-CE)A;(;QEEL HEAT (R) CELL -_— COOL {R) CELL HEAT (R} AND (C) CELL PURGE (C) SYSTEM | STARTUP CHECKLIST (C) = COOLANT PURGE FUEL SYSTEM (D) = DRAIN TANK l (F) =FUEL (R) = REACTOR LEAK-TEST (FYSYSTEM (FV) = FREEZE VALVE Fig. 1.1. Principal Activities in MSRE Operations, August 1964— February 1965. The circulating loops, which had been cooled down in November for heater and insulation improvements and other work in the reactor cell, were heated up in early January and pressure tested at 62 psig and 1200°F. The coolant loop was filled and circulation was started on January 9. On the 12th, circulation of salt commenced in the fuel loop. Operation with salt circulating in both systems continued uneventfully for eight days. Both loops were then drained to test the drain action. The coolant loop was refilled in a careful calibration of volume vs level, and cir- culation was commenced again. The fuel system was also calibrated, and the salt was circulated for several days. Then a week was sSpent in tests and adjustment of the controls on the fuel system freeze valves. On Feb- ruary 3, circulation was resumed in the fuel loop. Salt was circulated in both loops continuously through the end of February, except for two interruptions of a few hours each when the coolant loop unintentionally drained. (Once the cocling air pressure to the coolant drain valves was reduced too low; the other time, the air supply was interrupted due to a partial loss of electric power in the area.,) The extended period of oper- ation was for the purpose of conducting an experiment in which the salt and graphite were first saturated with 85Kr and then were purged to deter- mine stripping and transfer rates. By the end of February, salt had been circulated for a total of 1167 hr in the coolant loop and 914 hr in the fuel loop. Samples of salt were taken before the salt was charged into the circulating systems and at frequent intervals during the periods of circulation. Results of the analyses of the samples and implications concerning the corrosion of the systems are reported on p. 152, this re- port. Component and System Performance This initlal period of operation included the testing of almost every component and system with the exception of the sampler-enricher, the contalnment, and those items which can be tested only by nuclear power operation. The most significant test results and operating ex- periences are discussed in the sections which follow. Inevitably a host of adjustments and some minor modifications were required, and a few problems of more importance were disclosed by the testing. Some of these problems demanded considerable attention and effort from operations and development personnel, and a few, such as the radiator enclosure, will need major modification before power op-= eration. The importance of the problems is put in proper perspective, however, when it is noted that they caused very little delay in the planned test program, and none appear to threaten the completely success- ful operation of the MSRE. Heaters and Insulation All the 54 removable heater-insulation units in the reactor and drain tank cells were examined, and their electrical resistances were measured before they were put into service. Minor modifications were made in about half the boxes to eliminate possible short circuits in bead-insulated power leads. Removal and replacement of the boxes showed that four or five will require slight modifications to permit remote maintenance., One of the hairpin heater units in the reactor vessel furnace re- peatedly grounded when it was fully inserted in its tubes. This was remedied by slightly shortening the heaters. Some of the coolant system heaters installed under permanent in- sulation showed resistances to ground slightly lower than is desirable. These resistances increased to acceptable values when the system was heated. During the initial heatup of the empty salt systems, it was found that the equipment temperatures could be brought up to an acceptably high range (1000 to 1250°F) with a few exceptions. The most important exception was the radiator. This situation is described in the sectlon on the radiator enclosure, p. 10. Temperatures at the freeze flanges and on the coolant lines at the cell penetrations were, as expected, be- low the liquidus temperature of the salt, but only over short sections. Localized low temperatures were also observed at thermocouples under a short heater box between freeze flange 100 and the fuel pump furnace. These couples could not be gotten above 700 to 900°F until heat losses were reduced by closing some of the gaps around the heater. The coolant outlet nozzle on the heat exchanger could not be heated into the desired range without overheating the inlet nozzle, because the two nozzles, both under the same heater box, differed by 330°F. A similar situation existed on the coolant between the radiator and the pump, where there was 400°F difference among temperatures along a section of line under a single heater control. The spread was reduced to about 175°F by replacing part of the permanent insulation to reduce air leakage. When salt filled the system, temperatures on the piping became much more uniform and approached truly isothermal conditions when the salt was circulating. The heater input required to maintain the circulating salt at 1200°F was about half the installed capacity of 900 kw. During operation at high temperature for over 1500 hr, only two heater elements failed. Both of these were in removable heater units on the main fuel circulating lines and had spare elements embedded in the ceramic heater plates. The spare elements were put into service readily by out-of-cell wiring changes. Freeze Flanges The five freeze flanges on the 5-in. salt lines were sealed to meet leakage specifications without undue difficulty. The leakage rates all stayed within tolerance during the initial heatup, the salt fill and cir- culation, and after the salt was drained. Table 1.1 lists leakage rates measured under these conditions. Table 1.1. Freeze Flange Leakage Rates" Leakage Rate (std cm3/sec) Freeze Flange Initial Salt System Hot, Heatup Circulating Drained X 103 X 10~3 X 10—3 100 2.0 0.57 0.7 101 1.3 0.40 - 0.25 102 0.5 0.30 0.30 200 1.0 0.21 0.41 201 0.6 0.22 0.21 aRates of leakage of helium from the ring groove, pressurized to 100 psig. Freeze Valves A number of problems were encountered in the operation of the freeze valves, but none seriously hindered the overall operation of the reactor, and all appear amenable to improvement by modifications which are planned. Freezing the fuel drain valve while salt was being supported in the fuel loop by gas pressure in the drain tank proved to be very easy and took less than 5 min. The freezing times for the other valves were un- desirably long, because of inadequate air flows. This is being remedied by modifying the air piping. These modifications are described in Sect. 2 (Component Development). Thaw times depend on the temperature range in which the salt plug is held by control of the cooling air. The cooling air flow is controlled by the shoulder temperatures, and in the reactor, problems arose because shoulder temperatures were often unequal. In the case of the drain valve, FV-103, the salt 1s drained from the pipe on one side after the valve is frozen, giving rise to a very uneven temperature distribution. This prob- lem will be met by proper settings of the control modules (see Sect. 2). The other valves comnected to the fuel tanks often developed unequal shoul- der temperatures, apparently because the procedure for establishing the frozen plug did not always leave the line full of salt on both sides of the valve. In leak-testing frozen plugs we found that helium in contact with the plug would leak through, but molten salt would not. We also found that salt flowing from one drain tank to the core ex- cessively heated the freeze valve in the line which tees off to the other drain tank. This heating made it difficult to keep the second valve fro- zen. Indeed, during one drain, the frozen valve thawed, allowing salt to run to both tanks simultaneously. This is a nulsance because the salt must be transferred so that all of it is in one tank before we can fill the re- actor again. The alir supply lines to these valves and other similar valves will be modified to reduce pressure drops so enough air can be supplied to keep the valves frozen. As a temporary expedient, higher pressure com- pressed air was used to cool the valves. Tests showed the necessity of modifying the coolant drain valves to ensure rapid freezing. They will also be made to thaw without use of emergency power in the event of loss of normal power. These modifications, t00o, are described in Sect. 2. Reactor Access Nozzle The first attempts to establish frozen salt seals in the annulil around the 10~ and 2-1/2-in.-diam plugs in the reactor access nozzle were unsuccessful because of insufficient air flow through the cooling pas- sages. The flows were increased from about 3 to 15-20 cfm by replacing the air control valves with temporary lines containing hand valves. When the salt was again forced up into the neck by pressurizing the loop, it was found that even with the increased flows the salt was not frozen as low in the neck as expected. The exact situation is not well defined because of the wide spacing of the thermocouples and the uncertainties in calculating salt levels. Superposition of the calculated levels on the approximate temperature profiles does indicate that some salt was probably frozen in each annulus. During subsequent operations the neck temperatures were well below the freezing point of the salt for at least a foot below the access flange, which was below 200°F. This situation is considered adequate. The control valve trim has been modified to give approximately the flow obtained through the temporary lines, Fuel and Coolant Pumps Except for the control of the air flow to the fuel pump cooling shroud and the temperature distribution under the shroud, the operation of both pumps has been completely satisfactory. The design of the cool- ing air system called for control of the flow in the range O to 400 cfm and a normal flow of 200 cfm. Temperatures observed on the MSRE pump indicated that the optimum flow is considerably less than 100 cfm. In this low flow range, the control of flow by the valve proved unsatisfac- tory and a valve with a smaller C, is to be installed. The distribution of air flow in the cooling shroud is unsymmetrical, and this produces an unsymmetrical temperature distribution on the top of the pump bowl. The resulting effects are not believed to be serious enough to require modi- fication of the shroud. 10 Radiator Two major problems were encountered during the initial heatup of the radiator: insufficient insulation and Jjamming of the doors. The doors proved inoperative when the radiator was hot, because of inward bowing and warping of the surfaces facing the hot tubes. The warp- ing was enough to Jam the doors tightly agalnst the enclosure. New doors of improved design will be installed before power operation. Meanwhile the doors were left closed. Excessive temperatures on the top of the enclosure and low tempera- tures at the lower ends of the tubes led to the discovery that insulation had not been installed in some of the closed compartments of the enclosure. Tube temperatures ranged from 400 to 1000°F, and heater adjustments could not adequately narrow the spread, After the proper insulation was in- stalled and gaps in the vicinity of the lower end of the tubes were caulked, satisfactory temperatures were obtained. Weighing Systems The pneumatic weighing systems on thée coolant salt tank and the fuel drain tanks were calibrated by the use of lead weights before the initial heatup. These tests showed quite linear relations, which have been used since 1n measurling amounts of salt transfers. The weighing system on the coolant tank proved to be stable, showing no long-term drifts or effect of external variables. With 5756 1b of salt in the tank, at about 1200°F, the extreme spread of 40 indicated weights, taken over a period of a week, was *22 1b. This is only *0.4% and is quite satisfactory. The indicated weights of the three tanks in the fuel system have ex- hibited rather large, unexplained changes. In some cases there were changes of 200 to 300 1b over a period of several hours. There is some evidence that these were the effects of changes in forces on the sus- pended tanks as temperatures of attached piping and the tank furnaces changed. The mechanism has not been definitely established, however. During periods when the tanks were undisturbed, there were much smaller variations in indicated weight. Over a period of several days, with about 9200 1b of salt in FD-2, the spread in indicated weight was 262 1b, or +0.8%. This is typical for all three of the fuel tanks. The weighing systems are useful in observing transfers, and under controlled conditions can give sufficiently accurate measurements of weight changes. Absolute determinations of inventory over long periods of time, however, must rely on computations involving salt densities and volumes at certain reference points (the circulating loops filled to the pump bowls or the tanks filled to the level probe points) to eliminate the effects of extraneous forces on the indicated weight. 11 Drain Tank Heat Removal Systems The systems for removal of fission product decay heat were tested and proved operable after some minor piping changes to eliminate air locks in the gravity-feed water supply. The performance was measured in a cool- down test on FD-2, where heat was removed at a rate of 140 kw with the salt at 1150°F. This is more than enough to meet the after-heat removal requirements. Thermal Shield Cooling Water The thermal shield removable segments, through which the cooling water flows in parallel with the main parts of the shield, would not fill prop-~ erly at first because of trapped gas. Vents were installed in the con- necting lines to allow complete filling. The flow through the segments was not enough to keep temperatures low, however, so piping changes will be made to supply water directly to the segments, thus ensuring adequate flow. The pressure drop in the line returning water from the thermal shield was greater than expected, and the flow required during nuclear operation could not be attained within the pressure limitations of the thermal shield. (The water from a 49-psig supply is throttled to less than 18 psig before entering the thermal shield, and a rupture disk protects the shield tank. ) The return piping was modified to permit an adequate 50-gpm flow without excegsive pressures. It was found that closing the radilation block valve on the thermal shield water return line caused the rupture disk to burst, because the inlet valve dld not close in time. Changes in control circuits or valve operators will be made to eliminate this situation. Helium and Off-Gas Systems The oxygen content of the helium supplied to the salt systems has consistently been below 1 ppm (either in moisture or as 02). This was the goal of the purification system that consists of dryers, preheaters, and titanium sponge for removing oxygen and moisture. The capability of the purification system was not really tested, however, because the helium delivered from a tank trailer to the system was already below the oxygen limit. Four helium control valves failed in the first month of operation by severe galling of the close-fitting trim (17-4 PH plug and Stellite seat). Other valves in the system with similar trim (including the four replacements) have operated satisfactorily. These failures of trim which has proved satisfactory in other applications are believed to be related to the valves belng extremely clean and used in nonlubricating, dry-helium service. 12 One of the control valves in the helium system, PCV-528, which throt- tles the off-gas to control pressure in the coolant pump, was replaced with a valve having a larger flow coefficient. Other pressure drops in the off- gas line were appreciable, and the original valve was too small to hold the pressure down to the specified 5 psig. The check valves in the off-gas system were designed for very low pressure drops. One such valve was removed after it was observed to re- main open when the pressure drop was reversed. When a bit of debris was found to be the cause, all the other similar valves were removed for in- spection and cleaning. The valves were tested for positive checking action before being re- placed. A forward pressure of at least 5 in. Hy,0 was necessary to open the valve, and the valve reclosed by the time the forward pressure drop was reduced to 2 in. Hy0. The off-gas lines contain filters upstream of the control valves which are designed to stop particles down to 40 4. The filter in the coolant off-gas line was replaced twice because the pressure drop had become so large that the pump bowl pressure could not be controlled at 5 psig. The first cartridge plugged after 24 days of service with salt in the coolant loop and twice as much helium throughput before salt was added. The second cartridge was replaced after 20 days of service with salt in the pump. Inspection showed the filter cartridges to be covered with a black substance which has not yet been identified. Operations Analysis Behavior of Noble Gas in the Fuel System The neutron poisoning by 13%¥e will be affected by the rates of sev- eral transfer processes. These include 1. transfer of xenon from the circulating fluid to the unclad graphite, 2. transfer of xenon from the circulating fluid to the gas phase in the pump bowl, and 3. removal of xenon from the gas space in the pump bowl by the flow of helium into the off-gas system. The purging from the pump bowl depends to some extent on the gas mixing, but can be calculated with reasonable confidence from the helium flow rate and the free volume in the pump bowl. The transfer of xenon from the circulating fluid to the gas phase can be described in terms of a stripping effectiveness. This includes the efficiency with which dissolved xenon is stripped from the circulat- ing liquid and the rate of replacement of circulating helium bubbles with 13 bubbles containing a lower concentration of xenon. The volume fraction of gas circulating in the loop plays an important role in this mechanism, because the noble gases are only slightly soluble in molten fluoride salts. Consequently, most of the circulating inventory of 139%e will be associated with circulating bubbles if as much as 1% by volume of helium is entrained in the circulating fluid. The rate of xenon transfer to the graphite depends on the mass-transfer coefficient across the boundary at the salt-graphite interface and on the diffusion coefficient in the graphite itself. (Xenon-135 that enters the graphite is lost by neutron absorption or radioactive decay, so the graph- ite is, 1In effect, an infinite sink for 135%e.,) Preliminary calculations indicated that resistance to transfer through the fluid film would be the controlling factor. The only information on stripping effectiveness was from some tests on the MSRE prototype pump using water and carbon dioxide. The calculation of the transfer to the graphite is also fraught with uncertainties. It was to help remedy this situation that an experiment on noble gas behavior during the operation with flush salt was conducted in cooperation with the development group. Radiocactive &°Kr (lO-year half-life) was used as the noble gas. This gas was bled, at a rate of 6 curies/day, into the helium stream goling to one of the bubblers in the fuel pump, until the 85Kr concentrations in the pump bowlé the circulating fluid, and the graphite had time to build up. Then the ®°Kr inflow was stopped, and the concentration in the off-gas was monitored as the krypton was purged from the system. The decreasing off-gas activity was represented by a sum of exponentials whose time con- stants are related to the transfer processes described above. (One aif- ference is that the experiments observed transfer to krypton from the graphite to the salt, whereas in power operation the 13°¥e will be trans- ferred in the opposite direction.) At the beginning of the experiment the equipment and techniques were checked by a 7-hr injection followed by 6 hr of stripping. A somewhat longer test, 2-1/2 days of injection and 2-1/2 days of stripping, pro- vided estimates of the time required to bring the graphite to 90% of sat- uration. The third and main part of the test brought the graphite con- centration above this level in ll—l/2 days of injection. The decreasing off-gas concentration was followed for 6 days of stripping. The final part of the experiment consisted of 5-hr injections and 5- hr stripping periods with salt in the pump bowl at three different levels. The object here was to observe the effect of salt level on salt-gas strip- ping effectiveness. Preliminary analysis of the 2-l/2~day injection and stripping experi- ment gave four time constants: 9 min, 1.2 hr, 5.7 hr, and about 75 to 95 hr. The shortest agrees fairly well with the calculated time constant for sweeping the gas space in the pump bowl. The second time constant is prob- ably that for stripping from the flulid. The value indicates that the strip- ping effectiveness was somewhat less than 50%. The longer time constants 14 are within the range of estimates for the graphite-to-salt transfer. One postulated explanation for there being two long time constants is that transfer rates are different in different parts of the core (e.g., in laminar~ and turbulent-flow regions). The longest experiment has not yet been analyzed. The results, when available, will be used to predict the 135%e poisoning and to help analyze the poisoning actually observed during power operation. Gas 1n Circulating Loops When the fuel loop was filled there was no indication of any trapped gas other than that in the reactor access nozzle. After circulation was established, however, there was evidence of a much larger compressible volume in the loop. When the pressure in the pump bowl was reduced or when the pump was stopped, lowering pressures around the loop, the rise in pump bowl salt level reflected the expansion of this gas. Tests showed that when the pump was operated with a lower salt level, the compressible volume was larger. At the normal level (2.5 in. above the volute center line) the compressible volume was equivalent to 0.42 ft2 of gas uniformly distributed around the circulating loop. This is a volume fraction of 0.6%, which is consistent with the experience in the pump development loop, where the Jets from the stripper ring were observed to produce bubbles which got into the circulating strean. The situation was quite different in the coolant loop. Each time this loop was filled with salt, about 0.3 t2 of gas was trapped (prob- ably in the heat exchanger). Upon starting the pump, the trapped gas was swept to the pump bowl, causing a drop in salt level. Thereafter, tests showed no pressure effect on salt level, implying no more gas in the loop. (The coolant pump has no spray ring. Overflow from Fuel Pump Continuous, slow accumulation of salt in the overflow tank beneath the fuel pump was observed when the pump was operated with high salt levels in the pump bowl, but the pump could be started and operated sat- isfactorily with a salt level below that at which transfer occurred. The transfer occurred when the level indicated by the bubbler was well below the lip of the overflow line, ceasing only when the indicated level was more than 2.8 in. below the lip. At 2.2 in. the rate of accumulation in the overflow tank was about 1 gal/day. The mechanism for the transfer is not known. Practically complete recovery of salt from the overflow tank was effected repeatedly by simply pressurizing the overflow tank to push the salt back into the pump bowl. 15 Fuel and Coolant Loop Cool-Down Rates Tests were made to determine the rates at which the fuel and coolant loops cooled down under various abnormal conditions. A simulation of complete loss of electric power to all heaters and pumps, with emergency power to pumps and radiator heaters coming on after a 10-min delay, resulted in a decrease of 75°F in 3 hr. Cutting off all heaters, but leaving the pumps on, gave a decrease of 60°F in 2 hr. When all heaters and the pumps were cut off for 40 min, the temperature of the large piping in the loops decreased about 30°F and the radiator tube tem- peratures decreased about 90°F. These rates give plenty of time for a salt drain or remedial action in case of partial or complete power failure, Drying Out the Salt Systems Before salt was charged into the reactor, the salt systems were carefully dried by evacuating and purging with dry helium. The effective- ness of the procedures is attested by the salt analyses, which showed that the salt met with very little oxidizing impurities. (See Sect. 6.) Figure 1.2 shows the procedure followed in removing the molsture from the fuel circulating system. Throughout the period shown, the in- jection point for the dry helium was the fuel pump. The pump was op- erated to circulate helium from November 2 on, except when the system was at a vacuur. During the operations at room temperature, gas was dis- charged from the circulating loop through the drain tank piping and out at line 110, between the drain tank system and the chemical processing cell. Just before the reactor cell was heated up, the discharge was changed to the normal off-gas line from the pump bowl (Line 518) to avold sending the moisture driven out of the graphite through the drain tank piping. The temperature in Fig. 1.2 is from a thermocouple on the lower head of the reactor vessel, which is believed to reflect the temperature of the circulating helium. The graphite temperature would lag somewhat while the helium temperature was changing. The increase in off-gas moisture content during the heatup of the core showed that a considerable amount of moisture began to come off when the helium temperature reached about 250°F. After the system had been evacuated to help get rid of this moisture, and with the temperature at 500°F, the off-gas moisture subsided to a low level. It rose agaln sharply when the helium temperature was raised above 650°F, and again the system was evacuated. No further releases were seen when the temperature was raised from 800 to 1130°F. An accounting of the total moisture removed was not made because the exit moisture could not be measured during the evacuations, which no doubt eliminated a large part of the total. When the fuel drain tanks were heated up, Jjust after the period shown in Fig. 1.2, helium from the circulating loop again flowed through the drain line and out at line 110, to prevent moisture released from the drain tanks from entering the circulating loop. ‘H20 {ppm) 9T FLOW (liters/min) (°F) TEMPERATURE ORNL - DWG 65-4169 1000 800 OFF-GAS MOISTURE A 600 ! 60 - I A i ! 50 ‘ AN . 400 f \ 40 | e | e READINGS AT LINE 110 A | | 30 L A READINGS AT LINE 510 l / \ '\. 200 0 ? N Tt 20 1% \ \ e / A A ®* ok ANK / ] b T '.“ J N | OUT OF SERVICE R A 0 St @ ® A h#-'l ok e -5 psig EVAC. TO 15-in. Hg (-2 EVAC. AND | 5 O psig ( EVAC. E SILEAK TEST| 8K || _Evac.To | 5 | FFTAND L5 11y 5o6ig w2 1O 555 70 5 psig ] 3 AT S |25 in.Hg | psig | psig | REACTOR [Tsig PSI9 1% in. He 9 725 in | wiz! 25 in Hg o OPENED) Hg 10 R u HELIUM INFLOW 5 IJ | (—t— ) , M, L 1250 1 - 1000 — e e 750 REACTOR VESSEL TEMPERATURE 500 // 250 /// 0 20 22 24 26 28 30 | 3 5 7 9 1 13 15 OCT 1964 NOV 1964 Flg. 1.2. Initial Removal of Moisture from Fuel System. 17 The coolant system was dried out by a procedure similar to that fol- lowed in the fuel system, except that it was not necessary to evacuate during the heatup. Instrumentation and Controls Design and Installation General Design, procurement, fabrication, installation, and check-out of the MSRE Instrumentation and Controls Systems are now essentially complete. All systems necessary for operation of the reactor during criticality and low-power experiments have been completed as originally designed and require only minor revisions and modification to improve performance or to conform to changes in system design criteria. Ixcept for a small amount of instru- mentation on the vapor condensing system, some additional safety instrumen- tation required to protect the contalinment system from excessively low pressures, and possibly some revision of the radiator door control system, the design of all systems required for high-power operation is complete, Installation and preliminary check-out of equipment and circults for these systems are complete in all areas where the design is complete. Final check~out is proceeding as the systems become operational. Reactor Process Instrumentation With the exceptions noted above, the reactor process instrumentation is complete and ready for operation. This equipment was checked out prior to and during precritical operation of the reactor. In general, perform- ance of process instrumentation was very good; however, a number of minor difficulties were encountered. Revisions and modifications to correct these difficulties were or are being made. No difficulties were encoun- tered which required major changes in instrumentation design or control philosophy. The most serious difficulty encountered was a failure of the excita- tion and signal cable leads on the fuel flush tank level probe. This failure was caused by excessive temperature which caused oxidation and embrittlement of the copper-clad, mineral-insulated copper-wire cables. These cables were designed on the assumption that they would be routed in air above the tank insulation and that their operating temperature would not exceed 200°F; however, in the actual installation, the cables were covered with insulation and the temperature at the point of attach- ment to the probe was probably in excess of 800°F. A replacement cable as- sembly, capable of continuous operation at 1200°F, has been designed. The defective cables will be replaced during the precritical shutdown in March. 18 Other, less serious, difficulties encountered include: 60-cycle noise pickup and drift in the thermocouple scanner system, set-point drift in the electrosystems (magnetic amplifier type) temperature switch mod- ules, and failure (due to galling between plug and seat) of low-flow weld- sealed helium control valves (see Sect. 2). The performance of the thermocouple system has been very encouraging. Although there are 1033 thermocouples in the MSRE system, and 858 of these couples are installed on heated pipes and vessels, there were very few thermocouple fallures during the startup and prenuclear operation of the reactor system. (Experience with similar thermocouple installations on the MSRE Engineering and Pump Test Loops indicated that most failures would occur during the initial heat-up and cool-down of the system and that the failure rate would be very low during subsequent 0perations.) Reactor Nuclear Instrumentation Design of the MSRE nuclear instrumentation and control system is complete, and installation of equipment is nearing completion. Fabri- cation, installation, and preliminary check-out of panel-mounted equip- ment, rod drive, and interconnecting signal and control wiring are com- plete. OSome work remains to be done on the installation of chambers and chamber drives; however, this work is expected to be completed by April 1. Electrical Control Circuits The largest single effort during the past year was the completion of the electrical control system. During this period the design of con- trol relay cabinet wiring, jumper-board wiring, console wiring, control- and safety-circuit interconnection wiring, instrument power distribution Interconnection wiring, and annunciator interconnection wiring was com- pleted, and the wiring was installed and checked out. A comprehensive functional check-out of the control, safety, and alarm circuitry was made prior to the start of prenuclear operation of the reactor systems. Al- though some design and wiring errors were found during the check-out, the errors were of a minor nature and were easily corrected. In general, the quality of the installation was excellent. During prenuclear operations the need for some circuit revisions and additional controls became apparent. Some minor changes have been completed, and the design required for circuit revisions and additions needed for critical experiment operations is in progress. All revisions are expected to be complete before the start of power operation. None of the revisions or additions made to date or presently under considera- tion involve major changes in control circult design or philosophy. 19 Control Panels and Cabinets Design and fabrication of all panels required in the MSRE reactor and auxiliary instrument systems is now complete. Design, fabrication, and installation of the main control console, control circuit jumper board, control relay cabinets, and process safety instrument panels were completed prior to the start of prenuclear operations. Installation of nuclear and Health Physics (personnel monitoring) panels is now complete. Fabrication of the fuel system sampler-enricher and chemical processing system panels was completed in February, and installation of these panels is in progress. Table 1.2 lists the panels and cabinets presently in- corporated in the MSRE instrument and control system. Figure 1.3 shows the completed main control room panels and console. Fig. 1.3. MSRE Main Control Room Panels and Console. Table 1.2. MSRE Instrument Panels and Cabinets Length . . . Number Height Depth Density of type Location of Panels per(?:?el (£t) (£t) Components Console Main control 1 ~10 ~3 ~2 High room Main control Main control 10 2 7 2 Medium board room _ Control circuit Main control 1 4 7 2 Medium Jumper board room Auxiliary process Auxiliary control 6 2 7 2 Medium room Thermocouple patch Auxiliary control 1 4 7 2 High panel room Process safety Auxiliary control 2 2 7 3 High room Nuclear control Auwxiliary control 2 2 7 3 Very high and safety room Process radiation Auxiliary control 2 2 7 2 High monitors room Personnel and Auxiliary control 1 2 7 2 Medium stack radiation room monitors Control relay Auxiliary control 1 6 7 2 High cabinet room Safety relay Auxiliary control 1 4 7 2 High cabinet room Auxiliary process Transmitter room 7 2 7 2 High Solenoid rack Transmitter room 1 6 7 2 High Transmitter rack Transmitter room 1 9 g 1 High Lube o0il system Service tunnel 2 1 2 1 Low (Local) Iube oil system Service room 2 2 7 2 Medium (remote) 0¢ Table 1.2 (continued) Length . ) . Number of Height Depth Density of Type Location Panels per(?i?el (ft) (£t) Compongnts Contalnment air Outside building 1 2 7 2 Meddium system south Containment air High-bay area 1 2 7 2 Low system Coolant sampler High-bay area 1 l-l/2 2 l/2 Medium Fuel sampler- High-bay area 3 2 7 2-1/2 High enricher system Chemical process High-bay area 2 2 7 2 Medium system Chemical processing High-bay area 3 2 7 2 High system sampler (future) Cell containment North electric 1 2 7 1 Medium block safety service area systen Thermocouple scanner Basement 2 2 7 2 High Cooling water Water rcom 1 2 7 2 Medium system Cover gas Diesel house 2 2 7 2 Medium system T 22 Data System Design of all signal interconnection and power wiring required for the data logger-computer is now complete, and installation of the wiring is in progress. Vendor fabrication of the data logger-computer is com- plete, and preliminary check-out and company acceptance tests are in progress at the vendor's (Bunker-Ramo ) plant. Delivery of the equip- ment to the MSRE is presently scheduled for about April 1. We plan to have much of the on-site work (intercomnection and power wiring, cabinet bases, access holes, ete.) completed before delivery of the equipment. Installation, final check-out, and acceptance tests are expected to be completed in April. Fuel Sampler-Enricher and Chemical Processing System Sampler A review of the design of the panels used for testing of the prototype sampler-enricher system revealed that extensive wiring changes would be required for the panel wiring to meet the standards of the ORNL recommended practices for safety system wiring. Since a new sampler was needed for the chemical processing system, and since the safety requirements on the chemical processing system are much less stringent than on the reactor fuel system, we decided to build new panels for the fuel sampler-enricher system and to use the existing panels for the chemical processing system sampler. All panel and installation design required for the fuel sampler- enricher is complete. Procurement and panel fabrication are also complete, and installation of this system is in progress. Installation and check- out of the fuel sampler-enricher system is expected to be complete before the start of criticality experiments. Revisions of the existing panel and installation design for the chem- ical processing system sampler will be deferred until other, higher pri- ority, work is completed. Coolant Salt Sampler Design, procurement, fabrication, installation, and check-out of in- strumentation and controls on a coolant salt system sampler were initiated and completed during this report period. Since the requirements for con- tainment on the coolant salt sampler were much less exacting than those of the fuel sampler-enricher system, and since most of the interlocks were mechanical rather than electrical, the instrumentation and controls on the coolant salt sampler are much simpler than those on the fuel sampler-en- richer. The activity of the coolant salt samples should be low; so the instrumentation was designed for direct maintenance and operation. Di- rectly actuated pressure switches and gages were used to obtain the re- quired indications and control signals. 