ORNL-3419 UC-80 =Reactor Technology MOLTEN-SALT REACTOR PROGRAM SEMIANNUAL PROGRESS REPORT FOR PERIOD ENDING JANUARY 31, 1963 OAK RIDGE NATIONAL LABORATORY operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION DISCLAIMER This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency Thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. DISCLAIMER Portions of this document may be illegible In electronic image products. Images are produced from the best available original document. Printed in USA. Price: $2.75 Available from the Office of Technical Services U. S. Department of Commerce Washington 25, D. C. LEGAL NOTICE This report was prepared as an account of Government sponsored work, Neither the United States, nor the Commission, nor any person acting on behalf of the Commission: A. Makes any warranty or representation, expressed or implied, with respect to the accuracy, completeness, or usefulness of the information contained in this report, or that the use of any information, apparatus, method, or process disclosed in this report may not infringe privately owned rights; or B. Assumes any liabilities with respect to the use of, or for damages resulting from the use of any information, apparatus, method, or process disclosed in this report. As used in the above, ‘‘person acting on behalf of the Commission’’ includes any employee or contractor of the Commission, or employee of such contractor, to the extent that such employee or contractor of the Commission, or employee of such contractor prepares, disseminates, or provides access to, any information pursuant to his employment or contract with the Commission, or his employment with such contractor. Contract No. W-7405-eng-26 MOLTEN-SALT REACTOR PROGRAM SEMIANNUAL PROGRESS REPORT For Period Ending Januvary 31, 1963 R. B. Brigge, Program Director Date Isscued JUR 3 - 1963 OAK RIDGE NATTONAL LABORATORY Oak Ridge, Tennessee operated by UNION CARBIDE CORPORATION | for the U.S. ATOMIC ENERGY COMMISSION ORNL-3419 THIS PAGE WAS INTENTIONALLY LEFT BLANK A} & iii SUMMARY Part 1, MSRE Design, Engineering Analysis, and Component Development 1. Reactor Analysis Selective freezing of fuel salt in the drain tanks could permit the MSRE core to be filled with concentrated fuel and premature criticality could occur. This situation was studied for the case of fuel containing no thorium. If the core were filled completely with concentrated fuel, fluid temperatures in excess of 2500°F could result even if all control rods were inserted. Insertion of rods near the time the core first be- came critical would, however, keep the reactor subcritical for a time suf- ficient to stop the fuel addition; stopping the addition would limit the fuel temperature rise to less than 25°F. Estimates were made of coolant salt activation by absorption of de- layed neutrons for both the normal coolant LiF-BeF, (66-34 mole %) and for LiF-NaF-KF (46.5-11.5-42 mole %). After 10-Mw operation, the activity of the normal coolant will be about 2.7 pc/cm® and that of LiF-NaF-KF would be about 2.5 pc/cm’. Expressions were obtained relating the reactivity effects of Xel3? to the reactor conditions and the xenon spatial distribution. For the MSRE and highly enriched uranium fuel, a reactivity decrease of about 0.6% was associated with an effective xenon poison fraction of 1%. Calculations were made of the reactivity effects associated with graphite shrinkage, fuel soakup, and uncertainties in salt and graphite densities. TFor highly enriched uranium fuel, a 0.34% increase in re- activity was associated with a 1% increase in average salt density; a 1% increase in average graphite density increased reactivity by 0.53%. The increase in average salt density associated with a 1% decrease in graphite volume as a result of shrinkage caused the reactivity to increase by 1.18%. 2. Component Development Two INOR-8 freeze-flange joints for 5-in.-diam sched.-40 pipe were successfully tested, one in the thermal-cycling test loop and the other in the pump-testing facility. The joint in the thermal-cycling test loop was thermally cycled a total of 103 times between 100 and 1275°F, at the bore, with a final helium leak rate of less than 5 X 107° cm?® (STP)/sec at the lower temperature. The Jjoint tests were completed, but the joint in the pump-testing facility will be left in place for observation after extended operating intervals. An MSRE prototype freeze valve control circuit was installed in the valve test loop for operational testing, and it performed satisfactorily- iv in the control of the transfer valve, which is typical of all MSRE valves, except the reactor drain valve. The reactor drain valve differs in the orientation of its flat surface and in the presence of a concentric emer- gency drain line inside the valve proper. The extra conduction of this line resulted in a larger temperature difference between the center of the valve and the surface during the freezing transient, requiring that the automatic controls be set at a lower temperature than desired. In- sulation of the thermocouple area is being tried as a solution to this problem. The design and fabrication of a prototype control rod drive was sub- contracted, and work on the drive was started. Two types of flexible con- trol rods were tested under realistic operating conditionsg. Excessive wear at the bends of the guide thimble was solved by the installation of rollers. Exact positioning of the bottom of the rud rewmains a problem. Two all-metal prototype heaters for 5-in. plpe were received, and testing was initiated. One heater was in excellent condition after oper- ating for 2600 hr above 1200°F. The other failed after 1260 hr, probably because of either abnormally high thermal stress or the presence of a con- taminant that will not be encountered in the reactor. The prototype cooling bayonels for removing fuel afterheat from the MSRE drain tank were thermally shock tested a total of 2257 without fail- ure. Six of the ten reactor-grade thermocouples were still intact after this very severe test. : Fabrication of the sampler-enricher system mockup and installation into the engineering test loop is nearing completion. Difficulties with the sealing of the flanged disconnect were traced to shresses produced during fabrication and were solved by annealing. Operation of the full-scale MSRE core model with water at 85°F was continued to make preliminary measurements of the internal flow distribu- tion and pressure drop through the corc., The difference in flow between mutually perpendicular channels was reduced by drilling holes in the sup- port lattice directly under the starved channels. The small radial depend- ence of flow was deftermined as beneficial in giving more nniform tempera- ture rise of the fuel. The over-all pressure drop through the MSRE core model was measured and was found to be essentially independent of the Reynolds number. - The engineering test loop was placed into operation after successful pretreatment of graphite with wvacuum and heat, and after treatment of the operating salt with H, and HF to remove oxygen. The graphite container access Jjoint seal was operated in a manner to reduce the heat removed from the joint. After 1540 hr of operation, the loop was shut down and graphite sam- ples were removed for metallurgical examination. Treatment of the salt in the drain tank with HF indicated the presence of at least two separate ) phases of oxide. Although some improvements were made in the sampling technique, the chemical analyses for oxide were not reliable and consist- ent. Measurements were made of the permeability of ETL graphite samples with area-to-length ratios ranging from 69 to 99 cm. The permeability to helium ranged from 1 X 10-2 to 3 x 10-? cm?/sec. A procedure for the remote replacement of a freeze flange gasket was demonstrated. Two men completed the replacement in 8 hr, and it was esti- mated that the replacement would require an additional 24 hr in the reac- tor. Additional tools were designed and are being fabricated for the maintenance of freeze flange FF-100 under its special conditions. A 1/6-scale model of the reactor cell was constructed for use in studying maintenance problems. Operation of the prototype fuel pump in molten salt was resumed, and the testing program was reinstated. A water test of the fuel pump was also run to determine the value of the radial force on the impeller for several pump operating conditions of interest. Endurance testing of the lubrication pump and the PKP fuel pump with molten salt were continued. A prototype model of a two-level single-point molten-salt-level probe has operated successfully for six months. Although the signal-to-noise ratio obtained is adequate, an effort is being made to reduce the noise level. Developmental testing of a continuous liquid-level-indicating ele- ment for use in measuring the molten salt level in the MSRE pump bowl was continued. Performance of two units is still satisfactory after one year of operation at temperatures between 900 and 1300°F. Performance of a bubbler-type molten-salt-level-indicating system, which simulates a sys- tem to be installed on the MSRE, has been satisfactory. A developmental temperature-scanning system was operated satisfacto- rily for 3000 hr. At the end of this period, the system began generating excessive noise pulses. The noise was determined to be due to oxidation of mercury in the lower switch deck. Negotiations were completed for the procurement of a special high- temperature, NaK-filled, differential-pressure transmitter. Thermon X63 was determined to be unsuitable for use as a heat-conduct- ing bond on the MSRE radiator thermocouple installation. Eight MSRE pro- totype surface-mounted thermocouples continued to perform satisfactorily after 3000 hr of operation on the MSRE Experiment Test Loop. Drift of six similar thermocouples remained at less that *2°F after 8000 hr at 1200 to 1250°F. Five out of ten thermocouples are still functioning after 2050 severe thermal cycles. vi Data obtained from tests of MSRE prototype surface-mounted thermo- couples on the MSRE pump test loop indicated that the thermocouples are greatly int'luenced by the heaters. The thermocouples could not be used for computation of reactor heat power or for precise measurement of the mean reactor temperature unless the heater power was maintained constant and a correction was made for bias in the thermocouple reading. Several sealing and potting compounds were tested for use in seal- ing the ends of mineral-insulated thermocouples and copper-tube-sheathed thermocouple extension cable. Excellent seals were obtained with Araldite epoxy compound and with a glass-to-metal hermetic solder seal. No accept- able seals were obtained with low-temperature-setting ceramic-base com- pounds. A compound which requires high-temperature curing is being in- vestigated for posgsible use in sealing the ends of indivlidual wetal- sheathed thermocouples. 3. MSRE Design, Procurement, and Installation No significant changes were made in design concept or in detail of any component or system. Design work, except for instrumentation, was essentially completed, and a design report giving all engineering calcu- lations and analyses of the system is being compiled. The layout of the instrumentation and control system remains essen- tially the same as previously reported. Three panels were added, location of wireways was determined, and containment penetrations were assigned. Additional instrument application flow diagrams for the chemical pro- cessing system, the fuel loading and storage system, aud the instrument air distribution system were completed. ‘Tabulations aud application dia- grams were revised to incorporatée recent design chauges., - Preliminary control system block diagrams were prepared, and crileria for control and safety circuitry are being reviewed. The design of instru- ment and control system panels 1is approximately 85% complete, and panel fabrication is 50% complete. The design of lonslrument air interconnec- tione was completed. Interconnection wiring designs for the annunciator system and for the Foxboro Electronic Consotrol Instrumentaltion System are nearing completion. Design work on thermocouple interconnections 1is under way. Location and attachment drawings for thermocouples in the fuel and coolant system and a tabulation of the 819 thermocouples in the reac- tor system were completed. Fabrication drawings for drain tank salt level probes and for oil system venturi flow elements were completed. Requirements for process ra- diation monitors were established, and design work on field installations for these monitors is in progress. Panel design is nearing completion. Design work on personnel monitoring installations is under way. Approximately 90% of the commercially available standard components for the reactor system instrumentation was received. Approval was ob- tained for procurement of the data system. vii An analog simulation of the reactor fill and drain system indicated that the use of restrictors in the bypass lines between drain tanks and the pump bowl would reduce pump bowl pressure transients during a dump to an acceptable level without an objectionable increase in the time re- quired to complete a dump. All modifications to Building 7503, which will house the MSRE, were completed. TFabrication of the fuel system flush tank, the coolant salt storage tank, and the steam dome and bayonet tube assemblies for the drain tank coolant system was completed. Other components are approxi- mately 85% complete. Difficulty in the manufacturing of the MSRE graphite was encountered that will delay delivery until after July 1, 1963. Procurement of miscel- laneous equipment arnd material for MSRE auxiliary systems is approximately 85% complete. Part 2. Materials Studies 4. Metallurgy A full-scale sample heat exchanger was successfully fabricated to test previously developed welding and brazing procedures. The 52 welded- and-brazed tube-to-tube sheet joints exhibited good weld soundness and complete brazes. Ultrasonic inspection techniques for the tube brazes were correlated with metallographic studies, and a 3/32-in-diam flat- bottom reference hole was selected as the standard for evaluating braz- ing of the MSRE heat exchanger. The MSRE heat exchanger core was suc- cessfully assembled and welded. Mechanical properties studies of random heats of reactor quality INOR-8 to be used in the MSRE indicated that these materials have signifi- cantly better properties than the design values established with the pre- viously available INOR-8. INOR-8 was approved for code construction by the ASME Boiler and Pressure Vessel Code Committee. The allowable stresses are reported. The CGB graphite bars produced for the MSRE moderator were found to meet specifications except that there were cracks and some bars had densi- ties as low as 1.82 g/cmB. Despite these conditions the graphite was found to have good mechanical strength and low permeation when exposed to salt. The salt permeation of cracked CGB graphite was tested at 150 psig and 1300°F and was found to be less than 0.1% of the bulk volume. Rapid thermal cycling between 390 and 1300°F did not damage the graphite or cause salt-impregnated cracks to propagate. The tensile strength was found to range from 5440 to 6500 psi when tested as a round bar or a ring. The strength of severely cracked specimens was as low as 1500 psi; how- ever, the material did not demonstrate notch sensitivity. viii N Sample control rod elements were tested in the control-rod-testing rig. They were thermally and mechanically stressed for 600 hr through approximately 11,000 cycles. The hot-pressed Gdp03-Al,05 cylinders cracked during this testing but did not crumble, and the metal container was not distorted. Cold-pressed and sintered cylinders of Gd,03-A1,0; mixtures contain- ing 30 wt % Al,0; were prepared by working with prereacted powder that had 95% of the calculated density. Shrinkage behavior at successive sintering temperatures caused distortion, apparently because of the for- mation of intermediate compounds by a peritectic reaction. The use of a prereacted Gd;03-Al,03 holding fixture was found to resolve the distor- _ tion problem. ] . 5. Radiation Effects : v The source of the previously encountered F,; in the cover gas of sealed capsules examined after an exposure of 1070 neutrons/em® wus dis- covered in recent irradiation experiments. The fluorine was evolv=d from the frozen fuel at room temperature as a consequence of radiation damage to the crystals. No evidence of radiation damage to the molten fuel was found, and the evolution of F, at low temperature appeared to be easily avoidable without appreciable changes in present plans for MSRE operation. Examinations of sealed capsules from earlier cxposures were continued. The results served chiefly to confirm previously reported preliminary findings and surmises. Analyses of gas samples that were accumulated in capsules operating at MSRE conditions gave reassuringly negative results in regard 1.0 the evolution of CF,. Results from all experiments, includ- ing gas analyses from a varilety ol uperaling conditions, make a strong case for the presence of CF, only as a secondary consequence of F, pro- duction. The absence of evidence of unusual loss of [fluorine in any form from the fissioning fuel at high temperatures was firmly established and confirmed bhoth thermodynamic predictions and the conclusions from irradia- \ tion tests since the earliest work on molten salt reactors. The post- exposure evolution of F, from the solids, which even yet has uol always 8 occurred when it might be expected, was generally not identified when- present in earlier hot cell examinations, presumably because the gas phase usually escaped and was never analyzed; fluorine was undoubtedly associlated with an earlier instance of an unidentit'ied smoke that was noted when loop sections ¢ontaining lrradlaled Luel werc scgmented. 6. Chemistry - Continued phase equilibrium studies of MSRE-type fluoride mixtures showed that in none of the complex solid compounds so far encountered in the LiF-rich region of the LiF-ZrF,-UF, ternary system does UF, partici- pate in a solid solution; however, the simple compounds UF, and ZrF, oOr UF; and UF3 do form solid solutions. The solubility of UF3 in MSRE fuels was studied and found to be more than adequate. Revised and lower values for the solubility of oxides in fuel mix- tures were correlated with recently established solubility products of 7Zr0, and UO,. Saturation limits at as low as 80-ppm oxide for precipi- tation of Zr0O, were indicated for melts like the MSRE fuel; the scaveng- ing action of ZrO, in protecting against precipitation of UO, was con- firmed. Favorably low vapor pressures for MSRE melts were determined experi- mentally. The low vapor pressures are in part explained by strong com- plexing of potentially wvolatile constituents, such as ZrF,, as reflected by calculated activity coefficients in the system LiF-ZrF,. Further confirmation was obtained of the adequacy of proposed MSRE startup procedures for removing moisture from graphite. Physically held water was driven off near 100°C, and a small burst of more strongly bound water was evolved near 400°C in a roughly reproducible pattern that was only slightly altered by changes in the heating rate. An experimental study of the reaction of CF,; with fuel mixtures re- duced by the addition of zirconium turnings showed that the CF, reacted with the reduced fuel at an accelerated rate when admitted to the fuel through a hollow graphite cylinder immersed in the fuel. The reasons for this behavior are not yet fully established. Experimental investigations involving fluorine were facilitated by the use of a new manifold for regulating and controlling the flow of fluo- rine gas. For example, XeF, was prepared as required., A facility for the production of pure UF; was improved. Four approaches to out-of-pile studies of irradiation-induced evolu- tion of fluorine from solid fluoride fuels are being investigated. Con- sideration is bein% given to the use of Van de Graaff electrons, beta radiation from Sr°°-Y%9, gamma radiation from a Co®® source, and x-rays from a high-capacity x-ray machine. All these radiation sources are amen- able to use over extended salt composition and temperature ranges, and most could be employed on molten salt 1f necessary. Work continued on the development and evaluation of methods for analysis of the radioactive MSRE fuel. A hot-cell mockup was used in most of Llls wurk, dlong with actual high-level-radiation hot cells, in order to simulate as closely as possible the conditions under which it will be necessary to analyze highly radiocactive materials, Improved means of analyzing fuel for oxides were studied. 7. I'uel Processing Work on the detailed design of a fuel-processing system for the MSRE was essentially completed. The locations of the equipment were estabw lished, and the flow sheel was changed to route the exit gas from the fluorine disposal system through the caustic scrubber. | THIS PAGE. WAS INTENTIONALLY - LEFT BLANK x1i CONTENTS SUMMARY &ttt it ittt ittt i e e e i e e e e iii PART 1. MSRE DESIGN, ENGINEERING ANALYSIS, AND COMPONENT DEVELOPMENT REACTOR ANALY SIS + it it ev et nnorcnanasnenessossscoessnsases 3 Analysis of Filling Accident .........ccuiit i innrnerennnns 3 Coolant Salt Activation .......c.eiii it iiin et veoeiononnesans 4 Reactivity Effects of Xel3% it ien s 4 Reactivity Effects of Changes in Salt and Graphite Densities .......iiviiivn e nnenans P e e e e e 6 COMPONENT DEVELOPMENT Freeze-Flange Joint Development ..........v.cu... e e e e e 10 Thermal-Cycling Test ..ttt ntoteeoerstesostneassonansas 10 Joint Tests in Pump-Testing Facility ....... ... nn 10 'lests of Freeze-Valve Control Circuit ........ccciiiiennan. 12 TransSter Valve i ittt o ittt estneoneenerseeneneeeneasnenas 12 Reactor Drain Valve .....ieiee e enenneeenrosoeraneonanss 14 Reactor Control Rods and Drives ........ . veiivoeereneaeennes 15 Control Rod Drive Prototype ..... ..ot ivivinnrnennneecanass 15 CONErOL ROA t'vvveevnvnoseensennennassennn e 17 Heater Tests v ii it it tni e et ot tovererocasocosenoesenosonoacesss 17 Pipe Heaters with Reflectlve Insulation ..... e e e 17 Drain Tank COOlers .e.eeeeeeeeeas R N T TR 23 Sampler-Enricher System MOCKUD «ovt ittt eenereorooaronoeoeoes 23 Flange DisCONNECtS v vttt it ittt it et et et ee et ennenennseneenens 27 Core DeveloDmMent . vttt ittt ittt ettt te st neneoetnntoeeeanneens 27 Internal Flow Distribution ..o v vt it it it i ineeeesennnees 27 COTE PreSSUTE DIUP v vv e e s tteenneneseenneetonnseeeannss 28 Engineering Test LooP (ETL) «vvvrerrnvennoneeunenneneennnonns 28 Results of the Graphite Pretreatment ..................c... 29 oo} o T 6; oTc ol = v I J 30 Operation of the Graphite Container Access Joint ......... 30 Graphite Sampling Via Removable Dry BOX ...vevevunrrnneens 32 Analysis of Solids Removed from the Graphite Container ACCESS JOINL .. vt v in it vt nnrnneossnnsesnenns 33 Chemical Analysis and HF Treatment of ETL Flush Salt ..... 33 ETL Graphite Permeability ...... i nnns 35 Xenon Transport in the MORE ... .t ittt treroneereneneenns 35 MSRE Maintenance Development ............ e e ae e e Veeo 35 Maintenance Study Model .....ivieeeernrennn. e e 37 2 Pump Development .......c.cviiiinrencnnes e ettt e e e 37 L xii Prototype Pump Operation and Testing .......covvveereennnn 37 PKP Fuel Pump High-Temperature Endurance Test ............ 40 Lubrication-Pump Endurance Test ....ivt i iiiiirnnnnnnnons 40 Instrument Development ..o v vt vt it vt ittt et et bttt et e nsneses 40 ' Single-Point Liquid-Level Indicator .........eveeveveneenn. 40 Pump-Bowl Liquid-Level Indicator ..........vuceeenennven.. 41 Bubbler-Type Liquid-Level Indicator .........cvvvvvvneenn. 43 P eMPETratUre SCaNNEY + vt vt ee ettt v e oe tsonvsnennoenenennenns. bdy High-Temperature NaK-Filled Differential-Pressure B = T 7 v Y 45 Thermocouple Development and Testing .....covvvevvvennan.. 46 MSRE DESIGN ....... B b e et e e s e e e e s e s e e e e e s e 49 Design Status .....cuvveeunnns N e e 49 REACEOY SYSteM vt vttt it ient tene cnntsnnnnnnnanns e 49 Instrumentalion and CONBIULlE .....vviiiiorannenen.. - uuds 49 Procurement and Installation Status ............. .0 iiunoen, 54 Subcontract Work at Reactor Site ......cveiiiennnnnnann, 54 Fabrication of Major Reactor Components ........ Ceh e - 54 MSRE Cost Estimate Revision ............ R R 62 PART 2. MATERIALS STUDIES METALIURGY «@vvevverenns e e P 65 Heat Exchanger Fabrication .........iiiie it i nneenenns 65 Sample TUDE BUNGALe 4 v v vt o trse s eroeonenssoonennernenas 65 Tube Joint Inspection ....... S e e ea e et acus s st eerren ‘e 65 MSRE Tube BUndle ....euovveeereernenennneas fe e et d e e 67 Mechanical Properties Of INOR-8 ... vt ittt ie v tnieennsennnness 68 ASME Boiler and Pressure Vessel Code Allowable Stresses £Or INOR=8 tivv e it neneieeonsnnensronasonns . 68 Reactor-Quality INOR- 8 ................................ o 68 Evaluation of MSRE Grabhite ...vvrtivenererennnnonnenn. R 70 Salt Permeation of MSRE Graphite .........civuivennnenn 71 . Thermal, Cycling of Salt- Tmpregnated Graphite et e .. 71 Tensile Strength of CGB Graphite ...vveiivreeerer e ennan T4 Nevelopment of Gd;03-A1,0; Control Rod Elements ............. 76 Sample Control Rod Element Testing ... v vt i iine e ninss 76 Sintering Characteristics of Gd;03-A1,05 ' Specimens .............. e et ettt 76 RADTATTION EFFECTS ittt v vt aetr oo seeeoneeennnennneoeanans SR 80 Postirradiation Examination of Experimental Assembly ORNL-MIR=47=3 .ttt erunvernoeeeneeeneennenneenanns 80 Postirradiation Examination of Experimental o ' o Assembly ORNL-MIR =47 =4 vttt eneneenneensnensennennisensse 80 xiii GAS ANBLYSES vttt tratonesaetootoets ot tasersoroesasonaes 80 Metallographic Evaluation of the Wall of a Capsule from Experimental Assembly ORNL-MTR=47-4 .. ...uiveeeeennn 84 Autoradiography of Unopened CapsSuUles ...oeereenruresvennas 87 Bubble Formation in CapsUles ....v vt vt et veneneneeenos 88 Chemical Analyses of Materials from Experimental Assembly ORNL-MIR=47 <4 vt nr ittt ettt neeeneeeeenennenns 88 Observations of Sectioned Capsules .........vevivrenunnnss 91 Control Experiments on the Fluorination of Graphite ...... 93 Conclusions Drawn from the ORNL-MIR-47-4 Experiment ...... 93 Molten-Salt Irradiation Experiment ORNL-MIR-47-5 ............ 94 Capsule Irradiation and Sampling ........c.cvt v iveecnnn. 04 Analyses of Gas Samples Taken During Irradiation ......... o7 Pressure Increase During Reactor Shutdown .........cvc0vvn 101 Effects of Flssioning in Frozen Fuel ...........couivnn.. 104 Pressure Rise Following Termination of Exposure .......... 106 Conclusions Drawn from Experiment ORNL-MTR-47-5 .......... 106 CHEMISTRY vt vttt cneennannenesnnsas f e bttt e e 108 Phase Equilibrium Studies ... ..t iri ittt in e inenenennns 108 The System LiF=ZrF =UF, -+ttt ittrtnnneseeennnnenecnnnnns 108 The System UFg-UF, « vttt entnutennnetnnesesenennaeness 108 The System LiF-UF3=UF, ... ittt iirrtnnecnnneenas 109 The System LiF-BeFo-UF3 «itueeu ittt cneneennooneeesesnnns 109 The System LiF-BeFo-ZrF4-UF3 «. ittt iennonreoenens 109 Core and Blanket Fluids for Future Reactors .............. 109 Crystal Structure of Xenon Fluoride ........ i ivinunnn. 110 Oxide Behavior in Mixtures of Flush Salt and Fuel Salt ...... 110 Physical Properties of Molten Fluorides ........coveevenecens 116 Vapor Pressures Of Fluoride MiXtures .......eeeeveeeeennns 116 Densification Of LiF Powder ... v ve ittt tiieoeronerenronnns 117 DENSIEY OF CrF o v v vt evnner it tn s st tanesenssensnsosess 119 Activity Coefficients of ZrF,; in the LiF-ZrF, System ........ 120 Graphite Investigations .......ii it it iiir it neannnnnneeeeens 122 Bvolution ul Volatile Impurities trom Graphite ........... 122 Behavior of Carbon Tetrafluoride in Molten Fluorides ......oviiiiin.as e e e s e ee e et b e e e e e 127 Production of Purified Materials .........c.ov i ininnrnennss 131 Pure Uranlum Trifluoride .......c.. ittt it einnnnn. 131 Preparation of Xenon Tetrafluoride .......cci i i 131 Fluorine Gas Manifold ... ..t iiinitiennneetnnnnenenens 133 Out-of-Pile Irradiation Studies .....c.cvveiiir i innnnennns 134 Fluorine Evolution from Solid Fluoride Salts Under Irradiatlon by Van de Graat Electrons ................... 134 Effect of Beta Radiation on Fused Fluorides .............. 135 Gamma Irradiation Experiments with a 10,000-Curie Cobalt~B0 SOUICE vttt it ittt e te e et eeie st nennnnnnas 135 X=-ray Irradiation of Inorganic TFluorides ,......ecvovveu... 136 Reaction of Fluorine with Reduced Fuel ...........c..c.... 136 7. Analytical Chemistry Methods for the Analysis of Radiocactive MSRE Fuel Analytical Hot-Cell Mockup for MSRE Fuel Analysis FUEL PROCESSING Xiv ----------------------------------------- --------- ---------------------------------------------- 138 138 . 