3 wish somenns ORNL-1556 This document consists of 14) poges. Copy ?3 of 249 copies, Series A. Contract No. We7405-eng-26 AIRCRAFT NUCLEAR PROPULSION PROJECT QUARTERLY PROGRESS REPORT for Period Ending June 10, 1953 R. C. Briant, Director A, J. Miller, Assistant Director Edited by: W, B. Cottrell DATE ISSUED O AK RIDGE NATIONAL LABORATORY Operated by CARBIDE AND CARBON CHEMICALS COMP ANY A Division of Union Carbide and Coarbon Corporation Post Qffice Box P Oak Ridge, Tennessee MARTIN MARIET TA ENERGY §YS8TEMS LIBRARIES TR 3 4456 0349535 3 ORNL.-1556 Progress INTERNAL DISTRIBUTION 1. é:"«::;_;. . Adamson 42. R, N. Lyon 2. R. O, 43, W. D. Manly 3. C. R.Baldock 44, L. A, Mann 4. C. J. Basgp 5. W. B. McDonald 5. E. S i1 J. L. Meem 6. D. S, Bl‘hng n A, J, Miller 7. F. F. Blanken K. Z. Morgan 8. E.P.Blizard = % E. J. Murphy 9. M. A, Bredig . H. F. Peppendiek 10. R. C. Briont P. M. Reyling 11. R. B. Briggs H. W, Savage 12. F. R. Bruce £. D. Shipley 13. A. D, Callihan 0. Sisman 14. D. W. Cardwell L. P, Smith (consultant) 15. J. V. Cathcart A. H. 5nell 16. C. E. Center R, W. Stoughton 17. J. M. Cisar C. D. Susano 18. G. H. Clewatt J. A. Swartout 19. C. E. Clifferd E. H. Taylor 20. W. B. Cottrel! F., C. Uffelman 21. D. D. Cowen B, M, Uthe 22. F. L. Culler E. R. VanArtsdalen 23, W. K. Eister ; F. C, Venderlage 24, L, B, Emlef (Y~12 # J. M. Warde 25. W. K. Ergen ; A, M. Weinberg 26. A, P. Fraas J. C. White 27. W. R, Goll & 3. E. P. Wigner {(consultant) 28. C. B. Grahamg 5, H. B, Willard 29. W. W, Grigogfeff (consultant) . C. Williams 30. W. R, G”. ,.:':;' ue, Wi!scm 31, A. Holloffder 32. A S f'jfiu&seholder 33. L Ty 1:- owe 34, W._fi Humes (K-25) 35. G¢W. Keilholtz 36. ,C. P, Keim M. T. Kelley F. Kertesz E. M. King C. E. Larsen R. S. Livingston 93-94, . Reactor Expe;:gmentuu Engineering Le;bmry Central Research Library s e , i "‘:i;'_ 5 95-97. 98. ) 99, \%“::;_._,_ ]00. 10%]112, 113. 114-121% 122. 123, 124-128. 129. 130. 131-132. 133-138. 139, 140, 141, 142, 143. 144-147. 148, 149, 150-152., 153-156. 157. 158. 159.166. 167. 168-175, 176-177. 178-179. 180. 181. 182. 183-185. 186-189. 190. 191, 192-193. 194-195. 196-201. 202 EXTERNAL DISTRIBUTION Air Force Engineering Office, Qak Ridge Air Force Plant Representative, Burbank Air Force Plant Representative, Seattle ANP Project Office, Fort Worth Argonne National Laboratory {1 copy to Kermit Anderson) Armed Forces Special Weapons Project (Sandia) .Atomic Energy Commission, Washington "B.c_’:tfelle Memorial Institute Bechtel Corporation Brookhaven National Laboratory Bureau of Aeronautics (Grant) Bureau of Ships California Research and Development Company Carbide and Carbon Chemicals Company {Y-12 Plant) Chicago Patent Group Chief of Naval Research Commonwedlth Edison Company Department of the Navy - Op-362 Detroit Edison Company duPont Company, Augusta duPont Company, Wilmington Foster Wheeler Corporation General Electric Company, ANPP General Electric Company, Rlcfiland Hanford Operations Office USAF-Headquasters, Office of Assistant for Atomic Energy ldaho Operahans Office (1 copy to Phl“:ps Petroleum Company) lowa State College Knolls Atomic Power L.aboratory Lockland Area Office Los Al cmos Scientific l_aboratory Massachuseffs Institute of Technology (Benedict) qusac:hu setts Institute of Technology (Kaufmann) Mon sanio Chemical Company Mound L aboratory National Advisory Committee for Aeronautics, Cl evefcmd (3 copies to A. Silverstein) National Advisory Committee for Aeronautics, Washmgtfin Naval Research Laboratory % New York Operations Qffice North American Aviation, Inc, Nuclear Development Associates, Inc, (NDA) Patent Branch, Washingten 204-215. 216-217. Jigtidn (1 copy to V. G. Henning) 218. San Francisc@iiperations Office 219. i erations Office, Augusta 220, Bt | 221, U S, : iolo@ical Defense Laboratory 222-223. iversity fCahforn,Radl ation Laboratory, Berkeley 224-225, i :!5:'" ':71"-_1:_ diation Laboratory, Livermore 226. & 228-233. Weg nghou se Electric Corpomfion 234-248. T:hmcal information ‘Service, Odk. Ridge, Tennessee 249. ;fbrh s-Wright Corp., Wright Aeronoutlccxl Division (K. Campbell} vi Reports previously issued in this series are as follows: ORNL-528 ORNL-629 ORNL.-768 ORNL -858 ORNL-919 ANP-60 ANP-65 ORNL-1154 ORNL-1170 ORNL.-1227 ORNL-1294 ORNL.-1375 ORNL-1439 ORNL-1515 Period Ending November 30, 1949 Period Ending February 28, 1950 Period Ending May 31, 1950 Period Ending August 31, 1950 Period Ending December 10, 1950 Period Ending March 10, 1951 Period Ending June 10, 1951 Period Ending September 10, 1951 Period Ending December 10, 1951 Period Ending March 10, 1952 Period Ending June 10, 1952 Period Ending September 10, 1952 Period Ending December 10, 1952 Period Ending March 10, 1953 FOREWORD ........... CONTENTS -------------------------------------------- PART |, REACTOR THEORY AND DESIGN INTRODUCTION AND SUMMARY .. i i it i i i e e et ittt e e nans 1. CIRCULATING-FUEL AIRCRAFT REACTOR EXPERIMENT . ... .. it in. Fluid Circuit ......... Reactor ., .,......... Instrumentation . .. ..... Dff-Gas System .. ...... Reactor Control . .. .. ... Electrical System .., .... Fuel Loading Facility ... Fire-Fighting Preparations -------------------------------------------- -------------------------------------------- ------------------------------------------- -------------------------------------------- -------------------------------------------- -------------------------------------------- -------------------------------------------- ............................................ 2, EXPERIMENTAL REACTOR ENGINEERING .. .. ittt i i e i e ee e Pumps for High-Temperature Liquids © . .o i i it ittt ittt i i e i ie et te s nnenn Cenfrifugal pump with combination packed~frozen flucride seal ... . ... .. ... ... ... ARE-size sump pump ... -------------------------------------------- Frozen-sodium-sealed pump for ARE moderater-coolant circuit . . . .o v vt it it v s ARE packed-frozen sealed fluoride pump . . . i it it it i e e e e e s Allis-Chalmers conned-rofor pump . v v v v it vt s it ot vttt o s ot ssnnaes e te s Frozen-lead-sealed fluoride pump . . o o ittt i et e e e e Laboratory-size sump-type pump with gas seal ... i it ittt ittt i e e Rotating Shaft ond Yalve Stem Seal Development . .o 0 i i ittt ii it n e Graphite-packed seals .. -------------------------------------------- Grooved-shaft packed-frozen seal fests . .. L i il i it i e e Bronze wool and MoS, packed-frozen seal tests .. ... ... ..., et et Frozen-lead seal tests . . Voringseal ......... BeF2 seal tests , ... .. Packing penetration tests Instrumentation .., ...... -------------------------------------------- -------------------------------------------- -------------------------------------------- -------------------------------------------- -------------------------------------------- -------------------------------------------- -------------------------------------------- llllllllllllllllllllllllllllllllllllllllllll OO OSSNy O n rmal = B ) BRI B el oend s bomd med e i) wwd cemd el mwemd el pmed mad met WKIMNMRO VOGNS NN e =d OO b B B M2 Lo vii 4. REACTOR PHYSICS ..o i i ittt ittt i it e et e saenannnaaneesas Kinetics of a Circulating-Fuel Reactor ., . . . ittt ittt ittt eanerans Behavior at long time after adisturbance .. . .. it ittt i e e e e e Oscillations .. . it ittt i i ittt e ttettetensaaraeaaeaeaeee e The first overswing following adisturbance ... ... ... . i it nn. Statics of the Circulating-Fuel ARE Critical Experiment .. .. . . 00 it ittt it oo 5. REFLECTOR-MODERATED REACTOR CRITICAL EXPERIMENTS . ............... Power Distribution . . .. ... ittt ittt i it it it ettt i ettt Leakage Flux . oo i it it it it s i e s vttt sttt as s eeonennnseesasnnsenss Control Rod Measurements .. ... i ittt ie ittt ennertonoseasenseesononas Danger Coefficient Medasurements . . ... ..ot ittt et s e esaennoneansansensonssas Correlation with Theory . ... . . ittt ittt it toteotnetnssntosesasssnnnsassss PART il. MATERIALS RESEARCH INTRODUCTION AND SUMMARY . .. .. i i ittt i i ittt 6, CHEMISTRY OF HIGH-TEMPERATURE LIQUIDS .. .. .. ... it ieennens Fluoride Mixtures Containing UF , . ... .. o i i i i i e A e U T RbF-ZrF -UF o o e e e e NaF-ZrF -BeF -UF , o i i s Fluoride Mixtures Containing UF, ... .. ittt Chloride Mixtures Containing UCL, .. ... .t it ittt e KCl-Ucl, Licl-ucl, TICl-uCl, ------------------------------------------------------ ------------------------------------------------------- ------------------------------------------------------ KC-LiC-UC e e i i e Coolant Development . .. .. i it ittt it ittt et ittt eieonnsseetonessonanss NGF-ZrF rBeF ) 4 vt e v et et s e et LiF-ZrF4 ----------------------------------------------------- Differential Thermal Analysis . . v v i it it it ittt it it et e e e s s nnoasnanaenens Production and Purification of Fluoride Mixtures .. . .. i i i i it ittt ittt ee e o e nnn Removal of HF from fuel batches .. ... .. . . i i it it e it it et Reduction of metal fluorides with Hy, ... .. o o i oo, Pilot-scale fuel purification . . .. ... . .0 it iiie it iteetneeroeesnesnns Fluoride production facilities . ... i ittt ettt it et nennonsnnensensas Preparation of Complex Fluorides . .. ... ... i i ittt it X-ray Diffraction Studies . ..o i ittt ittt ittt it et e e e e e e e viii NaF-UF, NaF-ZrF ------------------------------------------------------ oooooooooooooooooooooooooooooooooooooooooooooooooooooo 26 26 26 26 27 28 30 30 31 31 31 31 37 39 39 39 40 41 42 42 42 42 42 43 43 44 44 45 45 46 47 48 48 49 49 49 7. CORROSION RESEARCH Fluoride Corrosion of Inconel in Static and Seesow Tests ... ........ s Effect of exposure time .. ....... e e e e C e e Residual hydrogen Huoride . ... ........ e e e e e Fluoride pretreatment . .. ... e e e e e s et e e Structural metal fluoride additiens . ... ........ ... c e c e Structural metal oxide additions . , . ... ... . . . oL, e ‘e Chromium metal addition . ... ... ... i i i e e Zirconium hydride addition .................. e r e ey Carbon addition ... .. . ittt ittt n et s s ater et r e Crevice CoMmOSION & .t it et i it s sa e snseessononronnnsnennennn Inconel container pretreatment . .. ... ....... e e Static Corrosion of Stellite by Fluorides e e e Fluoride Corrosion of Inconel in Rotating Test .. ... .. .0 eon.. e o Fluoride Corrosion in Inconel Thermal Convection Loops . ... ... .. ... ... Effect of exposure time .. ......... Temperature dependence . .. ... ... . Chromium and nickel fluoride additives . Fuel purity . ........ Loopoxide films ., ... .,.......... Nonuranium bearing Huorides ... .... Fluoride Corrosion of Nickel Thermal Convection Loops ------ - s » e 8 e » u & s & @ * . . 2 v a w < * 2 = ®F N 8 W A A B *T B & O n ----- 4 5 &8 2 v & w B & . . “ 8 4 s » . a LI 4 + 8 B @ & & & & ¥ ¥ B s & 3 4 s Fluoride Corrosion of Type 316 Stainless Steel Thermal Convection Loops . . . . Liguid Metal Corrosion .. .......... Mass transfer in liquidlead ... ... ....... e e e e . BeO in sodiumand NaK . . . . ... .. .o oo, C e e e e e e Coated beryllium in sodium . .. .. ot i i e e o Carboloy in lead and sodium .. ... ... .. .. f e e s e 8. METALLURGY AND CERAMICS ..... Welding and Brazing Research ., .. ... Welding of heat exchanger tube bundles Brazing of air radiators .. ... . ... ... Dilution and diffusion of brazing alloys . Brazing Carboloy to stainless steel . .. -------- B4 A sy s High-temperature brazing alloy evaluation tests , .. Creep and Stress-Rupture Tests of Structural Metals Fabrication Research ... ........... Extrusion of high-purity Inconel tubing . Air oxidation of columbivm ., . ... .. Control rods for the G-E reactor . .... Hot-Pressed Pump Seals . . . Fabrication of small compacts with Mo$ B4 a oW s 4o « 8 8 P 2 e » 2 % & » 4 » v a a2 %« 3 * & N s Metallographic examination of compacts with MeS, . -------- 50 51 51 51 52 53 53 54 54 55 55 55 56 57 58 58 62 62 62 62 64 64 64 - 68 68 69 71 72 72 73 73 74 76 76 77 77 77 77 79 79 80 Fabrication of face seal rings . . . .. .. it i i i it e Surface sulfurization of molybdenum , . High-Conductivity Metals for Radiators . Chromium-clad QFHC copper ...... Chromium and nickel electroplate on copper Inconel claddingon copper .. vttt ittt e Type 310 stainless steel cladding on copper ... ...... .. Type 310 stainless steel cladding on sil Copperalloys . .............. . -------------- oooooooooooooo ver ------------------------ * 5 5 & ® s a s Solid Fuel Formed in Spheres .o .ttt i it it ittt s et enietannonnonnnnss Heating under vacuum . ..o i ittt e ittt i et et aann Cee s Three-phase carbon GrC . v it i i it ittt et tn e st e onsenaonsans Molten alloy passed through a small orifice ,....... e e Cere s e Settling of particles through o molten salt Inflammability of Sodium Alloys . .. . .. i ittt ittt ittt e s ot amiCS o v v it i it v e v et o s o s s s e s oseesseesonsnessensons . Development of cermets , ... ...... Effect of heating rate on the beryllium oxide blocks for the ARE 9. HEAT TRANSFER AND PHYSICAL PROPERTIES ------------ ooooooooooooooooooooooooooo Thermal Conductivity . . i ittt i iiientte et ensennnnsesenans et e . Density and Viscosity . ........... Heat Capacities of Liquids ... ...... Surface Tensions of Fluorides . . . . o ittt ittt ittt ittt et i s ennnss e Vapor Pressures of Fluorides .. . . ..o ittt it it ittt st taesronnnnnn ON LooOPS . vttt ittt ittt e s e e e Velocity Distributions in Thermal Convect Forced-Convection Heat Transfer with NaF-KF-LiF High-Temperature Reactor Coolant Studies -------------- Turbulent Convection in Annull . . 0 Lttt it i it it it e sttt oe o enononoernsnens Circulating-Fuel Heat Transfer , .. ... Bifluid Heat Transfer Experiments ., .. 10, RADIATION DAMAGE .. . it it ittt ettt e e i e eaaaens Irradiation of Fused Materials .. ..... fnpife Circulating Loops + . oo i i ittt i it it it it e et e st cessas et onaas Cene Creep Under Jrradigtion . 0 i i ittt ittt ittt it is s s e e ennannonsnns v Radiation Effects on Thermal Conductivity 11. ANALYTICAL STUDIES OF REACTOR Chemical Analyses of Fluorides and Their Determination of oxygen in metallic oxides with bromine trifluoride Volumetric determination of zirconium Determination of uranium trifluoride and Colorimetric determination of zirconium --------------------------------- MATERIALS ...... ... vt coas Contaminants . . v v v v vt o v o enas e ee e aees -------------- ----------------------------------- metallic zirconium ... ..... ------------- 80 80 81 82 82 82 82 82 82 83 83 84 84 84 84 85 85 85 86 86 88 88 88 89 90 91 91 92 92 94 94 94 95 96 97 97 97 98 98 98 ““Tiron' as a reagent for the determination of uroniom . .. .......... Petrographic Examination of Fluorides .. ... .. 0oy Summary of Service Chemical Analyses ... ..................... 12. FLUORIDE FUEL REPROCESSING ... ci v v it it neenntanonsns Dissolution of ARE Fuel .. . . it ii i i it ittt et s et neraesanns Solvent Extraction . v v v v it ot v o vt oo et o ranonsnsasanss o Corrosion in Dissolution of ARE Fuel ... .. e e h e Plant Processing .+ . v v v v v ot v vt ot oot a asansnenesansoscnssos INTRODUCTION AND SUMMARY .. ... ...t iiinnienenns 13. BULK SHIELDING FACILITY .. .. ittt et ii i e i e e Neutron Spectra for the Divided Shield . ... ....... .. ... ot Energy per Fission and Power of the Bulk Shielding Reactor ... ....... Gamma-Ray Air-Scattering Caleulations . ... .. oo oo v e Differential results . ... .. ... ... ... e e e e Integral results ., . . . i e e it e e e i e M., LIDTANK FACILITY ... it ittt i i s et n s e asas . Reflector-Moderated Reactor Shield Tests . ... ........ . .00 Shield configurations . . . ... i v it ittt it n i e s e Gamma-ray attenuation data . . o . v oo it b i i i e s e Neutron attenuation data . .. ... v v v s ek e r s s n e e ma e e n e e as e Removal Cross Sections . . .. 0o i i i n it ittt i it i Facility Modification .. . it vt e e i ettt i as v annosaons 15. TOWER SHIELDING FACILITY . ...van i iionn, PART IV. APPENDIX 16. LIST OF REPORTS ISSUED DURING THE QUARTER . ........... llllllllllll ------------ oooooooooooo ------------ ------------ nnnnnnnnnnnn oooooooooooo ------------ ------------ llllllllllll llllllllllll ------------ ------------ QQQQQQQQQQQQ oooooooooooo uuuuuuuuuuuu oooooooooooo ------------ ------------ ------------ llllllllllll ------------ 107 108 108 108 109 111 113 119 119 120 120 121 122 123 124 127 xi ANP PROJECT QUARTERLY PROGRESS REPORTY FOREWORD This quarterly progress report of the Aircraft Nuclear Propulsion Project at ORNL records the technical progress of the research on the circulating-fuel reactor and all other ANP research at the Laboratory under its Contract W-7405-eng-26. The report is divided into three major parts: |. Reactor Theory and Design, [l, Materials Research, and lll. Shielding Research. Each part has o separate introduction and summary. _ The ANP Project is comprised of about three hundred technical and scientific personnel engaged in many phases of research directed toward the nuclear propulsion of aircroft, A considerable portion of this research is performed in support of the work of other organizations porticipating in the national ANP effort. However, the bulk of the ANP research at ORNL is directed toward the development of a circu- lating-fuel type of reactor. The nucleus of the effort on circulating-fuel reactors is now centered vpon the Aircraft Reactor Experi- ment -~ a 3-megawatt, high-temperature prototype of a circulating-fuel reactor for the propulsion of air- craft. This reactor experiment is now being assembled; its current status is summarized in section 1, However, much supporting research on materials ond problems pecuhar to the ARE will be found in other sections of Parts | and |l of this report, The ANP research, in addition to that for the Aircraft Reactor Experiment, falls into three generdl categories: (1) studies of aircraft-size circulating-fuel reactors, (2) material problems associated with advanced reactor designs, and {3) studies of shields for nuclear aircraft. These three phases of research are covered in Parts |, 11, and 1], respectively, of this report. INTRODUCTION AND SUMMARY The Aircraft Reactor Experiment (sec. 1) is now being rapidly assembled; most of the component parts have been received ond installed in the ARE Building. There were no significant design modifi- cations during the quarter, but a supplementary off- gas system to handle contaminated pit atmosphere is contemplated. Installation of the fuel-circuit piping is nearly complete, and the detailed testing and cleaning procedures for both the fuel and the sodium circuits are being established. Integral parts of the system, including the water system, the main off-gas system, and the rod cooling sys- tem, have been completed. The fill and flush tanks are installed, wired, and insulated. The reactor core has been completely assembled, welded, and is ready to be shipped to the ARE Building. All instrument indicators have been installed, and the interconnecting lines to the control room are es- sentially complete, Routine and emergency pro- cedures for handling sodium and the fluorides have been established. The development of a satisfactory ARE pump seal has required major research effort (sec. 2). Conventional centrifugal pumps have the required pumping characteristics, but the aftainment of a reliable seal which will retain the high-temperature fluorides with leskage rates of less than 35 cm? per day continues to be an outstanding development problem for the ARE. This seal problem is being investigated in several pump loops, as well as in numerous seal testers, with the result that several seal-types now either appear to be potentially capable of fulfilling the leakage requirement or have actually dome so. The frozen-lead seals for fluorides have operated with vertical shafts with negligible leakage for extended periods of time; gas-sealed pumps have also been satisfactorily operated. At present, however, the ARE plumbing arrangement requires a horizontal shaft pump for which the following packed-frozen seals appear promising: {1) alternate layers of copper (or bronze) wool and a grophite-MoS, mixture, (2) triangular rings of Graphitar bearing on sintered Cu + MoS,, and (3) foyers of NaBeF, packing. I has been demonstrated that oxygen has a deleterious effect on all seals, so they are now blanketed with helium. The frozen-sodium seal pumps for the ARE reflector-coolant (sedium) circuit have been fabricated. Recent studies of an aircraft-size reflector-mod- erated- reactor system have included analysis of the control of a 200-megawatt power plant and some parametric studies of fuel volume in the heat ex- changer and in the reactor for vorious core diame- ters and power levels (sec. 3), From the 200- megawatt power plant control study, it appears that the reactor temperature will not tend to rise ot a rate greater than 100°F/sec, even if cooling of the engine radiators is sfopped instantaneously. Insuch a case, the negative temperature coefficient will adequately control the reactor, The kinetics of circulating-fuel reactors and the interpretation of the ARE critical experiment com- prise the recent reactor physics studies (sec. 4). Introduction into the core of 1% A%, if not compen- sated by control rods, would cause a 100°C rise in the reactor equilibrium temperature. On the other hand, power oscillations damp out quickly and no serious overswing in power or in tempera- ture is expected. The ARE critical experiment dato are being extrapolated to provide additional information on the statics of the ARE. The critical assembly of the reflector-moderated reactor has provided measurements of power distri- bution, leakage flux, and control rod effective- ness, as well as danger coefficient measurements of materials in both the fuel and reflector regions {sec. 5). The end leakage of fast neutrons is six times the side leakage when the end fuel ducts are shielded with boral sheets, The boral reduces the ke” by 2% and correspondingly modifies the power production in its vicinity. Extrapolation of the uranium required in the experiment to an air- craft reactor with Inconel structure indicates that the critical mass of the latter is about 30 pounds. ANF PRQJECT QUARTERLY PROGRESS REPORT 1. CIRCULATING-FUEL AIRCRAFT REACTOR EXPERIMENT E. S. Bettis, ANP Division Duwring this quarter, effort on the ARE Project has been concentrated on the installation of equip- ment in Building 7503, No significant design modifications or alterations were made, and the work has progressed steadily with no serious inter- ferences. The major emphasis was on the installa- tion of the fuel-circvit piping, which is now neorly completed. The checking out of auxiliary systems was started. The water-cooling system has been com- pletely checked and put in stand-by condition for actual operation, All water lines, water heat ex- changers, valves, tanks, and lines were flushed, the flows were checked, and the system was filled with a rust inhibitor to keep it in readiness for use, A check has been started on the off-gos system, but this has not yet been completed. All major components, except drive motors for the pump, have been received. All volves have been received, except two of the 2-in. sodium valves, A large amount of effort went into preparing the testing and preliminary operating procedures. The first draft of these procedures has been completed and is being carefully studied for revision, As was expected, some bosic changes in the original con- cept must be made, ond during the next few weeks all possible effort will be made to complete these procedures. Much research that bears directly on the experi- ment is contained in other sections of this report. Of particular significance are the following: pump seals {sec. 2), kinetics of circulating-fue! reactors (sec, 4), composition and production of the fluoride fuel (sec. 6), corrosion of the fluoride-Inconel system (sec.7), high-temperature strength of Incone! in fluorides (sec. 8), physical properties of the fuel (sec. 9), effect of radiation on creep, thermal con- ductivity, and comosion of Inconel (sec. 10), and reprocessing of the fuel (sec. 12). FLUID CIRCUUT G. A, Cristy Engineering and Maintenance Division The fuel piping system is almost completed and initial testing is expected to begin the middle of June. This work has proceeded satisfactorily and according fo schedule, Approximately 5% of the welds have foiled o pass inspection and have had to be reworked. This represents less trouble than waos anticipated, since the standard for an accept- able weld is high. The hot fuel dump tank is completely insulated and ready for use. The fill and flush tanks have been installed, and installation of the heaters and the insulation is almost completed. Figure 1.1 shows the status of the tank pit as of the middle of the quarter, Installation of the piping for the sodium circuit will not be started until the fuel piping is com- pleted. The sodium-to-helium heat exchangers are now in place and the sodium pipe lines can be installed as soon as the fuel piping is completed, The final installation of the sodium piping must await installation of the reactor; according to the present schedule, the reactor will be installed during the latter port of June. Pumps for the sodium circuit have been completed and should be ready for installation about the middle of June. seal, These pumps employ a frozen-sodium - which has been thoroughly tested by the Experimental Engineering Group. No decision has yet been made as to the type of seal to be used for the fluoride pump (cf., sec. 2), Strain gages have been installed on the fue! piping system and the calculated prestress has been verified by measurements made with these gages. Additional checks will be made as the system is warmed up. A complete report on the stress analysis of the system is being compiled. REACTOR The reactor, including the beryllium oxide block, the fuel tubes, and the pressure shell, is completed, except for the final closure welding (Fig. 1.2). The reactor core will be transferred from the Y-12 shops to Building 7503 about the latter part of June, but it will not be installed in the system until the prechecking of the fuel circuit has been completed. Prior to installation in the system, the reactor will be helium checked at a temperature of 1200°F to maoke sure that the fuel circuit is tight with respect to the moderator volume, Wi @ o HOT FUEL PERIOD ENDING JUNE 10, 1953 - j PHOTO 1412107 "DUMP TANK Fig. 1.1. ARE Tank Pit. INSTRUMENTATION R. G. Affel, ANP Division All instrumentation has now been completed and most of the pneumatic lines have been checked. The leak detection system has been further simpli- fied, but no other alterations are contemplated. Liquid-level and flow instruments have been com- pleted and tested insofar as is possible without actual operation. Interconnection of the control room with instrument panels, relay racks, amplifier boards, and control actuators is 98% complete. A very complete labeling system, with a permanent wall-mounted check print for each control point, has been effected. checked as rapidly as an integral part is completed, and by the middle of June, the control room will have been completed and checked. The control wiring is being The annunciator points have been checked and tested insofar as the preliminary status of the sending points allows. The same condition exists with the connections to the graphic panel. No significant problem remains, so far as the instrumentation of the ARE is concerned. OFF-GAS SYSTEM The main off-gas system as originally designed has been installed and partly checked. A con- siderable amount of thought and calculation has been given to a secondary system for handling pit gas in the event of pit atmosphere contamination. The evaluation of this system has not been com- pleted, but it is fairly certain that it will contain a charcoal adsorber maintained at liquid-nitrogen This adsorber will probably consist of two tanks located outside the building and temperature. ANP PROJECT QUARTERLY PROGRESS REPORT PRESSURE SHELL S FISSION CHAMBER THIMBLE REFLECTOR & COOLANT TUE!E‘% y ; b BeO MODERATOR BLOCKS B A MODERATOR CAN FUEL TUBE INLETS PHOTO {1208 JFISSION CHAMBER THIMBLE | FISSION CHAMBER THIMBLE . BRREFLECTOR CODLANT TUBES Fig. 1.2. ARE Reactor Core. (a) BeO moderator blocks partly stacked. (b) BeO moderator blocks in place. arranged to permit alternate adsorption and re- generation cycles. This auxiliary off-gas system is designed pri- marily to handle contaminated pit atmosphere, but it can also be used for system-gas disposal in the event of trouble with the primary off-gas system. REACTOR CONTROL The active fission chamber has passed several hundred hours at 800°F with no increase in back- ground or deterioration of the octive cylinder. This chamber therefore appears to be satisfactory for ARE use., The control console and its associated servo, safety rod actuator mechanisms, efc. are completely installed and almost entirely checked. The control rod ‘‘igloo’’ assembly is nearly complete. It was necessary to rewind the safety rod drive motors in order to make them reversible without limitation of the number of reversals per~ mitted in any unit of time. All safety rod drive, control rod servo, fission chamber, and source drive mechanisms have been assembled. A dummy control console for checking the ac- tuator assembly has been built, The control actu. ator assembly is 75% complete. ELECTRICAL SYSTEM The electrical work in the building is on schedule. All major electrical circuits have been wired; however, the fluid circuit pumps have been changed from 5 to 15 hp, and therefore new circuits will be needed for these motors. |Installation of annulus heaters is proceeding wherever completion of the lines permit., The motor generator sets have been moved to a shed outside the building. FUEL LOADING FACILITY G. J. Nessle, Materials Chemistry Division Preliminary work has been started to provide adequate facilities for loading the processed fluo- rides into the ARE. A platform is to be built inside the storoge tank pit for installation of the loading furnace to be used in transferring the processed fluorides into the storage tanks. PERIOD ENDING JUNE 10, 1953 FIRE-FIGHTING PREPARATIONS D. R. Ward, ANP Division In view of the use of hazardous materials and the high-temperature operations connected with the ARE, some fire-fighting preparatims have neces- sarily been made. The materials and techniques to be employed in case of fire are based upon the considerable experience acquired during the past two years by the Experimental Reactor Engineering Group. The test pits will be open while the fuel system is being cleoned and leak tested so