23 Personnel. and Stack Monitors Except for the installation of some equipment and tie-in of lines for retransmission of data to the central ORNL radiation safety and con- trol system, the design, procurement, installation, and check-out of the MSRE personnel radiation monitors (Health Physics) and stack radiation monitors are now complete. Reliable Instrument Power System During the design of the instrument power distribution system an estimate of instrument power requirements was made which indicated that the 25-kva generator would be grossly overloaded if all equipment then assigned to the reliable power system was on line at one time. The needs for reliable power were reviewed, and some equipment (notably the Health Physics instrumentation) was transferred to the TVA-diesel bus. Since the estimate of reliable power requirements was still much more than 25 kva, a decision was made to expand the capacity of the reliable power system 1f measurements made during prenuclear operations verified the estimate. A review of reliable power requirement was completed recently using measured data for equipment on line and revised estimates for equipment not installed. It indicated that a minimum of 40 kva of single-phase, 120-v ac power and 3 kva of three-phase, 208-v ac power is needed if the system is perfectly balanced and is operated at full capacity. The addi- tion of 25 kva of three-phase, 208/120 v ac supply or replacement of the present single-phase, 120/240 v ac supply with a 50-kva three-phase, 208/ 120-v ac supply has been recommended. ©Several methods of obtaining the additional power requirements are being investigated. Final selection will be based on considerations of cost and delivery. Installation of the additional capacity is expected to be completed before the start of power operations. Remote=Maintenance Closed-Circuit Television Specifications were prepared for three radiation-resistant closed- circuit television cameras for use in remote maintenance of the MSRE, and for associated nonbrowning zoom lens, pan and tilt units, monitors, and miscellaneous components. Procurement and acceptance testing of all equipment specified are now complete. Design of the television system installation is in progress. 24 MSRE Component Fabrication Sampling and Enriching Systems Fabrication of equipment for obtaining samples of the circulating coolant salt was completed by the Paducah machine shop. The apparatus, shown in Fig. 2.4, was fabricated of stainless steel in accordance with the ASME Boiler and Pressure Vessel Code for Secondary Nuclear Contain- ment. Fabrication of the fuel salt sampler-enricher was completed by the Paducah machine shop. This dual-purpose machine will enrich and sample the fuel salt during power operations. The machine was fabricated mostly of stainless steel in accordance with the ASME Boiler and Pressure Vessel Code for Primary and Secondary Nuclear Containment Vessels (see Fig. 1.4). PHOTO 69409 Fig. 1.4. Enricher-Sampler Installation at the MSRE. 25 Vapor-Condensing System Fabrication of the water tank for the vapor-condensing system was completed, and fabrication of the gas tank was nearly completed at ORGDP. Both tanks are 10 ft in diameter. The water tank is approximately 23 ft long, and the gas tank is approximately 67 ft long. The tanks will condense gases and contain all the fission products released by the reactor system if an accident should occur that results in fission product release at high pressure in the reactor cell. The tanks are fabricated of 1/2- and 3/8-in.-thick carbon steel plate in ac- cordance with Sect. III, Nuclear Vessels, of the ASME Boiler and Pressure Vessel Code. MSRE Construction and Installation Ixcept for the installation of the vapor-condensing system, MSRE construction work was completed during the period. Preliminary check- out of components and systems uncovered the need for minor modification to some components and systems. These modifications were made prior to starting the preoperation testing program. Pump Installation Following hot shakedown tests in pump test loops, the fuel and coolant pumps were delivered to the MSRE site. The coolant pump was installed in the coolant cell. The fuel pump was assembled with the heat exchanger on the fuel system jig. After fitting was completed and locating points pre- cisely determined, the pump was assembled in the fuel circulating system. The pump rotary element was removed and replaced using remote maintenance techniques, in order to establish that the system can be remotely main- tained after power operations. Heat Exchanger Modification and Installation Water flow tests revealed excessive vibration of the heat exchanger tubes and excessive pressure drop through the shell side of the exchanger. Further investigation suggested that these faults could be corrected by (1) inserting close-fitting spacer bars in two directions between the triangular pitched tubes, (2) installing a baffle at the entrance nozzle to prevent direct impingement of the fuel stream upon the tube bundle, (3) removing four outer tubes, and (4) enlarging the outlet nozzle. These modifications were completed, and the heat exchanger was re- tested and installed in the reactor cell. The heat exchanger installation completed the installation of the fuel circulating loop, permitting the final commection of components and piping and the installation of the freeze flange clamps. 26 Heater and Electrical Installation The component and pipe heaters were installed, including remotely replaceable and permanently installed heaters. All electrical system in- stallation was completed. Thermal insulation was installed as required, and all heating systems were tested. High-Bay Containment The north and south walls of the high-bay containment, including the installation of access doors, were completed. The change room was com- pleted. Drain Tank System Installation of components and piping in the fuel drain tank cell was completed. Heaters, thermal insulation, and electrical circuits were completed. All systems were checked prior to the addition of flush salt. Charcoal Beds The installation of the charcoal beds and associated vent-house piping system was completed. These systems were checked out. Control Rod Drives The three control rod drive units were installed and checked. Sampling and Enriching Systems Installation of the fuel system sampler-enricher was nearly completed. The coolant system sampler was installed, and its performance was checked. Vapor-Condensing System Installation of the vapor-condensing system was started. Excavation was completed and the water tank set in place. Piping work was started. 27 2. COMPONENT DEVELOPMENT The installation of the reactor system of the MSRE was essentially completed in August. Since then, most of the development effort has been devoted to checking out and assisting at the reactor in the startup of various components with which there was previous development experience. I.ife tests on several of the components are being continued to provide advance information for MSRE operation and maintenance planning. The following is a description of the results of work that we have done at the reactor and of other related tests in our development facilities. Life Tests Pipe Heaters The prototype of the removable heater for 5-in. pipe has operated for over 8020 hr with an internal temperature of 1250-1350°F without difficulty. Examination after 4920 hr of operation showed the unit to be in good condition. Drain Tank Heater The prototype of the drain tank heater has operated through 7552 hr at 1200°F without daifficulty. Drain Tank Cooler The drain tank cooler testl’? was shut down after a total of 2551 thermal cycles from 1200 to 200°F. The unit had been examined after 1632 cycles and both 1/2-in. INOR-8 water tubes were intact. However, after 2551 cycles, we found that one of the tubes had cracked into two sections at a point 3 ft 4-3/4 in. above the bottom of the tube. The cracked tube was the one that has its intake 7 in. above the bottom of the steam dome. The fracture was Jjust above the small center spacers which are welded to the 1/2-in. tube. The three spacer pleces had broken off. The 1/2-in, tube in the low-inlet (1 in. above the bottom of the steam dome) test unit was intact. The 1l-in, sched-40 pipes which enclose the l/2-in. tubes were intact but were badly warped. The 1/2-in. tubes and 1l-in. pipes were examined with dye check methods, and no additional cracks were found on either the inside or outside surfaces., Selected specimens were examined metallographically, and the results are reported on p. 80. Since the l/2—in. tube which failed is separated from the salt by two additional tube walls, there is no safety hazard involved, and the net effect of a similar failure in the reactor would be a slight change in the heat removal capacity of the cooler. A life of 1600 cycles is adequate for the MSRE based on an estimate that less than ten such cycles will be required after 28 each prompt drain from full power, and this number is further reduced by holding the fuel in the circulating loop for some period after the reactor power is reduced. These tubes will be replaced and testing will be continued to determine the conditions required to produce significant damage to the 1l-in. pipe. Freeze Valve Thermal cycling of a prototype of the reactor drain valve was started. The test consists of alternately raising and lowering the temperature at the center of the valve at a rate which produces an approximation of the temperature distributions of a normal thaw-freeze cycle., 1In another series of tests, one freeze valve had been subjected to over 200 complete freeze-thaw cycles without a detectable change. We plan to continue cycling the prototype until some change is found, so that operational limitations of the valve can be established. Check-Out and Startup of Components Freeze Flanges We assembled the five freeze flanges in the 5-in. pipe in the MSRE, and measurement of the leak rates at the ring joint seals indicated that the seals were acceptable. After small corrections were made in loca- tions of the heater-insulation units, the temperature distributions across the flanges agreed with that found during the testing of the freeze-flange prototype. Several very small leaks (<10™* std cm® of helium per second) were found in the flange leak detector system, and preparations were made to repair them during the maintenance shutdown. A method was devised and demonstrated for repairing the sealing surface of the cone seal disconnect® used in the leak detector lines. The method consists of using a tapered tool for expanding the nose of the male cone member and then removing the imperfection by lapping. While the method requires direct contact with the disconnect, it is adaptable to remote operation should such be necessary after the reactor system becomes radiocactive. Freeze Valves All the freeze valves in the drain tank cell, coolant cell, and reactor cell were tested as part of the prenuclear operation. We found that the cooling air flow was insufficient for all except the reactor drain valve. Changes are being made to reduce the pressure loss and thereby increase the air flow to the other valves. The freeze valves in the coolant system included an electric heater at the center of each valve to supply the extra energy needed for a fast thaw. Because of the provision for maintenance of this heater, this type 29 of valve was of open construction and the heat losses were large. As a result there was not enough stored energy to thaw the valve by simply turning off the coolant air, and it was necessary to depend on emergency power to the center heater during a failure of normal power. These valves will be reworked and made similar to the other valves, which thaw on total loss of power, Control Rods Fabrication of the operational units was completed. The rods were operated at the test facility through 4000 cycles and 500 full scrams at temperature (1200°F) before shipment to the reactor. Recent calculations of the heating of the rods in the reactor at full power indicate that some in-core parts could possibly be heated to above 1600°F if operated without cooling. In view of this, materials in those parts will be changed from stainless steel to alloys with better high-temperature strength and/or oxidation resistance. The lower 60 in. of hose which is enclosed by the poison elements will be replaced with Inconel hose. The two type 347 SS braided wire retaining cables (1/8 in. in diameter) will be replaced with a single solid 1/8—in.—diam rod of INOR-8., 1In tests designed to demonstrate the integrity of the 1/8—in. rod, the control rod was dropped 2500 times without failure. Flexure testing is continuing. Prototype Control Rod Drive Test The prototype control rod drive was operated through 124,400 cycles (102 in. travel per cycle) in 150°F ambient temperature. During the last period, two types of gears were tested. 1. Tool steel "Stentar" Carpenter steel worm gear, Rockwell hardness R, 55.0, operated against a "Vega" Carpenter steel worm, R, 53, through 29,485 cycles. These gears were regreased at approximately 5000-cycle intervals. At the completion of the test, the weight loss of the worm was 13% and weight loss of the worm gears was 2%. The gears were still in operating condition. 2, Stainless steel ASTM 4276 type 440C fully hardened worm (R, 58) and type 440C worm gear (RC 58.0) were operated through 40,000 cycles without regreasing. At the completion the gears were in good condi- tion. Weight loss of the worm was 6% and weight loss of the worm gear vwas l1.5%. These gears were regreased and are still in use. This gear material was recommended for use in the operational units., The 25-w drive motor was replaced with a 10-w motor to lower the starting load and the running temperature of the front motor bearing. The bearing operating temperature for the 25-w motor was 220°F; the temperature for the 10-w motor bearing was 156°F, The grease in the front bearing (APL, NRRG-159) becomes stiff and lumpy when overheated and can cause bearing failure and subsequent motor failure. After the 25-w motor had 30 operated through 55,000 cycles, examination showed the grease to be tarry and stiff; it appeared that failure was imminent. Testing of the 10-w motor is continuing. During periods between tests, all moving aluminum parts of the prototype were replaced with steel and other modifications were made to prepare this drive as a replacement unit for use in the reactor. Control Rod Drive Units The four control rod drives for the reactor were received from the vendor, The Vard Corporation, Pasadena, Calif. The units were accepted with a variance permitted in surface finish specified for the worm and worm gear, Prior to the functional check-out, these gears were removed from each unit and jig lapped to improve the surface finish. The lapping operation was only partially successful due to the extreme roughness of the worm gear teeth, These gears will be replaced with fully hardened type 440C stainless steel gears prior to the criticality tests. Each unit was then installed on the test stand for functional checking. Each unit was operated for one week with no load, and for three days assembled with a control rod and in a 150°F ambient tempera- ture. Some of the measurements made during the tests are reported in Table 2.1. Table 2.1. Control Rod Drive Data Drive Unit Number Function 1 2 3 4 Clutch gap, in. 0.010 0,010 0.010 0.010 Clutch slippage, with 28 v dc, 1b Up 70 65 75 55 Down 200 >150 >150 >150 Clutch current, amp 0.01%9 0,020 0.020 0.019 25-w drive motor, v 119 119 119 119 Running current, amp 0.5 0.5 0.5 0.5 Stall current, amp 8.5 8.0 8.6 7.8 Buffer stroke, in. 3.0 3.4 2.8 2.7 Full rod stroke, in. 51.18 51.0 51,07 51.0625 Rod speed, in./sec Up 0,527 0.536 0.532 0.53 Down 0.531 0.538 0.534 0.542 31 CRNL-DWG 64-4470 FISHER GOVERNOR el VENT MODEL 67H PRESSURE REGULATOR MOISTURE 250 psig 40 psig ANALYZER X (-] REGULATOR BYPASS Fig. 2.1. Moisture Diffusion Test — Cover Gas System. Diaphragm Leakage A study is being made of several different makes of helium pressure regulators to determine relative susceptibility to diffusive inleakage of moisture through the regulator diaphragms. The Fisher governor type 67H, which is used presently in the MSRE helium supply system, was tested with a MEECO model W moisture analyzer (see Fig. 2.1l). The results indi- cate diffusion of moisture through the diaphragm of about 1 x 10™% std cm?/sec. At the normal system flow of 6 liters/min, this rate of inleak- age would add 1 ppm of moisture to the helium. Three different makes of regulators were subjected to a comparative test where the interior was evacuated and the exterior was flooded with helium, and the leak rate was checked using a CEC model 24-120A helium leak detector. Results are given in Table 2.2. The Victor model VTS-201 regulator is less complicated than the Grove regulator and apparently will reduce the inleakage of moisture to an acceptable level, so it will be installed in the reactor cover gas system and tested in place. Table 2.2. Results of Diaphragm Diffusion Test Diaphragm Leakage (cm?/sec) Regulator Helium Moisture a w— Fisher governor model 67H 2 x 107 1 x 1074 Victor model VTS-201 g8 x 107 ) Grove model RBX 204-015 2 x 10710 *Maximum range., bMinimum sensitivity. 32 Samplers Temporary Drain Tank Sampler A temporary sampler (Fig. 2.2) was designed for use during prepower operations to isolate a 10-g sample from the fuel drain tank FD2. The line used during the initial filling provided access to the tank. Sampling procedures are similar to those used for a proved method of sampling on the Engineering Test Loop. Five samples of flush salt were taken without difficulty with this sampler prior to the initial filling of the fuel system loop. PHOTO sez64 SAMPLER BALL VALVE SLIDING SEAL SAMPLE CAPSULE S ATMOSPHERE FITTING FOR ATTACHMENT _ TO DRAIN TANK CONTROL FITTING Fig. 2.2. Drain Tank Sampler for FD2. Temporary Fuel System Sampler A temporary sampler that was devised for use on the fuel system during the initial flush salt operation is shown in Fig. 2.3. It will be replaced by the sampler-enricher system as soon as that installation is complete. The sampler has a cable drive unit of the same type used in the permanent installations to move a capsule into and out of the pump bowl. The box which encloses the drive unit serves as both carrier for the sample and seal for the pump during sampling. A hand crank extends through a buffered double O-ring seal into the box and operates the drive unit. There is a ball valve at the bottom of the sampler assembly to seal the sample during transport to the analytical laboratory. The box is connected to the pump bowl by a flanged joint in a temporary transfer tube, which also contains a ball valve to seal off the pump bowl while the sampler is removed. An evacuation system permits purging of oxygen and moisture prior to opening the isolation valve to the pump bowl. Coolant Salt System Sampler The coolant salt system sampler consists of a dry box connected to the coolant pump bowl gas space by a transfer tube. Two ball valves are 33 CABLE DRIVE HOUSING = BUFFER CONNECTION ATMOSPHERE CONTROL FITTING SAMPLER BALL VALVE SAMPLER FLANGE TRANSFER TUBE BALL VALVE Fig. 2.3. Fuel System Temporary Sampler. used to isolate the dry box from the pump bowl. Figure 2.4 shows the installation. The 1-1/2-in.-diam transfer tube is similar to that used for the fuel sampler-enricher system except that no expansion joint is required because the coolant pump does not move with changes in system temperature. The dry box has two ports in it: one for direct viewing and one for lighting. Necessary manipulations inside the box are done by one hand through a glove port. A cover over the glove port protects the rubber glove against damage from pressure changes inside the box 34 during sampling and system cleanup. The cover is closed at all times except when the glove is in use. The sample carrier, mounted above the dry box, has a sliding gas seal at the top which permits lowering a rod and sample capsule assembly through a ball valve into the dry box. The seal keeps the atmosphere in the box from being contaminated with 0, vwhile the capsules are being inserted or removed. It, together with the ball valve, retains an inert atmosphere in the carrier while the sample is being transported to the analytical facilities. A drive unit assembly, similar to the one for the sampler-enricher system, is located in the top of the dry box. No position indicator is included since movement of the cable can be observed through the viewing port, and it can be visually determined when the sample capsule latch has reached the stop at the pump bowl. No shielding is required for this sampler since the induced activity in the coolant salt has a very short half-life. Also, since there is no radioactivity, only single containment is necessary during transport. BALL VALVES| VIEWING PORT TRANSFER TUBE Fig. 2.4. Coolant Salt System Sampler. 35 A system of key interlocks is installed to assist in assuring that the proper sequence of operation is followed. With this system, the five valves opening into the dry box are locked in position. A key is used to unlock one valve, which can then be opened. A second key locks the valve in the open position, sealing the first key in the lock and releas- ing the second key. This second key is used to unlock another lock on a valve or switch to permit the next step in the sequence., Switches are used to check the dry-box pressure prior to taking a step. An alarm will sound when the switch is locked {or unlocked at times) if the box pressure is not proper for the next step to be taken. This mechanical restraint method of interlocking appears to be a satisfactory alternative to the electrical method used on the fuel sampler-enricher. The sampler has been installed and is now in routine operation by the reactor operating personnel, Engineering Test Loop After 15,400 hr of trouble-free operation, the Engineering Test Loop (ETL) was shut down because of excessive oil leakage at the pump shaft seal. While this repair is not a major one, the loop was put into standby for the duration of the reactor check-out and startup. The following is the result of an experiment run just prior to the shutdown. Zirconium Oxide Cold Trap One object of following the oxide concentration in molten-salt systems is to prevent the buildup of undesirable precipitates within the circulating streams. Due to uncertainties in the method used for sampling and analyzing for oxides, results are obtained which at times appear to have a bias about equal to the oxide solubility limit. An indirect, after the fact, method of determining oxide concentration has been to measure the amount of water formed during subsequent HF treatment of the salt. However, since the treatment is performed in a special tank, the oxides which were not transferred from the primary loop with the salt would not be included in the inventory. Such a holdup might occur in the heat exchanger because of the strong tempera- ture dependence of the oxide solubility and because the heat exchanger is the coldest zone in the circulating stream. One method of preventing the random buildup of such temperature-dependent solubility deposits would be to insert a colder zone into a region of low salt flow rate. By observing the change in the heat transfer characteristics of the zone as an indication of the solid buildup, it would be possible to process the salt before the oxide concentration became high enough to start deposition at the heat exchanger. A test was run in the ETL to determine the oxide accumulation rate in such a cold zone. The test assembly proper consisted of a 1/2-in.-OD INCR-8 tube in- serted into the pump bowl through the enricher-sampler access nozzle, the lower end of the tube being extended about 4 in. below the surface 36 ORNL-DWG 65-4171 4 cfm AIR TO VENT SURFACE ( -+ OF SALT . THERMOCOUPLES {(TYPICAL) Fig. 2.5. Detail of Lower End of Cold Trap. | | % | | | | | | 1 ;—EXISTING SPIRAL SHRQUD AT SAMPLER STATION of the salt. Cooling air was introduced at the bottom through a smaller, concentric tube and exhausted through the annulus between the two tubes (see Fig. 2.5). The unit was operated for 48 hr with the mean salt temperature in the loop at 1200°F and the tube-salt surface temperature at 1050°F. TUpon removal from the pump, the portion of the tube which had been submerged in the salt was covered with a crystalline coating 1/32 to 1/16 in. thick, Subsequent petrographic examination indicated the crystals to be predominantly zirconium oxide. The running of additional tests has been held up pending reactivation of the ETL, at which time the tests will be extended to include other possible contaminants. Maintenance The program to demonstrate remote maintenance tools and techniques was continued in conjunction with the installation of reactor components. Because of the construction schedule and the status of some of the main- tenance equipment, some portions of the program were delayed or elimi- nated. The description below gives the details of some of the work and the changes resulting from the experience with the tools. Large components, such as the pump bowl and the heat exchanger, were picked up from their installed positions and moved vertically until they were clear of adjacent equipment, thus simulating the most difficult phase of the remote handling procedure. A log was kept of any physical obstructions and the action required to clear them. Several days of effort were required specifically on the process of removing and stowing the meny auxiliary lines preparatory to the pump bowl removal. 37 ORNL—DWG 65-4172 TEFLON PRESSURE SEAL e INSIDE TUBING CONNECTOR GASKET PLATES i TO SEAL TOOL FOR BOOT — TOOL AND SEAL LIGHT ACCESS PLUG — HOSE CLAMP [oajE o] PLEXIGLAS WINDOW j=m FOR LIGHT AND VIEWING N : == PLASTICBS'CI)_TEXIBLE GRAPHITE SAMPLER p= WORK SHIELD ; ‘fl¢,¢UWfl%fl\\\\\ !\fi \ O-RING SEALS 7 AN Fig. 2.6. Tool and Light Access Plug for Graphite-Sampler Work Shield. The portable maintenance shield was set up over the center line of the pump to practice the handling of the pump rotary element, the pump motor, and other components in this area., The tooling for freeze flange 100 was tried through the portable shield as this is by far the hardest flange to maintain. Finally, all five freeze flanges were made up using the long-handled tools but not the complete remote procedure., Some difficulties were encountered with the freeze-flange clamp operators.4 They have been redesigned, the new parts have been fabricated, and they will be tested after assembly. After the reactor was assembled, the graphite sample assembly and the control rods were installed with remote tooling. Preparations are now in progress to test the atmosphere control system to be used during graphite sampling. A special shield plug, shown in Fig. 2.6, with a window and tool penetration, was designed and built. It provides a flanged boot gas seal to allow room for horizontal movement of the tool. The basic remote operation utilizes an external-light-source viewing device, which incorporates a mirror and lens system to keep the viewer out of the direct radiation field. 38 Design and fabrication were completed on an offset socket wrench to operate specifically on the bolts hidden by the control rod thimbles, and the design was completed on a long-handled heater tool used to thaw out the salt frozen around the graphite sample access joint. Design and testing of cooled lights to view inside thermally hot vessels were con- tinued. Instrument Development Ultrasonic Single-Point Molten-Salt Level Probe During the past report period we have initiated a program of assist- ance in the design, fabrication, and testing of an ultrasonic molten- salt level probe, This probe, which was developed for the AEC by Aero- projects, Inc., with ORNL assistance, provides a single-point indication of molten-salt levels inside closed weld-sealed vessels. ORNL partici- pation in this project consisted of reviewing the Aeroprojects probe design and incorporating such revisions as were required to satisfy the metallurgical, containment, and environmental requirements of a reactor- grade installation; fabrication of those parts of the probes that required special materials and fabrication techniques; providing a test facility; and providing assistance in testing the device, In principle the ultrasonic probe is basically an acoustical im- pedance device. Level is determined by detecting the increase in energy transmission (increased loading) which occurs when salt contacts the probe. The probe system consists of a tank probe assembly, an excita- tion rod, a transmitter, and a receiver, The tank probe assembly con- sists of a vertical (1/2-in.-diam) rod (which has a horizontal flat sensing plate attached at the bottom or tip end) and a "force-insensitive mount, " which supports the rod and forms a helium leak-tight containment seal. The design of the force-insensitive mount (which is a proprietary item) is of particular interest since its use permits ultrasonic energy to be transmitted through one or more vessel walls from a transducer located in a hospitable environment outside reactor shielding and contain- ment. The tank probe is an all-metallic (INOR-8) weld-sealed assembly. The excitation rod, which is used to transmit energy to the probe, is a 1/2-in.-diam stainless steel rod. The transmitter is a magnetostrictive transducer, located at the ouber end of the excitation rod and excited by an electronic power oscillator. The length of the probe and excita- tion rod and the dimension of the sensing plate are chosen so that the system is resonant at the power oscillator frequency. The receiver consists of two piezoelectric crystals attached to the excitation rod at points (near the outer end) which are equal to or less than one- quarter wavelength apart at the frequency of operation. The outputs of the crystals are applied to the inputs of a differential amplifier. The output of the differential amplifier is used to operate a relay, which in turn can be used to operate high- and low-level signal lights or interlocks. 39 Since the probe excitation system is resonant, an increase in load- ing, such as occurs when salt is in contact with the sensing plate, will result in a decrease in the amount of reflected energy and consequently in the magnitude of the standing wave present on the excitation rod. The receiver functions as a standing wave ratio detector, which will operate a relay when a change in loading, caused by salt rising above or dropping below the probe tip, produces a change in the standing wave on the excitation rod. Review of the contractor's design and preparation for installation of the ultrasonic probe in the level test facility were started in May 1964. A probe assembly design submitted by the contractor (Aeroprojects, Inc.,) was reviewed and revised as required to satisfy reactor systems design and containment criteria. A probe assembly was then fabricated in ORNL shops and shipped to the contractor for testing. After the contractor's tests were completed, all component parts were shipped to ORNI, and installed in the molten-salt level test facility. Final test- ing of the system was begun in December. An Aeroprojects engineer as- sisted with the initial tests. During the initial tests the frequency of the driving signal was readjusted to obtain optimum performance at each test temperature. Temperature during the test ranged from 1000 to 1500°F, and the excitation frequency changed as shown in Fig. 2.7. The equipment operated well, was easily installed and adjusted, and had no measurable hysteresis, Following these tests a long-term test period was started to deter- mine whether the system parameters would change with time. Data taken over the last three months have indicated that some changes have taken place. During the initial test period, the highest resonant frequency (the system has three resonant frequencies within the range of the excitation oscillator) gave the greatest signal output when molten salt touched the sensing plate. However, for reasons unknown at this time, the greatest signal output is now obtained when the lowest resonant frequency is used. The force-insensitive mount and the sensing plate were maintained at a constant temperature for this test, ORNL-DWG 65-4473 49,700 Q\ Fig. 2.7. Change in Excita- tion Frequency of Ultrasonic Level Indicator vs Temperature of Sensing Plate, 49,680 \\\\“‘42 49,660 \\\T 49,640 1000 1100 1200 1300 1400 1500 TEMPERATURE (°F) FREQUENCY (cps) 40 ORNL-DWG 65-4174 49,800 \\x \Fx 49,700 T ] \x\ - RESONANT FREQUENCY NO. 1 49,600 1 | | | = 49,400 a Q \ > \o\ o = N o 49,300 - § \Q\o- &« RESONANT FREQUENCY NO. 2 49,200 ' ' ' 49,000 \.\ \\ 48,900 ""‘-e‘._. RESONANT FREQUENCY NO. 3 48,800 I l ‘ 60 65 70 75 80 85 AMBIENT TEMPERATURE (°F) Fig. 2.8. Variation in Reso- nant Frequencies of Ultrasonic Level Indicator with Changes in Room Am- bient Temperature., Force-insensi- tive mount. Temperature constant. In addition to this change in sensitivity, there have been constant changes resulting from changes in ambient temperature. This was expected, as the system is resonant and any change in ambient temperature would change the length of the system and therefore its resonant frequency. These changes have repeated very well between 65 and 85°F, as shown in Fig. 2.8. Aeroprojects is investigating the feasibility of replacing the present oscillator with one which would automatically adjust to the natural frequency of the probe system. An ultrasonic probe similar to the type tested will be installed Because of the high expected corrosion rate in the fuel storage tank during chemical processing system opera- tions, some dimensional changes were required in order to provide For example, the thickness of the sensing in the MSRE fuel storage tank. adequate corrosion allowance. plate has been increased from 0,078 to 0.5 in. and the design of a transition joint has been modified. MSRE probe will be 25,000 cycles. probe was 50,000 cycles.) progress at the manufacturer's plant. The excitation frequency for the (Excitation frequency of the test It is expected that the oscillator circuit for the MSRE probe will be revised to incorporate the automatic frequency control feature discussed above, Design of the MSRE probe assembly has been completed, and fabrication is in progress at ORNL. Design and fabrication of other parts of the probe system are in Installation of the probe in the fuel storage tank is scheduled for completion before the start of operation of the chemical processing system. Testing of the prototype system installed in the MSR level test facility is continuing. 