140 141 PART 1. MSRE DESIGN, ENGINEERING ANALYSIS, AND COMPONENT DEVELOPMENT THIS PAGE WAS INTENTIONALLY LEFT BLANK 1. REACTOR ANALYSIS Analysis of Filling Accident The filling accident described previouslyt was reanalyzed2 because it is presently planned that the first charge of fuel will not contain thorium. The accident postulated consisted of premature criticality while the core was being filled with fuel salt containing more than the design amount of uranium. The increase in the uranium concentration could result from selective freezing of the salt in the drain tanks. The accident would be more severe without thorium in the fuel salt because, if present, it would be concentrated with the uranium and would act as a neutron poison. : In the reanalysis it was considered that the fuel contained fully enriched uranium (0.15 mole % U) and no thorium. With this fuel, it was found that control rod action alone was not sufficient to prevent a power excursion for all the cases considered. It would be necessary to stop filling the core because rod insertion would not always prevent the re- actor from becoming supercritical and developing excessive temperatures if the filling were completed. Even with 39% of the fuel frozen in the dump tanks, the core could be completely filled with the remaining fluid. If the frozen fraction contained no uranium and the core were filled with the remaining fluid, the fuel temperature would exceed 2500°F, even with all control rods in- serted. Power and temperature transients were therefore predicted, using an analog computer, to test various corrective actions. The results in- dicated that if the fill rate were limited to 0.5 cfm, excessive temper- atures could be prevented if the control rods were inserted and the gas- control valves were operated at the time the power level reached ~15 Mw. Inserting the rods limited the fuel temperature rise during the initial excursion to less than 25°F, and operation of the gas valves stopped the filling in time to prevent a damaging second power excursion. In an ex- treme case in which it was assumed that one of the three control rods failed to drop and two of the three gas control valves failed to operate, a second excursion resulted from '"coast-up" of the fuel level. In this case, with an assumed initial fill rate of 0.5 cfm, the fuel temperature rise at the hottest point was 150°F, and the final temperature was 75°F above the starting temperature. 1l0ak Ridge National Laboratory, "MSRP Semlann Prog. Rep. Aug. 31, 1962," USAEC Report ORNL-%369, p. 21. 27. R. Engel, P. N. Haubenreich, and S. J. Ball, "Analyeis of Filling Accidents in MSRE," USAEC Report ORNL-TM-497, Oak Ridge National Laboratory (in preparation). Coolant Salt Activation The activation of the coolant salt by delayed neutrons in the fuel- to-coolant heat exchanger was estimated using TDC, a multigroup neutron transport code. Important activities in the normal coolant, that is, LiF-BeF, (66-34 mole %), were 2.0 pc/cm® from 7.3-s N1€ and 0.7 pc/cm® from 1128 F2°. If LiF-NaF-KF (46.5-11.5-42 mole %) were used as cool- ant, the activities would be 1.1 pc/em® from N'€, 0.4 pc/cm® from F=°, 0.7 ue/em® from 15-h Na2%, and 0.3 pc/cm® from 12.4-h K*2, The present design provides adequate shielding for the radiation from the coolant system during operation with either coolant. After shutdown there would be no problem of radiation from the Lil- Bel,, cool- ant. With LiF-NaF-KF as the coolant, radiation 6 ft from the coolant drain tank shortly after 10-Mw operation would be only 100 to 200 mr/hr. Dispersal of coolant through a leak in the radiator would present a health hazard with either type of coolant. The toxicity of NaF de- termines the maximum permissible concentration of LiF-NaF-KF in the air, and the radiocactivity is relatively unimportant. Because of the beryl- lium, the maximum permissible concentration of the LiF-BeF, in air is about one tenth that of LiF-NaF-KF, Reactivity Eifects of Xel®S The total reactivity loss that will result from Xe'®> poisoning during high-power operation of the MSRE will depend both on the total amount and the spatial distribution of xenon in the fuel salt and in pores in the graphite. The reactivity loss and the poison distribution can be related theoretically, and the relation is most conveniently ex- "pressed in terms of a reactivity coefficient and an importance-averaged xenon concentration.® According to first-order perturbation theory, the weight function for the poison concentration is proportional to the product of the thermal flux, ¢_,, and the thermal flux adjoint, ®5 thus 2 jNg N aV+fN LOre_ AV * N (1) Xe ? £¢2¢2 av 3B. E. Prince, "Methods of Computing the Reactivity Effects of Distributed Xenon, Graphite Shrinkage, and Fuel Soakup in the MSRE," USAEC Report ORNL-TM-496, Oak Ridge National Laboratory (in preparation). where N is the importance-averaged concentration per unit reactor vol- ume, ®and Ns are the local concentrations per unit volume of graphite and sa{%, respectively, and the integral limits g, s, and R refer to graphite, salt, and reactor, respectively. The quantity N. is also the uniform equilibrium concentration of xenon that produces thé same re- activity change as the actual distribution. In determining the total reactivity loss, a third quantity is often utilized, the effective thermal poilson fraction, Py.. This is the number of neutrons absorbed in xenon per neutron absorbed in U235, also weighted with respect to neutron importance, and is given by N. [ o e_av . o-Xe-}(e R 2 2 P, = ) (2) Crsl o 2 4% % (21,070, + Z5_0,0,) AV whexre Z;;E = macroscopic absorplion cross sectlons of U3 for fast (L) and thermal neutrons (2), respectively, OXe = xenon microscopic thermal-neutron absorption cross section. * The relation between total xenon reactivity and PXe is given by: (zl 0 ¢ + X2 o7 ) av 25 2 2 % p PXe 2 (3) R 1 = é(vthbld)l + vthb 0 ) dv where v is the number of neutrons produced per fission and 1’2 are the macroscopic fission cross sections for fast (l) and thermal neutrons (2), respectively. Thus, given knowledge of the xenon distribution, deter- m%ning*the xenon reactivity involves three steps; namely, calculation of Nyos P> and &/k from Egqs. (1), (2), and (3), respectively. Alterna- tively, these relations may be used in the reverse sense in attempting to infer the distribution from reactivity measurements at power. The coefficients in the above equations were evaluated for the case in which the fuel salt contained 0.15 mole % UF4 as highly enriched uranium. The results were Sk * S Sl 0.634 Poe (L) * Pre 4.08 x 10 Ny, s ¥ . . . - where N is the atoms of xenon per cubic centimeter times 10 =%. The axial and radial weight functions are plotted in Figs. 1.1 and 1.2. UNCLASSIFIED ORNL-LR-DWG 78435 7 - - — CORE BOUNDARIES - LT IN SRR RELATIVE REACTIVITY IMPORTANCE larkitrary units) -10 0 10 20 30 40 50 60 70 80 ‘AXIAL POSITION (in.) Fig. 1.1. Relative Reactivity Importance of Xenon Concentration Versus Axial Position. Reactivity Effects of Changes in Salt and Graphite Densities Factors that influence the reactivity effects of graphite and salt density changes are graphite shrinkage, fuel soakup, and uncertainties in measured values of the densities at design conditionsg. When changes in the over-all dimensions of the core can be neglected, reactivity changes with density may be calculated using a reactivity coefficient and a weighting function that expresses the relative importance oil' the density change with respect to position.® The latter is useful when the density changes occur in a nonuniform manner throughout the core. The calculation is similar to that for determining the temperature coef- ficient of reactivity;4 in fact, the temperature coefficient is -the sum #0ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962," USAEC Report ORNL-3%369, p. 33. UNCLASSIFIED ORNL-LR-DWG 78436 5 LN\ RELATIVE REACTIVITY IMPORTANCE (arbitrary units) 0 5 10 15 20 25 30 RADIAL POSITION (in.} Fig. 1.2. Relative Reactivity Importance of Xenon Concentration Versus Radial Position. of the effects of density and thermal spectrum changes. The results of calculations for fuél salt containing 0.15 mole % UF4 were N x ok ) © = 0.343 (5—) S (5) aN _ x Sk _ 8 k - 0'533 (N ) 2 g where NS and N_ are the effective homogeneous densities of galt and graphite, respgctively, per unit rcactor volume, and (SN/N)" _1is the importance-averaged fractional change in density. The latt8?Cis given by 8 J Gy (r,2) % av * R 5s8 J'GN(r,z)dV R (29) (6) where G, is the spatial importance function. The axial and radial weight functions for fuel ‘and graphite density changes are plotted in Figs. 1.3 and 1..4. UNGLASSIFIED ORML~LR=DW} 7R437 | i N - CORE BOUNDARIES — - n (63 (o] 7 \ / ‘!_‘: o N / \\ GRAPHITE (’ \ 7 N Y FUEL SALT \ \ \ A n 0 /% by \ -{0 0 0 20 30 40 50 69 70 80 AXIAL POSITION (in.) o [$]] \\\ N \\ RELATIVE REACTIVITY IMPORTANCE (arbitrary units) o N ~ \\ Fig. 1.3. Relative Reactivity Importance of kractional Increases in Salt and Graphite Densities Versus Axial Position. - The results given above can be used to calculate the reactivity effects of uncertainties in the measured densities of the materials at design conditions. In order to calculate the effects of graphite shrink- age and fuel soakup, however, some specific situation must be considered. If shrinkage were uniform in the transverse direction across a graphite stringer and if the center of the stringer remained fixed during con- traction, gaps would open between stringers and fill with fuel salt.. The homogeneous density of the graphite would remain constant; however, the effective salt density, N, would increase. If \A and v_ represent vol- ume fractions of salt and graphite in the lattice, the f%actional change in salt density would be given by BVS = —0v_, and the associated changes in salt density and reactivity would be N ov v \'2 v e—=-£ £ K (7) N v v v v 1 s S s g S UNCLASSIFIED ORNL-LR-DWG 78438 2.5 @ . & pos | . > - - 5 2.0 ’,’ \-\ = / N 5 // \\ w o~ / N\ s Y \\pRAPHWE 2 15 < S \ 8 \ 2 \ \ - ‘\\\ \ > 10 \ e FUEL SALf\\\\\\ \ 3 N [ w \ \\ > N % 0.5 N\ _ N w ‘\ N\ o N\ AN AN N 0 \\\k 0 5 10 15 20 25 30 RADIAL POSITION (in.) Fig. 1.4. Relative Reactivity Importance of Fractional Increases in Salt and Graphite Densities Versus Radial Position. and 8k g N * = = 0.343 < v fl> =1.18 £, (8) where f_ is the fractional decrease in graphite volume as a result of shrinkage (a function of position), and the starred quantities are importance-averaged values of the indicated functions. Equation (8) also applies to the case of fuel soakup if fl is replaced by f2, the fraction of the graphite volume filled with salt. 10 2. COMPONENT DEVELOPMENT Freeze-Flange Joint Development Two INOR-8 freeze-flange joints for 5-in.-diam sched.-4O pipe were successfully tested, one in the thermal-cycling test loop and the other in the pump-testing facility. Both joints were subjected to the antici- pated extreme reactor conditions. The results of these tests are de- scribed below. Although no specific future testing is planned, the flanged joint in the pump-testing facility will be used to observe the behavior of freeze flanges over extended operating intervals. Thermal-Cycling Test The freeze-flange joint installed in the thermal-cycling test loopt was thermally cycled 103 times; an oval ring gasket was used for the last 60 cycles. Typical cycle temperatures are indicated in Fig. 2.1. Repre- sentative leak rates measured during these cycles are listed in Table 2.1. The performance of the freeze-flange joint during thermal cycling indicated the following: 1. The leak rates at both the upper and lower cycle temperatures were less at cycle 103 than at cycle Lk. 2. Both oval and octagonal rings make acceptable gas seals. 3. The leak rate at the lower cycle temperature always exceeded the leak rate at the upper cycle temperature for both types of ring gaskets; that is, the leak rate decreases with increasing temperature. 4. The joint maintained acceptabie salt and gas seals under high- temperature operation (1300°F), repetitive cycling, and severe tempera- ture transients (100°F per minute for 6 min). No future operations are being planned for the thermal-cycling test loop. Joint Tests in Pump-Testing Facility A 5-in. INOR-8 freeze-flange joint with an oval ring gasket was thermally cycled six times in the pump-testing facility.€ A cycle con- sists of preheating the facility for 24 hr, introducing the molten salt in 1 hr, circulating the molten salt, stopping the circulation for 1 hr, ) 10ak Ridge National Laboratory, "MSRP Quar. Prog. Rep. July 31, 1960, " USAEC Report ORNL-301L4, pp. 24-25. 20ak Ridge National Laboratory, "MSRP Sémiann. Prog. Rep. Feb. 28, 1961, " USAEC Report ORNL-3122, p. 51. ' 11 UNCLASSIFIED ORNL-LR-DWG 78803 1400 CYCLE NO. 90 SALT BULK TEMPERATURE = A 1385 °F o—eo ° 1200 " } [ ] 1000 \F‘fl—-BORE TEMPERATURE — [ J ¢ 800 \ " \ g ° : \ oy -9 ° 2 600 S \ - ® ./ \ 400 // \s. RING GASKET \\\\ TEMPERATURE PREHEAT [ ] / ! =g i SALT OSCILLATION 1 l \' « PIPE HEATERS ON ~n§::::::::8 PIPE HEATERS OFF o L_ | _%(;} 0 4 8 12 TIME (hr) 20 ' 24 Fig. 2.1. Typical Temperature Cycle During Thermal-Cycling Tests of an INOR-8 Freeze~Flange Joint in 5-in.-diam Sched. -40 Pipe. and dumping the salt. The last cycle had an operating time of 82 days with a bulk salt temperature of about 1225°F. than 4.0 x 10-6 cm3 of helium per sec. All leak rates were less The test in the pump-testing facility was less demanding on the flanged joint than the test in the thermal-cycling facility because of the fewer cycles and generally lower operating temperatures. The flange temperature distributions obtained in the two testing facilities systems are compared in Fig. 2.2. The data indicate that the oscillating fuel flow used to heat the flanged joint in the thermal-cycling loop produced a good simulation of the conditions obtained with continuous flow in the pump~testing loop at reactor flow conditions. 12 Table 2.1. Helium Leak Rate During a Thermal-Cycling Test of a Freeze-Flange Joint Bore Temperature® (°F) At Minimum At Maximum Total Helium Leak Rate Cycle No. Cycle Temperature Cycle Temperature {em® (STP)/sec] Ly 100 7.5 x .1o‘LL 1080 3.1 X 10'L+ by | 100 2.2 % 1077 5 150 | 1.8 x 10”7 67 - 1280 6.2 x 1071 8l 105 1.5 x 1077 86 1275 1.2 x 1077 103 1245 4.5 x 107" 100 2.3 x 1077 %a11 temperatures taken with system at thermal equilibrium and represent an average of temperatures at two points 180° apart on the bore. Tests of Freeze-Valve Control Circuit The prototype control circuit for a prototype freeze valve was in- stalled in the valve test loop for operational testing. The coutrol unit consists of a basic circuit into which several individual "off'-on" modules may be plugged. Each of the five modules for each valve is actuated on a signal from a thermocouple mounted on the valve and serves as a relay to control the valve heater and cooling-air supply. The module set point is adjustable uvn each unit. The control of each valve is adjusted to produce a specific temperature distribution near the valve for on and off positions, respectively. The results of testing this control circuit on two quite different freeze valves are reported below. Transfer Valve Results of tests indicate that the type of valve to be used to con- trol the transfer of salt between storage tanks will have a melting time of less than 5 min and a freezing time of less than 15 min. The valve 13 UNCLASSIFIED ORNL-LR-DWG 78804 1400 1200 1000 \ \ SALT FREEZE TEMPERATURE 800 Vfis 600 = ‘ W\\\dfl:;_—,g ® DATA OBTAINED IN PUMP-TESTING FACILITY WITH A BULK SALT 400 |— TEMPERATURE OF ~ 1225 °F — TEMPERATURE (°F) A DATA OBTAINED IN THERMAL -CYCLING TEST LOOP WITH A BORE TEMPERATURE DURING SALT OSCILLATION OF {300 °F V DATA OBTAINED IN THERMAL-CYCLING TEST LOOP WITH A BORE 200 — TCMPERATURE DURING SALT OSCILLATION OF 1200 °F — 0 2 4 6 8 10 12 DISTANCE FROM CENTER LINE OF PIPE (in.) Fig, 2.2. Comparison of Temperature Distributions of Freeze-Flahge Joints Tested in the Thermal-Cycling Test Loop and in the Pump Testing Facility. - is oriented so that the flat surface is in the horizontal position. The loop salt temperature was held at 1225 to 1250°F during the tests. The heat input required for melting was 1500 w, the "fast-freeze' air flow required to establish the frozen zone was 35 scfm at 8.5 psig, and the "hold-freeze" air flow required to maintain the frozen seal was less than 7 scfm at the same pressure. The temperature distribution across a valve in the normal "hold-freeze' condition is shown in Fig. 2.3 for two dif- ferent air flows. Tests have shown that if the air flow is increased without a compen- sating change in the heat input to the pipe on either side of the frozen zone, the length of the frozen zone will be increased and there will be an increase in the time required to thaw. If such compensation is made, however, only a small increase in the time required to thaw will result. For example, the ‘time required to thaw increased from 3.75 to 4 min for an increase from 3.6 to 5.4 sefm, with heat inpul compensation. 14 UNCLASSIFIED ORNL-LR-DWG 78805 1200 - N | - A4 1000 \\%fl - 3.6 scfm AIR FLOW: / . NERY% 600 - -\ ERRNY4 \/ 6 4 2 0] 2 4 6 LONGITUDINAL DISTANCE FROM VALVE CENTER LINE (in.) TEMPERATURE (°F) Fig. 2.3. Temperature Distribution Across a Freeze Valve in the Normal "Hold-Freeze' Condition. Reactor Drain Valve Difficulty was encountered in operalluu ul Lle mockup of the rcactor core drain valve with the control circuitry described above. This valve is located within the core thermal shield, and there is no means for directly controlling the heat on the adjacent piping. It ic installed with its flat surface in the vertical plane to facilitate remote mainte- nance of its heater. There is also a l/Z-in.-diam emergency drain line installed within this valve that extends from the core ltank to a point just below the valve. Preliminary tests indicated an excessive freezing time, and an x-ray photograph taken with the valve in the frozen condition showed a large vold above the frozen zone of the salt. Gas was trapped in this pocket by gravity. The condition was corrected and the normal 15-min freeze time was attained by reshaping the upper edge of the valve to eliminate the gas pocket. The emergency drain line that terminated just below the frozen zone of the valve conducted heat from the upstream molten salt into the freeze- plug area, created an unbalanced temperature distribution, and caused the freeze plug to be displaced. It was necessary therefore to direct the alr to the upstream side of the valve center line and to modify the air nozzles to offset the extra heat flow. 15 Conduction of heat by the emergency drain line into the frozen zone at the valve center also created some difficulty in adjusting the valve shoulder temperature controls in the "hold-freeze" condition. The ther- mocouples at this position are surface mounted and exposed to the cooling- air flow and therefore do not indicate the true salt temperature. In the transient from the thawed to the frozen condition the indicated temper- ature may be as much as 300°F below the freezing point of the salt before the valve is frozen. The thermocouple signals are transmitted to the electronic switches which provide "on-off" control action about a predetermined set point. If the set point is 750°F, the "fast-freeze" air will reduce the pipe surface temperature to this point rapidly and the control circuit will falsely indicate that the valve is frozen. The circuit will oscillate from the '"fast-freeze'" to the "hold-freeze" con- dition several times in the 10 to 15 min until the salt is frozen. If the set point is lowered from TSOOF, the freeze plug is excessively large. Insulation and air shields were applied to the thermocouples to obtain better indication of the salt temperature, and further tests are under way. Reactor Control Rods and Drives Control Rod Drive Prototype The Vard Corporation of Pasadena, California, was given a contract to design and build a prototype rod drive mechanism and to fabricate additional drives after approval of the prototype. Completion of the prototype is scheduled for May 1963. A diagram of the drive is shown in Fig. 2.4. The diagram does not show the limit switches and shock absorbers. A continuously running fan will be provided to cool the motor, but its use is to be avoided if pos- sible. The motor, tachometer, and gear reducer No. 1 may be incorporated into a single integrated assembly. The drive is expected to have the following performance characteristics: Rod speed, maximum, in./sec 0.50 Rod stroke, in. 60 Position indication (referred to a point on pitch circle of the drive sprocket), in. Coarse 10.20 Fine +0.030 Acceleration during scram, minimum, 16 ft/sec2 Clutch release time, maximum, sec 0.050 16 UNCLASSIFIED ORNL-LR—DWG 78806 SYNCHRO NO. 2 60° PER INCH OF ROD MOTION -» ~=-[ACHOMETER SERVO POTENTIOMETER MOTOR SYNCHRO NO.1 5°PER INCH ; -w— GEAR REDUGCER NO. 2 'BRAKE OF ROD MOTION-»~ 4 11 ~INPUT SPROCKET _ T ELECTRO- o MAGNETIC (000) GCLUTCH fi ‘\EQUAL RPM GEAR : REDUCER NO. 4 = i , l SPROCKET GHAIN ' [ AlIR IN\ OVERRUNNING CLUTGCH", FLEXIBLE TUBULAR ROD SUPPORT v=0.5in./sec - REACTOR VESSEL CELL -l «THIMBLE | REACTOR CORE POISON ELEMENTS 1t He~TEMPERATURE v 4400° F Fig. 2.4. Diagram of Control Rod Drive. L The potentiometer provides the independent rod position information required by the safety specification that the rods be withdrawn a mini- mum amount during reactor filling. The overrunning clutch provides an alternative drive path in the rod insert direction only and can be used to assist gravity if the rod or its associated chain and sprockets stick when called on to scram. Preliminary tests by the Vard Corporation indicate that dynamic braking of the motor may be a satisfactory substi- tute for the electromechanical brake shown on Fig. 2.4. This would per- mit the elimination of one in-cell component, along with its associated wiring and space requirements. Control Rod The control rod, which must be flexible to travel through the off- sets necessitated by the crowded conditions at the top of the reactor, is being developed at ORNL. Two types of flexible metal hose have been tested: a helically wound hose with two l/8-in. braided restraining cables, and a convoluted mctal hose with a single wire-mesh sheath (see Fig. 2.5). Each hose was 14 ft long from the drive chain connector to the bottom of the poison elements, and it was tested in a z2-in. test thimble which was maintained at 1000 to 1300°F during the tests. Two reactor-quality stainless steel poison elements (Gd,0z-Al.0s) were in- cluded in the assembly, which was operated through 5000 full cycles of 60-in. stroke and 1500 scrams with accelerations near 16 ft/secz. The poison elements were removed periodically for examination. The results of the examination are discussed in Chapter L4. Both types of flexible hose ran freely in the two offsets and thim- ble. The maximum acceleration of the rods in the scram operation was approximately 16 ft/secz. There was considerable backlash because of the large clearance between the thimble and the control rod. Position- ing within *0.030 in. was difficult because of the combination of the inherent backlash and stretching of the metal hose at temperature. The lower 16-in.-radius bend of the thimble, with the mesh-covered convoluted metal hose in position, is shown in Fig. 2.6. Rollers were installed in the thimble at this point and on the inside of the upper bend to reduce the friction and abrasion of the thimble walls, which had been as much as 0.022 in. in these agreas. Heater Tests Pipe Heaters with Reflective Insulation Two all-metal prototype heaters, with a joint closure piece, were obtained from the Mirror Insulation Company, and are being subjected to endurance tests. The units, shown in Fig. 2.7, are 30 in. long and are made up of a 1lb-gage austenitic stainless steel outer shell which con- tains six spaced laminations ol 0.006-in.-thick stainless steel sheets. UNCLASSIFIED PHOTO 39338 CCNVOLUTED HOSE HE_ICALLY WOUND HOSE WIRE MESH SHEATH DISCONNECT e ......... e B / e . | crpe e L U ‘ F s FEACTOR-GRADE POISON ELEMENTS DUMMY ELEMENTS Fig. 2.5. Flexible Control Rod. 81 19 UNCL ASSIFIED PHOTO 39337 Fig. 2.6. Lower Bend in Control-Rod Thimble. =5 LIFTING BAILS FOR REMCTE MAINTENANCE ¥ : Pig. 2.7 ELECTRICAL CONNECTIONS T4ROUGH INSULATORS SILVER FLATED SURFACE Heater with Reflective Insulatioa. UNCLASSIFIED PHOTO 38924 HEATER CLIP . 0c 21 The total thickness of the reflective insulation is 4 in.; the inner 16-gage face sheet is flash plated with silver. ments are clipped to the inner surface for easy removal. The formed heating ele- The heat characteristics of the units were studied in tests conducted with the units mounted as they would normally be positioned on the 5-in. pipe. The test data are presented in Fig. 2.8. One unit, which has not been disturbed while hot or exposed to high thermal stresses, has operated for 2640 hr above 1200°F. blistering of the silver plating on the unheated base section. The silver oxidized uniformly and presented a white crystalline surface throughout. It is in excellent condition, with only minor The heating elements, which are totally imbedded in ceramic material (Cooley Electric Mfg. Co.), are also in excellent condition. The other unit operated 1260 hr before failure. This heater was fre- quently moved and opened for inspection while hot, and it was also used as an oven to burn out insulation that contained organic material. ure occurred when the lead wires shorted to ground at the lead-wire exit ports on the inside of the unit, as shown in Fig. 2.9. were shielded with 5/32-in.-OD porcelain heads. Fail- The lead wires The hcat generated during the short to ground was sufficient to burn off one lead wire from each of the two curved heating elements and to melt the 3/8-in.-tubing lead-wire ports at these points. Fig. sulation. TEMPERATLRE RISE OF THE HEATER ELEMENT ABOVE AMBIENT (°F) 1400 1200 1000 800 600 400 200 2.8 The exposed silver in this unit blis- tered both on the base section and the removable unit; however, the UNCLASSIFIFD ORNL —LR—DWG 78807 o= - / _~0 THERMOCOUPLES MOUNTED ON INNER FACES OF HEATERS L P P 7 O o0 P~=THERMOCOUPLES NOT ATTACHED IN FREE LAMINATIO N SPACE BETWEEN BACK OF HEATER AND FIRST 7 — / ’0/’ / r Y y O 7 * o / s / o) /- /—» /i o/”/' o P 100 200 300 400 500 POWER REQUIRED (w/linear foot) 600 350 300 250 200 150 4 100 AMBIENT (°F) TEMPERATURE RISE OF THE OUTER SURFACE ABOV Heat Characterislics of Pipe Heaters with Reflective In- - INSULATOR UNCLASSIFIED | PHOTO 39335 : PLATING PEELED OFF & LOCATION OF ELECTRICAL SHORT Fig., 2.9. Inner Surfsces of Pipe Heaters with Rzflective After 1260 hr of Operation at Temperatures Above 1300°F, BLISTERS IN PLATING Insulation ce 23 silver under the heating elements was in good condition. There was severe oxidation of the heater retaining clips and lower inside edges of the removable section below the plating. Along the latter edge, there were several points of complete penetration of the first lamination. The failure occurred within a U8-hr period after the unit had been used as an oven to burn out organic material from another type of insu- lating material. The period of burnout was accompanied with off-gassing and visible smoke, some of which plated out on the end closures. An analysis of the material on the end closures indicated high sulphur con- tent. The degeneration of the silver finish and the porcelain lead wire beading might have been caused by the sulphur. Breakdown of the insu- lators where the lead wires leave the unit could have caused a ground. All materials for use in the reactor are consequently being checked to eliminate those with volatile binding constiluents. The unit was opened frequently at one end while being operated at temperatures above 1000°F. The low heat capacity of the reflective sheets could have caused high thermal stresses when the heated unit was opened to the atmosphere, with resultant blistering of the silver plating. ©Such opening of the unit would not occur during reactor system operation. These units are, in general, structurally strong, nondusting, and quite efticient. Therefore they were recommended for use in the MSRE based on new specifications to prevent the severe oxidation problems mentioned above. Drain Tank Coolers The prototype cooling bayonets for removing afterheat from the drain tanks were thermally shock tested a total of 2257 times without failure. Ten prototype wall-mounted thermocouples were installed in the test facil- ity, and four of these failed aftcr 1689 cycles. Three of the thermo- couples which failed were the l/8-in.-OD, sheathed, duplex type and were located at thc high stress area at the bottom of the 1l-in. bayonet. The number of cycles completed before failure was a blg improvement over the performance in previous tests. There is a l/l6-in.-OD thermocouple of the same type in the same location which has not failed. The tests are continuing. Sampler-Enricher System Mockup Fabrication of the sampler-enricher system mockup and its installa- tion into the Engineering Test Loop are nearing completion. Some minor modifications of the initial design were made to facilitate fabrication. 24 The cable drive mechanism is shown in Fig. 2.10 prior to installa- tion. The motor has a 1 l/2-rpm output speed and 28-in.-1b output torque. There are four position switches on the motor. The cable, its spring- loaded storage drum, and drive gear were purchased as a unit from Teleflex, Inc. Two synchro transmitters geared directly to the drive-gear shaft supply signals to the position indicator. One transmitter is geared so that one revolution of the shaft represents 1 ft of cable passing the drive gear; for the other, one revolution of the shaft represents 48 ft of cable movement. The inner compartment (area 1C) is pictured in Figs. 2.11 and 2.12 prior to installation of the wiring. The cable drive mechanism is located in the upper box. The electrical conneclions for thc motor, posi- tion indicator, and position switches will be bruought oul through and hermetically sealed in the three tubes localed vn the top of the box. Front and rear views of the box are shown in Figs. 2.11 and 2.12. UNCLASSIFIED PHOTO 39085 SYNCRO TRANSMITTERS DRIVE MOTOR Fig. 2.10. Sampler-Enricher System Drive Unit Assembly. 23 UNCLASSIFIED PHOTO 39086 DRIVE UNIT BOX ACCESS PORT OPERATORS T'ig. 2.11. OSampler-Enricher System Inner Compartment.,, Front View. 26 UNCLASSIFIED PHOTO 39084 ELECTRICAL PENETRATIONS _~ PRESSURE El FMFNT - T DRIVE U NIT BOX CHECK VALVES Fig. 2.12. Sampler-Enricher System Inner Compartment, Recar View. 27 Flange Disconnects Two sets of integral dual-sealed ring-joint-flange disconnects, that are planned for use in the line which connects the pump bowl and the sampler were tested. The flanges were machined individually using gages and procedures to make them interchangeable. The first set did not fit properly and failed to seal. A helium leak rate of less than 10-6 cm3 (STP)/sec was obtained with a second set of flanges that had been fully annealed and remachined. Further testing is in progress. Core Development Internal Flow Distribution Operation of the full-scale MSRE core model with water at 85°F was continued to make measurements of the internal flow distribution and the pressure drop through the core. The distribution of flow through the moderator assembly is shown in Fig. 2.13. The flow was measured in 77 more or less randomly chosen fuel channels. There are two regimes of flow distribution which represent fuel channels that are mutually perpen- dicular to each other. The two regimes are characterized by different inlet conditions. A plan view of a section of the l-in.