that the usuadl direct fire-fighting methods can be employed if NaK is used and a fire occurs. Graphite powder will be the preferred extinguishing material, with Ansul extinguishers filled with Metal-X on hand for additional safety. Several buckets of graphite will be strategically located in each of the three pits. Standard safety items, such as proper gas masks, fire-fighting suits, etc., will be on hand. During the reactor operation period, the test pits will be closed and access for fighting fires will be impossible. During this time, any fires must be combated remotely. Reservoirs of graphite powder will be located over the most vulnerable parts of the system and will be fitted with dumpers which may be operated from the control room, It is expected that the (approximately) 80% helium plus 20% qir atmosphere in the pits during reactor operation will help to retard fires, but combustion tests have not been conducted in such atmospheres. Metal catch-trays containing graphite powder will be placed in wulnerable locations, such as beneath pump seals and heat exchangers, to catch leakage both during leck testing and during reactor operation. All temporary flammable material, such as wooden stairways, will be removed from the pits before high-temperature testing is started. Exposed pit wiring will be wrapped with asbestos tape ond coated with flame-retardant paint. A flexible exhaust hose will be in readiness to reach any fires to withdraw noxious fumes from their points of generation. ANP PROJECT QUARTERLY PROGRESS REPORT 2. EXPERIMENTAL REACTOR ENGINEERING H. W. Savage, ANP Division During this quarter, the developmental effort of the Experimental Reactor Engineering Group has been devoted mainly to determining final specifi- cations for the pumps and the pump seals to be used in the fuel and moderator-coolant circuits in the ARE. Numerous seal testers and several pump loops are being used concurrently to obtain the necessary information. Fabrication of three 100-gpm frozen-sodium-sealed horizontal-shaft centrifugal pumps for use in the moderator-coolant circuit of the ARE has been completed. The operational parameters of these pumps are being determined in detail. A 100-gpm double-cell a-c electromagnetic pump has been delivered to the ARE building for use in the pro- posed NaK precleaning operation. A frozen-sodium seal tested on a 21/2-in. (ARE-size) shaft has given in excess of 1,000 hr of continuous, reliable, and predictable performance. Successful stop-start procedures have been established. The seal is formed by freezing sodium in the seal gland. Also, since the seal is the coldest part of the system, sodium oxide from the circulating sodium precipi- tates and accumulates in the seal. An extensive program is under way to determine within the next few months whether a horizontal shaft seal can be proved sufficiently reliable for operation with fluoride fuel in the ARE. The three types of packings tested to date that show promise of sufficiently low leakage and reliable, low-power consumption are (1) alternate layers of copper or bronze wool and a gmphii‘e-MoS2 mixture, (2) an arrangement of rings of triangular cross section in which clternate rings are Graphitar and sintered copper impregnated with MoSz, and (3) a series of annular cells packed with a material of high BeF, content, such as NaBeF,. Materials that are high in Bef,_ content are glasses, and the temperatures at which they soften vary over a wide range. To provide o steep thermal gradient over the seal length, one end of the seal is heated and the other end is liquid cooled. Two alternate types of pump systems for fluo- rides are being constructed for tests that are to begin in June. One system will have a frozen-lead- sealed pump with o moiten lead—molten fuel inter- face above the seal. In a test run, this type of seal 10 operated successfully for 500 hours. The other system will have o gas-sealed sump pump, the sump of which would replace the ARE expansion tank, An Allis-Chalmers 5-gpm canned-rotor pump is being set up for test, and a small 2-gpm 100-psi- head air-driven pump for inpile operation is being modified for seal testing. The use of thin-walled bellows (0.005 in.) has been proved entirely practicable by tests operated in excess of 2500 hr, with many cycles, provided the temperature of the contacting fluoride is held below 1100°F, It has been demonstrated that an external oxygen atmosphere is deleterious to the successful opera- tion of frozen-sodium, frozen-lead, and frozen-fluoride seals. Consequently, all such seals will be ex- ternally blanketed with inert gas. Parameters of operation of inert-gas-blanketed seals are calculable on the basis of coolant temperature, liquid film thickness, seal length, and the pressure differen- tial across the seal. PUMPS FOR HIGH-TEMPERATURE LIQUIDS W. B. McDonald A. G. Grindell W. G. Cobb G. D. Whitman W. R. Huntley A. L. Southern J. M. Trummel P. W. Taylor ANP Division Centrifugal Pump with Combination Packed- Frozen Fluoride Seal. Three tests were conducted in which the 50-gpm centrifugal pump(!? was used to circulate the fluoride NaF-ZrF,-UF, (50-46-4 mole %). The operating conditions for these tests were: pump speed, 900 to 1400 rpm; pump suction pressure, 4 to 6 psi; pump discharge pressure, 30 to 60 psi; fluid flow, 10 to 30 gpm; fluid tem- perature, 1200°F. The shaft of the pump was con- structed of 2'/2-in.-dia stainless steel coated with Stellite No., 6; the fuel leakage rate at the post seal was 40 to 150 g per 24 hr; and the seal L/D ratio was ¥. In the first of these tests, the gland was packed with Dixon’s ‘‘Microfyne’’ flake graphite powder retained by close-fitting APC graphite rings. {Dy, B. McDonald et al., ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p, 18. During this 340-hr test, the seal leckage was low and performance was generally satisfactory., The gradual loss of graphite from the seal caused the termination of the test, For the second test, the seal was packed with braided copper sheathing filled with MoS, powder, The power required to rotate the shaft was quite low, but operation was characterized by consider- able leakage of MoS, powder from the seal. After 260 hr of operation, sufficient MoS, powder had been lost to cause fuel leakage through the seal. Post-run examination of the seal showed the copper sheathing to be generally worn at the shaft surface. Stop-start tests were difficult to make during this run, and they always involved the danger of a massive leak, because the temperatures in the seal could not be measured accurately or readily con- trolled. For the third test, the packing gland was filled with strands of copper rope impregnated with Mo3,. Smooth operation was experienced in the early stages of this run; however, power fluctuations became greater, ond more power was required to drive the shaft as the test progressed. The test was terminated after 385 hr because of o gross sea! Jeak that resulted from frictional overheating of the seal. ARE-Size Sump Pump. All parts for the high- temperature model of the ARE-size sump pump(z) have been received, and the pump is being as- sembled in a test loop. A sectional view of this pump is shown in Fig. 2.1. The sump tank of this pump is designed to serve as one of the ARE surge tarnks, and low-temperature tests show that priming, vortexing, and fluid degassing will not be problems. The gas seal for this pump consists of a Graphitar ring rotating against a stationary hardened-tool- steel surface, The gas-seal wear faces will be maintained at a low temperature during operation by circulating coolant in three different regions at or near the seal. Hydraulic tests of a mockup of this pump with water, water and glycerine, and tetrabromoethane indicated that the design provides for stable pump operation and adequate degassing. Tests planned for this pump during the month of June will determine the operating characteristics at ARE tempercture and flow conditions. (Dipid., p. 20. PERIQOD ENDING JUNE 10, 1953 Frozen-Sodium-Sealed Pump for ARE Moderator- Coolant Circuit. The ARE mode! FP pump(®) with a frozen-sodium seal has pumped sodium for more than 1000 hours. Operation has been at near the ARE moderator-coolant pump design conditions of 100 gpm at 20-psi developed head with sodium temperatures of 1100 to 1250°F., The pump suction pressure has been varied from 10 tc 50 psi to get data on seal leakage. Six pump shutdowns have been made to correct loop difficulties or to make loop changes., Shutdowns and startups have been made without difficulty, although considerable time (1 to 2 hr) and care are required to effect g safe startup, For a safe startup, it is necessary to warm the frozen seal. The most critical element of this pump, as in most other high-temperature liquid pumps, is the shaft seal. The sodium freezing gland for this seal is 513/16 in, long on a 21/2-in.-dia shaft with a 0.030-in, radial cleorance in which the sodium is frozen. Earlier difficulties with seal seizure ond roughness have been eliminated by mointaining a helium gos blanket between the cold end of the seal ond the atmosphere, Seal operation has been sufficiently reliable and reproducible to permit taking quantitative data on seal power and leakage while varying speed, pressure, and seal coolant temperature. An analysis of seal data indicated that viscous liquid film theory(*) can be used to at least partly describe seal characteristics, in particular, the scul power requirements and the seal leckage rate. In the viscous film model, the equilibrium thickness of the liquid sodium film at the melting point is determined by the rate at which the heat energy of viscous friction is removed from the liquid film. In turn, the rate of addition of heat energy is the mechanical power supplied te the sec!, and this must be equal to heot removed from the seal by the coolant. The rate of leakage by viscous flow is related to the film thickness and to the pressure differential. The viscous film theory can be used to obtain significant dimensionless numbers which may be used to describe seal characteristics. Two such numbers are the power number, Nps and the leak (pid., p. 21. )H. Rouse, Elementary Mechanics of Fluids, Chap. 6, Wiley, New York, 1946, 11 ANP PROJECT QUARTERLY PROGRESS REPORT UNCLASSIFIED DWG. ESK 160234 -—-GAS SEAL LABYRINTH SEAL EVEL INDICATION PUMP SUMP AND SYSTEM SURGE DISCHARGE SUCTION N Fig. 2.1. ARE-Size Sump Type of Pump (Centrifugal Pump Model DAB Assembly). 12 number, NQ, which are defined in the following: N P P AT UrL and ouU3 (AT) L e T T .. ! AP +7 208 where P = power, AT = temperature difference between the melting point of sodium and the seal coolant tem- perature, U = over-all coefficient of heat transfer from liquid film to coolant per unit area of film, r = shaft radius, ' L. = seal length, O = volume leakage rate, AP = pressure difference across seal, i = viscosity of sodium at its melting point, n = speed of shaft, According to viscous film theory, the numbers N, and N, should be constants. The experimental data obtained by varying AP and AT are shown in PERIOD ENDING JUNE 10, 1953 Figs. 2.2 and 2.3, The slopes of the lines agree fairly well with theory. The scatter may be due to error in instrumentation ond failure to observe the presence of pertinent factors, such as shaft surface roughness and contamination of seal sodium, which were included in the analysis. Shaft surface rough- ness may explain the fact that the absclute value of the leakage number (N ) determined experi- mentally is several times larger than the theoretical value, The presence of oxygen around the frozen-sodium seal may account for scme of the discrepancy When the cold end of the seal was open to air, operation deteriorated over a period of three or four days and usually ended in sudden shaft seizure., That this was due to oxide accumulation in the seal oppears obvious, since the condition was alleviated by addition of the helium gas blanket. However, the seal also accumulates oxygen from within the system, since it contains the coldest sodium of the system. Seal roughness and relatively high leakage rates have been observed following periods of no flow through the bypass filter in the loop. The between theoretical and experimental data. 1O e [T peeemeees T | SEAL POWER (hp) LINE DRAWN FOR N, = —— Z ....................................... | 120 DWG. 19895 2Yp-in.- dia SHAFT, 5'%g-in.-LONG COOLING GLAND. ' WATER COOLANT WITH ABOUT S-gpm FLOW PUMP FLUID AT ABOUT t2Q0°F AT MOST POINTS | —" SPEED AT MOST POINTS, 1700 rpm ST | C = CONSTANT o (ATYUrL 180 200 220 140 160 MEAN GCOOLANT TEMPERATURE (°F} Fig. 2.2. Seal Power vs, Seal Coolont Tempercture for Frozen Sodium Secled Pumps, 13 ANP PROJECT QUARTERLY PROGRESS REPORT Ay DWG. 19896 2Ys-in.-dic SHAFT, 1700rpm, 5'%g-in-LONG GOOLING GLAND. FP PUMP WITH FROZEN SODIUM SEAL LINES DRAWN FOR LEAK NUMBER 3 3 N, = QUIATYL CONSTANT (AP r P a® WATER COOLANT WITH FLOW OF ABOUT 5gpm PUMP FLUID AT ABOUT 4200°F 1000 SEAL PRESSURE DIFFERENCE, AR 50 psi 5 25 psi 10 psi = 5 psi o 9 2 ~ M & © ~ 100 L o <1 & = 5 - I < 10 20 50 100 200 SODIUM MELTING POINT MINUS MEAN COOLANT TEMPERATURE (°F) Fig. 2.3. Seal Leckage vs. Seal Cooling Temperature Difference for Frozen Sodium Sealed Pump. 14 deterioration of operation resulting from this condi- tion is best alleviated by mointaining clean sodium in the system aond, if necessary, by infrequent clean-outs of the sodium in the seal. ARE Pocked-Frozen Sealed Fluoride Pump. An ARE model pump®® circulating the fluoride NoF- ZrF ,-UF, (50-46-4 mole %) with a packed-frozen seal was operated for o period of 49 hours, Shaft seizure occurred, and the Carboloy sleeve under the seal was found to be cracked. The seal packing consisted of strands of copper wire rope which were socked in on oil suspension of MeS, for a period of 24 hr before installation, During the test, the pump operated at 1250 rpm with a flow of 45 gpm and produced o head of 55 psi. Power input to the pump drive motor was 5.5 kw, with an estimated 3-hp input to the pump shaft and 0.75 hp absorbed in the seal. Fluoride chip leakage from the frozen end of the seal was checked for each 24-hy period of operation, ond measurements of 133 and 132 g were obtained, The seal configuration is shown in Fig. 2.4, Heat was supplied to the seal by electrical tubular heaters on the outside of the shaft housing and a cartridge heater inside the shaft, as shown. Cool- ing of the seal area was achieved by introducing compressed air between the anchor flange and the packing gland Hange and allowing the air to pass out through multiple holes around the nose of the gland slightly oblique to the shaft axis. This cooling was supplemented by two small centrifugal blowers directed against the rear of the packing gland and seal. The pump and the test loop of ll/é-in.-lPS pipe were constructed of 316 stainless steel. Flow was measured with a water-calibrated venturi, Pres- sures were measured with Moore Nullmatic trans- mitters, which have air balancing the loop pressure across a bellows, The transmitters contained triple-ply Inconel bellows and were operoted in a downward position so that the transmitters were flooded and free surfaces were eliminated, When pump operation begon, the frozen seal ap- parently established itself chead of the packing area, but as the test progressed, the frozen zone moved rearward until, after approximately 12 hr, the seal existed in the annulus between the shaft and the packing gland, However, after an 18-kw power surge, a small amount of heat was ploced on the packing and the shaft housing next to the im- peller, and no other power surges occurred until PERIOD ENDING JUNE 10, 1953 the shaft seized. At the time the shaft seizure occurred, all heat had again been off the seal for 30 minutes. The Carboloy sleeve was found to have a crack approximately 3/&4 in. wide that originated in the corner of a driving notch at the cool end of the sleeve. Subsequent heating and cooling tests on the sleeve-shaft assembly alone indicated thot a Carboloy sleeve cannot withstand any apprecicble axial temperature gradient, A stainless steel sleeve with Colmonoy hard surfacing will be tested in the next pump assembly, since the Carboloy sleeve does not appear satisfactory for this appli- cation, Allis-Chalmers Conned-Rotor Pump. A 5-gpm conned-rotor pump built by Allis-Chalmers is being assembled in a test loop and will be tested with NoK af temperctures approaching 1500°F., This pump has a hydrostatic bearing, pressurized by o smoll pump impeller at the rear of the pump, which bypasses some of the system fluid through the bearing. The motor windings are made of small- diometer copper tubing with fiberglas insulation. The windings are cooled by circulating a coolant through the tubes. The pump is completely in- strumented so that temperatures at all critical points can be determined during operation, Frozen-Lead-Secled Fluoride Pump. Sea! tests have demonstroted that a frozen-lead seal has smoother operating characteristics than a frozen- fluoride seal and that molten lead and molten flucrides do not intermix when in intimote contact with each other. Consequently, o centrifuga!l pump of ARE size has been redesigned as a frozen-lead- sealed pump for fluorides. In this redesign, the pump is mounted in o vertical position with the seal on the hottom side of the pump. A chamber equipped with immersion heaters located immedi- ately below the pump impeller housing contains both molten lead ond molten fluorides at 1000°F, Because of the difference in the densities of the two mixtures, the fluorides float on the lead, and a sharp fluoride—lead interface is maintained. A frozen-lead seal is accomplished in the bottom of this chamber in a liquid-cooled seal annulus that is approximately 1 in. long ond has approximately 0.030-in, radial clearance around the 21’2-in.-diu shaft, This redesign permits the modification of the frozen-fluoride-sealed pump to a lead-sealed pump that utilizes oll the parts of the original pump, except the seal. 15 91 UNCLASSIFIED DWG CSK-15993A 2¥%s 3 i 3 —— NOTE: ALL DIMENSIONS ARE IN INCHES far==1e }//4 oD TUBE 45° { ! 1 : ~COOLING DUCT { %4DIA SCREW 3 EACH SIDE Y32 THICK SHEET ?fi . 7 TUBULLAR HEATERS J-11 STUD AND NUTS—__ ¢ == 78°TYP&CAL§ boss 3ip IPS STAINLESS STEEL PIF’E/7 ‘ / Y NE [ P—% 3 -2'/32 SEAL GLAND o | T | L L o403 2 Scoows (I | S - §§E ‘:_fl_{"e CARTRIDGE HEATER Ry ""’""‘\3&;%1'1"/8 o ! ' REEEY % 5 ’ ! ! N bbb —_ | ! : }\\‘\\\X N M 1 ; N N i T T | o Tt N SESESES TINESESES | (it .f.{‘,“.‘"‘.'-‘w e by o I|II'||"|’ e i , ; L AN S THERMOCOUPLE HOLE LEAK CATCHER ——=! e ‘L% ° e e e e - TWIST LOCK JO:NT\E ! 4 a4 TUBE DOWN TO CATCH POT Fig. 2.4. Seal Gland for ARE Pump with Packed Frozen 3eal. 130d 3 SSF¥908d AT4ILIVND LIIF0¥d INY L.aboratory-Size Sump-Type Pump with Gas Seal. It was reported previously“) that the laboratory- size sump-type pump operated 965 hr with a stuffing-box seal ond 2300 hr, after redesign, with a rotary face seal of Graphitar Ne. 30 running on hardened tool steel, An additional 1000 hr of operation has been accumulated on another pump of the some design; however, Morganite MYIF, a silver-impregnated graphite with self-lubricating properties, was substituted for Grophitar No, 30 in the seal. Speeds of up to 3600 rpm were used with a flow rate of 5 gpm that produced a 55-psi head to pump the fluoride NaF-ZrF ,-UF, (50-46-4 mole %). Suction pressure varied from 10 to 30 psi, and the fluid femperature ranged from 1200 to 1500°F, There was regular dropwise oil lubrica- tion at a rate of 6 to 8 drops per 25 hours ot first, but later, lubrication was used only when power fluctuations occurred in the drive-motor input, Gas leakage during the test was negligible up to 24 hr prior to termination of the test, when a marked increase was noted. The graphite seal piece was found to be pitted and scored. The stationary tool steel foce had a work depression. Operation of the pump will be continued with continuous oil-mist lubrication on the seal exterior. ROTATING SHAFT AND VALVE STEM SEAL DEVELOPMENT W. B. McDonald W. R. Huntley W. C. Tunnell L.. A, Mann R. N. Mason D. R, Ward P. G. Smith J. M. Cisar ANP Division Graphite~-Packed Seals. A seal with layers of natural-flake grophite and MDS2 retained by Graphitar rings and machine turnings of an Ag-Mo$, compact was operated to seal against the fluoride NaF. ZrF -UF, (50-46-4 mole %) at 1300°F and 10 psi for 53 hr before termination of the test. Thermeo- couples were attached within the rotating shaft, Cperation was smooth; the power requirements were low; and consequently, the temperatures were low. It was possible to make 2-min stops without it being necessary to add heat, The indicated seal temperature was above the melting point of the fuel (515°C). However, there was no leakage when the shaft stopped and froze, which indicated that some fluoride was leaking into the sea! area und freezing, The test was terminated when continuad heating caused gross ligquid leckage. PERIOD ENDING JUNE 10, 1953 In an attempt to toke advantage of the apparent nonwetting characteristic of grarhite with fluoride, several dry runs were made with graphite as the packing material. In this series of tests, ths shaft rotated in o right-hand direction and o left- hand Y thread was machined on the shaft. This arrangement tended to pack the graphite ot the flucride end of the seal. An air cylinder wos con. nacted to the gland so that gland pressure could be controlled, and with this st up, it was determined that the packing pressure must be less than 30 psi to avoid excessive friction between the graphite and the rotating shaft, Another seal comprised of bronze wool as o retainer and artificial graphite plus 5% MoS, as the packing materia! sealed helium ot 10-psi pressure. This seal also satisfactorily sealed the fluoride NaF-ZrF UF, (50-46-4 mole %). There was some indication frem stop-start tests that an interface or meniscus wos established at the hot end of the seal. However, in the start-stop tests, performance was erratic. In some instances, heat was required for starting ofter stepping, and ot other times the shaft started freely, After more than 500 hr, operation was continuing ot 1200°F and 20-psig pressure, but there was the slight leakage of granulated material that is typical of frozen seals, Grooved-Shaft Packed-Frozen Seal Tests. Various packing matericls hove been tested in a sealing gland in which both the shaft and housing sides of the gland contain meshed annulor grooves to minimize loss of scal moterial. Three tests were made of seals with different packings but with similar geometry and eoperating conditions. The tests were operated with the fluoride NaF-ZrF -UF (50-46-4 mole %) at temperctures ronging from 1050 to 1350°F. The shaft diameter was ]3/16 i, and the seol glond length was 1% inches. la all these tests it was noticed that the first leakage of fuel from the seals was preceded by leckoge of the packing material, which indicates that there is need for improved retention of the packing material, Start-stop tests werz made saotisfactorily only after the addition of sufficient heat to compensate for frictional heat loss when the shaft was stopped, These thiee tests are summarized in Table 2.1, after the first test showed the arooved section of the shaft to be badly worn to the extent that there was o single wide groove covering about half the widih of the original grooved Examination 17 ANP PROJECT QUARTERLY PROGRESS REPORT TABLE 2.1. GROOVED-SHAFT PACKED-FROZEN SEAL TESTS GROOVE FLUID FLUID DURATION SHAF T DIMENSIONS PACKING PRESSURE | TEMPERATURE | OF TEST | RCASONFOR MATERIAL . . o TERMINATION (in.) (psi) (°F) (hr) Type 316 stain- | % by ko by % | Dixon's No. 2 grophite 7.5 1050 to 1225 92 Bearing failure less stee! plus Inconel braid impregnated with MoS, Stellite-coated 3/16 by 1/a by l’a Dixon's No, 2 graphite 10 1100 to 1200 53 Scheduled type 316 plus Inconel braid stainless steel impregnated with MoS, :}‘16 by ]/8 by 1/8 Dixon’s Na, 2 graphite 10 1200 1o 1350 148 Loss of seal plus superflake material during graphite start-stop test section and approximately as deep as the original grooves, Neither of the other seals showed as severe damage, although some wear occurred at the retainer end of the packing used in the second test, Bronze Wool and Mo3, Packed-Frozen Seal Tests, Several packed-frozen seals in which the packing gland contains bronze wool and a lubricant, such as MoS, or MoS, and graphite mixture, have been tested with the molten fluoride NaF-ZrF -UF, (50-46-4 mole %). A leakage rate on the order of 1 g/hr has been obtained with a ]3’]6-in.-diq shaft and 10 psi pressure across the seal. The leakage rate is not only a function of the pressure across the seal but also of the location of the actual point of seal within the gland, The attempt to adapt this seal geometry to a larger shaft (2]/2 in. in diameter) has not been entirely successful. Applications of successful small-diameter-shaft packings to larger shafts have usually been ac- companied by perturbations attributable to the increase in surface speed, higher friction, and increased area for heat conduction from the hot liquid being sealed. A summary of design and operating conditions for the four seals tested is given in Table 2.2, TABLE 2.2, SUMMARY OF BRONZE WOOL AND M052 PACKED-FROZEN SEAL TESTS Shaft material: Stellite-coated type 316 stainless steel SHAFT SEAL FLUID FLUID DURATION | bEasON FOR DIAMETER | DIMENSIONS PACKING PRESSURE | TEMPERATURE | OF TEST | 1 ERMINATION (in.) (ind) (psi) CF) (hr) 1:}16 Y by 5 Y=in, layers of bronze 10 1200 to 1275 78 Seal leak wool plus M052 z-in. layers of bronze 10 1200 to 1275 620 Still operating wool plus M052 plus graphite 2} % by 5 | Vi, layers of bronze 10 1200 to 1275 Seal leak wool plus Mc>S2 plus graphite ]{l-in. layers of bronze 10 Still operating wool plus graphite plus 5% M052 18 Although the first secl remained free ofter stop- start tests of up to 10-min duration, the seal lecked excessively when raised to a temperature above the melting point of the fluoride, and the test was terminated. The power requirement for this seal was approximately ¥ kw, with slight surges. Start-stop tests of up to 101/2 hr have been made in the second test, but heat was added to maoke up for friction heat loss. Although the seal is believed to have moved up inte the glond region, the test is operating smoothly with a shaft rpm of 1500, The initial leakage rate of 1 g/day increased to 1.5 g/day, and it was necessary to raise the seal temperature about 100°F above its operating tem- perature to enable the motor to restart the shoft seal without assistance. In the first test with the 2)-in.-dia shaft, the seal lecked liquid which appeared to be an MoS, mixture, Power fluctuations were greater than hitherto observed. During a 3-hr stop, the shaft remained free. Although the seal usedin the second test had to be repacked after excessive initial graphite leakage, operation is continuing smoothly with a low leakage rate and o somewhat variable, but usually smooth, power requirement, Frozen-Lead Seal Test. A series of tests has been conducted to determine the feasibility of sealing a high-temperature fluoride pump with a frozen-lead seal. These tests show this type of scal to be feasible, since it has been determined that molten lecd and molten fluorides do not inter- mix when in intimate contact with each other. Rather, the fluorides Hoat on top of the lead, and there is, apparently, a sharp fluoride-lead interface. The test equipment consists of a 13/1 -in.~dia type 316 stainless steel shaft powered by a 2-hp motor. The shoft was operated in avertical position at a speed of 1500 to 2500 rpm and the seal was below the shaft, A helium gas blanket was pro- vided for the cold end of the seal, and above the seal was a pot contuining lead at 1000°F, A series of five test runs was conducted for a total operating time of 550 hours. by extremely smooth operation and low power input during the first 30 10 100 hr of operation. Continued operation resulted in shaft seizure in the seal region, and the test was terminated. The conditions which appeared most likely to be contributing to seal foilure were oxidation of the lead in the frozen seal because of the air at the cold end of the seal, insufficient shaft clearance at the hot end of the lead sealing annulus, and impurities in the lead. These runs were characterized PERIOD ENDING JUNE 10, 1953 In the new seal test rig design, which eliminates the difficulties mentioned, the frozen seal region was shorteried to approximately }’2 in, in length and liquid coeling was substituted for gas cooling. High-purity lead was obtained and pretreated by bubbling hydrogen through it for approximately 100 hours. A frozen-lead seal has operated in the new rig for upproximately 500 hours. The power input to the shaft has been low, there have heen no power tluctuations, ond the lead leckage rate has been negligible. The shaft speed during this test was varied from 2000 to 4200 rpm, and performance over the entire speed range was excellent, V-Ring Seal. rings machined from Graphitar No. 14 and a hot- A seal consisting of alternate pressed compact of copper and 14% MoS, is being tested. The rings are assembled as shown in Fig. 2.5, A 13/16-in.-diu hardened shaft rotates against the Graphitar. A gas cylinder is used as the means of applying force to the gland to compress the rings, The seal was designed to operate at ap- proximately 1100°F. A dry run of this seal indicated that helium leak- age through the seal was almost zero with the seal area at a temperature of above 1000°F. Operation during the dry run consisted of rotating the shaft at approximately 1650 rpm with the seal area at a temperature of above 1050°F, Power dissipation in the seal is approximately 50 watts, With these conditions, an ottempt was made to force helium through the seal with pressures of up to 30 psig. Flow of helium to the seol was read on a gas rotnmeter. The unit was operated for one day with indicaoted flow and then disassembled for inspection. When reassembled, the unit was oper- ated for three days with an indicated flow of about 0.4 cm3/sec, so preparations were made fo intro- duce fluorides. Calculations indicate that for o gas leakage of about 2 em 3/sec of helium the cor- responding fluoride ledkage will be about 0.5 cm3/day, BeF, Seal Tests. A series of tests has been conducted with a stuffing-box type of seal pocked with BeF , to determine whether the high viscosity ere of BeF, at its softening temperature would give a satisfactory szal against the fluoride mixture NaF- ZrF4-UF4 (50-46-4 mole %) at high temperatures. A summary of the seal design and operating data is given in Table 2.3, In the first test, a !