41 Float-Type Molten-Salt Level Transmitiers Testing of the prototype float-type molten-salt level transmitters? has been terminated. The two systems installed on the level test facility operated satisfactorily for 29 months. Measurements made at the end of the test indicated a maximum change in calibration for both systems of 0.2 in. After the final calibrations were completed, the transmitter in which the differential transformer core was suspended below the graphite float was removed from the test loop and inspected. A visual check of the graphite float showed no deterioration of the graphite, There was no visible indication that molten salt had penetrated the graphite, and no granulation of the surface had taken place. Machine marks made when the surface of the float was being finished were still visible, The INOR-8 core tube, which had been immersed in the salt for this entire period, was still clean., There was some indication that the nickel wire used in winding of the differential transformer may have become brittle, since one lead wire broke at the point where it entered the transformer when the transformer was disassembled. Twisting the end of the broken wire caused it to break again, The transformer will be sectioned and examined to determine the extent of the embrittlement, The transmitter in which the differential transformer was mounted above the float was not disturbed after termination of the tests. This transmitter is being used routinely to check the performance of an ultrasonic level probe, which is being tested in the level test facility. The pump bowl level indicator installed on the MSRE prototype pump test loop has operated satisfactorily since installation, This system is sensitive and repeats very well, Some difficulty was encountered in obtaining an accurate field calibration of the transmitter, A laboratory setup was made and used to determine the procedure necessary to eliminate the calibration difficulty. Installation of a float-type level transmitter on the MSRE coolant pump is now complete except for final calibration and adjustments. This transmitter has been at system temperature since the start of precritical operations and except for errors in calibration has performed satisfactorily. Design, development, and testing of a high-temperature differential transformer for use with a ball-float-type molten-salt level transmitter on the Mark IT MSRE fuel circulating pump have been completed. This transformer showed no indication of physical deterioration or change in performance characteristics after two months operation at 1200°F, Conductivity-Type Single-Point Molten-Salt Level Probe Installation of conductivity-type single-point molten-salt level indicator prdbes6-7n8 in four MSRE drain tanks was completed. (Plans 42 to install a fifth probe in the fuel storage tank were canceled because of the high corrosion rate expected in the fuel storage tank during operation of the fuel processing system.) Each installed probe system was checked out and adjusted when the tanks were empty and at operating temperature. Under these conditions and with the excitation current to each probe adjusted to 6 amp, the noise level on each signal system (measured at the input terminals of the alarm transducer) was less than 3 my. When salt was added to the coolant drain tank, the signal levels (measured at the input to the alarm transducers) were 87 mv for the "high-level probe and 280 mv for the low-level probe., An alarm point of 35 mv was selected, and all alarm transducers were adjusted accord- ingly. All four probes performed very well during initial filling operationm, and three probes have continued to perform satisfactorily. However, the probe installed in the fuel flush tank failed during January. The cause of the failure was found to be a break in the excitation circuit produced by severe oxidation and embrittlement of the copper-clad, mineral- insulated copper wire excitation cable in the region adjacent to the point of attachment to the probe head assembly. From examination of the cable it was obvious that the operating temperature was greater than the 400°F maximum temperature anticipated during design of the probe, The probe head and cable design was based on the assumption that the head and cable would operate in open air; however, in the MSRE installations, the head and part of the cable were covered with insulation. The probe head and cable assembly will be modified during the March shutdown. All sections of copper cable which might operate at high temperature will be replaced with stainless-steel-sheathed, ceramic- beaded nickel wire cable. Design of the modified replacement assembly and fabrication of necessary replacement parts and cables have been completed. Similar modifications will be made on the other three installed probes if inspection indicates that the operating temperature of their cables has been excessive. MSRE Bubbler-Type Molten-Salt Level Indicator The bubbler level systems,’ which measure the level of molten salt in the fuel and coolant system pump bowls, were placed in operation when the fuel and coolant loops were filled with salt. Performance of the bubbler systems was generally satisfactory; however, some difficulty is being experienced in obtaining a steady purge flow rate to the bubblers, and there 1s a slight offset in the zero of the fuel pump bowl bubblers. 43 The purge flow instability 1s due to poor throttling characteristics of the hand valve used to control the flow., These valves will be replaced as soon as possible, The zero offset is due to unexpectedly high pres- sure drops in the purge line between the differential pressure transmit- ters and the bubblers. This offset is apparently a fixed characteristic of the system and cannot be eliminated without major piping changes. It can, however, be compensated by adjustment of the differential pres- sure transmitter zero if the purge flow i1s held constant, and should cause no trouble once the purge flow instabilities are corrected. There has been no change in status of the developmental bubbler system installed in the pump test facility. Testing and observation of the performance of this system have been discontinued. Thermocouple Development and Testing Drift Tests. Testing of eight metal-sheathed, mineral-insulated Chromel-Alumel thermocouples made of material selected from MSRE stock was continued at a reduced level of effort. The maximum drift observed during 16 months of operation at MSRE temperatures was in the range of +2.6 to +3,5°F, The calibration drift during the period from April 20, 1964, to September 28, 1964, was in the range of +1.9 to +2.1°F. The maximum drift noted during the past five months has been less than 1°F. Thermocouples on Engineering Test Loop. Performance of eight MSRE prototype surface-mounted thermocouples on the ETL continues to be satisfactory. After approximately three years of operation, all couples are still functioning. Routine observation of these couples has been discontinued. Thermocouples on the Prototype Pump Test Loop. Ten MSRE prototype surface-mounted couples on the PTL continue to perform satisfactorily. Except for short periods of operation, this lcop was shut down from January through August of 1964; routine observation of performance and logging of data were discontinued. Radiation Damage Tests of Thermocouple Lead Wire, Disconnects, and Sealing Materials. Testing of a typical copper-sheathed multiconductor extension cable, discommect, and thermocouple assembly!? was terminated July 30, 1964. The assembly was exposed to the 6000 gamma source for a period of eight months and accumulated a total exposure equivalent to 5.6 x 10° r. The insulation resistance between a typical thermocouple circuit and ground (measured with a 500-v megger) was 2 X 10% ohms at the end of the test. Pressure buildup in the sheathed cable assembly continued to the end of the test. Irradiation of specimens of Mica-Temp and Super-Temp radiation- resistant, ceramic-insulated wire, sealed in copper tubes, was started in February and discontinued on July 30. Gas pressure buildup was observed in these assemblies until the end of the tests; however, no change in the resistivity of the wire insulation was observed. 4d, End Seals for Mineral-Insulated, Stainless-Steel-Sheathed Copper Wire Cables. Tests were conducted to determine whether mineral-insulated copper vwires sheathed in a stainless steel tube could be sealed against molisture with a radiation-resistant material. Ceramacite C-100, a ceramic-vitreous enamel sealing material produced by Consolidated Electrodynamics Corporation, appears promising. Although this material requires a curing temperature of 1200°F, seals that were leak-tight to helium and moisture were obtained when the copper wires were protected by a helium atmosphere during the curing. Three seals of this type, which were subjected to water pressures up to 60 psig, withstood a 500-v insulation breakdown test after moisture was dried from the outer surface. Coolant Salt Radiator AT Thermocouples. Laboratory tests were conducted on thermocouple and extension lead-wire materials used in the differential temperature thermocouple installation on the MSRE coolant salt radiator to determine how much mismatch of materials could be tolerated without incurring excessive error in the computed reactor heat power signal. Since a 5% accuracy in the overall heat power measurement was required, and since a number of factors, including the flowmeter accuracy and the accuracy of various electronic components, as well as the accuracy of the thermocouples, contribute to the overall accuracy of the heat power computation, an arbitrary maximum inaccuracy of 2-1/2% was assigned to the AT measurement. A 2-1/2% error in temperature measurement corresponds to an error in the emf produced by the thermocouple circuits equivalent to 2°F. Test results showed that, under certain conditions of mismatch between thermocouple and extension lead-wire materials, error voltages equivelent to as much as 2°F can be generated in a single Jjunction when the temperature of the Jjunction is varied over the range from 32 to 150°F. Since several Jjunctions are involved, the need for careful design and matching of material was apparent, and the design of the MSRE installation was revised to obtain the maximum possible cancellation of Jjunction effects. Additional error voltages can be produced by variations in ambient temperature if the thermocouple lead-wire material is not perfectly homogeneous. Tests are being performed at the MSRE to determine whether such effects exist in the thermocouple lead wire which was installed for use in the radiator AT thermocouple circuit. Temperature Scanner Installation of the MSRE temperature scanner systems'! was completed prior to the start of precritical operation of the reactor. The scanner systems were used during the initial heat-up of the components and piping and during subsequent operations with salt circulation. Considerable trouble was experienced with series-mode 60-cycle ac pickups in the thermocouple input circuits during initial operation of the systems. The origin of the pickup noise was traced to the scanner cabinet input signal wiring, and the cause of the pickup was determined to be an error in the design of the input wiring, which resulted in the separation of input wire pairs and the consequent formation of a large pickup loop. The trouble was corrected by rerouting the wiring and by 45 adding some magnetic shielding in the scanner cabinets, There was no appreciable series-mode pickup in the external thermocouple wiring; however, considerable common-mode pickup was found to be present on all of the thermocouple input signals. Since the scanner system was designed to have a high common-mode rejection, the presence of this common-mode voltage does not present any problems, After correction of the noise problems, further complaints were received from operations personnel. Investigation showed that the dc amplifiers in the salvaged 17-in. oscilloscopes, used to display the scanned signal, had poor stability and were drifting badly. The oscil- loscope was replaced with a new 21-in, model having adequate stability; however, complaints about calibration error, drift, and signal identifi- cation continued. Another investigation showed that the drift had been substantially reduced by the use of the new oscilloscope, that the gains of the amplifier were not changing, and that a bi-stable condition existed which would cause ambiguity in the signal identification. No apparent reason for the complaints about calibration errors and drift could be found except for the possibility of misadjustment of front panel amplifier gain and oscilloscope controls. A revision in the mode of operation and method of connecting the marker generator, which supplies the identification signal, corrected the signal identification trouble. Removal of unnecessary control knobs and locking of critical adjustments apparently eliminated the changes in calibration and drift. Circuit revisions have been made in three 17-in. oscilloscopes (two scopes are required on the scanner system). These revisions eliminated the drift and generally improved the performance of the scopes. A cover was installed over all controls on these scopes which were not required for routine operations.,. The five scanner systems installed at the MSRE are now operational and are performing satisfactorily. The life of the mercury switches used to scan to signals has been much longer than was expected. Four of the five scanner switches have operated continuously without main- tenance, excessive electrical noise, or vibration since September. One switch required cleaning and repair after a nitrogen purge line connection was accidently broken. It had been expected that the mean life of the switches between routine cleaning or repair would be about 1000 hr. Radiation Thermocouple Scanner Reference Voltage Supply. A stable, adjustable millivolt reference supply, equipped with automatic cold- junction compensation, was required for use with the scanner for the thermocouples on the MSRE radiator. Since neither inquiries nor a literature search revealed a commercially available device which met the MSERE requirements, a special reference supply, shown in Fig. 2.9, was designed, developed, and fabricated. Basic components of the device include a Minneapolis-Honeywell Zener diode constant voltage supply, a TLeeds and Northrup bridge circuit panel, and a Minneapolis-Honeywell circular slide-wire. The device can be operated as a preset fixed voltage supply or as a variable voltage supply. When operated as a Fig. 2.9. Thermocouple Reference Voltage Supply (Side View). variable supply, the temperature equivalent of the output voltage can be read directly on the front-panel-mounted microdial which is used to adjust the voltage. Final packaging and testing have been completed. Test results indicate that the change in output voltage will be less than the voltage equivalent of 1°F for line voltage variations between 105 and 125 v ac and ambient temperature variations between 73 and 114°F. Single-Point Temperature Alarm Switches Some difficulty has been experienced in obtaining relisble operation of the Electra Systems single-point tempersture alarm switch modulesl? installed at the MSRE. The number of cases where the modules became inoperative has been small; however, numerous cases of set point shift have been reported, and several modules have been found to have dual trip points. No consistent pattern to the shifts has been found, and in some cases the reported shifts could not be verified; however, during the course of testing and inspection the following component defects were discovered: 47 1. The printed circult board contacts were not gold plated and had become oxidized, producing an open or high-resistance contact. (The switches tested and evaluated before purchase of the MSRE switches were gold plated.) The low potentials available in the input circuits are not adequate to break down such oxidation, In many cases it was found that cleaning the input circult contacts on the printed circuit board with a pencil eraser was sufficient to restore the original set point. 2. A number of calibration jacks (which have a contact in series with the input signal) also had developed open circuits or high contact resistance. The contacts on these jacks are gold plated, and repeated operation did not improve the performance of jacks having high resistance or degrade the performance of jacks having low resistance, 3. GSeveral trim pots, installed at ORNL to obtain a control hysteresis characteristic, became open circuited or erratic after a few adjust- ments, 4. The bias point of a transistor in the modules having dual (ambiguous) set points had shifted. The bias shift is believed to have been caused by aging of the transistor. A group of modules was selected for use in a test., Printed circuit contacts on these modules were gold plated, the calibration jacks were permanently shorted out with wire jumpers, the trim pots were replaced with fixed resistors, and resistor values in the modules having ambiguous set points were changed to restore the proper bias levels. The set points on these modules were checked routinely for one month, None of the set points on the modified modules shifted during this period. The modifica- tions made on the modules tested are being made on the remainder of the modules installed at the reactor. It is expected that these modifica- tions will eliminate the set point drift and restore the reliability of the switches to a level comparable with that experienced with the switches which were tested and evaluated prior to purchase of the MSRE switches. An attempt will be made to find acceptable jacks and trim pots; however, no components will be replaced in the MSRE modules until the reliability of the replacement component has been proven. High-Temperature Resistance Thermometers Four resistance thermometers, rated for short-term operation at 1400°C, were operated at 1350°F to determine the feasibility of using resistance thermometers to obtain precision temperature measurements under conditions of long-term operation at MSRE temperatures., Two of the four thermometers became erratic shortly after the start of the tests. One of the remaining two thermometers, supplied by a different manufacturer, also became erratic after 1250 hr of operation. The performance of the fourth thermometer remained stable after 1850 hr of 48 operation. These tests have now been terminated. An attempt will be made to determine the cause of the three thermometer failures. High-Temperature NaK-Filled Differential Pressure Transmitter Both MSRE coolant salt flow channels performed satisfactorily during the initial startup and operation of the MSRE coolant salt system; how- ever, several days after the start of coolant salt circulation, the output of one channel started drifting down scale. The output of the other channel remained steady. The trouble was isolated to a zero shift and possibly a span shift in a NaK-filled differential pressure transmitteri? in the drifting channel. The exact cause of the shift has not been found as yet; however, it has been determined that the body of the transmitter in the defective channel is extremely sensitive to ambient temperature, while the transmitter body in the other channel is not. It was also determined that the defective transmitter output signal was not exces- sively affected by variations in system pressure. The spare transmitter was tested to determine that it was operable and that it was not sensi- tive to ambient temperature variations. No excessive temperature sensi- tivity was found., Since two of the three transmitters are not excessively sensitive to ambient temperature, we presently believe that the drift is caused by a defect in fabrication and not by an error in design. Also, since the defective transmitter does not appear to be sensitive to system pressure, we believe that the NaK-filled part of the transmitter is intact and that there have been no NaK leaks. Further tests and observation will be made with the hope of finding the cause of the trouble and of correcting it without having to cut the Transmitter out of the system. If the cause of the trouble cannot be found, the defective transmitter will be replaced with the spare trans- mitter during the March shutdown, and the defective trensmitter will be tested, inspected, and disassembled as required to determine the reason for the drift. Helium Control Valve Trim Replacement Four helium control valves failed in service during initial opera- tions of the MSRE. The initial complaint in each case was that the valve had stuck and would not respond to the control air signal. The defective valves were replaced with spare valves and were disassembled. Visual inspection revealed severe galling between the plug and seat. The galling was thought, at the time, to have been caused by small metal particles which either had not been removed when the valves were cleaned or had been blown into the valves from the helium supply system, New trim was fabricated in ORNL shops and installed in the valves after all observable metal particles and burrs had been removed from the valves and after the valves had been carefully cleaned and inspected. The first valve assembled stuck during the first cycle of operation. Subsequent examination showed that, as in the case of the initial 49 failures, severe galling of the close-fitting trim combination (17-4 PH plug and Stellite seat) had occurred and that the trim had been properly fabricated and hardened. A second valve was carefully assembled to ensure that there was no misalignment. This valve also failed 1In the same manner during the first cycle. A third set of trim stuck while the plug was being manually inserted into the seat. After discussions with ORNL metallurgists and Mason-Neilan Company engineers, it now appears that, although a number of similar valves are presently operating satisfactorily in the MSRE, and although similar trim has previously been operated successfully in water systems, the 17-4 PB-Stellite trim material combination may not be satisfactory for the MSRE conditions of extreme cleanliness and nonlubricating dry-helium service. Other trim combinations including 440C stainless steel to Stellite and Stellite to Stellite have been recommended and will be evaluated. Control Valve Actuator Motion Multiplier A motion-multiplying device was designed and developed for use in obtaining a 1-in. stroke, required by some of the valves in the MSRE, from all-metal pneumatic bellows-sealed 1/2-in.-stroke valve actuators. These bellows actuators, which were tested and proved in HRT operation, were available as surplus at ORNL. The motion multiplier is basically a rolling-vwheel device in which the actuator motion is applied to the axis of the wheel., The motion is transmitted to the valve stem by an Flgiloy tape. Since the device is required to operate inside the MSRE containment vessel under conditions of high-level nuclear radiation and a dry nitrogen atmosphere, it was designed to operate without lubricants. An attempt was made to use a commercially avallable flexure pivot assembly for the wheel axis. Although this flexure was operated within the manufacturer's published specifications, it failed repeatedly after a small number of test cycles. Subsequent discusslons with the manufacturer revealed that the published load capacities applied to static loads (no cycles) and that the load capacity decreased with the number of cycles. Since the manufacturer could not supply a flexure that would meet the requirements, the flexure was replaced with a needle bearing assembly. After replacement of the flexure, satisfactory performance was obtained in shop tests. Two motion multipliers were installed on valves in the MSRE, and the performance of these devices under field conditions is being observed. 50 Pump Development MSRE Pumps Molten-Salt Pump Operation in the Prototype Pump Test Facility. The spare rotary assemblies for the fuel and coolant pumps were sub- Jected to hot shakedown tests in the prototype pump test facility circulating salt LiF-BeF,-ZrF,-ThF,-UF, (66.4-27.3-4,7-0.9-0.7 mole %) at 1200°F for 452 and 1024 hr respectively. During operation with the spare fuel pump, the effectsl% of cooling alr on the temperature distribution in the upper shell of the pump tank were investigated. A temperature asymmetry as great as 100°F was ob- served before the test was halted because of increased pump power requirements. Posttest inspection revealed a deep groove in the suction shroud of the impeller and a buildup of metal on portions of the mating surface of the volute casing. The metal buildup was found by spectro- graphic analysis to be INOR-8. Room-temperature measurements indicated that the axial running clearance between impeller and volute casing was much smaller than had been anticipated.l4 Additional measurements will be made during fabrication and assembly to ensure the presence of ade- quate axial running clearance, and the design of the cooling shroud for the upper shell of the pump tank will be revised to provide a more uni- form distribution of cooling air flow. The coolant pump spare rotary assembly is being prepared for delivery to the MSRE, and the fuel pump spare rotary assembly was reassembled and installed in the prototype pump test facility for further tests with circulating salt. Lubrication Systems and Lube Pumps. Tnitial operation of the lubri- cation systems in the MSRE was hampered by failure of the stator insula- tion in one of the pump motors and by low electrical resistance to ground in the stators of others of the pump motors. Intrusion of moisture into the stator was suspected. In order to eliminate this difficulty the stator was rewound, a potting compound (Dow Corning Silastic RTV 731) was used to seal motor housing joints, and a moisture-resistant coating of paint (Sherwin Williams epoxy white B69W6) was applied to the exterior surface.'” Four of the motors were treated, and two additional motors will be treated as soon as practicable. Entrained gas was observed in the circulating oil during proof tests prior to delivery of the lubrication systems to the MSRE. A problem of low flow and harsh strident noise during startup of the standby pump was noted.l® A transparent mockup of the reservoir used in the lubrication systems was fabricated and is being assembled into a circulating oil test stand. The flow passages and baffles in the reservoir will be revised as indicated by the tests to minimize the entrainment of gas into the circulating oil. 51 The lubrication pump which is circulating oil in an endurance testl? has been operated without incident for 13,800 hr. Recently the pump has been operated for 1 hr and then turned off for 1 hr to subject it to thermal cycling conditions. A total of 370 thermal cycles of a planned 1000 cycles has been accumulated. Measurement of the Concentration of Undissolved Gas in Circulating Liquid. Two gas-liquid systems are being used in the investigation of the potential problems of entrainment of gas in salt in the MSRE fuel pump. One is air and water in a Lucite pump tank,18 which models the MK-2 pump tank, and the other is helium and molten salt in the prototype pump and test loop,1?:%0,21 which is analogous to the MSRE fuel loop. A radiation densitometer has been designed and installed on each system. The densitometers have identical electronic and readout systems; they differ from each other in the radiation ener and the sizes of the sources and detector shields. A 600-curie *7Pm source of 0.223-Mev beta rays, which yields 1.8 curies of 0.038-Mev x rays, is being used on the water system, and a 40-curie '37Cs source of 0.662-Mev gamma rays is being used on the molten-salt system. Previous use of a densitometer produced interesting but questionable results because the instrument was unstable,20 and this prompted the investigation of new detector and electronic components. The RCA 2020 electron multiplier phototube has demonstrated stabllity during several weeks of continuous 100-pa current output without noticeable fatigue. No other phototube has been obtained as yet that has matched this per- formance. Organic and inorganic phosphors of several types were studied. The response of each phosphor was obtained during static measurements and compared to its response during dynamic measurements of identical density variations. All phosphors tested exhibited undesirable tran- sients when subjected to near-step density changes; however, the duration of transients in the response of organic phosphors was one to three orders of magnitude less than that of the inorganic phosphors. Most transients were lags, but one organic phosphor was observed to overshoot. Organic phosphors were chosen for the densitometers. An operational amplifier was chosen to receive the phototube output current. The transfer function of the circuit converts current input (0 to 1 ma) to voltage output (O to 8 v), and it has a response time of 0.12 msec, The average value of the input current can be suppressed to zeroj thus only the changes in input current are observable at the amplifier output. The amplifier feeds an integrator which has five output integration times; they are 0,001, 0.250, 1.0, 9.4, and 25 sec. These periods were chosen with regard for the fuel circulation parameters of the MSRE. The short periods permit examination of fine structure, that is, rapid variations in density, whereas the longer periods yield average values. The short amplifier response time (0.00012 sec) inhibits equivocation of detector excursions before integration; thus the inte- grator output indicates changes in density with fidelity. 52 The integrators will be monitored with Visicorders (response time is less than 0.001 sec) and strip-chart recorders, The 0.00l-sec inte- grator will feed a Visicorder, the 0.250-sec integrator will feed a 1/4- sec strip-chart recorder, and the balance will in turn be monitored with a l-sec strip-chart recorder. Provisions can be made to monitor all but the 0,00l-sec Integrator with an in-line data acquisition computer; the logic has been determined. MK-2 Fuel Pump. The design and fabrication?? of a prototype rotary assembly for hot tests have been completed. The design of the pump tank awaits the completion of water tests with mockups of the tank to develop an arrangement of internal baffles and flow channels, which will minimize the entrainment of gas in the circulating liquid. It has been delayed by the efforts to produce radiation densitometers with stability and response characteristics suitable for measuring concentrations of un- dissolved gas in water and salt. A 4-in. section of the MK-2 pump tank was fabricated of transparent plastic. The various baffles and flow channel devices used with it in preliminary water tests did not provide any noticeable reduction in the entrainment of gas in the circulating water. Fabrication and Procurement of Pump Components for the MSRE. The second spare coolant pump drive motor,<” the fourth motor in an order for four motors, passed its acceptance tests and was delivered. Other Molten-Salt Pumps The startup of test operation for the pump with the molten-salt- lubricated bearing?®4 and the PKP molten-salt pump was delayed by the emphasis on delivering reactor pumps to the MSRE. 15. 16. 17. 18. 19. 20. 21, 22. 23. 24, 53 References MSR Program Semiann, Progr. Rept. Jan. 31, 1964, ORNL-3626, p. 24. MSR Program Semiann. Progr. Rept. July 31, 1963, ORNL-3529, p. 23. MSR Program Semiann. Progr. Rept. Aug. 31, 1961, ORNL-3215, p. 65. R. Blumberg, Maintenance Equipment Tested in the Reactor Cell of the MSRE, MSR-64-45 (internal use only). MSR Program Progr. Rept. Feb. 28, 1962, ORNL-3282, p. 61. MSR Program Progr. Rept. Aug. 31, 1961, ORNL-3215, pp. 72-76. MSR Program Progr. Rept. Aug. 31, 1962, ORNL-3369, p. 66. MSR Program Progr. Rept. Jan. 31, 1963, ORNL-3419, p. 40. Ibid., p. 43. MSR Program Progr. Rept. Jan. 31, 1964, ORNL-3626, p. 43. MSR Program Progr. Rept. Aug. 31, 1962, ORNL-3369, p. 71. Ibid., p. 68. MSR Program Progr. Rept. Jan. 31, 1963, ORNL-3419, p. 45. Letter from P. G. Smith to R. B. Briggs, Impeller Rubbing Incident, MSR-65-71 (Jan. 25, 1965) (internal use only). Letter from D. L. Clark to R. B. Briggs, Recommendation to Relieve FElectrical Insulation Problem with the MSRE Lubricant Pump Motors, MSR-64-64 (Oct. 20, 1964) (internal use only). Letter from P. G, Smith to R. B. Briggs, Operation of the Lubrication System for the MSRE Pumps on the Prototype Pump Test Facility, MSR- 64-10 (Feb. 28, 1964) (internal use only). MSR Program Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, p. 41. MSR Program Semiann. Progr. Rept. July 31, 1964, ORNL~3708, p. 166. MSR Program Semiann. Progr. Rept. July 31, 1963, ORNL-3529, p. 50. MSR Program Semiann., Progr. Rept. Jan, 31, 1964, ORNL-3626, p. 40. MSR Program Semiann. Progr. Rept. July 31, 1964, ORNL-3708, p. 164, Ibid., p. 166, MSR Program Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, p. 41. Ibid., p. 40. 54 3. MSRE REACTOR ANALYSIS MSRE Stability Analysis The analysis of the inherent stability of the MSRE was completed. It was found that the system is inherently stable for all operating condi- tions. Low-power transients persist for a long time, but they will even- tually die out because of inherent feedback. Previous studies® have shown that this sluggish response at low power can be eliminated by the control system, which suppresses transients rapidly. Methods Used The MSRE stability study included calculations of transient response, frequency response, and root locus. The transient-response results were obtained with an analog computer simulation and with a digital computer. The digital computer was used for freguency-response and root-locus cal- culations. The most complete representation of the system used included the fol- lowing features: 1. The core was represented by 27 lumps (or nodes) with weighted re- activity feedback from temperature changes in each lump. 2. ©ix delayed-neutron groups were used, with allowance for the dy- namic effects of delayed-neutron-precursor circulation. 3. Exact transport delays (e~T8) were used. 4. The primary heat exchanger and radiator were represented by 50 lumps (or nodes) each. 5. Heat transfer in the pipe runs was included, ©. The effect of delayed power generation by gamma emitters was in- cluded. 7. Xenon reactivity effects were included. Results The closed-loop frequency response (Bode diagram) calculated for the MSRE at 1 Mw and 10 Mw with allowance for these effects is shown in Fig. 3.1. The peak amplitude at 10 Mw is broad, indicating that system per- turbations will be rapidly damped. The peak amplitude at 1 Mw is sharper, indicating that the system will oscillate more before the transient dis- appears. Also, note that the peak shifts to lower frequencies at lower power, indicating that the characteristic frequency of oscillation de- creases with lower power. The effect of power level on the period of oscillation and on the phase margin is shown in Table 1. PHASE (deq) 55 4 ORNL—DWG 65-4i{75 10 | L 1T 1l I T | = —t— MAGNITUDE RATIO A/BK 5 N. =10 Mw NA N 0 2 /x I / /RN = / ™ x E - = \ - 5 x S { ‘N\ No—" MW = y’ M -——- 2 > {10 Mw ’f‘ - s/ '02 poonet 5 . i L { Mw / T FUEL LOOP | N TiMeE =24 sec 2 p i H (@) 7 =10 min =1{min a 0 LLLL LU 0% 2 5 1070 2 5 1072 2 5 o' 2 5 0% 2 5 10 FREQUENGY (rcdicns/sec) 20 80 LA \,\ PHASE N/8K |} —30 \\ /r N vy —-40 A/’ (b) —50 0% 2 5 0> 2 5 1072 2 5 ! 