-thick bars of the core-support lattice with the fuel channel orientation superimposed is also shown in Fig. 2.13 to illustrate the different inlet conditions of the two regimes. The flow, which must pass through the small square orifice formed by the plane of contact of the two layers of support lattice, diverges into the channels between the bars of the upper layer. From there, part of the flow continues up into the fuel channels directly above, while the rest of the flow must undergo a change in direction before entering the "perpendicular" fuel channels. The effect of this change of direction is to rcduce the flouw Lo Lhuse channels. This reduction in flow will be eliminated by drilling holes in the lattice bars directly under the starved channels. It is also shown in Fig. 2.13 that the flow rate decreased slightly with radial distance from the center of the vessel. This is because of the radial pressure gradients in the vessel heads, particularly the lower head. Since the flow rate is lowest where the power density is lowest, this radial variation has the beneficial effect of making the temperature rise of Lhe fuel more uniform. 'the scatter of the data is primarily due to buildup of tolerances in the assembly of the core-support lattice. The fuel passes through constricted orifices in the lattice before it goes to the moderator. The equivalent diameter of these orifices varies 30ak Ridge National Laboratory, "MSRP Prog. Rep. March 1 to August 31, 1961, " USAEC Report ORNL-3215, pp. 64-65. 28 UNCLASSIFIED ORNL-LR-DWG 78808 r [ 4sl‘NDAgD DEVIATION e e O — 4.2 LEAST SQUARES UNE_OO (o) o i T o N 320 \ \STA.I\EA? ——a0_q % 8 ® o ° | o [ O__ 1.0 Py BOTH Lines —81 020 2 . —— ——— _.-_o N R asT SQUARES LINE o © Qe O [ STANDARD DEVIATION 8- = 0.8 B FUEL CHANNELS DESIGNATED BY O T FUEL CHANNELS FUEL CHANNEL FLOW RATE {gpm) DESIGNATED BY @ 3 VIEW FROM JUST ABOVE TOP OF CORE-SUPPORT LATTICE BARS WITH FUEL CHANNELS SUPERIMPOSED - | | I 4 6 8 10 12 14 16 18 20 22 24 RADIUS O FUEL CHANNEL IN CORE (in.) Fig. 2.13. TFlow Distribution of Fuel Among Core Fuel Channels for a Total Core Flow Rate of 1200 gpm. with the summation of several ltolerances. Since the prcssure drop through these orifices is high compared with that through the moderator, the ori- tices are important in conlrullluy the flow distribution, cven though cross flow channels are provided 1lu Lhe cure blocks. Thio ocatter will be much less in the reactor because closer tolerances will be used than . were used in fabricating the model and also because many of these clear- ances decrease as the reactor heats up to operating temperature. Core Pressure Drop The over-all pressure drop through the core model from the inlet pipe to the 10-in. outlet pipe was measured with water. The data arc pre- sented in Fig. 2.14. The line has a slope of 2.0, indicating that it is almost independent of the Reynolds number; therefore, the samc preccure drop in terms of feet of fluid will be experienced with molten salt. Engineering Test Loop (ETL) The engineering test loop% was placed into operation in August 1962 after off-gassing the graphite in the graphite container and treating the 40ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Feb. 28, 1962, " USAEC Report ORNL-3282, pp. 34-LO. 29 UNCLASSIFIED ORNL—-LR— DWG 78809 20 T T T T T 1 T l O MEASURED WITH HIGH —RANGE OIFFERENTIAL PRESSURE CELL \ ® MEASURED WITH LOW—RANGE DIFFERENTIAL PRESSURE CELL v e SLOPE=2.0 ™ O ’ C o O / OVER-ALL PRESSURE DROP IN MSRE CORE (ft of fluid) '0.2 \ (ON 2000 1000 500 200 100 FLOW RATE (gpm) Fig. 2.14. Pressure Drop Through MSRE Core. operating salt in the drain tank with HF and H, (o remove oxygen. After 1540 hr of continuous operation the loop was shut down (in October 1962), and samples of graphite were removed via a dry box for metallurgical examination. The subsequent treatment of the salt in the drain tank with HF and H, produced information about the quantity and original phase of the oxide removed. Additional information was obtained on the operational characteristics of the graphite container access joint and on the forma- tion of solids at the salt liquid level. Details of these tests are pre- sented in the following sections. Results of the Graphite Pretreatment As described previously,® the graphite was given a vacuum and dry helium purge treatment at 1300°F prior to the introduction of the salt. 50ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962, " USAEC Report ORNL-3369, pp. 53-59. 30 In order to determine the effect of this treatment, as-received sample ETL-1 and a sample (ETL-2) removed from the loop after this treatment were degassed at elevated temperatures, and the evolved gas was analyzed. The results of the gas analyses are presented in Table 2.2.6 Table 2.2. Comparison of Gases Evolved from As-Received and Treated ETL Graphite Temper- Volume of Gas Composition (vol %) 0. Removed 2 Sample ature Gas (ppm by weight) (°c) Evolved® H, HC -H,0 Ny CO CO, ‘PPHPY Vel ETL-1 600 8.1 14 21 54 4 05 -2 ‘ 19 (agm- - - cl000 0 O34, .72 8 8 2 11 10 received) T T T s e e e ETL-2 600P bl 5 4 5 1 3 2 2 (treated) 1000 7.5 bz 1 1 2 50 Lk 17 &Volume is given in cm3 (STP) of gas per 100 cm3 graphite. Gas evolved from ETL-2 at this temperature contained 80% helium. Loop Operation Pretreated salt® was circulated through the pretreated graphite at 1200°F with a velocity of 1 1/2 to 1 3/4 ft/sec in the flow channels. The loop was filled initially to the minimum operating level of the DANA pump to provide a minimum salt-to-graphite ratio and a maximum resolution of the effect of ‘the graphitc on the salt samples. The volume of salt cir- culated was estimated to be 1 3/4 I't2; the gruphite volume was 1 1/8 ft3; and thus the salt-to-graphite ratio was 1 1/2. Thc pressure on the graphite during operation was 11 to 12 psig at the inlet and 8 to 9 psig at the outlet. In October, after 1540 hr of uperation at the above con- ditions, the loop was shut down for removal of samples. Operation of the Graphite Container Access Joint The graphite container access joint wae operated differently than reported® for the previous period of operation in which the liquid salt 8Letter from J. P. Blakely of Reactor Chemistry to J. L. Crowley, "Gas Evolution from ETL Graphite,” August 29, 1962. 31 level was maintained against the frozen salt zone. During filling of the loop with salt prior to operation during the current period, too much of the trapped gas in the joint was bled off, and the liquid level rose higher than was intended. After the frozen salt cake was established, as indicated in Fig. 2.15, the heat removed by the cooling air required to maintain the frozen zone was excessive and the loop temperatures could not be maintained at 1200°F with the installed heater capacity. The heat transfer in the liquid salt portion of this joint was much greater than could be accounted for by conductance alone and was possibly due to some small flow in the annular region below the frozen zone. By increasing the gas pressure in the region above the frozen zone, it was possible to lower the liquid level to the location shown in Fig. 2.15 and thus leave the ring of frozen salt at the upper location, thereby arranging to take the large temperature drop in the gas space. Apparently the salt cake UNCLASSIFIED ORNL-LR-DWG 78810 < = o '—. 5 7N S - Q . o] N\ = w | 78 w g NN 5 B AR /Q (@] (@] 21 O = 4N g g ] I —rrozen & F N SALT ':E N CAKE 16 A -————— —8F ° ! ;\. _______ 1C 4 .. N 1 AN \ AN —— — ———- 8E * ZIN ' ZN 12 N ZI\ ‘\\\\ MNy _ [ —] / \. ”B 10 . N LIQUID LEVEL DURING OPERATION ~ 8 —1 i ool s S o \I ) VERN : [ ] ZgmN ;:Q \\ AR 4 . INSULATIONSIP-N \ ERSLIZEN L /-\ FN 2 Y777877727,77 78N [~ ___ SALT FLOW o — ¢ — 100 300 500 700 900 1100 - TEMPERATURE (°F) N PR 5 - k) O S 0 0% .a%$5$%%f¢? o &5 PRI RIS BELIXE Fig., 2.15. Temperature Profile of ETL Graphite Container Access Joint During Operation at 1200°F. Desired liquid level maintained hy trapped gas rather than by frozen-salt seal. 32 was porous to gas pressured from the "dry" side, while an increase in pressure from the liquid side tended to seal the pores with liquid. The gas trapped above the liquid level maintained the level in the position shown in Fig. 2.15 during 1500 hr of operation without further attention. Although cooling air flow was maintained to assure sealing of the joint during operation, the salt-to-gas interface temperature remained above the temperature of the molten salt. The presence of the liquid-to-gas interface during operation was also indicated by a constant differential between the Jjoint pressure and the changing pump bowl gas pressure. The actual position of the salt-to-gas interface was determined by examina- tion of the Jjoint following shutdown. Indications are that a similar type of joint could be operated in the MSRE. The frozen-zone backup plug would protect against level [luc- tuations resulting from pressure transients in the reactor. 1In partic- uwlar, it would protect against the effect reported previously® of the salt rising in only one sector of the annulus as the liquid began to freeze in the other sectors. The method of establishing the joint in the reactor would be to have a low fill pressure in the reactor before the salt started in and then to raise the salt to the level desired for the frozen-salt backup ring. After freezing the ring, the level would be allowed to drop to below the bottom of the joint, and the pump bowl pressure would be increased. The salt level would then be raised to trap the required amount of gas and thus obtain the proper liquid level for the designed operating conditions. Some small manipulation of this final level would be possible by regu- lating the gas pressure in the salt pump bowl. Graphite Sampling Via Removable Dry Box Twelve, small, cylindrical graphitc samplee, two graphite prototype core elements (2 x 2 x 50 in.), and three metal specimens were removed for metallurgical examination. A description of the problems encountered in the sampling operation is given below. The dry-box facility,® which enabled samples to be removed and re- placed in the loop without admitting atmospheric contamination, was pre- pared and attached to the graphite access joint after shutdown of the loop. The small samples were easily removed and isolated in carriers, but the graphite prototype core elements® appeared to be stuck and could not be removed with the equipment avallable in the dry box. A 200-1b spring scale had been included for measurement of the lifting force required, but it was inadequate. Removal of the two elements was finally accomplished by rapidly heating the graphite container to 500°F and there- by expanding the container and loosening the elements. The lifting force used was probably between the 200-1b limit of the scale and the 500-1b rating of the hoist. When extracted, there was evidence of small globules of salt remain- ing between contacting surfaces of the graphite elements; however, from 33 the appearance of these globules it is not believed that these were the major cause of the removal difficulty. The original installation of these elements was difficult because irregularities in the rolled and seam- welded container made the fit very tight. The new elements were machined to a slightly smaller size and were placed in position without difficulty. The two elements removed had been inspected prior to testing and designated class I and class IV, respectively. Class I denotes the best and class IV the worst with respect to surface faults of all the elements of the graphite obtained for the ETL experiments. When weighed upon re- moval from the dry box, both elemente had net weight gains of 7 g. Most of this gain, however, was from salt that had entered the threaded open- ing made for the INOR-8 lifting attachment. When this salt was removed and weighed separately, the net increases in weight were 0.86 g and 1.04 g, 0.0185% and 0.022%, respectively. The MSRE specification allows a 0.5% net increase after operation with 150-psig pressure. Analysis of Solids Removed from the Graphite Container Access Joint When the core plug of the graphite container access joint was removed :+ .4 for graphite sampling, a ridge approximately 1/32 in. thick of metallic- g appearing material was found at the position of the former liquid level. . «. s - Most of the joint that was below the liquid level was covered with a very o thin dark coating down to the salt outlet of the container. A sample of v st the metallic-appearing material was chipped off and was found to be mag- . netic.” A chemical analysis of the sample indicated 7.55 wt % Fe, 2.34 wt .- % Cr, and 52.5 wt % Zr. Salt samples from the loop contained only * 0.13 wt % Zr. Indications were that the zirconium was in the form of ox- ide, since an HF acid solution was required to dissolve the sample rather N than the usual nitric and sulphuric acid solutions used for the fluoride.? : The collection of corrosion products and zirconium oxide was apparently caused by the cold-trapping effect of the temperature gradient in this area and the very low salt velocity just below this area. Additional tests will be conducted to determine the characteristics of this cold- trapping effect. Chemical Analysis and Hb Treatment of ETL Flush Salt The flush salt used during operation of the ETL with graphite in- stelled was treated with Hy, and HF for two periods of 48 and 52 hr during November and December to remove oxygen. In each case, the treatment was contimied until oxygen wac no longer carried out in the effluent. Oxy- gen removed by both of these treatments amounted to 18.75 g, or an equiv- alent of 78 ppm in the total salt inventory. Less than half of the in- ventory was circulated in the loop. Chemical analyses of 36 salt samples taken during 1500 hr of operation with the graphite installed averaged 300 ppm in the loop and 200 ppm in the drain tank, for an average of 236 ppm in the total inventory. Since the Ho-HF treatment appeared to 7W. F. Vaughan, private communication, November 1962. 34 - remove all the available oxide in the salt, it is concluded that the chem- ical analyses contained some error. This error could result: from contam- ination of the sample or a fault in the procedure. The product removal rates of three treatment periods during August, November, and December have a similarity in form which suggests a method of determining not only the total oxide content but also the quantities present in different forms, such as soluble ZrO, and precipitated ZrO.. The results of the November treatment (following graphite operation), shown in Fig. 2.16, indicate two phases of oxide removal. The curves of _ UNCLASSIFIED ORNL- LR -DWG 78811 PHASE 1II i \ N—— 08 \\ — N\ \ 0.6 \ N 0.4 WATER COLLECTED AT 32°F (g/fime unit) \‘/PHASE I \x\ .0.2 - L 0] 2 4 6 ‘ 8 10 12 TIME (4-hr collection units) Fig. 2.16. Removal of Hy0 During Third HF Treatment in Right Drain Tank of ETL During November 1962. o 35 Fig. 2.16 are based on data collected from the first cold trap at 32°F, since rates could not be determined from the final trap at -109°F. If, however, the total collection rate could be determined, integration of the individual slope equations would yield amounts of individual phases. Additional data will be obtained by adding oxide to the tank and measur- ing the rate at which it is removed by the HF treatment. ETL Graphite Permeability Samples of the ETL graphite were tested for permeability at room temperature.® The permeability tests were performed on prototype sec- tions by measuring gas flow from the flow channel surface to a l/h-in. hole drilled at the axial centerline of the element. For this configu- ration, which has an area-to-length ratio ranging between 69 and 99 cm, helium permeability of from 1 x 10-3 to 3 x 10-3 cm2/sec was measured on three samples. Xenon Transport in the MSRE A series of experiments is being planned to obtain a better under- standing of the mechanism controlling the distribution of xenon in the MSRE. Preliminary studies in support of these experiments have shown that the gaseous fission products tend to be transported into the graphite or removed at the pump bowl. The studies have also shown that, within the range of knowledge of the critical parameters, this division is strongly dependent on the core flow characteristic, as well as the diffu- sivity of the graphite and the removal rate at the pump bowl. Presently, experiments are being planned to determine the xenon diffusion coeffi- cient for reactor-grade graphite, to determine the removal rate in the pump bowl, and to determine the over-all xenon mass transfer coefficient. MSRE Maintenance Development A procedure was prepared for the remote replacement of a freeze-flange gasket. This task requires most of the anticipated in-cell manipulation of the flanges. Eighty-eight steps are involved Lhat include assembling the clamp onto the flange (sce Fig. 2.17), stowing the clamp on in-cell brack- ets, spreading the flanges apart with the pipe line jack system, overcoming misaligmment, and handling the flange covers and the ring gasket. This procedure was demonstrated in the maintenance mockup while observing as many of the reactor restrictions and limitations as possible. The tools, handled with a jib crane and a special A frame, were operated through a wooden mockup of the portable maintenance shield and were observed re- motely with a sheathed periscope. The complete task required two men for — 8Letter from R. B. Evans, "Helium Leak Rates — ETL-MSRE Graphite, " August 14, 1962. 36 UNCLASSIFIED PHOTO 39120 i E : i SHEATHED SCOPE ==3= HYDRAULIC CLAMP OPERATOR Fig., 2.17. Remole Assembly of Freeze-Flange Clamps. 37 one 8-hr shift in the mockup, and it is estimated that four 8-hr shifts would be required in the reactor. The workability of the maintenance sys- tem was established. These tools and techniques are applicable to all freeze flanges in the system, with the exception of FF-100 (between the reactor and the pump), where interferences with a pump support beam and line 101 hinder the place- ment and operation of the presently designed tools. For this case, a special offset clamp operator tool, which is the tool most affected by the interferences, was designed and is being fabricated. Also, the mockup is being completed in this area so that the new tool and all others may be tested. Maintenance Study Model A 1/6-scale model of the reactor cell was constructed (Figs. 2.18 and 2.19) to assist in evaluating maintenance problem areas both during design and later during maintenance periods. The model is complete with respect to the components that will be removed during maintenance. It is dimen- sionally accurate and can be disassembled in the same fashion as the reactor system. Color coding has been used to distinguish the fuel salt, coolant salt, air, oil, helium, and water lines. Pump Development Prototype Pump Operation and Testing High-Temperature Operation With Circulating Molten Salt. The rotary element was reassembled with more running clearance for the shaft, and testing of the prototype pump was resumed. It has operated 2150 hr circu- lating the salt LiF-BeF,-ZrF,-ThF,-UF, (70-23-5-1-1 mole %) at 1200°F, 1200 rpm, and 1200 gpm. Operation was interrupted by a short that de- veloped in the stator winding of the alternator of the drive-motor power supply, but testing was resumed with the pump drive motor directly con- nected to 60-cycle electric power. The pump has been running 300 hr at the above-mentioned conditions. Instruments are being calibrated for measuring the back diffusion of gas up the pump shatt by using Kr®5. Difficulties encountered in measur- ing gas flows of the order of 0.1 to 1.0 cm3/min and with leakage of gas from the counting chambers were satisfactorily resolved. Calibration and testing of the Kr®> counting equipment are under way. Test Operation With Water. During further water tests of the fuel pump, measurements were taken of the impeller loading over a range of head and flow at three speeds. The results of the measurements are pre- sented in Fig. 2.20 and compared with data supplied by the designer of FUEL PUMP Y ccror Fig. 2.:18. Maintenance Model — Elevation View. IUNCLASSIFIED PHOTO 39419 8¢ & UNCLASSIFIED PHOTO 37417 REACTOR THERMAL SHIE 6€ 40 UNCLASSIFIED ORNL- LR -DWG 78812 300 L T \\1300 rpm T~ 150 \\ — — ALLIS- CHALMERS DATA \~\ ~ ORNL EXPERIMENTAL DATA 200 S N \\\\ \\\\ 860 >~ Ny ~ 5 N = 100 B e— — Y 1300 rpm e, . w | | N 8 \\ v\»\' ~, \\\ N £ B e S N 3 = A i ™ N = e 1150 " W @ 860 \\\ \ \\\ ) e '\ \\\ N A~Aa_| -\.%Q{ N N L . PN -100 S ~s -200 : 200 400 600 800 1000 1200 1400 1600 1800 2000 FLOW (gpm) Fig. 2.20. Comparison of Design and Experimental Data on Fuel Pump Impeller Loading. Experimental data obtained in tests with water and a 11.5-in.-diam impeller. the impeller and volute hydraulic configurations. The ORNL experimental results indicate—that the impeller loads at low pump capacity are as much as a faetor of 3 1nwer than those predicted, and thus bearing life should be greater than previously anticipated. PKP Fuel Pump High-Temperature Endurance Test The PKP test pump continued to operate; it has accumulated a total of 9500 hr of operation at 12250F, 1950 rpm, and 510 gpm circulating the salt LiF-BeF,-ThF,-UF, (65-30-4-1 molc %). Leakage across the upper seal has continued to be approximately 10 cm3/day, and the leakage across the lower seal has recently averaged 2.5 cm3/day, which is considered to be small. Lubrication-Pump Endurance Test The test lubrication pump has now operated 5900 hr at l6OOF, 70 gpm, and 3500 rpm circulating a turbine-type oil. Instrument Development oingle-Point Liquid-Level Indicator The prototype two-level conductivity-type molten-salt-level probe de- scribed previously has been operating successfully for six months. Several 41 attempts to increase the signal-to-noise ratio were made, with limited success. This ratio is already high enough, better than 3 to 1, for depend- able operation, but a method of reducing the noise level would be desirable. To date, the instrument has operated successfully with an excitation current as high as 20 amp at 1 kc, and as low as 0.1 amp at 1 kc. The very low current excitation was obtained from the oscillator of a Foxboro Dynalog, with the Dynalog being used both as an excitation source and an indicator- recorder. This was considered to be one of the most successful test runs made, because it indicated positively that no special instrument systems were needed to use this single-point liquid-level indicator. During this reporting period, the probe connectors to be used in the MSRE system were also tested. These connectors, which are 12-pin units of the type used for thermocouples,® operated satisfactorily and gave no indi- cation of failure or introduction of noise into the system. Some inter- mittent difficulties that started when these connectors were first installed were found to be caused by poor contacts at compression connectors else- where in the circuits. Replacement of the compression connectors with brazed connections eliminated the trouble. Fabrication drawings of the units to be installed in the MSRE drain tanks were completed. The final level-indicator design and a diagram of the electrical equivalent circuit of the device are shown in Fig. 2.21. Additional tubes that give structural rigidity to the assembly are not shown. Pump-Bowl Liquid-Level Indicator Developmental testing of continuous liquid-level-indicating elements for use in measurements of molten-salt levels in the MSKE fuel and coolant pump bowls was continued. The two systems undergoing developmental test- ing completed one year of operation. During the year they were operated at temperatures between 900 and l3OOOF, with no objectionable operational characteristics being observed. For approximately 11 months of this period the system operated at 12500F. In addition to the planned temperature cycles previously reported, power was off twice for modifying electric wiring in the building. There were no changes in the characteristics of the two systems as a result of the temperature cycling associated with these power failures. It is still not known whether there has been salt-vapor deposition on metal surfaces above the liquid-salt level. One adjustable spark plug probe is locked in place by solid material, but there has been no indication that the level indicator system, with the core tube above the liquid salt, has any salt deposit on the core or in the tube. It was assumed when this system was designed that only a small amount of solids deposited on the core tube, or in the core tube container, would lock the core tube in place or restrict its movement. There is no indication that this has taken place. 90ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962, " USAEC Report ORNL-3369, pp. 82-83, esp. Fig. 3.25. 42 UNCLASSIFIED ORNL—LR-DWG 78813 EXCITATION v, — . ! SIGNAL, LOW- ‘ - SIGNAL, HIGH- . LEVEL LIQUID Vst A V2 LIQUID LEVEL V, EXCITATION o \ ['—0—. - I MWWV SIGNAL LEAD, COMMON NN NT- \ N \ SIGNAL LEAD, HIGH-LIQUID LEVEL w Py ™ N V\ N\ NI Nog .k\ V\WV\W i NOTE: S, AND S, ARE ! USED TO INDICATE THAT THE MOLTEN SALT GROUNDS THE CONTACT PLATES IN ORDER TO PRODUCE THE DESIRED SIGNAL ). >~ SIGNAL LEAD, LOW-LIQUID LEVEL —\WWWWAWWWWY ~ /(/ 1 2 ‘/7777777> .|||_\\ Fig. 2.21. Single-Point Liquid Level Indicator. The third system, mentioned previously, is now being constructed for installation on the prototype pump. In this system, the float is a-hollow INOR-8 ball, rather than graphite, and the iron core has been increased in diameter from 0.125 to 0.170 in. 43 Bubbler-Type Liquid-Level Indicator A bubbler type of liquid-level-indicating system was developed for use in measurements of molten salt levels in the MSRE pump bowls. This system is an alternate for the float-type system discussed above. A simplified diagram of the bubbler system is shown in Fig. 2.22. The basic operation of the system is the same as that of conventional dip-tube bubbler systems in that the level signal is obtained by measuring the dif- ferential between the pressure in the gas space above the molten salt and the pressure inside the dip tube. When the tube is purged with a small gas flow and the salt density is maintained constant, the differential pressure produced is proportional to the height of salt above the bottom of the dip tube. The MSRE installation presented special problems not found in a con- ventional installation because (1) the bubbler system piping communicates with the primary system and forms a possible escape path for radioactive gases and (2) the system could fail if pressure transients in the reactor pump bowl forced molten salt into the cold purge piping external to the pup bowl. 1In the present system design, contaimment of radioactive materials is effected by placing double check valves in each purge line and by enclosing the check valves and all lines and equipment downstream UNCLASSIFIED ORNL-LR-DWG 78844 SHIELD SECONDARY CONTAINMENT PUMP HELIUM - VESSEL 1 SUPPLY ] TO SECOND | PURGE R V PRESSURE SWITCHES SYSTEM ( | THROT TLING CHECK VALVES —— ‘ —L VALVES}, \ — PURGE_LINE ] Yy YL 2028 ] L 1 | L N~—— (\\- &psi (Efi> REFERENCE o SHUTOFF CAPILLARY VALVES RESTRICTORS / (D|FFERENTIAL SOLENOID VALVES PRESSURE TRANSMITTER PUMP BOWL TRANSMITTER SPECIAL EQUIPMENT ROOM ROOM REACTOR CONTAINMENT . CELL Fig., 2.22. MSRE Bubbler;Type Level-Indicating Systemn. 