@-in.-déa shaft was sealed by a series of surrounding annular compartments con- taining BeF , powder. During operation of the shaft 19 ANP PROJECT QUARTERLY PROGRESS REPORT LS LY N A ‘I"‘ ‘fi;’?&z s £ et R i N ey i e ! AR\ P a) PR ac N e g A Z ey »'. . »‘l', e e P L-._._H 0.002 +0.0005 (COLD) 2 UNGCLASSIFIED DWG. ASK 16050A —0.0000 SHAFT SURFACE 1 CONTAINER WALL Fig. 2.5. Y-Ring Seal at 440 rpm, the power needed to overcome friction was too lowto read on a 10-kw full-scale wattmeter. Leakage through the seal wos zero. A packing of l/4 in. of graphite-impregnated asbestos was used at the cold end to contain the dry powder. The shaft was stopped and started many times for periods of up to 1 hr, and only motor power was used for restarting. After fermination of the test, the seal was cooled and sawed lengthwise. The fuel had penetrated only into the second compart- ment, that is, less than ?’/4 inch. None of the pack- ing material beyond the second compartment had melted. Chemical analysis showed 0.96 mole % BeF, in the first compartment and 50.2 mole % BeF , in the second compartment, In the second test, a 21/2-in.-dia shaft was sealed by o packing glond which was packed with copper- braid sleeves filled with BeF, powder. One turn of graphite-impregnated asbestos was used on the air end of the seal to contain the dry powder. This seal showed none of the characteristics of a BeF, seal, but operation and failure were similar to the experience with frozen-fluoride seals. 20 In the third test, a vertical ]3/]6-in.-dic1 shaft was packed with BeF, in annular compartments, as in the first test. Failure after ”’2 hr of operation at 880 rpm appeared to have been the result of the air end of the seal becoming too hot (870°F), The seal ran well until it failed. For the fourth test, a horizontal 2]/2-in.-diu shaft was assembled with annular compartments filled with BeF,. Improper design caused thermal dis- tortion and mechanical binding. The shaft was started 15 to 25 times, but each time it was stopped by metal to metal contact in the seal. In the fifth test, a 23/8-in‘-diu shaft, which simu- lated a horizontal ARE pump shaft, extended through a packing gland into a body of the fluoride fuel NaF-ZrF -UF, (50-46-4 mole %). Provision was made for selectively heating the end of the packing gland adjacent to the hot fluorides and for cooling the gland at its outer end. A metal sleeve around the shaft, sealed at the inner end of the gland, provided a narrow annulus ]/1 in. wide and obout 2 in. long for the fluorides to pass through and be cooled somewhat before they entered the PERIOD ENDING JUNE 10, 1953 TABLE 2.3. TESTS OF Ber-PACKED SEALS ON ROTATING SHAFTS OPERATED IN NnF—ZrF4-U Fy {50-46-4 mole %) SEAL PACKING CHAMBER PACKING - T PACKING MET SHAFT H N TEST MATERIAL Over-all Llflr:zgoridz Ne. of ANNULUS gfl‘fmf? spégn PRE[s"slfigE Dgrf '}qur REASON FOR No. COMPOSITION Length | oo " | Compart- | THICKNESS (in) ( . L TERMINATION o . 9 ; in, rpm) {psig) {hr) {mole %) (in) (in.} ments {in.} 1| 100% BeF, 4 2 5 0.25 Y 440 30 50 Scheduled 2 | 100% BeF, 1.25 0.B5 5 In coppar 2k 550 10 1 Senl leak braid sleeves 3 | 100% BeF, 3.1 2 6 0.25 % 880 30 1% | Seal l=ak 4 | 100% BeF, 4 2 7 0.375 2} 650 5 Off ond on | Mechonical worpage 5 | 75% BeF, plus 25% 5 2 1 0.4375 2% 380 ta 1460 1110 3 flucride mixture 6 Various percentoges 3.1 2z 7 0.25 Hfllfi 1350 to 3800 5t % 104 Failed during of BeF , plus ZrF stop-start tests 3 7 100% NaBeF 2.1 2 7 0.25 J 1%, 1350 5 J principal packing zone, Strands of copper rope were then placed in the seal annulus to a depth of ]/2 in. to serve as the principal deterrent to entry of the fluorides into the seal proper, which was formed by a 2-in. copper sleeve. The seal cavity was filled with 75 mole % BeF, ond 25 mole % fluoride fuel granules, and as a result there was 25 to 30% void space in the cavity. The seal was heaoted and instrumented so that the temperature gradient along the seal could be controlled. Pene- tration of the fuel proceeded until the resulting mixture in the seal cavity was too cold or too viscous to penetrate further. Shaft speed wos maintained at 1400 rpm for approximately 400 hr, and the test ended at 448 hr because of the failure of the external seal heaters and plugging of the coolant line. Power consumption in the seal be- cavse of friction {primarily at the ocuter end of the seal) was of the order of 400 watts. Many per- versities were encountered, as well as long periods (days) of stable operation with low friction in the seal. During the stable periods, fine powder leaked from the seal at a rate of 10 to 15 cm® per day. Shaft wear was light except under the outer copper, which was cold. Sufficient powdered material accumulated in this location to act as a grinding compound for the shaft material. Nearly all the major perversities of this seal could be attributed to the accumulation of fluoride powders under the external copper, augmented by oxidation of the fluorides at the cold end. The sixth test was run in the same equipment as was the third test, but the following alterations were made. (1) Two turns of ]/A-in.-dia copper cooling coil was wrapped (not soldered) arcund the top of the seal housing to carry cooling water, (2) The fluoride packing was divided into seven compartments by wusing type 316 stainless steel spacers and washers, and copper-rope packing was placed at the hot and cold ends of the seal. {3) The fluoride packing consisted of 8- to 20-mesh grains of BeF, plus ZrF, in the following weight per- centages of ZrF, in the layers from the hot to the cold end, respectively: 0, 15, 25, 40, 55, 70, 100. The unit was operated 104 hr at o sea! temperature of 1200 to 1250°F on the hot end and approximately 650°F on the cold end. The speed was varied from 880 to 3800 rpm. The helium blanket pressure was 5 to 10 psig. The main purpose of this test was to see whether ZrF, would prevent leakage (be- cause of its high melting point)., Although the leakage was not great, it waos fairly continuous, and the stop-start test was unsuccessful for a 10- min stop. Exomination of the shaft after the fourth test revealed that a high percentage of NaBeF, was possibly the most effective sealant. Therefore it was decided to operate a seal with NdeF3 packing in the same equipment as was used in tests 3 and 6. This seventh seal was similar to that used in the sixth test, except that all seven compartments 21 ANP PROJECT QUARTERLY PROGRESS REPORT were packed with NaBeF,. Gronules of 8 to 20 mesh (Tyler Standard) were used to pack the seal, and the copper cooling coil was soft soldered to the seal housing. The shaft speed was increased from 1500 rpm, that is, until the shaft peripheral speed was at ARE design point (15,3 fps), and was maintained there for the balance of the 160-hr test. Several successful stop-start tests were made with stop periods of up to 10 minutes. On most restarts, some assistance to the 2-hp motor was required; however, several restarts were made with no as- sistance. There was no detectable leakage of sealant, The friction drag was estimated at about 150 to 200 wotts. Temperatures along the seal remained constant. The temperature range was from approximately 1250°F at the hot end to ap- proximately 200°F at the cold end. The seal was cooled and separated from the unit and sectioned axially. Inspection revealed that the fuel had penetrated four of the seven compartments and that, in the remaining three compartments, the NaBeF3 had fused or sintered and changed color from almost white to a gray-brown. The only shaft scoring found was at points of contact with the stainless stee!l washers used as separators between com- partments. Packing Penetration Tests, Packing penetration tests have been continued in which graphite or graphite mixed with another material is used as the primary sealant, The use of graphite has been emphasized in these tests because it appears to be one of the few materials capable of preventing fluoride leakage. A test in which Baker Chemical Company pow- dered graphite was used to seal against the fluo- ride NuF-KF-LiF-UF4 (10.9-43.5-44.5-1.1 mole %) was terminated after 240 hr with no leakage, as reported previously.’3) Since other tests of similar seals used 1o seal against the fluoride NaF-ZrF - UF, (50-46-4 mole %) had leaked, it was thought that the sealing might be dependent either on the presence of LiF or on the absence of ZrF,. How- ever, during this quarter, another test was operated for 690 hr with the same type of graphite and zirconium-bearing fluoride mixture, and there was no leakage. This test was terminated at the end of the 690-hr period, because excessive oxidation occurs if the graphite tests are allowed to run for extended periods of time., The oxidation begins (5w, B. McDonald et al., ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 28. 22 at the opening in the bottom of the container and then extends upward toward the fuel region. Upon postrun examination, a large cavity was found in the graphite around the opening. If the test had continued, there would have been leckage due to the oxidation of the graphite rather than to pene- tration, All the graphites tested this quarter, except the Baker Chemical Company graphite, have been of the artificial type, which should have a very low amorphous carbon content, Tests of these groph- ites have had one thing in common: leakage began at a very slow rate and could only be detected by the small amount of corroded metal that fell from the furnace in which the packing containers were heated. The tests were continued and terminated only when fuel was detected in the receptable beneath the furnace. Because of the slow leakage and continuation of the tests, the outer surfaces of the containers were very badly corroded; they were covered with o heavy scale that could be removed easily. Results of these tests are summa- rized in Table 2.4, One combination of packing material being tested is a mixture of 80% artificial graphite powder (ob- tained from the Y-12 Carbon Shop) and 20% pow- dered BeFZ. The mixture was compressed ond then heated to 1150°F and further compressed. Heating and compressing were done three times, and the packing lost approximately 40% of its original volume, The test of this packing material is being run at the usual pressure, 30 psi, but the test temperature is 1150°F instead of the usual 1500°F, The test has been running for 600 hr with no feakage. INSTRUMENTATION P. W, Taylor D. R. Ward ANP Division Pressure Measurement. A large pressure trans- mitter, pattemed after the Moore Nullmatic instru- ment, () that utilizes a triple-ply Inconel bellows as the pressure-sensing element has been in inter- mittent service on a fluoride pump loop. The transmitter operates upside down to measure pump suction pressure (about 5 psi) and it is completely filled with 1100°F molten fluorides., The tempera- ture of the fuel in the loop is approximately 1200°F, The total operating time to dote is 360 (6)y, B, McDonald et al., ANP Qugr. Prog. Rep. Dec. 10, 1952, ORNL-1439, p. 30. PERIOD ENDING JUNE 10, 1953 TABLE 2.4, SUMMARY OF PACKING PENE TRATION TESTS PERFORMED DURING THIS QUARTER Pressure: 30 psi Temperature: 1500°F, except as noted TEST DURATION PACKING MATERIAL OF TEST REMARKS NO, (hr) n Stainless steel wool impregnaoted with MoS,, 0 L.eaked immediately 12 Boker Chemicol Company powdered graphite 6%0 Did not leak 13 MoS, in copper sheath ]/2 14 Graphite from Y-12 Carbon Shop 262 Scale indicated slight leakage 15 National Carbon Company graphite 2301 191 Scale detected at 72 hr 16 80% graphite from Y-12 Carbon Shop and 20% Be F2 Operating temperature 1150°F; had not leaked at 600 hr; test continuing 17 50% graphite from Y-12 Carbon Shep and 50% M052 40 Scale detected at 30 hr hours., The fuel froze in the transmitter during twe pump shutdown periods, and it was found that subsequent remelts and startups did not damage the transmitter. A small nickel transmitter operating completely filled with fluoride fuel on a static test (10 to 30 psi, one cycle every hour) at 1100°F failed after 2500 hr of continuous operation. Postoper- ative inspection revealed two small holes in one of the inner bends of the 0,005 in. (wall thickness) bellows. Throughout the 2500-hr period, operation was excellent, The combined range and zero shifi over a 1500-hr period wos less thon 1]/2% of full scale (30 psi). Two other type 316 stainless steel transmitters that are being operated on a similar static test are still functioning smoothly, One of these trons- mitters has logged 3400 hr, to date, in fuel ot 1100°F; the other has operated 3500 hr in lead at 700°F, The initial test of the single-diophragm force- balanced pressure transmitter is being run at 1050°F with no fuel. This instrument, constructed entirely of Inconel with a 0,014-in, Incone! X diaphragm, is being operated at a static pressure of 50 psi for 700 hr, with o cycle every 4 hr to 20 psi. Upon heating to 1050°F from room tempera- ture, there was an approximate 2% increase in output ot 50 psi. After 1200 hr of operation at 1050°F, the output has increased 3%. Response to pressure chonge is excellent, with ne visible delay. This fast response is obtained at the ex- pense of a very large, balancing, «ir flow. The instrument uses approximately 24 cth of air, NaK Leak Detection Methods, Since NaK may be used os the final leak-checking material for the ARE fluid circuit, a religble NaK leak detector for small leaks is desirable. Some preliminary testing has been done to determine the effect of NoK vapor on copper wool, With a pure helium atmesphere at temperatures from 250 to 650°F, the copper wool gave a substantial basic indication in phenol-red indicator solution (in H,O) after an exposure time of 1 hour. The addition of about 5% oxygen to the helium atmosphere decreased the strength of the indication but did not eliminate it entirely. 23 ANP PROJECT QUARTERLY PROGRESS REPORT 3. REFLECTOR-MODERATED CIRCULATING-FUEL REACTORS A. P. Fraas, ANP Division An extensive description of a proposed reflector- moderated circulating-fuel reactor suitable for powering an aircraft was presented in the previous | Subsequent studies of the full-scale power plant system have included an examination of reactor control, and a revision of calculations on the basis of new data from the critical assembly of this reactor configuration. The reflector-moder- report, (1 (DA, P. Fraas ot al., ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 41-84, ated reactor critical experiment is discussed in sec. 5, and shielding configurations applicable to this reactor are described in sec, 14, REACTOR CONTROL The control problems of the full-scale power plant are being examined. The fluid flow rates, transit times, and thermal capacities for the various ele- ments of the full-scale 200-megawatt sea-level aircraft power plant are presented in Table 3.1. These values should be considered as those for a TABLE 3.1. FULL-SCALE 200-MEGAWATT SEA-LEVEL AIRCRAFT POWER PLANT SYSTEM DATA MaK in each radiator core circuit NaK in radiator header drums NaK in radiator tubes NaK in 3.3-in.-ID 20-ft-long outlet line NaK in 3.3-in.-ID 20-ft-long inlet line MNaK in intermediate heat exchanger NaK flowing in enfire circuit NaK in header tank Total NaK in sach system Total NoK for sach engine Total NeK in complete power plant 81b 8 50 56 28 i 150 30 180 720 2880 Ib NaK flow rote through each radiator circuit (for a 400°F Az and a 12,500-kw heat rejection rate) MNaK transit time through system NaK transit time through intermediate heat exchanger NaX transit time through radiator tubes Fuel in core circuit (flowing) Fuel in header tank Fuel in core Fuel flow rate Fue!l cireuit transit time Thermal capacities of power plant Radiator cores (1b) Lines and pumps MNaK (flowing) Intermediate heat exchanger Total for NaK systems Fuel Total for power plant 24 2.5 #t3/sec or 120 Ib/sec 1.25 sec 0.23 sec 0.067 sec 6 ft3 1 £e3 1.2 #3 12.0 #13/sec 0.5 sec 600 Btu/°F 250 600 250 1700 Btu/°F 300 Bru/°F Ao r——— 2000 Btu/°F typical case; considerably less favorable values might be required with some designs (for example, NaK line lengths would be considerably greater for designs with engine nacelles in the wings), while on the other hand, more favorable values might be obtained for some of the other factors through design changes, with small penclties in turbojet engine weight or performance. It is important to note that the NaK transit time for the intermediate heat exchanger is 0.23 sec, which shows that even if a step change in NaK temperature could be effected in the secondary circuit, such a temperature change would be damped out over a period of more than 0.23 sec in the intermediate heat exchanger. It is interesting to note, too, that the combined thermal capacity of the fluoride and NaK circuits is 2000 Btu/sec; hence, even if the reacter were operating at 200 PERIOD ENDING JUNE 10, 1953 megawatts and heat removal in the engine radiators was stopped instantaneocusly (something that would be impossible to effect), the temperature in the reactor could not rise at a rate greater than 100°F per second. Thus the negafive temperature co- efficient of the reactor should be able to cope with a condition even so impossibly severe as this with a wide margin of safety, GEMNERAL DESIGN PARAMETERS Several parameters of importance in both control and shielding calculations have been compiled in Tables 3.2 and 3.3 for ¢ variety of reactor power outputs and core diameters. A constant power density of approximately 10,000 kw/ft3 in the intermediate heat exchanger was assumed for all cases. TABLE 3.2, FUEL VOLUME IN CORE AND FAST-NEUTRON ESCAPE FOR VARIOUS CORE DIAMETERS Constant Heat Exchanger Power Density: 10 I‘a.'iegcmrflfls/f’r3 Core digmeter, in, 14.3 Escape factor for neutrons* 0.40 Escape factor for 3-Mev gammas 0.47 Fuel volume in core, > 0.625 18 0 0 1 22.7 28.5 36 45.3 .35 0.32 0.29 0.26 0.23 .43 0.39 0.35 0.30 0.27 .25 2.5 5 10 20 *Fraction of neutrans escaping from reactor cors as fast neutrons. TABLE 3.3, FRACTION OF FUEL IN HEAT EXCHANGER FOR VARIOUS REACTOR POWERS AND CORE DIAMETERS Constant Heat Exchanger Power Density: 10 Megawafls/fi‘3 REACTOR FUEL IN HEAT FRACTION OF FLOWING FUEL IN HEAT EXCHANGER POWER EXCHANGER Core Diameter (in.) (megawatts) (1) 14.3 18 22.7 28.5 36 45.3 50 1.25 0.67 0.50 0.33 100 2.5 0.80 0.67 0.50 0.33 200 5 0.89 0.80 0.67 0.50 0.33 400 10 0.99 0.80 0.67 0.5 0.33 25 ANP PROJECT QUARTERLY PROGRESS REPORT 4. REACTOR PHYSICS W. K. Ergen, ANP Division A review of the status of the work on kinetics of the circulating-fuel reactor brings out three main points: (1) permanent introduction into the reactor of more than 1% of the total uranium investment in excess of the amount of uranium required to make the reactor critical, has to be avoided or quickly compensated for by control rods or the reactor will be damaged by the resulting permanent rise in the equilibrium temperature, (2) oscillations in reactor power will damp out quickly if the reactor can be regarded as rigid and the flow pattern remains constant, and {3) following a sudden introduction of excess reactivity, no serious over- swings in power and in temperature are expected. The possibility of an excessive pressure surge has not yet been ruled out, A number of results were obtained by further analysis of the critical experiment data for the ARE. The reactivity values of both the regulating and the control rods, as well as the axial and the radial cadmium fractions of the reactor, have been determined., The critical mass is, however, still uncertain. Since the earlier estimate(!! of 28 Ib was made, the coolant composition was changed to eliminate the highly absorbing potassium. This would allow a reduction in critical mass, but other revisions in the calculations, such as revision of the data on the absorption of Inconel, increase the critical mass, KINETICS OF A CIRCULATING-FUEL REACTOR W. K. Ergen, ANP Division If a disturbance is introduced into a reactor, the kinetic behavior of the reactor introduces three problems: (1) the behavior at long times after the disturbonce, (2) the question of whether oscil- lations initiated by the disturbance build up or are damped, and (3) the extent of the first overswing of reactor power, temperature, and pressure. Behavior ot Long Times After a Disturbance, [t is known that an excess reactivity Ak introduced into a reactor with a negative temperafure coef- ficient —a, sets up a new equilibrium temperature which exceeds the original equilibrium temperature by AT = Ak/a. Thus if a = 1074/°C and Ak = 1%, “)W. B. Cottrell (ed.), Reactor Program of the Air- craft Nuclear Propulsion Project, ORNL-1234, p. 45 (June 2, 1952). 26 AT = 100°C, which is barely tolerable. However, if o were smaller or Ak greater, the reactor would be in danger. A Ak of 1% con be caused, for instance, by introducing into the reactor 3 to 4% more uranium than is required for criticality, that is, less than 1% of the total uranium investment, The HRE, with its much larger temperature coef- ficient, can tolerate much more excess fissionable material in the core; but, on the other hand, the large concentration differences occurring in the HRE fuel ore mainly due to the reprocessing of the fuel circulation, and such reprocessing will not be attempted during the operation of the ARE. Though the possibility of bringing excess fission- able material into the reactor is more or less a specific difficulty of the circulating-fuel reactor, other aircroft reactor proposals have similar prob- lems, such as the possibility of the deformation of voids or the compression of supercritical low- density water, Oscillations, It has been shown that the circu- lation of the fuel causes damping of power oscil- lations of the reactor, This damping can now be demonstrated, even in cases in which the power distribution varies along the fuel path and perpen- dicularly to it and in cases in which different fuel particles have different fransit times through the reactor. Also, this domping is added to the damping afforded by whatever delayed neutrons are left, and there is no destructive interference between the damping by delayed neutrons and the damping by circulation, However, the present theory is as yet incomplete, because it dees not take info account possible mechanical deformation of the reactor or variation in the hydrodynamic flow of the fuel, Professor Cornelius Weygandt has obtained, on the ditferential analyzer of the University of Penn- sylvania, a few numerical solutions to the kinetic equation of a circulating-fuel reactor, and some of them are shown in Fig. 4.1, The solutions refer to the constant-power-distribution constant-tfransit- time case without delayed neutrons, [t was as- sumed that at times ¢ < 0, the transit time is equal to the period p of the oscillation, At t = 0, the transit time changes, while all other parameters of the equation remain constant, Figures 4,1a and 4,16 refer to a small oscillation; for the undamped UNCLASSIFIED DWG 19087 099 (a) 8=09p P/P, tCQ e *—-J!‘ e 120 | 104 e — —_—— t 0se”” | b B=11p PP ‘U e {30 PO e 6.3 0.0t (€} 9=09p P/Pq 10 mm—m——— A {30 85 e} oo | @ 8=11p P/P, 1< == {30 6.5\-“ ollel] (8) 8215 Fig. 4,1, Differential Analyzer Solutions to the Kinetic Equation of a Circulating-Fuel Reactor, PERIOD ENDING JUNE 10, 71953 condition at £ <0, the maximum and minimum power deviate by only 1% from Py, At ¢ = 0, the transit time in Fig. 4.10 abruptly gets 10% smaller than the period of the undamped oscillation, ond in Fig. 4.1b, 10% greater, The oscillation is practi- cally sinusoidal, and the damping is rather weak, though present, Figures 4.1c, 4,1d, and 4.1 use a different scale and refer to a more violent oscil- lation, In the undamped condition, the moximum of the oscillation is 6.5 times P and the minimum is only 0.01 Py; that is, the reactor almost shuts itself off, At ¢ = 0, the transit time jumps to 0,9, 1.1, aond 1,5 times the period of the undamped oscillation in Figs. 4.1c, 4.1d, and 4.le, respec- tively, Figures 4.l1c and 4.1d show that the damping is small initially because of the small deviation of the transit time from the period, As the amplifude decreases, the period decreases. In Fig. 4,1c, this means that the period more closely approaches the fransit time, and hence the damping decreases. In Fig. 4.1d, the opposite is the case. In Fig, 4.1e, the large deviation of the transit time from the pericd results in strong damping during the first cycle, Then the period happens to have decreased to one half the transit time, and the oscillation continues essentially undamped. It should be noted, however, that undaomped oscil- lations of this kind are unstable, because a small disturbance which decreases the amplitude and hence the period would ‘‘detune’’ the system and cause further domping. More important, the possis bility of undamped oscillations is largely a conse- quence of the simplifying assumptions used in the computation, in porticular, the assumptions of constant transit time and of constant power and flux distribution, Under more general conditions, the undamped oscillations are much less likely to occur, and with proper design, their occurrence can be prevented, The First Overswing Following a Disturbance, Sufficient work has been done to show that under conditions envisaged for the aircroft reactor, even a sudden application of considerable excess reac- tivity will only result in a moderote temperature overswing. However, since this temperature in- crease occurs in a short time, it is not casy to show, and it has not yet been shown, that the fuel expansion occurs without accompanying large pres- sure surges, 27 ANP PROJECT QUARTERLY PROGRESS REPORT STATICS OF THE CIRCULATING-FUEL ARE CRITICAL EXPERIMENT C. B. Mills, ANP Division The analysis of the data of the ARE Critical Experiment is not yet complete, However, several interesting results have been obtained. 1. The multiplication constant of the reactor with a simple loading pattern of about 61 tubes in the 70 fuel-coolant tube holes(? in the core was recalculoted, and aoll the correction factors now known were used. The result was 1,003, as com- pared with the experimental value of 1,000, As reported previously,(s) the assembly also went critical with a higher U235 concentration and about 42 fuel tubes. Recalculation of this arrangement gave o multiplication constant of 1,009, These slight errors are consistent with a slightly in- correct value for neuvtron chamneling from the un- filled fuel-tube holes., No large errors have as yet been found, 2. The danger coefficient of Inconel, the struc- tural material to be used in the hot ARE, was inconsistent with the assumed dbsorption cross section of the material. An error was found in the basic data which increased the values of the thermal macroscopic absorption cross section from 0.310 em™! to 0.356 cm™!, The reactivity coef- ficient for Inconel thus changed from 0,170 to 0.231, and the predicted critical mass of the ARE was increased by about 4 Ib, Previously,!V a value of 28 |Ib was predicted as the uranium re- quired in the ARE core. However, since that prediction was made, the fuel and reflector coolant were changed in composition and the highly ab- sorbing potassium was eliminated. The Inconel correction and the elimination of the potassium almost compensate for each other, but because of the possibility of other corrections, the critical mass is still somewhat in doubt, 3. The computed and the experimental cadmium fractions, both radial and axial, for the ARE (CA-8) are given in Fig, 4.2, The discrepancy at the core boundary is largely due to the assumption of core homogeneity, with the boundary specified by the mean first neutron collision distribution. (2)D. Scott and C, B, Mills, ANP Quar. Prog. Rep, Sept, 10, 1952, ORNL-1375, p. 43. B3)p, seott, C. B. Mills, J. F. Ellis, D, V. P. Williams, ANP Quar. Prog, Rep, Dec. 10, 1952, ORNL-1439, p. 54. 28 4, The assumed cos? @ for control rod sensi- tivity and the cos? @ 40 for control rod value are compared to the experimental values in Fig. 4.3, DWG. 12897 REACTOR RADIUS (in) 25 0 25 1 : FUEL-TUBE SPACING, 3}, in, &3 TUBE DIAMETER 1Y, in. EXPERIMENTAL NS h RADIAL \' \ | / } _ z > . £ £ O = | = = | 2 & 5 THEORETICAL —_ Z w o0—% . - o0 4 [ae] @ z | ° 2 | g = o b a AXIAL o g & - e 3 y_..fl - CORE - &3 '&J o Q hd I 7 INCONEL” 7~ 7 7 7 7] 1 25 1 0 CADMIUM FRACTION — Fig. 4.2, Rodicl and Axial Cadmium Fraction in the ARE Critical Experiment, A DWG. 19898 o A & = o S & °© o o H OO ®© 9 \?Cu ¥ o n o el END OF REACTOR ARBITRARY SCALE -0.4 -0.6 o EXPERIMENTAL COMPUTED -0.8 -1.0 5C 40 30 20 10 0O RADIUS, LENGTH FROM REACTOR CENTER (cm) 10 20 30 40 5C 60 70 Fig. 4.3. Control Rod Values for the ARE Crit- ical Experiment, PERIOD ENDING JUNE 10, 1953 00 ow'e'.||9“g|33 5. Computed ond experimental values for the cadmium fraction vs, computed values for per cent thermal fission are shown in Fig. 4.4, ~ 6. The three safety rods are worth 13% in k_ /e = This is twice the value required to reduce tf;e 9 ) o . 3 operating temperature 1000°F, The preliminary B 50— N — guess for the value of the rods had been 15%,(4) L ‘_XJ e ,,-_._.,,,,,,,4___L\_* = COMPUTED | CADMIUM FRACTION £ THERMAL 1 ca CL’LATEDt EXPERIMENTA (4) * FISSIONS (%) “A-CY XPERIMENTAL W, B. Coitrell (ed.}), Reactor Program of the Aifr- T sar | o329 | craft Nuclear Propulsion Project, ORNL-1234, p. 49 58 0 0345 | 036 (June 2, 1952). 59.6 0353 | 0.355T0 0.39 90.5 0.662 | 066 0 0.50 100 CADMIUM FRACTION FOR THE BARE ARE,C.A-8 Fig. 4,4, Per Cent Therma! Fissions as a Fune- tion of the Cadmium Fraction for the Bare ARE, 29 ANP PROJECT QUARTERLY PROGRESS REPORT 5. REFLECTOR-MODERATED REACTOR CRITICAL EXPERIMENTS A, D. Callihan D. V. P. Williams R. C. Keen J. J. Lynn Physics Division D. Scott C. B. Mills ANP Division The second critical assembly of the reflector- moderated circulating-fuel reactor was described previously. (1) The fuel, a mixture of Zr0,, NaF, and C (66-24-10 wt %) with sufficient enriched UF , added to make the U235 density 0.2 g/cm3, was placed around a central ‘‘island’’ of beryllium and was, in turn, surrounded by a beryllium reflector. The critical mass of this system was 7.7 kg of U235_ POWER DISTRIBUTION The power distribution and the method used in obtaining it in a direction parallel to the reactor axis were given in a previous report.?) The power BERYLLIUM REFLECTCR FuU BERYLLIUM (SLAND RELATIVE FISSION FRAGMENT ACTVITY o -5 G 5 distribution has been remeasured in the same location, with the fission rate reduced in the ends of the fuel coolant channels by shielding the chan- nels on four sides with boral sheets 0.25 by 2.87 by 9 inches. (Boral is a mixture of aluminum powder and boron carbide sandwiched between aluminum sheets; the boron thickness is 0.25 g/ecm?.) Figure 5.1 gives the two power distri- butions normalized at a point 5.25 in. from the mid-plane, “)D. V. P. Williams et al,, ANP Quar. Prog. Rep. Mar. 10, 7953, ORNL-1515, p. 53. pid., p. 58. DWG. 19900 0.25-in.