2 5 10° 2 5 FREQUENCY (radians/sec) Fig. 3.1. MSRE Fregquency Response — Complete Model and Current Data. 56 Table 1. Periods of Oscillation and Phase Margin Power Period Phase Margina (Mw ) (min) (deg) 0.1 22 12 0.5 10 29 1.0 7 41 2.0 5 57 5.0 2.5 79 10.0 1.3 29 ®Phase margin is a measure of stability ob- tained from the open-loop frequency response. Larger phase margins imply greater stability. Instability occurs for negative phase margins, In addition to the analysis based on this detailed model and best estimates of system parameters, the effects of changes in the theoretical model and the parameters were studied. This included a systematic de- termination of the worst stability performance to be expected within the range of uncertainty on the design parameters. The analysis showed that the MORE 1s stable not only at the design point, but for any combination of parameters within the predicted range of uncertainty. Suppression of Criticality in the MSRE Cell in Event of the Maximum Credible Accident An estimate was made of the effectiveness of borosilicate Raschig rings in suppressing criticality in mixtures of fuel salt and water that might be present in the bottom of the reactor cell in the event of the rupture of the primary system (maximum credible accident). The borosili- cate rings considered were 1-1/2 in. OD by 1-1/4 in. ID by 1.7 in. long, and contained 4 wt % natural boron. A considerable number of experimental criticality measurements have been made at ORNL on the effectiveness of these rings in solutions of uranyl nitrate 90% enriched in 235U (ref. 2). Mixtures of glass with aqueous solution concentrated to 415 g of uranium per liter of solution have been found to be subcritical as an infinite medium, provided the glass occupies greater than 22% of the mixture volume. A volume fraction of 24% corresponds roughly to close packing on a square pitch of rings of the above size. Other experimental data are summarized in ref. 3 for lower glass volume fractions. TFor rings with the boron con- tent given above, 325 g of 23°U per liter of solution and 17.8 vol % oc- cupied by the glass, the thickness of an infinite critical slab is 26.5 in. 57 If the hemispherical bottom of the reactor cell is filled with these rings, and if the entire contents of the primary circulating system were dumped into the cell, the maximum height the liquid level could attain above the lowest point of the hemisphere is about 1.7 ft. The geometrical configuration would be that of a dish, reflected by water beneath the vessel and possibly also by water above the fuel salt. The concentration of 235U in the salt is about 27.6 g of 22°U per liter (fuel B, 9% enriched uranium) and 46.1 g of 2357 per liter (fuel C, ~35% enriched in 235y). Thus, even if all the salt were effectively replaced by water maintaining the same uranium concentrations, a considerable safety margin from a crit- ical configuration should be maintained. Mixtures of water and salt would be expected to be further subcritical. Tt was concluded that the addition of borosilicate Raschig rings con- taining 4 wt % natural boron will be effective in suppressing criticality even though a most extreme condition should take place. Practical con- siderations lead to conditions which are less reactive than the ones con- sidered above. Effectiveness of a Radial Molten-Salt Blanket on the Breeding Potential of a One~Fluld Graphite-Moderated Reactor Potential advantages of the use of an unmoderated radial molten- salt blanket in improving the breeding capability of a single-fluid thermal molten-salt reactor were studied. The core was a graphite- moderated right cylinder with diameter and height in the range of 10 %o 15 ft and volume of salt between 10 and 20 vol %. The concentration of thorium in the salt was varied between 5 and 13 mole % ThF,. These con- ditions correspond roughly to a 2500-Mw (thermal) reactor with an average power density of 400 w per cm® of core salt. The core was surrounded with o radial blanket of molten salt, identical in composition with core salt. The thickness of the radial blanket was varied between 1 and 2 ft. The neutronics calculations were based on six-group one-dimensional diffusion theory in cylindrical geometry. The GAM-2 (ref. 4) and the MODRIC® pro- grams were used in these studies. A reflector savings of 1 ft for the top and bottom of the cylinder was assumed, to account approximately for the effect of any end reflection on the neutron economy in the core. The study was concerned mainly with the sensitivity of the breeding potential to geometry and graphite-salt volume composition of the core. No neutron losses due to fission products or 233py were considered; hence, the nu- merical values of the breeding ratios obtained should be considered as %gger 1imits. The uranium isotopic composition was assumed to be 92/6/2 3U/235U/236U) . Results For the range of conditions given above, the neutron economy in the core can be optimized independently of the blanket conditions. These re- sults are sumarized in Fig. 3.2, which shows the effect of the carbon- to-uranium ratio on the production and absorption of neutrons in the core. 58 ORNL—DWG 65-4176 (2) 20 VOL % SALT IN CORE _ (6) 10 VOL % SALT IN GORE ‘ ’ 7 -1 | 7 -Iq 1] 1.2 o= — BRmax—] e — b T~ BRmax — 1.0 R S * \-\ * \\- h— f \h.. 0.8 T~ CORE CONDITIONS: COMPOSITION OF BASE SALT: 66/29/5 mole% LiF/BeF,/ThF, ISOTOPIC COMP URANIUM: 92/6/2 (233y/235 /236y) 0.6 ISOTOPIC COMP LITHIUM: (0.005% 5|_|) EFFEGTIVE CORE BUCKLING: 52= (0™ * MEAN CORD LENGTH: 2¢m 0.4 0.2 0% 2 5 104 2x10° 5 10* CARBON/ URANIUM ATOMIC RATIO Fig. 3.2. Effect of ¢/U Ratio on Neutron Economy in a Large Graphite—-Molten-Salt Core. For volume fractions of salt of 20 and 10% (Fig. 3.2a and b respectively), curves are given for the quantities n— 1, £*, and BRpay . These are de- fined as follows: - _ fission neutrons produced in core N = Deutrons of all energies absorbed in #23U + %°°U in core °’ * = total absorptions of neutrons in uranium (?33U + 235y 4 23675) total absorptions in uranium + graphite + diluent salt (°Li + ‘Ii + Be + F) _nfx — 1 BRma.x_ f* - productions — absorption (uranium + graphite + diluent salt) absorptions in uranium in core The base salt composition, 66/29/5 mole % LiF/Bng/ThF4, was an arbitrary choice for these calculations and probably corresponds to a lower limit of useful thorium concentrations. However, supporting calculations in- dicated very little sensitivity in the shape of the curves in Fig. 3.2 59 to the base salt composition, if the thorium content is varied between 5 and 13 mole %. The mean chord length of 2 cm was chosen to correspond roughly to a minimum of resonance self-shielding in the core thorium, that is, to small salt channels. From Fig. 3.2, as the carbon-to-uranium ratio is decreased, a larger fraction of the neutron spectrum lies in the eplthermal energy range, and an optimum C/U ratio occurs when the loss in 7 balances the reduction in thermal parasitic neutron absorption in graph- ite, lithium, beryllium, and fluorine. The optimum ratio lies in the range of 2500 to 4500, and in this range the production of excess neu- trons, available for absorption in thorium, is insensitive to the C/U ratio., The optimum ratio is also insensitive to the volume fraction of salt in the core, For the optimum core moderation conditions given above, further cal- culations were made of the distribution of neutron absorptions in critical reactors with unmoderated radial blankets of molten salt. The resonance self-shielding of the thorium contained in the blanket corresponded to that of the homogeneous ummoderated fuel salt mixture. The results of these calculations are summarized in Fig. 3.3. The influence of the C/U ratio and blanket thickness on both the breeding ratio and fraction of thorium absorptions in the blanket is indicated in this figure. The thorium-to-uranium ratios required for criticality ranged between 14 and 20 for 20 vol % salt in the core (Fig. 3.3a) and between 33 and 46 for 10 vol % salt in the core (Fig. 3.3b). Figure 3.3 indicates that, while some improvement in breeding ratio can be obtained by use of the unmod- erated radial blanket, the improvement is small for thicknesses greater than 1.5 ft. ORNL-DWG 65-41{77 - 1.2 l I 1 ] | T T T Y T 0.24 2 (b)40VOL%>SAUWNCORE‘ o (g) 20 VOL % SALT IN CORE b ——— + —— :::-m._zn = i | S — 1.5 "IJ o RADIAL BLANKET ~ 0 x R = THICKNESS (ft)} | ' 0.20 ) — [ & £ — \Q 2.0 ) S ™10 @ & 1.0 2 0] 0.16 2 o) P2 CONDITIONS Y @ / ‘ CORE: 15 ft DIAM RIGHT GYLINDER & N PR BLANKET: UNMODERATED SALT @ ) // < S 0.3 . & 0142 = = ~ |~ 2 a x s - : 2 aol - m 1.0 p z ; 0.8 e —— / //1/ 1.5 - 0.08 .c:) 2 x ~ GONDITIONS =====;—” > Lol g g CORE: 1t DIAM RIGHT CYLINDER " i I BLANKET:UNMODERATEDSAET’ — g:: 0.7 | | l | l 40'04 103 2 5 10* 2x10° 5 10 CARBON/URANIUM ATOMIC RATIO Fig., 3.3. Effect of C/U Ratio and Blanket Thickness on Breeding Ratio. quired for the blanket. 60 To determine whether any real gain is obtained from a radial salt blanket, it is necessary to consider the additional fuel inventory re- For this purpose, the ratio of the breeding gain (BR-1) to the total system uranium inventory is plotted in Fig. 3.4, Two limiting cases are compared: that of a two-region reactor with an unmoderated salt blanket and that of a single-region reactor with the dlameter of the core extended by an amount equal to the blanket thick- ness of the two-region systen. Practical external cooling system inven- tories were assumed to be in the range of 100 to 400 ft3 of salt. The total system inventory is that of the core, radial blanket, and external cooling system. For the range of blanket thicknesses studied, it may be concluded that the increase in breeding ratio obtained by use of the un- moderated blanket 1s not enough to compensate for the required additional uranium inventory. Thus there appears to be little or no advantage in the use of an unmoderated radial blanket. o & ORNL-DWG 65-4478 [ l W (@) 20 vol % SALT IN CJ:ORE | l l (£) 10 vol % SALT IN CORE ! r | | -] 1 ==-=BLANKET COMPOSITION SAME ‘ AS CORE COMPOSITION —UNMODERATED SALT BLANKET 3 " EXTERNAL COOLING T EXTERNAL COOLING — | =3 SYSTEM INVENTORY SYSTEM INVENTORY N 2 ft3 E 0.44 i )'__ 100 — __(____) _____ 100 '9 "w - z ” et w Q4 L > 7 = 200\ T T ] 200 v g = = 040 2 o e > / 400 v - [ 400 ? o.08 /_’_::\ e 1 E ,f’, gg% i 0.06 //——-\ 400 :—\Q\ioo 1 CORE: #t DIAM Nizoo « 0.04 RIGHT CYLINDER —— CORE: {5t DIAM —400 m RIGHT CYLINDER 1.0 1.5 2.0 1.0 1.5 2.0 RADIAL BLANKET THICKNESS (ft) Fig., 3.4. Effect of Radial Blan- ket Thickness and Composition on Fuel Yield. References S. J. Ditto, Control of MSRE Between 1 Megawatt and 10 Megawatts, MSR-63-23 (June 4, 1963) (internal use only). J. T. Thomas et al., Critical Mass Studies — Part XIII, Borosilicate Glass Raschig Rings in Agqueous Ur anyl Nitrate Solutions, ORNL-TM-499 (Feb. 6, 1963]). Critical Dimensions of Systems Containing 23SU‘,, 239Pu, and 233U, TID- 7028, pp. 4647, G. D. Joanou and J. S. Dudek, GAM-II: A B3 Code for the Calculation of Slowing Down Spectrum and Associated Multigroup Constants, GA-4265 (1963]. J. Replogle, MODRIC : A One-Dimensional Neutron Diffusion Code for the IBM-7090, K-1520 (Sept. 6, 1962). Part 2. MATERIALS STUDIES 4. METALLURGY Reaction of INOR-8 with Impure Nitrogen The reactor cell atmosphere in the MSRE is designed to be nitrogen containing less than 5% oxygen. This atmosphere of low oXygen content serves to eliminate the hazards of explosion that could result from a leakage of o0il from the lubricating system of the fuel-salt circulating pump. Since INOR-8, the reactor structural material, will be exposed to this environment at temperatures up to 1300°F, tests were made to determine the compatibility of INOR-8 with 0,-N, mixtures. The composition of INOR-8 was formulated to resist oxidation and molten-salt corrosion, but not nitridation. Furthermore, the possibility existed that the proposed atmosphere could accelerate the oxidation rate of the alloy several orders in magnitude due to "oxygen starvation." Briefly, this surface phenomenon arises from the formation of volatile and corrosive molybdenum oxides rather than the adherent films based cn chromium oxide. It is expected that the prcposed service temperature will be too low for either of these reactions to cccur. The experiments described below verify this coneclusion. INOR-8 sheet specimens 0.010 x 1-3/8 x 2 in. were heated for periods up to 700 hr in a predetermined 0,-N, mixture that ranged from 0.03 to 5.6% 0. The gas composition was controlled by metering air and tank nitrogen into the reaction chamber and was analyzed with a gas chromato- graph. The system pressure was held at 300 mm Hg by a bleed-off arrange- ment, while the reaction rate was measured with an automatic recording balance. The reaction rate of the alloy with nitrogen ccontalning three con- centrations of oxygen at 1300°F and two concentrations of oxygen at 1400°F 1is shown in Fig 4.1. These data show that the reaction rate at 1300 and 1400°F increased as the oxygen content of the nitrogen was increased. The characteristic sharp decrease in the reaction rates after a short exposure time indicated that the reaction product formed an effective barrier between the gases and the alloy. The uniform small rate of weight increase of the specimens subsequent to the initial reac- tion period is the normal behavior of a corrosion-resistant alloy. The above results support our original conclusion regarding resist- ance of the alloy to the proposed atmosphere., Although the surface films have not been identified, they are expected to be oxides. Based upon our correlations between weight gains vs depth of oxidation, the estimated extent of reaction in the worst case amounts to an oxidation depth of about 0,05 mil in 700 hr. The reaction product found was a tenacious film with hints of interference colors ranging from green to dark brown. Such surface features suggest that the film is very thin, 63 64 ORNL-DWG €5-4479 20 I ._'__—-.._ 1400 °F, 5.6 % 02 "—’.‘_———___._-—-—" ® . ,/')/M = 1400 °F, 4.2% O, 16 P PUBES e me P figfili 1300 °F, 4.6% O, — 4:_.& NE Ad “‘-‘,:313 A % / I.( °© o 12 7 s |/ = N 1300 °F, 0.97% O, g o} A S o M%Mfi—-fi—{_ 7\ a E Z M 1300 °F, 0.03% O, = ay [ ] r g .’-..—/.-——-— [ ] -..’ /.I 4 V o 0 1 2 3 4 5 6 7 TIME {10C hr) Fig. 4.1. Reaction Rate of INOR-8 with N, Containing 0,. Alteration of MSRE Heat Exchanger Tube Bundle During preinstallation flow testing of the MSRE primary heat ex- changer, the pressure drop through the shell side was found to be sig- nificantly greater than was desired.l! To correct this situation, it was decided to increase the free space around the shell-side inlet and outlet nozzles by removing the four outer U-tubes and sealing the eight tube stubs remaining by plug welding. This welding operation is described below. Due to the low pressure differential (50 psi) between the tube and shell sides of the heat exchanger, the plug was not required to have great strength; however, a leak-tight joint was essential. With these facts in mind and taking into account the close proximity of the adjacent U-tubes, an edge~type weld preparation was made on the plugs as shown in Fig. 4.2, Each plug was positioned flush with the tube end. We believed that a qualified welder with a small gas—tungsten-arc torch could more easily make a high-quality weld on this type of joint than on any other. Since the edge-type weld design also readily allows the tube end to be joilned to a member of similar thickness, adequate weld penetration is more easily assured. The plugs were tapered and machined for each individual tube to provide a slight interference fit of 0.0000 to 0.00002 in. With such a tight fit, none of 30 quelification samples exhibited any indication of root cracking when examined metallographi- cally. 65 Fig. 4.2. BSchematic of Tube Plugging Design. SN ORNL -DWG 6411563 WELD /TUBE WALL S With the joint and plug designs fixed, the welding parameters of arc current, travel speed, and electrode orientation were varied to determine the optimum combination. In addition, a copper chill bar was fitted to the tubes at a distance of about 0.070 in. from the top of the joint to decrease the thermal gradients between the tube and plug. The conditions giving the best penetration and weld configuration were: Electrode material Electrode diameter Inert gas Gas flow rate Welding amperage Welding speed Electrode position Tungsten + 2% thoria 1/16 in. Argon (99,995% purity) 17 cfth (34 £ 1) amp 40 to 55 sec/joint Outer edge of tube The welder was qualified on joints in a setup which simulated the configuration of the actual heat exchanger. He was also qualified for gas—tungsten-arc welding of INOR-8. After preparation and thorough cleaning of the tube stubs and plugs, the plugs were pressed into the tube stubs, the welding sequence was followed, and the Jjoints were visually inspected. 67 Visual examination revealed two welds that had areas of questionable penetration. After demonstrating repair operations on sample welds, the welder rewelded the two areas at full amperage to assure sufficient pene- tration. All welds were then inspected by dye-penetrant and radiographic methods. No imperfections were revealed., A photograph of four of the sealed tubes is presented in Fig. 4.3. A more detailed description of this work has been reported.2 INOR-8 Welding Studies Mechanical Properties of INOR-8 Weld Metal A vital part of the overall welding study on INOR-8 has been the determination of the mechanical properties of welds at elevated tempera- tures. Testing was completed using transverse samples machined from l-in.-thick welds that were made under highly restrained conditions. The manual gas—tungsten-arc welding process used for fabricating MSRE components was also used to fabricate the test weldments; the filler metal matched the composition of the base metal. The heats of INOR-8 material used in this study were taken from the stock purchased for construction of the MSRE. This material consisted of 16 heats which had exhibited satisfactory weldability in the tests required by the pur- chase specification. The transverse weld specimens containing weld metal, heat-affected zones, and base metal were tested at elevated temperatures in tension and creep tests. Tensile tests were performed at 200°F intervals between 600 and 1800°F, and creep testing was done at 1100, 1300, and 1500°F. The specimens were tested in both the as-welded and stress-relieved conditions with stress relieving being performed in both argon and hydrogen atmospheres for 2 hr at 1600°F. Samples were also machined from the as-received base metal for the determination of elevated- temperature hot ductility after being subjected to simulated heat- affected-zone thermal cycles. Tensile testing of these transverse specimens of INOR-8 at room and elevated temperatures showed that this material possessed a good combination of strength and ductility. The data obtained from these tests are shown in the curves of Fig. 4.4. Stress relieving produced the general effect of lowering the yield point and raising the minimum ductility, with this minimum generally occurring at 1600°F. 1In the case of the short-time tensile properties, argon was found to be a better stress-relieving atmosphere than hydrogen., Yield-point reduction was observed to be less for samples stress relieved in argon, and those samples exhibited somewhat better overall ductility. The creep studies of as-welded samples, summarized in Table 4,1, indicated that reasonable stress-rupture properties, comparable with those of the base metal,? were obtained at all testing temperatures. 120 @® o 60 STRENGTH (psi x1073) 20 ELONGATION (%) 68 ORNL.-OWG 65- 4180 AS WELDED-YIELD STRENGTH STRESS RELIEVED -HYDROGEN - YIELD STRENGTH AS WELDED - ELONGATION AND ULTIMATE TENSILE STRENGTH STRESS RELIEVED-HYDROGEN-ELONGATION AND ULTIMATE TENSILE STRENGTH STRESS RELIEVED-ARGON - YIELD STRENGTH STRESS RELIEVED-ARGON-ELONGATION AND ULTIMATE TENSILE STRENGTH . - - I -- - 400 800 1200 1600 2000 TEST TEMPERATURE (°F) Fig. 4. 200 4‘ 400 600 800 1000 TEST TEMPERATURE (°F) for Transverse Specimens of INOR-8 Welds. 1200 1600 1800 Results of Room- and Elevated-Temperature Tensile Tests Table 4.1. Results of Elevated-Temperature Creep Tests on INOR-8 Transverse Weld Specimens Average Time to Rupture (hr) Average Elongation (%) Test Applied Temperature Stress As Stress Stress As Stress Stress (°F) (°c) (psi) Relieved, Relieved, Relieved, Relieved, Weldea Hydrogen Argon Welded Hydrogen Argon 1100 594 74,000 1.3 1.7 14,1 13.0 1100 594 54,000 197.8 188.3 2.5 8.2 1100 594 49,000 308.4 570.5 2.2 5.3 1300 704 45,000 3.7 6.4 5.5 3.9 8.2 5.4 1300 704 24,000 158.4 337.8 185.4 3.7 8.8 7otk 1300 704 20, 000 472.3 936.7 4522 406 10.8 3.7 1500 816 22,000 12,7 12.1 16.9 20.9 1500 816 13,000 172.1 117.5 14,4 2.8 1500 816 10,000 446,9 314.5 8.3 8.1 Sstress relieved at 1600°F, 2 hr in atmosphere specified. 69 70 ORNL-DWG 65- 4181 80 | | 80 0.4 . .\I AN 80 60 ¢ 0.3 \\\\\ TESTED ON 40 5‘40 e 0.2 - HEATING 2 N i & 20 i = 20 \ < Od {\4. o * a \ et \\ [ : N, |z 5 N ;:’ O ' — o O E O 1 'Y S z 80 & 80 » 0.4 A o o 3 - [ W O k = @] 2 60 / | 4 60 F 03 o w L J T Z / \ e | TESTED ON 30 | 40 \\ 0.2 > COOLING FROM q\\“ 2300°F T ] '/i& 20 20 \\ 0.4 \\ N\ : - | "\. 0 0 0 — 1800 2000 2200 2400 180C 2000 2200 2400 1800 2000 220 2400 TEST TEMPERATURE (°F) Fig. 4.5. Results of Hot-Ductility Tests on MSRE Reactor-Grade INOR-8, Showing the Nil-Ductility Temperature for the Material To Be 2300°F. Stress relieving, using either hydrogen or argon atmospheres, created a significant effect on the creep properties of samples tested at 1300°F. Data collected at this temperature revealed that stress relieving resulted in a factor of 2 improvement in the time to rupture and total strain properties., However, the creep-test results on stress-relieved samples at 1100 and 1500°F showed little or no improvement over as-welded properties. It was generally observed that the minimum creep rate of these composite specimens was slightly lower then that found for the base metal at all test temperatures and stresses. Hot-ductility experiments using synthetic heat-affected-zone speci- mens revealed the elevated-temperature nil-ductility point for this material to be 2300°F. A summary of the data from these experiments is shown in Fig. 4.5. It can be seen in these curves that the mechanical properties recovered reasonably after heating to the nil-ductility temperature, although some damage occurred as a result of the welding thermal cycle. Metallographic analysis of the microstructures of alloy that had been thermally cycled through these temperature ranges revealed some grain-boundary liquation. It appears from this study thet the room- and elevated-temperature mechanical properties of welds in INOR-8 compare favorably with base-metal properties. This behavior indicates that the weldments made of this reactor grade of INOR-8 for use in the MSRE should be of high quality. 71 INOR-8 Welding Microstructure Study In the early welding studies on certain experimental heats of INOR-8, a large number of microfissures were found in the heat-affected zones of highly restrained welds.*™ At that time, it was observed that cracking invarisbly occurred in a eutectic-like structure existing in the grain boundaries. This structure was found to be formed during the welding thermal cycle and originated from a stringer-type phase existing in the INOR-8 base metal. Since the thermal damage originated in a brittle structure in the grain boundaries, the morphology of the heat-affected zone was of obvious interest, so a rigorous microstructural analysis was begun on the grain- boundary phase. The material used in this study was taken from those early experimental heats of INOR-8 which were found to contain large quantities of the suspect structure in heat-affected areas. A micro- structure similar to that found in the heat-affected zone of a weld was produced in the metal by treating specimens in a Gleeble hot- ductility machine. Data obtained from these studies were supplemented by information obtained from hot-stage metallography, from hot-ductility tests previously done for these INOR-8 heats, and from quantitative analysis of the grain-boundary structure using microprobe techniques. A view of the grain-boundary structure in question is shown in Fig. 4.6, vhich was produced by a welding thermal cycle with a peak temperature of 2300°F in INOR-8 heat No. SP-19. Currently, only INCHES 8 1000% 3 Fig. 4.6. Simulated Heat-Affected-Zone Microstructure in INOR-8 Heat No. SP-19 Using a Welding Thermal Cycle with a Peak Temperature of 2300°F. Note the eutectic-type structure in the grain boundaries. Etchant: H3PO,, HpO electrolytic. Oblique lighting. 1000X. 72 preliminary data are available from the microprobe analysis investiga- tions. These studies were performed on the sample shown in Fig. 4.6, and the results are presented in Table 4.2, As shown in these data, the eutectic-type structure contains considerably more aluminum and silicon than does the matrix. A trace across this eutectic-type structure was made and is presented in Fig. 4.7. This trace shows the marked differences in concentrations of these two elements in the matrix and grain-boundary regions of the heat-affected zone. These studies are continuing, and further analyses will be made of the data and microstructures. Table 4,2. Preliminary Results of Microprobe Spectral Analysis of INOR-8 from Heat No. SP-19 Approximate Composition™ (wt %) Nominal i ic- Composition of Froment strasture erucours, . Heat SP-19 (w5 %) Ni 68 68 69.62 Mo 18 12 16.10 Cr 7.8 6.1 7 .04 Fe 4.8 4.3 4 .60 Mn 0.7 0.4 0.52 Al Not detected 3.3 0.06 Ti Not detected Not detected 0.02 C Not analyzed Not analyzed 0.024 Co 0.5 0.5 Not analyzed Si 0.6 2.5 0.16 aThese are semiquantitative analyses with an approximate correc- tion for absorption effects. P 73 ORNL-DWG 65-4182 2.5 T I 1 l 1 MATRIX EUTECTIC- TYPE MATRIX STRUCTURE STRUCTURE STRUCTURE - A i3 - 1O~ —e—Si B2 - —O— Al ._ [ 1% 3 N n | bl = 10 NG \ ’ . DISTANCE TRAVERSED (u) Fig. 4.7. Aluminum and Silicon Concentration Profiles Across the Eutectic-Type Structure. Graphite-to-Metal Joining Development Transition Joints The joining of graphite to structural metals such as INOR-8 is of interest for use in advanced molten-salt reactor concepts as well as for general high-temperature engineering systems that might require the use of graphite. The basic obstacle encountered in attempting such a joint is the very large difference between the thermal expansion coef- ficients of graphite and many metals. Due to this difference, a brazed joint of graphite to INOR-8 cracks upon cooling from the brazing tempera- ture. One possible method of circumventing this problem is to introduce one or more materials with expansion coefficients intermediate between those of the graphite and INOR-8 parts. In order to determine optimum techniques for making such a graphite- to-INOR-8 transition joint, INOR-8-to-dissimilar-metal joints and graphite-to-metal Jjoints were brazed with various fluoride-resistant brazing alloys, and T-joints of these combinations were examined metal- lographically. The results from the tests of metal-to-metal joints are listed in Table 4.3. T4 Table 4.3. Metallographic Observations of Dissimilar-Metal Joints Metal Brazing . . . Combination Alloy Metallographic Observations of Joint W-Nb NB-50% Much transverse cracking W-Nb Au-Ni Some transverse cracking in phase next to niobium W—INOR-8 NB-50 Some transverse cracking W—INOR-8 Au-Ni Sound joint W—INOR-8 Cu Sound joint Mo-Nb NB-50 Much transverse cracking and joint separation Mo-Nb Au-Ni Transverse cracking in phase next to niobium Mo-Nb Cu Complete lack of bonding Mo—-INOR-8 NB-50 Complete separation of joint Mo-INOR-8 Au-Ni Sound Jjoint Mo—INOR-8 Cu Sound joint — incompletely brazed “NB-50 — Nominal composition Ni—13 Cr-10 P (wt %). Pru-Ni — 82 Au-18 Ni (wt 4). The most significant observation from Table 4.3 is that metals with relatively wide differences in coefficients of thermal expansion exhibited sound joints when brazed with a ductile alloy (gold-nickel or copper). A second observation is that the tungsten-niobium and molybdenum-niobium Jjoints brazed with gold-nickel cracked in a brittle phase next to the niobium. The other joints brazed with gold-nickel were sound. The results from the tests of graphite-to-metal joints are listed in Table 4.4. In these joints, the ductility of the alloy seemed to make little difference, and the limiting factor was the difference between the coefficients of expansion of the graphite and the metal. Thus, the molybdenum-to-graphite and tungsten-to-graphite joints showed no transverse cracking, while the metals with the higher coefficients, niobium and tantalum, showed considerable cracking across the brazed Jjoints. As a result of this study, several possibilities for joining graphite to INOR-8 appear to be of interest. They are listed in Table 4,5, and work is currently in progress with those compositions. 75 Table 4.4. Metallographic Observations of Graphite-to-Metal Brazed Jolnts Metal Brazing . . . Member Alloy Metallographic QObservations cof Joint Nb ANM-5" Much transverse cracking Nb ANM—16b Much transverse cracking Ta ANM-5 Much transverse cracking Ta ANM-16 Much transverse cracking Mo ANM-5 Sound joint Mo ANM-16 Sound joint W ANM-16 Sound joint — occasional cracks in carbides dispersed in braze ®ANM-5 — 70 Au—20 Ni—10 Mo (wt %) (very ductile). PANM-16 — 35 Auw-35 Ni~30 Mo (wt %) (slightly ductile). Table 4.5. Promising Materisl Combinations for Graphite-to-INOR-8 Transition Joints Intermediate Braze Material Braze TNOR-8 Graphite ANM-5 or Mo Au-Ni% INOR-8 ANM-16 or Cu Graphite ANM-5 or W Au-Ni INOR-8 ANM-16 or Cu pu-Ni — 82 Au-18 Ni (vt %). Brazing Alloy Development The most successful alloy that has been developed for brazing graphite and that is also compatible with molten salts is the 35 Au-35 Ni—30 Mo (wt %) alloy. In the gold-nickel-molybdenum alloy, a compatible, carbide- forming element (molybdenum) is added to a corrosion-resistant and reason- ably low-melting alloy system (gold-nickel). Since the use of a gold- containing alloy in a high-flux region of a reactor might be inadvisable, 76 8 study was begun to develop a gold-free graphite brazing alloy for this application. The palladium-nickel system was found to be similar to the gold-nickel system and was selected for this study. This noble-metal alloy is expected to be corrosion resistant to fused fluorides. In addition, palladium has the added advantage of a lower thermal-neutron cross section than gold (8 barns vs 99 barns). Also, the daughter product of palladium would be silver, which is more desirable in the molten-salt system than mercury, the daughter product of gold. Evaluation of MSRE Graphite Oxygen Contamination of the MSRE Core Bars and Lattice Bars There is approximately 69 ft> of graphite in the MSRE, of which the vertical core bars constitute about 98% and the supporting horizontal lattice bars the remaining 2%. The lattice bars were fabricated with higher permeability than the core bars in order to secure more structural integrity by eliminating or decreasing the number of cracks. To aid in this, the bars were also halved longitudinally and given the final processing operations as nominally 2-1/4 X 1-1/2 x 60 in. bars. After their final graphitization, the bars were machined to the basic dimensions required for the lattice bar shapes, 1 X 1-5/8 in. cross section and 57 in. long. The standard shape of the specimen used in the oxygen determination is a right circular cylinder 1-1/4 in, in diemeter and 1 in. long, and the specimens from the lattice bars were machined in this shape. The specimens from the core bars were 1-1/2-in.-long quarter sections cut from machined bars, as shown in Fig. 4.8. A specimen of this shape was used so that the volume would be approximately equal to that of a standard cylinder, and the composition would be representative of that of the cross section of the core bar. Geometry has not been found significant in the outgassing results.S8 ORNL -DWG 65-4183 Fig. 4.8. Cross Section of Typical Core Bar. {a) Quarter section 1-1/2 in. long removed for oxygen contamination determination. 77 Table 4.6. Results of Qutgassing Tests on Specimens from Graphite Bars of the Core and Lattice Bars of the MSRE Superscript numbers in parentheses indicate the number of values averaged Bar Bulk Volume of CO (cm?® STP Type N Density per 100 cm? O« (g/cm?) of graphite) Core 23 1.860%) 3.103) Core 788 1.86¢%) 2.20?) Core 1148 1.85(4) 3.0(3) Lattice 1195 1.86%) 11.6() Lattice 1379 1.8,¢%) 14,91 Lattice 1559 1.g7(1) 7.2() As required by the specification,9 the tenaciously held oxygen was determined by placing each of the specimens in a closed system, evacuating the system to T10™ torr at room temperature, and then measuring the STP volume of carbon monoxide evolved from the graphite at 1800°C (3270°F). The results of these determinations are summarized in Table 4.6. It is of interest to note that the bulk densities of the lattice bars were essentlally the same as those of the core bars even though the lattice bars are supposed to be more permeable. The results of the oxygen determinations indicate that both the core and lattice bars have oxygen contaminations well below that permitted for the MSRE graphite. The reason for the lattice bars showing a higher oxygen contamination than the core bars was not determined; however, one would first suspect differences in actual exposed surface area. For this reason, the surface areas of three specimens from each of the outgassed specimens from the oxygen determinations were measured by the BET method by Analytical Chemistry. Specimens from both the core and lattice bars averaged 0.7 m?/g in a range from 0.6 to 0.9 m?/g. These areas have no relationship to the quantities of oxygen contamination found in the specimens. Accessible Voids Content of MSRE Core Graphite The molten fluoride fuel does not wet the graphite under any standard operating condition expected for the MSRE. However, if it should wet for some unknown reason, the nature of the accessible voids becomes important. Some initial studies with wetting fluids, xylene and molten sulfur, sug- gest that such a problem would be somewhat less severe than originally 78 supposed. Measurements of the accessible volds with xylene were made by determining the quantity of xylene picked up by specimens that were initially evacuated, flooded with xylene, and allowed to set for 1 hr under atmospheric pressure. Since the xylene wets the graphite, it should have penetrated all the accessible pores. The measurements were determined as a function of four different sizes. After each measurement with xylene, the Xylene was removed by vacuum distillation and each graphite specimen was reduced in size for the next measurement. The first measurements were made on the 1.58-in.- long transverse shapes taken from three different machined bars. Each of these full sections was progressively subdivided into specimens having the following shapes and nominal dimensions in inches: One cube, 1.56 X 1.56 X 1.56 One cylinder, 1.50 diam x 1.50 long Five rods, each 0.375 diam x 1.000 long The cylinders and rods were all machined with their axes parallel with the extrusion direction of the original bar. The five rods were taken one from the center line of the original bar and the other four symmetrically from the volume around it. The accessible voids (porosity) and bulk densities that were measured after each stage of machining are summarized in Table 4.7. The bulk density wvalues in general indicate that the graphite bars have uniform properties transversely. Therefore, one does not have to resort to special machining to maintain desirable properties within a bar; however, the accessible voids vary some from bar to bar, and also differ from the overall average of 4.0% reported previously.!® The accessible voids remained low and about the same for the larger specimens for each bar. The higher values obtained for the 0.375-in.- diam specimens are consistent with the picture that the wetting liquid fills essentially all the 15% void volume at the surface, but that the voids interconnect for only short distances into the interior of a completely sound piece. The study was extended to determine the nature and configuration of the accessible voids. Molten sulfur at 150°C was used to impregnate the accessible volds by a technique adapted from work reported by Nelson.l! The specimens were evacuated at room temperature, heated to 150°C, submerged with molten sulfur, and an overpressure of 150 psig was applied. The conditions were maintained for 15 hr at 150°C. The sulfur wets the graphite but does not react with it; therefore, the 150 psig overpressure probably was not necessary to ensure that all the accessible voids would be filled. The excess sulfur was drained away, and the impregnated specimens were cooled to room temperature. The results are summarized in Table 4.8. The volumes filled by sulfur are consistent with xylene measurements made on specimens of this general size and shape. The grade AGOT graphite was present as a control. 79 Table 4.7. Accessible Voids and Bulk Densities of Specimens of Grade CGB Graphite as Functions of Specimen Shapes and Sizes Type of Specimen Bar Number and Dimensions in Inches 23 788 1148 a Accessible Porosity (%) Full sectionb 4,9 1.7 2.6 Cube (1.56 X 1.56 X 1.56) 3.3 1.5 2.2 Cylinder (1.50 diam X 1.50) LA 1.6 2.5 (10) (10) Rod (0.375 x 1.000) 10.0 6.2 Bulk Density (g/cm3)c Tuil sectiofib 1.863 1.861 1.850 Cube (1.56 X 1.56 x 1.56) 1.863 1.962 1.951 Cylinder (1.50 diam x 1.50) 1.862 1.849(?) 