44y of the check valves in a secondary containment vessel. Plugging of the purge lines with salt during pressure transients is prevented by maintain- ing the volume of the purge lines downstream of the check valves as small as is permitted by pressure-drop considerations and by installing a surge pot having a volume ten times that of the purge lines in the heated zone adjacent to the dip tube. The present system is designed to withstand a 60-psi pressure increase from 6 psia without plugging the purge lines. The surge pot is a toroidal pipe inside the pump bowl. Capillary restrictors located upstream of the check valves serve a double purpose of providing a means of flow measurement and of preventing back diffusion of radiocactive gas into operating areas. The purge rate is controlled by throtlling valves. Deviation of the purgc rate from the design value is detected by pressure switches ingtalled upsbream and Le- tween the capillaries. Solenoid valves installed downstream of the capil- laries provide a means of checking the zero calibration of the differential pressure transmitter. The solenoid valves may also be used to block the lines on a signal from the safety control circuitry. A mockup of the bubbler-type system was installed on the prototype pump and was in operation approximately two months during this reporting period. The system was then modified temporarily so that the bubbler pene- trations could be used for back-diffusion studieg with krypton. During the period that it was in use, the bubbler system operated satisfactorily. One mishap occurred when an improperly installed fitting leaked and allowed the pressure in a bubbler line to become less than the pump-bowl pressure. Molten salt was forced into one of the dip tubes and associated gas lines, and it completely blocked them. This situation could not be corrected, so both differential-pressure transmitters were connected to the other dip tube and were operated this way until the systems were shut down. Since much higher quality construction will be used in the MSRE, the probability - of a similar failure occuring in the reactor installation is very small. Two types of dip tubes were tested but there were no observable dif- ferences in their operating characteristics. One was open ended, with a small V notch cut in the side; the other had a closed end with a l/8-in. hole through the side of the tube just above the bottom closure. Temperature Scanner Development testing of the temperature scanner system installed on the liquid level test facility was continued. A test was started August 6, 1962, and the scanner operated satisfactorily until December 3, 1962. This was approximately 3000 hr of continuous operation. On December 3, it was noted that the system was generating excessive noise pulses. The input thermocouples were checked and found to be the cause of some of the noise. The rest of the noise apparently was being generated by the switch. The noise became worse with time, and the system was shut down December 15. The switch was disassembled and examined. The mercury in the bottom deck was quite dirty because of oxidation, and the signal pins were wetted by the mercury. In contrast the top switch deck was comparatively clean, 45 and there was little evidence of mercury oxidation. It is believed that the difference in mercury oxidation of the two decks was caused by an in- adequate nitrogen purge through the lower deck. The lower deck bearing was also found to be bad. Mercury had leaked through the bearing and onto the top of the motor. The switch was thor- oughly cleaned; the lower bearing was replaced; and the switch was then reassembled. The nitrogen purge system was modified to ensure equal nitro- gen flow through each deck of the switch. The scanner was placed back in operation December 28, 1962. On January 7, 1963 the switch was again stopped because of excessive audible noise. Examination of the switch showed that the top bearing had failed. The switch was again cleaned and reassembled. Operation started again January 8, and the switch has operated satisfactorily with little noise since that time. It is planned to continue the test to determine switch operating life and to obtain data for determining more closely the oscil- loscope, amplifier, and discriminator stability and the precision and accuracy of the total system. High-Temperature NaK-Filled Differential-Pressure Transmitter A high-temperature differential-pressure transmitter to be used in measuring molten salt flow in the MSRE coolant system was ordered from Taylor Instrument Company, the only bidder on requests sent out in May 1962. After some clarification and a few changes in specifications, which reduced the cost of each transmitter to one-third the original proposal, the purchase order was placed in December. Delivery is expected in June 1963. The transmitter is a modified version (see Fig. 2.23) of Taylor's standard instrument Model 2z25T. The two major changes required were that the pressure-sensing head be made of INOR-8 and that the standard pneu- matic output system be replaced by an electrical strain gage. The range of the transmitter will be 0-50 to 0—750 in. H,0 at reactor operating ltemperatures. Two diaphragms are used in each pressure-trans- mitter system. One is in the pressure-sensing head to isolate the coolant salt from the NaK. NaK is used to transmit the pressure from the pressure- sensing heads to the differential-pressure transmitter. The second dia- phragm is between the NaX system and the interior of the differential- pressure transmitter, which is filled with silicone oil. Three transmitters were ordered and will be tested at ORNL prior to installation on the reactor. It is expected that ORNL participation in the development of this transmitter will be limited to reviewing the design prior to fabrication, INOR-8 welding and inspection, and testing and eval- uating the completed transmitters. 46 UNCLASSIFIED ORNL-LR—-DWG 78815 NoK FILLED PROCESS FLUID ARMORED CAPILLARY STAINLESS STEEL e PROCESS INOR-8 SEAL CONNECTION ELEMENT (HIGH) il N e T A R N A S g Y o b A Sicone R 2670 STRAIN GaGE ADJUSTOR TRANSDUCER DIFFERENTIAL~ PRESSURE-SENSING DIAPHRAGM FORCE BEAM 7 SEALING / SRR '.-‘-_-:..‘-,-'..‘.‘:':::. PR YASINSARO BELLOWS N RANGE . ADJUSTOR SEAL ELEMENT (LOW) VENDOR: TAYLOR {NSTRUMENT CO. MATERIAL: INOR-8 AND STAINLESS STEEL . - PROCESS RANGE: 0-50 TO 0-750 in. H,0 CONNECTION DESIGN TEMPERATURE: 1250°F ' QUTPUT: 0-25 mv Fig. 2.23. All-Welded, High-Temperature, NaK-Filled Differential- Pressure Transmitter. . .Thermocouple Development and Testing Radiator Thermocouple Tegst. It was determined from tests that Thermon - X63, a heat-conducting cement, is not suitable for use with radiator tube thermocouples because of its corrosive effect on INOR-8. A typical radia- tor tube thermocouple attachment was coated with this cement and heated to 1250°F for a period of five weeks. Metallographic examination revealed corrosion to a depth of 1 mil on the surface of the INOR-8 tube in contact with the Thermon X63 cement. _ Thermon will not be used on the MSRE radiator thermocouples and the reduced accuracy will be accepted. Engineering Test Loop Thermocouples. Eight MSRE prototype surface- mounted thermocouples installed on the ETL facility for testing under simu- lated reactor operating conditions accumulated an additional 1500 hr be- ~ tween 1040 and 1200°F for a total of 3000 hr. All thermocouples were still functioning satisfactorily as of October 24, 1962, when the loop was shut down. Test thermocouples and respective reference thermocouples continued to agree within the limits previously reported. 47 Drift Test. The maximum observed drift in the calibration of six Inconel-sheathed MgO-insulated Chromel vs Alumel thermocouples continued to be *2°F equivalent in emf output. These thermocouples have operated in 1200 to 1250°F air for over 8000 hr. Bayonet Thermocouples. Thermocouples being tested in the drain tank bayonet test facility continued to be subjected to thermal shocks in the temperature range between 1350 and 200°F at l-hr intervals. Five thermo- couples have failed to date, one at 6 cycles, one at 900 cycles, two at 1600 cycles, and one at 2049 cycles. Five other thermocouples are still functioning. Thermocouple Tests in MSRE Pump Test Loop. Tests were conducted with MSRE prototype surface-mounted thermocouples installed on the pump test loop to determine how accurately thermocouples located on the walls of pipes adjacent to heaters would measure the temperature of the molten salt inside the pipe. Temperatures indicated by wall-mounted thermocouples and a well thermocouple in the salt stream were compared at various operating conditions with heaters turned on and off. Data obtained from these tests indicated that the temperature readings of the surface-mounted thermo- couples were influenced by the heaters to such an extent that the thermo- couples could not be used for computation of reactor heat power or for pre- cise measurement of the mean reactor temperature unless the heater power was maintained constant and a correction was made for the bias in the ther- mocouple reading. Thermocouple End Seals. Several sealing and potting compounds were tested for use in sealing the ends of mineral-insulated thermocouples and copper-tube-sheathed thermocouple extension cables terminating in dis- connects located inside the reactor and drain tank cells and junction boxes located near penetrations outside the cells. Ceramic-base compounds were tested for use inside the cells, and epoxy compounds were tested for use outside the cells. Test seals were made with the following compounds: Araldite epoxy compound, Ames Technical G copper oxide cement, Sauereisen No. 30 ceramic-base cement, and Thermostix ceramic-base cement. Several of the ceramic-base compounds required curing temperatures too high for usc with copper. Preference was given to cements that cured and set at room temperatures. No acceptable seals were obtained with the ceramic base compounds; either the gas leak rate was too high or the electrical re- sistivilty was too low. Several test seals were made with Araldite in which no leaks were de- tected with 60-psig helium applied from the inside. One test seal was made by doping the Fiberglas insulation on the individual wires with Duco cement before potting the Araldite and then pressurizing with 60-psig helium for a period of two weeks. No leaks were detected. The insulation was removed from the wires in the other test seals made with Araldite. Glass-to-metal multi-terminal header seals were also tested for use in sealing the ends of the copper-tube-sheathed thermocouple extension cables terminating in disconnects located inside the cell. Test seals of this type were pressurized to 60-psig helium and no leaks were detected. 48 This method of sealing the ends of the thermocouple extension cables ap- pears promising. The glass-to-metal header seals under test were supplied by the Hermetic Seal Corporation. Further tests will be conducted with this type of end seal, including irradiation of typical assemblies to de- termine susceptibility to radiation damage. Formula 0900 glaze compound, manufactured by the Physical Science Corporation, was tested for use in sealing the ends of individual metal- sheathed thermocouples terminating in disconnects located inside the cells. This material is a water-mix glaze compound that is completely nonhygro- scopic when cured at 1550°F for a period of 15 min. The suitability of this material for the application has not yet been determined. The mate- rial adheres well to metals, but it is rather brittle when cured. 49 3. MSRE DESIGN Design Status Reactor System The design work on the Molten-Salt Reactor Experiment is nearly completed, except for the instrumentation and controls. During this reporting period, the design of the fixture for the assembly of the drain tanks and the flush tank was completed, and the designs of the electrical distribution system for the heater network, the conduit and cable tray layouts, and the modifications to the existing auxiliary diesel power system were more than 90% completed. There were no changes in design concept or significant changes in detail of any component or system. Based on test evaluations of two different types of insulation, specifications for the design, construction, and testing of all-metal reflective-type removable insulation with integral ceramic heaters were prepared for bid invitations. This insulation will be applied to salt- containing pipes. Work was continued on the compilation of the engineering calculations and system analyses on which the design was based. This compilation will form the basis for a design report. Instrumentation -and Controls System ILayout. The layout of the instrumentation and controls re- mains essentially the same as described in the previous reports. Three instrument panels, all located in the high-bay area, were added. Two of these panels will serve the fuel salt chemical processing system and the third is panel No. 2 of the containment air system. ILocations of the wire ways between panels and field-mounted equipment, signal-cable routing, and locations of terminal boxes in the vicinity of the reactor and drain tank cells were determined. Locations of thermocouple and instrument discon- nects within the reactor and drain tank cells are now considered to be Fliwm. Asslgnment of contalnment vessel penetration locations for instru- ment lines and cables is essentially complete. ¥low Diagrams. Instrument application flow diagrams for the instru- ment air distribution system, fuel salt chemical processing system, and the fuel loading and storage system were completed. The chemical processing system drawing and the fuel loading and storage system drawings replace and supercede a previously prepared drawing entitled "Fluorination System." Eleven diagrams previously approved are novw being revised in accordance with recent design changes. The tabulations of all instruments shown on the flow diagrams and the input signals to the data system were revised and corrected in accordance with reactor instrumentation and control system changes. ' 50 Control Circuitry. Criteria for control and safety circuitry are being reviewed, and preliminary block diagrams of the circuits are to be modified, as necessary, to incorporate recommendations resulting from the review. Detailed designs of circuits that are obviously consistent with recommended practices were started, and two schematic drawings of the annunciator circuits for the main and auxiliary control boards were com- pleted. The designs of the 48-v de power system for the safety control circuit and the remote control circuitry for switch gear and motor control centers were approved for construction. Control Panels and Cabinets. The designs of all instrument and con- trol system panels are approximately 85% complete, and fabrication in the ORNI. shops is 50% complete. Fabrication was completed of the eight modular instrument panel sec- tions required for the main control board. The three remaining main con- trol board panels, as well as the control console, were partially designed and fabricated and will be completed when nuclear instrument and control circuit regquirements are known. A complete design for one panel section includes front panel layout drawings, instrument mounting hole cutout drawings, pneumatic tubing diagrams, and electrical wiring diagrams. The design of the main control board, including the console, is 80% complete, and fabrication is 70% complete. Design drawings for three auxiliary instrument panels for temperature alarm switches were completed. Designs of the six nuclear instrument panels. to be located in the auxiliary control room are 30 and 60% com- plete, respectively. Fabrication of the process instrument panels is 35% complete. The thermocouple cabinet. design was revised to increase the number of patch. plugs: The designs of the control panel for the sump level indicator and the two instrument racks, one supporting the solenoid valves and the electric signal transmitter power supplies and the other supporting the electric transmitter signal amplifiers, were completed. These panels are to be located in the transmitter room. The designs of .panels and racks to be located in this area are now complete, and fabrication is 50% complete. : The design of containment air panel No. 1 was revised and contain- ment air panel No. 2 was designed. The designs of all auxiliary instru- ment panels to be located in the field, except the two fuel salt chemical processing system panels, are now complete, and fabrication is approxi- mately 60% complete. o Interconnections and Field Installations. Nineteen detail design drawings of the instrument air tubing interconnections and thermocouple extension cable routes through penetrations in the reactor and drain tank cell walls were completed. The thermocouple and other electrical signal cables originating in the reactor and drain tank cells terminate in the junction boxes just outside the cells where a transition from 51 copper-sheathed cables to standard wiring is made. Designs for the fabri- cation and mounting of 16 junction boxes were completed. Preparation of interconnection wiring and cable drawings for the an- nunciator system and for the Foxboro Electronic Consotrol Instrumentation System is nearing completion. Designs for interconnection wiring for the instrument power distribution system, miscellaneous indicator lamps, and the thermocouple and fabrication is being prepared. Installation design details for cell penetration safety block valves and for field-mounted instruments on the containment air system and the 0il systems for the fuel and coolant salt pumps were completed. Fabri- cation drawings for four salt-level-indicator probes on the drain tank were completed. Designs of four venturi flow elements for the oil system were completed and fabrication drawings were issued. Four drawings that were approved for construction completed the de- sign of the reactor cell nuclear instrument penetration tube. The design of the cable tray layout was completed. Process and Personnel Radiation Monitors. Requirements for process radiation monitors were reviewed and revised to incorporate changes re- quired by system revisions. At present there are 23 monitors located at 12 stations. Where the monitors perform an alarm function only, a single channel is provided. Monitors that are interlocked in the reactor safety system are duplicated or triplicated, with their contacts arranged in 1 out of 2 or 2 out of 3 coincidence. The selection of detectors and as- sociated electronic equipment was completed for all approved monitor applications. In general, ORNL Q-1916 count-rate meters with GM tube detectors are specified for low radiation level applications, and ion chambers and electrometers are specified for high radiation level appli- cations. Provisions are being made to permit remote testing of some of the process monitor channels. Detailed design of field-installation hardware, wiring, and shielding is under way. Panel installation design is nearing completion. Personnel-monitoring instrument requirements remain as previously reported. Individual instrument locations were specified, requirements of power and alarm circuitry were determined, and the designs of power and signal junction boxes were completed. Several prototype alarm modules for the master alarm panel were fabricated and tested. Panel installation design is under way. Thermocouples. Five drawings showing the thermocouple locations and attachment details for the fuel and coolant salt circulating systems, flush and drain system, and the coolant salt radiator tubes were completed. A list giving all thermocouples in the reactor system and the proposed method of readout was prepared. There are presently 819 thermocouples. Of these, 169 will be connected to the data logger, %65 will be connected to tempera- ture scanners, 42 will be connected to a precision temperature indicator, 42 will be connected to multipoint recorders, 78 are permanently assigned 52 to miscellaneous process instruments, and 123 are unassigned. The latter are presently terminated at the patch panel but can be readily connected to indicating or recording instruments by means of patch cords. Designs of attachments and methods of installation of the 120 thermo- couples on the radiator tubes were completed, and the thermocouples and attachment devices were fabricated. Installation of the thermocouples on the radiator tubes is under way. Analog Simulation of Control-Rod Servo Mechanism. The servo control- ler was modified as shown in the simplified diagram of Fig. 3.1l. Prelimi- nary results of analog computer simulation using this controller to hold the reactor fuel outlet temperature constant are encouraging, and additional development studies are under way. ~_ Simulation of Reactor Fllllng and Draining Transients. An analog compuuer model of the 1ill and drain system was constiructed for otudying UNCLASSIFIED ORNL -LR-DWG 78439 _9® T T 7o v N I 7 =T e L F 3 ERROR = ¢ = /:_i ~Tysp = T3) + 7c->sp N € > 38 INSERT ROD € < -8 WITHDRAW ROD 8 = [ [£ - -7)dr = 2 [ [ 2~ (5-7- AT ot 17 ~ 0.5 min = RESET TIME CONSTANT ¢ = NEUTRON FLUX F = FUEL SALT FLOW ¢ =7 AND AT CONSTANT FLOW IS 0.5°F OR 1% OF DESIGN POINT FLUX S 7o = MEASURED REACTOR OQUTLET TEMPERATURE 7. = MEASURED REACTOR INLET TEMPERATURE 7;sp= OUTLET TEMPERATURE SET POINT Tir = RESET INLET TEMPERATURE Fig. 3.1. Computer Diagram for MSRE Servo Controller Simulation. . 53 the following process control problems: (1) finding the maximum valve tem- peratures resulting from normal venting of a pressurized drain tank; (2) designing a system to prevent excessive rates .of pressure rise in the pump bowl; (3) determining reactor draining and filling characteristics for normal and ebnormal conditions; and (4) finding optimum control para- meters for the pump bowl pressure control system. It was found that passing helium at 1200°F through the drain tank vent valves during normal venting of a pressurized tank would not cause excessive valve temperatures. The valve plug and body temperatures would not exceed 500 and 250°F, re- spectively. ' Pressure surges in the pump bowl are undesirable because they could damage the pump bowl oil seal, tend to reverse the direction of purge gas flow, and tend to back salt up into the level detector dip tube, which could cause an erroneous low salt level indication. A probable worst- case incident for pressurizing the pump bowl was assumed in which im- mediately after the reactor had been filled and the freeze valve frozen, a dump demand occurred, the vent valve on the pressurized drain tank failed to open, the drain tank pressure was relieved through the bypass valve in- to the pump bowl, and the pump bowl pressure increased. In order to pre- vent an excessive pressure transient in this incident, flow restrictions were added to the model. ' When it was assumed that the bypass valve to the spare drain tank was closed, the simulation.showed that a restriction large enough to limit the pressure transient satisfactorily would have the following effect on the dump time: ‘ No Restrictor = With Restrictor Time to complete normal dump, 31 54 min Time to empty 25% of core, min 12 20 When it was assumed that the bypass valve to the spare drain tank was open and that the tank was empty, a restriction large enough to limit the pressure transient satisfactorily would have only a small effect on the dump time as shown below: With Nestrictor Vent Valve Vent Valve No Restrictor . Open Closed Time to complete 31 : 32 38.5 normal dump, min Time to empty 25% 12 12.5 14.8 of core, min Draining and filling simulations were made for a wide variety of operating conditions. Of particular interest was the case where the by- pass valve failed to open, and the draining of the salt tended to pull a 54 vacuum in the pump bowl. Resulting drain times, assuming 0.3 scfm purge flow, wvere: Spare Drain Tank Spare Drain Tank Bypass Valve Open Bypass Valve Closed Time to complete dump, 43,5 68 min Time to empty 25% of core, ik.5 18 min ' Investigations are to be made of the alternative configurations. A preliminary investigation of the pump bowl pressure controller indicated that the small purge flow rates and the large volumes would make the control quite sluggish. The control should, however, be satisfactory. for long-term steady operation. Approximate optimum controller settings are to be determined from further simulation work. Prbcurement and Installation Status Subcontract Work at Reactor Site The Kaminer Construction Corporation completed the major modifications to Building T7503. These modifications consisted of a 7-ft extension to the reactor containment cell, with the necessary penetrations and nuclear radi- ation shielding; the construction of a drain tank cell of magnetite con- crete lined with stainless steel; the installation of building exhaust ducting; modifications to the radiator ducting; construction of a contain- ment enclosure inside the high bay area; installation of miscellaneous electrical, water, and compressed air lines and equipment; and the con- struction of a remote-maintenance control room. The R. S. Hixson Company completed the major modifications outside of Building 7503. They consisted of the erection of a 100-ft steel stack, with associated blowers and ducting; the construction of a filter house for ventilating air supplied to the building and a filter pit for exhaust air; and the installation of a cooling tower. The work of both contractors passed all specified performance and leak tests. Fabrication of Major Reactor Components The machine shop at the Y-12 plant of UCNC has the prime responsi- bility for fabricating the major reactor components, consisting of the "reactor, primary heat exchanger, radiator and enclosure, salt storage tanks, and salt pumps. These components are all being made of -INOR-8, 55 except the radiator enclosure, which is being fabricated from carbon steel and stainless steel. Fabrication of the components is approximately 85% complete. Reactor Vessel and Control Rod Thimble Assembly. Fabrication of the support grid for the graphite moderator was completed. The reactor vessel and the control rod thimble assembly are approximately 80% com- plete. Fabrication of work on the control rod thimble assembly was temporarily halted pending the outcome of development work. A contract was awarded to the Westinghouse Electric Corporation for the fabrication of the control elements, which are to be hollow cylinders made of a mix- ture of gadolinium and aluminum oxides and canned in stainless steel. A contract was also awarded to the Vard Corporation for the detail design and fabrication of a prototype control rod drive. Moderator Graphite. The vendor has not yet been able to manufacture graphite to meet all the requirements of the specifications. Approximately 1000 out of the 1200 pieces needed have been formed, baked, and graphitized, but most of the pieces have small cracks in excess of specifications and are being evaluated for possible use (see Chap. 4, this report). The vendor is attempting to manufacture the remaining pieces to specification. The graphite was originally scheduled for delivery in September 1962 but will not be available until after July 1963. Thermal Shield. The 35-ton stainless steel thermal shield for the reactor vessel was completed by the UCNC Paducah Machine Shop and de- livered to the reactor site. The approximately 7O tons of steel balls that will be poured into the thermal shield for gamma-ray attenuation were also received. Heat Exchanger. The primary heat exchanger is approximately 85% complete. The shell, head, and tube bundle are shown in Fig. 3.2 before final assembly. Fabrication of the mounting for the heat exchanger was completed by the Taylor Engineering Company. A mockup of the tube bundle with 9 full-scale tubes and a partially machined full-size tube sheet was assembled prior to fabricating the unit for the heat exchanger. Problems associated with forming, fitting, welding, and back-brazing the tubes into the tube sheet were solved on the mockup, and methods were developed for visual and radiographic inspection of the welds and for ultrasonic inspection of the brazed joints. Techniques were also developed for handling, positioning, and shipping the unit to the Wall Colmonoy plant in Detroit for brazing in a hydrogen atmosphere. When the full tube bundle was assembled, the 165 U-tubes were welded into the tube sheet by use of semiautomatic equipment without any weld repairs being required. The back-brazing of the complete unit was accom- plished satisfactorily. Fuel and Coolant Pumps. Fabrication of the pump bowls, shield plugs, and bearing housings is approximately 85% complete, and the pump supports Fig. S & Primary Heat Exchanger UNCLASSIFIED PHOTO 39450 9¢ 57 were completed by the Taylor Engineering Company. Two pump-bearing lubri- cating-oil package units were completed by the ORNL Machine Shop. Fabri- cation of the drive motor vessels is under way. Radiator and Radiator Enclosure. The salt-to-air radiator is approxi- mately 80% complete. The ten banks of 12 tubes each were completed, the tube banks are being assembled to the main headers, and thermocouples are being installed on the tubes. The venturi and 5-in.-pipe subassemblies for attachment to the radiator inlet were fabricated, and the venturi was calibrated. This venturi will assure flow in the coolant piping system. The radiator enclosure is approximately 80% complete. Work was started on installation of the electrical heating equipment, but most of the work remaining on the enclosure cannot be completed until the radi- ator coil has been installed. Salt Storage Tanks. The fuel salt flush tank, the coolant salt storage tank, and the two steam domes and bayonet assemblies (shown in Fig. 3.3) for conling the two fuel salt storage tanks were completed. The tanks were pressure tested, and mass spectrometer helium leak testing is in progress. Fabrication of the two fuel salt storage tanks is approximately 85% complete. A set of thimbles for one of the tanks is shown in Fig. 3.L. The remaining work consists of welding the shell and lower head in position on both tanks. The mountings for the four salt storage tanks were com- pleted by the Taylor Engineering Company. Salt Piping and Component Heating Equipment. The reactor freeze valve heater, the coolant pump furnace, and the coolant salt line anchor sleeve furnaces were completed, and work is under way on eleven heater control cabinets, the fuel pump furnace, the coolant salt drain tank furnace, and four salt storage tank furnaces. Procurement was initiated for most of the wiring, heaters, junction boxes, cable trays, etc., and delivery was made on all items, except special units for heating the salt piping and heat exchanger in the reactor and drain tank cells. A contract is presently being negotiated for the fabri- cation of these heaters. Procurement is approximately 75% complete on electrical material for miscellaneous electrical work outside the cells. Remote Maintenance Equipment. A jig for use in assembling the heat exchanger, fuel pump, reactor and all associated piping in the reactor cell was completed, as well as a jig for attaching piping to the fuel drain tanks. The pipe-alignment brackets for freeze flanges were partially fabricated. The arrangements were completed for the purchase of thrce specially de- signed tongs to 1lift the shielding plugs and special optical tooling equip- ment. A contract was awarded for the fabrication of the large, portable, sliding shield for the maintenance facility. 