- THICK BORAL i 0.002-in.- THICK URANIUM DISK PLUS ALUMINUM CATCHER SOIL DISTANCE FROM REACTOR MID-PLANE (in)} Fig. 5.1. Power Distribution Parallel to Axis of the Reactor. 30 LEAKAGE FLUX The fast neutron leakage through the end of the reactor, both with and without the boral around the fuel coolant channels, was measured by a fission chamber placed outside the end reflector and perpendicular to the axis of the reactor. The chamber was lined with almost pure U238 gnd covered with boral. neutron leakage through the side of the reactor, the chamber was placed in the mid-plane outside For measurements of the fast- the reflector and parallel to the axis of the reactor. The fast-neutron leakage flux through the end was 30 times the corresponding flux through the sides without the boral around the fuel-coolant channels, and only about six times the flux through the sides with the boral shield in place. Of course, this reduction in fast-neutron leckage couses some loss in reactivity; the introduction of the boral on one end reduces the keff by 2%. CONTROL ROD MEASUREMENTS The loss in reactivity as a function of the po- sition of g %’lé-in. stainless steel tube filled with boron when inserted along the axis into the island of the reactor is shown in Fig. 5.2. The boron density was about 0,14 g per inch of the 0.1375 in.-1D tube, or about 50% of the theoretical density. A solid stainless steel rod of the same size in- serted along the axis to the mid-plane gave o loss in reactivity of 5 cents, DANGER COEFFICIENT MEASUREMENTS A large number of danger coefficient measure- ments were made with the test material placed in both the fuel and reflector regions. The results DG 9t 1 REACTIVITY LOSS (cents; DISTANGE FROM REACTOR MID-PiLANE (in) Fig. 5,2. Change in Reactivity Produced by Boron-Filled Rod us a Function of Its Position Along Reactor Axis, PERIOD ENDING JUNE 10, 1953 are reported in Table 5,1. The last two columns list the change in reactivity which occurred when an empty test space was filled with the sample. The samples of Li7, KF, NoF, Zr, Cr, ond RbF were in aluminum containers 5/1‘5 by 2]/2 by 9 in., which had ]/32—in.-thick walls. These containers were assumed to have no significant effect on the reactivity. The sample of sodium was contained in The reactivity value of the can when empty was compared with that obtained with the sample in place to obtain the reactivity value for sodium. The other materials required no containers. The long dimensions of the samples were parallel to the reactor axis, and the samples were symmetrically locoted about the mid-plane. When in the reflector, the samples were placed about 12 in. from the axis. In this location the total neutron flux was greatest, (3) The reactivity change incurred by the substitution of a sample of D,0 for beryllium was measured at three locations in the assembly, The sample was 99.2% D,0 and weighed 2.65 kg. It wos placed at the assembly mid-plane in the island adjacent to the fuel, in the reflector adjacent to the fuel, and in the reflector 3 in. from the fuel layer, In each of the first two locations the gain in reactivity resulting from the substitutions was 12 cents; in the latter position the reactivity was decreased 24 cents, In the present design of a three-region reflector- moderated reactor, the fuel-coolant flows in an Inconel shell.) To evaluate the loss in reactivity caused by the insertion of such a poison around the fuel region, Y-in.-thick stainless stee! sheets 9 in. long were placed around one-quarter of the fuel annclus. The section covered was symmetri- cal cbout the mid-plane. The addition of this stainless steelresulted in o 1.4% loss in reactivity, which was compensated for by the oddition of 468 g of U235 distributed in the fuel region, a thin stainless stee!l con. CORRELATION WITH THEORY Theoretical extrapolation of the various measure- ments, as well as of the danger coefficient meas- urements presented in Table 5,1, indicates that the distribution of 1 vol % of Incone! throughout the moderator and the addition of a !g-in.-thick fuel- coolant container would increase the critical mass Lid., p. 60, Fig. 4.22. ipid., p. 62, Fig. 4.25. 31 ANP PROJECT QUARTERLY PROGRESS REPORT TABLE 5.1. DANGER COEFFICIENT MEASUREMENTS S AMPLE SAMPLE WEIGHT SAMPLE REACTIVITY CHANGE (cents/g) (9) SIZE* In Fuel In Reflector Li” 152.4 A 0.0000 ~0.0415 KF 198.0 A ~0.0162 ~0.0578 NaF 97.2 A —0.0058 --0.0354 Zr 287.7 A +0.0150 —0.0498 Cr 249.4 B ~0.0237 ~0.0680 RbE 156.3 B +0.0118 +0.0026 Inconel 1956.4 A ~0.0276 ~0.0448 Ni 2061.0 A ~0.0274 —0.0437 P 2544.8 C ~0.0008 ~0.0007 Pb 4592.7 D ~0.0012 Stainless Steel 1990.4 C ~0.0243 ~0.0385 Stainless Steel 3195.0 D ~0.0233 Stainless Steel 995.2 E ~0.0509 Be 753.6 D +0.0090 Na 366.9 D ~0.0190 Bi 3931.8 D +0.0006 Graphite 376.2 A +0.0108 +0.0043 Graphite (Density, 1.72 g/em®) 700.1 D +0.0064 Graphite (Density, 2.0 g/cm>3) 825.6 D +0.0065 BeO 375.5 F +0.0055 A = 5/‘]6I:;y5by9in. B = C = 32 %16 bY 2}2 by 9 in. 1/4by 5374 by 9 in. 1 by 27/8 by 85/8 in, ]/B by 5% by 2 in, 7/8 by 27/8‘by 5in. to about 40 |b, compared with 17 Ib for the critical experiment not containing the Inconel. A critical mass of only 30 Ib results if the pressure shell is only ys in. thick and the Inconel structure in the reflector decreases rapidly from 1% at the core- reflector interface to 0.04% ot the outside of the reflector, with the average being 0.2%. This distri- bution is probably o good approximation to the distribution in on actual reactor. The cadmium fraction for one point at the core- reflector interface of the reactor was previously reported(3) to be significantly different from the computed value. The discrepancy has now been resolved by shifting the radic!l coordinate in the computation to properly take into account the void which existed ot the boundary. Figure 5.3 shows the revised computed curve; some difference still exists between the shapes of the experimental and the theoretical curves. These differences may be due to (1} the deviation of the experimental, flat geometry from the spherical geometry assumed in the computation, (2) boundary effects not quite accurately described by simple diffusion theory, or (3) the radial orientation of the foils in the experi- ment. The radial orientation makes locating the effective centers of the foils difficult in the rapidly varying flux. PERIOD ENDING JUNE 10, 1953 A— DWG. 19902 0.7 fw-— FUEL COOLANT~— i - Be REFLECTOR -] 3 ' FIT AT 19, in. 0.6 =4 o "— ) a [ong u. = 2 = 0 < [ QA [ 4 ® EXPERIMENTAL o CALCULATED 0 — e ) 3 2 1 0 1 2 3 INCHES - Fig. 5.3. Computed vs, Experimental Indium Activation and Coadmium Fractions at the Fuel. Coolant-tg-Reflector Interface of the Reflector- Moderated Reactor. 33 INTRODUCTION AND SUMMARY The research on high-temperature liquids has been primarily concerned with the detailed study of the complicated NaF-ZrF -UF, system, the production of the fluorides required by the ARE, and, to a lesser extent, with other halide systems containing UF, or UCI, (sec. 6). The pseudo- eutectic that has been shown to exist at or neor the composition 65-15-20 mole % is now designated as the ARE fuel concentrate. Although its melting point (515°C) is comparable to that of the previ- ously considered 50-25-25 mole % mixture, the new mixture is the more desirable, since it does not segregate on cooling and, in addition, has o lower vapor pressure. Use of this new fuel- concentrate in conjunction with the NquF5 carrier results in a fuel composition of 53-43-4 mole % of NaF-ZrF ~UF . Construction of the facilities for production of these fluorides in the quantities required by the ARE is complete., These facilities provide for the necessary purification by hydrogen- ation-hydrofluorination to effect the removal of such corrosive contaminants as NiF,, FeF, and CrFa. The recent corrosion studies have been almost entirely devoted to the general problem of the corrosion of Incenel by fluorides (sec. 7). Other studies include secondary systems associated with the ARE, such as fluorides on Stellites ond be- ryllium oxide in sodium, as well as the longer range problem of mass transfer in lead, The cor- rosion of Inconel by the fluoride fuel NaF-ZrF .- UF4 (50-46-4 mole %) haos been reduced to around 5 mils in 500 hr, and the corrosion mechanism is now fairly well established. The corrosion rate decreases with time ond is primarily a function of the fluoride contaminants NiF and FeF,, which react with the chromium in the mefal. At present, the attendant formation of UF,; limits the extent to which the NiF ond FeF, contaminants may be removed from the fuel. With regard to beryllium oxide in sodium, it has been fairly well established that the corrosion mechanism.is in reality only the mechanical erosion of the beryllium oxide surface, In convection loop tests of various metals in lead, only molybdenum and columbium did not show any mass transfer, while type 446 stainless steel showed a slight amount, Other metals, including Inconel, Armco iron, ond types 304 and 347 stain- less steel, showed extensive mass transfer, Welding, brazing, creep-rupture tests, develop- ment of cermets, and the fabrication of various pieces of equipment constitute the metallurgical research program (sec. 8). Manual inert-arc welds have been successtully applied to tube-to-header heat exchanger welds which did not lend them- selves to the semiautomatic cone-arc technique. The flowability of the Nicrobraz brazing alloy used to assemble the liquid-to-air radiator section is adversely affected by the presence of nitrogen, The use of high-conductivity oxidation-resistant fins for these radiators is obviously desirable, Coating copper with chromium, nickel, Inconel, and stainless steel has been undertaken. Severdl copper alloys fin material. Creep and stresse-rupture tests in air, argon, hy- drogen, and fluorides indicate that Inconel ond nickel are more sensitive to environmental changes than are the austenitic stainless steels, The creep rate of Inconel in tluorides is comparoble to that in argon, Of the several methods for the fabrication of spherical solid fuel elements that were investigated, forcing the molten alloy through a small orifice produced the most uniform and least oxidized particles. MNumerous compacts with 14 vol % MoS5, have been hot pressed for use as pump seals, show promise as While the physical properties of molten fluorides and hydroxides are being measured at temperatures of up to 1000°C, the heat transfer characteristics of these liquids are being studied in vorious systems (sec. 9). Among the physical property measurements pertinent to the ARE are (1) the viscosity of the recently designoted ARE fuel NaF-ZrF ,-UF, (53-43-4 mole %), which decreases from 13.5 cp at 620°C to 8.8 ¢p of 757°C, (2) the density of this fluoride, which is 3.5 g/cm’® at 653°C, and (3) its vapor pressure, which ronges from 4.5 mm Hg at 790°C to 39 mm Hg at 958°C. The experimental velocity profiles which have been determined in a convection system differ significantly from the parabolic characteristic of isothermal laminar flow, Forced-convection heat transfer for the NaF-KF-LiF eutectic—inconel system in 0.175.in.-ID tubes is one half that for a comparable fluoride-nickel system, probably be- cause of a corrosion layer found on the Inconel surface. A mathematical analysis has been made of the effectiveness of a reactor coolant with 37 ANP PROJECT QUARTERLY PROGRESS REPORT regard to duct dimensions ond spacing, amount of heat to be removed, coolont temperature rise, and coolant physical properties, The radiation damage studies are primarily con- cerned with the evaluation of the results of the fluoride-Inconel somples irradiated in the LITR and MTR, as well as the effect of radiation on creep and thermal conductivity, An inpile fluoride loop is being constructed (sec. 10). The irradiated fluoride-containing Inconel capsules show an inter- gronular attack of up to 3 mils, which does not occur in control samples, Chemical analyses of these fluoride mixiures indicated uneven distri- bution of uranium. Cantilever-type creep measure- ments made on Inconel in a helium atmosphere in the LITR indicated no serious change in creep properties as a result of irradiation; also, there was no significant change in the thermal conduc- tivity of an Inconel specimen irradiated in the MTR. The analytical studies of reactor materials in- clude the development of chemical, petrographic, and x-ray anclyses of fuel compositions and/or identification of fuel corrosion products (sec. 11). A procedure, which uses BrF as a reagent, has been developed for the determination of oxygen in fluorides. The determination of zirconium in fuels may be made volumetrically by the use of p- 38 bromomandelic acid as a reagent, or colorimetri- cally with reference to zirconium alizarin sulfonate, ““Tiron'" is shown to be a suitable reagent for the determination of uranium, The concentrations of UF3 and Zr° in Nc:l"'-Zrl:4--UF:4 (50-46-4 mole %) have been determined by the evolution of hydrogen, upon treatment of the mixture with hydrochloric and hydroflueric acids, Optical data, from petrographic examination, are reported for many new fluoride compounds. Studies of reprocessing of fluoride fuels indicate that reprocessing may be accomplished readily in existing facilities (sec. 12). Although the NaF- ZrF4-UF4 (50-46-4 mole %) fuel was used in these studies, the information obtained is generally applicable to other compositions within this system. After an aqueous solution of the fluoride that is suitable for solvent extraction is obtained by reaction with dilute aluminum nitrafe—nitric acid solution under reflux conditions, the uranium can be recovered and decontaminated by solvent extraction in batch countercurrent runs with losses of less than 0.01%. Processing of the fuel in this manner is possible in the present ORNL Metdl Recovery Plant without major equipment additions, olthough the corrosion of existing equipment will be severe and some precautions will have to be taken to avoid criticality. PERIOD ENDING JUNE 10, 1953 6. CHEMISTRY OF HIGH-TEMPERATURE LIQUIDS W. R. Grimes, Materials Chemistry Division An effort is being made to obtain a better under- standing of the solid-phase relationships in the very complicated NaF-ZirF -UF, system because of its importance to the aircraft reactor program. Some additional thermal data were obtained, but the greater number of samples prepared was for solid-phase studies. A pseudoeutectic has been shown to exist at or near the composition 65-15-20 mole %, and it has been fentatively designafed as the ARE fuel concentrate rather than the previously considered 50-25-25 mole % mixture. Data on a few compositions in the RbF-Z¢F -UF, system indicate that this system may possible be of interest as « fuel material, Because of the current interest in BeF, pump seals, the change in the melting point of the 50 mole % NaF—46 mole % ZeF -4 mole % UF, mixture when various amounts of BeF, are odded was restudied. Studies of the phose diagroms of mixtures of metal and uranium halides, other than those con- taining UF,, have continved. Several binary fluoride systems containing UF, have been ex- amined, but no satisfactory phase diagrams have yet been obtained. Lack of sufficiently pure UCH, has hampered the thermal analysis of chloride system; however, low-melting-point compounds are indicated in several mixtures. Modifications of the hydrogenation-hydrofluori- nation process previously described for the fuel preparation have been studied on laboratory and pilot scales. There has been considerable study of the kinetics of removal of HF from the melts by stripping with helium and hydrogen, and efforts have been made to evaluate the kinetics of re- duction of NiF,, FeF, FeF, CiF; and UF, by hydrogen in these melts. [t appears that stripping of the HF by use of hydrogen rather than helium is to be preferred, since more complete removal of the structural metal fluorides will be effected at the same time, Construction of the equipment for production of 3000 Ib of ARE fuel solvent (NaZrF,} is compleate, and preliminary testing of the equipment has been satisfactory. Production of this material is sched- uled for June and early July. X-ray diffraction studies have been made on two fluoride mixtures to establish the compounds and phases found in the respective melts. The mixtures studied were NaF-UF, (75-25 mole %) and Nak- ZrF , (50-50 mole %). Several ternary phase regions were also examined, and the x-ray data, together with petregraphic observations for compounds found in these regicns, are presented. FUEL MIXTURES CONTAINING UF4 L. M. Bratcher J. Truitt C. J, Barton Materials Chemisiry Division NoF.ZrF -UF . The discovery that UF sepo- rates from the mixture NaF-ZrF4»UF4 (50.25-25 mole %) when cooled without efficient stirring made this mixture undesirable as a fuel concentrate for the addition of enriched uranium to the ARE, The need for a nonsegregating composition with a high uranium concentration was also demonstrated in radiation damage studies. X-ray diffraction data for compesitions on the NaUF ,-Na,ZrF, line indicated that these com- pounds may form the pseudobinary system for which thermal data are shown in Fig. &.1. The data seem to indicate o minimum composition at about 45 mole % NGUFS (22.5 mole % UF4). How- ever, petrographic examination of the 40 and 45 mele % compositions showed that both compositions had the appearance of o eutectic (aggregates of very fine-grained crystals), and the results of settling tests (cf., “"Analytical Studies of Reactor Materials,”” sec. 11) seem to indicate that the lower concentration of NaUF, is closer to the sutectic. This composition (NaF-ZrF -UF ,, 85- 15.20 mole %} is, af present, tentatively designated as the enriched-uranium composition for ARE startup. It seemed undesirable to change the composition of the uranium-free fused salt (NQZrFS) that will be circulated in the reactor before the addition of uranium. Mixing the new uranium-rich composition with NaZrF o in the proper proportions results in a fuel with the composition NaF-ZrF -UF , (53-43-4 mole %). This mixture shows a melting point of 220°C, which is approximately the same as that of the 50-46-4 mole % fuel composition previously considered. Vapor pressure data for the new fuel composition are reported in sec. 9, ‘"Heat Transfer and Physical Properties.’’ 39 ANP PROJECT QUARTERLY PROGRESS REPORT B50 prrereeeeens 800 | 750 |- - 700 TEMPERATURE (°C) 650 ~_DWG. 19903 600 NaUFg {moaole %) Fig. 6.1. The System Na,ZrF_-NaUF . RbF-ZrF4-UF4. The low melting point of RbZrF,, recently redetermined to be 405 * 5°C, makes this material seem gttractive as a possible base for a fuel preporation. Mixtures containing various proportions of RbZrF . and RbUF . (melting point, 710 + 10°C) were prepared, and data on thermal effects were obtained from cooling curves (Table 6.1). Thermal data for the 92 mole % RbZrF ~8 mole % RbUFS mixture were checked on samples pre- pared in sealed capsules. This material showed a halt only at 402°C, which is the lowest melting point yet recorded for o nonberyllium fluoride mixture containing as much as 4 mole % UF ;. For mixtures with higher uranium concentrations, the rather sharp rise in the melting point probably indicates the formation of a higher melting complex, such as Rb3UF7 (melting point, 990°C). Further work on this system will be undertaken as the supply of RbF permits NaF-ZrF ,-BeF,-UF,. Data on mixtures resulting from the addition of BeF, to the binary compo- sition with NaF.-ZrF UF . (50-46-4 mole %) were 40 TABLE 6.1. THERMAL EFFECTS FROM COOLING CURVES FOR MIXTURES OF RbZrFs AND RLUF, COMPOSITION THERMAL EFFECTS (mole % RbUF ) (°C) 4 400, 385 (halt) 8 400 (hal1) 12 560, 400 (halt) 16 505, 400 (halt) 20 575, 400 (halt) given in a previous progress report;(? however, o more detailed study of such mixtures was carried out during this quarter, The thermal effects noted are given in Table 6,2, At the higher BeF, con- centrations, the thermal effects were not well marked on the cooling curves. The dato indicate, My, s, Coleman and W. C. Whitley, ANP Quar. Prog. Rep. Sept, 10, 1952, ORNL-1375, p. 78. however, that liquid probably exists in mixtures with 60 or 70 mole % BeF, at as low as 430°C, The uranium concentration in these mixtures is very low, and the data are of interest only in TABLE 6.2. THERMAL EFFECTS FOR MIXTURES OF NaF-ZrF4-UF4 {50-46-4 moie %) AND BeF2 CONCENTRATION | THERMAL EFFECTS FROM OF BeF, COOLING CURVES {mole %) (°C) 1 510 (halt), 490 2 520 (hal?) 5 493, 445 10 476 (halt), 435, 365 20 465 (halt) 50 602, 543, 444 {halt) 60 505, 430 70 620, 430 80 880(?), 415 PERIOD ENDING JUNE 10, 1953 connection with the possible use of BeF, in pump seals, FLUORIDE MIXTURES CONTAINING UF3 V.S, Coleman C. J. Barfon Materials Chemistry Division T. N. McVay, Consultant, Metallurgy Division Thermal analyses of binary mixtures of NeF- UF, and KF.UF,, supplemented by examination of the solid phases, have not, to date, provided satis- factory information for the preparation of phase diagrams of these systems, A number of compo- sitions in other systems were prepared to aid in the identification of the solid phases occurring in reduced zirconium-base fuels. The complex com- pound 27rF -UF,, mentioned in a previous re- port,® has been definitely identified. Table 6.3 shows the thermal data obtained from cooling curves and the optical data obtained from petro- graphic examinations for severa!l preparations in the UF -UF ;-ZrF , system, (2w, C. Whitlay, V. S. Coleman, and C. J. Barton, ANP Quor. Prog. Rep, Dec. 10, 7952, ORNL -1439, p. 109. TABLE 6.3. THERMAL AND OPTICAL DATA FOR UF:;«Z:-FJ1 AND UF -ZrF, MIXTURES OPTICAL DATA MOLECULAR COMPOSITION | THERMAL EFFECTS (°C) UF 4-27:F 715, 631 UFS-ZrF4 UF UF,-2Z¢F 730 UF -UF -4 ZF, 742 UF -3UF,-8ZrF, 712 UF -UF Z/F, 735 AUF -UF .. Z:F, 847 3UF -8UF 5 ZeF , 784, 773, 600 15% UF ,~35% UF ;-50% ZrF 725, 590 Orange-red compound; RI* = 1,556 Yellow and brown crystals; Rl = 1.616; free UF, Predominantly brownish, some olive drab Predominontly olive drab; biaxial +, alpha = 1.556, gamffld = ].568, 2V = 80 deg Mixture of olive-drab and red-orange striated crystals Brown isotropic phase; elpha = 1.560, gamma = 1.568; same UF3 Predominantly UF4 surrounding brown-orange crystals of Rl = 1.56 %0 1.57 Homogeneous clive-drab phase different from the 1:1:4 preparation; uniaxial +, plecchroic, algha =1.57Q, gamma = 1.578 Predominantly olive drab, trace of UF 4 and brown- orange crystals *Refractive index. 41 ANP PROJECT QUARTERLY PROGRESS REPORT CHLORIDE MIXTURES CONTAINING UCl4 R. J. Sheil C. J. Barton Materials Chemistry Division A considerable part of the work with uranium tetrachloride during the past quarter was devoted to efforts to obtain pure material. Neither sublis mation of product from hexachloropropylene chloris notion of UO; nor from chlorination of UH, pro- duced material of the desired purity., Although the accepted volue for the melting point of UCIH, is 590°C, the highest melting point obtained to date from these preparations was 567°C. Plans have been made to procure pure UCl, and pure UCI, prepared by vapor-phase chlorination of UO; with ccl, (and reduction of the UCld with Hz) from the Y-12 plant, Although one significant experiment in the KCi- UCl, system is reported below, little study of the NaCl-UCl, aond KCI-UCl, systems waos attempted during the quarter, This work will be resumed for final exploratory checking when sufficiently pure UCI, is available. Studies of the LiCI-UCI,, TICI-UCH,, and LiCI-KCI-UCI, systems were in- itiated during this period. The best material avail- able was used. KCI-UCi,. Thermal data obtained in this labo- ratory$3) indicated a eutectic between K,UCI, and KUCI which melted at 320 + 10°C, while the previously published(‘” diagram for this system shows a melting point for this composition of about 600"C. To ascertain whether a solid phase sepao- rates from the melt without giving o detectable thermal effect, a mixture containing 57 mole % KCl and 43 mole % UCI,, the approximate eutectic composition, was heated to 355°C, and a portion of the liquid was drawn by suction through a fritted- glass filter. |t was not possible to filter the liquid at lower temperatures, probably because of its high viscosity. The comparison of observed and calcu- lated volues for composition of the liquid are shown below: Found (%) Calculated (%) Potassium 10.9 10.8 Uranium 48.8 48.1 (B)R. J. Sheil and C. J. Barton, ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 108. (4)C. A, Kraus, Phase Diagrams of Some Complex Salts of Uranium with Halides of the Alkali aond Alkaline Earth Metals, M-251 (July 1, 1943). 42 The agreement between the calculated and the analytical results shows that no significant amount of material separated from the melt at 355°C and indicates that the previously accepted value for the melting point of this composition is erroneous. LiCi-uct,. Thermal data for compositions in the Lill-UCl, system containing 25, 40, and 75 mole % UCl, were reported in a previous study, (4} On the basis of the few compositions studied, it was reported that there were no compounds and that the eutectic temperature was 410 to 430°C, The thermal data obtained in this laboratory with the use of impure UCI, prepared by hexachloro- propylene chlorination of UO, are shown in Fig. 6.2. The data are considered tentative and subject to revision when purer UC|4 becomes available. The values obtained thus far ogree reasonably well with those reported in the previous study, but they seem to indicate that two eutectics melting at 415°C (approximately 29 mole % UCI,) and 405°C (approximately 48 mole % UCI )} exist in the sys- tem. The compound Li,UCl, appears to melt con- gruently ot 430 + 10°C; however, the flatness of the liquidus curve suggests that this compound is rather unstable, The existence of the compound is not yet supported by petrographic examination, since compositions in this region were too poorly crystallized to permit microscopic characterization of the solid phases. TICI-UCI,. On the basis of data obtained to date, it appeors that there is a eutectic in the TICI-UCI, system at approximately 16,5 mole % UCI, which melts ot 362 £ 10°C and a congruently melting compound TIZUClé which melts at about 480 + 10°C. The data are subject to the reser- vations stated above regarding purity of materials. KCI-LiCl-UCl,, The five compositions studied to date in the KCI-LiCl-UCl, system were prepared by adding increasing amounts of UCI, (HCP product, as received) to the KCI-Lill eutectic (41 wmole % KCI; melting point, 352°C), The ad- dition of 2 mole % UCI, to the binary eutectic raised the melting point about 16 deg, and subse- quent additions of up to 20 mole % raised the melting point further. For UCl, concentrations of up to 10 mole %, the cooling curves indicated the probable existence of a eutectic whose compo- sition is not yet known which melts at 345°C and contains a very low concentration of UCIH,, At the higher UCl, concentrations (30 and 35 mole % PERIOD ENDING JUNE 10, 1953 DWG. 19304 700 e : 600 — 5QQ oot 5 - Lud o ;{3 o <( - E 3OO b e e el Lt a = L) o 300 1 . et > 200 e 1 8 -~ o 2 o 1001 e e Licl t0 20 30 40 50 60 70 80 90 ucl, UCI4 (mole %) Fig. 6.2. The System LiCl-UC|4 (Tentative), UCl,), the cooling curves showed halts at approxi- mately 275°C. The halts are probably due to a high-uranium eutectic in the vicinity of the KCl- UCI, binary eutectic (43 mole % UCl,; melting point, 320 % 10°C). Work on this system is con- Yinuing. COOLANT DEVELOPMENT L. M. Bratcher V.S, Coleman C. J. Barton Materials Chemistry Division Most of the work with coolants during the past quarter was devoted to checking some of the results of earlier work and to preparing possible compounds for petrogrophic ond x-ray diffraction examination. An effort was also made to extend the data on the binary ZrF, systems to higher ZrF, concentrations through the use of sealed capsules. It seems likely that a part of the earlier data on systems containing KF, RbF, and CsF with ZrF, was erronecus because of the lack of sufficient care in keeping water out of contact with these hygroscopic materials, Work on a number of systems is still in progress and will be reported at a later date. NoF-ZrF,-BeF,. Although some data on the NaF-ZrF ,-BeF, system were reported earlier, renewed interest in the system as a result of the recent use of BeF, in pump seals made further work with compositions in this ternary system desirable, The lowest melting point found was 345°C for a mixture containing 15 mole % ZrF, 42.5 wole % NaF, and 42.5 mole % BeF,. Halts on the cooling curves were observed with this composition and other compositions in its vicinity at temperatures ranging from 290 to 320°C. It is not clear from the avaiiable data whether this thermal effect is due to a solid transition or fo a evtectic of unknown composition. On the NaZrF .- BeF., line, the lowest melting point observed in the range of 4 to 45 mole % BeF, was 487°C at 12.5 mole % BeF,. 43 ANP PROJECT QUARTERLY PROGRESS REPORT Lif-ZrF,. Preliminary data on the LiF-ZrF, system were reporfed in a previous progress ree por‘r.(s) Most of the early data have been subse- gquently proved to be incorrect, prohably because of the presence of oxide or oxyfluoride in the mixfures. 1he minimum melting point in the system was reported more recently.