1.851 Rod (0.375 X 1.000) 1.8600%) 1.8610%) *A11 values are averages of two determinations with the exception of those with superscripts of (10); the values with these superscripts are averages of duplicate determinations on five specimens. bA section 1.58 in. long cut transversely from a machined MSRE graphite core bar. CEach value is a single determination except for those with the superscripts of (5), which are averages of determinations made on five specimens. Table 4.8. Results of Tmpregnating the Accessible Voids of Graphite with Molten Sulfur at 150°C ) Bulk Specimen Percent of Bulk Volume Graphite Density Dimensions of Graphite Grade (8/Cm3) (in.) - Filled with Sulfur a AGOT 1.68 0.500 diam x 1.500 21.3(3) CGB 1.83 0.500 diam X 1,500 9.4(3) CGB 1.87 b 5.401) aSuperscript numbers in parentheses indicate the number of values averaged. bThis was a 2-in.-long transverse section cut from a machined MSRE graphite core bar; nominally, the specimen was 2 X 2 X 2 in, Fig. 4.9. Rediographs of 0.025-in.-thick Sections Machined from Impregnated MSRE Graphite Showing the Penetration by (&) a Wetting Fluid, Molten Sulfur, and (b) a Nonwetting Fluid, a Molten Fluoride Salt. The vhite phases are the impregnating fluids. Radiographic examinations of the specimens indicated that the 0.500- in.-diam X 1.500-in.-long specimens were completely penetrated by sulfur with the exception of a few small zones in the grade CGB graphite. However, the 2-in.-long transverse section from the machined core bar was not completely penetrated. The major penetrations were limited to about l/é—in. penetrations below the exposed surfaces whether they were external or planes of cracks. This is shown in detail in Fig. 4.9a, which is a reproduction of a radiograph of a transverse section 0.025 in, thick cut from the middle of the impregnated specimen. This suggests that a wetting fluid would not grossly penetrate grade CGB graphite, but it would penetrate much deeper than the nonwetting fluoride fuel, which is shown for comparison in Fig. 4.9b. Metallographic Examination of Bayonet Tube in Drain Tank Cooler Test Section 2 (Component Development) of this report describes a failure of an INOR-8 bayonet tube in the Drain Tank Cooler Test. The failure occurred close to some welded spacers in the lower part of the tube. The tube had been thermally shocked more than 2500 times from 1200 to 200°F with cooling water. Tests with Inconel tubes in which the tubes, in a somewhat more constrained arrangement, were subjected to a large number of similar thermal cycles resulted in distortion of the tubes, complete cracking of tube walls at spacers and thermocouples, and general transverse cracking to a depth of about 0.020 in. on both the inner and outer surfaces. 81 Y-62161 r 0.035 INCHES T> 100X = Fig. 4.10. Crack Through Transverse Section of INOR-8 Tube from Drain Tank Cooler Test. Metallographic examination of the INOR-8 tube revealed one crack (Fig. 4.10) that extended across the tube wall., This crack occurred at a position adjacent to a tack weld in the region of maximum distortion. Several transverse cracks approximately 0.010 in. in depth were also found in this region. However, the general transverse cracking observed in the Inconel tubing was not present, and areas away from the distorted region were completely free of cracks. 82 An oxidized layer that penetrated to a depth of several mils was observed throughout the tubes, including the crack surfaces. The unusual thickness of this layer indicated that oxidation had been accelerated by the thermal cycling. The cycling probably caused the normally protective oxide layer to crack and allow the oxidation to penetrate. The oxidation in the crack indicated that it had existed for some time prior to shutdown of the test. We concluded from the examination that the crack and accompanying distortion had probably been caused by a concentration of stresses at a flaw or stress riser at the edge of the weld. Mechanical Properties of Irradiated INOR-8 A study'? of the effects of irradiation on the tensile properties of INOR-8 at high temperature showed that the stress-strain relationship was not affected by irradiation but that the high-temperature ductility was reduced. This reduction became more severe as the temperature was ralsed. Ductility was also observed to decrease as the strain rate was reduced from 0.2 to 0.002 in./min. Although the tensile ductilities measured at the MSRE operating temperatures were still adequate to accommodate any tensile-type loading expected in the MSRE, the trend of decreasing ductility as a function of strain rate suggested that additional information was needed on the effects of irradiation on creep properties, and in particular on creep ductility. To determine this information, a joint study was started by investigators at GE MPO and at ORNL. This study is being sponsored by the Fuels and Materials Branch, AEC, In this study, INOR-8 creep specimens from heats used in the MSRE will be irradiated at approximately 1200°F to dose levels ranging from 10'® to 2 x 10°! nvt. Postirradiation creep tests will be run at 1100, 1300, and 1500°F for times ranging to 1000 hr. In-reactor creep tests are also planned, and an attempt will be made to correlate these data with the postirradiation creep data. A correlation also will be attempted between the effects of irradiation on creep properties and tensile Properties. A creep test experiment (ORR-135) was designed and built for ir- radiation in the ORR poolside facility, P-5. This test is scheduled to be installed in April and is expected to operate for two reactor cycles. The creep experiment rig is shown in Fig. 4.11. It contains five tensile creep bars made from heat 5081, INOR-8. The creep bars will be kept at temperature by wire-wound furnaces that surround each bar, and loads will be applied through a bellows and counter-level arrangement. Creep testing will be performed during irradiation, and simultane- ously five similar INOR-8 specimens will be irradiated for postirradia- tion creep tests by GE MPO. FroTo w7 Fig. 4.11. Experiment for Creep Testing INOR-8 in ORR. MSRE Material Surveillance Testing Surveillance specimens were fabricated of INOR-8 and type CGB graphite for placement in the central position of the MSRE core. These specimens will be used to survey the effects of reactor operations on the material from which the reactor and moderator were constructed. INOR-8 specimens were made from heats 5081 and 5065 in the shape of miniature tensile bars approximately 2 in. long and with a gage length 1.0 in. long and 0.125 in. in dieameter. The tensile bars were welded end to end to form two rods approximately 64 in. long. Included were four tensile bars made with transverse sections of welds in the gage length. The graphite specimens were made from sections of MSRE moderator bars selected as being crack free by radiographic methods. These speci- mens were machined into rectangular bars of various dimensions. Testing will be conducted to get data for both the longitudinal and transverse directions of the graphite. In the final assembly, the graphite specimens will be arranged into a long subassembly having a rectangular cross section of 0.47 X 0.66 in. Three of these graphite subassemblies will be placed in an arrangment shown in Fig. 4.12, and these will be surrounded by six INOR-8 specimen rods and three INOR-8 tubes containing flux monitor wires. The entire assembly will be held in a 0.200-in.-diem perforated tube. After six months of reactor operation, two INOR-8 specimens, a flux monitor, and one graphite subassembly will be removed from the reactor and examined, and new specimens will replace those removed. The frequency of removel of additional specimens will be established by the results of the first set of specimens. 84 ORNL - DWG 65-4184 GRAPHITE SPECIMEN INOR-8 SPECIMEN FLUX MONITOR 1 | | INCH Fig. 4.12. Surveillance Specimen Assembly. Cross section. The analysis of the INOR-8 specimens will include (1) metallographic examination for structural changes, corrosion effects, and possible layer formations; (2) tensile properties and creep properties, with emphasis on creep ductility; (3) a general check for material integrity and dimensional changes; and (4) chemical analysis for composition changes and fission product deposition. The graphite specimen analysis will include (1) metallographic examination for structural changes and material deposition; (2) radio- graphic examination for salt permeation and possible wetting effects; (3) flexural strength tests; (4) dimensional checks for shrinkage effects; (5) chemical analysis for salt and fission product deposition; (6) a general inspection for integrity; and (7) possibly physical properties such as electrical conductivity, thermal conductivity, and Hall coef- ficient. Control Test for Surveillance Specimen Control specimens, of shape and source identical to the reactor specimens, will be exposed in fuel salt to the reactor thermal history in a control test rig shown in Fig. 4.13. The chamber, containing 85 ORNL-0OWG 65-4185 Hf i !’ = ! A - = - . __..-COOLING AND 1 '} CARRYING CHAMBER “t-—-— FUEL SALT LEVEL i: % i - = SPECIMEN HEATING I ; ‘; : CHAMBER I S TN RN 1| J) CR— Fige. 4.13. Controlled Test Rig for Surveillance Program. 86 specimens submerged in salt, will be heated by .three zones of clamshell heaters. These will be controlled by the signals generated from the inlet and outlet temperature of the reactor through the Logger-Computer. In this manner, the control test specimens should automatically be ex- posed to the approximate temperature profile and major temperature fluctuation experienced by the reactor specimens. Specimens will be removed from the control test rig by pulling them up into the cooling chamber, where they may be isolated and removed. It 1s planned to examine these specimens in the hot-cell area at the same time that the reactor specimens are tested. References 1. MSR Progrem Semiann. Progr. Rept. July 31, 1964, ORNL-3708, pp. 174-77. 2. R. G. Donnelly, Tube Plugging in the MSRE Primary Heat Exchanger, ORNL-TM-1023 (in press). 3. MSR Program Semiann. Progr. Rept. July 31, 1964, ORNL-3708, pp. 348-52. 4. P. Patriarca and G. M. Slaughter, MSR Program Semiann. Progr., Rept. Oct., 31, 1957, ORWL-2431, pp. 18, 21-23. 5. G. M. Slaughter, MSR Program Semiann. Progr. Rept. Jan., 31, 1958, ORNL~2474, pp. 65-71. MSR Program Semiann. Progr. Rept. Feb. 28, 1961, ORNL-3122, pp. 81-86. MSR Program Semiann. Progr. Rept. Aug. 31, 1961, ORNL-3215, pp. 107-9. 8. L. G. Overholser and J. P, Blakely, "The Degassing Behavior of Commercial Graphites," p. 196 in Proceedings of the Fifth Conference on Carbon, Macmillan, New York, 1962. 9. Tentative Specification for Graphite Bar for Nuclear Reactors, MET-RM-1 (May 10, 1961). 10. W. H. Cook, MSR Program Semiann., Progr. Rept. July 31, 1964, ORNL-3708, p. 377. 11. J. B. Nelson, "X-ray Stereo-Microradiography of Carbons,"” pp. 438-55 in Proceedings of the Fifth Conference on Carbon, vol, 1, Macmillan, New York, 1962, 12. W. R. Martin and J. R. Weir, The Effects of Elevated Temperature Irradiation on the Strength and Ductility of the Nickel-Base Alloy, Hastelloy N, ORNL-TM-1005. g7 5. RADIATTION CHEMISTRY Introduction The last report! reviewed the more significant features of irradia- tion experiments MIR-47-1 through MIR-47-5 on the compatibility and sta- bility of MSRE materials in a fissioning environment and briefly described preliminary results from experiment MIR-47-6. The chemical, radiographic, and metallographic examinations of the 47-5 specimens have since been com- pleted, and only a few of the scheduled 47-6 postirradiation examinations remain to be completed. In the early irradiations of the MIR-47 series, the INOR-&-fuel— graphite test assemblies generally survived irradiation without gross damage to the test materials, but evidence was observed of slow fluorine gas evolution from the irradiated fuel that occurred after solidification and cooling to room temperature. The 47-5 test, in which two of the INOR-8& capsules were equipped with gas lines to sample the cover gas during ir- radiation, showedl that fuel radiolysis occurred only during reactor shut- downs when the fuel solidified and cooled to 35°C. Cover gas samples, taken while the salt was molten and fissioning, showed only minute traces of C¥,, which probably originated from F, generated during previous shut- downs. The other indications of material incompatibilities observed in the early runs were thought to be related to the radiolysis of the fuel during reactor shutdowns. Graphite specimens exposed to the capsule cover gas were sometimes badly corroded. Thin deposits of uranium, up to 4 mg per square centimeter of exposed graphite, were observed on submerged graphite specimens. The 47-3 molybdenum specimens were badly corroded. These un- desirable reactions probably occurred during the sudden heating accompany- ing reactor startup after a shutdown period during which fuel radiolysis had produced fluorine and a chemically reduced fuel salt. Such chemical premises as the above explained the remaining results from the 47-5 examinations and guided the design of the 47-6 experiment. Experiment MIR-47-3 The results of the gas analyses, the study of postirradiation fluorine generation, the analyses of the cover gas from the four sealed capsules, and the visual observations of the dissected capsule specimens have been previously reported.2'3 Chemical, metallographic, and other detailed ex- aminations of the INOR-8, graphite, and salt specimens are reported below. Table 5.1 summarizes the capsule designs, loadings, and exposure conditions for the 47-5 experiment. Table 5.1. or details of the capsule designs, see ref. 2 Design and Conditions for ORNL-MIR-47-5 Irradiation Experiment Capsules Capsule position Weight of graphite, g Weight of U (fuel type, see note below) Graphite-salt interface area, Cm Metal-salt interface area, cm® Fission power density, w/em 7% 5.696 1.012(1) 12.5 35 90 3 2 5.586 0.536 0.525(2) 1.025(1) 12.5 1 35 46 50 75 A 1 /" Front Rear 11.288 3.6353 3.6655 0.770(1) 0.021(1) 0.013(2) 27 Fuel- Fuel- impregnated impregnated graphite graphite 7 65 20 5 Fuel type analysis: Fuel . Bulk Constituents (wt %) Corrosion Products (ppm) Type Nominal U Content (mole %) i T o2 T o e S 1 0.7 4,07 12.4 6.56 10.2 29 <5 94 5 2 0.35 2.08 12.3 6.58 10.8 17 < 60 13 88 89 Graghite Analyses. Chemical analyses of capsule 3 and capsule 4 specimens showed 13 and 18 mg of uranium per gram (or 5 and 7 mg of uranium per cm? of exposed area) of graphite respectively. The total weight of other fuel salt cations per gram of capsule 3 graphite was 12 mg. Extensive fuel radiolysis was known to have occurred in those capsules. The graphite from capsule 1 (graphite crucible) contained only 2 to 4 mg of uranium, along with 0.2 mg of other fuel cations and 0.7 mg of iron, molybdenum, and nickel (combined) per gram of sample. A relatively small amount of F> and CF, had been found in this sealed capsule. The small graphite wafer specimen from capsule 2 was not analyzed chemically in order to preserve the sample for other tests. The pre- viously reported weight gain3 suggested uranium deposition on salt pene- tration. Curiously, the fuel-impregnated graphite rods from capsules R and F contained only 2 and 5 mg of uranium per gram, respectively, accord- ing to the chemical analyses. More uranium was expected from the amount of other fuel components found by analysis in the specimens. (The initial loading was 140 and 180 mg of uranium per gram of graphite.) No radiolytic fluorine or CF, was found in the cover gas of these small capsules. Qualitative gamma scans of the graphite specimens (without radio- chemical separations) generally revealed much higher 106Ry concentrations than in irradiated fuel samples (see below). A tendency of ruthenium to accumulate at the graphite-salt interface is thus implied. The major ac- tivity in most graphite specimens was 957p-9°Nb. The *%*Ce activity was curiously low, only 10% of the 927r-2°Mb in many specimens, whereas it was the major activity in bulk salt samples. The 1270y activity in sev- eral specimens from the large graphite crucible in capsule 2 exceeded the 957r-2Nb activity, possibly due to the diffusion of the gaseous precursor 137¥e into the graphite pores. Metallography. Graphite specimens from each capsule were mounted in epoxy resin, polished, and examined metallographically. The samples from capsules 1, R, and F, in which little or no fuel radiclysis had taken place, showed no metallographic evidence of graphite damage (Fig. 5.1). No impregnated salt was visible in the pores of the R and F samples; probably the salt had been dissolved by the ethylene glycol polishing slurry. The graphite from capsules 3 and 4 showed rough surfaces and cracks (Figs. 5.2 and 5.3). A damaged region of capsule 4 graphite is shown in Fig. 5.4. The visually apparent damage to the wafer sample from capsule 2 was evident metallographically only in the rough surface, Fig. 5.5. The metallographs failed to reveal any evidence of the visually apparent shiny films or deposits on the graphite surfaces. Likewise, x-ray-diffraction examination of the few flat graphite surfaces and of scraped graphite samples yielded only graphite lines; however, lines consistent with LiF and metallic Cr, Fe, Ni, and Nb were obtained from scrapings from the cap- sule 4 core. The x-ray examinations are questionable, since some opera- tional malfunctions of the newly installed hot-cell diffractometer had not yet been corrected. Fig. 5.1. Metallograph of Transverse Section of Capsule 1 Graphite Core. Reduced 8%. Fig. 5.2. Metallograph of Transverse Section of Capsule 3 Graphite Core. Reduced 8%. Fig. 5.3. Metallograph of Transverse Section of Capsule 4 Graphite Core. Reduced 8%. Fig. 5.4. Metallograph of Damaged Region of Capsule 4 Graphite Core. Reduced 5%. 2 Fig. 5.5. Metallograph of Transverse Section of Capsule 2 Graphite Wafer. Reduced 8%. Antoradiograp_g. Autoradiographs were taken of graphite specimens from each capsule. Transverse sections of the R and F impregnated rods and of the central graphite core of capsule 1 are shown in Figs. 5.6, 5.7, and 5.8 respectively. Some escape of fuel salt from the surface region of the capsule R sample was indicated. A similar effect was discernible in the capsule F picture, but the individual spots of highest activity were near the surface. The capsule 1 graphite produced an autoradiograph very similar to previous ones from the 47-4 experiment, showing a narrow band of high activity on the salt-exposed surface. The autoradiographs of transverse sections of the capsule 3 and 4 cores (Figs. 5.9 and 5.10) showed spotty activity concentrations and penetrations into cracks in the interior of the graphite. The autora- diograph of the graphite wafer from capsule 2 (Fig. 5.11) showed deep penetration of activity and a region where the surface apparently cracked off. Also the interior region was more generally radiocactive than for the other specimens. The autoradiographs confirm the other indications that uranium and salt penetration of the graphite was more severe in those specimens ex- posed to highly radiolyzed fuel. R-20194 Fig. 5.6. Autoradiograph of Transverse Section of Capsule R Im- pregnated Graphite Rod. R-20190 Fig. 5.8. Autoradiograph of Transverse Section of Capsule 1 Cen- tral Graphite Core. 93 R-20196 Fig. 5.7. Autoradiograph of Transverse Section of Capsule F Impregnated Graphite Rod. R-20198 Fig. 5.9. Autoradiograph of Transverse Section of Capsule 3 Graphite Core. Fig. 5.10. Autoradiograph of Transverse Section of Capsule 4 Graphite Core. Transverse Section of Capsule 2 Fig. 5.11. Autoradiograph of ‘ Graphite Wafer. X Radiography. The x radiographs of thin transverse slices of the graphite cores from capsules R, F, and 1 are given in Figs. 5.12 to 5.14. These confirm with better definition the conclusions drawn from the au- toradiographs. The spotty penetration of heavy elements deep into the core of capsule 1 is of particular interest. 95 R-20427A Fig. 5.12. X Radiograph of Transverse Section of Capsule R Graphite Core. Fig. 5.13. X Radiograph of Transverse Section of Capsule F Graphite Core. Fig. 5.14. X Radiograph of Transverse Section of Capsule 1 Graphite Core. 9% The x radiographs of capsules3 and 4 specimens likewise generally confirm the autoradiographs, Figs. 5.15 and 5.16 respectively. The salt- filled cracks are clearly apparent in Fig. 5.15, and the loading of salt into the surface of the capsule 4 core is shown in Fig. 5.16. The x radio- graph of the graphite wafer section, Fig. 5.17, is interesting for contrast with the autoradiograph, Fig. 5.11. The dense regions of the former do not correspond to the high-activity regions of the latter, contrary to all other observed cases. Thus, this is the first example of transport of R-20411A Fig. 5.15. X Radiograph of Transverse Section of Capsule 3 Graphite Core. Fig. 5.16. X Radiograph of Transverse Section of Capsule 4 Graphite Core. 97 R-20417B Fig. 5.17. X Radiograph of Transverse Section of Capsule 2 Graphite Wafer. uranium after the in-pile exposure to be encountered in the in-pile tests; the implication that UFe was responsible is strong. Summary of 47-5 Graphite Behavior. The chemical and physical examina- tions consistently indicated heavy uranium deposition, surface roughening, cracking, and fuel penetration in graphite samples exposed to highly ra- diolyzed fuel salt. These unfavorable phenomena were absent in capsules where little or no fuel radiolysis took place. INOR-8 Longitudinal cross sections of the capsule wall, including the region of the salt-vapor interface for capsules 1 through 4, were mounted in epoxy resin, polished, etched with aqua regia or picral-HCl, and examined metal- lographically. Similar examinations were made of longitudinal sections of the INOR-8 bottom centering pins in capsules R and F, the INOR-8 in- serts between which the graphite wafer of capsule 2 was sandwiched, and samples of the capsules 3 and 4 gas lines. Only minor signs of corrosive attack were noted on the capsule wall specimens. The metallograph of the capsule 3 wall at the vapor-salt in- terface region, Fig. 5.18, was similar to those of the capsules R and F samples (including the centering pin specimens) and shows no evidence of damage to the metal. Figure 5.19 shows a slight roughening of the wall at the vapor-fuel interface of capsule 4. The degree of attack on the capsule 2 wall was intermediate between the cases shown. Thus, signifi- cant corrosive attack on the INOR-8 capsule wall was not observed, even in those cases where extensive fuel radiolysis took place. R-20298 g / o y R e Fig. 5.18. Metallograph of Longitudinal Section of Capsule 3 INOR-8& Wall at Vepor-Fuel Interface. Reduced 5%. Extensive carburization of the surfaces of the capsule 2 metal in- serts adjacent to the graphite wafer was noted (Fig. 5.20). The salt- exposed regions of the inserts were undamaged and not carburized. The absence of carburization on the centering pins of capsules R and F may be ascribed to a loose fit of the pins into the graphite rods. The INOR-8 gas lines leading from capsules 3 and 4 showed no sign of corrosion. However, the stainless steel reducer on the capsule 4 exit line, about 1 ft from the capsule, showed voids and intergranular attack in the stainless steel which were characteristic of fluorine corrosion (Fig. 5.21). The corresponding reducer on the capsule 3 exit line was relatively unattacked. An effort was made to detect and identify metallographically the shiny and dark films visually observed on the INOR-8 specimens. No films or deposits of any kind were discernible with the metallograph, suggest- ing either the evanescent nature of the deposits or their removal by the ethylene glycol polishing slurry. An attempt was made to determine chemically whether uranium plated out on the INOR-8 capsule walls. Large samples of the capsule side walls were scraped to remove the bulk of the adhering fuel salt and leached first 99 R-20296 T 100X I Fig. 5.19. Metallograph of Longitudinal Section of Capsule 4 INOR-8 Wall at Vapor-Fuel Interface. Reduced 5%. with concentrated sulfuric acid then with hot nitric acid. The separate leaches were analyzed for uranium and the other bulk fuel constituents. The small quantities of uranium that were found (0.1 to 30 mg total) could be accounted for as adhering fuel salt in those cases where reliable anal- yses were obtained. Thus, no evidence for uranium deposition on the cap- sule walls was found, although reduced uranium would be expected to deposit on INOR-8 as readily as on graphite. Summarizing the observations on the irradiated 47-5 INOR-8 specimens, the metal withstood exposure to fluorine and reduced fuel, as well as ex- tended contact with fissioning fuel salt, with only slight indications of corrosive attack. Normal carburization of INOR-8 was observed only in the capsule 2 specimens which were in close contact with graphite. No chemical or metallographic evidence for uranium deposition on INOR-8 was found. Fuel Salt The fuel salt from each of the four large capsules was divided into top, middle, and bottom portions, and each was separately pulverized and chemically analyzed for the bulk fuel constituents. The uranium was de- termined amperometrically, the lithium by flame photometry, and the be- ryllium and zirconium spectrographically. The major fission product ac- tivities in each sample were determined by gamma spectrometry. Corrosion 100 Fig. 5.20. Metallograph of Longitudinal Section of INOR-8 Insert Adjacent to Graphite Wafer in Capsule 2. Reduced 5%. products were determined by solvent extraction and spectrographic analysis for the large middle salt sample in each case. The same analyses, except for corrosion products, were carried out on samples of the salt-impregnated rods from capsules R and F and on the droplets of salt which had escaped from the impregnated graphite. The agreement of the chemical analyses for bulk constituents was dis- appointing, with more than half the analyses varying from the original cation concentrations by more than 10%. The individual analytical methods had a tested accuracy of 3 to 5%. Some of the discrepancies could be ac- counted for by segregation of the various salt phases during feeezing. However, the overall material balances for each bulk constituent were sim- ilarly poor; the amount found in well-mixed samples representing all of the capsule charge was usually low. Also, the sum of the cation percent- ages for each sample was invariably lower, in several cases by 5 wt % out of a total of 33 wt %, than for the original salt. Some additional checks of the analytical methods failed to reveal the sources of the apparent discrepancies in this particular set of analyses. The general trend of the analyses for individual constituents was in line with evidence from petrographic observations that no gross chem- ical changes occurred in the fuel. The top, middle, and bottom salt anal- yses indicated no single vertical pattern of segregation in all capsules, 101 R-20303 Fig. 5.21. Metallograph of Longitudinal Section of Stainless Steel Reducer on the Capsule 4 Gas Exit Line. Reduced 5%. and the implied range of segregation was unexceptional. The analyses in- dicated that at least most of the uranium and zirconium remained in solu- tion. If solid compounds of these elements had formed, it is not likely that they would have remained suspended in the top region of the frozen melt. The droplet material and the impregnated fuel in capsule R (low power density) were approximately the same in composition as the original fuel. However, the droplet material in capsule F (high power density) was high in beryllium and low in uranium compared to the original fuel, with the inverse true for the remaining impregnated fuel. This suggests that the fuel left the capsule F graphite partly by volatilization, which has been observed to yield a distillate high in beryllium and low in uranium. The temperature of the capsule F graphite should have been high since it con- tained the 4 wt % 2357 fuel and was positioned at the high-flux end of the in-pile assembly. The gamma spectra of the 47-5 salt samples appeared different from those of previous runs because the 6- to 9-month delay between irradia- tion and observation enhanced the 30-year 13705 and the 1-year 106Ry peaks relative to those of shorter-lived fission products. The predomi- nant peaks in bulk salt samples were l“’Ce, 95Zr-‘951\1'b, 13705, and 106y with relative intensities (dis/min) of approximately 100, 40, 30, and 1 102 respectively. The 106Ry showed the most variation, from 0.1 to 90 on the above relative basis (*%%Ce = lOO), reaching the higher values for samples taken near the top surface of the fuel salt. (Later experience has shown that much of the ruthenium activity may be lost in the usual salt dissolu- tion by boiling sulfuric-nitric-boric acid without a condenser.) The drop- lets of salt from capsule F showed a *37Cs activity four times the 1%%Ce activity, possibly indicating volatilization of CsF from the fuel-impreg- nated core of this high-flux capsule. The 1440e to 2°7Zr-°°Nb ratios for most salt samples varied little from the average value of 2.5; a few samples of material adherin% to the INOR-8 wall in the vapor phase con- tained less *%%Ce and more °°Zr-°°Nb activity. INOR-8 corrosion products were determined in the middle portions of fuel salt from the four large capsules. The concentration of iron varied from 40 to 230 ppm, molybdenum from 20 to 120 ppm, and nickel from less than 70 to 160 ppm. Chromium and manganese were not detected to limits of 4 and 2 ppm respectively. The corrosion product levels were highest in capsule 3 salt. Since chromium is the most active metal in INOR-8, the chromium con- centration should have been higher than nickel, for example, but similar puzzling observations were made in previous runs. In any case, the low levels of corrosion products, particularly of chromium, indicate insig- nificant attack on INOR-8, in agreement with the metallographic evidence. Summary of MIR-47-5 Postirradiation Examinations The results of the detailed examinations of the graphite, INOR-8, and fuel salt specimens from the six irradiated capsules of experiment MTR-47-5 confirmed the previously suggested relation between material in- compatibilities and fuel radiolysis at low temperature. Heavy uranium deposition, surface roughening, cracking, and fuel penetration were ob- served in graphite specimens from capsules in which the fuel salt was ex- tensively radiolyzed. Much less damage to graphite was observed in cap- sules where little or no fuel radiolysis took place. The INOR-8 withstood exposure to fluorine and reduced fuel, as well as extended contact with fissioning fuel, with only slight indications of corrosive attack. Car- burization was observed only in two specimens which were in close contact with graphite. The chemical analyses of the salt, though imprecise, in- dicated no gross chemical changes in the fuel. Gamma scans of salt samples showed a tendency for fission product ruthenium to concentrate in regions exposed to the capsule gas spaces. Experiment MI'R-47-6 Ob jectives The general objective of the 47-6 test was to demonstrate that the undesirable effects noted in previous experiments could be avoided by 103 maintaining an elevated fuel salt temperature during the reactor shut- down periods, thus eliminating periodic fuel radiolysis during the total exposure. The main particular objectives were to determine CF, concen- trations in the capsule cover gases with high sensitivity and to examine the graphite surfaces thoroughly for uranium deposition. Another ob- Jective of the 47-6 test was to measure the rate of radiolysis of CFy. The traces of CF, found in the 47-5 gas samples might have been the remnants of much larger original quantities. OSince past results of the usual postirradiation examinations of salt, metal, and graphite specimens were clouded by fuel radiolysis effects, these routine tests on the 47-6 specimens were more directly pertinent to the MSRE. More attention than in previous tests was paild to determining the fate of fission products such as iodine and tellurium, whose volatility was of interest, and ru- thenium, which was expected to deposit as metal on the capsule walls. The uranium concentration in the fuel was varied from 0.5 to 4.0 mole % in the four 47-6 capsules to determine whether this variable affected ra- dioclysis or corrosion. DMolybdenum coupons were included in two of the capsules as possibly more sensitive corrosion indicators and to test whether the previously observed corrosion occurred when fuel radiolysis was suppressed. Experimental Capsule and Heater Design. The principal innovation of the 47-6 design was the provision of individual heaters on the four capsules to keep the fuel salt molten during reactor shutdowns. This change required a redesign of the complete nose end of the in-pile assembly, with a modi- fied capsule shape and fission-heat removal by means of individual water jackets around each heater and capsule (Fig. 5.22). Four of these assem- blies were mounted vertically in a horizontal array in the nose cone and were numbered 1 through 4 in order of increasing flux position. Fach capsule heater consisted of a spiral of l/8-in.-OD Inconel- sheathed, MgO-insulated heater wire embedded in a cylindrical solid copper body by a flame spraying technique. The power rating of each heater as- sembly was 8000 w at 110 v ac. The heater inside diameter was slightly less than the capsule outside diameter, providing good thermal contact when the capsule was shrunk by cooling and allowed to expand into posi- tion. The heaters operated automatically when the reactor shut down to maintain the fuel salt molten during the complete 12-week irradiation period. For additional capsule temperature control, heat leaving the capsule and heater was made to pass through a narrow gas gap to reach the cooling water jacket. The gap was purged with helium, nitrogen, or an adjustable mixture of these gases. Thermocouples were provided to measure the tem- perature of the water entering and leaving each cooling jacket; from these readings, total capsule power could be calculated and fission power esti- mated. As in previous experiments of this series, the nose end of the in- pile assembly could be mechanically moved 13 in. toward or away from the MIR core during irradiation, permitting controlled variation of the cap- sule fission power density. 104 ORNL-~DWG 64-4010 THERMOCOUPLE~\\\\\ l s ¢ ] ' //-PURGE TUBE _‘ I m B < (@] o Q c o 2 m ] ’ | \\ COOLANT WATER COVER GAS SUPPLY TUBE SPACE (~ 3 cm3) L WATER JACKET - HEATER ANNULAR GAS GAP ——| AN SN AN NN NN NN ESSONSISSS CGB GRAPHITE CORE MOLTEN SALT FUEL 1 | L INOR-8 CAPSULE BODY i) | ) I II | i S ' I 1 ll v e : I I I' ||!I|III - ‘::. | ] I [} II I' ' LIl - S s s S S S N N AN A I NN N S AN N S S S S SAOSS S SESSSIN K VOISO ) | ] | N imm——WATER FLOW CHANNEL SOV N SASISNIISSSSNSSNSN HEATER LEADS — GASCAPPURGESUPPLYTUBE—”//’/ Fig. 5.22. MIR-47-6 Capsule Design. 5 2 ] Capsules 1, 2, and 4 each contained a 1.35-in.-long, 1/2-in.-diam graphite core held in position by the thermocouple and a bottom pin. The core of capsule 3 was only half as long, to test the effect of exposed graphite area on CF, evolution. Two flat surfaces were milled on opposite sides of each cylindrical core to implement postirradiation x-ray diffrac - tion examinations for surface deposits. Each core was submerged in about 105 27 g (12 em?® at 1200°F) of fuel salt, leaving a capsule gas space of about 4 em®. Capsule 1 fuel salt contained 4.0 mole % UF,, that in capsules 2 and 3 contained 0.9 mole % UFy (like MSRE fuel), and the fuel in capsule 4 contained 0.5 mole % UF,, the uranium being highly enriched (93.26% 23515) in each case. The sealed capsules 1 and 4 each contained three molybdenum corrosion specimens, 1 in. X 1/8 in. X 0.020 in., mounted vertically in slots in the capsule side and bottom walls. Capsules 2 and 3 were equipped with gas lines for purging the capsule gas spaces with purified helium and for sampling the capsule effluents. The first 7 to 9 in. of the capsule inlet and outlet lines was 0.305 in. ID to minimize plugging by volatilized fuel salt. The remaining 50 £t of each line to the sample station was l/8-in.