58 UNCL ASSIFIED PHOTO 39404 Fig. 3.3. ©Steam Dome and Bayonet Assembly for Fuel Storage Tank. Fuel Salt Storage Tank Thimbles. UNCL ASSIFIED PHOTZ 39353 69 60 INOR-8 Piping Fabrication. Fabrication of the anchor sleeve assem- blies and 12 freeze valve subassemblies is complete. The freeze flanges, pipe bends, and welded pipe subassemblies are partially complete. Fuel Salt Sampler and Enricher. The transfer-tube and positioning- jig assembly is approximately 90% complete, and miscellaneous tools were completed. Several special bellows assemblies, miscellaneous stainless steel stock, and cable drive units were procured. Powell Valve Company is manufacturing special motorized valves. Fabrication of the main unit of the sampler-enricher, which will be located on the ground floor, has not been started. Stainless Steel Pipe and Fittings. Approximately 85% of all stain- less steel pipc, tubing, valves, and fittings is on hand. Most of the pipe, tubing, and fittings were purchased in accordance with ORNL speci- fications that require more rigid inspection than is required by ASTM specifications. These specifications included liquid penetrant and ultrasonic inspections that were pertormed at the various vendors' planls. Numerous special adaptors and disconnects were fabricated in UCNC shops. Support Structures. In addition to the special mountings mentioned above, valve hangers and miscellaneous pipe and equipment supports were fabricated. Reactor Auxiliary Systems. ©Special equipment was fabricated that included helium dryers for the cover gas system, oil catch tanks, air and helium peretration plugs for the drain tank cell, the penetration plug for the sampler-enricher, the blower and filter frame, and freeze valve cooling air nozzles. Three stainless steel heat exchangers, activated charcoal for the charcoal beds, two 75-hp motors and blowers for compo- nent air cooling, two stainless steel feedwater tanks, and two component air cooling vessels built to Boiler Code specifications were obtained from vendors. The Young Radiator Company is fabricating three space coolers for cooling the cell air. Instruments. Approximately 909 of the commercially available standard components required for the process instruments were received, as well as some components that required special development or procurement effort. The special components include the drain tank weigh system, the venturi flow element for the coolant system, thermocouple wire and exten- sion cable, and the thermocouple disconnects. Vendors fabrication drawings were reviewed and approved for weld-sealed transmitters, valves for radio- active helium gas service, and helium flow elements. Purchase orders were placed for an addition to the thermocouple patch panel and for a high- temperature NaK-filled differential pressure transmitter. A proposal for the acquisition of a data system was prepared. This proposal recommended that a specific data-handling system selected on the basis of preliminary proposals be acquired. The recommended method of acquirement was to lease the system for one year and then to purchase it. Table 3.1. MSRE Expenditures and Cost Estimate Costs in Thousands of Dollars Expenditures FY 1963 Cost Presernt New ‘Contingency Through Through Working Estimate Portion of FY 1962 December 1962 Estimate ’ New Estimate I. Building and equirment modifi- 1011 554 1724 1775 32 cations , ‘ IT. Reactor installation 0 - 109 557 856._ 196 III. Thermal shield and squipment -9 159 121 236 14 support structures IV. Reactor materials 447 ‘315 830 954 33 V. Reactor equipment 239 390 1256 1235 198 VI. Instrumentation and controls 33 186 513 698 126 VII. Engineering, inspection, and 1883 566 1804 - 2831 180 overhead Contingency 287 Total 3622 2279 7092 . 8585 779 19 62 AEC approval of the proposal was received on December 5, 1962, but per- mission to start procurement was deferred until a Firm operation date is established. An order for freeze flange and freeze valve temperature alarm switches was placed with the Electra Systems Corporation. The equipment is to be shipped in February 1963. The selection of the Electra switches was based on the low bid and the satisfactory operation of test units on a freeze valve test facility. Major instrumentation items for which procurement has not been initiated include nuclear instruments, components for the temperature- scanning system, weld-sealed solenoid valves, control circuit relays, area surveillance and remote maintenance television, instrument emer- gency pover static inverter, personnel radiation monitors, and the main operating console. ' MSRE Cost Estimate Revision A new estimate was made of the cost of the Molten-Salt Reactor Experiment, and the total cost for engineering, inspection, procurement, construction, and installation of the reactor increased from $7,092,000 to $9,364,000. This estimate includes a contingency of $779,000. The new estimate is compared with the present working estimate in Table 3.1. THIS PAGE WAS INTENTIONALLY LEFT BLANK 65 L. METALLURGY Heat Exchanger Fabrication Sample Tube Bundle A full-scale tube bundle was fabricated to test welding and brazing procedures. This assembly, which simulated closely the mass and actual design of the tube bundle for the MSRE heat exchanger, constituted a final proof test of the fabrication procedures. The sample was assembled, welded, shipped, and brazed in the manner prescribed for the actual unit. The sample tube bundle is shown in Fig. 4.1 in the welding fixture. Several full-length U tubes were included to determine whether unsup- ported tubes would distort and oxidize during brazing. The full bundle of tubes, of which 52 were welded and back-brazed, was used at the tube sheet to obtain gas and heat baffling effects that might influence the quality of the braze. The 52 welded and back-brazed Jjoints were located so that half were near the center of the tube sheet and the remainder were on the perimeter at 3, 6, 9, and 12 o'clock positions. Stainless steel tubes were tack-welded in the remaining holes. The welds were made semiautomatically using the prescribed conditions, and no difficulties were encountered. Visual and radiographic inspection indicated that all the welds were sound. During the heating cycle of the brazing operation, a temperature difference of approximalely 4OO°F was observed between the tube sheet and the extreme end of the U tubes, and there was a difference of 150°F between the center and the perimeter of the tube sheet. The planned holding period at 1650°F reduced these differences to less than 100°F, however, and at the brazing temperature all thermocouples read well with- in the specified 1835 to 1885°F range. After brazing, visual inspection of the sample heat exchanger showed that all INOR-8 components were bright and all brazed joints had good fillets. No distortion or tube sagging was evident. Ultrasonic inspection revealed all joints to be completely brazed and to have only minor, scattered porosity. This was later confirmed by metallographic examination of selected brazes. A typical tube-to-tube sheet weld and back braze is shown in Fig. 4.2. This photomicrograph illustrates the uniform weld contour obtained and the flow of brazing alloy to the root of the seal welds. Tube Joint Inspection Seventy-nine brazed tube-to-header joints in two different mockups of the Lube bundle for the MSRE heat exchanger were inspected using the Lamb-wave technique described previously. Thirty joints were evaluated in the first mockup with the use of a 1/16-in.-diam flat-bottom hole as a reference for calibration sensitivity. One tube was found to be 66 UNCLASSIFIED PHOTO 38928 Fig. 4.1. Sample Tube Bundle in Welding Fixture. completely unbonded. Metallographic examinations and ultrasonic findings were correlated, and the correlalion indicated that thc test was very sensi- tive to small pores that would not affect braze integrity. Therefore a 3/52-in.-diam flat-bottom reference hole was used as a primary standard to evaluate 49 brazes in the MSRE sample tube bundle. The 1/16-in.-diam refer- ence hole was used as a secondary standard, and five of the joints were re- evaluated at this sensitivity. None of the joints showed unbonded areas when compared with the 5/52-in.-diam primary standard. One of the five 67 UNCLASSIFIED Y-46953 Fig. 4.2. Welded-and-Back Brazed Joint Showing Sound Weld and Good Flow of Brazing Alloy to Root of Weld. joints inspected with a 1/16-in.-diam reference hole had an unbonded area. This area was found by metallographic examination to contain only minor porosity in the braze. The MSRE tube bundle will be inspected using the 5/52-in.-diam refer- ence hole. Scanning of the joints will be accomplished with the mechanical device described previously, which, as modified, allows automatic x-y re- cording of unbonded areas. MSRE Tube Bundle Using the procedures demonstrated on the sample, the MSRE tube bundle was successfully assembled and all 326 tube-to-tube sheet joints were welded. 68 The tube weldments were inspected dimensionally by using plug gages and for soundness by using radiography. All welds permitted passage of a 0.4125-in.-diam plug gage, and radiography revealed no porosity in any of the welds. Mechanical Properties of INOR-8 ASME Boiler and Pressure Vessel Code Allowable Stresses for INOR-8 The properties of INOR-8 were reviewed by the ASME Boiler Code Com- mittee, and this material was approved for code construction in code case 1515, subject to the allowable stresses llsled In Table 4.1. These values now supersede previous design values established for the MSRE. lable 4.l. Maximum Allowable Stresses for LNUK-8 Reported by ASME Boiler and Pressure Vessel Code Committee Maximum Allowable Stress (psi) Tem?f;?ture Material Other . Than Bolting g 100 25,000 10,000 200 24,000 9,300 300 23,000 8,600 40O 21,000 8,000 500 20,000 7,700 600 20,000 7,500 700 19,000 7,200 800 18,000 7,000 900 18,000 6,800 1000 17,000 6,600 1100 13,000 6,000 1200 6,000 3,500 1300 3,500 1,600 Reactor-Quality INOR-8 Mechanical properties are being determined on random heats of INOR-8 from which MSRE reactor components have been fabricated to evaluate the effects of large-scale production and improved quality requirements. Some tensile tests and stress-rupture tests were completed. The tensile properties of plate material made from heats 5075 and 5081 were determined in the range 7O to 1800°F, and the data are given in Table 4.2. The values quoted for ultimate tensile strength and 0.2% yield strength are averages for four specimens, two from each heat of Table 4.2.- Average Tensile Properties of INOR-8 Ultimete Tensile 0.2% Yield Elongation Reduction in Area Temperature Strength® Strength@ (%) (%) (°F) (psi) (psi) Heat 5075 Heat 5081 Heat 5075 Heat 5081 T0 113,600 46,500 532 548 600 103,220 36,000 55 50 800 100,120 35,000 5% 52 1000 96,020 33,200 5% 46D 50D 1200 74,820 30,600 ppb 360 28 36 14C0 61,620 31,800 21 30 23 30 16C0 56,400 31,600 23 Lo 25 b3 1800 20,300 20,000 28 28 29 29 ?Average values for four specimens consisting of two from heat 5075 and two from heat 5081; one specimen of each heat was cut parallel to the plate rolling direction and the other normal to the rolling direction. bAverages for two specimens. 69 70 material, with one specimen being cut parallel to the plate rolling direction and one normal to the rolling direction. The two heats behaved similarly except with respect to elongation and reduction in area at high temperatures, and no variation in properties could be attributed to direction in the original plate. Stress-rupture data for heat 5055 tested in air are listed in Table k4.3, A few longer tests on this heat are presently under way, and stress-rupture testing was started on heats 5075 and 5081. Table L4.3. and 'l'estéd in Air Stress-Rupture Data for INOR-8 Specimens Cut Parallel to Plate Rolling Direction (Heat 5055) Temperature Stress Time to Rupture Strain at Hupture (°F) (psi) (hr) (%) 1100 81,000 5 57 70,000 33 20 61.,000 140 5 50,000 1050 6 Lk, 000 2430 2 1300 52,000 6 30 39,000 30 16 %k, 000 68 16 31,000 160 38 27,500 350" 29 22,000 510 15 25,000 860 50 20,000 1700 28 18,000 2700 25 1500 2%,000 14 48 15,000 90 L3 12,500 190 3 10,500 390 23 8,200 910 21 The results of these tests indicate that.the mechanical properties of these heats are significantly better than design values established for the MSRE based on the previously available material. Evaluation of MSRE Graphite Graphite bars selected from material produced for the MSRE moderator were evaluated to establish the acceptability of the lot for reactor use. 71 The material was reported to meet specification requirements except that there were cracks and some bars had low bulk densities. The effects of these deviations from the specifications on mechanical properties and on salt permeation were determined, and the cracked condition was studied. Examinations were made of two base-stock graphite bars and four bars of completely processed graphite supplied by the vendor. The base-stock graphite had a bulk density of 1.66 g/cms, which increased to 1.82 to 1.88 g/cm® in the final fabrication operations. Most of the bars had bulk densities of approximately 1.85 g/cm®. The microstructure of the completely processed graphite was found to be relatively complex. It appeared to be fine-grained graphite fabricated from at least four different sources of carbon. Visual and radiographic techniques were used to evaluate the cracks, which tended to be near the longitudinal axes of the bars. No relation- ship existed between the size of opening and the lengths of the cracks, which ranged up to 0.004 in. in opening and up to 9 in. in length. Crack widths ranged from 1/8 to 1 1/4 in. and averaged 5/8 in. Most of the cracks had openings of 0.001 in. (25 microns) to 0.002 in. (50 microns) and were less than 3 in. long. It was calculated that cracks of the latter size that intersected the external surfaces of the bar would fill with molten salt at a pressure difference of 20 psi. Fabrication information for this graphite indicated that the cracks developed during the final graphitizing operation. The base-stock graphite examined was essentially free of cracks; only two, short, closed cracks were detected radiographically. A microscopic examination of completely processed graphite showed impregnating material in some of the cracks, however, and thus some cracking must have occurred at pro- cessing steps earlier than the final one. A more detailed study of this is being made. Salt Permeation of MSRE Graphite A comparison was made of the permeation by molten salt and mercury of base-stock graphite and finished graphite under the conditions given in Table 4.4. The results are summarized in Table 4.5. The data show that permeation of CGB graphite by mercury or by molten salt is within specification limits. It is significant to note that the 150-psig molten salt pressure used in the standard permeation screening test for graphite is approximately three times the maximum expected in the MSRE and that the molten salt that permeated the graphite in the screening test was found only at the external surfaces and in the cracks. The latter finding indi- cates that the pore structure of the graphite is fine enough to prevent impregnation by the molten salt. Thermal Cycling of Salt-Impregnated Graphite Laboratory thermal cycling tests were run on CGB graphite with salt- impregnated cracks to determine whether damage would occur under conditions simulating the draining, cooling, and reheating of the MGRE core. 72 Table 4.4. Test Conditions for Standard Graphite Permeation Screening Tests Mercury Molten-Salt® Test Test Temperature, °F 70 1300 Test period, hr 20 100 Pressure, psig 470 - 150 i F-BeFg-ThF,-UF, (67-18.5-14-0.5 mole %) . : | Table 4.5. Bulk Density and Salt Permeation of Base- Stock and Completely Processed MSRE Graphite Mercury o Bulk Volume Bulk Density Permeation Permeated (g/cm®) (wt %) by Salt (%) Specification 1.87¢ . 3.5 0.5 CGB finished material 1.83 1.85 0n.02 CGB finished material 1.87 0.09 CGB base stock 1.66 : n1.kL AGOT graphite 1.68 13.9 ®specimen size: 0.125 x 1.50 X 1.50 in. bSpecimen size: 0.5 in. diam x 1.500 in. C,... . Minimum value. The thermal cycling tests were made on 2-in.=-long transverse sections cut from a machined CGB graphite bar in a zone having a relatively high concentration of cracks. All the specimens had a bulk density of 1..87 g/cm>. Two specimens were impregnated with LiF-BeFo-ThF,-UF, (67-18.5-14-0.5 mole %) at 1300°F during a 100-hr exposure at 50 psig, and two were im- pregnated at a pressure of 150 psig. The lower pressure was selected be- cause it is approximately the maximum pressure expected in the reactor. The higher pressure was selected to ensure that the cracks would be more deeply penetrated by salt in the test than in the reactor. The average 73 bulk volume of the specimens permeated by the salt was 0.08% at 50 psig and 0.13% at 150 psig. The salt impregnation appeared to be limited to the fissures, as shown in Fig. 4.3. UNCLASSIFIED PHOTO 60329 (a) I | I I | | | Y-48961 Fig. 4.3. CGB Graphite Specimens Impregnated with Molten Salt at 50 psig. (a, b, c) As radiographed, (a) After 100 thermal cycles. T4 The specimen with the greatest salt permeation at each pressure was tested by thermal cycling in an argon atmosphere at a pressure that varied from 10 to 20 psig, depending on the temperature. The specimens were sub- Jected to 100 thermal cycles in which the temperature was increased from 390 to 1300°F in the first 9 min of each cycle. After this treatment there were no detectable changes in the appearance of the graphite or in the lengths of the cracks. Each specimen lost approximately 38 wt % of the original impregnating salt. Additional studies will be made of the effects of thermal cycling on graphite with salt-impregnated cracks under a qualitative simulation of a hot spot in the MSRE. Specimens will be impregnated at 1300°F and then rapidly healed from 390 to 183%2°F. Tensile Strength of CGB Graphite Tensile strengths were delermined f'or CuUB graphite 1n bolli Lhe Liaus- verse and parallel directions by a ring technique similar to that reported by Bradstreet and his co-workers® and by conventional tensile bar tests. Two sets of rings were machined, with their axes of revolution rcspectively perpendicular to and parallel with the extrusion direction of the graphite bar. The rings were 1.250 in. OD, 0.500 in. ID, and 0.500 in. long. Radi- ography was used to identify cracks in the rings that were oriented perpen- dicular to the directions of maximum stress during testing. The average tensile strengths perpendicular to and parallel with the length of the bar were 4800 and 6500 psi, respectively. There was no significanl trend that could be related to the presence of cracks. The graphite was apparently not notch sensitive, since stress failure cracks frequently terminated with- in the body of the rings. A more comprehensive tensile test of CGB graphite was made using sixteen 5-in.-long cylindrical specimens (Fig. 4.4a). Nine of the tensilc specimens were not appreciably affected by the cracks and yielded an aver- age strength of 5440 psi and a modulus of elasticity of 3.0 x 10° psi. A typical fraclure is shown in Fig. l.lkc. The specimens that demonstrated a definite effect of internal cracks had a minimum strength of 1510 psi and an average strength of 2940 psi. These test bars all fractured in a stair-step fashion (Fig. 4.4b), and in some cases the axial crack was longer than 1/2 in. The specimens that were greatly reduced in strength failed in a two-step process that was quite audible. Apparently the axial crack caused one side of the specimen Lo fail and create the first audible snap. The specimen continued momentarily to support the load before the eccentric loading caused sufficient bending to fracture the olher side and causc the second audible snap. The tensile strength of the graphite parallel to the bar length was 5440 psi in the round-bar tests as compared with 6500 psi in the ring tests. 1S. W. Bradstreet el al., WADC-TR-59-708, Wright Air Development Center, pp. 85-10T7. 75 UNCLASSIFIED Y-48954 . , P ‘ A B C Fig. 4.4. Tensile Specimens of CGB Graphite. (a) As machined. (b) Typical stair-step fracture. (c) Normal fracture. Since the bulk densities of the specimens were essentially the same (1.85 and 1.86 g/cm®), it is believed that the higher strength value obtained with the ring specimens is primarily due to the type of stress distribution im- posed by the ring configuration. The 5440O-psi value for tensile strength in a direction parallel to the bar length is considered to be more accurate for flaw-free CGB graphite. Based on the ring-test data, transverse strength would be approximately 75% of the longitudinal strength, or 4000 psi. These data indicate that the weakest cracked bar tested met the speci- fied tensile strength requirement of 1500 psi and that even this bar was stronger than the commonly used nuclear-grade AGOT graphite. The CGB graphite that was essentially free of cracks was three to four times stronger than AGOT graphite. 76 Development of Gd-03-Al-05 Control Rod Elements Sample Control Rod Element Testing Prototype control rod elements consisting of Gds03-Al-05 (70-30 wt %) ceramic bodies canned in austenitic stainless steel were procured to de- termine the fabricability of the rod design. These elements were subjected to general inspection and to testing that included exposure in the control- rod testing rig described in Chapter 2 in order to determine soundness, dimensional stability, and general suitability for MSRE use. The elemente were designed to fit over a flecxible hose that would move through a crooked path. Difficulty in meeting dimension specit'ications and density requirements caused the supplier (Dresser Products, Inc.) to abandon cold-pressing methods for fabricating the ceramic pieces. The elements were produced by a hot-pressing technique, and deviations from specified dimensions were less than 0.008 in. on the diameter and 0.010 in. in the length. Densi- ties ranged from 96 to 99% of theoretical. Radiography of uncanned ceramic pieces revealed no evidence of cracks or lack of soundness. The canned ceramic bodies showed some axial cracks on the ends that were probably caused by local heating during welding of the end caps. Two canned elements were exposed to repeated thermal and mechanical shocks in the control-rod testing rig, as described in Chapter 2. They were examined radiographically and dimensionally after 24 hr (one cycle), 350 hr, and 600 hr of testing that included approximately 11,000 cycles and 1700 scrams. Holes were made in one can to expose Lhe Gds03-Al505 to air for the final 250 hr of test. Dimensions of the metal cans were not changed by the testing. Axial cracks were observed radiographically after the first cycle, and severe cracks in both the axial and transverse direction were observed at later stages. No crumbling or ratcheting was evident, however, and no condition was observed that might affect the nuclear or mechanical performance of the control rod elements. Sintering Characteristics of Gd>03-Al>03 Specimens Sintering studies were continued on Gdz03-Al503 cylinders. The compo- sition 70 wt % Gds05—30 wt % Als03 was used in all experiments run during this report period. The Al>03 powder used was a high-purity commercial product (ALCOA grade A-303). The major impurities in the Gd,0s; powder were 200 to 500 ppm of europium, 200 ppm of silicon, and 100 ppm of calcium. In preliminary experiments it was observed that a cold-pressed mechani- cal mixture of this compositon melted when heated to 1750°C. A similar body was heated to 1650°C and became severely distorted, with the diameter at the top larger than at the bottom; in addition, the top portion was visibly porous and the bottom portion was dense. Apparently liquid had formed at 77 1650°C and was able to flow and settle in the lower regions. No evidence of liquid loss was detected on the spheroidal, high-fired Gd>05 particles (-35 +100 mesh) used as setter material, however, and no weight change was detected after sintering. A series of experiments was then carried out with "prereacted" powder in an attempt to overcome the problems encountered when using the rather standard fabrication procedure described above and to better understand the sintering characteristics in this system. The prereaction step con- sisted of heating a dry-blended and cold-pressed (7500 psi) mixture to 1650°C for 1/2 hr in hydrogen. The product was then ground to a -325 mesh size ( for capsules 6, 24, and 36. The gas samples were obtained by a puncturing operation which released the gas 1nto an evacuated collection manifold. After puncturing the first -two capsules, the collection manifold was thoroughly prefluorinated. Usually, four, *Oak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Feb. 28, 1962," USAEC Report ORNL-3282, p. 105. ®Tvid., pp. 110-112. S0ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962," USAEC Report ORNL-3369, pp. 112-116. Table 5.1. Results of Mass Spectrometer Analyses of Cover Gas From Capsules Irradiated in Experimental Assembly ORNL-MTR-47-4 Gas Quantities In Large Capsule No. 12 In Small Capsule No. 4 In Large Capsule No. 45 Consi?iuents vol % cm® (STP) vol % em® (STP) vol % em® (STP) Total® 153 98 270 3 47.0 71.9 90.L 88.5 88.3 238.4 CO + No 2.3 3.5 0o 4.9 » 7.5 0.5 0.5 0.34 0.92 Ar 0.02 0.03 0.8 0.8 He L.5 6.9 3.3 3.3 2.96 7.97 COz 1.2 1.8 0.4 0.k 1.8 CF4 35 53.5 .2 k.1 8.1 21.87 Kr 0.16 0.245 0.29 0.29 0.08 0.22 Xe 0 0] 0 SiFs 1.0 1.5 0.04 - 0.04 0.1 0.27 Other . 3.8 5.8 0.1 0.1 0.24 0.65 %Total volumes were based on pressure measurements. I ‘l + 8 Table 5.2. Data on Fo and CF4 Produced Following an Exposure of Experimental Assembly ORNL-MIR-47-4 to an Integrated Flux of 102° Neutrons/cm® Fissiorn Krypton Xenon Milliequivalents : - Total F Power Found Found of F in Gas Decay Total F Converted Capsule Densuya (% of theo- (% of theo- Per Millimole c¢f Period in Gas to CFq4 Identification (w/cm®) retical) retical) U Burned (days) (meq) . (%) N . Small capsules . « 6 43 145 ' 140 0.2 €3 0.012" 100 L 260 & 0 23.4 137 8.6 8.5 Large capsulesb 2l 83 140 0 28.0 71 8.2° 31.4 36 & 1254 0 59.6 95 17.2 18.2 45 : 117 65 0 €1.5 134 25.3 15.5 12 €T 130 0 72.9 141 16.7°¢ 57.2 aCalculated from cobalt activity in stainless steel dosimeter wirz. bCapsules listec in order of sampling. “Fluorine analyses were corrected to include fluorine reaction products formed during sampling. An average based on Kr-to-He and Kr-t>-CF4 ratios in three confirmatory ‘samples; direct analyses from the first samples were higher by a factor of 2. . (A2 83 evacuated, 200-ml collection bottles were opened and closed in succession so that a little more than one-half the remaining gas was collected each time at successively lower pressures. The higher pressure samples were analyzed promptly by mass spectrometry, and a few of the lower pressure samples were analyzed later for confirmation. Unfortunately, samples that were allowed to age showed evidence of reaction of the fluorine, and the data obtained were usually not regarded as quantitative. The data of Table 5.2 are tabulated in chemical terms to facilitate assessment of radiation-induced chemical alteration of the fuel. There were about 875 milliequivalents of fluoride ion in the large capsules, of which 2% could be accounted as UF4. In the extreme case, about 3.5% of the total fluoride was converted to Fo or CF4 in the gas phase. The second column in Table 5.2 reflects the different flux levels at different capsule locations within the sodium metal bath used as a heat transfer or cooling medium during irradiation; the high value for capsule I arose from a higher uranium concentration. The dosimeter wires indi- cated about twice as great a range of flux levels as was anticipated. Comparison of the krypton and xenon analyses with expected yields reveals a holdup of xenon that is attributable to the formation of the recently discovered, relatively nonvolatile, compounds XeF- and XeF4. The literature provides a recent reference? to a reaction of xenon with Fo at room temperature to produce, primarily, XeF,, in which the activa- tion energy was supplied by a mercury-arc lamp with an effective wave length of 2500 to 3500 A (~0.25 ev) that presumably produced fluorine atoms. In view of the analytical precision available for the measured krypton production, the agreement with the theoretical prediction was as good as could be expected. More abundant gaseous yields, such as Fy and CF4, were amenable to measurement to within probably better than 3%, even though these samples were analyzed under less than ideal conditions. The sixth column (Table 5.2) affords an indication that, within cap- sule types, longer decay periods corresponded to larger relative yields of fluorine in the gas phase, although the correlation is ragged. The total fluorine data include both Fp and CF4, which was interpreted as a secondary product from the reaction of fluorine with graphite. An assess- ment of fluorine generation as a function of estimated decay energy has not yet been attempted, but the variation in total fluorine released per unit of uranium burned indicates that no consistent correlation with de- cay energy will be found. Where small but significant amounts of 0o were encountered, the fluorine content was corrected to include this oxygen, which was also considered to be a reaction product of the fluorine which came from the collection apparatus. The air in-lecakage, recognized by the presence of Ny, never exceeded a few percent, and thus did not sig- nificantly affect the analyses. 4J. L. Weeks, C. L. Chernick, and M. S. Matheson, "Photochemical Preparation of Xenon Difluoride," J. Am., Chem. Soc., 84:4612 (1962). 84 The amount of fluorine in the gas phase per unit of energy generated ‘during the exposure (column 5, Table 5.2) was a possible basis for com- paring the results from different experiments. Two of the four capsules irradiated in experiment ORNL-MIR-4T7-3 included relatively porous graphite boats that were prepermeated with fuel;l these capsules yielded relative- ly little fluorine, all in the form of CF4, and the data agreed with the results from capsule 6 of experiment ORNL-MIR-4T7-L4, even though the power density in experiment LT-3 was more than four times as great. The other two capsules in experiment LT7-3 gave ten times as much fluorine, again as CF4, but they were still low in total fluorine production by a factor of more than ten in comparison with the nmodal trend in experiment L7-4. The gas-collecting system used in experiment 47-3 was not compatible with Fj and, in retrospect, was recognized to have consumed some fluorine, although probably not enough to negate the rough comparisons cited above. 