(®) Publication of an equilibrium diagram for the system has been de- layed by an apparent conflict between the thermal data and the results of petrogrophic and x-ray diffraction identifications of the solid phases in the 75 mole % LiF—25 mole % ZrF, mixture. Since it appears that a rather detailed examination of this region will be necessary to clear up the dis- crepancy, the thermal data are presented in Fig. 6.3 as a tentative equilibrium diagram, (3)y. P. Blakely, L. M. Bratcher, R. C. Traber, Jr., ond C. J. Barten, ANP Quar, Prog. Rep. Mar. 10, 1952, ORNL-1227, p. 104. (6). M. Bratcher and C. 1. Barton, ANFP Quar. Prog. Rep. Dec. 10, 1952, ORNL-1439, p. 112. The liquidus line shown in the diagrom indicates LijZrF, to be a congruently melting compound, but the solid material of this composition was found to contain Li,ZrF,, LiF, and a crystalline phase of lower refractive index than Li,ZrF . This latter phase has not been prepared in pure form, but it predominates in the mixtures con- taining 15 and 20 mole % ZrF,. The thermal data also show some indication of a compound con- taining more than 1 mole of ZrF, per mole of LiF. An unknown phase, in addition to Li,ZrF, and ZrF,, was reported to be present in the x-ray diffraction pattern of the 67 mole % ZrF , mixture, DIFFERENTIAL THERMAL AMNALYSIS R. A. Bolomey, Materials Chemistry Division Experience during the quarter with the differ- ential thermal analysis apparatus has shown that both the precision and the accuracy of the data obtained with this apparatus are increased by 1000 B DWG. 19905 900 e e— . e e ——— — - — _— - fg.-—i,--%b / / ¢ | // 800 |-—- 1--— _— e N P o — 7 O 2. E:J 700 ——— O e & —1 e = = < O o o ™ ; = / o w 600 f----———— S N | B ] f“——"f—J‘I— — 1 \po by < . 5 . L) o™ . !L- Cromeer s (ymrroonl) rarmeums el smmen 300 00 o o0 o o o L rTTT T T o 400 e e 300 . v — LiF 40 50 &0 70 80 90 Zrf, ZrE (mole %} 4 Fig. 6.3, The System LiF-ZrF, (Tentative). referring the differential temperature to the sample temperature rather than to the Al /O, reference material, Further improvement might be effected by using reference materials which have the same thermal properties as those of the sample but which do not exhibit transition temperatures in the temperature range of interest, PRODUCTION AND PURIFICATION OF FLUORIDE MIXTURES F. F. Blankenship G. J. Nessle Materials Chemistry Division Removal of HF from Fuel Batches (C. M. Blood, F. P. Boedy, R, E. Thoma, Jr., Materials Chem- istry Division). Poor corrosion performance of the fluoride mixtures could be caused by HF that re- mains in the finished fuel, Although there is no reason to believe that HF is appreciably soluble in NaZrF, at 800°C, further experimentation to check this point ond fo evaluate the rate of re- moval of HF by stripping techniques were per- formed during the quarter. Generally speaking, vacuum techniques for removing the HF must be avoided because of the risk of introducing air and water vapor through undetected small leaks. The HF content of gas which has been bubbled through the molten salts is best measured by the method of White and Manning{?) i which the con- ductivity of a dilute HBO, solution is measured as g function of the HF it cbsorbs from the strip- ping gas. In this way, the concentration of HF in the effluent gas can be followed almost con- tinuously, it has been shown that HF can be removed from the molten salts ot 850°C much more rapidly than from the unheated valves, gages, traps, and con- nections of the purification assembly, Conse- quently, numerous blanks have been run ic ascer- tain the rate of removal of HF from empty as- semblies under various conditions; these runs served only to indicate that the lost fraces of adsorbed or trapped HF are almost impossible to remove, Diminishing returns were reached at a level of about 107> mole of HF per liter of strip gas when routine procedures were applied to an empty apparatus with the usual lines aitached, and 107% mole of HF per liter of strip gas was regarded as close to the limit of removal. (7)J. C. White and D. L. Manning, ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 173. PERIOD ENDING JUNE 10, 1953 A mathematical analysis (L. Alexander, private communication) based on models resembling the system showed that a first-order process can be expected. |t o first-order process is actually followed, the amount of HF remaining in the melt at any time during the stripping operation can be calculated and then combined with the equilibrium vapor concentration to provide the Henry's law constant for solubility of HF in the melt. When helium is used as the stripping gas, the concentration of HF in the effluent gas decreases with time in a manner agpproximating that expected for a first-order process. When 2 kg of NaZrF is stripped at 800°C with a 5-cm bubble path, the half life appears to be 10 liters or less, Larger values seem to be due to siow removal of residual and adsorbed HF from cooler sections of the apparatus; with the most rigorous efforts to avoid extraneous contributions, half life values between 5 and 10 liters have been obtained with NaZrF ¢ as the melt. If 10 liters is used as the half life volume and 0.6 as the fraction of equilibrium con- centration maintained in the effluent gas, an upper limit estimate of 0.25 mole of HF per liter of strip gas per mole of HF per kg of fuel is obtained for the Henry’s law constant. When the stripping rote is 1 mmole of HF per liter of strip gas, the equis librium concentration is 4 mmoles of HF per kg of fuel; this amount of HF would produce about 80 ppm of CrF, by the reaction 2C¢° + 6HF‘::A"2CI‘F3 + 3H2 . The conclusion has been reached that melts siripped to a concentration of 1074 mole of HF per liter of strip gas connot contain HF in amounts sufficient to cause appreciable corrosion; pre- cautions must, of course, be taken to avoid recon- tamination of the melt from tramp HF from other portions of the system, When hydrogen is used to remove HF from the molten salts, the stripping operation is compli- cated by the production of HF by reduction of various fluorides. The differences observed in the use of hydrogen ond helium for stripping o uranium-containing fue! in Fig. 6.4, which shows that helium stripping occurs at a rate corresponding to a half-life volume of about 20 liters but that the change to hydrogen results in the production of HF ot a higher rate, The steeper portion of the hydrogen curve dppears fo be related to the reduction of structural metci are given 45 ANP PROJECT QUARTERLY FROGRESS REPORT 0_3 DWs. 19908 3 e e e e e ws. 19208 { T T j i . : [ 40-liter HALF LIFE | | | 5 . ] u ! L . . ! ‘ | | J . | o UG + Y H,m URg+HF ] ( _CHANGE FROM H, \ 2 |- T TO HELIUM ‘ - | | N | 5/ \ ‘{‘--*HZ TREATMEPJTM-—E = | MOLES OF HF per liter CF STRIP GAS T } o , ! o Lo P 3kg OF 50 50 NoUFy-NeZrFy AT 700°C [ 5-in. BUBBLE PATH i 2o e I . ; T ] e s \ { f | | | | [ TR I ‘7,, L } J 50 100 150 200 250 300 STRIP GAS (liters) Fig. 6.4, Stripping of Hydrogen Fluoride from a Molten Fluoride with Helium and Hydrogen, ions; a leveling takes place at about 1.8 x 1074 mole of HF per liter of gas, which corresponds to the rate of reduction of UF, to UF .. In the larger scale apparatus for treating 50-lb batches of material, the flow rate has been in- creased from 1 to 5 liters/min, and the bubble path is 10 em. Stripping rates from the 50-lb apparatus are about the same as those for the smaller rigs, Experimental data for four runs in the 50-1b apparatus, in which hydrogen was used for the stripping operation, are shown in Fig. 6.5. It appears from these data that HF concentrations in the gas phase can be brought to moderately low values by using hydrogen, even in large apparatus, without excessive stripping times, Reduction of Metal Fluorides with H, (C, M RBlood, F. P. Boody, R. E. Thoma, Jr., Materials Chemisiry Division). Nicke! is one of the metals that is resistant to HF ot high temperatures be- cause of a protective film of NiF, that forms on the metal. However, when nicke! is used to con- tain a melt capable of dissolving this film, the 46 3" x 5\ - _!u | — | . _! | ] | ~ | — i 2 _k b l - |1 S - _ L [ (MELTS PREVIOUSLY TREATED WITH HF FOR TWO HOURS. NICKEL CONTAINER) \ REACTION MONITORED BY MEASURING o A RATE OF HF EVOLUTION } L A .\ 815°C 10-in. BUBBLE PATH - ! i [ i o | I = TN i \ g | \e | | | ¥ o - e \g o [ | | - { o\ * | Ué I T N Li, . - _ - . | | l_j ,,,,, 17 A RO Afl J‘,io e l —— | ‘ g | ~N ! 2 NG e | AN | R R S N : \ AN A I f Ak \lr ’ 5 e s b i 1 — F— = {\ — ! ‘ : ‘ ’ t axyo L - j 7,\,,,_,7,fi,._% | A 200 400 800 800 1000 1200 H2 {liters) Fig. 6.5. Stripping of Hydrogen from Hydrogen Fluoride in 4- to 50-Ib Batches of NuF-ZrF4-UF4 (50-46-4 mole %). metal is slowly attacked by MF to the extent that 0.1% or more of Ni' " can be dissolved by the melt during a 2-hr HF treatment at 800°C, Unless this Ni** is subsequently removed, the me!t is quite corrosive to Inconel because of the nearly gquanti- tative replacement of Ni*" by chromium from the metal , Accordingly, several attempts have been made to study the rate of reduction by hydrogen of structural metal fluorides in NaZrF . In general, the NaZri, contained in graphite was siripped with hydrogen until the HF concentration in the exit gas concentration reached 107 mole/liter, After the melt had cooled, the predetermined quantity of metal fluoride was added. The sample was reheated to B0O0°C while being stirred with helium; hydrogen was then readmitted and analysis of the exit gas was begun, Data from some preliminary runs with NiF, are shown in Fig. 6.6, The trial run with T wt % Ni*" indicated that a clear-cut first-order reaction was involved., However, the data shown by 0.1% ad- ditions were poorly reproducible and did not agree well with those obtained with the higher concen- tration. In most cases, the integrated total yield N b | . { e e g tm{i\ { NiF, + H af\fu + 2HF \ I romrmea e et R T L _ ,,,,J..___,,,, - dmme s ermmesaan .. | 10’24% { . | i 10wt % Nitt f , 35-liter HALF LIFE _ 4 1 5 40 -liter f 3-kg BATCH AT 800°C ’ HALF LIFE _ 5-in. BUBBLE PATH - o e 2’;; T it A . f \# % . i Y J . \ MGLES GIF HF per fiter OF HYDROGEN 4 &0 -liter a;\ j T HALFLIFE e 1 S \ f \\ . } \L a l RN ‘ . 1G5 _‘A"fiAw‘_LMN~ 50 100 150 200 250 300 H2 (iiters) Fig. 6.6. Rate of Reduction of NiF in NaZrF by Hydrogen. of HF was about 25% low; this is well beyond the limits of accountable experimental errors, It seems likely that some reduction by the graphite from the container is responsible for these dlff;- culties. Preliminary data for other structural metal fluo- rides reveal that Fe*" is reduced about four times less rapidly than Ni** and Cr*** about 10 times less rapidly if equal quantities of these ions are present during the usual fuel-production procedures, in practice, no significant quantity of chromium is present in the starting materials, About 500 to 700 ppm of iron is present and, as described above, significant concentrations of mckel are added from apparatus, Data from analyses for iron, chromium, and nickel in the finished fuel are plotfed as a function of hydrogen treatment in Fig. 6.7. It is apparent that substantial benefits are obtained by hydrogen treatment for reasonable mfervals of time, [t has been established that the reduced metuls tend to accumulate as spongy masses on the hydrogen delivery line, on the vessel walls, and at the surface of the melts. Although the accumu- lation is valuable in prolonging filter life, it may prove to be a serious problem in equipment which cannot be opened for cleaning after preparation of the molten mixture. To date, however, there is no PERIOD ENDING JUNE 10, 1953 e BWE, 19908 T [¥p] - £ £ TG a = a < = & o) oy W Q2 L = o | > - 2 a uwi = 2 3 HYDROGEN STRIPPING (hr) Fig. 6.7. Hydrogen Reduction of Structural Metal Fluorides in NaF-ZrF -UF, (50-46-4 mole %) at 1500°F. evidence that continued re-use of the equipment leads to less pure fuel mixtures. Attempts have been made to prepare very pure fuel by using graphite-lined nickel equipment. A batch of NaZrF. so handled and given 20 hr of hydrogen treatment showed 30 ppm of iron and less than 20 ppm each of chromium and nickel. A more thorough study of the effect of UF, concentration on the rate of reduction of UF4 by hydrogen is planned. The preliminary indication from trials with compositions containing 20 ond 25 mole % UF, was that the rate of reduction is approximately the same gs with 4 mole % UF or, in other words, that the reaction is of zero order with respect to UF ; concentration, Pilot-Scale Fuel Purification (G. J. Nessle, J. P. Blakely, J. E. Eorgon, Materials Chemistry Di- vision). During this quarter a fotal of 318.9 kg of mixed fluorides was processed. This quantity of fluorides comprised 12 small batches of approxi- mately 2 kg eoch and 11 large batches of approxi- mately 27 kg each. 47 ANP PROJECT QUARTERLY PROGRESS REPORT The gas treatment of the fluorides during prepa- ration has been altered to effect a more thorough removal of retained HF in the melt and to yield a final product lower in metal impurities. The treat- ment consists of the following steps: 1. melt fluoride charge under HF atmosphere, treat with hydrogen for 1 hr, treat with HF for 11/2 hr, flush with hydrogen for 3 to 4 hr, transfer molten fluorides to receiver with helium, flush with helium during cooling and solidifi- cation. This oltered procedure substitutes a hydrogen flush for the helium flush; hydrogen will strip out HF as efficiently as helium aond, in addition, will reduce more of the metal impurities. With the uranium-bearing mixes, the longer hydrogen flush may produce a small amount of UF,. Careful con- trol of the amount produced will be necessary to avoid precipitation of the UF,. An aodditional change in the treatment procedure is the use of a continuous flow of gas of approxi- mately 5 liters/min during the hydrogen-flushing period., Previously, intermittent flow had been obtained by pressurizing to 10 Ib and then bleeding off to atmospheric pressure in a 2-min cycle. The data concerning the preparation of niixed fluorides in the pilot-scale facility during this quarter are given in Table 6.4. Fluoride Production Facilities (G. J. Nessle, Materials Chemistry Division). The installation of process equipment to be used for production of the nonuranium fluoride base for the ARE fuel has been * O hWwho completed, and preliminary tests are under way. The first ““dry run’’ in one of two process cubicles proceeded with no difficulty, ond data were ob- tained concerning heating time and gas flow; no leaks were found either before or after the test, During a second dry run in this cubicle, a short circuit developed in the receiver furnace and necessitated a complete shutdown for two weeks for installation of new elements in the furnace, A dry run for the other process cubicle is sched- uled in the near future. It is hoped that some valuable information concerning the behavior of HF in the process equipment can be obtained simul- taneously with the trial runs. The installation of the process equipment to be operated by the Y-12 Production Division for the production of the enriched-fluoride mix for the ARE is nearly complete. Arrangements have been made to train the personnel directly concerned with this operation. This short training program will probably be started during May. PREPARATION OF COMPLEX FLUORIDES B. J. Sturm L. G, Overhol ser Material s Chemistry Division Methods of preparation and partial listings of properties of complex fluorides of the types re- sulting from interaction of alkali and structural metal fluorides have been reported previously, (8 During this quarter, additional batches of the @) ANP Quar. Prog. Rep. Dec. 10, 1952, ORNL-1439, p. 115, TABLE 6.4, PILOT-SCALE FLLUORIDE PREPARATIONS COMPOSITION MIXTURE per cent ppm pH F U Ze Ni Cr Fe NaF-ZrF ;-UF , (46-50-4 mole %)% 42.1 8.52 | 38.8 125 20 375 2.03 NaF-ZrF -UF , (50-46-4 mole %){®) 42.3 8.92 38.0 135 30 345 2.32 NaF-ZrF , (50-50 mole %)) 43.8 (d) 43.2 37 20 245 2.38 NaF-ZrF ;-UF ; (50-25-25 mole %){¢) 33.6 | 41.2 16.3 20 20 40 2.92 NaF -UF , (50-50 mole %){) 27.3 | 67.4 20 120 190 2.85 (a)Average of 4. Average of 16, (e) <) Average of 3. 43 (d)Nof reported. Average of 1. complex fluorides, as well as batches of the simple structural metal fluorides, were prepared. These materials were identified or characterized by the use of chemical analyses, x-ray examination, and crystallographic data. ' The foliowing list gives the materials prepared during this quarter. Similar batches of some of these materials were prepared previously by this group, but some other batches represent new syntheses made during this period: NiF, KNiF, Na.FeF CsNiF BT N tentatively identified LiCrF RbNiF, CrFy NiCl, NaCrF, FeCl, Fusions were made that consisted of NaZrF with each of the following: CF, NoCrF, and NoCrF,. In each case, the chromium compound lost its identity; x-ray analysis indicated a possi- ble solid solution of Na CrF, and NaZrF ofter fusion. Fusions of mixtures of KF and MnFB, corre- sponding to KMnF ,, K, MnF , KBM”F(H and K MnF_, all yielded compounds with the same x-ray pattern, The product approximated K,MnF, which may be a solid solution of KMnF ; and KMnF . X-RAY DIFFRACTION STUDIES P. A. Agron, Chemistry Division In order to resclve some of the complexities in the NaF-ZrF -UF, system, examinations have been made of several incongruently melting binary com- pounds and a number of ternary regions. ' NaF-UF, (75-25 mole %). The binary region in the vicinity of the NaF-UF (75-25 mole %) compo- PERIOD ENDING JUNE 10, 1953 Cooling curves, aleng with x-ray examination of the solid phases, defi- nitely established that Na,UF is an incongruently melting compound. The tetragonal N03UF7{9) ap-~ pears to be stabilized by zirconium ion, sition was investigated. small additiens of Another phase, which may be an allotropic form of Na,UF,, will be investigated further. ' NoF.ZrF, (50-50 mole %). Further study of the NaF.Z:F, (50.50 mole %) region indicates strongly that NaZrF . has an incongruent melting point and Preparations of this binary salt generally yield the normal hexagonal compound(m) on cooling the melt with adequate stirring. On examination of large-batch prepardtions of NaZrFs, the appearance of an unknown phase is marked in many instances. Two new phases in the NqF-ZrF4 (50-50 mole %) composition region have been observed in quenching experiments aft 600°C and on fusing small quantities of NiF, and CrF, with normal NquFg. Further work will definitely establish the nature of these two new crystal structures. The phase induced by rapid observed earlier ond erroneously aftributed to o reduced form of zirconium fluoride. This unidentified phase was found to be unstable when exposed to atmospheric conditions. The lack of maintenance of equilibrium the attendont complexities, cooling was conditions on possing through the incongruent freezing point is doubtless responsible for the muitiplicity of phases found in the solid. * W’W. H. 2aahariasen, The Crystal Structure of NQ3UF7, AECD-1798 (Mar, 3, 1948). : (10)p pgrom, ANP Quar. Prog. Rep. Sept. 10, 1952, ORNL-1375, p. 86. 49 ANP PROJECT QUARTERLY PROGRESS REPORT 7. CORROSION RESEARCH W. D. Manly, Metallurgy Division W. R. Crimes, Materials Chemistry Division H. W. Savage, ANP Division During this quarter, the static and seesaw test facilities have been used almost entirely for the study of the corrosion of Inconel by a proposed ARE fuel mixture, NaF-ZrF -UF . Among the corrosion parameters examined in 100-hr tests at 1500°F, were exposure time, crevice corrosion, container cleanliness, and the effect of various additions, including potential corrosion inhibitors, Neither crevices nor container cleanliness seems to have a significant effect on corrosion. The corrosion rate decreases with time and is shown to be « function of the fluoride contaminants, particulorly Nif and FeF,. Hard-facing alloys of the Stellite type thot were deposited on Inconel and static tested in the fluorides at 1500 and at 1300°F were rather severely aottacked, Several corrosion tests have been completed with fluorides in Inconel on the NACA type of rotating apparatus. |his type of apparatus is advantageous becouse velocities can be obtained that are higher than those available in thermal convection loops or in the seesaw tests; however, no large temperature gradient can be attained and the flow of liquid is not continuous. In tests conducted so far ot 1500°F for 100 hr, there has been no indication of plugging, and the corrosion is similar to that obtained in the static tests, even though a flow velocity of 10 fps is employed. as well as fuel-processing contaminants. Fluoride corrosion studies in dynamic systems provided by thermal convection loops have been continued, The loops are normally operated for 500 hr with a hot-leg temperature of 1500°F, Most of the work during this quorter was with Inconel loops circulating the fluoride NaF-ZrF -UF (50-46-4 mole %), which was the proposed ARE fuel. It has been shown that the depth of attack increases quite rapidly during the first 250 hr of operation but that thereafter the attack is essen- tially constant, These effects are also found with the fluoride NaF-ZrF , ~ the ARE fuel carrier - but the attack is not so deep. The depth of attack is not temperature sensitive; only a slight increase in depth occurred when the temperature was in- creased from 1250 to 1600°F, The type of attack, 50 The depth of attack was reduced when the small amount of weld scale found on the loop was removed by precleaning with NaF-ZrF , (30-50 mole %). Some purification of the fuel is desirable, however, does change, but the present production- purification facilities seem to be adequate. It was shown that the depth of attack is a function of the fuel and not of the diffusion of chromium from the solid-solution [nconel, since two batches produce almost twice the penetration of ¢ single batch. Addition of NiF, to the fuel increased the depth of attack., Chromium metal added to the fluoride NaF-KF-LiF-UF , (10.9-43.4-44.5-1.1 mole %) had no effect on the attack. In a nickel loop which circulated the fluoride Nc1i7:-Z:'F4-UI::4 (50-46-4 mole %) there was practi- cally no attack; however, a small amount of metallic mass transfer was found. Two type 316 stainless steel loops were operated for 500 hr with this same fluoride composition; these loops showed considerable mass transfer but no plug formation. The study of the mass transfer of various metals in high-purity lead is being conducted in quartz thermal convection loops. So far, the following container materials have been examined: Inconel, molybdenum, columbium, Armco iron, and types 304, 347, and 446 stainless steel. In these tests, molyb- denum and columbium did not show mass transfer, while type 446 stainless steel showed a slight amount of mass transfer, The remaining metals showed extensive mass transfer and premature plugging of the thermal convection loops. The coirosion resistance of treated beryllium oxide specimens has been studied in both static and dynamic corrosion tests, The corrosion mechanism is, apparently, only the mechanical erosion of the beryllium oxide surface. Beryllium samples plated with chromium and nickel have been tested in sodium contained in Inconel tubes. The plating did not adhere to the specimens. Several Carboloys have been tested in sodium and lead at 1500°F and, with the exception of Coarboloy 608, these alloys appear to have good corrosion re- sistance. FLUORIDE CORROSION OF INCONEL N STATIC AND SEESAW TESTS H. J. Buttram C. R, Croft R. E. Meadows Material s Chemistry Division D. C. Vreeland L. R. Trotter E. E. Hoffman J. E. Pope Metallurgy Division The test conditions of the static tests in which the Inconel specimen is sealed with the fluoride mixture in an Inconel capsule are 100 hr at 816°C, unless otherwise stated in the following sections. The conditions of the seesaw tests in which the copsule containing the specimen and the fluoride is rocked in a furnace are 4 cpm with the hot end of the capsule at BO0°C and the cold end ot 650°C. Effect of Exposure Time. Although a large number of studies have been reported in which fluoride mixtures in capsules of Inconel have been exposed in the tilting furnace for intervals of up to 500 hr, very little information has been available on the course of the attack during the first few hours of such exposures. To supplement previous data, copsules containing the luoride NaF-ZrF -UF {50-46-4 mole %) with and without zirconium hydride odditions hoave been exomined ofter exposures of 1, 5, and 10 hours. ' Metallographic examination of the capsules with- out zirconium hydride after these short exposures indicated only very slight attack, Chemicaol analy- sis of the fuel after tests (Table 7.1) revealed that PERIOD ENDING JUNE 10, 1953 the chromium content increased in direct proportion to the exposure time during these short exposures but that the iron and the nickel contents of the material dropped to nearly constant values after 5 hours. In the tests made with zirconium hydride added to the fuel, somewhat less corrosion was found upon metallographic examination. With zirconium hydride added to the fuel, the iron and nickel com~ pounds ore apparently reduced guite rapidly, and they remain at low concentration levels for the durotion of the tests., The linear increase in chromium content with exposure time is prevented. In previous experiments“? with fuel to which zirconium hydride had been added, the chromium content increased by about 100 ppm in the interval between 100 and 500 hr of exposure. Effect of Hydrogen Fluoride. The value of previous work(?2) on the effect of HF on the cor- rosion of ARE fuel was somewhat in doubt because of the possibility that the presence of UF, in the melt might have, according to the reaction of 2UF3 + QIHF:W_T“2UF4 + H,, masked the dele- terious effect of small additions. Additional studies have been performed with a mixture which should have contained fess UF, than was present in the previous mixtures. To produce hydrogen fluoride concentrations of 0.016, 0.032, 0.097, and 0.29%, “)D.C, Vreeland et al., ANP Quor. Prog. Rep. Mar, 10, 1953, ORNL-1515, p, 118, 2pid., p. 120. TABLE 7.1. EFFECT OF EXPOSURE TIME ON CORROSION OF INCONEL BY NoF.ZrF ~UF , {50-46-4 mole %) IN A TILTING FURNACE EXP OSURE zem, METAL CONTENT AFTER CHANGE IN METAL CONTENT TIME ADDITION TEST* (ppm) DURING TEST {(mea/kg) (hr) (wt %) Fe Cr Ni Fe Cr Y 1** None 215 45 65 ~5.1 +1.5 +0.4 Fx* None 145 120 35 76 +6.0 0,6 10** Mone 155 190 40 ~7:3 +9.8 ~0.5 1= 0.5 a5 30 <25 ~2.9 +0.8 -0.,9 5 0.5 55 25 <20 -10.8 +0.4 -1.0 10** 0.5 75 <25 <20 -~10,2 +0.3 ~1.0 *Metal content of fuel before test: **Average of two tests, Fe, 380 ppm; Cr, <20 ppm; Ni, 50 ppm. 51 ANP PROJECT QUARTERLY PROGRESS REFPORT NaHF ., was added to the mixture. The tests were conducted in the tilting furnace at 800°C, The data shown in Table 7.2 reveal that, as before, the two smallest additions were not harmful; however, the 0.097 and the 0.29% HF additions caused very marked increases in the chromium concentration, Thus the previous results were confirmed. There is no obvious explanation for this behavior. Similar studies will be conducted with uranium-free fuel solvent to eliminate uranium as a possible source of difficulty. It should be pointed out that the lowest concentration of HF used in these tests is probably higher than the amount that will be left in the fuel during its preparation. Accordingly, it seems likely that residual HF will not cause enhanced corrosion by carefully hydrogenated fuel, Flvoride Pretreatment. Since the corrosion characteristics of the fuel might be strongly de- pendent on the manufacturing process, the effect of some of the process variables has been evaluated by the standord tilting-furnace technique. The standard processing for fuel mixtures in- cludes high-temperature treatment with anhydrous HF. Since HF couses increased corrosion, at least when used in large amounts, parallel tests have been conducted on otherwise comparable materials from which the HF was stripped by purging with helium for various intervals of time. The data obtained by chemical analysis of three batches of fuel after 100-hr seesaw tests are shown in Table 7.3. Since the initial concentration of soluble structural metal compounds was similar in the three batches, the decrease in chromium con- tent with increasing stripping time is probably due to the effective removal of HF. Metallographic observation of the test capsules showed moderate corrosion to a depth of 1 mil for the shortest strip- ping time; corrosion by the other two batches was TABLE 7.2, EFFECTS OF chHl"'2 ADDITIONS ON THE CORROSIOM OF INCONEL BY I~lc=F-ZrF4-UF’4 (50-46-4 mole %) DURING A 100-hr EXPOSURE METAL CONTENT CHANGE IN METAL NaHF, ADDITION FINAL HF AFTER TEST* CONTENT DURING CONCENTRATION (ppm) TEST (meq/kg) (wt %) {wt %) (meq of HF/kg) Fe Cr Ni Fe Cr Ni None 0 0 70 830 60 | ~8.6 47 | +0.5 0.05 8 0.016 45 755 | <20 | -9.5 43 | -0.8 0.1 16 0.032 110 850 | <20 ~7.3 49 | -0.8 0.3 48 0.097 80 1705 | <20 | ~8.4 97 | ~0.8 0.9 144 0.29 75 | 3065 85 | -B.4 | 176 | +1.2 *Metal content before test: Fe, 310 ppm; Cr, <20 ppm; Ni, 45 ppm. All data are average of two tests, TABLE 7.3, EFFECT OF HELIUM STRIPPING ON CORROSION OF INCONEL BY NaF-ZrF4-UF4 (50-46-4 mole %) STRIPPING METAL CONTENT METAL CONTENT CHANGE IN METAL CONTENT TIME BEFORE TEST (ppm) AFTEFE-I_EST* {(ppm) DURING TEST (meq/kg) (hr) Fe Cr Ni Fe Cr Ni Fe Cr Ni 1 410 <20 30 g0 930 20 ~11.4 +57.8 ~0.3 3 680 <20 110 135 665 25 ~19.6 +41.0 ~3.0 5 470 <20 35 75 580 20 ~15.4 +35,5 -0.5 *Average of iwo tests. 