-OD stainless steel tubing, shielded by 6 in. of lead between the reactor plug and the sample station. To avoid the possibility of cross contamination of gas samples, completely separate purge gas supplies and sample collection systems were provided. The supply helium passed through a Grove pressure regulator to one of a set of three capillary flowmeters calibrated to measure flows between 50 and 5000 cmB/hr. Just prior to entering the sample station adjacent to the reactor beam hole, the helium was purified by a Linde 13X molecular sieve drying column and a titanium sponge trap maintained at 600°C. A simplified schematic drawing of the flow circuit for the capsule 3 part of the sample station is shown in Fig. 5.23. The sample station cu- bicle, shielded by at least 6 in. of lead, enclosed a similar circuit for capsule 2. During purging, the capsule effluent was led directly to the shielded carbon trap and thence to the off-gas line. During sampling, the Dewar flask containing the Linde 13X molecular sieve trap was filled with liquid nitrogen, and the gas leaving the capsule was passed through the trap. After collecting all the other gases from the helium for the desired time (1L to 48 hr), the helium was evacuated from the trap, the trap was warmed to 250°C, and the released gases were swept with helium into an evacuated sample bottle, furnishing a 24.6-cm® gas sample at 600 mm pressure and room temperature. The concentration effected by this col- lection method was more than a factor of 100 for the usual sampling times and purge flows. The gas samples in their valved and shielded nickel containers were shipped to ORNL for analysis by two separate mass spectrometry groups. Also, those samples showing high radiocactivity (about 1 in 4) were ana- lyzed by gamma spectrometry for possible carry-over of volatile fission products such as iodine. Concentrated gas samples were taken over a wide range of the experi- mental variables, representing capsule temperatures from 800 to 1440°F, _ 3 . . [P purge-gas flow rates from 50 to 5000 em /hr, and fission power densities from O to about 70 w/cm3. For fear of possibly plugging the capsule exit lines with volatilized fuel salt, the early gas samples were taken at low temperatures, power densities, and flows. The later samples were taken at the higher values of these variables. 106 ORNL-DWG 65-2530 F—-—-—-- TO PRESSURE RECORDER DIFFERENTIAL PRESSURE CELL CHECK VALVE g THERMAL T\ \J=—PURIFIED He _______ CONDUCTIVITY CELL > | 1 GAS CHROMATOGRAPH Y ISkt Elj AND DETECTOR IRRADIATED r——-—-TO PRESSURE RECORDER CAPSULE ' G‘IS DIFFERENTIAL PRESSURE CELL HEATER Sk - ] ¥ El] CARBON POWER - TRAP [ Y Y LINDE 43X MOLECULAR | VACUUM SIEVE r—-— PUMP A 24.6 HEATER l cc BOTTLE Fig. 5.23. Gas Flow Circuit. Qualitative information on the rate of radiolysis of CF, was obtained from the regular concentrated gas samples taken at different purge flow rates, but more definitive measurements were obtained by passing helium- CFs4 mixtures through capsule 3 and determining the change in thermal con- ductivity with a sensitivity of 0.005% CF,. Gas mixtures containing 0.1% CF, and 5% CF, were passed through the capsule 3 flow circuit for these tests. BSince the helium purification system could not be used with these mixtures, some contamination was unavoidably introduced into the flow cir- cuit during these tests. The variation of radiolysis rates with capsule power density and temperature was observed. A gas chromatographic method for detecting parts per billion levels of impurities in helium, using a newly developed sensor which measured impurity-caused change in helium breakdown voltage, was applied to the detection of CF, in direct unconcentrated samples of capsule 3 purge gas. A third special purge-gas mixture, containing 1 ppm CF,, was passed through the capsule 3 circuit to calibrate the detector and to measure the change when the l-ppm mixture was passed through capsule 3. When the four MIR cycles of irradiation, accumulating 1500 hr at full reactor power, had been completed, the capsules were allowed to decay for 32 hr with the fuel salt molten, in the hope of minimizing fuel radiolysis 107 during dismantling. The dismantling was carried out as rapidly as possible in an MIR hot cell. The nose end of the in-pile assembly was sawed off, the individual capsule-heater-water jackets were sawed apart, and finally the capsules themselves were milled open. The fuel from the last capsule opened was isolated from the other capsule components within 48 hr of the freezing of the salt. Qualitative tests with KI-starch paper indicated a slow evolution of fluorine from the salt specimens. The fuel salt, graph- ite, and metal specimens from the capsules were packed in separate metal containers and shipped to ORNL for postirradiation examinations. Results and Discussion In-Pile Operation. The in-pile assembly and the sample station per- formed according to design in all crucial respects. Due to the new nose- end design, estimated fluxes were only about half those actually experi- enced, resulting in higher power densities than planned. To avoid boilling of the water in the individual cooling Jjackets, the maximum insertion was limited to 10.1 in. rather than 13 in. The ball and metal-diaphragm-sealed valves used throughout the sample station were found to leak slightly during actual operation, though not when left in open or closed position. By minimizing valve manipulations and by appropriate flushing operations, significant system and sample con- tamination was avoided. An instrument power failure for 20 min during a reactor shutdown period permitted a momentary freezing of the fuel salt and cooling to 250°F. Subsequent gas samples indicated no measurable fuel radiolysis during this failure. Two of the thermocouples measuring cooling water Jjacket temperatures failed, introducing uncertainties in estimating heat balances and capsule power densities from these readings. Satisfactory estimates were derived, however, from the remaining thermocouple readings. The 48-hr delay in separating the fuel salt from the other capsule components after the final cooling of the in-pile assembly was longer than planned, but subsequent examinations indicated that no serious damage was done. Flux and Power Density Determinations. The most reliable indications of burnup, average flux, and average power density were obtained from 236U/235U ratios determined on salt samples from each capsule. The re- sults are given in Table 5.2. The results for power density and neutron flux are in excellent accord with estimates® based on cooling water temperature measurements (Fig. 5.24). Self-shielding by the in-pile assembly explains the differences in the slope of flux variation with insertion position between capsules as compared with the slopes for individual capsules. Flux results from ccobalt dosimetry Table 5.2. Burnup, Average Flux, and Average Power Density for the 47-6 Capsules B a A F1 Average Range of Power Capsule u?ggp ( tverzg: C_Ex 2) Power Density Density Variation neuLtron e c w/cm3) (W om 1 0.835 2.7 % 1.01% 54 38-57 2 1.45 46 X 1012 23 17-33 3 2.86 9.2 x 1012 47 2968 4 5.53 1.8 x 10%3 53 36175 aBurnu.p by fission; does not include burnup by capture. bEstimated from cooling water heat balance data. 80T 109 ORNL-DWG 65-4186 10" ‘ : : : : : — LINES ARE CALCULATED FRCM COOLING WATER | HEAT BALANCE DATA DURING THE FIRST REACTOR] | CYCLE.POINTS REPRESENT AVERAGE FLUXES | - DETERMINED FROM 238U/ 239y MEASUREMENTS 'o 5~ IN IRRADIATED FUEL FROM EACH CAPSULE AND | . | ARE PLOTTED AT THE AVERAGE INSERTION .. k) POSITION FOR EACH CAPSULE. g I w0 = £ CAPSULE 2 2 4 c ] Fig. 5.24. Variation of Flux « » . . - ] with Position in Beam Hole. 2 ///6;§m£ = 1013 ,/ g 'y |_ ] E 7 _~TAPSULE 2 § 5 //// » z e z i w " CAPSULE 4 2 1 z l > 2 < 1012 -8 -4 ) 4 8 12 16 20 PLUG INSERTION PGCSITION (in.) were unaccountably much higher, and the analyses are being repeated. These average results correspond to the average insertion position of 4 in. Gas Samples. No CF, or other volatile fluoride was detected by mass analysis in any of the regular concentrated gas samples (detection limit 2 to 10 ppm). The gas chromatograph analyses on direct unconcentrated samples of capsule 3 effluent likewise indicated no CF,, with a detection limit of about 0.2 ppm. The most sensitive determination involved a 48- hr sample collection from capsule 3 operating at a power density of 68 w/émB. The sample analysis showed less than 2 ppm CF, and over 2000 ppm of fission product xenon. For fuel radiolysis to be of significant prac- tical effect in the operation of the MSRE, the CF, generation would have to exceed the xenon production. ©Since the radiolysis results reported below indicated a negligible degree of CF, radiolysis under the sampling conditions used, the negative CF, analyses may be taken as definitive proof that molten MSRE fuel is stable to radiolysis in a fissioning or highly radioactive environment over the whole range of power densities and tem- peratures investigated. This demonstration accomplished one of the major obJjectives of the 47-6 experiment. The voluminous other gas analysis results are described in detail in a forthcoming report and will be only briefly summarized here. The levels of impurities such as N, Oz, Hp, H20, and CH,; were generally gratifyingly low, indicating satisfactory leak-tight operation of the sampling system. Water was detected only in the first few concentrated 110 samples. The total xenon and krypton analyses showed a wide scatter and a low bias by a factor of about 2, leading to a suspicion that these fis- sion products were released in occasional bursts. Xenon isotopic analyses revealed low 131%e and 132%xe percentages com- pared to fission ylelds. These particular isotopes had relatively long- lived precursors, 8-day 1317 and 77-hr 122Te. It is thought that the volatilized precursors were swept out of the system during purging, re- sulting in low quotas of 131ye and 132Xe in subsequent gas samples. It was also noted that the 3®Xe/1?%Xe ratios were intermediate between those calculated from fission yields for no conversion of *2°Xe to *3%Xe by neu- tron capture and for conversion corresponding to the exposure of 135%e to the neutron flux for its complefe 13.3-hr mean life. From these data it was calculated that the average residence time of 12°Xe in the high-flux region was about 7 hr. The gamma spectrometer results on the more radio- active gas samples showed 133%e as the only detectable activity. The de- tection limit for *3'I and 1?°Te activities was 0.05% of the 1?3Xe activity. CF, Radiolysis. ©Since no CF, was detected in capsule 3 effluent at any purge flow rates, the regular concentrated gas samples yielded no quantitative information on the rate of radiolysis of CF,. During the third MIR cycle of irradiation, special radiolysis tests were performed, passing helium containing 0.1% CF, or 5% CF, through capsule 3. The mix- ture was admitted through one side of a thermal conductivity bridge, passed through the capsule, and exited through the other side of the conductivity cell. The reading was compared to that when the same gas was passed through both sides of the cell without passing through the capsule containing fis- sioning molten salt. With the instrument set at highest sensitivity, l-mv output change (read on a 10-mv recorder) corresponded to 0.05% CF, concen- tration change. A second, more sensitive but less accurate, method in- volved isolating the capsule containing a known CF, concentration for sev- eral hours, then flushing this gas through the thermal conductivity cell with the original mixture. Radiolysis rates obtained by this static method, reading the minimum of the effluent concentration, were uniformly less than rates determined by the flowing method, since some mixing of flush gas with capsule gas was bound to occur. The first dynamic tests were made with helium containing 0.1% CF, until it was determined that the radiolysis rate was too low to be meas- ured with any accuracy. A static test, isolating 0.1% CF, in the capsule for 10 hr, indicated a radiolysis rate of 4.2%/hr, with capsule tempera- ture at 1077°F, minimum insertion (power density of approximately 29 w/cm3), and a flushing rate of 192 cm®/hr. The 5% CF, mixture was then flushed through the system, and radiol- ysis runs gave the results shown in Table 5.3. The tabulated data indi- cate a CFy radiolysis of less than 4%/hr above fuel fissioning at the de- sign power density and temperature for the MSRE (14 w/cm3 and 1200°F). The strong inverse temperature dependence of the radiolysis rate may mean that back reaction of the radiolysis products is accelerated by increasing temperature. The decomposition rate observed during reactor shutdown, presumably a "chemical” blank since the beta-gamma decay power density is Table 5.3. Radiolysis of 5% CF, Mixture Tygznof Flow (em?/ar) Temp?f§§ure In?:;?§on Pow?;/gzgiity (gifii) Unc?ifiii?ty Dynemic 1000 1075 0 29 5.0 +3 Static 1000% 1075 0 29 3.1 2, =0 Dynamic 100 1075 0 29 4.1 +0.4 Static 100 1075 0 29 3.6 +1, =0 Dynamic 100 1275 10 68 8.8 0.5 Dynamic 100 1290 2 35 4ol +0.4 Dynamic 1.00 1042 2 35 6.5 £0.5 Dynamic 100 831 2 35 8.6 +0.5 Dynamic 100 1047 2 QP 2.0 £0.5 Static 100° 1049 2 o 1.6 0.3 Dynamic 100 1049 2 o 1.6 +0.4 %Flush rate. Testing during reactor shutdown. TTT 112 a small fraction of the total power density associated with fission, was considerably larger than expected from out-of-pile CF, decomposition studies. The 5% Cl, radiolysis results taken as a whole indicate that the nega- tive CF, analyses from the 47-6 gas samples may not be ascribed to rapid CF, radiolysis. However, considerable corrections for radiolysis are in- dicated for the 47-5 gas samples, since the purge gas stagnated in the capsules for several days before sampling. Some puzzling observations related to radiolysis of CF, were made while the l-ppm CF, mixture was passing through capsule 3. With the re- actor operating, both the gas chromatograph and gas samples indicated the destruction of 1 ppm of CF, on passage through the capsule. With the re- actor down, gas samples indicated the survival of CF,, while the chroma- tograph indicated its conversion to an unidentified product. Since mass analysis of CF, actually measures a peak for CF3 , a particle produced also from other fluorocarbons, these observations may not be contradictory. It is suspected that the observed changes were due to radiation-catalyzed re- actions of CF, with gaseous impurities such as I or CH,. Analyses for Uranium in Graphite. Uranium was found to have deposited on the surface of graphite cores from earlier trials.” Since the evolution of fluorine was also encountered in these same capsules, there was a strong likelihood that the transport of uranium was associated with the fluorine. However, the mechanism remained obscure, and the possibility that the ura- nium deposition was inherent in the exposure of graphite to fissioning fuel lingered as a question of considerable concern. The importance of the question was such that the uranium in the graph- ite was determined by several methods. Two types of quantitative analyses were employed. Samples of graphite, cylindrical cross sections weighing approximately 1 g, were taken from the upper and lower portions of the cores. These were dissolved in a mixture of nitric and perchloric acids, and the solutions were analyzed fluorometrically for uranium. Other por- tions of the same solutions were analyzed spectrographically for the rest of the fuel constituents. The results on uranium, shown in column 5 of Table 5.4, corresponded roughly to the trace amounts of fuel that adhered to the graphite, as evaluated from the amounts of other fuel constituents found. Adjacent samples of graphite, cylindrical cross sections about 20 mils thick, were analyzed by counting delayed neutrons after activation in the ORR. The results, shown in column 7 of Table 5.4, applied only to 23°U and were generally lower than found by fluorometric analysis, which meas- ured the total uranium. Since the two types of analyses had not been applied to the same samples of graphite, the different results might in part have been due to irregularities in the distribution in the graphite. To check this point, samples of the solutions used for fluorometric anal- yses were analyzed by the delayed-neutron method. The results, shown in column 6 of Table 5.4, were usually lower than indicated by fluorimetry, and were regarded as meriting more weight than either of the other sets of figures. Table 5.4, Uranium in Graphite from In-Pile Experiment 47-6 Ursnium in Fuel Estimated Graphite Samples Capsule Type .(mole %) Burnup Dissolved Solid (%) Fluorimetric Delayed Neutron 1 Sealed 4 0.8 40 30 10 96 89 40 2 Purged 0.9 1.5 7 4.8 8 5.2 3 Purged 0.9 2.9 60 19.4 6 12 4 Sealed 0.5 5.5 51 12.2 6 175 8.5 130 Blank 20 0.1 0.11 (16)% (16)® ®The delayed neutron analyses applied only to 235U; the figures in parentheses are corrected to give the total uranium in the blank, which had the normal isotopic distribution. eit 114 CAPSULE 3 CONTROL Fig. 5.25. X Radiograph of Transverse Section of Capsule 3 Graphite Core. In spite of the scatter in the results, the amount of uranium found is inconsequential, particularly in light of the evidence that it arose from traces of adhering fuel. The search for uranium by means of x ra- diography of thin sections of graphite gave negative results, in contrast with the results from earlier exposures (Fig. 5.25). Also, x-ray diffrac- tion from flat surfaces of the graphite gave only graphite lines. The amounts of uranium were lower by about two orders of magnitude than those found in earlier experiments where F, was involved. Although questions about detailed mechanisms that produced the earlier deposits were unresolved, the prospects are excellent that operation of the MSRE will not be impaired by the deposition of uranium in the graphite. Fission Product Volatilization. In previous postexposure examinations, little emphasis was placed on determining the fate of iodine, tellurium, and ruthenium fission products because the liberation of F, had disturbed the chemical picture in the capsules. Since fluorine evolution was sup- pressed in the 47-6 experiment, information on the behavior of fission products was more pertinent to the MSRE. The possibilities of releasing iodine and tellurium from the fuel as volatile compounds and of plating of ruthenium on the graphite and metal walls were of particular practical concern. Salt samgles from each of the four capsules were analyzed radiochem- ically for x3 I; samples from capsules 2 and 3 were also analyzed for 10%Ry and *2%Te. In addition, three sections of inlet and exit gas lines from capsules 2 and 3 were analyzed quantitatively for 21T and 1?°Te and qualitatively for *°°Ru. The four 7-in.- to 9-in.-long, 3/8-in.-OD tubes attached to the capsules, four 3-ft sections of l/B-in.-OD tubing 30 ft from the capsules, and four 1-ft sections of l/8-in. tubing at the sample station were taken as representative samples of the gas lines. 115 The analytical results for the salt and for the gas lines are given in Tables 5.5 and 5.6, respectively, expressed as percentage of total fis- sion product calculated from the known uranium contents, fission yields, and neutron fluxes. The iodine evidently remained in the salt in the sealed capsules, but volatilized to an appreciable extent from capsules 2 and 3. Correspondingly, sizable percentages of the total ilodine were Table 5.5. Fisgsion Products in 47-6 Fuel Salt g lea Percent of Percent of Percent of I Total 1311 Total 12°Te Total 193Ru 1FS 6.4 1FS 109.0 1F18 44 .0 173 (black) 85.0 1F3 (green) 96.5 2FS 19.0 34.0 36.0 2F18 8.6 3FS 2.0 26.0 0.34% 3FS 33.0 3rF18 53.0 PATS 120.0 4F18 74..0 aThe first numeral is the capsule number; FS represents a composited bulk fuel sample; F18 represents a sample of fuel from the bottom of the capsule; F3 refers to special colored fuels from capsule 1. Table 5.6. TFission Products in 47-6 Gas Line Samples . Percent of Percent of 103 Capsule and Location Total 1311 Total 1297 Ru 2, near capsule 8.0 0.004 Some 2, 30 £t from capsule 0.2 0 2, 50 £t from capsule 1.0 0 , near capsule 6.0 0.07 Some , 30 ft from capsule 5.0 0.006 50 ft from capsule 1.0 0 116 found by leaching the gas lines. About 70% of the 129%7e had left the fuel salt in capsules 2 and 3, but only traces were found in the gas lines. Presumably the volatile tellurium compound did not react with the metal walls and passed through to the carbon trap. The 103gy also mainly left the salt phase; its activity was detected in the gas lines by qualitative gamma scans of the leach solutions, but much of it may have plated on the capsule walls. Examinations of the latter for *°3Ru and 196Ru are planned. As mentioned above, the low 131Xe and 13%xe percentages found in the isotopic analyses of the gas samples are consistent with volatilization of tellurium and iodine from the fuel salt. Qualitative gamma scans of the graphite specimens (and their solutions) indicated higher 103Ry than 957r-2°Np activities, a reversal of the situation in gamma scans of bulk salt samples., Thus plating of ruthenium on graphite is suggested. Similar observations were made on some salt samples taken from the top meniscus of the fuel or from the material deposited on the capsule walls above the fuel level. Examination of Salt Samples. The salt specimens from the 47-6 test were generally gray-green in color rather than black as in previous tests. The salt from capsule 1 had segregated into two clearly separated regions, one pale green and the other black. Chemical analyses of the green and black portions revealed a significantly lower uranium content in the green material. The paler color of the 47-6 salt specimens may be related to the suppression of radiolysis during reactor shutdowns or to the heating for 32 hr after the final shutdown. Petrographic examination of selected salt specimens yielded limited results since most samples were microcrys- talline due to rapid cooling. The LiyBeF, phase was identified in all samples and the 3LiF-UF, phase in a few (notably in a black sample of capsule 1 salt). Worthy of note was the absence of pink-bronze colored . materials found previously in radiolytically reduced salt specimens. In spite of difficulties in mounting x-ray diffraction samples of salt re- motely and operational troubles with the hot-cell x-ray diffraction in- strument, some further results of interest were obtained. The presence of the above-mentioned phases was confirmed, another uranium-containing phase, 7LiF-6UF,, was observed, and very curiously no zirconium-containing phases were detected with certainty. The absence of the latter and the presence of 3LiF-UF, indicated that the cooling occurred under conditions quite different from equilibrium. However, no indications suggesting fuel instability were observed. A large number of gamma scans were obtained on salt samples taken from various regions of the capsules. The types of sample examined in- cluded top, middle, and bottom samples of bulk fuel salt, volatilized beads, and adherent scale from the INOR-8 wall in the gas space, green and black salt from capsule 1, middle salt adjacent to graphite and to INOR-8, and salt from the top meniscus of the fuel. The principal ac- tivities were 95Z:|:'-95I\Tb‘, 1410e, 140Ba—l40La, and 193Ru in these relatively short-cooled samples. As for samples from previous runs, there were large variations in fission product distributions even for adjacent crumbs of salt, and a 117 number of exceptions were observed to the following generalizations. The principal activities in bulk fuel samples were 937r-7°Nb and 141Ce, usually in that order, but with L4lce often predominating in bottom salt samples. In the salt bead samples from the purged capsules, 103Ry was the maJjor ac- tivity, and it was prominent in the samples of scale on INOR-8 in thel%as- phase region. The middle salt adjacent to the capsule wall showed an 3Ru activity higher than that of 14lce or 140Ba-140La, The gamma. scan of an oxalate solution leach of the capsule 4 gas lines adjacent to capsule 4 showed high 103Ry and 95Zr--951\Tb, low 140Ba-l4OLa, and no 14*Ce. Fuel meniscus samples generally showed higher 103Ru activities than bulk fuel samples. The green salt from capsule 1 contained relatively more 103Ru than the black portion. These results generally indicate a tendency of ruthenium to concentrate at interfaces between the fuel and other phases. For chemical analyses, the fuel salt from each capsule was divided into a bulk salt sample (~25 g) and a bottom salt sample (1 to 2 g). From capsule 1, additional samples of segregated pale green and black salt were taken. FEach sample was pulverized, composited, and analyzed for uranium (amperometrically), lithium (flame photometrically), beryllium (spectro- graphically), and zirconium (spectrographically). The average deviation of the analyses of all the samples (Table 5.7) from the original cation concentrations by weight was 13.5% for beryllium, 10% for lithium, 9% for uranium, and 5% for zirconium. The deviations cal- culated for the large bulk samples alone were slightly less. The direction of deviation was usually positive for beryllium, lithium, and zirconium, and negative for uranium. The analyses of bulk salt and bottom salt usu- ally differed considerably, but consistently only in that the beryllium content was higher in the bottom salt of each capsule. For most samples, the weighted average of the bulk and bottom salt analyses would not cor- relate significantly better with the original cation concentrations. It is clear that improvements in analytical techniques are required in order to observe small changes in fused salt composition. The uranium and lith- ium results tabulated were obtained by methods planned for use on MSRE salt samples. Examination of Graphite. The observations pertaining to uranium deposition on the graphite cores have been discussed previously. Visual examination with low-power magnification revealed no differences between the irradiated and unexposed blank cores. Preliminary results from metal- lographic examinations of graphite specimens from each capsule likewise showed no indications of radiation or chemical damage. The graphite cores were weighed and dimensioned after a light brushing to remove most of the adhering salt. The dimensions had not changed within error of measurement (+0.002 in.). The slight weight gain of two cores could be attributed to visible adhering salt in the positioning holes. Aytoradiographic examinations of transverse cross sections of the cores remain incomplete. Table 5.7. Chemical Analyses of MIR-47-6 Fuel Salt Capsule 1 Capsule 2 Capsule 3 Capsule 4 Bulk Bottom Black Green Bulk Bottom Bulk Bottom Bottom Li, wt % a 10.8 10.3 9.5 11.9 10.6 10.9 11.6 11.0 Deviation, % +30 +20 +12 +8 -5 —2 +5 -1 Be, wt % 5.63 7.0 4.7 6.3 7.6 6.8 7.3 7.0 Deviation, % +10 +40 -10 -3 +1.7 +7 +13 +8 Zr, wt % 9.45 10.4 9.05 10.2 10.6 10.4 9.2 11.0 Deviation, % +5 +15 0 +2 +6 +4 -8 ~3 U, wt % 18.3 15.2 17.9 10.4 5.05 % .69 4.30 5.57 3.32 Deviation, % -2 ~19 —4 =4y -1 -8 ~16 +10 +11 8TT a L} - * . > L] - Percentage deviation from cation concentration in original salt. 119 Examination of INOR-8. Visual examination of the inner capsule sur- faces with low-power magnification revealed no evidence of damage or cor- rosion. The gas-exposed areas were covered with salt scale and volatilized salt beads near the fuel meniscus and with a very thin black film in the higher regions. In all cases, the salt-exposed areas were shiny and ap- peared sound. Preliminary results of metallographic examination indicate no change in wall thickness. Etching patterns are being studied. X-ray examination of flattened portions of the capsule walls is scheduled to identify the observed surface films. Leaching tests are also planned to search for possible ruthenium deposition on the capsule wallg, Examination of Molybdenum Specimens. The molybdenum specimens from capsules 1 and 4, on visual low-magnification examination, showed no damage except for scratches probably made during disassembly of the capsules. The specimens from capsule 4 were mottled in appearance, as if by the plat- ing of a noble metal. Gamma scans of specimens from both capsules showed the major activities were 103Ry and '%®Ru. The only other activity noted was °°Zr-2°Nb. A crude hot-cell bend test indicated marked embrittlement of the mo- lybdenum by the exposure. The specimens broke cleanly after bending 5°. No oxidation or surface films could be detected visually to which the em- brittlement could be attributed. Preliminary results of metallographic examinations indicate no change in thickness, no oxide or other films on the surfaces, and no carburization or oxidation in the interior of the metal., The survival of the specimens with no visual evidence of corrosion was in marked contrast to the gross attack observed on the 47-3 molybdenum specimens. The difference was ascribed to the suppression of fluorine evolution during the 47-6 exposure. Summary of Experiment MIR-47-6 In the MTR-47-6 test four capsules containing INOR-8, MSRE fuel salt, and MSRE graphite were irradiated for 1500 hr at fuel power densities from 17 to 75 w/cm3 and temperatures up to 1440°F. The capsules were eguipped with individual heaters to keep the salt molten and suppress radiolysis during reactor shutdowns. No evidence of CF, generation was found in the gas samples taken from the two capsules equipped with gas lines. The mass analyses of the gas samples showed low and variable Xenon and krypton yields, low helium impurity levels, and low 131%e and 132xe isotopic percentages. The latter support other evidence that long-lived iodine and tellurium fission products were partially volatilized from the purged capsules. The observed 136Xe/134.Xe ratios suggest that the mean residence time of 13°Xe in the capsules was about 7 hr. The removal of 120 1 ppm of CF,; in helium when passed over molten fissioning fuel may have been due to the radiation-catalyzed reaction of CF, with H; or CH, im- purities in the helium. The rate of CF, radiolysis in contact with molten fissioning MSRE fuel was determined by thermal conductivity measurements and found to be less than 4% per hour at MSRE. power density and temperature. The net ra- diolysis rate was slower at higher temperatures, suggesting accelerated recombination of radiolysis products. The low rates observed preclude radiolysis effects on CF, concentration in the 47-6 gas samples. Sur- prisingly high radiolysis rates were observed with only beta-gamma ra- diations when the reactor was off. A gas chromatograph using a helium breakdown voltage detector proved valuable in detecting trace impurities in helium in the presence of highly radicactive gases. The device confirmed the absence of CF, in purge gas which had passed over molten fissioning MSRE fuel. The graphite specimens from the 47-6 capsules were thoroughly analyzed for uranium content by a chemical fluorometric method and by delayed-neu- tron counting on activated samples. Negligible quantities of uranium were found by both methods, confirming presumptions that uranium deposition oc- curred only if radiolysis of the fuel salt was allowed. Radiochemical analyses of salt and gas-line samples showed that sizable fractions of the iodine and tellurium fission products volatilized from the fuel. Most of the ruthenium also left the fuel, and some was qualita- tively detected in leaches of gas line samples. INOR-8 and molybdenum specimens showed no visible evidence of cor- rosion, but the molybdenum was severely embrittled. Metallographic ex- amination of the metal samples is incomplete. On the basis of the MIR-47-6 test results, it appears that CF,; gen- eration and uranium deposition on graphite or metal surfaces will not be serious problems in the operation of fluoride-fueled reactors. References 1. MSR Program Semiann. Progr. Rept. July 31, 1964, ORNL-3708, pp. 252-87. 2. Ibid., pp. 271-8l. 3. MSR Program Semiann. Progr. Rept. Jan. 31, 1964, ORNL-3626, pp. 86-98. 4. W. R. Mixon, "Analyses of ORNL-MIR-47-6 Heat Balance Data to Determine Capsule Fission and Gamma Power Densities," intralaboratory correspond- ence to S. S. Kirslis (Mar. 2, 1965). 5. Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1964, ORNL-3591, p. 17. 121 6. CHEMISTRY High-Temperature Fluoride Phase Equilibrium Studies Fuel System for the Molten-S5alt Reactor Experiment The System LiF-BeF,-ZrF,. ILiguid~solid phase equilibria in the sys- tem LiF-BeF,-ZrF, are now well defined except for compositions of less than approximately 20 mole % LiF. Within this composition region, accu- rate determination of the phase boundaries of the immiscible liq_uidsl is incomplete because of the volatility of ZrF, at high temperatures and be- cause of the high viscosity of the liquid phases. Until the boundary limits of the immiscible liquids are established, the phase diagram re- ported previOuslyl serves as the most accurate description of phase be- havior in this system. A three-dimensional model of the LiF-BelF,-ZrF, phase diagram was constructed (Fig. 6.1) as a means of affording a sim- ple, graphic display of the crystallization behavior of the MSRE fuel and coolant salts. The model is now on display at the MSRE site. Except for the crystallization reaction in which the uranium phase 7LiF+6UF,; is produced, all equilibrium crystallization reactions of the fuel and cool- ant salt are evident from the model. The System LiF-BeF,. Although numerous investigations of the system ILiF-BeFs> have been conclucted,z"'8 the phase diagram is still uncertain, especially the liquidus curve for compositions greater than 33 mole % BeF,. The liquidus temperatures have been reexamined for compositions from 33 to 70 mole % BeF,, the concentration range of greatest signifi- cance to MSRE performance. Very pure component salts were used for these measurements: crystalline LiF selected from slowly cooled melts of analytical-grade material and very pure BeF, obtained by distillation. The total cationic and oxide contaminant concentrations in the salt mix- tures used for these determinations did not exceed 0.2%. For the compo- sition range 33.5 to 52 mole % BeF,, liquidus and solidus temperatures were readily obtained by conventional thermal analysis techniques. These temperatures were measured with NBS-calibrated thermocouples and have a precision of 0.3°C. Such a precision was not achieved for mixtures in which BeF, is precipitated as the primary phase, probably because of the high viscosities of the liquid phase. Liquidus data for compositions 52 to 70 mole % BeF, were obtained by using the thermal-gradient quenching method. Rapid cooling of LiF-BeF, mixtures in which crystalline BelF, 1s the saturating phase retains the liquid phase as a true glass. This be- havior facilitates determination of the liquidus in quenched specimens by x-ray and petrographic methods. Figure 6.2 shows the typical appear- ance of BeF, that has precipitated out of the LiF-BeF, system. The LiF-BeF, phase diagram, constructed from the cooling curve and the quench data (Table 6.1) as well as from data selected from refs. 2 to 5, is shown in Fig. 6.3. 122 PHOTO 67435A LiF + LIQUID 3LiF-Zrf, + LIQUID BeF, + LIQUID Fig. 6.1. Three-Dimensional Model of the LiF-BeFp-ZrF, System. PHOTO 68702 Fig. 6.2. LiF-BeF, (37-63 Mole %). (a) Quenched from 483°C. Glass and quench growths. Circular areas are bubbles. Halos typical of crystalline BeF, are absent. 160X; white light. (‘Q) Quenched from 475°C. Halos surrounding crystalline BeF, denote occurrence of saturat- ing phase. 160X; white light. (c) Quenched from 475°C. White areas denote location of birefringent BeF, crystals. Glass is isotropic and invisible in polarized light. 160X; polarized light. Reduced 16%. 124 Table 6.1. Phase-Transition Data for the System LiF-BeF, . . (=] Composition Phase-Transition Temperature (°C) (mole %) Thermal Analysis Thermal-Gradient Quenching LiF Bel Liquidus Solidus Liquidus Solidus 67.03 32.97 457.8 65.5 34.5 457.6 65.0 35.0 456 .5 64 .01 35.99 454.0 63.0 37.0 451.9 61.02 38.98 442 .8 59.02 40.98 431.7 56. 44,0 411.5 55. 