'I'hese comparisons indicate that neither power density nor total energy was di- rectly controlling in a simple manner. The predominance of CF4 in con- trast with unreacted F, in the gas space of the W'(-3 capsules could have been influenced by the large area of accessible graphite surface (a fea- ture of the design) in the presence of a radiation field. Recent control experiments in the absence of radiation (see later section, this chapter), as well as evidence from the literature, indicate that Fy at room tempera- ture and moderate pressures is not necessarily reactive toward graphite. Since 11.2 cm® (STP) of Fs, or 5.6 cm® (STP) of CF4, are one milli- equivalent, the values in the last two columms of Table 5.2 can easily be converted to volumes. ‘The amount of CF4 produced, which might be ex- pected to depend on the Fs pressure, as well as on the interrelated decay energy and time, appeared to increase with decay time in a sporadic man- ner, suggesting an influence of the accessibility of the graphite core, which is dependent on the conditions in the frozen salt ingot. The indi- cated amounts of gas accumulated in the large capsules in a vapor space of about 4.5 cm3, which was not occupied by frozen salt or graphite. Intercommunicating voids within the graphite were estimated to have a volume of 0.25 cm~. Two of the capsules produced fluorine in an amount corresponding to the contained UF4; this was thought to be a coincidence, although in the liquid state, where equilibrium could have prevailed, such a coincidence would have been regarded with suspicion, since uranium is the most easily reduced of the fuel components. Metallographic Evaluation of the Wall of a Capsule from Experimental Assembly ORNL-MTR-47-4 When the large amounts of fluorine evolution were first recognized upon puncturing capsule 24, a detailed metallographic examination of* the capsule was made. The INOR-8 wall, nominally 50 mils thick, was examined at magnifications up to 1000X for evidence of Fo attack and for possible irradiation-accelerated corrosion. The inner surface was exposed to fuel for 1500 hr at a flux of 2 x 10° neutrons/cm®-sec and temperatures up to T50°C. 85 A preliminary evaluation of the capsule at low magnification, shown in Fig. 5.1, indicated no general attack of the magnitude expected from Fpo at elevated temperatures. All surfaces, including the salt-vapor in- terface, appeared to be uniformly unaffected. No bulging or distortion of the container was noted, such as might be caused by the high pressures of Fo potentially present at elevated temperatures. A thin nonmetallic film was found on parts of the inner metal surface that was easily re- moved by scratching; it was not present in sufficient quantity for analy- sis. The surface of the metal did not contain uranium, in accord with observations from autoradiography and chemical analyses of material leached from the capsule welds, as discussed below. " - E B e > : UNCLASSIFIED - : o . 3 R-11213 Fig. 5.1. Cross Section of Capsule 24 Imbedded in Resin. High-magnification examinations were made of sections of the capsule wall in the vapor region and at various locations exposed to salt, in- cluding the interface where maximum corrosion was expected and the weld- ment. No attack, either uniform or preferential, was observed in any of these specimens. A transverse section, as shown in Fig. 5.2 along with 86 UNCLASSIFIED PHOTO 60623 S w - . ? ool 0.03 ETCHED CONTROL ETCHED IRRADIATED Fig. 5.2. Transverse Sections of Walls of Capsules 24 and Its Con- trol Specimen. a control specimen, measured 0.050 in. in thickness, as compared with 0.0495 in. for the control, indicating no loss of wall thickness. It was evident from the metallographic examination that the severe attack normally caused by Fo- gas on nickel alloys at these high test tem- peratures was not present. This lent confidence to the view that high- pressure fluorine was not generated by the molten salt. The examination also confirmed previous experimental work which indicated that irradi- ation does not cause accelerated attack in this system. 87 Autoradiography of Unopened Capsules A pinhole camera was developed and used for obtaining photographs of all unopened capsules by their own gamma irradiation, except capsules 2L and 36, which were opened before the camera became available. The auto- radiograph shown in Fig. 5.3 of capsule 12 required an exposure of 22 hr. UNCLASSIFIED HCO-1817 Fig. 5.3. Gamma-Ray Autoradiograph of Capsule 12. 88 The photographs were scanned for evidence of precipitation of uranium on solid surfaces, where tell-tale evidence of imbedded fission products might have been discovered; none was found. The distribution of fission products appeared to be uniform throughout the salt, except for bulbous regions of low activity that were distinguishable at the rounded bottoms of capsules 12 and U45; these were found to be voids, resembling gas bub- bles, that had stuck to the metal wall. Some of the specimens were photo- graphed in a tilted position, as in Fig. 5.3, so that the surface of the salt was visible at an angle. Autoradiographs of recovered graphite cores were also made with the same camera; here the activity was less uniform, but any pronounced concentrations were associated only with a few visible remnants of salt that had not drained or separated cleanly and that could be explained by circumstances other than the wetting of graphite by the melt. If loss of fluorine by the salt had occurred in the liquid state, recognizable deposits of uranium should have been in evidence; the absence of such deposits is a further indication that the partitioned fluorine was not present prior to or during the last time the fuel was molten. Bubble Formation in Capsules The fuel contracts about 20 to 25% on freezing, and consequently the presence of contraction voids is not unusual in frozen ingots. The voids, first noted as a peculiarity in the autoradiographs of capsules 12 and 45, were examined after the capsules were sectioned in a longitudinal plane. The voids were unusual in that they were located at the bottom of the cap- sules and in that their shape and size were those of approximately hemi- spheric bubbles 1 to 3 mm in diameter. Further, they were positioned on bare metal, as shown in Figs. 5.4 and 5.5. Re-examination of sections from capsule 24 and the smaller capsules led to the conclusion that bubbles had occurred in all capsules but that no metallographically detectable effects were associated with their presence. They were presumed to have been filled with helium of the samc composition as the cover gas. The inner metal walls at the bottom of the capsules were, nominally, 30°C colder than the hottest metal, and the gas was less soluble in cold fuel than in hot. Although several relatively plausible mechanisms for the growth of these bubbles are available, the apparent metastability against buoyancy remains as a curiosity. The contraction voids in control cap- sules, as revealed by x-ray photographs, were normal. No deleterious indications with respect to compatibility were found in the exposure- induced bubbles, and no bubbles of appreciable size were found adjacent to the graphite cores; against the bottoms of the graphite liners in the smaller capsules, however, there were a number of flattened voids, or bubbles, that were 0.5 mm in height and more than three times that in diameter. Chemical Analyses of Materials from Experimental Assembly ORNL-MTR-47-L4 Chemical analyses were made which indicated that no uranium had de- posited on the capsule walls during experiment 47-4. There was concern on this point, mainly because of the considerable loss of fluoride from the irradiated salt. These results conform with previous indications that UNCLASSIFIED HCO 1854 Fig. 5.4. Bubbles Near Bottom of Capsule 12. The graphite core was re- moved before this photograph was taken. There were no bubbles on the sides of ‘the capsule. UNCLASSIFIED HCO-1856 e Fig. 5.5. Matching Half of Capsule 12 Showing Arrangement of Bubbles in Salt. 68 20 there is no loss of uranium from the fuel either with or without irradi- ation. A section of the capsule wall, approximately 6 cm® in area, which had been exposed to the molten salt in capsule 24, was selected for checking for uranium deposition. The adhering salt, which might have obscured the results, was removed by dissolution in a dilute ammonium oxalate solution, and the specimen was then rinsed with distilled water. The specimen was leached for 2 hr with hot 6 M HNO3 to dissolve any deposits in the surface of the metal. The leach solution was analyzed for uranium by a fluoro- metric method, which had a limit of detection of 10 "9 g of uranium per milliliter. If any uranium was alloyed with the surface of the INOR 8, the amount was evidently less than 2.0 x 10 . g of uranium per cm® of sur- face. A monatomlc layer of uranium metal would contain about 5 x 107 of uranium per cm® of surface. As originally planned, the salt from two of the capsules was to be removed by melting under an inert atmosphere so that samples suitable for reducing-power determinations could be preserved. As previously indi- cated,> through no fault of the apparatus, the plan did not work. The melting allowed reduced phases to form that probably included metal. Evi- dently this led to a slurry that would not drain completely; the one-half to two-thirds which did drain was pale green in color from the presence of UF4, which showed that a uniformly complete readjustment of the reduc- tion had not occurred. The portion melted out of capsule 36, which was the only capsule on which a melt-out was attempted, gave an analyzed re- ducing power of 0.34 milliequivalent per gram as compared with the 0.69 milliequivalent per gram expected on the basis of the fluorine content of the cover gas. It was therefore concluded that most of the reduced spe- cies had not drained. The uranium concentration was 2.56 wt % compared with the 3.89 wt % expected for the fuel as a whole. For comparison, the fuel from capsule 24, even though it had been exposed in lump form to air, was also analyzed. A reducing power of 1.5 milliequivalents per gram was found, compared with the 0.34 value expected on the basis of the cover gas analysis. There is a possibility that ad- ditional fluorine escaped during the 4O-day interval for capsule 36 and the 51-day interval for capsule 24 between puncturing and chemical analy- ses. The longer delay might explain the discrepancy for capsule 24, at least in part. In capsule 24, the uranium analysis gave 3.82 wt % com- pared with the nominal value of 3.89 wt %. The isotopic distribution of the uranium was that shown in Table 5.3, but further work on the isotopic distribution was postponed until samples from all capsules could be eval- vated at the same time in an effort to clarify discrepancies in the amount of burnup. ®0ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962," USAEC Report ORNL-3369, p. 11k, 9L Table 5.3. Distribution of Uranium Isotopes in Samples from Experimental Assenibly ORNL-MIR-L4T7-4 Quantity of Uranium Isotope Present (at. %) Capsule 3A Mass of Uranium (unirradiated Capsule 24 Capsule 36 Isotope control) 233 <0.05 0.25 + 0.05 0.33 * 0.05 23L 401 £ 0,05 1.02 £ 0.06 1.03 £ Q.05 235 93.07 * 0.10 91.48 + 0.1 90.49 + 0.10 236 0.20 = 0.05 1.34 £ 0.05 1.63 £+ 0.05 238 12 X 0,10 5.92 = 0.10 6.53 + 0.10 The reducing power of the fuel salt was measured by dissolving the salt in a boric and hydrochloric acid mixture, which reacted with the re- duced constituents to yield hydrogen. The amount of hydrogen, 2 to 9 cms, gave the reducing power, after a correction of about 0.1 cms, as deter- mined by an immediately subsequent blank run on the same solution for radiolytic hydrogen. The weighed sample of fuel, about 1/2 g, was placed with an equal amount of boric acid in the dissolving flask. The flask was swept with carbon dioxide to displace all air and then partially evacuated so that the required 35 ml of 6 M HC1l could be added. When the sample had been dissolved with the aid of a magnetic mixer, the gas was swept through a KOH solution with a stream of COz. The CO- was absorbed, leaving the Ho, which was transferred to a thick-walled centrifuge cone that served as a gas-collection test tube. The test tube, which could be decontaminated and removed from the hot-cell walls, was closed with a rubber diaphragm that permitted hermetic transfers by means of hypodermic needles. Initially, air had been displaced from the test tube by water, and a measured 1 ml of helium was introduced through a hypodermic needle while displacing an equal amount of water through a second hypodermic needle. The measured amount of helium served as a standard against which the hy- drogen was assayed by ratios from mass spectrometry. Observations of Sectioned Capsules Capsule 24 was sawed through longitudinally, with the graphite core included, as soon as the presence of fluorine was recognized. The gener- ally good, unaltered condition of the contents, reported previously,® was confirmed by the complete dissection of all the capsules, except capsule 36, which was not available because it had been used in the melt-out at- tempt. The tops of the capsules were removed with a tubing cutter at the 92 vapor phase region to expose the frozen liquid level for inspection. Lit- le or no evidence of an accumulated scum was observable, but the visibil- ity conditions were unsatisfactory because of the blackened nature of the salt surface. A tubing cutter that was operable in an inert atmosphere had been developed, but it was used only on capsule 36. Removal of the capsule tops involved the extraction of the appended thermocouple wells in the case of the large capsules. Generally, this was accomplished easily, although some abrading and fracturing of the upper surface of the salt occurred in the vicinity of the shoulder of the well. As a result, fairly large areas of the upper surface of the graphite core were exposed. At some locations corresponding to the liquid level rise and fall during freeze-thaw cycles, a black powder had accumulated on the metal wall. Frequently, both white and discolored droplets of condensed salt were lodged on the powdery coating. The coating, although it obscured the metal surface, was present in too small an amount for satisfactory sampling. In general the conclusion was that the interface was relative- ly clean. There were, however, regions, as shown in Fig. 5.6, in which HCO-1871 Fig. 5.6. Fuel-Gas Interface of Capsule 45 Showing Nonwetting of Both Graphite and Metal. 93 there was a nonwetting miniscus of the salt with respect to the metal. Since salt wets clean metal, the nonwetting was attributed to the uniden- tified coating. Capsules 12 and 45 were opened without having been potted in Araldite resin, which was normally used to ensure that the contents remained in po- sition. 1In these cases, diametral cuts were made in the cylindrical metal walls, but the salt was penetrated only slightly, and the metal halves were pried apart. In capsule 45 the salt adhered to the metal in two halves, one of which retained the graphite core. Capsule 12 received less delicate manipulation, and the upper portion of the salt and the graphite fell out; the bottom portions of the halves remained intact. The exposed cross sec- tions were photographed and examined for surface layers, but none were found. The cause of the low activity noted at the bottoms of the capsules in the autoradiographs was not evident. The graphite cores that were re- covered intact were stored. Nothing unusual was seen, except for an indication of flattened bub- bles at the bottom, when the small capsules (4 and 6) were sawed longi- tudinally down the middle. The salt was sampled and a portion of capsule L, characterized by the highest power density, was prepared for metallo- graphic examination. The potting resin partially penetrated a thin annu- lus resulting from contraction of the salt ingot away from the graphite, and the salt thus cemented was difficult to sample. The results are not yet available. Control Experiments on the Fluorination of Graphite In preparation for a proposed hot-cell experiment in which graphite cores from the irradiated capsules would be baked under vacuum and the evolved gases analyzed, a similar experiment was performed on unexposed cores. Three control cores were fluorinated at room temperature under a fluorine pressure of 8 to 5 psig for 95 hr and, after sweeping away most of the excess Fp with helium, heated to 1200°C under vacuum. The evolved gases were continuously collected by a Toepler pump as they were released during the period of increasing temperature. No CF4 was found; the pres- ence of 3% SiF4 indicated that traces of free fluorine had bheen liberated in the quartz bakeout apparatus, while the bulk of the total of 3 to 6 em® of gas collected was Hp, CO plus Np, and COs-., As far as could be con- cluded from the unirradiated cores, a negligible amount of chemisorbed fluorine was retained by the graphite; no sudden release of gases was en- countered during heating. Both exposure of the control graphite to air and the effects of irradiation caused differences in conditions between the test trial and the proposed hot-cell experiment, but no contraindica- tions were found. Conclusions Drawn from the ORNL-MIR-L47-4 Experiment Continued examinations of the capsules from the LT-U4 experiment have confirmed and strengthened the evidence that Fo is released in large amounts by radiation damage to solid crystals at room temperature and that most, if not all, the CF4 that has been found arose as a reaction 9% product of the Fs or free fluorine with graphite. No evidence of chemi- cal incompatibility at high temperatures has been found. The irregular pattern of fluorine generation in response to slightly changed conditions in otherwise similar capsules suggests that one or more of the important factors has not yet been recognized or measured; one factor under suspi- cion is the freezing history. In phase-behavior studies, both the crys- tallization path and crystallite sizes have been recognized as virtually nonreproducible in supposedly duplicate tests, except in the case of ex- tremely slow cooling in which equilibrium behavior is closely approached. Future experiments are planned in which cooling rates can be varied. Molten-Salt Irradiation Experiment ORNL-MIR-UT7-5 Capsule Irradiation and Sampling The fifth capsule irradiation experiment was completed on January 23, 1963, after approximately 4 1/2 months of operation in the MI'R. This cap- sule, described previousl.y',6 was designed for studying the formation of gaseous products both during and following irradiation of molten salt fuel in contact with graphite and INOR-8 alloy. Four of the six capsules in the assembly were sealed, as in earlier irradiations. The previous set (ORNL-MTR-4T-4) of irradiation capsules was found to contain both CF4 and Fo, and it was necessary to know whether these gases were released to the capsule atmosphere during reactor operation or only while the fuel was at low temperature following reactor irradiation. Two capsules were there- fore equipped with purge gas lines for continuous monitoring of pressure and for sampling. At the time the capsules were designed, only CF4 had been observed in irradiated capsules, and consequently some of the materi- als of construction Were nol entirely suitable £or expusure to Fo. The capsules and the gas-purging system appear, however, to have functioned satisfactorily, indicating that the F, release probably was small during high-temperature operation. The power density in one of the purged capsules was twice that in the other. The sealed capsules were designed to include different ratios of - graphite and metal surfaces exposed to the fused salt fuel. By using var- ious ratios, a range of oxidation-reduction levels was obtained to bracket the conditions expected during the life of a reactor. Important proper- ties of the capsules are listed in Table 5.L. ' Equipment at the MIR was altered to include provisions for removing gas samples and to measure the pressure in the capsules both during opera- tion and in reactor shutdown periods. This equipment was calibrated pre- cisely to ensure that purge samples included a large and known fraction of the gas collected. A schematic diagram of the new equipment is shown in Fig. 5.7. o ®0ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962," USAEC Report ORNL-3369, p. 102. . Teble 5.4. Description of ORNL-MIR-47-5 Capsules a Fuel-Wetted Central Core Surface Ares, Capsule Weight-of Uranium Composition (cm?) Position Fuel Fue; Remarks Numbter (g) (wt %) . Material Qu?gglty INOR-8 Graphite 3 25,240 2.08 CGB graphite 5,587 35 12.5} Capsules provided with lines 4 24 .862 4.08 CGB graphite 5.696 35 12.5 for purging and gas sampling 2 25.188 4,08 Graphite wafer 0.50 46 1.0 Capsules sealed; variation 1 18.928 4.08 CGB graphite = 3.074 7 27 in graphite-to-metal wetted- ‘ surface-area ratios Front 0.5037P 4.08 AGOT graphite 3.6353 Capsules sealed; graphite Rear 0.6465b 2.08 AGOT graphite 3.6655 impregnated with salt fuel; no free salt #Tyradistion temperature and power density varied with operating conditions. Briel was 93.25%-erriched uranium. G6 96 UNCL ASSIFIED ORNL-LR-DWG 77838 £ ) - { ~—p-4—Pp<4 XXX XXX X DRIER <] < HELIUM —_— — X XXX XX X — —><-><><><><><—1 TO MTR -—-> X XX XX é | PR ) | COUPLING —-—— MOLTEN SALT SUBMERGED GRAPHITE | SAMPLE BOTTLES CARBON TRAPS xxxxxxx CAPILLARY TUBING PR PRESSURE RECORDER Fig. 5.7. Gas-Collection System for ORNL-MTR-47-5 Experiment. The aperating conditions of the experiment were varied throughout the test to include as nearly as possible those anticipated in dirferent regions of the MSRE. Temperalure and power werce voried independently bhut were held constant for the periods required to collect gas samples. Fuel temperatures during reactor operation ranged from a few hundred degrees to 1500°F, with power densities in the fuel varying from 3 to 80 w/cm . The gas samples were taken by isolating the capsule for a period of time and then purging the accumulated gases into an evacuated sample bomb. The gas samples may be classified by five types related to the condltlons maintained during the accumulation period: 1l. reactor at full power with the fuel molten, 2. reactor at an intermediate power with the fuel frozen, 3. reactor shut down with the fuel cold, 4, periods spanning fuel meltout during reactor startup, 5. periods spanning fuel freezing following reactor shutdown. 97 Most of the samples were of the first type, as may be seen in Table 5.5. These samples were taken with accumulation periods of from one to ten days at fuel temperatures from 900 to 1500°F and at power densities of from 15 to 80 w/cm® in the fuel. The samples taken with the reactor at an intermediate power and the salt frozen provided information for evaluating the effect of intense ion- izing radiation on the evolution of CF4 and Fp from the fuel with the salt frozen but at various temperatures. Gas accumilation times were 1 to 3.5 hr with the fuel temperatures varied from 180 to L80°F. The power densi- t}esawithin the fuel are estimated to have been in the range of 3 to 10 w/cm”, Gas samples taken during reactor shutdowns were for determining the composition of the gases evolved when the capsules were cold. The over- all release rate was measured as a function of time by monitoring the pressure during sample collection. The other two types of samples which were taken during the freezing or thawing of the fuel were for evaluating the effect of a phase transition on the release of gases, The pressure rise in the capsules that occurred during reactor shut- down appeared to be a function of the irradiation history immediately be- fore the shutdown. In general, a delay period that varied from hours to days occurred before the first indication of pressure rise. The longer delays and lower rates of pressure rise were associated with a previous history of low-power operation. The initial rates of pressure rise indi- cated gas evolution rates of 0.l1l1 to 0.51 cma/hr per capsule. The rate of pressure buildup decreased slowly with time, and in no case was a shut- down long enough for the pressure to reach equilibrium. The shutdown gas semples contained considerably more radioactivity than those taken with the reactor at power and the salt molten. Gamma-ra{ spectrometer analyses of these samples showed the presence of Tel®2 and I'®2, whereas the other samples showed only noble gases, During the accumulation periods with the reactor at low power and the salt frozen, there was usually no indication of pressure rise; however, the accumulation times were only 1 to 3 hr, and the data are not consider- ed to be definitive. Upon reactor shutdown, the capsule lines were purged with helium for 21 hr to remove gaseous products. Residual activity, probably deposited in the 1/8-in.-diam tubing, produced activity readings of 2 to 4 r/hr at 3 in. Tollowing removal from the beam hole, the experi- mental assembly was sectioned, and valves were installed on the severed ends of the purge lines in the MIR hot cell., The experimental assembly was then returned to Oak Ridge for further observation of gas evolution from the fuel and for postirradiation examination, as described in the following sections. Analyses of Gas Samples Taken During Irradiation Of the six capsules irradiated in experimental assembly ORNL-MIR-L7- 5, two were like the capsules in assembly L4T7-4, except for the presence 98 Table 5.5. Gas Sampling Conditions of Experiment ORNL-MIR-LT-5 Duration of Capsule Temperature at Sample Gas Accumu- Time of Sampling (°F) Number Jdation Remarks (hr) Capsule U4 Capsule 3 1 29 1050 900 2 0 1050 900 Residual of sample 1 3 50 1248 1100 4 24 1250 1090 5 2 © 1380 1200 6 96 1400 1230 T 16.3 96 90 Sample taken after 4.3 hr at _ 40 Mw and a 12-hr shutdown 8 73 1205 1050 . 9 0.75 150 150 Sample taken during salt freezing following shut- ' down 10 56.75 105 (a) 11 40.5 85 Pressure rise measured during shutdown 11 97.25 85 Pressure rise measured during shutdown 12 2.75. 85—-1061 85-998 Sample taken during salt melting upon startup 13 . 23 1247 - 1148 . ‘ ' 14 1.8 T70—-1060 70983 Sample taken during salt : melting upon startup 15 31.78 1492 1200 16 0.50 1228-320 1115—-320 Sample taken during salt freezing upon shutdown 17 19.5 122 122 Pressure rise measured : during shutdown 18 2.33 1200 (a) 19. 18 1212 (a) 20 2L .67 1470 1230 21 73 1510 1250 ) 300 285 Sample taken with reactor ‘ operating at 10 Mw 23 380 350 Sample taken with reactor operating at 15 Mw 24 T2 1185 1070 25 70.67 -1308 1132 26 111 1490 1260 : 27 29.5 - 81 80 Pressure rise measured _ during shutdown 28 3.5 199 180 Sample taken with reactor operating at 5 Mw 29 3.5 480 410 Sample taken with reactor ‘ operating at 20 Mw 30 .75 1480 1270 31 80 1505 1320 32 9.25 110 110 Pressure rise measured during shutdown a Not sampled. 99 of inlet and exit lines for gas sampling.7 Valves on these lines, outside the shielding, isolated volumes of a little over 100 cm® in which the evolved gases were allowed to accumulate in a 12-psia helium cover gas. Sampling was accomplished by sweeping the gas with helium into 250-cm3 evacuated nickel collection bottles that were replaceable in groups of six for each of the two capsules (see Fig. 5.7). The sampled capsules contained 25 g of fuel from which thorium had been omitted and which had the composition shown in Table 5.6. The flux did not permit matching both the power density and the operating temper- ature of the MSRE in the same capsule; accordingly a twofold difference in uranium concentration was used in the two capsules. The purified fuel was originally loaded in the form of molded ingots, rather than by liquid transfer, and after the graphite cores were cleaned with a molten-salt flush, they were handled only in a dry helium atmosphere. The graphite was selected from a preliminary delivery of graphite bars for the Engi- neering Test Loop (ETL). Cores without visible fissures or striations were obtained, but there was much spoilage during their selection. Table 5.6. Fuel Compositions Irradiated in Purged Capsules 3 and 4 in Experimental Assembly ORNL-MIR-L4T-L Specific gravity at 1200°F: 2.13 Liquidus temperature: 8L42°F ' Thermal conductivity at 1200°F: 3.21 Btu/hr.ft2(°F/ft) Specific heat at 1200°F: 0.455 Btu/1b.°F Composition Fuel In Capsule b4 In Capsule 3 Mixture mole % wt % cation mole % wt % cation LiF . 67.36 12.4 67.19 12.3 BeFo 27.73 6.56 27.96 6.58 ZrF 4 4,26 10.2 4,51 10.8 UF4 0.66 4,07 0.34 2,08 The gas samples, as received at ORNL, were, in general, too radioac- tive for examination in a mass spectrometer, but a transfer of a portion 70ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962," USAEC Report ORNL-3369, p. 102. 