52 considerably less severe, and there were no ap- preciable differences between the results for the 3- and 5-hr stripping times. Hydrogen should be more advantageous than helium for stripping the HF, since reduction of FeF,, FeF,, and NiF, to the metallic state and reduction of some UF, to UF,; would be simul- taneously dccomplished. In Table 7.4, the resulis of tests with untreated fuel, fuel treated at 750°C for ]]/2 hr with hydrogen, and fuel treated at 750°C for 1‘/2 hr with hydrogen and then filtered are shown for comparison. Results of metallographic exami- indicate that in 100 hr at 800°C in the seesaw apparatus, the untreated material caused heavy attack, fo a depth of 4 mils, while the fuel which was treated with hydrogen and filtered caused slight attack, to o depth of 1 mil. The material which was freated with hydrogen but not filtered caused moderate attock, to o depth of 3 mil s, Structural Metal Fluoride Additions. In additional studies, FeF,, CrF,, and NiF, were added to molten fluorides in an attempt to evaluate the importance of these materials in the selective leaching of chromium from Inconel. Previous experience had shown that FeF., increased the corrosion of Inconel by fluoride mixtures, and in the recent experiments in which FeF, was used, there was severe attack, as was expected. How- ever, the recent experiments indicated that in- creasing the FeF, concentration from G.1 to 0.9 wt % had very little effect on the quantity of nations PERIOD ENDING JUNE 10, 1953 chromium dissolved or on the severity of attack during the 100-hr exposure. Additional experiments will be required to confirm and explain this be- havior, The addition of large amounts of CrF, (0.5 to 3.0 wt %) to the ARE fuel mixture prior to exposure caused moderate to heavy subsurface void formation to a depth of 6 mils in the hot porfion of the cap- sule. It is possible that the reaction CrF3 + Cr® = ZCer is responsible for the very high corrosion observed. Structural Metal Oxide Additions. A series of static tests of Inconel in the fluorideNaF-ZrF ,-UF (46-50-4 mole %) with additions of Fe,O,, Cr,04, and NiO have been completed. The tests were run for 100 hr at 816°C. In these tests, additions of Fe 0, and Cr,0, did not affect the depth of attack, but NiQO increased the attack, The metallographic results of these tests are listed in Table 7.5. In all tests in this series, a bluish-gray phase was detected on the surface of the Inconel in what have in previous tests been considered voids. It was found on repolishing some of the specimens that by eliminating the last two polishing steps on the long-nap cloths, a large amount of this phase could be retained. The material in place in the voids is shown in Fig. 7.1, enlarged 2000 times. In order to confirm the presence of material in the voids, a Bergsman microhardness determination was run on this bluish-gray phase. It was found to have a DPH value of 892 (inconel matrix 135). TABLE 7.4, EFFECT OF HYDROGEN STRIPPING ON CORROSION OF INCONEL BY I‘\h:tl""-Zer;-lJF4 (50.46-4 mole %) L CONTENT ME DN CHANGE IN N SLroRE TEST AFTER TESTY ME TAL CONTENT PROCESSING e f ) DURING TEST VARIATION ppm ppm (mea/kg) Fe Cr Ni Fe Cr Ni Fe Cr Reference (untreated) 2100 35 200 75 4000 30 w74 229 Hydrogen stripped 690 70 190 265 2035 -4 118 Hydrogen stripped and filtered A70 20 | 35 205 610 ~-10 35.2 *Average of two tests, 53 ANP PROJECT QUARTERLY FPROGRESS REPORT TABLE 7.5. EFFECT OF OXIDE ADDITIONS ON STATIC CORROSION OF INCONEL IN NuF-ZrF4~UF4 AFTER 100 hr AT 816°C ADDITIVE METALLOGRAPHIC NOTES None Specimen attacked 3 to 6 mils, average 4 mils; tube attocked less than 0.5 mil 1% Fe203 Specimen and tube attacked 3 te 4 mils 1% NiO Specimen attacked 5 to 7 mils; tube attacked 7 to 8 mils 1% Cr203 Specimen attacked 3 to 4 mils; tube attacked 2 to 3 mils UNCLASSIFIED PHOTO Y-8708 Fig. 7.1. Static Corrosion of Inconel by NaF-ZrF ,-UF, (46-50-4 mole %) After 100 hr ot 816°C. Note material in voids., 2000X Microspark spectrographic examination of these specimens indicated that the material in the voids contained Fe, Ni, Cr, Zr, and probably U, although only a faint trace of uranium was detected. Chromium Metal Addition. Molten ARE mixture was pretreated in a tilting furnace by exposure to chromium metal in Inconel capsules. Most of the excess chromium was then removed by sedimenta- tion, and the pretreated mixture was transferred to another capsule for testing, In o series of experi- ments, this pretreated mixture caused only a slight roughening of the surface at the hot end of the capsule and a slight deposit on the cold wall, The addition of NiF2 to samples similarly pretreated increased the corrosion considerably; in these 54 tests the chromium content of the melt increased linearly with the NiF, addition. Zirconium Hydride Addition. Zirconium hydride reacts with UF , in fluoride melts to form uranium trifluoride. A certain amount of the trifluoride can be retained in solution, but precipitation of this material occurs if extensive reduction takes place. An effort has been made to check previous data on the tolerable limit of ZrH, addition by allowing the insoluble UF, to separate by sedimentation after exposure. The additional data obtained (Table 7.6) indicate that the uranium concentration of the melt is decreased considerably with addi- tions of 0.7% or more of ZtH,. These data, in general, agree with data from previous experiments performed with other techniques. The chromium content of the melt is sharply decreased by the addition of as little as 0.1% of ZrH,; further increases in the ZrH, concentration have only o slight effect. [t appears that although up to 0.7% of ZrH, may be folerable in this system, much lower concentrations, of the order of 0.2% may show the beneficial action desired without the risk of precipitation of uranium from the melt, Carbon Addition. Static tests were run for 100 hr at 816°C with o mixture of 25% powdered graphite and 75% NaF-ZrF -UF , (50-46-4 mole %) in type 316 stainless and Inconel tubes to determine carburization of these moterials would No evidence of carburization was detected in these tests. After testing, the grain boundories of the type 316 stainless steel had large amounts of carbides present, but the deposits were uniform throughout the specimen and no hardness change could be detected from the inner to the outer surface of the tube wall, | whether oCCur. PERIOD ENDING JUNE 10, 1953 Crevice Corrosion. Crevice corrosion tests have been run in the seesaw furnace with the tapered, or V-shaped, crevices described previous]y.“) Corrosion in these crevices seemed to be somewhat erratic, with some surfaces being apparently un- attacked and others having the usual subsurface voids. No acceleration of corrosion was noted in these crevices, and the depth and the extent of afttack in the crevices was not beyond what might be usually expected for the materials tested, Table 7.7 summarizes the results obtained in these tests. Similar results for crevice corrosion were obtained in the rotating tests reported in a following section. Inconel Container Pretreatment. The descaling properties of the fluoride NaZrF _ have been checked on oxidized inconel at 1000 and at 1300°F. As can be seen in Fig. 7.2, the oxidized Inconel was cleaned up after 4 hr at 1300°F, but it was not cleaned up after 4 hr at T000°F, It was thought that perhaps o passivation treat- ment, similar to that given fo stainless steels, TABLE 7.6, EFFECT OF VARYING QUANTITIES OF ZrH, ON CORROSION OF ~ INCONEL BY NaF-ZrF -UF ; (50-46-4 mole %) ZiH, ADDED URANI;M CONTENT AFTER METAL CONTENT AFTER TEST* (wt %) rH, ADDITION (ppm) (wt %) Fe Cr Ni None 2. 11 95 560 ‘ 35 0.1 .01 70 195 B8O 0.3 8.94 125 150 50 0.5 9.17 220 175 90 0.7 8.65 405 145 155 0.9 6.76 350 130 125 *Metal content before test: Fe, 310 ppm; Cr, 20 ppm; Ni, 45 ppm, All results average of two fests. TABLE 7.7. RESULTS OF SEESAW TESTS OF INCONEL WITH TAPERED CREVICES IN NaF-ZeF UF , (50-46-4 mole %) FOR 100 hr WITH A HOT-ZONE TEMP ERATURE OF 800°C AND A COLD-ZONE TEMPERATURE OF 600°C POSITION OF CREVICE METALLOGRAPHIC NOTES Hot zone Hot rone Cold zone 2 mils of subsurface voidz in upper poart of crevice, increased to 3 mils in lewer part Subsurface voids te 2.5 mils Apparently no attack 35 ANP PROJECT QUARTERLY PROGRESS REPORT !nnhm!nu?;.;;; AS POLISHED gfi.&. LR imETen gy g - ¥ . 4 - {9. f UNCLASSIFIED PHOTO Y-9192 ’u!rmimslnnl:zzzizm AS OXIDIZED Fig. 7.2. Temperature-Dependence Tests on the Descaling of Oxidized Incone! by NaZr Fs. would be beneficiol in increasing the corrosion resistance of Inconel to the fluoride mixtures. Therefore, Inconel tubes were hydrogen fired and the specimens were electropolished, as usual, and both the tubes and the specimens were treated in a 25% solution of nitric acid at approximately 120°F for 30 minutes. The tubes and specimens were then used for static corrosion testing with fluoride NoF-ZrF -UF, (50-46-4 mole %) for 100 hr at 816C. The pretreatmant did not appear to be especially beneficial. Attack inthese tests seemed more erratic than is usual. In two of the tests, subsurface voids to a depth of 5 mils were noted; 56 in the other test, subsurface voids were noted 1o a depth of 3 mils. STATIC CORROSION OF STELLITE BY FLUORIDES D. C. Vreeland E. E. Hoffman, L. R, Trotter Je E. Pope Metallurgy Division In conjunction with the hard-facing problems of the ARE, Stellite No. 6 (27.5% Cr, 4% W, 2 to 3% Fe, 1% C, balance Co) was deposited on pieces of Inconel, and these composite specimens were static tested in the fluoride NaF-ZrF4-UF4(46-50-4 mole %) in Inconel tubes at both 816 and 538°C, The structure of Stellite No. 6 has been described as ‘‘dendrites of a cobalt-rich solid solution sur- rounded by a mixiure of carbides.'”” Apparently it is the carbides which are attacked by the molten fluorides., There appeared to be no acceleration of corrosion on either the Stellite or the Inconel as a result of the presence and contact of these dissimilar metals. The extent of attack appeared to be the same in the tests at 538°C as in the tests at B16°C. In all cases, the Inconel showed about 1 mil of subsurface voids, while the Stellite showed an average attack of 3 to 4 mils and a maximum attack of 9 mils on the carbide phase in the Stellite. Figure 7.3 shows the type of attack which occurred in these tests.. FLUORIDE CORROSION OF iNCONEL IN ROTATING TEST D. C. Vreeland L. R. Trotter E. E. Hoffman J. E. Pope Metallurgy Division Several tests have been completed with the fluo- rides in Inconel on the NACA type of rotating ' UNCLASSIFIED { PHOTO Y-8951 A . 1 3 o - - ( - . - .‘ - Ve . - e * ! - ’ - R v l < L % wn LsJ X O =z ‘ K . 0.005 "'-:-. ,0.: "‘. ” T ? — v ’ # -(x' . X C e bes v oE s .Y oo ) ' s *Y . I?) .o o ..’ fl r - ‘ - 0010 RN )"’ . PERIOD ENDING JUNE 10, 1953 apparatus.t3) This type of apparatus is advan- tageous because velocities can be obtained that are higher than those available in thermal convec- tion loops or in the seesaw tests, In the tests conducted to date, there has been no indication of plugging, and the corrosion results have been similar to those obtained in static tests, even though a fluid velocity of 10 fps is being employed, The depth of penetration has not exceed=d 2 mils in any of the tests in which leaks hav> not oc- curred. Atypical 100-hr test with NaF-ZrF 4-UF , (46-50-4 mole %) in which a temperature of 816°C was maintained throughout the apparatus resulted in uniform attack in the form of subsurface voids to a depth of 2 mils. Additions of sodium have re- sulted in the development of surface layers on the inconel, as was also noted in static tests. The fluoride, with 2% sodium added, was tested for 94 hr with a hot-zone temperature of 816°C and a cold-zone temperature of 783°C., Although the attack was in the form of subsurfoce woids, a surface layer was visible in many places.. The total depth of voids and ldyer did not exceed 0.5 Ghbid., p. 121, Fig. 7.3. Static Corrosmn of Stellite No. & After 100 hr at 816°C in NoF-ZrF 4-UF ; (46-50-4 mole %). 250X 57 ANFP PROJECT QUARTERLY PROGRESS REPORT mil, nor did the attack vary in extent or nature from the hot zone to the cold zone. Some of the crevices near welded joints tested in the whirligig opparatus in fuel mixture with the 2% sodium addition were examined metallograph- ically for evidence of crevice corrosion, and no accelerated corrosion could be found. The greatest depth of attack noted was 0.5 mil in the form of subsurface voids. On the other side of the crevice with the greatest depth of attack, there was attack to 0.5 mil that was definitely intergranular. As is usually noted in these crevices, the attack was somewhat erratic, with some sections unattacked. FLUORIDE CORROSION IN INCONEL THERMAL CONVECTION LOOPS G. M. Adamson, Metallurgy Division The use of thermal convection loops for determin- ing dynamic corrosion by liquids has been pre- viously described.’#) Unless otherwise specified, for the tests described in the following sections, the temperature of the hot leg of the loop was maintained at 1500°F and the temperature of the vninsuloted cold leg was approximately 1300°F, With the fluoride salts, this temperature difference results in a fluid velocity of aboyt 6 to 8 fpm, The usual testing period is 500 hours. Effect of Exposure Time. A series of loops has been placed in operation to determine the effect of time on the depth of maximum penetration. The loops were filled from a single batch of fuel at as (4)D. C. Vreeland, R, B, Day, E. E. Hoffman, and L. D. Dyer, ANP Quor, Prog. Rep. Mar. 10, 1952, ORNL - 1227, p. 119, near the same time as possible and were then allowed to circulate for times varying from 10 to 3000 hours. The tests of short duration have been terminoted, and the loops have been examined. The data obtained are given in Table 7.8, along with the data obtained from loops run previously for periods of 100, 250, 500, and 1000 hours, The data show that the depth of attack increases quite rapidly for about 250 hours., After 250 hr, the depth of attack remains almost constant, but there is some increase in the intensity of attack. When data are available from the other loops filled from the same fuel batch, a time curve will be prepared. Since chromium must diffuse to the surface of the hot leg before it can be removed, the diffusion rate could be the limiting factor in the corrosion rate. Two Inconel loops were operated for 500 hr to test this hypothesis, The first loop operated for two 250-ht periods, with g new chorge of fuel for the second period. The second loop operated for 500 hr with a single charge from the same fuel. If dif- fusion was the limiting factor, the attack should have been the some. The loop which had two charges showed heavy subsurface void formation of 8 to 17 mils, with the voids primarily in the grain boundaries. The loop with a single charge showed moderate to heavy attack of from 3 to 8 mils, with general attack of 5 mils, The hot-leg sections from both of these loops ore shown in Fig. 7.4. The doubling in depth of attack for two charges waos also confirmed by the chemical analyses, The chromium content of the fluorides increased more during the second 250-hr period in Loop 281 than it did in the entire 500-hr period in Loop 28Q. These lcops show that the decrease in attack rate TABLE 7.8, EFFECT OF EXPOSURE TIME ON THE CORROSION OF INCONEL BY NaF-ZrF4-U Fy (50-464 maole %) EXPOSURE TIME MAXIMUM DEPTH (hr) OF ATTACK REMARKS (mils) 10 1 Widely scattered grain boundary attack 50 3 Scattered groin boundary attock 100 4 l.ight, typical subsurface voids 250 9 Moderate attack 500 9 Moderate attack 1000 11 Moderate to heavy attack 58 PERIOD ENDING JUNE 10, 1953 UNCLASSIFIED PHOTO T.3037 "> T ’ “ * | 04 % e : o - | ! “. ‘ _\\ ’ . S . - - 3 —8 o @ . Q. Son o ® o - . W ) . Y . ‘ O < . 9 . ’ " - e - - - . - g \ oy —?_w . > - o FES . -, ;a * ‘O ST ~ : T ' Iy | - MILS . .- > - e Y £ : e . 7 - ¢ 4 - ;'_’ ’,/ . * . L £ . “’ - . e i / ! & - . oo B f : ’ P Lt ey . LA - L % ’ - : / =, t 3 v ¥ - e . . . -~ i * i * ‘ .‘ P ¥ bt & UNCLASSIFIED tPHOTO T-3042 Fig. 7.4. Corrosion of Inconel by NaF-ZrF -UF (50-46-4 mole %) After 500 hr at 816°C. (a) 500 hr with same fuel charge. (b) 250 hr on each of two fuel charges. 250X 59 ANP PROJECT QUARTERLY PROGRESS REPORT with time is caused by changes in the fuel rather than by the limiting action of diffusion, The in- crease in attack is probably coused by an increase in impurities with a double charge. Temperature Dependence, In the previous re- port, (5} results for a single loop operated at a hot-leg temperature of 1650°%F were compared with those for a loop operated at a hot-leg tempera- ture of 1500°F, An additional series of three loops has now been operated to confirm those results, The loops in this series were all filled from a single fuel pot on the some day. The results of this temperature study are presented in Table 7,9 and illustrated in Fig. 7.5. These results show that the change in depth of penetration of the attack is not very temperature sensitive. The maximum penetration at 1300°F is slightly less than at 1500°F, but it is within the usual spread in re- sults, and the results for all the loops are definitely within the experimental variation. The nature and distribution of the ottack does change with temperature. At the lower temperature, the voids are small and evenly distributed. As the temperature increases, the voids concentrate in the grain boundaries and become quite large. The growth and concentration in the woids are caused by the increasing mobility of the fluid as the temperature increases. In contrast with the pre- vious temperature tests, the recent tests showed no systematic distribution of impurities in the fluorides at any temperature. (5)G, M. Adamson, ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 123. Chromium ond Nickel Fluoride Additives., One more attempt was made to reduce the depth of attack by pretreating the fluoride NoF.KF-LiF-UF, (10.9-43.5-44.5-1.1 mole %) with chromium meta! powder. The powder was added to the fluoride while in the charging pot. The mixture was held at 1300 to 1400°F, agitated with helivm for several hours, and then blown through a stainless steel pressure snubber before the sample was taken, The chromium content of the fluorides as charged to the loop was 3500 ppm. After the 500-hr test, the chromium content was 2400 ppm. The attack in this loop was to a depth of 10 to 14 mils (which is a standard depth of attack), and a corrosion layer was found in the cold leg. This was the fifth loop operated with the fluoride NaF-KF-LiF-UF to which chromium metal was added by some Of the five loops, four definitely did not show any improvement in corrosion resistance, and the improvement of the fifth is doubtful, An attempt will now be made to add chromium metal to NaF- Zr'FA-UF4 (50-46-4 mole %) to determine whether an improvement can be obtained in this system. The snubber served os a very coarse filter, means. Three loops were operated with material from the same batch of NaF-ZrF4-UF4 (50-46-4 mole %), but NiF, was added to two of them. The additions were made in the fill pot, which was agitated for 1 hr before the loop was filled. The attack in- creased with increasing amounts of NiF,. The control loop showed moderate aitack of up to 8 mils. With the addition of 0.05% NiF,, the attack was reported as heavy, with a moximum depth of TABLE 7.9. EFFECT OF TEMPERATURE OM CORROSION OF INCONEL BY I«h::F-ZrFA-UF-"4 (50-46-4 mole %) HOT-LEG TOEMP ERATURE HOT-LEG ATTACK °F) 1650* Light to moderate subsurfuce voids of 4 to 11 mils; voids larger than normal and concentrated in grain boundaries 1500* Light to moderate subsurface voids of 3 1o 10 mils; general attack fo 4 mils 1300 Moderate to heavy attack of 4 to 9 mils; voids very small and well distributed 1500 Moderate to heavy attack of 7 to 13 mils 1650 Moderate attack of 3 to 12 mils; voids very large and concentrated in groin boundories (5) *Results reported previously. 60 PERIOD ENDING JUNE 10, 195 TUNCLASSIFIED PHOTO T-3088 1§ - P T . . y : k L e ’ ? A el ey 9 3 . - K B o~ - i & - . . By . T * P . - A - h , S 1 ‘ T il Y . 5 s 5 4 9 - - - . e . : - » } i - S e ] . - o 5 - ' - 2 . . . . . - e : s % . 7 —2 ‘. Y o M""‘s . ‘v i & . - MY ' . s s} o’ N Sl po 4 e P e} 3 » . .. -'~.$ - ARSI '_‘r‘ s » 's,, _‘_‘V 5 I o e 2 cet ¥ ~_‘*a ‘14_}: N . v A ‘.‘ - . v . \!’-; ® ’ e e ol )\. . . ‘o‘ . -7 T ' - - i s . . ‘ o . » ¥ o Y » . e a W% ,. . Sy v @ . ’ & Tl e BT oA N ! SN ., _— . e e -'.f.*fi Y 'n.l-.:‘,._ . ‘.’“. o, fl;'\‘., vt Ty "‘,0" 5 eV PEN 2 e e i S SN " - e . Lo S e . . T ~ o .‘- ¢ "“_6 N . W -8 ". e . o [ s, Y e e 7 - . . . ‘ “" - ' \A~ - : ;' ;:. =’ \" e ""- i ‘ ‘. . 4 - l“ ) .’ . » - .1 . ¥ ; - . Lo T . s b . T e - . ’ . L s . —7 ™ . \ - "" o sy‘ » > 4 - : ","\ ’é 7 . oy » : " L ., - - ® » - ' . ,‘ L 2 " s o2 (- w . o~ . S - e g o e sl " ] | v M ' - . _ __..9. % ! % - ! ‘(0) v MILS UNCLASSIF [ED : PHOTO T-3071 - Crecepr @t ‘ | LW TN —_ x\ fui\. e T A N ; S L. ’r ‘ g . - !‘ _ v 3 » . . . 1 g e o - S e . - ..* ¢ A, e e e aN - T §: -~ 4 * T N VyoenT £ s € Y L8 ,: . ) . fr 'J}‘. L A R .. o 0 0. ’”,o . . P v re #a. —3 T « 0 : . o ‘ UNCLASSIFIED PHOTO Tf3083 » 9§ : » \. ¢ . e ‘ # ‘ , 4 . - . . ‘ ,f . = % ’g‘. - ‘ .'q. ’ - \ ‘w 5 » ’ /‘ ¥ ' ’ : * — . ¢ ¢ 7 - . 5 —8 o . ; * % . + ) ) ~4 ' : A" . .7 . ‘ L e ey, "o ® MILS Fig. 7.5, Corrosion of Inconel by NaF-ZrF -UF, (50-46-4 mole %) After 500 hr As a Function of Temperature, (a) 1300°F. (b) 1500°F. (c) 1650°F. 250X 61 ANP PROJECT QUARTERLY PROGRESS REPORT 10 mils. With a 0.2% addition, a rough surface and heavy attack of up to 11 mils was found. The nickel analyses of three samples taken as the loops were filled were reported as 120, 280, and 500 ppm, respectively. It has not yet been de- termined whether insufficient time was allowed for all the nicke! fluoride to go into solution. In the samples with the NiF, addition, the iron content increased from 620 to 1100 ppm, while the chromium, which is the usual reaction product, only increased from 230 to 620 ppm. Fuel Purity. Several loops were operated with high-purity botches of NaF-ZrF ,-UF, (50-46-4 mole %) prepared in graphite-lined vessels. One loop contained o batch of high-purity fuel prepared from standard materials, while two other loops were filled with fuel prepared from hafnium- free zirconium oxide, which was low in iron im- purity. However, the actual chemical analyses showed the iron and the nickel contents to be about the same in all three charges, that is, a total of about 300 ppm in each. In addition, one of the loops with fuel prepared from hafnium-free zir- conium oxide was cleoned by circulating NaZrFS. In the specially cleaned loop, the attack was less than normal, with @ maximum penetration of 5 mils; in the other two loops, the attack was normal, with a maximum penetration of 9 mils, However, the corrosion attack of the high-purity fuel batch was not measurably different from that of the standard fuel in the comparable test. In another test, two batches of fuel that had received no purification other than a vacuum drying during melting were run in two other loops., Heavy attack, with a maximum penetration of 15 mils, was found in both loops, This is the deepest pene- tration found in any loop in which the fluoride Nc1F-ZrF4-UF:4 (50-46-4 mole %) has been circu- lated. It is apparent from these tests that although purification of the fuel is desirable, the present standard production procedures appear to be ade- quate. Little, if any, additional reduction in attack can be obtoined with additional fuel purifi- cation processes, special However, the contaminants (FeF, and NiF,) that remain in the fuel may be further reduced by additions such as ZrH,,. Loop Oxide Films. To determine the effect of oxide scale on the maximum depth of attack, several loops that had been subjected to various surface treatments were operated for 500 hr at 62 816°C with fluoride NaF-ZrF ,-UF, (50-46-4 mole %). The various treatments included (1) circu- lating NaZrF_ for 2 hr ot 1400°F, (2) degreasing, and (3) oxidizing before filling. The maximum attack in the three loops was 6, 7, and 8 mils, respectively. The attack on the NaZrF .-cleaned loop and the attack on the oxidized loop are shown in Fig. 7.6. Although the treatment with NaZrF . removed the oxide film from the loop, there was a brown film on all sections of the loop which varied in thickness from thin in parts of the cold leg to granular in the hot horizontal section. A diffraction test showed the film to be over 90% Na,ZrF,, with some zir- conium oxide present. The effect of this compound on the fuel is not known. No measurable attack took place during the cleaning. small Removing the amounts of oxide film by treating with NaZrF, results in what oppears to be o slight reduction in depth of attack; however, a large oxide content did not cause a large increase in depth of attack. Nonuranium Bearing Fluorides, During the test- ing and filling of the ARE system, flucride mixtures which do not contain uranium will be used. Cor- rosion work on such mixtures was started in the previous quarter and was continued. The corrosion results contirm those previously reported in that nonuranium bearing fluorides are found to be less corrosive than the fuel mixture. With the fluoride NaZrF, alight attack with a maximum penetration of 5 mils was found. In the cold leg, the maximum attack was 2 mils. A maximum attack of 5 mils was al so found in a second loop, which was similar but was allowed to circulate for 1000 hours. A very thin corrosion layer was found in the cold leg of each loop. Such a layer is not usually discerni- ble with NaF-ZrF,.-UF, (50-46-4 mole %). The nature of the attack is the same with or without the uranium in the mixture, FLUORIDE CORROSION OF NICKEL THERMAL CONVECTION LOOPS G. M. Adamson, Metallurgy Division The fluoride NaF-ZrF4-UF4 (50-46-4 mole %) was circulated in a nickel thermal convection loop which was cleaned with dry hydrogen. After 500 hr at 816°C, o few metallic crystals were found in the trap and in the fluorides in the lower part of the cold leg. An extremely thin corrosion layer PERIOD ENDING JUNE 10, 1953 | UNCLASSIFIED | PHOTO T-3148 + . ' L S e Y MILS . . . - . - L - - t (£) Fig. 7.6. Effect of Oxide Scale on the Corrosion of Inconel by NoF-ZrF -UF , (50-46-4 mole %) Atter 500 hr at 816°C, (a) Deoxidized with NaZrFSV (b) Oxidized. 250X 63 ANP PROJECT QUARTERLY PROGRESS REPORT was found on the cold-leg surfaces by metallo- graphic exomination. The hot-leg surfaces were quite smooth, with no intergranular or subsurface voids. The wall of the hot leg was 0.002 in. thinner than that of the cold leg, but since the thickness was still within commercial tolerances, no definite conclusion may be drawn from this one measurement. Sections of both the hot leg and the cold leg of this loop are shown in Fig. 7.7. A small amount of mass transfer took place, but it was less than that found with type 316 stainless steel and many times smaller thon that found when the fluoride NaF-KF-LiF-UF, (10.9-43,5-44.5-1.1 mole %) was circulated in nickel. FLUORIDE CORROSION OF TYPE 316 STAINLESS STEEL THERMAL CONVECTION LOOPS G, M. Adamson, Metallurgy Division Two type 316 stainless steel thermal con- vection loops operated for 500 hr at 1500°F with NaF-ZrF ,-UF , (50-46-4 mole %)} without plugging. These were the first loops of stainless steel to circulate uranium-bearing fluorides at 1500°F with- out plugging. Typical hot- and cold-leg sections from these loops are shown in Fig, 7.8. The hot legs in both loops were very rough, with both the intergranular and the typical Inconel types of attack. In loop 133 the depth of attack was 5 to 11 mils, while in loop 134 it was 4 to 7 mils. Some grains had been removed from the surface, but the average thickness had not been reduced. In both loops, a heavy multilayer deposit was found in the cold leg. The top layer of the deposit consisted of fine, dendritic, metallic crystals, These crystals were also found in the fuel. Diffraction and spectrographic studies show these crystals to he mainly iron, with some chromium in solution. The zirconium content was higher than would be found if fuel particles were mixed with the crystals. Both the nicke!l and the molybdenum contents were very low. Under the first layer was another layer which appeared to be nonmetallic, and in some areas there was another metallic layer beneath the nonmetallic layer. The attack found in these loops appears fo be much more comparable to that found with Inconel than was the attack in the loops that plugged with the fluoride NaoF-KF-LiF-UF, (10.9- 43,5-44.5-1.1 mole %). 64 LIQUID METAL CORROSION . A, Knox E. E., Ketchen L.. G. Overholser Materials Chemistry Division J. V. Cathcart D. C. Vreeland G. P. Smith E. E. Hoffmon W. H. Bridges L.. R. Trotter J. E. Pope Metallurgy Division L. A. Mann D. R. Ward ANP Division Mass Transfer in Liguid Lead., Studies of the mass transfer of various solid metals in small thermal convection loops containing liquid lead have continued. These tests have been performed in thermal convection loops made of quartz tubing. Samples of the test metals are fastened in the hot and cold legs of the loops. Details of the con- struction of the loops and of the results obtained for Inconel, columbium, molybdenum, and type 304 stainless steel were reported previously, (67} Armco iron and types 347 and 446 stainless steels were tested during this quarter., The loop containing type 347 stainless steel was operated with hot- and cold-leg temperatures of 800 and 500°C, respectively, and failed after 140 ht because of the formation of a plug in the cold leg. Figure 7.9 is a cross section of the hot leg of this loop; the intergranular type of corrosion which occurred, in addition to the mass transfer, can be seen. A similar, but less marked, corrosive attack occurred in the cold leg. The loop containing Armco iron failed because of plug formation after 250 hr of operation. The hot- and cold-leg temperatures were 800 and 545°C, respectively, As may bhe seen from Fig. 7.10, which shows a cross section of the hot leg of the loop, the iron specimen suffered almost no inter- granular attack. A decrease in wall thickness of 0.003 in. was measured for the hot-leg specimen, There was no change in the wall thickness of the cold-leg specimen, A marked increase in resistance to mass transfer was noted in the loop containing type 446 stain- less steel. This test was terminated on schedule after 619 hr of operation with the hot leg at 810°C and the cold leg at 520°C. Although insufficient )G, p, Smith et al., ANP Quar, Prog. Rep. Mar. 10, 1953, DRNL-1515, p, 128, 7)Met. Prog. Rep. April 10, 1953, ORNL-1551 (in press)e PERIOD ENDING JUNE 10, 1953 UNCLASSIFIED PHOTO T-3149 » \‘s. 7 ™, ; ‘ \\ , : -9 Y, : ' -c;: \ N 4 - T R [+ % A e 5, o o ., s \% e - ‘°‘1\ / \."\ - M | L“S ! A o * ) // ’ » i £ //‘ 5 & & /,‘ r“‘d e & W . -// UNCLASSIFIED ' PHOTO T-3150 —9 —10 MILS (£) Fig. 7.7. Corresion of Nickel Thermal Convection Leop by NaF.ZrF .UF ( (50-46-4 mole %) After 500 hr at 1500°F. (a) Hot leg. (b) Cold leg. 250X 65 ANP PROJECT QUARTERLY PROGRESS REPORT UNCLASSIFIED PHOTO T-2709 v T e ., M_'j;,,“ K. . ¢ Fig. 7.8. Corrosion of Type 316 Stainless Steel Thermal Convection Loop by NaF-ZrF4-UF4 (50-46-4 mole %) After 500 hr at 1500°F. (a) Hot leg. (b} Cold leg. 250X 66 PERIOD ENDING JUNE 10, 1953 UNCLASSIFIED PHOTO Y-9255 INCHES 0.005 - (500 Xx) Fig. 7.9. Corrosion of Type 347 Stoinless Steel by Lead in Quartz Thermal Convection Loop After 140 hr ot 800°C. 500X : _ : : C . 3 s H}i - [ - ! e - s w £ , - 1 T - 5 e U . - %, - ,“" ‘:(:.-Bs ‘ > ’:\ » i% “a . A % - - _ % e ™, & t bt - 1&} ” : \'\,,\w‘*lk' « -v_‘_“ - - : T( i » . 