45 404.5 51.98 48.02 386.3 360.02 49 .97 50.03 374 .6 360.4 45.0 55.0 422 42.5 57.5 425 40.0 60.0 431.3 359.7 39.0 61.0 392 37.0 63.0 392 35.0 65.0 479 32.5 67.5 479 ORNL-DWG 65-2526 650 ‘ ‘ ‘ 600 S — | - 5 550 | - \ 548 < Hp e "1 Fig. 6.3. The System LiF~BeF,. € 500 | —— T g 4576 W 450 M : — o eF 5 + LIQUID T 400 1 - e It S B LiF +2LiF -BeF, 360.3 350 —T ] —] 2LiF -BeF, + BeF, 00 | , 1 LiF 10 20 30 40 50 60 70 80 90 Befp Bef, (mole %) 125 In principle, the liquidus temperatures at which BeF, crystallizes can be calculated from thermodynamic activities of BeF, in the melt, provided that the requisite BeFy thermochemical data (heat of fusion, melting point, and heat capacities of solid and liquid) are available. Trial calculations, using current measurements of activities® and the heat capacity and heat of fusion of BeF2,10 gave liquidus temperatures for the 52 to 70 mole % BeF, concentration range much lower than seems reasonable from the available phase equilibrium data. The System UF,-UF;. Previously, the collection of equilibrium phase data in UF3 systems has been impeded by the tendency of most experimental procedures to induce a shift toward the right in the reactions 3 UF3 = J—Z:UoJf 0% (1) UF; + H,0 = %UF4 + %Hz + HF + %UOE . (2) These reactions take place at elevated temperatures if the container metals form solid solution alloys or liquid eutectic mixtures with ura- nium metal at the temperature of the study or if the opportunity is available for water to adsorb on UFj3, as from exposure to laboratory atmosphere. The system UF,-UF; was (tentatively) described some time ago. Its further elucidation, however, appeared to be possible only through dynamic methods since equilibration periods resulted in &a con- version of UF3 to UF,. As a part of a study of the stability of UFj (see section on "The Stability of UF3," this report), the reaction .3 1 UF; = ZUF, +ZU0 , for which . (aUF4)3/4 (aUO)1/4 D aUF3 was examined by measuring partial pressures of HF produced when Hp; was passed through a bed containing UFs; and UF, to establish the equilibrium UF4+%H2$UF3+HF. In one series of measurements beginning with pure UF,, the reduction re- action was allowed to proceed to completion, and from observations as the temperature was altered at several stages of completion, some deduc- tions could be made about the phase behavior. For example, a change in pressure drop across the reaction vessel as temperature was increased indicated that the eutectic temperature was 895 * 5°C, while a melting point determination at 60% conversion to UF3 gave a thermal arrest at 892°C, which coincided with cessation of bubbling; all the observations 126 ORNL-DWG 65—2524 1500 . 1400 1300 1200 H00 Flg. 6.4. The System UF4-UF3. 1000 : e oo L A JE. TEMPERATURE {°C) 900 I S L S W 800 ; \ s | i . oo UF, * UF {SOLID SOLUTION) 'I 600 ce——— — _ UF4 10 20 30 40 50 80 70 80 0 UF:5 UF5 (mole %) were consistent with a eutectic temperature of 892 * 2°C. The UF, solu- bility in UF3 appeared to be approximately 38 mole % at the eutectic, in good agreement with the previous report. The value obtained for the disproportionation reaction equilibrium constant, log Kp = 0.328 — 3.30(1000/T), is sufficiently low to suggest that UF,-UF; liquids should not change composition during short exposures to the conditions that pre- vail in our thermal analysis experiments. Accordingly, we obtained cool- ing curve data for UF,-UF; compositions ranging from 10 to 40 mole % UF; and discounted all but the highest liquidus temperatures for repeated runs. On the basis of our previous data and from the results of the re- duction of UF; by Hp, we have constructed the phase diagram of the UF,- UF3 binary system shown in Fig. 6.4. MSRE Fuel Fractionation Experiments. The homogeneity generally characteristic of multicomponent salt mixtures in the liquid state is progressively destroyed as the mixture undergoes gradual crystallization (see Table 6.2). The possibility that the MSRE fuel mixture, LiF-BeF;- ZrF,-UF; (65-29.1-5.0-0.9 mole %), might, on cooling in the MSRE drain tanks, experience sufficient solid phase fractionation to create poten- tially hazardous conditions has been examined via laboratory-scale ex- periments. Some 650 g of the fuel salt was cooled at rates approximating that expected of the entire drain tank assembly and that expected of the fuel alone, 3.46 and 0.387°C/hr respectively. In both cases the radiative cooling geometry was controlled to simulate as nearly as possible that expected in the drain tanks, even though it is realized that the hori- zontal AT profile in the cooling radiocactive fuel mixture may be sub- stantially different from that prevailing in the laboratory experiment. The concentrations of uranium were found to be identical at the top and bottom fractions of each of the two ingots, irrespective of cooling rate. A photograph of the ingot resulting from the slower cooling rate experiments is shown in Fig. 6.5. Chemical analyses of the salt speci- mens from each of the locations designated in Fig. 6.5 were obtained. 127 Table 6.2. Eguilibrium Crystallization Behavior of the MSRE Fuel Mixture, LiF-BeF,-ZrF,-UF,; (65-29.1-5.0-0.9 mole %) Temperature . - Crystal Density (°c) S0lid Phases Crystallizing (g/cmB) 434 2LiF+BeF, 2.167 431 21iF-BelF, + 2LiF*ZrF, 2.167, 3.163 416 2LiF-BeF, + 2LiF-2rF, + 7LiF°6UF, 2.167, 3.163, 4.770 350 2LiF*BeF, + 2LiF-ZrF, + 7LiF°6UF, + 2.167, 3.163, 4.770, BeF, 2.234 Constitution of frozen salt: 77.07 2LiF+BeF,, 14.29 2LiF*ZrF,, 2.57 7LiF+6UF,, 6.07 BeFs (mole %) For areas 1 to 7, the uranium concentrations were found to be 2.94, 3.49, 4.15, 3.32, 4.28, 4.62, and 5.74 wt %. In comparison with the nominal concentration of uranium in the MSRE fuel, 5.13 wt %, these experiments show a maximum increase of 23.4% in uranium concentration on very slow static cooling. A fact which accounts for the small degree of segregation of the uranium phases in these experiments is that, at the onset of crystallization of 7LiF-6UF,, simultaneous crystallization of the three solid phases 2LiF-BeF,, 2LiF-ZrF,, and 7LiF-6UF, is proceeding. In addi- tion, the volume of the liquid phase is being reduced steadily, so sharply, in fact, that in this experiment as well as in previous ones, some of the liquid phase is apparently occluded among dendritic-like crystals of the solidified phase, a phenomenon which helps prevent compo- sitional variation in the mixture. In the freezing of multicomponent mixtures, maximum segregation of crystalline phases takes place under equilibrium cooling conditions. The segregation represented by the results obtained in the fractionation ex- periments reported here represents in a practical way the nearest approach to equilibrium cooling that the MSRE fuel salt may experience in a single crystallization sequence. Removal of Lanthanides from MSRE Fuel by Zone Melting. The scheduled tests of the MSRE in early 1965 have increased the interest in all chemi- cal methods for reprocessing the LiF-BeF,-ZrF,-UF, fuel salt. Among the nonvolatile fission products that will form during the operation of the MSRE are the rare earths, which are poisons because of their very large thermal neutron cross sections. Preliminary phase relationships of the lanthanide trifluorides (ILnF3) and LiF (ref. 11) suggested that the distribution coefficient, k (ref. 12) (the ratio of the concentration of the impurity in the freezing solid to that in the liquid at equilibrium), for the lanthanide tri- fluorides in LiF would be significantly less than unity. The composition INCH — POSITION WEIGHT (% U) N OB WwN - Fig. 6.5. 2.94 3.49 4.15 3.32 4.28 4.62 5.74 PHOTO 67747A MSRE Fuel Ingot Resulting from Slower Cooling Rate. 129 of the MSRE fuel salt, LiF-BeF,-ZrF,-UF, (65.0-29.1-5.0-0.9 mole %), was chosen to lie Jjust outside the IiF primary phase field. Consideration of the LiF-InFi3 and LiF-BeF;-ZrF, phase diagrams suggests that the solu- bility of the rare-earth trifluorides in LiF-rich compositions of these mixtures is not highly composition dependent. By inference, the MSRE fuel salt might possibly be amenable to zone melting purification. In initial evaluations of the potential applicability of this technique, we first investigated the removal of rare-earth trifluorides from LiF with zone melting. 3 Ingots of LiF doped with CeFs, GdFs3, and Lul; were zone refined, using conventional zone melting apparatus. The results of anal- yses for rare earths showed that half of the rare-earth trifluoride was removed in 1 to 3 passes, and nearly all of it was removed in 12 passes. From graphs given by Pfann,l2 the effective distribution coefficient, kg, was calculated to be less than 0.2 for these experiments. A similar experiment was conducted with the MSRE fuel salt, LiF- BeF,-ZrF,-UF, (65.0-29.1-5.0-0.9 mole %), instead of LiF. The results of incomplete analyses indicated that at the midsection of the ingot, gener- ally less than half the LnF5 was removed in ten passes, and therefore that kp for the MSRE fuel was much larger than ki for LiF. The molten- zone temperature for the MSRE fuel was chosen as 460°C, just above the ligquidus of the fuel at 435°C, because, for practical engineering reasons, the temperature of the molten zone should be as low as possible. The inefficient separation of LnFj3 from the MSRE salt may be attributed partly to the large differences in the temperatures of the molten zone used in the LiF and MSRE fuel experiments. The temperature of the molten zone for LiF was arbitrarily chosen at 1000°C, about 150°C above the melting point of LiF. This AT helps assure better mixing, more rapid diffusion, and more vigorous convection circulation. Future experiments will be conducted to see whether more favorable conditions can be found. Reactions in Molten Salt Systems HF-HoO Equilibrium with Molten Fluorides Previously I'eportec':llz*'15 transpiration measurements of the reactions of HpO-HF mixtures in H, carrier gas with molten fluorides have been con- tinued. Measurements on LiF-BeF,; mixtures from 500 to 700°C have been com.pletedi and further measurements on LiF-BeF,;-ZrF, mixtures are in progress. 6 In general, the results have indicated that the following two reactions occur and that equilibrium can be readily achieved between the melt and the sparging gas: PEF [02~] B ; (1) Ho0 HaO0(g) + 2F(d) = 0°~(4) + 26F(g), Q 11 0 Il = — Do — H,0(g) + F~(a) = oE(d) + HF(g) , Q, 5 130 When sufficient oxide is present to saturate the melt with either ReQ or ZrOp, the concentration of oxide is fixed by the melt composition and the temperature; hence the quotient me,/P becomes constant: H,0 P2 H0(g) + BeF2(d) = BeO(s) + 28F(g) , G = oo ; (3) H,0 P2 Ho0(g) + 32rF,(a) = 32r05(s) + 2HF(g), Q, = % : (4) ' 2 Values thus far determined for Q‘O ’ Q’A functions of temperature and melt composition, are summarized in Figs. 6.66.8 for 2LiF-BeF, mixtures containing ZrF,. The equilibrium PI?IF/PH2O values for 2LiF-BeF, melts saturated with both Zr0O, and ZrF, are based on free energy data in JANAF.17 These melt compositions correspond ap- proximately to mixtures of MSRE flush salt (2LiF-BeF,) with fuel salt (LiF-BeF,-ZrF,;-UF,, 65-29.1-5-0.9 mole %). Values of Qs Qs Qs 8nd Q, ) Q’B’ and QZ, all of which are strong have been summarized more completely in a previous report ,16 which also ORNL-DWG 65-2538 TEMPERATURE (°C) 101 700 €00 ; 5?0 l\‘ 1\\ T | 5 ~ <. Cp! ‘\\aé‘p 2™ ~ & N NG Fig. 6.6. The Variation of o - N g. 6.6. e Variation of Qy g N ~2 4 for ZrOp-Saturated (2LiF-BeF,) + . \\\ \ie,p ZrF, Melts with Temperature and ZrF, . < ™~ ~ Concentration. The dashed curve rep- 5 ok N N ] S resents equilibrium PI%IF/PHzo values Q;‘“ ] \ \ N h for 2LiF-BeF,; melts saturated with T SN e —oes| both ZrO, and ZrF,. The line la- S S — S s~ 2fa/ve| beled 2LiF-BeF, shows the variation T SN \\ \kzsfi of Qg with temperature for such melts N N b | ] saturated with BeO. \ \ 6 S SO \\\ - < N NO.479 ] > \\ RN \ SN \Qoez\ N “ \o.ozoz 10 AN, 1.00 110 1.20 1.30 ide Conc ORNL-DWG 65-2540 1072 /L 5 = 600°C7 2 // ol TA 10~3 e o4 7 g ° e Fig. 6.7. Variation of the Ox- ¢ Ion Formation Quotient with ZrF, Ij; N entration and Temperature in = e - if- + Z M . 3 — (2L1F Bng) rF, Melts é} . —"500°C ] o - Q_% 2 // %?105 7 $ J,/ 5[ o5 2 10°6 0 X 02 03 04 05 06 OF (Pue / Pgo) [027] (mole 7kg) e ORNL-DWG 65-2541 005 §00°C | .02 /// o 0.0¢ - . o] (o8 0.2 0.3 .4 05 0.6 o7 ZrF CONCENTRATION (mole/kg) 500°C T Fig. 6.8. " droxide Ion Formation Quotient with ZrF, Concentration and Temperature in o /0 & / // | (2LiF-BeF,) + ZrF, Melts. ZrF4 CONCENTRATION (mole/kg} Variation of the Hy- describes the use of these results to derive the enthalpy and free energy of formation of BeF,, the activity coefficient of BeF,; in 2LiF-Bel, melts, and the activity coefficients of ZrF, in LiF-BeF,-ZrF, melts. Other information obtained from these results, more directly related to the MSR Program, will be summarized here. Removal of Oxide by HF Sparging. One important application of these measurements to the MSR Program is the use of the equilibrium quotients QA, QO, QB’ ang QZ to predict the maximum efficiency of HF utilization in the removal of oxide during salt preparation and during chemical reprocessing of the fuel salt. The calculations involved, which have been described previously,16 were performed on the ORNL CDC 1604-A computer.18 132 The course of removal of excess oxide, as water, from 2LiF-BeF, (the flush-salt composition) by equilibrium sparging with 0.1 atm of HF in H, at 500°C is shown in Fig. 6.9 in which the effluent pressures of H,O and HF', as well as the melt concentrations of oxide and hydroxide, are shown as a function of the standard volumes of sparging gas passed per kilogram of melt: W= vol. of sparging gas -l -] = R x T % molt weight Wole atm™ kg™) (T is the absolute temperature at which the gas volume is measured). In- itially the effluent pressures of HF and H»O0 should be low, but in- creasing, as the formation of OH occurs in the presence of solid BeO. This should be followed by a steady-state period during which the efflu- ent H,O and HF partial pressures and the dissolved oxide and hydroxide concentrations all remain constant as long as solid BeO is present. Finally, after all the solid oxide has been consumed, a peak in the effluent PH 0 is expected, followed by a sharp drop as the last of the 2 dissolved oxide and hydroxide is removed. The rapidity with which P initially rises to a steady value is H,0 inversely related to QA. The HF utilization thereafter is determined by, and varies inversely with, the value of QB. Finally, the rapidity of the removal of the dissolved oxide after the solid has been consumed is inversely related to both QO and QA' Since QA, 0’ and QB all rise with ORNL-DWG 65-2543 E 040 f-==mimm——- i‘PH?: _— | - ' L W 0.08 - l ] & . i3 0.06 |— ! o —L a 2 ' = 0.04 2 oozl ) ] Fig. 6.9. The Removal of Oxide i | Pug | from 2LiF-BeF, by HF Sparging with ° ! 0.1 atm HF at 500°, Assuming Equi- = e | librium Between Melt and Gas Stream. °52 4 I i L — | Qp, Qp, and Q4 values are 7.91 X o % - 2 : ' B,4 ? A 6 1 525, o R0t N 0 ] 107% atm, 3.9 X 107° atm mole kg™, 8g§ | i J and 4.5 X 10~2 mole/kg respectively. 89 o : J \ E : y T ! [ | o : ! ! : ! %3 0.40 \\ — _A A S R F ! ! EE0.0S | - ‘ \\ ‘ m ; : ' : o : % i 0 0.5 1.0 1.5 20 25 30 35 4.0 V/RT (kg melt)=W (mole kg~! atm™*) 133 temperature, HF utilization in equilibrium sparging should therefore in- crease with decreasing temperature. This trend is shown by the curves in Fig. 6.10, which show, for various temperatures and influent PHF values, the calculated overall utilization of HF to be expected during the removal of oxide from 2LiF-BeF, until a final concentration of 0.001 mole/kg (16 ppm) is reached, as a function of the amount of oxide initially present. These curves indicate that, provided agitation is sufficient to maintain equilibrium between the sparging gas and the melt, the removal of oxide from flush salt should be readily accomplished. For example, with 0.1 atm influent HF pressure and with more than 0.1 mole of oxide per kilogram (>1600 ppm), almost 90% utilization of the gas is predicted at 500°C. With reduced influent HF pressure, with reduced initial oxide content, or with increased temperature, the utilization of HF is de- creased, but there is a wide range of conditions over which greater than 50% HF utilization is predicted. In general, these estimates are sup- ported by experience gained during the purification of flush salt at the Y-12 production facility. Under typical conditions used there — ~0.063 mole of oxide/kg (~1000 ppm) initially present and ~0.09 atm influent HF pressure at 600°C — ~46% HF utilization was obtained. While this value is lower than the HF utilization predicted for equilibrium sparging con- ditions (~70%), it indicates that equilibrium sparging conditions were at least approached in the simple equipment used and that great improve- ment in efficiency could not be attained by improvements in equipment alone. The removal of oxide from melts containing ZrF, is expected to be more difficult, since QA, Q., and QZ all increase with the ZrF, concen- tration (Figs. 6.6—6.8). Data are not yet sufficiently complete to per- mit the calculation of curves, such as those in Fig. 6.10, for the MSRE fuel composition; but, from the present estimates of QZ for Z2LiF-BeF, + 5 mole % ZrF,, estimates may be made (Table 6.3) of HF utilization during the steady-state period when excess ZrOp is present. The removal of dis- solved oxide after the solid ZrO, has dissolved cannot be accurately pre- dicted because accurate values of QO and QA for the fuel composition are not yet known. ORNL-DWG 65-4187 100 INFLUENT HF PRESSURE (atm) 0.2 i -~ 90 ——=0.05 +1 = DS T | — ® N @) T 1| Fig. 6.10. The Expected Effi- _ 70—~ Ess ciency of HF Utilization in Removal E‘w of Oxide from 2LiF-BeF,; with Equi~ v I librium Sparging Under Various Con- § 5 ditions. L 40} o | ] o 5 10! 2 5 100 INITIAL OXIDE CONTENT {mole/kgq ) S N M 134 Table 6.3. Conversion of HF to H,O at Steady State with ZrQO, Present HF Converted (%) T (°C) Qg - BHF = 0.05 atm P%F = 0.1 atm P%F = 0.2 atm 500 ~0.06 477 58 68 600 ~0.3 20 30 42 700 ~1.6 6 10 13 Experience at the molten salt production facility is reasonably con- sistent with the estimates in Table 6.3; for example, with LiF-BeF,-ZrF,, 65-30-5 mole %, saturated with ZrO,, the HF utilization with an influent HF pressure of ~0.09 atm at 600°C was 13-25%. This low oxide removal efficiency was circumvented, however, by transfer of the melt at 600°C away from the Zr0Op in the meltdown vessel, followed by a short sparging treatment of the salt to remove the dissolved oxide (see below). The present data indicate that this method of purification can be made even more effective by decantation and purification of the salt at 500°C, since at the lower temperature the oxide is less soluble and that which is dissolved is more readily removed by HF. (A lower temperature was not used during present salt production because, in the unstirred vessels used, the temperature gradients might have been great enough to cause melt components to crystallize in the cooler regions.) Oxide Tolerances in the MSRE. The maximum amount of oxide contami- nation which can occur in the MSRE fuel and flush salts — and in their mixtures — without the precipitation of an oxide solid (i.e., the oxide tolerance) as well as the identity of the oxides which ultimately do precipitate are important questions which have been of continuing concern in the chemical development program for the MSRE. Previously reportedl®—21 studies of oxide phase behavior in the flush-salt—fuel-salt compositions thus far have clearly established that: 1. the stable solid oxide produced on contamination of the flush salt is BeO; 2. with addition of fuel salt to flush salt sufficient to give ZrF, con- centrations greater than ~0.01 mole/kg (~1% fuel salt in flush salt), ZrQ, becomes less soluble than BeO and remains the least soluble oxide with increasing proportions of fuel salt; 135 3. U0, is not precipitated by oxide contamination of these melts until a considerable fraction of the ZrF, has been precipitated; in addition, no evidence has been found that UO,-Zr0, so0lid solutions will be formed under reactor conditions. However, relatively little information had been obtained about the "oxide tolerances' of these melt compositions except that they were low. Precipitation experiments have been reported in which the amount of dis- solved oxide was estimated from the difference between the amount of oxide added to the melt as BeO and the amount precipitated as Zr0Op and UO». These calculations indicated that the solubility of BeO in flush salt is ~0.06 mole/kg (~1000 ppm) at 600°C and that the oxide tolerance of flush-salt—fuel-salt mixtures decreased with addition of fuel salt in about the way one would expect from the equilibria U0, (s) = U (a) + 202~(4d) , (5) 7r0s (s) = zZr4t(4) + 20%~(4) . (6) It was recognized at the time that the estimates of dissolved oxide con- centration were indirect and therefore not as reliable as measurements which might be based on the direct measurement of dissolved oxide or the direct measurement of removed water. The present transpiration measurements have given a more accurate means of estimating the oxide tolerance of these melts, since the method involved a more controlled addition (or removal) of oxide as Hy0 in measured amounts. The amount of oxide present in mixtures of (2LiF-BeF2) with ZrF,, saturated with Zr0O;, corresponding approximately to flush- salt—fuel-salt mixtures, is given by the ratio QO/QZ (Fig. 6.11). Simi- larly, for those compositions close to 2LiF-BeF, which are saturated with BeO, the oxide concentration is given by QO/QB. It is seen that the ORNL-DWG 65-2542 6 107! = —— T m _——PREVIOUS ESTIMATE (800°C) N r | | I N I | 5 s ~o 600°C ~ ~BeD SAT#J\RATION 7 Fig. 6.11. Estimated Variation ¢ 7 B SYn y’ of Dissolved-Oxide Concentration with 7 2 i ZrOz SATURATION Y, . . . o ZrF, Concentration in (2LiF-BeFz) + = o2 - . Pd ZrF, Melts Saturated with ZrOz. 5 B : 1~ ] b % 3 BeO SATURATION o N b 500°C, A N - / H> }\i—yzmz SATURATION §oS z ] = \ | %,. | g —— L1710 > 1073 | 0.004 001 002 005 04 0.2 05 1 2 ' Zrf; CONCENTRATION (mole/kg) 136 present estimates of oxide tolerance at 600°C are much lower than the previous estimate for the compositions near that of the flush salt, but that they are higher for compositions increasingly rich in the fuel salt (~1.3 moles/kg of ZrF,). The solubility of BeO in the flush salt at 600°C appears to be ~0.011 mole/kg rather than ~0.06 mole/kg. The oxide tolerance drops at first as ZrO, is precipitated with increasing Zrl, concentration, as expected from reaction (6); but, with further in- creases in the ZrF, concentration, it apparently increases. This could be the result of the formation of complex ionic species such as Zr02+, or simply the result of a strong medium effect on Q. While values of Q3 have not yet been determined at a ZrF, concentra- tion corresponding to fuel composition (~1.3 moles/kg of ZrF4), a crude estimate of the oxide tolerance of the fuel salt may be given by the amounts of oxide removed during purification of the fuel solvent (LiF- BeF,-Zr¥F,, 64.7-30.1-5.2 mole %) for the MSRE. Before treatment with HF, this salt was decanted at 600°C from the meltdown vessel to the treatment vessel leaving the unsoluble ZrO, in the meltdown vessel. Since these salt batches were almost certainly saturated with ZrO,, the amount of oxide subsequently removed from the decanted salt could represent an upper limit for the oxide tolerance. The average amount of HpO recovered from these batches was equivalent to 728 ppm of oxide. This value — which would be low if all the H,0 were not removed from the off-gas during purification or high if some ZrO, were transferred with the salt from the meltdown furnace — was used (Fig. 6.11), in the absence of other information, to extend the transpiration results to a composition corre- sponding to that of the fuel. The present results not only indicate that the oxide tolerance of the fuel at 600°C is higher than previously thought, but that it has a strong positive temperature coefficient. This supports the view that excess oxide could be removed from the fuel salt by inserting a cold finger in the MSRE primary system. No conclusive explanation can be given at present for the lack of agreement between the present estimates of oxide tolerance and those made previously on the basis of U0O;-Zr0O, precipitation experiments. While the previous material-balance calculations were indirect and not very accurate, the magnitude of the discrepancy is so great it appears that systematic errors must have been present in the precipitation meas- urements; for example, it is possible that the BeO added did not react completely because the BeO particles became coated with the precipitated U0y or ZrOp. Our present estimates of the solubility of BeO in molten 2LiF-Bel', are considered to be subject to an error of no more than iZO% at temperatures of 500-700°C. Advantages of On-Stream HF-H, Sparging of MSR Fuels Continuous HF-H; sparging of the fluoride fuel in future MSR systems may offer advantages which are sufficient to outweigh the disadvantages of accompanying increases in complexity and cost. 137 As described above, oxide can be fairly readily removed from LiF- BeF, by HF sparging. In a large reactor system it should be possible, with acceptably low sparging rates, to remove oxide from the fuel at a greater rate than that at which it is entering the system, especially since the larger reactors may use a fuel composition approximating 2LiF- BeF, with a few tenths of a mole percent UF;. This composition is more amenable to oxide removal by HF than is the MSRE fuel composition. Such a treatment would also continuously indicate, by the rate of water evo- lution in the radioactive off-gas, the oxide content of the fuel. Second, continuous HF-H, treatment of the fuel would provide a means of controlling the oxidation state of the fuel. Limited reduction of UF, to UFs by the chromium in INOR-8 is expected to occur. Perhaps more important is the possible oxidizing effect of the fission process, since fewer equivalents of positive charge are represented by the fission products in their ultimate valence state than by U4t, The effects of these and other possible redox processes could be overcome by HE-Hp sparging, which, on the basis of laboratory studies (cf. the following section), should readily maintain the uranium in the tetravalent state. Finally, another possibly very important advantage of HF-H, sparging is presently being examined. The conversion of I~ to HI by the equilibrium HF (g) + I=(a) = F(a) + HI(g) (7) appears, in preliminary laboratory tests, to be quite favorable. If the stripping rates ?rove rapid enough, this reaction would afford a means of removing 6.7-hr 35I, the principal source of 135Xe, before appreciable decay to 135%e takes place. Concerning the disadvantages of such a chemical treatment, since it will be necessary in any case to include provision for gas stripping in MSR primary systems in order to remove fission product gases, it is necessary here to consider only the additional disadvantages of the use of HF-H, in such a stripping operation. The principal corrosive effect of HF on INOR-8 in the presence of hydrogen will be the oxidation of chromium: crl(s) + 28F(g) = crFo(d) + Ha(g) . (8) This reaction should proceed, as in the case of UF,; oxidation of chromium, until the rate falls and is limited by diffusion of chromium through the metal to the metal-salt interface. It would probably be desirable, how- ever, to use two sparging operations, one sparge of HF-Hp, followed by H, alone or Hp-He. Although the solubility of HF in oxide-free fuel will be low, with oxide present, OH will be formed by the reaction H(g) + 0°~(a) = om~(a) + F(4) , (9) 138 and this provides a mechanism for the solution of HF in the liquid phase. This reaction is readily reversed by H, or He stripping. The most important question in providing for HF-H, stripping of the fuel concerns the side-stream flow rates and the gas flow rates required to perform successfully the chemical treatments desired. While the re- moval of oxide and the reoxidation of the fuel should be readily ac- complished with the flow rates already accepted for fission-product-gas removal, the gas and melt flow rates required to remove *3°I with a re- moval half-time short compared with 6.7 hr may be greater. The equilibrium quotient of reaction (8) can presently be estimated to be >30 kg/mole from laboratory tests on 2LiF-BeF, at 480°C. Difficulties with absorption of HI in off-gas lines have thus far prevented a closer estimate. This lower 1limit indicates that, with equilibrium sparging by H, containing 0.1 atm HF, the volume of sparging gas required to remove half the iodide present in the salt will be less than 6 liters per kilogram of salt. To achieve removal half-times of the order of 1 hr, it appears that less than 6 liters of gas per kilogram of fuel in the primary system must be sparged through the salt in 1 hr. TFor a large reactor containing ~600 £t of fuel salt, the required gas sparging rate would be of the order of 100 ft’/min. Although this would be equivalent to only a small fraction of the total flow rate of the fuel through the primary system pumps (e.g., ~8000 ft3/min), it indicates that a gas-melt contactor of considerable dimensions may be required. The Stability of UFj The reduction of UF, to UF3 by hydrogen has been studied to provide better thermochemical information on UF3. Corrosion equilibria that occur in molten-fluoride reactor fuels involve the reduction of UF, to UF3 by the structural metals that are used to contain the fuel. Al so, the difference in the free energy of formation of UF3; and UF,; determines the stability of UF3; toward disproportionation to give UF,; and U°. Both the corrosion behavior and the deposition of U° are of vital interest for long-term reactor operation. The results obtained from reduction with hydrogen improved the thermodynamic basis for defining the limits within which favorable chemical behavior is to be expected in molten- fluoride fuels. The treatment of the data on the pressures of Hr and HF in equi- librium with UF, and UF; has been summarized previously,?? and a complete version that includes a description of the method will appear elsewhere.?3 A flow of Hp was used, and equilibrium concentrations in the effluent gas were determined. Preliminary results and conclusions drawn therefrom2%»25 were not altered significantly as the study was completed. When crystalline UF; was reduced by hydrogen, a saturated solid so- lution of UF, in UF; resulted initially, as indicated by the notation (ss) in the equation: vs the Reciprocal of the Absolute Temperature for the Reaction UF4(C) + 139 18, + UF,(c) = UFs(ss) + HF . | (10) A tentative phase diagram showing the solid solution appears in Fig. 6.4. The equilibrium data were appropriately represented in terms of the ratio QR(SS) = PfiF/(PH )1/2, where P is partial pressure at equilibrium, by 2 + 58 K J Logro Qp(gq) = 3+95 0.057 — 70030 (11) as shown in Fig. 6.12. The subscript R refers to reduction in distinction from D for disproportionation that is used later. Accordingly, between 840 and 1160°K the free energy change was M'R(SS) = =RT 1n Q’H(SS) T = (32.03 = 0.27) — (18.08 + 0.26) kcal/mole . (12) L 1000 ORNL-DWG 64-9775 950 850 800 750 700 650 600 ] ] T T 7 T Fig. 6.12. Plot of logio QR 1 > Ho(c) = UF3(ss) + HF(g). Qg (atm'72) 0.8 0.9 1.0 11 1.2 1000/ 7 ok 140 For the same reaction, but with UF; as a pure crystal, logio Ky = 3.57 - 6.87(1000/T) , (13) where Kfi is the conventional chemical equilibrium constant. This re- lation, when combined with data from the literature, gave AFg(UF3)-— AF?(UF4) = 97.0 — 15.6(T/1000) kcal/mole (14) and - AFg(UF3) = 351.4 — 52.8(T/1000) kcal/mole , (15) where superscript f designates the standard free energy of formation. Comparisons between these results and earlier estimates, in terms of numerical values at 1000°K, are given in Table 6.4. The calculated activities of uranium, as given in the table, are a measure of how far the disproportionation reaction, 4UF; = 3UF, + U°, might proceed before equilibrium is reached. Viewed from the standpoint of the tendency of UF3 to disproportionate, the present results are reassuring. This tend- ency, though greater than predicted by Brewer,26 is much smaller than that based on other figures.27’28 The free energy of formation of UF; at 1000°K agrees with that estimated by Brewer. The agreement with Rand and Kubaschewski for UF, was a result of choice of literature wvalues. Measurements of the equilibrium pressures of HF and H, associated with the reduction of UF, in MSRE-type melts were also made. The results could be summarized in the form logio QR(d) = 3.995 — 9.329(1000/T) +3.'7'7(10-2)XUF4 + 2.09(10"2)(xBeF2 - 30.0) , (16) where X is the mole fraction and the 4 refers to dissolved UF, and UFs;. The effects of changing the UF, and BeF, concentrations are shown in Figs. 6.13 and 6.14. The expression for QR(d) strictly applies only for UF, concentrations up to 5 mole % UF,. Nevertheless, the changes are so small that it can probably be applied to any solution containing up to 10 mole % UF, and between 25 and 35 mole % BeF, without introducing errors of greater than 10% in QR(d). The relatively low value found for Q implies that a relatively low concentration of corrosion products corresponds to egui- librium in the MSRE, where alloyed chromium, rather than H,, is the re- ducing agent. Table 6.4. Calculated Activity of Uranium in Equilibrium with Pure UF, and UF3 at 1000°K . 26 27 Rand and This Work Brewer Glassner Kubaschewski 28 —AFfooo(UF3), kcal/mole 299 299 281 290 _AF{OOO (UF4 ) 3 kcal/mole 380.3 375 373 381 AF{OO()(UF_—;) - AF{OOO (IJF'4), kcal/mole 8lL.3 76 92 Al Calculated uranium activity 5 x 10~ 3 x 10™16 8 x 1072 2 x 10— ™t 142 ORNL-DWG 64-10646R / -4.00 / s /i’// Fig. 6.13. Effect on Q4 at %910 r(a) 860°C of Varying the Initial UF, ///// Concentration with Constant Sol- —a10 ’/;/ vent Composition (66 Mole % LiF, 34 Mole % BeFs). | 0 { 2 3 49 5 5 mole % UFy ORNL-DWG 64-108B12R -3.4 T @ 860°C | / -36 [—- 7 Fig- 6-14- Effect on QR(d) at 10949 Gy 3-8 I / 860°C of Varying the BeF, Content e " of the Solution with Constant UF, cap Concentration (5.0 Mole %). 25 30 35 40 mole % BeF, Since KR = QR(d)yUF3/7UF4’ where y is the activity coefficient with saturated solution as the reference state, the activity coefficient ratio, 7UF3/7UF4’ is readily obtainsble. From the solubility of UF3%° in the MSRE fuel, LiF-BeF,;-ZrF,-UF, (65.0-29.2-5-0.8 mole %), the activity coef- ficient of UF3 is found to be log Yyp, = -1.62 + 3.77(1000/T) . (17) In the same solution, the interpolated value of QR(d) is log QR(d) = 4.20 - 9.33(1000/T) . (18) In accord with the AFg for pure solids from above, log K, = 3.57 - 6.87(1000/T) . (19) Solving for 7UF4’ the resulting expression is log TuF, = —0.99 + 1.31(1000/T) . (20) Values of 7UF4 as calculated from this relation are compared with other estimates in Table 6.5. 143 Table 6.5. Estimated Activity Coefficients of UF,; in LiF-BeF,-ZrF,-UF,; (65.0-29.2-5.0-0.8 Mole %) Temperature Phase Diagram This Work UF,-U0; Equilibria® (°c) (~25% UF,) (~5% UF,) (< 1% UF,) 500 9 5 1.0 700 2.5 2 0.6 %C. F. Baes et al., Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1964, ORNI-3591, p. 46. If, for the disproportionation reaction in the same molten fluoride solution, UF,(d) = %UF4(d) L 70O, (21) the equilibrium quotient is written as o, [ &0 D" e, \ Xur, where x is mole fraction, the relations for T uF and for QR(d) can be A 1 L Q ’ (22) combined with the free energies of formation AFf(HF) and.AFf(UF4) to give —0.57 — 0.56(1000/T) . (23) log Qp For comparison, 0.328 — 3.30(1000/T) . (24) log Kb Either of these expressions is a downward revision of the implied extent of disproportionation when compared with previous qualitative observations of the apparently labile nature of UF; as encountered in many experiments in molten-fluoride fuels. For example, a melt containing 0.5 mole % each of UF3 and UF, at 1000°K is in equilibrium, according to QD’ with uranium metal at an activity of 1.5 X 10~7. The expression for the dissolved state, QD, also has a smaller temperature coefficient than could have readily been anticipated. 144 Viscosity in the LiF-Bel, System Viscosities of the LiF-BeF, system are being measured to provide significant physical data in support of molten-salt reactor technology and to systematically explore a series of fluoride melts which vary from the very viscous (pure BeF,) to the very fluid (pure LiF). These viscosities have been measured by determining the torque re- guired to maintain constant angular velocity of a cylindrical spindle immersed in the test liquid. The relisbility of the instrument (Brook- field LVT), which both measures the torque and maintains constant rotation rates, has been thoroughly checked with oils of known viscosity purchased from the National Bureau of Standards. Twenty compositions, with BeF, concentration ranging from 36 to 100 mole %, have been investigated, and the data for these are summarized in Table 6.6. All compositions, with the exception of pure BeF,, appeared to fit the equation n = K exp (En/RT) ) (25) where n = the viscosity in centipoises, K = a constant (also in centipoise units), Efl = the energy of activation for viscous flow, R = the gas constant, H i the temperature in degrees Kelvin. For more convenience in calculations, Eq. (25) was recast in the form log n = A/T-B , (26) where ™ ]l En/2.303R , o) I The viscosity data are probably accurate to within 10%. The temperature dependence of the viscosity (reflected in the constant A), however, is accurate to within 3%. As illustrated in the last column of Table 6.6, the viscosity de- creases rapidly with increasing LiF concentration. Also declining rapidly with LiF concentration is the energy of activation for viscous flow, defined in Eg. (25) (see Fig. 6.15). If we Tirst assume that 145 liquid BeF, is a disordered three-dimensional network of tetrahedra?