100 of the sample gave a large decrease in the activity, most of which tended to remain on the container walls. Routinely, the samples were first scanned for F- in a laboratory at the Oak Ridge Gaseous Diffusion Plant and then analyzed there if ¥y were found. Otherwise, and in most of the cases, the remainder of the transferred sample was .analyzed in an ORNL laboratory at the Y-12 Plant. In assessing the requisite sensitivity of the gas measurements, it was necessary to consider how much Fp or CF4 evolution might be tolerable from the MSRE fuel. As a rough approximation, a minimum inventory, as with a fuel containing 0.15 mole % UF4, is 200 g-.moles of UF4. If con- version of 10% of the UF4 to UFs during operation for one year at 10 Mw (thermal) represented an allowable limit, a yield of Fp of 10 moles or 61 ml/Mwd could be tolerated with this much UF4. A factor of one-half is required to express the above amounls in termo of CFy. For 5000 liters per day of off-gas flow from the MSRE at the full power of 10 Mw, the equivalent concentration of Fp 1ln the off-gas to correspond to . the maxi- mum allowable rate of reduction of uranium would be 120 ppm. The hotter of the two sampled capsules, that is, the one with the higher uranium concentration, was designed to produce 900 watts. If it is assumed that the yield of F, in this capsule was proportional to the power density, the MSRE tolerance limit corresponded to a daily concen- tration increase of 550 ppm. The CF4 yield was expected to depend on the graphite area, as well as on the free fluorine concentration, which, in turn, was assumed to be proportional to the power density. With only these simplifying assumptions, the tolerable daily increase in CFs con- tent in the 100 ml of capsule cover gas was 110 ppm. Although the capsule gas-collecting system, containing as it did stainless steel in some regions at intermediate temperatures, was not suitable for preserving small amounts 6t Fo when the Cuel was molten, the theoretical limit of sensitivity to CF4 should have been of the order of 1 ppm per day of accumulation time under operating conditions. This estimate was derived mainly from the considerations that 5 ppm was measurable by mass spectrometry and that 5=-day accumulation periods were feasible. In practice, the concentrations of CFs4 in the capsule gas were generally undetectable, and in the 32 samples so far analyzed, the CF4 content never exceeded 12 ppm. These amounts, which are too low to be of absolute significance, were low in comparison with the estimated tol- erance limit by a factor of 100 or more. Xenon and krypton were found in the expected quantities. Two points should be noted in regard to the sampling for Fo. First, both previous tests at K«25 and recent tests carried out there in support of the current irradiation studies showed that F> scarcely attacks pre- treated stainless steel at temperatures below 200°C, but at temperatures above 200°C the rate increases rapidly. Even when F5 is contained in prefluorinated nickel, a loss 1s likely to be noted in a period of weeks at room temperature. Thus the capsule sampling apparatus was expected to give evidence of F> if present at low temperatures, although the 101 amounts might be in error in the low direction. Second, any Fo potenti- ally produced in the molten fuel was expected to react with the capsule contents, fuel included, before it reached the stainless steel. The absence of CF4 in the gas collected at operating temperature could be considered as presumptive evidence that Fy was not being produced in sig- nificant quantities. The peculiarly low rate of attack of metals by CF4 was noted in previous MSRE studies® and is implied in other work in this report (see Chapter 6). The life-time of the experiment was originally thought to be deter- mined by the rate of plugging by salt vapor deposited in the gas lines. Accordingly, high temperatures were generally avoided during the earlier MIR cycles. More freedom in choice of conditions was exercised in later cycles, however, and at termination the gas line from the hotter capsule was partially plugged. Sample accumulation periods of 3 days were orig- inally considered as feasible without interruptions; subsequently periods of 10 and 17 days were achieved with only minor interruptions. Samples taken following pressure rises (described in the following section) were expected to contain Fp, but this was at first difficult to prove, for two reasons. The collection apparatus, even after prefluori- nated nickel collection bottles were available, had not been completely pretreated with fluorine and thus consumed some of the evolved fluorine. Second, samples of the gas taken after the pressure increases raised the activity level in the sampling station to an almost intolerable extent. Accordingly there was a reluctance to collect fluorine samples because they jeopardized the exchange of sample bottles required for obtaining gas during operation, which was the main objective. After confirming a few failures to detect Fo in samples following a pressure rise, a similar sample was obtained in which the Fy content was 5.8%; this was little more than half the expected amount, but it was a definite indication that the gas was fluorine. Little CF4 was noted in samples collected during shutdown, presumably because the Fs did not have sufficient access to graphite over long periods of time. Fluorine that was trapped in the graphite should have reacted on heating, Pressure Increase During Reactor Shutdown Additional light was thrown on the source of the large amounts of Fo encountered in the 47-4 capsules when pressure rises were observed during shutdown periods in the two of the 4T-5 capsules that were equipped for pressure measurement and sample collection. An attempt was made to cor- relate the rate of pressure rise, attributed to radiolytic F, generation, and the rate of absorption of decay energy of the fission products in the fuel. Calorimetric data on fission-product decay power® were used, since S0ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Feb. 28, 1962," USAEC Report ORNL-3282, p. TT. 7S. Untermeyer and J. T. Weills, "Heat Generation in Irradiated Uranium," USAEC Report ANL-4790, Argonne National Laboratory, Feb. 25, 1952. 102 energy 8bsorbed rather than total energy released was effective in radi- olysis. The decay power curve was integrated graphically to give a plot of total energy absorbed as a function of cooling time. The multiplica- tion of the energy ordinate of this curve by a G value in appropriate units yielded a theoretical Fs generation curve for this particular value. The G value is defined in conventional units as the molecules of product per 100 ev of energy absorbed. In practice, a family of energy curves was prepared by multiplying the energy ordinates of the original decay- energy curve by appropriate constants, and these curves were compared with the pressure rise data plotted on the same time scale. The G value was then calculated from the multiplying constant that gave the theoreti- cal curve most closely fitting the pressure-rise data. The best fits for the pressure-rise data obtained during the shut- down following the first 3-week cycle in the MIR are shown in Fige. 5.8 and 5.9. The "induction period" during which there was no pressure rise required a translation of the origing downward and Lu Lhe right for the UNCLASSIFIED ORNL-LR-DWG 77460 30 — 0 CAPSULE NO. 4 PRESSURE 14 0 CAPSULE NO. 3 PRESSURE | — 14 2.5 3 — 12 L, 20 — ¥ ) o @ 2 w 1_|,l_1o - - - % 1.5 — 7 M W x| o =z & 8 10— % 3 & T Y P w— 6 o 0.5 |— o o o \Vi 1,8 0O t—-= u 0 — 2 - — 0 0 10 20 30 40 50 60 70 80 90 100 | I I I I I I I I J o 0 10 20 30 40 50 60 70 80 90 100 ' COOLING TIME (hr) Fig. 5.8. First Correlation of Pressure Rise with Energy Release. On the energy scale, one scale unit is 31,500 joules or about 2 X 1017 Mev. 103 UNCLASSIFIED ORNL-LR-DWG 77464 16 30 © CAPSULE NO. 4 PRESSURE o — 14 0 CAPSULE NO. 3 PRESSURE 2.5 — —{ 12 ,’u:’ o < 2.0 . 7 S — 10 2 E (§¥] .| L ) < 9 5 g (el a >__" “ 90 g — s (e —16 T ) 6 G S 0.5 — o DG — a4 —Gg f P oi\o—ogl—o-—oj—o— | I l ' l 45 10 20 30 40 50 60 70 80 90 — 2 ] > & b I | I I | | I I 0 B 10 20 30 40 50 60 70 80 90 | | I I I | I I | | 0 0 10 20 30 40 50 60 70 80 90 100 COOLING TIME (hr) Fig. 5.9. Second Correlation of Pressure Rise with Energy Release. On the energy scale, one scale unit is 31,500 joules or about 2 x 1017 Mev. best fit (Fig. 5.8), although a fair fit was also obtained with the iden- tical theoretical energy curves by shifting only the origins downward (Fig. 5.9). The energy scale was such that one scale unit was 31,500 Joules or about 2 x 1017 Mev. The G value for Fo generation during the first 47-5 experiment shutdown was calculated from these curves to be 0.02, as compared with a G value of 0.035 estimated for one of the cap- sules in experiment 4T-k. The close correspondence of the pressure data in Fig. 5.8 to the ex- pected shape of the selected energy curves may be largely artificial. The ratios of the energy ordinates remained constant at 0.57, however, which was as close as could be expected to the ratio of the power levels in the two capsules, and the existence of induction periods required a shift, at least in the ordinate direction, for any matching of fluorine production to decay energy. Translations in the ordinate direction ac- commodated both the energy absorbed in causing crystal damage, which pre- sumably must first occur to a critical extent before F, appears, and also the fluorine which was reconsumed or permanently held for any cause. The 104 translation along the abscissa implied a time delay such as might occur in a diffusion-controlled process in which the average time of translation of the diffusion species remains constant, as it would in a steady-state process. The diffusing species could have been fluorine atoms, but there seemed to be little precedent in solid-state theory to support such a possibility. Although the hypothesis seemed plausible that the release of Fp from the solid fuel, which was mainly Li BeF4, was diffusion con- trolled, further speculations were postponed until the results from other shutdown periods were considered. The data on the subsequent pressure rises were, in part, fragmentary, and they have not yet been analyzed in detail. There were more similarities than dissimilarities, however, when compared with the general pattern expected if decay energy were a control- ling factor. The occurrence of induction periods prior to the pressure rise was confirmed, and in the case of the lower power capsule the induction period extended throughout the shutdovn periods for the second, third, and tit'th reactor shutdowns. No simple relationship was found between the duration of an induction period and the energy released during that time, as might have been expected if the critical amount of damage prior to fluorine re- lease had remained constant. The literature'® describes a 22% decrease in the density of LiF as a result of a dose of 1 x 10'® to 2 x 10'® neutrons/cm®. It was inter- esting that in the irradiation described by Spaepeni© there was also an initial period of low G value, corresponding to an induction period. Nine-tenths of the density decrease took place between dose levels of 6 x 1017 and 1 x 10*® neutrons/cm®, where the estimated G value was 0.10, compared with 0.037 for the over-all exposure. The pressure rises that were encountered in the sampled capsules were sufficient to represent a pronounced alteration of the fuel beyond limits of reduction that were deemed safe in the MSRE. This meant that results from later cycles were possibly not applicable to the MSRE fuel, but the possibility that marked differences in the evolution of gases would occur was considered to be remote. Among the liberties taken with this altered fuel in experiment 47-5 were low-power operations, as de- scribed in the following section, in which fissioning was caused to occur in solid fuel at relatively low temperatures. Effects of Fissioning in Frozen Fuel The design of the unit for inserting the L47-5 experimental assembly into the reactor allowed for considerable variations in flux by adjusting the depth of insertion and for limited temperature control by regulating a gas gap that controlled the rate of heat loss. No independent vari- ations between capsules were possible, and the available range was insuf- ficient to reduce the capsule temperature below the freezing point of the 107, Spaepen, Density of Neutron Irradiated Lithium Fluoride Crystals, Phys. Rev. Letters, 1:281-282 (1958). 105 fuel at full MIR power (4O Mw). Therefore, periods of low-power operation were arranged during MIR startups, and the capsules were maintained at full retraction and minimum gas gap to observe the behavior at different levels of fissioning in the so0lid fuel. Four-hour periods were involved, and temperatures were steady during the last 2 or 3 hr of the period. The pressure was monitored throughout, and samples were drawn during the last 1/2 hr. In view of the copious evolution of Fp during shutdowns, the results, presented in Table 5.7, were somewhat surprising. They imply either a pronounced difference with temperature, possibly as an annealing effect that seals the crystals or as an accelerated rate of recombination, or a greatly decreased G value for Fo evolution from fission fragments. Table 5.7. Fissioning in Solid Fuel at Low Temperatures MIR Estimated Capsule Capsule Na Bath® Pressure Capsule Power Power Density Temperature Temperature Change’ Number (Mw) (w/em®) (°F) (°F) (psi) L 20 10 610 378 0 3 20 6 510 378 0 L 10 5 312 200 0 3 10 3 285 200 0 4 15 7.5 380" 260 o} 3 15 k.5 , 350 260 0 I 5 2.5 199 122 +0.43° 3 5 1.5 180 122 0 4 20 9 480 303 -0.1 3 20 5 410 303 +0.2 %The sodium bath temperature is the highest temperature occurring in the stainless steel tubing leading out of the capsules. Tests have indicated negligible attack of stainless steel by 1% F- at temperatures below LOO°F. bThe pressure readings, probably accurate to 0.1l psi, were taken during 2 to 3 hr of constant temperature operation. “The pressure rise occurred over a period of 65 min; for the subse- quent 2 hr of 5-Mw operation there was no further pressure rise. 106 The latter effect was noted in other crystals, such as chlorates, and ex- plained on the basis of microself-annealing from heat generated along the fission-fragment spike. The 0.43-psia pressure rise referred to in Table 5.7 was of interest as a possible instance of a time lag of the annealing effect, since following an initial 65-min period no pressure increase could be detected in the next 2 hr. This measurement was made immediately after a thorough helium flush at the end of a scheduled shut- down during which the evolution of F> had been proceeding as usual. Even for just the 65-min period, the G value was less than half that previous- ly encountered at 95°F, that is, at the estimated ambient temperature during shutdowns. The temperature of the sodium bath and hence of. the stainless steel gas lines was too low to have caused appreciable loss of Fo by reaction with the walls. Thus there currently appear to be no po- tential problems associated with Fp evolution in freeze valves and freeze flanges. Pressure Rise Following Termination of Exposure A monitoring manifold was successfully attached to the capsules in ORNL hot cells within 11 days after the end of exposure in the MIR. At this stage, the pressure increased steadily at a rate corresponding to about 13 and 6 ml per day of Fo (STP), respectively, in the two capsules, compared with about 4 ml per day sometimes found for the hotter capsule during shutdowns. Although the capsules and associated tubing in the monitored volume (~100 e¢m3®) had been evacuated at the beginning of the monitoring period in the hot cell, large amounts of Op and a few percent of CO- were found in the collected gas that was otherwise more than one- half fluorine. The Oz, which diminished sporadically in subsequent samples, was tentatively attributed to a fluorine-induced release follow- ing reaction of the walls, and possibly of the fuel, with air during the 2 1/2 hr that the gas lines were open in the MIR hot cells while the assembly nose was severed from the accessory equipment and isolating valves were installed on the inlet and outlet gas lines at about 4 ft from the capsules. The Jjacketed assembly that contained the capsules was provided in the hot cell with a specially constructed heater in order to study the effect of temperature. There were indications that a marked decrease in rate of gas evolution occurred between 50 and 85°C. Conclusions Drawn from Experiment ORNL-MIR-L7-5 The main conclusion drawn from the analyses of the cover gas is that CF4 was not evolved in significant amounts under operating conditions. The considerable variation in the extent to which CF4 was found in earlier experiments was only roughly explained; when the evolved fluorine was rapidly removed, as in experiment 47-5, almost no CF4 was found. The fluorine that accumulated to a pressure of 1/2 atm or more over a period of days at the end of irradiation in experiment 47-5 produced a concen- tration of about 0.2% CFs. There was no evidence of the presence of CFg4, except where Fp evolved from decay energy in frozen fuel was either found 107 or would have been expected. Since this was the case, no problem with CF4 in the MSRE is to be expected. ' An important result from experiment 4T7-5 was the confirmation of sur-. mises made in explaining the source of the Fs found in experiment 47-4. An unexpected bonus was the strong indication that frozen salt at temper- atures of 100°C or higher was not prone to evolve Fp, even with fission- ing occurring in the crystals. Additional experiments are planned that will provide a greater range of independent variation of temperature and dose, and postirradiation examinations are continuing. A point of possi- ble interest for future irradiations was the role played by the uranium cation concentration in providing a buffer in the form of a variable valence reservoir that might have helped control, in an inverse direction, the steady-state concentration of free fluorine in the fissioning melt and thus might have reduced the potential rate of production of CFa. 108 - 6. CHEMISTRY Phase Equilibrium Studies The System LiF-ZrF,-UF, Previous experience with MSRE fuel mixtures consisting of LiF, BeF,, ZrF,;, UF,, and ThF, indicated that U** ions do not substitute in Zr*t sites in the LiF-ZrF, compounds 3LiF:ZrF,, 2LiF-ZrF,;, and 3LiF-4ZrF,; however, the fact that 0ak Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962," USAEC Report ORNL-3369, p. 106. 60ak Ridge National Laboratory, "Reactor Chem. Div. Ann. Rep. Jan. 31, 1963," USAEC Report ORNL-3417. 111 (1) a mass action effect will cause Zr0O, to precipitate 1n1t1ally, and (2) zr4t may complex OX1de ion and thus produce an appreciable "oxide capacity" in the melt.® There was, however, need for more quantitative information on the solubilities of U0, ZrO,, and BeO, as well as on their mass action or equilibrium behavior in mixtures of flush salt and fuel salt. Accordingly, oxide precipitation studies, similar to those reported previously,7 were continued. In these experiments, melts of LiyBeF,, ZrF,, and fuel salt of known weight and composition, usually 2 to 3 kg, were purified with HF and Hy. Weighed increments of BeO were added as a fine powder that had been calcined at 800°C. Stirring was accomplished by sparging continu- ously with helium. After each oxide addition, samples of the melt, usually about 25 g, were withdrawn into hydrogen-fired copper-tubing filter sticks. In most cases, the weights of the samples taken were determined. Equilibration periods of one day proved adequate for an apparently complete reaction of the added oxide. These studies continue to be hampered by inconsistent results from attempts to obtain direct oxide analyses of the melts in the concentra- tion range of interest, that is, less than 1000 ppm. In the present ' experiments, in which known amounts of BeO were added to known weights’ of melt, caréful material balances between the oxide added and the oxide presumably removed by precipitation and by sampling were obtained as a means of estimating the oxide remaining in solution. In general, it was found that these calculated values were more self-consistent than . the oxide analyses. The MSRE coolant, which has also been prominently considered as a possible flush salt, has almost the same composition as the compound - Li,BeF,. A log-log plot is presented in Fig. 6.1 of the equilibrium U4t and Zr*t concentrations versus the calculated oxide-ion concentra- tion for a precipitation experiment in which BeO was added to an LizBeF, melt containing 5 wt ¢ fuel salt (LiF-BeF,-ZrF,-ThF,-UF,;, 70-23-5-1-1 mole %) at 600°C. The straight lines of slope -2 drawn through the data points correspond to simple mass-action behavior for the reactions: U0, (s) = U*t + 207" Zr0op (s) = zr** + 2077 70ak Ridge National Laboratory, "MSRP Prog. Rep. Feb. 20, 1961," USAEC Report ORNL-3122, p. 120, 80ak Ridge National Laboratory, "MSRP Prog. Rep. March 1 to Aug. 31, 1961," USAKC Report ORNL-3215, p. 124. 112 UNCLASSIFIED 0.1 ORNL—-LR-DWG 78007 \ 0.08 \ oos‘ \~ ‘\t =o 4+ NI\ 0.04 \. \ [ ) 4 ° > N 2 002 : E o \ - . & X \C = . W 0.0! o)\ Q _ I I OV W N - 8 0.008 ‘\ \ g \ @ \\ —.0.006 \ N = \ < a+ &\ ® = . . U \ 0.004 ‘ I N 0.002 0.001 — : 0.04 0.02 0.05 0.4 OXIDE ION CONCENTRATION (moles/kq) __ Fig. 6.1. Apparent Relation Between U** and Zr** Concentrations and 0"~ Concentration in an Li;BeF, Melt Containing 5 wt % Fuel Salt to Which BeO Was Added at 600°C. where the solubility products are Q;O- = [U4+][0“]2 = 1.2 x 107 moles:’/kg3 2 Q;rOQ _ [Zr4+][0-"]2 = 3.2 x 1077 moles?/kg’ If the scatter of the data were not obscuring a sllght curve with slopes more negative than -2 at higher cation concentratlons, the agree- ment with simple solubility behavior would imply that complexing of M* with 0~ in this medium is not an important effect in the range studied. 113 The solubility products indicate the following solubilities in LipoBeF, initially free of Zr**t, U**, and 0~ at 600°C: UO,, 0.014 moles/kg; and ZrOp, 0.020 moles/g. Observed precipitation behavior generally agreed well with that expected for simple mass action relations, as shown by the solid curves in Figs. 6.2 through 6.5. The addition of BeO to an LipBeF,; melt that contained 5 mole % ZrF, resulted in nearly quantitative precipitation of Zr0O, (Fig. 6.2). For the highest Zr** concentration level studied (~1.3 moles/kg), which corresponded to that proposed for MSRE fuel, there was no indication of an "oxide capacity" before precipitation began. The data of Fig. 6.2, as well as the calculated curve, were essentially consistent with quan- titative precipitation, except at the lowest zirconium concentrations (<0.1 moles/kg). The initial precipitation of ZrO,; from an Li,BeF, melt containing 5 wt ¢ fuel salt is shown in Fig. 6.3. While there was some indication from the 7r4t chemical analyses that precipitation did not occur until 0.04 moles/kg of BeO had been added, the scatter of the data from sub- sequent analyses (i2%) suggested that this apparent oxide capacity was not necessarily real. The dashed line represents gquantitative precipi- tation. A solubility product of 3.2 x 10-° for sr(, implies that ini- tial precipitation should have occurred when the oxide ion concentra- tion reached 0.005 moles/kg (80 ppm). Thus, even at Zr*" concentrations greater than 1 mole/kg (~5 wt % ZrF,), there was no conclusive evidence of zr4¥.o-- complexing. Also, there was no evidence of Ut precipitation (lower portion of Fig. 6.3), and no Thé+ precipitation was encountered in any of these experiments. The complete results of the experiment with the Li,BeF, melt containing 5 wt ¢ fuel salt (Fig. 6.4), from which the correlation in Fig. 6.1 was obtained, showed that 0.023 moles of oxide per kg of melt (370 ppm) could be added before Zr0O, precipitation began. Precipitation of UQ, did not begin until 0.092 moles of oxide per kg of melt (1500 ppm) had been added, that is, not until the protective screen of zr4t had been exhausted. In the region of ZrO, and UQ, saturation, the ratio of zr*t to U*' was constant at approximately 2.7. The horizontal por- tions of the three calculated curves of Figs. 6.4 and 6.5 represent the region where BeO saturation had been reached, which was estimated to occur at an O™~ concentration of 0.07 moles per kg of melt (1100 ppm). The results of a similar precipitation experiment in which 1 wt ¢ fuel salt was present in the Li,BeF, melt are presented in Fig. 6.5. Here there were two anomalies: (1) the analyses for the zr4* ana Ut ion concentrations were higher than expected, which suggested that some contamination had occurred, and (2) there was little or no oxide capa- city, although nominally about 0.05 moles of oxide per kg of melt (800 ppm) should have been necessary for ZrQ, precipitation, as indicated by the solid curves. Both these anomalies are resolvable if the contami- nation was such that in this experiment the melt initially contained approximately 0.05 moles of oxide per kg of melt. In such a case, the effect would have been that indicated by the dashed curves in Fig. 6.5. UNCLASSIFIED ORNL-LR-DWG 78009 .30 1.28 o o N S UNCLASSIFIED ORML - LR-DWG 78008 v N MELT CONCENTRATION (moles/kg) l 1.4 I . | ~ 1.2 E, \ (.20 , . S ' e \ i £ 1.0 - ] § \ . utt Z08 "\0 oo . ¢ o L ) e . S \ . 0. e o ) [ ] ° 2 06 * z N , | & N\ T g 0.4 N\, . ' o~ § 04 = \ } 0.005 o2 N \ \ | o Y N o 05 10 5 20 25 30 o | ocs 0.10 015 0.20 ADDED OXIDE (motes/kg) - i ADDED OXIDE (moles/kq) : i Fig. 6.2. Precipitation of ZrO; from an Fig. 6.3.4 Initial ZrO, Precipitation from LioBeF, Melt Containing 5.mole % ZrF; to Which BeO ~an LisBeF, Melt Containing 5 mole % ZrF, and 5 Was Added. wt % Fuel Salt 1to Which BeO Was Added at 600°C i ! 11 UNCLASSIFIED o4 ORNL-LR-DWG 78C10 008 . /4 0.06 \'\.j . A 2 ~ TN\NY ° ‘\/ - — \ . N\ \ / UNCLASSIFIED \ )«: ) o1 ORNL=~LR-DWG 78011 @ X 0.08 | X ) . | <~ \ : —_ & 002 K( \ 006 | — = 3 o' \ \. R . ’ l ‘4"/' 0-- // E \\ \.' ze | 0.04 L -~ ‘ — 1 = 04 - I 9 A A A A A \ C) > r' . g T o0t Ny > - a P NA b 3 oozl ™ & 0.008 AN N\ g = . 2 AW \ 2 e T~ O \¥ o Sy 269+ > 0.006 AN g R ™~ H O AN o e £ oo A - ™~ w ry . z ‘ — \ = \ S 0008 AN —— 0.004 g i Yo ————] \ + 0006 ] . o J Y = \ 0.004 . s b i nE R s A * > . udt 0.002 N sl TN _ ' 0,002 5 0.001 0.001 0 0.05 0.10 0.15 0.20 0.25 AN 002 004 006 008 ADDED OXIDE (moles/kg) : ADDED OXIDE (moles/kg) Fig. 6.4. Precipitation of ZrO; and U0, from Fig. 6.5. Precipitation of ZrO, and UO, an Li,BeF, Melt Containing 5 wt % Fuel Salt to from an LisBeF, Melt Containing 1 wt % Fuel Salt Waich BeO Was Added at 600°C. _ to ‘Which BeO Was Added at 600°C. GTT 116 It is apparent that oxide behavior in mixtures of flush salt and fuel salt can be predicted reasonably well from the simple solubility product behavior (Fig. 6.1). This means that a flush salt such as LisBeF, can accumulate the oxide ion that may be present in the drained reactor system up to a limit determined by the amount of fuel salt which the flush salt incorporates because of incomplete drainage. If the solu- bility product is exceeded, Zr0O, will precipitate before UO, precipitates. The addition of Zr*T to the flush salt would provide added protection against UQ, precipitation and should dissolve the small amounts of UO; that may be produced when the reactor is drained of fuel; at the same time, however, such a flush salt can tolerate less oxide contamination without precipitation of ZrO,. Whether the precipitate would remain in slurry form has not been established. These considerations suggest that the investigation of oxide behavior be extended to include experiments at other temperatures. The possible adsorption ot U*' by 4ruU, and also the possible formatilon of U(IV)-Zr (IV) oxide solid solutions could be investigated at the same time. Physical Properties of Molten Fluorides Vapor Pressures of Fluoride Mixtures " Recent vapor-pressure measurements provided information on the MSRE fuel and were also useful for calculation of thermodynamic para- meters in solutions of ions of widely disparate charge-to-radius ratios. The Rodebush-Dixon method? of vapor-pressure measurement was employed with an apparatus that was essentially a copper tubing version of the glass and metal equipment described by Beusman.t® Pressure differen- tials were observed by means of a differential-pressure cell in con- - Junction with a strip-chart recorder. Manometric pressures below 100 mm Hg were read on absolute pressure gages that were calibrated against a mercury manometer that was rcad by meane of a suitable cathetometer. Pressures above 100 mm Hg were determined directly with the manometer. The vaporization pots, welded from INOR-8 and nickel, were similar to those used by Cantor et al.+1 Temperatures were measured in a central thermocouple well by calibrated Pt vs Pt—10% Rh thermocouples whose signal was read on a K-3 potentiometer. The reliability of the appa- ratus was tested by measuring the vapor pressure of water at 14.25°C. The observed pressure was 12.07 mm Hg, in good agreement with a hand- book valuel? of 12.18 mm Hg. °W. H. Rodebush and A. L. Dixon, Phys. Rev., 26: 851 (1925). 10c. C. Beusman, "Activities in the KC1-FeCl, and LiC1l-FeCl, Systems," USAEC Report ORNL-2323, Oak Ridge National Laboratory, May 15, 1957. 1ls, cantor, et al., J. Phys. Chem., 62: 96 (1958). . 12N. A. lange, Ed., Handbook of Chemistry, 9th Ed., Handbook Publishers, Sandusky, Ohio, 1956. 117 The System LiF-BeF,-ZrF,-UF,. In order to determine whether the vapor pressure might set a limit on the amount of ZrF, in MSRE-type fuel melts, the vapor pressures of the two melts described in Table 6.1 were measured. The mixture containing about 4 mole % ZrF,; approximated the composition of MSRE fuel, and the other showed the effect of doubling the concentration of ZrF,. Extrapolation of these results to the MSRE operating temperature indicates that the second mixture has a vapor pressure about twice that of the present MSRE fuel composition; this is probably more a consequence of increased acidity, on a fluoride donor basis, than of merely doubling the amount of a volatile constitu- ent. The data are presented graphically in Fig. 6.6, and the constants for the vapor-pressure equation are listed in Table 6.1. The vapor pressures are sufficiently low to show that deposits in the vapor space of MSRE pumps will originate from mist or entrainment rather than from distillation. Table 6.1. Results of Vapor-Pressure Measurements in the System LiF-BeF; ~-ZrF,-UF, System Composition Equation Temperaturé Heat of vapor . (mole %) Constants® Range of . . Pressure Vaporization Measurement.s (kcal) (mm Hg LiF BeF, 2rF, UF, A B (°c) at 704°CP) 66.52 29.38 3.95 0.14 9.015 10,438 1003-1178 47.8 0.02 64.67 27.01 8.13 0.16 8.562 9,661 1011-1217 b4 2 0.05 aConstants for equation log P = A - B/T, where T is in °K. bHighest design temperature of MSRE. The System LiF-BeF,. Vapor-pressure measurements in the LiF-BeF; binary system were started, and the data obtained to date are presented in Fig. 6.7 and Table 6.2. Sufficient variance exists between the values for pure LiF obtained in this work and those obtained previously13 that further measurements need to be made for this compound, as well as for other compositions in this system. Densification of LiF Powder - As usually obtained by precipitation from aqueous solution, LiF is such a light, fluffy powder that it is not conveniently handled by the equipment for preparing MSRE fuel. An attempt was therefore made to densify the material by heating it at moderate temperatures (500-650°C) 135, Cantor, "ANP Quar. Prog. Rep. March 31, 1957," USAEC Report ORNL-2274, Oak Ridge National Laboratory (SECRET), p. 128. 