3, £ ? e . - N ey - 10.005 i E »» ) ~m L 3 / e Y flwwwrwn”a - | & ";“ o g : oy # - ; - ' - s » - — v § v R, k ’ . B / . : % e / % v - 1 ' o Fig. 7.10. Corrosion of Armco Iron by Lead in Quartz Thermal Convection Loop After 250 hr ot 800°C., 500X ' 67 ANP PROJECT QUARTERLY PROGRESS REPORT mass transfer occurred fo form a plug large enough to stop the circulation of lead in the loop, a small quantity of mass transferred material collected at the bottom of the cold leg. It is to be presumed that had the loop been dllowed to operate for a sufficiently long time, this deposit would have grown to sufficient size to stop the circulation of lead in the loop. The results obtained thus far indicate that alloys with a high nickel content — for example, |nconel and type 347 stainless steel ~ show little resist- ance to mass transfer or corrosion in lead. On the basis of the results with type 446 stainless steel, it appears that stainless steels containing no nickel suffer relatively little mass transfer as compared with the nickel-rich alloys, Columbium and molybdenum showed no mass fransfer in lead. BeQ in Sodium and NaK. The compatibility of beryllium oxide with sodium or NaK is of such concern to the ARE that the problem is being attacked simultaneously in seesaw, rotating, and thermal convection loop tests, with both treated and untreated beryllium oxide specimens. The treated specimens were sprayed with slurries of various fluorides, including CaF,, CaF -AIF ,, and CaF ;-MgF,, and then dried. In general, the cootings seemed to have some beneficial effect in reducing corrosion, Each of the tests was operated with a hot-leg temperature of about 1500°F for 100 hr in the seesaw and rotating tests and for 212 hr in the convection loop tests. The results in terms of weight change of the 1{‘- by ‘4- by ]/Q-in. beryllium oxide specimens vary from 1% in the seesaw tests up to 10% loss in the convection loop tests and to 71 to 100% loss in spinner tests., The sodium velocity in the rotating test was 400 fpm and, in thermal convection loop tests, it was only of the order of 6 to 10 fpm. The data obtained are summarized in Toble 7.10. It now appears that the corrosion mechanism is, in reality, a mechanical erosion of the hot-pressed BeO surface. Therefore the ARE has been designed so that the sodium in contact with the BeO will be virtually stagnant, In an attempt to determine where the missing beryllium oxide ends up, two convection loops were sectioned for analysis. Each loop had ope- rated for 212 hr ot 1500°F; one loop contained sodium and the other NaK, The distribution of the residual beryllium oxide in the liquid metal drained from the loop, in the wash solution, and on the wall is given in Table 7, 11. These data indicate that the beryllium oxide is mainly suspended in the liquid metal rather than being firmly attached to the walls, as was previously reported. Coated Beryllium in Sodium. A series of beryl- lium specimens, some of which were plated with TABLE 7.10. EROSION OF BERYLLIUM OXIDE BY SODIUM AT 1500°F TEMPERATURE (°C) WEIGHT LENGTH TYPE OF TESTY SPECIMEN CHANGE OF TEST Hot L.eg Ceold Leg (%) (hr) Seesaw A|F3~C0F2 treated BeO 825 760 -1.1 100 AIF3'CGF2 treated BeO 820 675 +0.20 100 Special grade of BeO 800 650 +0.89 100 Special grade of BeO 805 720 +0.32 100 Rotating A|F3-—C0F2 treated BeO 816 816 ~100 100 AIF3-COF2 treated BeO 816 816 -71 100 Special grade of BeD 816 816 —~94 100 Special grade of BeO 816 816 -90 100 Convection loop Untreated BeQC 816 600 -16.8 212 A|F3-C0F2 treated BeO 816 600 ~9.8 212 A|F3-C0F2 treated BeQ 816 625 ~1.8 212 C<:|F2 treated BeD 816 650 ~10.5 212 CaF, treated BeO 816 650 -7.3 212 MgFZ-Cch treated BeD 816 600 0.0 212 68 PERIOD ENDING JUNE 10, 1953 TABLE 7.11. DISTRIBUTION OF BERYLLIUM OXIDE RECOVERED FROM CONYECTION LOOPS RECOVERED BeO (%) LiQUID METAL CIRCULATED On Wall In Washings in Ligquid Metal NaK 1.2 36.9 61.9 MNa 2.4 77.2 20.4 chromium and others with nickel at Gerity-Michigan UNCLASSIFIED Company and at Y-12, were tested in sodium in PHOTO Y-8612 Inconel tubes for 100 hr in the seesaw furnace. Tubes were crimped so that the specimens were restricted to the hot zone. All specimens showed attack, and in only one nickel-plated specimen was the plating at all adherent. Since both chro- mium and nickel have good resistance fo sodium, ‘E! o e it is assumed that the poor showing of these i T A W AR specimens is due fo porous plating. The appear- '.(q .HQT ZE _ 10°C ance of some of these specimens after test is COLD 7ONE — 640°C shown in Fig., 7.11, Carboloy in Lead and Sodium, Static tests: of several Carboloys have been run with sodium and lead in Inconel tubes for 100 hr of 816°C. Except for one test with Carboloy 608 in lead, the Carbo- (b) HOT ZONE - 830°C loys (907, 779, 44A, aond 55A) appeared to have COLD ZONE —-T770°C good resistance both to sodium and to lead. While Carboloy 608 was satisfactory in sodium, when in lead, a light color phase appeared to be preferen- tially attacked to a depth of 26 to 28 mils. The ) brittleness of these alloys, however, appears to (c) HOT ZONE—830°C make them prone to cracking and spalling at the COLD ZONE—725°C edges, as can be seen in Fig. 7.12, It is not known whether this spalling takes place during testing or Fig. 7.11. Corrosion of Chromium-Plated Beryi. during metallographic preparation. fium by Sodium After 100 hr in Seesaw Test, 69 ANP PROJECT QUARTERLY PROGRESS REPORT R " . UNCLASSIFI%? i - kS PHOTO Y-8872 | » - G TR gy g P . P .. ' 4 e T TR ., N % , . s . ) . el - -7 - - S ¥ T ™ : * . . . * * Q - . 7 - * ‘ . > v' * * — 2 4 4 oy . s - e ~ 4 = ’ ‘. r ’ * ‘ ‘ . - . * - ’ ‘.‘ - . ‘ ' . ‘ - . . - g — ; " - . . ( . . ’ ¥ . . é t 4 , - . v . % - Q005 - - - . e et ‘ . . A P . . , . Lo . 4 - . - R » e . e ., 4 ", . .. . - % e . . % . - o — » , o, . . - - - - . - ¢ .\ o - -+ - ' - ’ » . - . oo ’ - z . o v~ v 0o N I & - ¥, folee TN e x 1 1N ‘ . ’ L , \' » ) o ’ ~ . » ¥ 1. ” . - - w7 : . . N e - ’ ' 0 SN - % 1 . $ - ~ * .,! TN .D o « 1 . poL , L a o . ., Nt L - “ . ' . . “fi. s + . ‘.l f !\ . 4 - , . . ’ , : - ‘ f‘l {\ e { o ' r : .‘ ’ .- ;c . =~ . ’ - 2 v ¥ &* " - -* < . AN Ul T ) . - e~ " 0010 ; * » ” + 4 - :'.. 2 : . " 3 . . " - 2 . t - 4 L - 4 . . o . .. ' - - r - ot P - - 13 “a (‘ L. ? ¢ - y . f’ LHoe -”“ . * - " - 4 - » ~ - . - * . - F » * . ” My . : - » * . € - “ ". . ~ ' s - il . , - e . i :‘ ‘. 4 ’ r . B . -, 5 a 1 - “ > » e ' -4 ! woN ¢ ‘n‘ 't ’ . - . Fig. 7.12. Carboloy 608 After Testing in Sodium for 100 hr ot 816°C. Note cracking and spalling at upper edge. 250X 70 PERIOD ENDING JUNE 10, 1953 8. METALLURGY AND CERAMICS W. D. Manly J. M, Warde Metallurgy Division The fabrication of an intricate liquid-to-liquid heat exchonger was effected by manual inert-arc welding of tube-to-header welds, The individual tube bundies were reinforced ond spaced through the use of grid-type spacers fabricated by cone-arc wire-to-header welding techniques. In a study of the effect of the various brozing vor- iables on the flowability of the Nicrobraz brazing alley, it was found that contamination of the brazing otmosphere with nitrogen has a serious effect on the flowability of the alloy. Dilution and diffusion studies on three high-temperature brazing alloys ~ Nicrobraz, 60% Mn~40% Ni, and 82% Au-18% Ni —~ have been completed. Brozed tube-to~fin joints were prepared to study the effect of the quantity of brazing alloy used, the bruzmg temperature, and the brazing time, A technique has been developed to braze ce- mented coarbide rings to shafts for use os pump seals. Standard brazing techniques could not: be used since there is approximately u factor of four differerce between the thermal coetficient of ex- pension of Carboloy ond that of stainless stesl. Additional information on the elevated temperature strength of Nicrobrazed Inconel and type 316 staine less steel joints hos been obtained. The study of the effect of different environments such as air, argon, hydrogen, and the fused fluoride salts on the creep and siress-rupture properties of Inconel has continued, and o wide variation of properties haos been found. This study is being extended to include austenitic stainless steels, ferritic stainless steel, ond cobalt-base alloys. Preliminary results indicote that Inconei and nicke! are much more sensitive to environmental changes than are the austenitic stainless steels, such as types 316 and 304. A design curve for Inconel tested in the fused fluorides at 1500°F has been completed over the stress range of 2500 to 8000 psi. Special high-purity alloys of the Inconel type are prepared in o vacuum melting furnace ond then extruded to form a draw blank. The drow blank is then reduced to produce 1/2-in.-c!ia 0.035-in,-wall tubing for corrosion tests. The activation energy for the oxidation of columbium in air was found to be 13,400 cal/mole for the temperature range of 600 to 900°C and 4350 cal/mole for temperatures above 900°C. The oxidation resistance of chro- mium on OFHC copper, chromium and nickel elec- troplate on copper, and Incone! and type 310 stain- fess steel clad on copper has been studied for the production of high-conductivity fins for an ANP type of radiator, Special oxidotion-resistant alloys of copper have been prepored, and their oxidation characteristics have been determined. Work on the fabricaotion of control rods for the GE-ANP program has continued. Aftempts to braze hot-pressed segments of B, C-Fe mixture fo stain- less steel tubes were not successful; therefore an alternate method of fabrication is being tried which consists of tilling a tube with a mixture of A_I-BAC or Cu-B,C powders and reducing to final size by swaging and hot rolling to consolidate the mixture, as well as to form a bond between the matrix and the tube wall, Powder metallurgy compacts of various metals with additions of MoS, were pre- pared for use as pump seals, Additional work on the preparation of spherical particles of uronium- bearing alloys has been performed with the use of the following methods: 1. heating in a vacuum to gbove the melting point, 2. dropping particles through o three-phase three- electrode arc, which provides the heat source, 3. forcing the molten alloys through a small ori- fice, 4. letting precut pqrtlcles of the alloy settle through o molten salt both with o temperoture gradient that extends beyond the liquidus and solidus of the alloy., Research has been initiated to determine to what extent the burning of jets of liquid sodivm cen be reduced by alloying the sodium with some of the heavy metals. In the initial phase of this project, a large number of alloys will be surveyed quali- tatively, Preliminary experiments indicate that alleying with lead, mercury, or zinc does not ap- preciobly change the inflammability of sodium at 800°C. A collection of this type of qualitative data will serve to indicate whether a detailed study of the numerous variables of the inflamma- bility process will be worth-while. 71 ANP PROJECT QUARTERLY PROGRESS REPORT A cermet has been developed in which 10% USi, is included in the central portion of a silicon car- bide fue! element. Other ceramics research has included @ study of the heating rate of beryllium oxide. |t appears that the safe heating rate is less than 200°C per hour, WELDING AND BRAZING RESEARCH P. Patriarca Y. G. Lane G. M. Slaughter C. E. Shubert Metallurgy Division Welding of Heat Exchanger Tube Bundles, Pre- liminary cone-arc welding experiments for deter- mining the feasibility of the consiruction of the six-tube-bundle liquid-to-liquid heat ex- changer were described in a previous reporh“) Further efforts to apply the process to this fabri- cation merely verified the evidence that the pro- duction of relichle, consistently leak-tight cone- arc tube-to-header welds was complicated by the curvature of the dished headers of the heat ex- changer. Rather than to delay the production of the heat exchanger, several experimental headers were fabricated by manual inert-arc welding., It was found that a qualified operator can fabricate consistently leak-tight tube-to-header welds by the use of manual welding techniques. Consequently, the manual welding of the six tube bundles come prised of 35 tube-to-header welds on each end of each bundle was undertaken, pending further de- velopment of the cone-arc welding process. One tube bundle consists of 35 Incone! tubes, 0.148 in. in outside diameter with 0.025-in.-thick walls, welded on each end into a 0.125-in.-thick dished (4-in. radius) heoder. FEach bundle was helium leak checked and Dy-Cheked prior to weld- ing of the nozzle assembly. The nozzle was at- tached by manual inert arc welding. Six of the 420 tube-to-header welds were found by the helivm leak detector inspection procedure to contain leaks, which were subsequently successfully re- paired, Since the tube bundles contained a free span of approximately 40 in., the design required that grid- type spacers and clamps be used at 8-in. intervals to provide reinforcement of the bundle and to ensure free flow of liguid around the tubes. A cone~arc wire-to-header plug-welding technique was lncone! ”)P. Patriarca, G. M. Slaughter, ¥. G. Lane, and C. E, Shubert, ANP Quar. Prog. Rep. Mar. 10, 1953, ORMNL-1515, p. 140, 72 developed and applied to the fabrication. Svitable jigs were used to provide the dimensional toler- ance required with respect to curvature and spac- ing, Figure 8.1 shows three completed tube bundles welded into a half section which will constitute the cylindrical vessel of the six-tube-bundle liquid- to-liquid heat exchanger when welded to its mate, : UNCL ASSIFIED s 2. PHOTO Y.90460 Fig. 8.1. Three Tube Bundles Welded into Half Section of Cylindrical Six-Tube-Bundle Liquid-te- liquid Heat Exchanger. Note spacers, Brozing of Air Radictors, In the fabrication of previous heat exchanger test assemblies by furnoce Nicrobrazing,(” experience indicated that furnace temperatures of 2300°F were necessary to obtain suitable flowability of the brazing alloy. At braz- ing temperatures below 2300°F, a granular braze was often obtained, and as o result, the joints leaked. PBrazing ot 2300°F wos found to cause severe dilution of the parent material and inter- granular dilution and/or diffusion, which resulted in severe embriftlement of the parent assembly, The design of the heat exchanger test units usu- ally required a multiple brazing operation. Conse- quently, leaky assemblies that were practically impossible to repair were sncountered on the sub- sequent rebrazing operations, possibly because of excessive dilution and/or cracking of previously embrittled joints as a result of the thermal stresses set up during the rebrazing operation. to remedy the deleterious effects of the high braz- ing temperature, a series of tests was run to de- termine the cause of poor flow at normal brazing temperatures. Brazing cycles were accurately determined by In an effort multiple temperature measurements made with chro- mel-alumel thermocouples encased in impervious ceramic tubes. The dew point of the hydrogen atmosphere was measured ond verified as being less than —70°F, The effect of rate of rise of temperature on diffusion of boron and the effect of boron on flowability were also experimentally discounted as major factors responsible for poor flow when the clloy is used in the form of a thick slurry and the contact area for boron diffusion is small. A determingtion of the effect of nitrogen as an impurity in the hydrogen atmosphere, however, produced significant results, There is indication that elimination of the hydrogen purge being used in the brazing operation is necessary. Controlled experiments performed with the use of a small muffle furnace revealed that the presence of ni- trogen impurity in the hydrogen resulted in a defi- nite impedance to the flow of Nicrobraz by very rapidly forming a poorly reducible nitride ond/or a new brazing alloy with a higher melting poini. These experiments led to the conclusion that in future can-brazing operations a helium purge should be substituted for the hydrogen purge. To verity this thinking, a smail 15-fin/in. heat exchanger was rebrazed after two previously unsuccessful operations at 2200 and at 2300°F. The joints were prepared with a fresh slurry of alley and brazed of PERIOD ENDING JUNE 10, 1953 2150°F for 40 minutes. tained, and helium ledk testing did not detect the presence of any leaky joints. Future work will further verify whether complex heat exchangers can be fabricated in a single brazing opsrotion at a reasonable temperature range (between 2050 and Z2150°F) in which the excessive dilution, embrittlement, and development of leaks experienced with the multiple brazing operations performed ot 2300°F can be avoided. Excellent flow was ob- Dilvtion and Diffusion of Brazing Alloys, It was indicated in a previous reporf(” that studies were being made of dilution and diffusion of three high- temperature brazing alloys — Nicrobroz, 60% Mn- 40% Ni, and 82% Au~18% Ni. Brazed fube-to-fin joints were prepared for a study of the metallo- graphic effects of (1) the quantity of brazing alloy used, (2} the brazing temperature for o given time, and (3) the brazing time for a given temperature, The joints were prepared from 0.010-in.-thick type 302 stainless stee! fin material and 0.150-in.-0OD 0.020-in,-wall type 304 stainless steel tubing. Smoll, medium, and large amounts of brazing alloy were investigated, while the time ot tempera- ture was varied from 10 min to 18 hr for an over- night brazing cycle. The brazing temperature was varied from 20 to 120°C above the melting point, The results of metollographic examination of the test specimens are presented in Table 8.1 as average percentages of the tubes wall thickness actually dissolved during the brazing. The specimens brozed with Nicrebraz alioy olso exhibited o diffusion zone which sometimes ex- tended completely through the wall thickness, The maximum penetration, which woas primarily intere granular diffusion of clioy constituents, was re- corded, but the scatter of the data was such gs fo No dif fusion zones wider than 0.5 mil were found in specimens brazed with the gold-nickel ond nickel- manganese olloys, make a definite correlation impossible, As a result of this investigation, it became op- parent that base-metal dilution increases rapidly with the use of increased amounts of Nicrobraz alloy. Severe alloying of the tube plus complete collapse of the fin is shown in Fig. 8,20, which is a photomicrograph of a joint brazed at 1220°C for 30 min with a large quantity of alloy, tube-wall and fin dilution can be seen in Fig, 8.25, which is a photomicrograph of a joint brazed under identical conditions but with less alloy. Dilution Much less ANP PROJECT QUARTERLY PROGRESS REPORT TABLE 8.1, RELATIVE AMOUNT OF DILUTION OF VARIOUS ALLOYS INTO 0.020-in.-WALL 0.187-in.-0D TYPE 304 STAINLESS STEEL TUBING TYPE AND QUANTITY EXTENT OF DILUTION OF STAINLESS STEEL OF BRAZING ALLOY TUBING (%) At 1220°C T At 1180°C At 1125°C Nicrobraz 1 ring of 31-mil wire* 8 3 2 2 rings of 31-mil wire* 28 15 10 3 rings of 31-mil wire* 26 26 5 At 1130°C At 1080°C At 1030°C 60% Mn—40% Ni 1 ring of 21-mil wire 3 1 0 1 ring of 32-mil wire 11 4 0 1 ring of 45-mil wire 15 7 2 At 1130°C At 1030°C At 980°C 82% Au~-18% Ni 1 ring of 21-mil wire 8 9 4 2 rings of 21-mil wire 13 10 6 1 ring of 37-mil wire 19 19 12 *Mixture of Nicrobraz alloy powder in acryloid cement binder. also increased rapidly with increasing brazing temperature, but it was virtually independent of time at temperature. Intergranular diffusion during Nicrobrazing was greatly increased with the use of high temperatures and/or the use of large quantities of brazing dlloy. Time seemed to affect the depth of the diffusion zone somewhat, but no definite relationship could be determined from this investigation, Dilution of the stainless steel by the manganese- nicke! and gold-nickel alloys increased substan- tially with both temperature and gquantity of alloy, but time ot temperature did not seem to be an important variable. In general, dilution was greater with the gold-nickel alloy than with the manganese- nickel alloy. Brazing Carboloy to Stainless Steel., It has be- come necessary to braze cemented carbide rings to stainless steel for use as rotating mechanical pump seals for operation at 1500°F in both fluoride and air environments, Standard brozing techniques re- sult in severe cracking of the brittle Carboloy upon cooling becouse of differences in coefficients of 74 thermal expansion. Carboloy, for example, has a representafive coefficient of thermal expansion of 5 x 107¢ in./in.-°C, while stainless steel has a representative value of 20 x 107° in./in..°C. Nicrobraz, which is highly resistant to high- temperature oxidation and moderately resistant to tluoride aottack, was used in preliminary investi- gations. However, severe distortion of the stain- less steel with subsequent fracture of the Carboloy occurs when standard techniques are used (Fig. 8.3). A small rod of the cemented carbide fractured completely along its length. A nickel shim, which was placed between the two parts to be joined to reduce the effects of contraction, was so em- brittled by the Nicrobraz that yielding was pre- vented and the Carboloy fractured. Thus it became evident that a ductile brazing alloy with the re- quired corrosion resistance was needed. An 82% Au-18% Ni alloy was selected, and actua!l as- semblies were brazed by wusing 0.020-in. rings of brazing alloy, together with various thicknesses of nicke! shim material. It was found that a nickel thickness of ]/8 in. was needed to consistently PERIOD ENDING JUNE 190, 15953 E UNCLASSIFIED ! E PHOTO Y-B981 § UNCLASSIFIED B | proTO Y-8982 § Fig. 8.2. Stainless Steel Tube-to-Fin Joint Nicrobrazed at 1220°C for 30 min, (o) With a large quantity of brazing alloy, severe tube wall dilution results. (b) With a small quantity of brazing alloy, little tube wall dilution is evident. Eiched‘ with aqua regia. 50X 75 ANF PROJECT QUARTERLY PROGRESS REPORT UNCLASSIEED PHOTO Y-8780 INCHES i \tl_’__LLE i l..l..l..l_l._l...i‘. ll’ i f_!_i 0 w 2 Fig. 8.3, Carboloy Rod Nicrobrazed to Type 316 Stainless Steel, produce cracksfree pump seals. Cooling rates of approximately 1000°F/hr were also employed to decrease cracking tendencies. One successfully brazed seal consists of a Carboloy ring brazed to a nicke! shim which is in turn brazed to the stoin- less steel body. It is expected that similar techniques may be applied advantageously in joining other materials with widely different coefficients of expansion, High-Temperature Brazing Alloy Evaluatien Tests. The results of several room- and elevated- femperature butt-braze tensile tests on Nicrobrazed Inconel and type 316 stainless stee! joints indi- coted that the low-temperature properties of stain- less steel joints were definitely superior to those of Inconel joints. However, the difference de- creased rapidly above 1200°F, as determined from a plot of tensile strength vs. testing temperature. A summary of the existing data is given in Table 8.2. Each value listed in the table is an average of five tests on 0.252-in.-dia test bars. The tests were shori-time tensile tests on buit-brazed joints that were machined under nearly ideal conditions. It should be remembered that the strength properties of actual joints may vary quite markedly from those shown for these more ideal cases. thermal CREEP AND STRESS-RUPTURE TESTS OF STRUCTURAL METALS R. B, Oliver C. W. Weaver D. A, Douglas J. W. Woods Metallurgy Division The effects of various environments, such as air, argon, flucrides, and hydrogen, on the creep and stress-rupture properties of Inconel were investi. goted, and wide variations in properties were found. This study is being extended to include the following additional materials: types 316, 304, and 405 stainless steel, nickel, and Hastelloy “C", Table 8.3 shows the results of the tests made to date, A stress was picked for each ma- terial which would cause rupture in an argon atmosphere in the time range of 700 to 1000 hours. The data seem to indicate that there is little environmental effect on either type 316 or type 304 stainless steel under the present testing con- ditions, An A nickel specimen which is still in test in an air atmosphere indicates that nickel may hove considerably more creep strength in air than in any of the other environments presently being used. If this is true, it is possible that the strengthening agent or mechanism is the same for nickel as Incaonel, which exhibits a similar effect. Fourther tests are in progress so that a more com- plete evaluation of the environmental effects can be made, As a part of this program, fine-grained Inconel was tested at 3500 psi in alternate cycles of argon and hydrogen. The first 200 hr of the test was in argon, ond a constant creep rate was reached before the first hydrogen cycle was started. Each hydrogen cycle resulted in on accelerating creep rate. Each argon cycle tended to maintain es- sentially the same creep rate as that observed at TABLE 8.2, TENSILE TESTS OF BUTT-BRAZED JOINTS TENSILE STRENGTH (psi) TEMPERATURE (°F) Stainless Steel Joint Inconel Joint Room 68,500 35,000 1000 60,500 33,000 1500 26,000 24,500 1700 16,000 13,500 76 PERIOD ENDING JUNE 10, 1953 TABLE 8.3. RESULTS OF CREEP AND STRES5S-RUPTURE TESTS OF STRUCTURAL METALS MATERIAL STRESS RUPTURE TIME (hr) CREEP RATE (%/hr) {psi) in Air In Argon | In Hydregen |In Fluorides In Air In Argon | In Hydrogen | In Fluorides Fine-grained Inconel 3500 Over 2500 730 275 750 0.002 0.007 0.020 0.015 Type 316 stainless steel 6300 855 830 770 865 0.009 0.009 0.010 6.020 Type 304 stainless steel 3500 1410 1100 1400 ¢.008 0.005 0.002 Nickel 1500 80O* 1004 935 0.0002 0.003 0.001 *Still in test. The test was dis- Metallographic exami- revealed profuse internal intergranular cracking, Other tests of fine-grained Inconel ot 3500 psi in argon only were discontinued after 200 and 450 hours. Microscopic examination did not reveal any intergranular cracking. It has been shown that, in both argon and hydrogen, failures will result from initiotion and propagation of inter- granular cracking. Hydrogen apparently tends to premote and accelerate the rate of nucleation and the propagafion of the cracking. Testing of Inconel in the fused fluoride NaF- ZrF ~UF , (46-50-4 mole %) is continuing, and the remaining tests are to be performed in the low stress range. A preliminary design curve for fine- grained material is shown in Fig. 8.4, The dotted portfions of the curves that are below 3500 psi were obtained by extrapolation. Data for a design curve for coarse- and fineegrained inconel will be completed to a stress of 2500 psi in opproximately three months. Tests at 1500 psi are scheduled, and it is estimated that about 12 months will be required for rupture of the specimens. the end of the hydrogen cycle. continued after 550 hours. nation FABRICATION RESEARCH E. S Bomar’ R. W. Johnson J. H. Coobs H. lnouye Metallurgy Division A. Levy, Pratt and Whitney Aircraft Division Extrusion of High-Purity Inconel Tubing. It was indicated in the previous repori(z)thof the principal problem involved in extrusion of high-purity cast (2, S. Bomar, Jr., et al., ANP Quar. Prog. Rep. Mar. 10, 71953, ORNL-1515, p. 152. lnconel was that of lubrication. During this period, several lengths of Inconel tubing was extruded by using glass wool on the billet ard rock woo! in the container. T1he results indicate that malleabilizing agents are unnecessary in these vaccum melts, The data are tabulated in Table 8.4. Analyses of the several heats show that the total impurities are about 0.4%, of which 0.3% is silicon and manganese from the ferrochromium used. The extruded tubes are being drawn to 0.50 in. in out- side diameter and 0.035 in., in wall thickness for corrosion testing. Air Oxidation of Columbium. The oxidation rate of columbium is linear between 400 and 1200°C. The logarithm of the rate constant plotted vs. 1/T is a straight line with a break at 900°C. This temperature corresponds with o change in the modi- fication of the Cb,0, from the low, or “'T", form to the high, or ““H’", form. The energy of activation between 600 and 900°C was found to be 13,400 cal/mole, Above 900°C, a value of 4350 cal/mole was obtained. The only oxide found in these tests was Cb,0,. Embrittiement of the metal occurs ot approximately 800°C, Control Rod for the G-E Reactor. The prototype contral rod, 30 in. long and Y% in. in diameter, was assembled and delivered for testing. This rod comprised several segments of hot-pressed 56% B ,C~44% Fe absorber canned in o stainless steel tube, However, brazing the hot-pressed mixture to the tube did not result in a satisfactery bond, so a decision was made to abandon this method of rod fabrication. An alternate method of fabrication that is being investigated consists of filling o tube with a mixture of Al-B,C or Cu-B4C powders and reducing the tube to 1/2 in. in diameter by hot 77 ANP PROJECT QUARTERLY PROGRESS REPORT CREEP RATE (% /hr) m 01 1.0 10 N T \RUPTURE /" CREEP RATE < o T N P "\\* ***** - . o NID N rofl ) NATINONS N < \\\ N, \ ® I & . X 4000 ; \\ A1 ™ \\ \\\ N = & e \ \ \ \ 2 d N NN e \ ® @ Q% @ A N SN ™R 4@ 3000 r b LR RN \-.‘\ & A N NONTNY @ ] \ \ ™ \:\ v ~d ™ Y .\\ \\\ ‘\\ \ 2000 L Jod \\\ \\;2&._ g ~L \_\\ \\ NN o R NN \\\‘\ 1000 . 1 10 {00 100QC 10,000 TIME {hr) Fig. 8.4, Creep ond Stress-Rupture of Incone! Sheet Tested in o Fluoride Fuel at 1500°F, Heat NX8004; cold rolled and annealed at 1500°F. Grain size: approximately 90 grains per square millimeter. Times were measured for 0.5, 1, 2, 5, and 10% extension and rupture vs. stress. Extension measured by diol gage. TABLE 8,4, EXTRUSION DATA FOR VACUUM-MELTED INCONEL BILLETS* BILLET EXTRUSION EXTRUSION UNIT INGOT NGO, TEMPERATURE®** PRODUCT RATIO EXTRUSION DIE (°F) PRESSURE (tsi) 7,9,10 2150 Tube 22:1 58 to 72 25 deg cone 8 2200 Tube 23:1 60 30 deg cone 8,2,10 2200 Rod 14:1 45 45 deg \ *Billets 3 in. in cutside diameter and 4 in. leng; front radius ]{4 in, **Heated in ND-15 salt bath. 78 swaging or hot rolling to consclidate the mixture, as well as to form a bond between the matrix and the tube wall. When about 50 vol % of fine B ,C powder is used, the core matericl should have nearly the same nuclear cross section as the hot- pressed B ,C-Fe composition. Mixtures containing 47% Al-53% B,C and 75% Al-25% B ,C were prepared by mixing ~100-mesh atomized alummum powder, grade ‘‘C'" electrolytic copper powder, and metallurgicdl-grade B C powder for 1 hr in an oblique blender. The B,C was ground for 16 hr ond contained about 60% -~325 mesh particles. The tap densities of these mix- tures were 63% for the B C-Al mixture and 50% for the B C-Cu mixture. The B ,C-Al mixture was used w:'rh % -in.~dia tubing and wcxs reduced a total of 40%. However, since the B ,C-Cu mixture re- quired more reduction for complete consolidation, it was loaded in a ¥-in.-dia tube and reduced o total of 58%. The finished rods were }/2 ine in diameter. The tubes used for both mixtures were AlS! type 304 stainiess steel with an original wall thickness of 0.049 inch. They were bright annealed in dry hydroger, loaded by tapping for 3 to 5 min, sealed, and evacuated through a porous sintered-metal plug. The 84C~AE rods were reduced at 600°C in two steps, while the B C-Cu rods were reduced at 1000°C in four steps. A total of seven experlmentai rods has been prepared, including one 20-in.