° held together by beryllium-fluorine bonds, the changes in 7 and Efl with the addition of LiF become explainable. When LiF is introduced, bond rupture occurs, causing the network to break down into clusters. The bond rupture may be pictured as | —ge—F4ge— + LiF —+-—fle—F + IiT + F’;BF— . (27) I | | Table 6.6. Summary of Data and Constants for the Viscosity-Temperature Equation Log 1 = A/T — B (n in centipoises; T in °K), for the System LiF-Bel, Composition Temperature Range A 3 Viscosi?y ?t 600°C (mole % BeF.) Measured (°C) (centipoises) 100 702-111.2 See below™ 63,800,000P 99.01 692— 967 11,390 5.955 1.2,300,000P 98.01 632— 917 10,300 5.135 4,710,000P 97.00 601~ 897 9,540 4.595 2,160,000P 96.01 601— 844 8,995 4.29 1,020,000P 94.91 557— 837 8,620 4.13 550,000 93.01 572— 842 8,185 3.96 258,000 91..02 545~ 832 7,690 3.77 110,000 90.02 59— 882 7,405 3.65 68,000 85.00 539— 747 6,580 3.37 14,500 79.99 558~ 745 5,950 3.185 4,250 75.00 490~ 705 5,405 3.085 1,275 70.00 480~ 704 4,695 2.695 480 65.00 451— 724 4,150 2.45 | 200 60.00 437— 584 3,775 2.35 92P 55.01 389 584 3,390 2.22 46D 50.00 376~ 577 3,065 2.08 27b 45 .00 419— 638 2,605 1.71 18.7 £0.00 4b1~ 637 2,185 1.35 14.2 36.00 462— 600 2,060 1.305 11.3 SThe equation for pure BeF, is: log n = 14,148/T — 18.345 — 3.382 log T . bExtrapolated. 146 CORNL-DWG 65-4488 60 50 \ 40 \\\ Fig. 6.15. Energy of Activa- 3 N tion for Viscous Flow (Efl) vs Com- - 30 \\ position in the LiF-BeF, System. 20 ‘*\\\ — \ 10 \\\““ 0 0 10 20 30 40 50 60 70 COMPOSITION {mole % LiF) As more IiF is put into the system, still more bonds break, and the average cluster size decreases. If we assume (1) that the flow unit is proportional to cluster size and (2) that the number of flow units in- creases as the number of clusters increases, then it follows that vis- cosity decreases with increasing LiF concentration. Furthermore, as the size of the flow unit decreases, the energy required to surmount the potential energy barrier to flow also decreases; hence Efl (which is a measure of this required energy) decreases with increasing LiF concen- tration. If we imagine, alternatively, that the unit of flow tunnels through the potential energy barrier, then the smaller unit should get through easier; hence, a tunneling mechanism leads to the same con- clusion: Efl decreases with increasing LiF concentration. Further viscosity studies in the LiF-BeF, system are in progress or planned. Liquids being or to be measured are: (1) pure BelFs, using speclal spindles for very high viscosity determination, (2) liquids having higher concentrations of LiF, (3) solutions of UF,, ThF,, or ZrF, in the LiF-Bel, system. Fuel, Coolant, and Flush Salts for the MSRE When fully operational, the MSRE will have an inventory of about 26,000 1b of fused fluoride mixtures. Prenuclear test operation of the reactor is currently in progress with the coolant-salt mixture and a simulated fuel mixture. This mixture, known as the flush salt, without UF,, will be retained for subsequent use to flush and clean the fuel circuit when maintenance on that system becomes necessary. The actual fuel-salt mixture is scheduled for loading into the reactor facility in April 1965. The production of fluoride mixtures for the MSRE was begun in March 1964. The coolant- and flush-salt mixtures were made available to the MSRE in September 1964. The various fluoride mixtures required to make up the reactor fuel will be available in March 1965. 147 The Production Process The method by which MSRE fluoride mixtures are produced is based on techniques previously used for preparing fluoride mixtures for the Air- craft Reactor Experiment31 together with refinements realized from process development programs during the intervening years. Individual fluoride salts are loaded in desired proportions into nickel vessels, melted under flowing hydrogen, and further purified by gas sparging with anhydrous HF and hydrogen. Impurities which can be volatilized are removed in the process-gas effluent stream; those which can be converted to insoluble particles are removed by filtration during transfer of the molten mixture to its storage container. A comprehensive description of the production method and equipment has been reported.32 Average concentrations of im- purities which remain in MSRE fluoride mixtures produced thus far are shown in Table 6.7. Table 6.7. Fluoride Production for MSRE — Average of Chemical Analyses of Salt Batches Average Concentration Chemical of Impurities "Oxide Salt Mixture . (ppm) Removed" Composition (ppm) Cr Ni Fe o Coolant "LiF-BeF, (66-34 19 26 166 <5 1460 mole %) Flush "1iF-BeF, (66-34 16 39 123 <5 1650 mole %) Fuel solvent LiF-BeF,-ZrF, 2L 15 77 <5 728 (64.7-30.1-5.2 mole %) Fuel concentrate 7LiF-UF, (73-27 24 35 49 10 1700 mole %) Coolant and Flush Salt Mixtures The coolant and flush salts currently in use in the MSRE are binary mixtures of ‘LiF (66 mole %) and BeF,. All "LiF used in the production of MSRE fluoride mixtures was at least 99.99% pure 'Li. However, that material having the highest 61i content was used for the reactor-coolant mixture. Approximately 16,000 1b of the binary salt mixture was prepared for the initial loading of the reactor. Fig. 6.16. Loading of Flush Salt into MSRE Fuel Drain Tank. During the latter part of October 1964, approximately 5755 1b of the mixture 7LiF-BeF, (66-34 mole %) was transferred from 22 salt storage con- tainers to the coolant drain tank in the MSRE. An additional 9200 1b of this mixture was transferred from 36 containers into a fuel drain tank during November 1964. A photograph of salt transfer equipment in place during this fill operation is shown as Fig. 6.16. All reactor loading operations were accomplished routinely and without detectable beryllium contamination to the reactor facility. Component Mixtures for the MSRE Fuel Approximately 11,260 1b of the mixture "LiF-BeF,-ZrF,-UF, (65.0- 29.1-5.0-0.9 mole % respectively) is required for the initial fuel load- ing of approximately 75 ft> in the reactor assembly. To facilitate its preparation and to accommodate an orderly approach to critical operation of the reactor, the fuel mixture is being prepared as a combination of three mixtures. Since fissionable 235U will comprise only a third of the uranium inventory, it is combined with 7LiF as UF, that is highly en- riched in 235U to form the binary eutectic mixture LiF-UF, (73-27 mole %). 149 The balance of the uranium required for the fuel is prepared as a chemi- cally identical mixture with UF, that has been depleted of 23°U. Re- maining constituents of the reactor fuel are combined as a third mixture having the composition LiF-BeF,-ZrF; (64.7-30.1-5.2 mole % respectively). These component mixtures are commonly referred to as the enriched fuel concentrate, the depleted fuel concentrate, and the barren fuel solvent. Since these mixtures will be combined in the MSRE fuel drain tank, the final fuel composition will, in fact, depend upon the amount of 2'255U re- quired to bring the system to the critical and then to the operating condition. Production of the barren fuel solvent, which is very near completion, has received the same processing procedure as that used for the coolant and flush salts. The preparation of about 600 1b of the depleted fuel concentrate will follow by the same production method. However, pro- duction of the enriched fuel concentrate required special apparatus and procedures to provide for nuclear safety and for planned reactor oper- ations. A part of this material will also be cast into fuel-enriching capsules for small incremental additions to the fuel. Preparation of Enriched Fuel Concentrate. MSRE operations, as cur- rently scheduled, will require approximately 90 kg of <2°U. The pre- paration of this quantity of fissionable uranium in six equal batches, containing 15 kg of 43°U each, satisfactorily met requirements for nu~ clear safety and reactor fueling operations and permitted the economical use of existing process equipment. The reaction vessel was constructed from a 36-in. length of 6-in. IPS, sched-40 pipe (stainless steel 304L). The vessel was provided with an inner liner fabricated from 1/8-in. nickel sheet. Storage containers for the finished batches of enriched fuel concentrate mixture were con- structed from 36-in. lengths of 4-in. IPS, sched-40 grade A nickel pipe. All vessels were of welded construction except for loading ports and gas line connections. The overall design of these vessels provided for a liquid depth of about 29 in. in the salt storage container and a dry-mix depth of about 26 in. in the reaction vessel. Batches of raw material were blended and loaded into the salt treat- ment vessels by the Special Processing Group of the Y-12 Plant. Fach batch contained about 4.9 kg of ’LiF and about 21.6 kg of 23°UF,. The loaded vessels were then transferred to the fluoride production facility, where the salt mixture was melted, further purified, and transferred to salt storage containers by procedures similar to those described for the production of other MSRE mixtures. A photograph of the process equipment is shown in Fig. 6.17. The processed batches of enriched fuel concen- trate are in storage in the Y-12 Plant. Preparation of Fuel Enriching Capsules. Incremental additions of 235y will be made to the circulating reactor fuel by dissolving a fused eutectic mixture of “LiF-22°UF, (73-27 mole %) from a small capsule that will be lowered into the bowl of the fuel pump. The fuel-enriching cap- sules are constructed from 6-in. lengths of 3/4-in.-0D x 0.035-in.-wall Fig. 6.17. Fluoride Production for MSRE — Process Equipment for Enriched Fuel Concentrate Mixture. nickel tubing with hemispherical bottoms. The top plug of each capsule is penetrated by two 1/8-in.-OD X 0.025-in.-wall nickel fill tubes. For filling purposes, seven capsules will be connected in series by their fill tubes and clustered within & 4-in.-diam heating chamber. Some 154 capsules will be filled from a prepared batch of the enriched fuel con- centrate mixture. Each fill operation will be monitored by radiography using a Norelco 160-kv, 6-ma porteble x-ray unit and a TVX camera to con- trol the salt liquid level in the last capsule. The filled capsule clusters will be sealed individually in watertight cans and placed in storage until needed at the reactor site. Chemistry of Prenuclear Use of Fuel and Flushing Salts in the MSRE Compositional Analysis of MSRE Salts A routine sampling method was employed for obtaining representative samples of each batch of purified salt used in prenuclear filling and 151 testing of the MSR. Specimens of the molten salt were obtained by with- drawing 20 to 50 g of molten liquid into copper tubes on completion of purification procedures. Specimens were obtained from each of 61 batches of LiF-BeF, (66-34 mole %) flush salt and 21 batches of LiF-BeF,-ZrF, (64.78-30.06-5.16 mole %) fuel solvent and chemically analyzed. Because numerous inferences concerning the performance of the MSR will necessarily be made on the basis of chemical analyses of the salts, all the available analytical data on the flush salt and fuel solvent were subjected to statistical analysis. Results are shown in Table 6.8. Although the nominal compositions of the flush and solvent salts are LiF-BeF, (66-34 mole %) and LiF-BeF,-ZrF, (64.78-30.06-5.16 mole %) respectively, the compositions calculated on the basis of statistical mean values from the analytical data are LiF-BeF, (63.80-36.20 mole %) and LiF-BeF,-ZrF, (62.60-32.16-5.24 mole %). The analytical data are particularly inter- esting in that analyses of both salt mixtures indicate that LiF is low by 2.2 and 2.18 mole % while BeF, is high by 2.2 and 2.1 mole % Table 6.8. Chemical Analyses of MSRE Salts Analyzed Concentration 95% Confidence Limits Nominal Concentration Element (wt %) (wt %) (ppm) Flush Salt: LiF-BeF, (66.0-34.0 Mole %) (C-101 to C-161) Li 13.83 13.12 +0.11 Be 9.26 9.68 +0.05 F 76.91 77.08 +0.07 Cr 17.2 +3.05 U 34.0 *+6.58 Fe 137.2 +10.30 Fuel Solvent: LiF-BeF,-ZrF, (64.78-30.06-5.16 Mole %) (F-162 to F-182) Li 11.35 10.86 +0.11 Be 6.84% 7.25 +0.08 Zr 11.92 11.95 +0.10 F 69.89 69.94 +0.13 Fe 22.9 12 .34 Cr 16.6 +3.81 Ni 95.2 1.1 152 respectively. Reasons for the disparity in analytical data have not yet been rationalized. There is evidence from the purification-plant in- ventory data that the composition of the salt delivered into the reactor system is of the design composition, 64.0 * 0.25 mole % LiF and 36.0 + 0.25 mole % BeF,. Our concern regarding the apparent bias in lithium and beryllium analyses gave impetus to the recently renewed investigation of the LiF-BeF, system (see "High-Temperature Fluoride Phase Equilibrium Studies," this report). The results of this reexamination show that ac- curate thermal analysis in the LiF-BeF, system is in some cases an ef- fective analytical method. A sample of the MSRE coolant, taken from one of the batches loaded into the reactor, showed a phase transition tem- perature of 457.6°C. This is within experimental error of 457.7°C, the temperature of the peritectic reaction, which can occur only in compo- sitions richer than 65.5 mole % LiF. Chemical analyses of this material indicated its composition to be 63.63 mole % LiF—36.37 mole % BeF,. The thermal data indicate, however, that the material contained at least 65.5 mole % LiF, and thus confirm the composition indicated by the weights of the materials used in its preparation. The average discrepancy between analytical values for LiF and BeF, in LiF-BeF,; and LiF-BeF,-ZrF, mixtures is ~ —2.2 mole % for LiF and ~ +2.2 mole % for BeF,. This variance, apparently related to an intrinsic bias in the analytical method, will necessarily be considered in subsequent estimates involving uranium inventories both in reactor operation as well as in chemical reprocessing. Chemical Analyses of Fuel and Circuit Salts During Prenuclear Tests of the MSRE Throughout the prenuclear test period, specimens of the flush salt and coolant salt were obtained for chemical and spectrographic analyses. Approximately 6783 kg of LiF-BeF, (66-34 mole %) was charged into the fuel and coolant circuits, 4173 kg into the fuel circuit and 2610 kg into the coolant circuit. From chemical analyses of iron, chromjium, and nickel, changes in the amounts of these elements deposited from or dis- solved into the test salt were computed. Results are shown in Table 6.9. The values shown here are particularly interesting because of their anomalous imbalance. A net rise of 4.16 redox equivalents of corrosion- product fluoride is noted, whereas there is an apparent reduction of 18.6 redox equivalents of iron and nickel. Mean values of iron and nickel in the charge salts, 34 and 137 ppm respectively, are considered to be anomalous on the basis of previous investigations of the kinetics of re- duction of structural metal fluorides by the methods used in the routine purification of MSRE salts. We consider the imbalance noted in Table 6.9 as further evidence that iron and nickel fluorides were not present in the charge salt to the extent shown by the analytical data. Rather, these elements were present partly in the metallic state, and probably passed through the sintered metal filters as colloidally dispersed material. 153 Table 6.9. Changes in Concentrations of Structural Metal Fluorides in MSR Coolant and Flush Salt Circuits During Prenuclear Tests Total Mass ey e (@) (Em) () Gm) (@) Iron Chromium Nickel Charge salt 6783 +137 +931 +17 +117 +34 +231 (61 batches) Coolant salt 2610 52 =136 +25 +65 0 o) Flush salt 4173 =50 —209 +10 +42 0 0 Net change during —345 +107 o) operation Redox equivalents —12.36 +4.16 —6 . 242 aNi concentration remained constant throughout prenuclear test op- erations at ~7 ppm as compared with mean concentration of 34 ppm in pro- duction batches; this corresponds to a loss of 183 g of Ni from the 6783 kg of salt delivered to the reactor, or —6.24 redox equivalents. Results of oxygen analyses of the flush and coolant salt show (Figs. 6.18 and 6.12) the greatest variance. This deviation is believed to re- flect minor contamination of the salt after it is removed from the re- actor. Moreover, the samples are unfiltered and could contain variable amounts of suspended BeO. If the oxygen represented contamination by water and if the HF which would have resulted from hydrolysis had been permitted to corrode the INOR-8 walls, then a substantial increase in dissolved chromium would have been seen, but, in actuality, was not. To test whether a large bias might have been involved in the fluorination assay of oxygen concentration, a large sample of salt was taken for analy- sis by fluorination and by the Hp-HF stripping technique; the results were consistent within approximately 50 ppm. The conclusion seems to be that the oxide levels, considerably higher than those considered possible for dissolved oxide, represent the presence of small amounts of suspended BeO. The results shown in Figs. 6.18 and 6.19 are graphic evidence that excellent procedures have been employed during the prenuclear test period. The concentration level of chromium, a sensitive indicator for presence of oxidizing impurities (including HF produced by hydrolysis), is re- markably low, sufficiently low, in fact, to suggest that no measurable corrosion of the reactor hardware occurred during the approximately 1000-hr prenuclear test period. IMPURITY CONCENTRATION (ppm) 154 T Fig. 6.18. Chemical Analyses & of MSRE Fuel Pump LiF-BeF, Speci- & '_ mens. & '_ Z 1] Q Z o Q )— e @x 2 [+ 2 ORNL-DWG 65-4190 {000 A o] i /N 500 / \\c ,/ \ OXYGEN //) 200 \\\V/ 100 r\& IRON N7 I\ A N TN 7\ Y a, A\ 50 A / \ SNo- \ . ‘-_-—.// L J [ J o-0-0-0r" e CHROMIUM \L 20 ///\\\ [ ] io IA\ i \\ n \ II ‘\ — \\ NICKEL 5 'u ‘I-l]—=/ \l—-——l'/ ] 0 200 400 600 800 1000 1200 TOTAL TIME IN PUMP BOWL (hr) ORNL-DWG 65-41489 1000 500 | \ [\ ’,)1\ OXYGEN lt! \v/ \\ / 8 8 > \ A \ / \ TN / \ 50 N / . \ B! P *—o \‘zk \* (‘ 20 10 X Z N\ NICKEL 5 -——-—.—-—_fi'—.fl———T| o 200 400 600 800 1000 TOTAL TIME IN PUMP BOWL (hr) Fig. 6.19. Chemical Analyses of MSRE Coolant Pump LiF-BeF, Speci- mens. 155 Development and Evaluation of Equipment for Analyzing Radiocactive MSRE Fuel Samples The development and evaluation of equipment for preparing and analyzing radiocactive MSRE fuel samples>> were continued. In July 1964, this equipment was transferred from the mockup cell in Building 45005 to the High-Radiation-Level Analytical Laboratory (HRLAL). At that time, the equipment to be used in preparing and that to be used in analyzing the samples were assembled in the decontamination cell and in Room 120 of the HRIAL respectively. In November, the equipment was transferred to cells 5 and 6 to establish the arrangement of it that would ensure maximum efficiency in the use of the two cells. Persons working in the HRLAL were trained in the analytical methods and techniques to be used on MSRE samples. The training program was com- pleted in November. During the program, it became evident that several changes in the designs of the equipment were needed to make it more adaptable to remotely controlled operations. ©Since standby equipment had to be fabricated, the design changes made were generally confined to this new standby equipment. Fabrication of the new equipment was completed in January 1965. With the exception of the decoupling device for the transfer container, the equipment was placed in cells 5 and 6; the original equipment is in reserve. The new decoupling device will be located in cell 5 by mid- February; cells 5 and 6 will then be ready to receive radioactive MSRE fuel samples. No further modifications to equipment are planned. Sample Preparation The design of the new equipment for receiving, crushing, and dis- solving the MSRE fuel samples was modified somewhat from the original. With the exception of the decoupling device, most of it is adaptable to remotely controlled manipulations. The decoupling device was very diffi- cult to operate with master-slave manipulators, because the transfer con- tainer had to be in a vertical position in order to be placed in it. Therefore, the new decoupling device (see Fig. 6.20) was so designed that the transfer container can be decoupled when it is in a horizontal posi- tion. The new device consists of two parts, one of which holds the trans- fer container while it is being transferred from the unloading cell to cell 5. Several minor modifications were made to the other equipment used in preparing the samples. The ladle cutter (see Fig. 6.21) was modified by mounting the cutting tool on the motor mount. This will allow the cutting operation to be carried out with one manipulator. Also, 250 copper capsules ("eggs") were fabricated to contain the samples while they are being crushed; this supply should suffice for the first 12 months of MSRE operation. 156 PHOTO 68465 Fig. 6.20. Sample-Transfer- Container Decoupler. Fig. 6.21. Ladle Cutter. Sample Analyses The equipment used in analyzing the salt samples was also modified. The most significant modification was to the electrochemical cell used for the coulometric determination of uranium and the amperometric deter- minations of chromium and zirconium. The modifications were necessary in order to eliminate cross contamination and to reduce the time required to change electrodes, condition electrodes, and maintain the cell re- motely. Separate titration assemblies were fabricated for use in each of these methods. Other modifications that were made are discussed below. Fig. 6.22. Q-1348B Pipetter. Dispensing of Sample Solutions. The solutions of MSRE salt will be dispensed with a remotely controlled pipetter (see Fig. 6.22). A new 3-ml pipetter was installed in cell 6 and is working according to specifi- cations. Cross contamination among samples will be eliminated by using expendable Pasteur capillary pipets prefilled with 0.01% solution of red oil in Amsco. Iron, Nickel, and Molybdenum. A new type of filterphotometer (see Fig. 6.23) was fabricated for use in determining the corrosion products — iron, nickel, and molybdenum. This instrument is yet to be checked out; calibration curves are to be prepared for each of these elements. Fluoride. A new apparatus (see Fig. 6.24) was fabricated for the remotely controlled pyrolytic determination of fluoride and was placed in cell 5. It appears to be working properly; further determinations must be made to substantiate the results obtained thus far. The original fluoride apparatus is being held in reserve. Beryllium. The apparatus for the determination of beryllium by the photoneutron method was transferred in December from Building 3550 to 158 Fig. 6.23. Densichron. Fig. 6.24. Tube Furnace (Byrohydrolysis Apparatus). 159 cell 5 of the HRLAL. A fresh 24Sb source was placed in the counting chamber of this apparatus in January. A calibration curve is now being prepared for this determination. Chromium. For the amperometric titration of chromium, a new type of assembly (see Fig. 6.25) was fabricated that can be repaired by using master-slave manipulators. Also, the titrant-delivery unit was modified to reduce repair time. The titration assembly and the delivery unit were installed in cell 6; both are working properly. Additional titration assemblies and delivery units were fabricated and are being held in re- serve. Zirconium. A new type of assembly for use in the amperometric titra- tion of zirconium was fabricated, and, as in the case of the chromium Fig. 6.25. Electrochemical Apparatus. 160 method, the titrant-delivery unit was modified. This titration assembly and delivery unit were also installed in cell 6 and appear to work properly. Additional titration assemblies and delivery units were placed in reserve. Uranium. In the coulometric titration of uranium, difficulties were encountered in achieving a 5-pa background current. Therefore, a new type of assembly was designed and fabricated. A background current of 5 pa was easily obtained in titrations done with the new assembly, which was installed in cell 6. Additional assemblies were fabricated and placed in reserve. Development and Evaluation of Methods for the Analysis of the MSRE Fuel Development work has been concentrated on hot cell methods for the determination of oxide and of the reducing power of the fuel. Since the powdered sample prepared for the determination of major fuel constituents and corrosion products is exposed to the moist atmosphere of the dry box, it has been necessary to study alternate sampling techniques for these methods. Because of the almost insurmountable problem of processing these samples under a dry, inert atmosphere in the hot cells, the analysis of individual unpulverized samples was chosen as the only practical tech- nique for these determinations. Oxide An inert-gas fusion method and a procedure based on the hydro- fluorination of the molten samples have been studied. In the inert-gas fusion procedure,?* a salt sample of about 200 mg is sealed in a threaded graphite crucible which is induction heated by means of an rf-field con- centrator to about 2400°C to evolve the oxide as carbon monoxide. The carbon monoxide is oxidized to carbon dioxide and determined gas chroma- tographically. It was found necessary to make two changes in the modified Leco furnace which has been described earlier;35 these changes are shown in Fig. 6.26. First, the threaded graphite spindle was replaced with a tantalum spindle to reduce the blank and provide increased mechanical strength. Second, the glass envelope was enlarged and shielded from the concentrator with a mica shield to eliminate the problem of occasional perforating arcs to the envelope walls. In a modification of technique, the loaded capsules are preignited at approximately 1000°C for 30 sec to remove moisture adsorbed on the capsule and then heated to 2400°C to evolve oxygen from the sample. The preignition reduces the blank to approximately 30 to 50 ppm of oxygen. With the above modification, reasonable agreement was obtained between analyses by the Leco technique and the established KBrF, method.?® Com- parative results are shown in Table 6.10, 161l ORNL-DWG. 65-248A TO 1,04 /fij CONVERTER AND GAS CHROMATOGRAPH N+—CAPSULE MICA SHIELD R. F. CONCENTRATOR= f HELIUM COOLING WATER Fig. 6.26. Leco Induction Furnace Modified for Oxygen in MSRE Samples. 162 Table 6.10. Determination of Oxygen Oxygen (ppm) Sample KBrF, Inert-Gas Fusion® LiF-BeF,-2rF,-UF, 2455 2350 + 170 925 1040 = 70 NaF-LiF-ZrF, 4115 3860 + 340 2070 2165 *+ 145 3315 3170 = 260 LiBeF; 475 545 + 60 AlF, 2205 2010 + 340 ?Average of 4 replicates. In a technique designed to provide an optimum sample for inert-gas fusion, a platinum ampul will be evacuated and sealed with a fusible alloy plug. When the ampul is immersed in the molten fuel, the plug will melt and permit the evacuated ampul to fill. The ampul is shaped as shown in Fig. 6.27 so that the middle link of the filled ampul can be snipped free and inserted in the capsule with a minimum of atmospheric exposure to the salt. Ampuls have been successfully spun from 0.020-in.- wall platinum tubing, and the Metals and Ceramics Division is investi- gating fusible alloys. A ternary alloy of gold, silver, and about 10 wt % germanium may meet specifications, which include low oxygen content, melting point slightly below the temperature of the salt in the pump bowl (1200°F), and compatibility with the fuel. A second approach to the oxide problem is based on the hydrofluori- nation of the molten fuel according to the reaction 0%~ + 2HF(g) = H,0(g) + 2F . (28) Baes>7 has found that this reaction proceeds quite rapidly at 500°C and that the oxygen can be quantitatively removed by bubbling anhydrous HF, at a few hundredths of an atmosphere in a hydrogen carrier, through a few inches of the molten salt. On a 50-g sample of Li,BeF,, evolution of water is completed on the passage of a few liters of the purge gas. Oxygen is determined either by measuring the water evolved or the HF con- sumed by the hydrofluorination technique. In the only sample of MSRE flush salt analyzed by this method (a relatively small sample), an oxygen concentration of 350 ppm was measured vs 265 and 335 ppm obtained by the KBrF, procedure. 163 ORNL-DWG, €5-1877 Fig. 6.27. Cross Section of Ampul and Graphite Crucible to be 1fifi Used for Sampling and Determination /' of Oxides in MSRE Samples. 3/16in, ID i | CRUCIBLE 1732 in. ID Moo CRUCIB 20 mil PLATINUM WALL AMPOULE FUSIBLE ALLOY A conductometric procedure is being tested for the continuous monitor- ing of HF in the effluent gas from the hydrofluorinator. Either a coulo- metric Karl Fischer titrator or an electrolytic moisture monitor will be used to measure water. Sealed ampuls containing solution titrated with Karl Fischer reagent were irradiated in a cobalt source, and no change in the excess of KF reagent was observed spectrophotometrically after ir- radiation at 10® r. The electrolytic moisture monitor would prove more convenient if the interference of HF can be eliminated. In sampling the reactor fuel, 50-75 g samples will be taken in a welghted enricher ladle. The tops of the ladles will be cut off, and the sample and ladle will be inserted in a tightly fitting nickel reactor with a bubbler tube spring loaded against the surface of the salt. The reactor will be heated and purged with hydrofluorinating gas mixture at a temperature below the melting point of the fuel. The temperature will then be raised to melt the salt and permit the bubbler tube to penetrate the surface. Thus only a relatively small surface of unpulverized salt will be exposed briefly to the atmosphere. Reducing Power A new method, a modification of the hydrogen evolution method, is being tested for the determination of the reducing power.38 The fuel sample is dissolved in tritiated 6 M hydrochloric acid to evolve tritium in proportion to the equivalents of_reducing species in the sample. The evolved tritium is freed of radiocactive contaminants by passing it through a liquid-nitrogen trap, then ignited to water over heated copper oxide and collected on an Anhydrone [Mg(Cl04)s] column. The adsorber is then removed from the hot cell for scintillation counting. The method is inherently quite sensitive; the limit of detection is 10 ppm for a 10-min count, but the sensitivity will probably be limited by the blank from radiolysis of the dissolver solution. A statistical evaluation of the method with respect to zinc metal, iron wire, and uranium trifluoride was made. In each series tested, the precision of the method was less than 3%. A sampling technique has not been developed; however, it will probably be similar to the fusible plug ampul technique. 164 Electrochemical Analyses Electroanalytical methods are, in general, adaptable to in-line analyses and appear to be especially attractive with regard to the direct analysis of impurities (e.g., corrosion products) and other electroactive species in the molten MSRE fuel. The initial stages of this work, aimed toward the attainment of the above goal, have been described in a previous report.>° At the present time, two experimental setups are in operation. One is for investigations in the MSRE fuel solvent, and the other is for the coolant salt. Until recently, the investigations involving the controlled-potential voltammetric measurements in the fluoride melts included the use of a platinum quasi-reference electrode. The instability of the potential of this "reference" led to an attempt to develop a more dependable reference electrode system. The first reference electrode which was investigated is shown in Fig. 6.28. The potential of the cathodic limit of the MSRE fuel solvent was measured vs this electrode and vs the platinum quasi- reference electrode over a two-week period. The limit varied over a range of 0.04 v vs the nickel reference, as compared to 0.3 v vs the platinum electrode. The potential of the nickel reference was also not affected by evacuation of the electrolytic cell or by the static or flow- ing conditions of the inert cover gas. However, the electrode did polar- ize easily with the passage of small currents. Although zirconium oxide becomes a solid ionic conductor at elevated temperatures, the resistance of the tube was still about 4000 ohms at 500°C, the temperature of the melt. ORNL- DWG, €5-140 NICKEL ROD 3/32" DIA Fig. ©6.28. Nickel Reference Electrode for Molten-Fluoride En- vironment. ZrO, TUBE - 3/46" DIA 7777 T | T 5% BY WT. NiO IN LiF-BeF, 7T 7, BORON NITRIDE PLUG—* 165 Another reference which is presently being studied involves the Zr-Zr(IV) couple. It was found that if a zirconium rod was introduced into the MSRE fuel solvent, it acquired a potential corresponding to the cathodic limit of the melt, which is the reduction of zirconium. Current- voltage curves were run using the zirconium rod as a reference, and it was found that the potential of the cathodic limit varied over a range of only 5 mv over a two-week period. It is still necessary to provide a compartment for the zirconium electrode, since the zirconium metal is capable of reducing the other ions of interest in the melt. This has been done as shown in Fig. 6.29. It is planmned to evaluate this electrode and to determine if its stability compares favorably to that of the un- sheathed zirconium rod. Another problem which has been encountered in this work is the con- struction of a suitable indicator electrode. Metal electrodes such as platinum are not entirely satisfactory, since the metal being reduced from the melt appears to alloy with the metal of the electrode. ©Several designs of indicator electrodes using commercially available pyrolytic graphite have been used experimentally but were found to be unsatisfactory. At the present time, the electrode shown in Fig. 6.30 is being used. The electrode is prepared by depositing pyrolytic graphite at a high tempera- ture and low pressure on a spectrographic-grade graphite substrate, which is then encased in a hot-pressed boron nitride sleeve. This electrode has been used for the reduction of chromium in the MSRE fuel solvent. The current-voltage curves are shown in Fig. ©6.31. The relationship be- tween the peak current and the square root of the scan rate indicates that a two-electron reduction is taking place. It is planned to continue this investigation to determine if the relationship is linear between the peak current and the concentration of chromium in the melt. ORNL-DWG. 65-142 NICKEL ROD Fig. 6.29. Zirconium Reference 1/8" DIA Flectrode for Molten-Fluoride En- vironment. ZIRCONIUM RQD ——— 7/32" DIA 7/32" DtA | Leteleteleteni | g-*—BORON NITRIDE 2277 77k [k Pt (LA T T T T PP T T T T TP 7 7 =—ATJ GRAPHITE 9/32" DIA LiF- Bef, - ZrF, T LT LD LT AT L A AT T & i T AL 4 t i 166 The investigations in the MSRE coolant salt are, at the present time, concerned with a reduction wave which occurs in both the fuel solvent and coolant salts at about —1.2 v vs the platinum quasi-reference electrode. Although the species being reduced at this point has not been positively identified, some evidence has been acquired which indicates that the re- duction of hydroxide is taking place. To aid in the elucidation of the reductions occurring in this potential range, the controlled-potential polarograph ORNL model Q-1988 (adapted for fast scan) was modified to enable the use of the derivative circuitry. The derivative current- voltage curves thus obtained indicate that the reduction at —1.2 v is even more complex than previously envisioned. Investigations of this wave will be continued. ORNL-DWG, 65-143 Fig. 6.30. Pyrolytic Graphite Electrode for Molten-Fluoride En- vironment, NICKEL ROD - 1/8" DIA ORNL-DWG. 65-144 SPECTROGRAPHIC GRADE L GRAPHITE 2.3mM CrFs in LlF-Bng—ZrF4 3/16" DIA Electrode Area ~8 mm PYROLYTIC GRAPHITE BORON NITRIDE 1t0 2 mil LAYER t/4" DIA 40 uA