118 UNCLASSIFIED ORNL-LR-DWG 78440 200 i I COMPOSITION (mole %) \ ' LiF Bef, A UF, J 00 AN ® 6652 2938 395 014 - AN A 6467 2701 813 016 80 \\ .Z‘:A ~ 60 @>;:: N D e S VAPOR PRESSURE {mm Hg) n (@) \; \ g ; .: N ,® & [ ) \\ ) : . 65 67 69 74 73 75 7 79 10,000/ 010, Fig. 6.6. Vapor Pressures of Two Mixtures in the LiF-BeFy -ZxF, -UF,, System. Table 6.2. Results of Vapor-Pressure Measurements in the System LiF-BeF, System Composition Equation Temperature a Heat of (mole %) Constan’gs Range of Vaporization , : Measurements (kcal) LiF BeFs A B (°C) 100 . ' 7.509 9,615 1002.5-1288 44.0 20 10 7.428 9,105 10071305 41.7 80 20 8.718 10, 787 1027-1317 49 .4 %Constants for equation log P = A - B/T, where T is in °K. well below the melting point.l4 In this preliminary experiment, a treat- ment at 650°C for 1 hr increased the bulk density of a sample from 0.85 to 2.00 g/cm?. ' 148, J. Sturm, "A Method for Densifying Lithium Fluoride Powder " November 20, 1962, unpublished internal communication. 119 UNCLASSIFIED ORNL-LR-OWG 78444 100 AN 80 _\\ 60 ~ 40 |-\ AN N (o] . ANIANEE N e \\\\ * >\\\ w \ %40 \ AN \ o NN g 8 ‘.\ \\ k § x \\\ \\‘\ a [ ) \ \.- - 6 . \‘T_. \ NI A COMPOSITION (mole %) \ N \ \.1 o o AN 22— o 100 (THIS STUDY) '\ ——= 100 (CANTOR, REF. 43) \ \\ A 90 10 N\ . 80 20 T | e [ 4 - L 6.2 64 66 68 70 72 74 76 10,000/7. (oK) Fig. 6.7. Vapor Pressures in the LiF-BeF, System. Density of CrFo Chromium fluorides are formed as corrosion products in the MSRE fuel and play an important-role in chemical studies of MSRE problems. 1302k Ridge National Laboratory, "MSRP Quar. Prog. Rep. Jan. 31, 1958," USAEC Report ORNL-2472, pp. 104-111. 16R.‘B. Evans, III, J. II. DeVan, and G. M. Watson, "Self-Diffusi of Chromium in Nickel-base Alloys," USAEC Report ORNL-2982, Oak Ridge National Laboratory, Jan. 20, 1961. 15,16 on 120 The system CrF,-Crfi; was described recently,t”? and this description, along with some groperties of CrF, and chromium (II,III) fluoride, has been reported.l Among the properties reported was the approximate density of molten CrF, at its melting point (894°C). When allowed to solidify and cool to room temperature in a cylin- drical crucible, a sample of CrF, formed an ingot having a toroidal- shaped internal void but no noticeable distortion of the outer surface. If this void was assumed to be due only to contraction on freezing and the small change in volume on further cooling was neglected, the total volume of the ingot, including the void, represented the volume of molten CrF, at its freezing point. ©Such an ingot weighing 736 g had a total volume of 216 cm’; therefore, the liquid density is about 3.40 g/cm’. Chromium (II,III) fluoride and CrF, slowly dissolve in water to form a blue solution, which, on standing, gives off hydrogen gas, turns green, and yields a nearly amorphous, green, gelatinous precipitate. The apparent reaction is 6CrF, + (3 + x)Hp0 — Crp03 xHpO | + 3H, T + 4CrF, Activity Coefficients of ZrF, in the LiF-ZrF, System An evaluation was made of the activity coefficients of ZrF, at low concentration (20 mole % and less) in a fluoride melt similar to the MSRE fuel. These activity coefficients should be useful in esti- mating activity coefficients, in the same melts, for UF, and ThF,, which are chemically similar to ZrFs- In order to obtain the coefficients, liquidus temperatures were carefully measured in the region (0 to 20 mole ¢ ZrF4), where crystalline LiF is the separating phase. From _ these temperatures, the activity coefficients of LiF were evaluated and used to calculate the activity coefficients of ZrF, by means of the Gibbs- Duhem equation in the form NL d1ln 7 =-==—4d1ln7 Z NZ L 2 where 7y and N, are the activity coefficient and mole fraction of ZrF,, and 71 and NL are the corresponding quantities for LiF. To calculate 1n 7y from Eg. (1), 7y at N= 0.2 was also required. This particular 17B. J. Sturm, "Reactor Chem. Div. Ann. Prog. Rep. Jan. 1962," pp. 13-14, USAEC Report ORNL-3262, Oak Ridge National Laboratory. 18, J. Sturm, "Phase Equilibria in the System Chromium (II) Fluoride-Chromium (III) Fluoride," Inorg. Chem., 1l: 655 (1962). 121 77, was evaluated from published vapor-pressure measurements for the LiF-ZrF, system.1® The calculated activity coefficients of ZrF, that were the objective of the investigation are shown in Fig. 6.8 to be of quite low magnitude. These low activity coefficients are a measure of the high degree of chemical interaction or complexing between ZrF, and LiF in these melts. By implication, the activity coefficient of UF, in dilute solution in alkali halides would be between 10~% and 10-° and in MSRE type fuels, between 10-3 and 107%. 19K. A. Sense and R. W. Stone, J. Phys. Chem., 62: 1411 (1958). UNCLASSIFIED ORNL-LR-DWG 78442 0.0010 0.0008 0.0006 \ 0.0004 \ 0.0002 \ fa “ N w (@] '—- . =2 w [ \\ L 00004 5 \ tooooos N\ = \\ > 5 2 000006 000004 \\\\ -y e 0.00002 \ 0.00001 21. 18 15 2 2] . 6 3 0 ZrF, CONCENTRATION (mole %) Fig. 6.8. Activity Coefficients of ZrF, in the LiF-ZrF, System in the Low ZrF, Concentration Range. ' A 122 Graphite Investigations Evolution of Volatile Impurities from Graphite Since the unclad graphite moderator of the MSRE will comprise about 80? of the reactor core volume, the volatile impurities in the graphite O are a possible source of contamination to the molten fluoride fuel mixtures. Of the various volatile impurities, water vapor will probably represent the greatest problem with respect to reactor opera- tion. The release of sufficient water vapor into molten fluoride sys- tems of the type to be used as MSRE fuel causes the precipitation of oxides?! and gives rise to corrosion. In the MSRE the introduction of water vapor into the fuel system would probably result initially in the deposition of Zr0O,. Consequently, the removal of water vapor and other volatile impurities from the graphite moderator before introducing molten fluorides into the reactor core is necessary. As part of the MSRE startup operations (prior to salt loading), it is planned to heat the reactor core to about 650°C while circulating helium. Accordingly, the experimental. program described previously for determining the characteristics of gas evolution from graphite under conditions simulating core heatup was continued. In these experiments, . the effect of temperature and heating rates on moisture removal were of specific interest. The results provided a basis for an estimate of the moisture that may be released while preheating the MSRE reactor core. Experimental Procedure. 1In order to simulate the helium purging operation to be carried out during startup, pieces of graphite (6 to 12 in. in length) of the same cross-sectional configuration as that designed for an MSRE moderator element were placed in a container with gas entry and exit ports at each end. This assembly was mounted ver- tically in a tube furnace as part of a recirculating gas system. Dur- ing operation, portions of the gas stream emerging from the graphite block were diverted through an electrolytic water analyzer. This in- " strument was adequately sensitive to changes in water-vapor content in the purge gas over the range O to 1000 ppm. A gas-sampling system was also provided for mass spectrographic analyses of carbonaceous gases evolved during the experimental procedure. As described previously, the main body of the circulating gas stream was passed through a mag- nesium perchlorate drying column and a cold trap to reduce the moisture content of gas entering the graphite container to very low values. Pro- visions were also made for passing wet helium (which had been saturated with water vapor at room temperature) over the graphite block surfaces to provide reproducible starting conditions for repeated tests of a single piece of graphite. Temperatures were read from a Chromel-Alumel 20L,. G. Overholser and J. P. Blakely, p. 194 in Proceedings of the Fifth Conference on Carbon, Vol. I, Pergammon Press, 1962. 210uak Ridge National Laboratory, "Reactor Chem. Div. Ann. Prog. Rep. Jan. 31, 1961," USAEC Report ORNL-3127, p. 158. 123 therfiocouple inserted in a thermowell which penetrated the central region of the graphite block. A schematic diagram of the experimental apparatus is shown in Fig. 6.9, UNCLASSIFIED ORNL~-LR-DWG 77959 GRAPHITE 7/;%7/ i | 1 COoLD 7 ' TRAP / TO VACUUM //// AND HELIUM GAS-SAMPLING SYSTEM MANOMETER FLOW METER ;.;7%%/ : 7 TUBE 77/ i FURNACE Y/ DRYING COLUMN MEECO MODEL W ELECTROLYTIC WATER ANALYZER -3 Fig. 6.9. BSchematic Flow Diagram of Gas-Purging Apparatus for_Ré- moving Volatile Impurities from Graphite Blocks. .- During preheating of the reactor core, the primary fuel pump is expected to circulate helium at a flow rate of about 75 liters/sec, and a maximum heating rate of approximately 30°C/hr is anticipated. If a uniform gas velocity through the reactor core is assumed, a flow rate of about 3 liters/min-through the experimental loop approximates the gas flow conditions for a moderator element. Experimental Results. Data that illustrate the evolution of moisture from a block of AGOT graphite and from an as-received piece of ETL* graphite were reported previously but have been revised as shown in Figs. 6.10 and 6.11. The predominant characteristics noted are the removal of physically absorbed moisture at temperatures up to 150°C and the removal of what appears to be chemisorbed moisture over the temperature interval 200 to 400°C. These results suggest that the evolution of moisture at temperatures in excess of 400°C will be in- consequential in MSRE operations. The evolution of gaseous impurities other than water from AGOT graphite, as determined by mass-spectrographic analyses of gas samples taken periodically from the loop, was shown previously.22 Since there *¥ETL denotes a specific grade of graphite purchased for experiments in the MSRE Engineering Test lLoop. The specifications for this graphite are similar to those for the purchase of the MSRE moderator elements. 220gk Ridge National Laboratory, "MSRP Semiann. Prog. Rep. Aug. 31, 1962," USAEC Report ORNL-3369, p. 125. UNCLASSIFIED UNCLASSIFIED ORNL-LR-DWG 74 258R ’ ORNL-LR-DWG 74359R 160 T | 6C — | | | . | OFF SCALE I | ‘ OFF SoALE GRAPHITE WEIGHT : 1.122 k >1000pPm GRAPHITE WEISHT : 695 g >10C0pem | UMES 620 em3 140 VOLUME: 448 cm3 - 14c : . o E— GAS VOLUME 2 96 iters (STP) , 1 - GAS VOLUME : 2.88 liters {STP) ~ i 2.96 hiters {STP € I FLOW RATE: 3 liters/min £ GAS FLOW RATE : 3 hiters/min a ° a , a i TEMPERATURE RISE : 30°C/hr 2420 TEMPERATURE RISE : 30°C /hr ~ 126 — s RUN K-6 = RUN K-5 S =} -} | . w : M g | T 100 z 100 2 ; ) z = ! = 3 = 80 T go : = 7 2 - & o 3] O =2 g AR : | 8 60 N z 60 . £ \ o T\\ S | o s o - > o0~ T 40 o : & 40 : © ot o \ v '\Q w ! N = N, . S ! : ° g \( = [x i o ° e 20 fo) ) 20 ; p. ° AN . . b«- o} c o) : 0 - 0 100 200 300 420 500 600 700 0 100 200 300 400 500 600 700 TEMPERATURE {°C) ! TEMPERATURE (°C) . ‘. ! . Fig. 6.10. Removal of Moisture from AGOT Fig. 6.11. Removal of Moisture from ETL Graphite by Helium Purging. Graphite by Helium Purging. | I | | | | 7cT 125 are no provisions for an initial evacuation of gases from the MSRE reactor core prior to heatup, and since the graphite will almost cer- tainly contain some entrapped air in its pores, it is expected that some products of the reaction of air with graphite will be produced during the preheating operation. In a similar experiment with ETL graphite, the accumulation of air and carbonaceous gases in the cir- culating gas stream at temperatures up to 700°C was found to be neg- ligible. No further examinations of these aspects of gas evolution were made. In several experiments, graphite specimens were presaturated with water vapor by exposure to moist helium. The moisture removal curves were similar to those shown in Figs. 6.10 and 6.11, and the reproduc- ibility of the moisture-removal pattern for the same graphite was quali- tatively good. Under some circumstances the evolution of chemisorbed moisture from graphite is probably diffusicon controlled. The consequent compli- cation of the temperature dependence, which could occur in the MSRE startup operation, was the subject of a series of experiments with a single piece of ETL graphite at different heating rates. Prior to each trial, the canned graphite block was exposed to a circulating stream of moist helium for 2 hr. (It was not established that a 2-hr exposure to the moist helium was sufficient to saturate the graphite completely, but the quantity of water removed during later trials was about the same as that obtained from the as-received specimens.) The temperature at which the chemisorbed moisture evolution reached a maximum value was found to increase from about 270 to 400°C as the heating rate was 'in- creased from 5 to 65°C/hr. Discussion. Unless the MSRE graphite is markedly different from that studied, the release of moisture from the MSRE core should occur in two steps at temperatures below 400°C. Physically adsorbed water should be removed while heating the reactor core to about 150°C; the chemisorbed water will probably be evolved over the temperature interval 200 to 400°C. Upon decreasing the rate of graphite heating, the more tightly bound water would apparently be evolved at lower temperatures. While the quantity of water present in the as-received moderator elements will probahly depend on their handling history, there has been no evidence that the chemisorbed water content is ever very large. Be- cause of the upper concentration limit (1000 ppm) of the moisture ana- lyzer used for these experiments, the relatively large quantity of physically held water in as-received ETL graphite has not been accurately measured. In two experiments, more than 25 cm’ (STP) of water vapor per 100 cm® of graphite was noted, The quantities of chemisorbed water in both as-received and presaturated pieces of ETL graphite are listed in Table 6.3. If it is assumed that the MSRE moderator graphite is like FTL graphite, no more than 100 to 150 g of chemisorbed water should be evolved from the 68 ft” of moderator elements. The specified allowable oxygen contamination in the MSRE moderator graphite (30 em® of CO per 100 em? of graphite at 1800°C) is, however, equivalent to about 465 g 126 Table 6.3. Chemisorbed Water Content of ETL Graphite WaterVContent of Graphite Specimen Chemisorbed Water Graphite Run No. Condition Removed (mg) - mg/cm3 mg/g K-12 As received 23.5 0.075 0.142 K-14 30-min saturation 14.0 0.047 0.089 with moist He K-15 =~ As received : 19.4 ' 0.063 0.119 K-16 2=-hr saturation 23.3 0.075 0.142 with moist He : K-17 2-hr saturation 20.2° 0.065 0.123 with moist He K-18 2-hr saturation 22.7 0.073 0.138 with moist He ' K-19 - 2-hr saturation 29.2 ' 0.09% 0.178 with moist He K-20 '2-hr saturation 21.8 0.070 . 0.133 with moist He of water pér reactor loading. Thus the evolution of chemisorbed water might correspond to about a 30% removal of the total oxide contamination in moderator elements that just meet specifications. In an attempt to evaluate the effectiveness of the MSRE startup procedure for removing volatile impurities from the graphite moderator elements, a high-temperature technique was used for analyzing the off gas. The sample was a small cylinder of ETL graphite previously re- moved and immediately reinserted in an as-received RTL graphite block from .which the moisture was then removed by heating to 700°C as in the startup procedure. The small sample was then raised to 1800°C, and the evolved gases were analyzed. Of the 45 cm® of additional gas removed, 30.4 cm’® was hydrogen and 13.9 cm?® was CO per 100 cm?® of graphite. If there were no gain in moisture or oxide during transfer of the small graphite sample, 469 of the specified allowable oxide remained after the simulated startup procedure. As-received ETL graphite, however, after a thorough room-temperature evacuation, produced less than 10 cm? of CO at 1800°C, so there may have been some contamination or a marked variation between the specimens studied. 127 In any case, the proposed MSRE procedure should be effective in removing water vapor from the unclad graphite moderator elements. Fur- ther, the effectiveness of this operation should be, within reason, in- dependent of the physically absorbed moisture content of the graphite. Behavior of Carbon Tetrafluoride in Molten Fluorides As part of the previously described investigation of the effect of CF,; synthesis in the MSRE, an experimental study is being made of the reaction of CF,; with fuel mixtures, both in the presence and in the absence of graphite. ©Since the formation of either elemental fluorine or CF; in the MSRE would deplete the fuel mixture of fluoride ion, there would probably be at least a partial reduction of uranium tetrafluoride to uranium trifluoride. Accordingly, studies of the reaction rate of CF, with a "reduced" fuel mixture have been emphasized. Earlier reports on this program were confused somewhat by lack of analytical confirmation of the extent of reduction in a reduced fuel mixture; however, better results were obtained in later experiments in- volving the reduction of the molten mixture LiF-BeF,-ZrF,-ThF,-UF, (70-23-5-1-1 mole %) with zirconium metal turninge. In these tests, reductions corresponding to an analyzed 4 wt % of trivalent uranium in the molten fluoride mixture were maintained for long periods at 600 to 850°C in nickel containers, probably by virtue of excess solid reducing agent. The presence of UF3 in filtered samples from these melts was confirmed by petrographic observations and by x-ray diffraction data. CF, Reactions in the Absence of Graphite. During one series of experiments, the reaction of CF, with a melt containing UF3 was studied . by recirculating a known mixture of CF,; in helium through a reduced mixture of LiF-BeF,-ZrF,-ThF,-UF, (70-23-5-1-1 mole %, respectively) contained in nickel at temperatures of 700 to 850°C.* Progress of the chemical reaction was followed by mass spectrographic analyses of gas samples periodically withdrawn from the reaction system. An attempt was made to characterize the rate behavior of the re- action. ©Since concentrations of UF3 in the fluoride mixture were main- tained at relatively high values (2 to 4 wt % as uranium metal), the reaction rate was assumed to bhe dependent only on the CF, concentration. The reacting gas mixture (initially about 209, CF, in helium) was recir- . culated through the system at a flow rate of about 1.7 liters (STP)/min; a total gas charge of about 3 liters (STP) was used in each experi- ment. For purposes of this evaluation, the loss of CF,; by the system was treated as a homogeneous, nonflow reaction. A fair correlation of the rate data with the first-order rate equation *The recirculated gas was bubbled through the melt by means of a nickel dip line extending nearly to the bottom of the reaction vessel. - 128 No 1ln N_ = kCt a was obtained; in this expression NO is the original mole fraction of CF, in the system and N, is the mole fraction of CF, present at time t. Cal- culated values of the relative reaction velocity constant, k,, are listed below: Temperature . (°c) ke (hr-t) x 10°% 700 6.4 750 11.4 e - 800__ .. _3l.4 . 850 47,1 From a temperature-dependence plot of these vélocity'constants (Fig. 6.12), an activation energy of about 32 kcal/mole for this assumed first- order rate mechanism was. obtained. , CF, Reactions at a Graphite-Salt Interface. In the MSRE, the syn- thesis of CF, and possibly a significant portion of the back-reaction of UNCLASSIFIED ’ ORNL—LR—DWG 77354A TEMPERATURE (°C) (x10™%) 850 800 750 700 . \e o (o) F o (%3 o nN (o] 5] d & ke, RELATIVE REACTION VELOCITY CONSTANT (hr™?) 85 9.0 95 10.0 105 1.0 10,000/ (k) Fig. 6.12. Temperature Dependence of the Reaction of CF, with UFs in "Reduced" MSRE Fuel in the Absence of Graphite. 129 CF, with the "reduced" fuel mixture could occur at a graphite-salt in- terface. Accordingly, an experimental assembly for determining the re- action rate at a graphite-salt interface was developed. In order to deliver CF, to the reaction site without prior exposure to the salt mixture, a hollow graphite cylinder was brazed to a nickel tube and immersed in the molten mixture LiF-BeF,-ZrF,-ThF,-UF,; (70-23-5- 1-1 mole ¢) to which zirconium metal turnings had been added. Carbon tetrafluoride was admitted to the reaction system through this tube at . very low flow rates, which were accurately controlled by a specially de- veloped gas pump that utilized mercury displacement. The CF, effused through the thin vertical wall of the cylinder. Since anticipated reactions such as CF4+ 4UF3 - 4—UF4+ C were expected to involve only one species in the gas phase, the rate was monitored solely by ascertaining the loss of CF, by the reaction system. The rate of consumption of CF, was deduced from measurements of changes in concentration in the gas phase above the salt. When using helium at a known constant rate to sweep the gas space above the melt, the steady-state concentration of CF, leaving in the gas swept from a nonreacting system can be calculated for any rate of CF, in- put. ©Since the total gas flow rate leaving the system equaled the com- bined inlet flow rates of CF, and helium, in a chemically inert system, the concentration of CF, in the gas phase (initially pure helium) above the melt at any time can be expressed by the equation dNCF4 . =— (A-DN,. ), dt By CFs where N is the actual mole fraction of CF,; in helium at any time t, A is the stéady—state mole fraction of CF, for a chemically inert system with perfect mixing, n_ is the total moles in the gas space, and f is the total gas flow rate in moles per unit time. Mole fractions as a function of time were obtained from a calibrated, recording, thermal-conductivity meter. From a plot of ANCF4/At as a function of NCF,, the value of A was determined from values of the slope and intercept of a line drawn through the data points. In experiments on a nonreduced fuel-salt sys- tem, values of A calculated from the rate data agreed with the CF, con- centrations expected from flow rates to within 1 cm?® of CF, per hour. These results could be considered to demonstrate the precision of the ex- perimental system and to imply that CF, did not react with the nonreduced fuel mixture at a detectable rate. As far as standard free energies of 130 reaction are concerned, the driving force for reactions with UF, is about one-seventh that for UF;. Since UF3; was present in relatively large con- centrations, the reaction rate of CF,, according to the proposed equa- tion, should not have been affected by small changes in the UF3 concen- tration during the course of an experiment. Further, under the flow con- ditions of the reaction system, the effective concentration of CF, from the graphite-salt interface, which was the proposed reaction site, should have remained constant. Therefore a zero- -order reaction with respect to CF, could be anticipated and, if present, should have been evidenced as a loss of CF, at a constant rate. By comparing the calculated A value with a value of the input CF, flow rate calculated from the flow rate data from one experiment that was considered to have provided reliable data (Fig. 6.13), a rate of CF, loss was obtained which indicated that CF; had reacted with the reduced fuel mixture at a rate of about 16 cm’ (STP) of CF, per hour. In this experiment CF, was admitted at a flow rate of 35 cm’ (STP)/hr into approximately 2 kg of a fuel mixture which contained about 4.8 wt ¢ uranium as UF3. Additional experiments were initiated to confirm this finding and to examine the effects of tempera- ture and UF3 concentrations on the rate process. "he tests described here were terminated because the graphite cylinder became plugged; in- vestigations of this phenomenon have not been completed. ' UNCLASSIFIED ORNL -LR-DWG 77350 0.06 | | l EXP: WJ =30 CF4 FLOW RATE: 34.9 cc/hr AT STP Q0% [ |1 FLOW NATEG, 00.C 08./he AT TP CFy INPUT (mole froction} FROM FLOW RATE DATA: 0.260 FROM EXP RATE DATA: 0.159 APPARENT RATE OF CF, LOSS: 16.1 cc/hr AT STP 0.04 — e (ANTICIPATED RATE OF APPEARANCE OF CF, AS - CALCULATED FROM "BLANK" RUNS 0.03 | b T - - 2 (mole fraction/hr) 0.02 DN A 0.04 B NORMAL COMP: LiF —BeFp—2rfy—Thf,-UF, (70-23-5-1—tmole %) REDUCTION: ~4 wt l% U3t o} 0.02 0.04 0.06 0.08 0.0 n_/CF {AVERAGE mole FRACTION) 4 Fig. 6.13. Rate of Appearance of CF, in the Gas Phase Above 'Re- duced" Fuel at 600°C. 131 Production of Purified Materials Pure Uranium Trifluoride The pure uranium trifluoride that was needed for the investigation of phase equilibria in MSRE-type fused salt systems was freshly prepared. Previously batches of 0.25 to 2.25 kg of UF3 with minimum purity of 959 and, in most cases, purity of better than 979 were made in a reactor ves- sel 10 in. long and 3 in. ID by reacting uranium metal with UF,. Since a solid-solid reaction is necessary for producing UF3 by this method, a ro- tating spherical reactor vessel, resembling a ball mill, was devised in ~an effort to assure more efficient mixing and reaction than that obtained in the revolving cylinder used previously. Uranium tetrafluoride (159.66 g), which was dried by heating with ammonium bifluoride, and 40.86 g of "pickled" uranium metal, 1.3% excess, were added to the spherical reactor, along with six 3/4-in.-diam stain- less steel balls. The rig was assembled in a specially modified pot fur- nace, and the charge was heated under vacuum to 450°C and held at this temperature for 2 hr. The temperature was then lowered to 275°C, hydrogen was added, and uranium hydride was formed as long as absorption occurred. The temperature was next raised to 500°C to decompose the hydride under a vacuum; this produced finely divided, highly reactive uranium metal. The hydriding and decomposition steps were repeated. The valves on the gas inlet and exhaust line were closed and the line disconnected to main- tain a vacuum. The reactor was held at 900°C and rotated at 60 rpm for 24 hr. It was then cooled, hammered to dislodge any particles adhering to the sides, and then retreated at 900°C for 16 hr. The total recovery was 190 g of UF3. Essentially complete reaction occurred; at least no UF,; or uranium metal could be detected by x-ray or petrographic exami- nations, which were the examinations employed in the phase studies of systems to which the UF; was added. Preparation of Xenon Tetrafluoride Xenon tetratfluoride crystals suitable for x-ray diffraction and mi- croscopic studies were prepared by a procedure based on the work of Claassen et al.?2 A diagram of the apparatus is shown in Fig. 6.14. A nickel reaction vessel was connected to the fluorine gas manifold (de- scribed in the following section) and passivated with fluorine at a pres- sure of 45 psia and a temperature of 600°C. Then, 0.42 g of xenon and 0.24 g of fluorine (giving a volume ratio of 1:3) were admitted to the reaction vessel. The mixture, initially at 24 psia, was heated to a maximum temperature of 480°C, which was reached after 1 hr. The reaction of xenon and fluorine appeared to start at 307°C and to be complete at 480°C, as shown in Fig. 6.15. The reaction vessel was 23§, H. Claassen, H. Selig, and J. G. Malm, J. Am. Chem. Soc., 84: 3593, 1962. 132 UNCLASSIFIED ORNL-LR-DWG 78443 Fa IN. Xe IN (1) NICKEL REACTION VESSEL; VOLUME, 200 cm?3 (2) HOKE DIAPHRAGM VALVE NO. 413, MONEL @ ASHCROFT "DURAGAUGE ", 30-in. Hg VACUUM TO 30 psig . (@) Ya-in.-DIA, O.QES-in.-WALL NICKEL PROCESS PIPING (® coLD TRAP (3) PYREX GLASS U TUBE, 11-mm-0D Fig. 6.14. Apparatus for the Preparation of XeF,. UNCLASSIFIED ORNL-LR-DWG 7RQ91A .4 : 60 .. AN s— 50 1.0 - 7 . / .‘.. ./ 40 10.6 o s ———— 7 0.2 // P - . 350 2.8 " / /] O/ ‘ . 9.4 /,‘ 10 | : 909 0 100 200 300 400 500 TEMPERATURE (°C) TIME (min.) PRESSURE (psig) Fig. 6.15. Reaction of Xenon and Fluorine as Functions of Time, Pressure, and Temperature. — ———— . = 133 cooled to -78°C, and the unreacted gases were pumped off. The condensate was then sublimed to a Pyrex U-tube. The yield, estimated to be approxi- mately the theoretical amount, 600 mg, was a colorless, crystalline mate- rial. The frost-like crystals, sealed in the Pyrex U-tube, grew to a few millimeters in size in a few hours. These observations are in agreement with those of Claassen, et al.?3 Fluorine Gas Manifold- A manifold for regulating and controlling the flow of fluorine gas was built. Its principal uses have been to passivate gas-sampling cap- sules and other equipment needed in connection with the MSRE radiation effects tests (see'Chap. 5, this report) and to furnish fluorine mixtures for various program needs. The unit consists of an arrangement of valves and gages in a semienclosed panel located in a hood. A schematic diagram of the system is shown in Fig. 6.16. The primary requirement of the sys- tem was that it withstand fluorine at a pressure of 3 atm. This limited the structural material to Monel, nickel, and bronze. To assure leak tightness, all valve-to-tubing connections were silver soldered, and the connectors used were of a special Hoke, Inc., ultra-high vacuum type. The manifold was pretreated by admitting fluorine gas slowly after evacuating.?* At 5 psia the fluorine flow was stopped, and after a 10- min passivation period the system was evacuated through the soda lime 1 24"Handling Elemenl Fluorine Gas in the Laboratory," General Chemical Division (Allied Chemical and Dye Corp:) Manual PD-TA-35413, revised Aug. 15, 1958, Lib. of Congress Number QD-181.FL-G-315. UNCLASSIFIED ORNL-LR-DWG 78090 SAMPLE CAPSULE P Lo —< CP_ f\) HF SCRUBBER GAS MIXING _ (NoF FILLED) CHAMBER EQUIPMENT . SODA TIME VACUUM PUMP TRAP 1/4-in. NICKEL TUBING x 0.065-in. WALL PROCESS PIPING D