-long rod of each material, for reactivity tests in the LITR. In addition, short sections about 6 in. in length of each type of rod have bzen prepared for thermal conductivity ‘and strength tests. In each cose the core material was consolidated to 95% or more of theoretical density and showed good distribution PERIOD ENDING JUNE 10, 1953 of B,C. The B,C-Cu core material showed good bondlng to the stainless steel tube wall in the as swaged condition, but the B C-Al mixture did not. However, after heating to 700°C for 15 min, the B,C-Al exhibited excellent bonding. The stainless steel tube wall was increased in the swaging process from 0.049 in. to a final thick- ness that varied from 0.053 to 0.057 in.; thus the final wall thickness was somewhat less than the required 0.065 inch., However, if the original wall thickness were 0.058 in., an increase of the same proportion during the swaging would give « final thickness very close to that required. HOT-PRESSED PUMP SEALS E. S. Bomaor - R. W. Johnson J. H. Coobs H. Inouye Metaliurgy Division A, Levy Pratt and Whitney Aircraft Division Fabrication of Small Compacts with Mo5,. Addi- tional small experimental compacts with 14 vol % MoS., were prepared by hot pressing. The data from these experiments are given in Table 8.5. The first four experiments showed that the 95% Ag~5% Cu alloy composition could not be prepared by using coarse silver powder. In all cases in which the pressing temperature exceeded the eutectic temperature of 780°C, o portion of the ailoy was lost by extrusion of the eutectic composition past the rams. A single compact pressed at 770°C with no loss of material was heated at 820°C in helium for a total of 36 hr in an attempt to produce a homogeneous alloy matrix. This treatment was not successful. Thus, it was necessary to prepare TABLE 8,5. EXPERIMENTAL COMPACTS OF FACE SEAL COMPOSITIONS MATERIAL HOT~PRESSINC; DENSITY REMARKS TEMPERATURE (VC) (% of theoretical) (Coarse 95% Ag—5% Cu)-MoSz. 790 94.5 8% loss 280 97.3 T0% leoss a50 95.9 4% lass 770 94.6 No lass Type 304 s;tuiniess steef-MoSz 1120 84.0 Highly magnetic 1220 94.0 Highly magnetic Molybdenum ‘ 1530 92.0 79 ANP PROJECT QUARTERLY PROGRESS REPORT this composition by using the ultrafine precipitated silver powder. Metallographic Exaemination of Compacts with MoSz. The compacts of stainless steel and MoS, are being studied metallographically to determine the extent of reaction, if any, between the MoS, and stainless steel. The sulfide phase seems to be interspersed with a metallic phase, which becomes more prominent as the pressing temperature is increased. X-ray diffraction analysis revealed only ferrite; the identity of the sulfide phase could not be established and austenite could not be de- tected. |t thus seems that molybdenum is dissolved or nickel is removed from solution, or both, with the consequent formation of ferrite. Thus the com- position may be unsatisfactory for the pump seal tests. Metallogrophic examination showed that the Cu-MoS, compact and the Ag-MoS, and (95% Ag- 5% Cu)-MoS, compacts prepared with precipitated silver powder had excellent structure. The flake- like particles of MoS, were oriented perpendicular to the direction of pressing and were distributed at random in the continuous, homogeneous, highly dense matrix. The coarse silver compact, on the other hand, had a rather poor structure in which the Mo3, was distributed exclusively at the particle boundaries, This compact showed rather poor interparticle bonding and had poor physical properties. Fabrication of Face Seal Rings. On the basis of results obtained with the small compacts, face seal rings 37/8 in. OD by 23/8 in, iD by 7/8 in. long with the compositions shown in Table 8.6 were fabricated by hot pressing at 2500 psi. The Cu-MoS, composition seemed quite satisfactory except that the density was not so high as desired. A slightly higher pressing temperature might have increased the density, but during the second hot pressing operation the die broke. Since the compact ob- tained could be used, the operation was not repeated, The Ag-MoS, compacts were prepared with coarse silver powder, and it was necessary to approach quite close to the melting temperature of silver to obtain the low porosity desired. Difficulty in accurately controlling the temperature of the large die assembly resulted in a substantial loss of material from one of the compacts by extrusion past the roms ond infiltration of the graphite die body. The (95% Ag—5% Cu)-MoS, compact was prepared with the fine precipitated silver powder which gave excellent results in small compacts. However, the density of the first compact was rather low, and in an attempt to obtain a density of 95%, the maximum temperature was exceeded ond again a subsiantial amount of the material was lost by extrusion ond infiltration. All compacts were judged to be satisfactory for testing and were delivered. An additional request was received for a number of smaller rings of the Cu-MoS, composition to be used in a packing gland to seal the shaft of a fluoride pump. A total of 20 rings, ]]}/16 in. OD by ]3/16 in. ID by ¥ in. thick, is required for the first test, Six cylinders, 13/4 in. QD by ”/8 in. 1D by 2 in, long, from which sufficient rings could be machined, were prepared from the 92% Cu--8% MoS$, composition, |t was found that densities as high as 97.5% of theoretical could be attained by hot pressing the rings at 950 to 960°C and cooling under pressure. Surface Sulfurization of Molybdenum. Efforts to form MoS, on the surface of molybdenum have been continued, The pyrex reaction vessel used initially was replaced with a vessel of material that permits TABLE 8.6. FACE SEAL TEST RINGS P A - NSI COMPOSITION TEM EOR TURE DE TY REMARKS (wi %) ") (% of theoretical) Cu—-8% M052 910 89.5 200 ~00 Die broken Ag-9% M052 260 92.5 15% weight loss 955 95.0 3% loss (95% Ag~5% Cu)-9% MaS, 835 85.5 No loss 880 86.5 20% weight loss 80 the use of a higher temperature. Also, H,S was used in one experiment in place of the sulfur vapor., For the experiment in which H,5 was used, the molybdenum was introduced in the form of an 18-in.- long by 0.057-in.-dia wire. Thewire was suspended in a vertical tube furmoce that had a temperature gradient of from room temperature to 1020°C, The H,S was passed through the tube at the rate of 1 liter/min during an exposure of ]!/2 hr at tempera- ture., The exposure was terminated because of the escape of H,S§ when the gas outlet became plugged with condensed sulfur, Data on the free energy of formation of H_§(3) indicate that at 1000°C the equilibrium pressure of sulfur gas is about 0,08 atomic weight. An appreci- abl e portion of the decomposed H,S must have been swept out of the reaction chamber before recombi- nation could take place. The molybdenum wire reacted with the gaseous atmosphere at tempera- tures of from 535 to 1020°C. X-ray diffraction patterns showed the reaction layer to be MoS,. However, the difficulty described here plus the psychological and physiological hazards associated with the escape of H,S make the use of sulfur vapor more attractive, _ Two furnaces were arranged to permit the heating of a molybdenum sample contained in a tube to one temperature and the heating of a sulfur reservoir at the other end to a lower temperature. Type 446 3y, . T. Ellingham, J. of the Soc. of Chem. Ind. 63-65, 125-133 (May 1944). PERIOD ENDING JUNE 10, 1953 stainless steel fubing was used for the container. Data from the four exposures made by using this arrangement are summarized in Table 8.7. These tests indicate that a MoS, layer can be formed by exposure to sulfur vopor and that a degree of con- trol may be exercised by varying the molybdenum temperature and the sulfur vapor pressure. Further checks will be made to determine whether the temperature at which the MoS, is formed has an appreciable effect on the nature of the compound formed. HIGH-CONDUCTIVITY METALS FOR RADIATORS E. S. Bomar R. W. Johnson J. H. Coobs H. Inouye Metallurgy Division A. Levy Pratt ond Whitney Aircraft Division Finned tubes could be advantageously used for the air radiators that will be required by most nuclear aircraft reactors to transfer the reactor heat to a turbojet air stream. The design of the fins requires that they be 0.010 in. or less in thickness, that they be oxidation resistant at 1500°F, and that they possess high thermal con- ductivity. The object of the initial investigation has been to determine the most promising method of fulfilling these specifications. The mu’ior effort, to date, has been directed toward the protection of copper by cladding it with TABLE 8.7. REACTION OF MOLYBDENUM WITH SULFUR VAPOR SUL.FUR SULFUR VA.POR MOLYBDENUM TIME AT TEMPERATURE PRESSURE TEMPERATURE | TEMPERATURE REMARKS (°C) (mm Hg) (°C)* {min) 445 760 800 60 No Mc:oS2 formed 490 1441 900 120 Complete conversion of 4-g molybdenum sample; layer of Fel formed on inner surface of tube 490 1441 900 B Powdery MoS, % _ in. thick on molybde- num compact . 460 948 900 10 Porous molybdenum sample, ]/Ié'in.‘ layer of MoSz; dense molybdenum sample, slight reaction *Data from Lange's Handbook of Chemistry, Sixth Edition, p. 1421, 81 ANP PROJECT QUARTERLY PROGRESS REPORT the specified maoximum permissible thickness of 0.002 in. of oxidation-resistant materials. Some attention has been given fo the use of oxidation- resistant copper alloys as the fin material or as a cladding on copper. Type 310 stainless steel clad silver was also investigated. The results, to date, are of a preliminary nature. Chromium-Clad OFHC Copper. Both vapor-de- posited and electroplated chremium claddings on oxygen-free high-conductivity (OFHC) copper were tested at 1500°F for 100 hours. The results indi- cate that this combination offers protection to the copper if the plate does not peel off or crack during temperature cycling. Diffusion does not appear to be o problem. Samples of ‘‘ductile’’ chromium on copperhave been requested from the Ductile Chrome Process Company. Chromium and Nickel Electroplate on Copper. A composite that was intended to be 0.0005 in. of chromium plus 0.0015 in. of nickel on copper was prepared; however, measurements indicated that the composite tested actually had 0,002 in, of nickel and less than 0.0001 in, of chromium. When tested, as plated, at 1500°F, the surface showed a few areas of failure after 100 hours. A 500-hr test showed no failure. After a pretreatment of 24 hr at 1000°C in hydrogen, a test for 100 hr showed no surface failure, but a test for 500 hr showed com- plete failure because of separation of the plate from the core. Metallegrophic examination of the specimens tested for 100 hr showed that the copper core had suffered exireme diffusion and that numerous voids had appeared near the original interface. The amount of unaltered copper was approximately 25% of the original core, Incone! Cladding on Copper. In 100-hr tests of Inconel-clad copper, diffusion occurs that affects a zone that is from 0 to 4 mils thick. In 500-hr tests, the characteristic color of the copper nearly disappears, and the cladding is penetrated by a copper-rich phase thai extends to the surface in some locations., The 100-hr tests showed that a core of commercial copper develops substantially more voids near the interface than does OFHC copper. In all instances, surface failure (that is, oxidation effects) is limited to the development of copper oxide nodules, This type of failure is 82 attributed to pinholes developed in the cladding and rolling of the composite. Suitable rolling tech- niques reduce the number of pinholes developed to the extent that this type of failure becomes in- consequential. The wide variations in thickness of the diffusion zones in the several different composites tested suggest that the method of fabrication is responsi- ble for the diffusion. It is thought, at this time, that assembling the composites with both the copper and the cladding as clean as possible promotes diffusion and that consequently a ‘‘not so clean” interface is desirable. Type 310 Stainless Steel Cladding on Copper. Tests of type 310 stainless steel cladding on copper have been completed, These tests were conducted in air for 100- and 500-hr periods. The results indicate that the diffusion which occurs is pre- dominantly of a different type than that found in specimens of Inconel cladding on copper. In most cases the interface becomes highly irregular, to the extent that islonds of o copper-rich phase appear in the cladding., In only one instance did a test piece show alteration of the copper core. The number and the size of the copper-rich islands that appear in the cladding are greater when com- mercial copper is used than when OFHC copper is used. All surface failures were pinholes such as found with Inconel cladding. Type 310 Stainless Steel Cladding on Silver, Tests of type 310 stainless steel cladding on silver were made at 1500°F for 100- and 500-hr periods., The material tested showed a cladding thickness of 5 mils. There was no failure in the cladding, and diffusion at the end of 500 hr was negligible. Copper Alloys. Copper alloys were tested for 100-hr at 1500°F with the results given in Table 8.8. The thermal conductivity at 20°C of the copper-aluminum alloys is about 0.20 cgs units, with a positive temperature coefficient. Extrapo- lated data indicate a value of about 0.4 cgs units at 1500°F {for Inconel and type 310 stainless steel the value is about 0,06 cgs units). These copper- aluminum alloys may prove to be adequate as they are, but the present plans are to clad copper with both the 6% and the 8% aluminum-bronze alloy. PERIOD ENDING JUNE 10, 1953 TABLE 8.8, TESTS OF COPPER ALLOYS IN AIR FOR 100-kw PERIDDS ALLOY LOSS OF COPPER CEMARKS {%) 99% Cu—1% Be 7.0 Uniform block oxide that scaled freely 98% Cu~2% Al 28.5 Thick uniform black oxide 96.5% Cu~3.5% Al 1.0 Numerous areas of localized oxidation 95% Cu—5% Al | 0.16 Few areas of lacalized oxidation 94% Cu~6% Al | No change Polished sucface bacame dull 92% Cuwm8% Al No change Polished surface remained bright 90% Cu~10% Al Weight increased Visible coating of Al,O, found Copper 42.4 Thick black oxide SOLID FUEL FORMED IN SPHERES E. S. Bomar R. W. Johnson J. H. Coobs H. Incuye Metatlurgy Division A. Levy Pratt and Whitney Aircraft Division The initial work on the preparation of spherical particles of solid fuel was reported pre\fiousl),f.(“‘3 For the previous work, alioys containing 5% uranium in copper, nickel or molybdenum were subjected to four different methods of processing. Only one method, spraying from a metallizing gun, gove an appreciable yield of particles that approached spherical geometry. : Additional work has been bused on modifications of earlier methods or the use of additional methods, Some success has been achieved. Briefly, the work has included: 1. heating to above the alloy melting point in a vacuum, ' 2. substitution of a three-phase three-electrode arc for o single-phase double-electrode arc as a heat source for particles dropped through the arc, 3. forcing the molten alloy through a small orifice, 4. letting precut particles of the alloy settle through a molten salt bath having a temperature gradient extending beyond the liquidus and solidus of the alloy. Each of these methods of preparing spherical particles is discussed below, and additicnal de- tails, including photomacrographs of the particles (4YE_S. Bomar, J. H. Coobs, and H. Inouye, ANP Quar. Prog. Rep. Dec. 10, 1952, ORNL-1439, p. 155. prepared by these and the previous methods, are presented in other reports.(sfl'_“ There are no plans at present for pursuing further the problem of fabri- cafing spherical particles., However, two odditional methods which might prove of interest are: 1. passing of a high velocity stream of inert gas through an electric arc formed between elec- trodes of the desired alloy, () 2. forming spherical geometry from small right cylinders by random impacts in g hammer mill, The second method was suggested in ¢ private communication from Doyle Rauch, Research Divi- sion, NYO-AEC. This method and the more con- ventional ball-bearing manufocturing techniques are used o produce the spheres used in ballpoint pens. Heoting Under Yacuum. Particles clipped from a 0.010-in. wire of 5% U-95% Ni alloy were degassed under a vaccum of 1077 wmm Hg ot 900°C while resting on an alumina plate. They were withdrawn from the hot zone of the furnace ond examined through a glass port, still under vacuum. The particles appeared bright and clean. The furnace temperature wos then raised to 1450°C, and the charge was gradually moved into the hot zone. The vacuum did not exceed 5 x 107¢ mm Hg during any of the heating cycles. After opproximately 5 min at temperature, the charge was wifhdrfiwn, alfowed to' conl, and removed from the vacuum Slg, 5. Bomar and H. Incuye, Fubrication of Spherical quflcfes, ORMNL CF-52-11-7 {Mov, 1, 1952). }A Levy, Fabrication of Spherical Forticles, ORNL CF.53.3-183 (Murah 19, 19531, (Y)H. H. Hausner and H, Mansfield, Avomization Method of Making Urdgnivm Powder, NY0-1133 {Aug. 7, 1950). ANP PROJECT QUARTERLY PROGRESS REPORT furnace for examination. Melting had occurred, but the formation of a surface film dgain prevented spheroidization. Three-Phase Carbon Arc, Three carbon arcs were symmetrically arranged about a common axis in a housing purged with tank argon. Passage of cut particles of 5% U-95% Cu and 5% U-95% Ni alloys through the zone of the arc produced a very low yield of melted particles ot currents of up to 35 amperes. At 40 to 45 amp, the major portion of the particles melted. Many particles were flattened when they struck the bottam of the heusing because there was insufficient free fall. As in the eorlier two-electrode set up, alignment through the arc was very critical. Examination of the melted parti- cles revealed surface oxidation, depletion of uranium, porosity, and Any or all of these effects may have been due to inadequate control of the atmosphere, Molten Alloy Passed Through a Small Orifice, Spherical particles were produced, with moderate success, by forcing molten uranium-copper alloy through o small orifice in the base of a refractory container. A positive pressure of argon was re- quired to force the alloy through 0.020., 0.025., or 0.031-in. holes in alumina crucibles. For a run, a crucible 10 to 15 in, long and ]/2 in. in diameter was passed through a rubber stopper into a quartz tube containing argon. A small graphite sleeve placed over the lower 3 in. of the crucible served as a heat source when expased to the field of an induction coil. Upon heating to approximately 1200°C, particles ranging in size from 0.010 to 0.020 in. were ejected from the crucible, The orifice size appeared to have no effect on the resulting particle size. Particle surfaces showed little oxidation. Macroexamination revealed a cast dendritic structure of primary copper plus UCu ,-Cu eutectic. Some of the surface defects were found to be due to shrinkage upon solidification, Settling of Particles Through a Molten Salt. BaCl, was charged into a graphite crucible, which was, in turn, positioned in a quartz tube that was being continuously purged with argen. The upper half of the crucible was heated by induction to produce a temperature gradient in the molten salt that ronged from 962 (mp of BaCl,)) to 1200°C. Particles of 5% U-95% Cu alloy clipped from 0.010- in. wire were partly spheroidized under these con- ditions, When the maximum temperature of the salt was increased to 1300°C, most of the particles were spheroidized, Slight surface oxidation of the particles occurred. INFLAMMABILITY OF SODIUM ALLOYS G. P. Smith L. L. Hall Metallurgy Division Studies have been initiated to determine to what extent the inflammability of jets of liquid sodium can be reduced by alloying the sodium with some of the heavy metals. In the initial phose of this project, a large number of alloys will be surveyed qualitatively. The qualitative data will serve to indicate the worthwhileness of a detailed study of the such as porticle-size distribution, rate of evaporation, ond gas-phase reaction kinetics, which probably influence the numerous variables, inflammability of jets of liquid alloys. The following method of measurement is being used. A 15-cm® sample of alloy is sealed in a metal capsule under 1 atm of argon. The capsule is made so that one end may be broken off to form a small hole. After the capsule is heated to a temperature of 700 to 900°C, the end is broken off, and the argon pressure developed by heating at constant volume ejects the molten alloy through the small hole to form a jet. Visual observations are made of the extent of combustion. One unalloyed sodium jet was burned at 750°C and three jets were burned at 800°C. The dif- ference in temperature did not make any significant difference in the flammability. In all cases, the jet burned with a brilliant yellow flash. Copious quantities of white smoke were produced. Motion pictures were taken of a speed of about 700 frames/sec of one of the burning jets as it left the capsule. The only source of illumination for these pictures was the light produced by the burning sodium, It was observed that the jet consisted of many, small, brightly glowing droplets accompanied by a less luminous cloud. This cloud may have consisted of very fine particles or may have been sodium vapor lighted by resonant radiation. When the cluster of droplets forming the jet had moved a few centimeters away from the capsule, there was a sudden increase in light intensity. Three jets of unalloyed lead at 800°C did not take fire, and most of the lead was recovered in an unoxidized condition. At 950°C, a jet of unalloyed lead oxidized considerably but did not take fire. Alloys of sodium containing 10 and 50 wt % lead appeared to burn more intensely at 800°C than did pure sedium. Additions of 5 and 10 wt% mercury did not seem to change the burning of sodium jets to an appreciable extent at 750 and 800°C. A single test was made at 800°C with sodium con- taining 10 wt % zine. This alloy burmed at about the same intensity as did pure sodium. The apparatus being used for these measurements is of a very simple design. A small furnace for heating the capsules and the mechanism for break- ing off the ends of the capsules are mounted on a transite board which rests on top of a 55-gal steel drum. The drum, which serves to confine the burning alloy, has a long narrow safety glass window down one side for observation. This ap- paratus is enclosed in a transite hood of temporary construction, After sodium or one of its alloys has been bumed, the dense cloud of white smoke (probably a mixture of sodium oxide, hydroxide, and carbonate) which is produced is removed through the exhaust system of the building. A new apparatus is being constructed in which the combustion of sodium dlloy jets can be studied in air at low pressures with controlled water-vopor concentration. The pumpose of this apparatus is to simulate more nearly atmospheric conditions at high elevation. CERAMICS L. M. Doney J. R. Johnson J. A, Griffin A. ) Taylor Metallurgy Division Development of Cermets. Further developments of the silicon carbide—silicon fuel elements in- clude the following: 1., The inclusion of 10% USi, in the center portion of a cross-shaped element has been suc- cessfully achieved. A layer of material opproxi- mately l/16 in. thick that does not contain fuel completely surrounds the fuel-bearing portion. 2. lsostatic techniques for pressing the porous carbon buse material have been initiated. PERIOD ENDING JUNE 10, 1953 3. A new means of fabrication has been developed that utilizes a dipping technique and induction heating for melting the silicon, 4, Preliminary autoradiographic tests have been developed to show the location of the fuel after impregnation with silicon has faken place. Effect of Heating Raote on the Beryllium Oxide Blocks for the ARE. Experiments were carried out to determine the effect of hieating rate on the full-size hot-pressed beryllium oxide blocks in the as-received condition. All heating was carried out in a large, Harper, glo-bar furnace; the samples were shielded from the direct radiation of the heating elements by silicon carbide plates 3/4 in, thick. Several of the blocks were checked for incipient cracks before the experiment was started, but no cracks were found in any of the blocks checked. In the first experiment, two blocks were placed in the furnace and heated at 100°C/hr to 1500°C; the furnace was then shut off with no holding time at the peak temperature. The cooling rate was slightly over 100°C/hr down to 1300°C, but it was less thon 100°C/hr from 1300°C down to room temperature. Both blocks were intact at the end of the experiment and had no visible cracks. Two unfired blocks and one from the first test were then placed in the furnace, heated to 1500°C at 200°C/hr, and cooled as before. Both new pieces were cracked through ot the center, per- pendicular to axis; the lower half of one was also split parallel to axis. However, the piece which had been heated previously was unchanged, In a third test, the second heot treatment was repeated, but this time, no pieces were broken. It cannot be claimed that enough samples were used to allow definite conclusions to be made; the results should be considered only as indicative of a irend., However, there is some evidence that these pieces do have internal strain and that their thermal endurance might be improved by an initial heat treatment at a very slow heating rate. On the basis of these few tests it would appear that a safe heating rate for the as-received blocks is below 200°C /hr. ANP PROJECT QUARTERLY PROGRESS REPORT 9. HEAT TRANSFER AND PHYSICAL PROPERTIES H. F. Poppendiek Reactor Experimental Engineering Division Some preliminary density and viscosity measure- ments were made on the new ARE fue! NaF-ZrF . UF, (53-43-4 mole %). The viscosity was found to vary from about 13,5 cp at 620°C to about 8.8 cp at 757°C. The density was found to be about 3.35 g/em® at 653°C. The enthalpies and heat capaci- ties of lithium hydroxide and sodium hydroxide eutectic and of fuel mixture NaF-KF-LiF-UF, (10.9-43,5-44.5-1.1 mole %) have been obtained. The heat capacity of LiOH-NaOH eutectic was determined to be 0,60 + 0.02 cal/g-°C over the temperature range 260 to 850°C; the heat capacity of the NaF-KF-LiF-UF, was found to be 0.44 +0.03 cal/g-°C over the temperature range 500 to 1000°C. The surface tension of NaF-ZcF -UF, (50-46-4 mole %) was determined to vary from 160 dynes/cm at 530°C to 115 dynes/cm at 740°C. Experimental velocity distributions have been determined in a pyrex convection harp system; the profiles differed significantly from the parabolic profile characteristic of isothermal laminar flow, At Reynolds' moduli as low as 85, large scale viscous eddies were observed that were considered to be due to thermal turbulence. Forced convection heat transfer data for NoF-KF-LiF flowing in a small Inconel tube were found to lie 50% below the data previously obtained in q nickel system. Some evidence of a thin, green corrosion layer was found at the surface of the Inconel. Such a layer or a nonwetting phenomena could account for additional thermal resistance in the thermal circuit, reactor coolant is a function of coolant duct dimensions and spacing, the amount of heat to be removed, the total coolant temperature rise, and the physical properties of the coolant. A mathematical onalysis of the effective- ness of reactor coolants has been made with the aid of IBM machines; this analysis can be used to optimize coolant systems in solid-fuel systems. An apparatus which transfer The effectiveness of a reactor simulgtes heat in a circulating-fuel piping system by passing an electric current through a flowing electrolyte has been successfully operated. An experiment has been analyzed in which heat was transferred in an annular system from a molten fluoride to NaK, The data are compared to Hausen’s and to Colburn’s equations, 86 THERMAL CONDUCTIVITY W. D. Powers S. J. Claiborne R. M. Burmnett Reactor Experimental Engineering Division The longitudinal thermal conductivity apparatus for measurements on solid and liquid metals has been completed, and preliminary runs have been made. The solid samples used are in the form of rods, the liquid samples ore placed in cylindrical, thin-walled capsules. The heat used is passed through a heat meter before it flows through the sample being studied. The conductivity of the sample is determined by measuring the temperature gradient within the sample and the heat flow through the heat meter. To eliminate radial heat losses, guard heating is provided. All heaters are individually controlled by variacs. The uninsulated and uninstrumented conductivity device is shown in Fig. 9.1. The two top calrods guard-heat the sample heater, and the other eight calrods are aftached to the eight annular guard disks. A copper disk, which is cooled with water (the heat sink), is located at the bottom of the apparatus. When in operation, the entire device is surrounded by a metal shell which is filled with fine Sil-o-cel powder. Platinum~platinum-rhodium thermocouples are welded to the sample at the l[evel of each guard heater disk; similar thermo- couples are attached to the guard heaters. After a sample heater has been turned on and o tempera- ture gradient has been established in the sample, the guard heaters are turned on and adjusted so that no temperature difference exists between the sample and the guard heater disk at each level. At present, the radial temperature differences can be held to within less than 1°C when a total axial temperature difference of 300°C exists. Preliminary thermal conductivity measurements have been made on type 316 stainless steel. A value of 10 Btu/hr-ft2.(°F/ft) at 750°F, which compares favorably with the literature value, was obtained, Before the thermal conductivity of liquid lithium is determined, the apparatus will be checked with molten sodium. PERIOD ENDING JUNE 10, 1953 - .. UNCLASSIFIED : ' PHOTO 202" SPECIMEN HEATERS SPECIMEN GUARD HEATERS HEAT SINK Fig. 9.1. Longitudinal Thermal Conductivity Device. 87 ANP PROJECT QUARTERLY PROGRESS REPORT DENSITY AND VISCOSITY S. |. Cohen T. N. Jones Reactor Experimental Engineering Division A new drybox designed to contain instruments for measuring densities, viscosities, and surface tensions by each of two methods has been com- pleted. Included in the system is an elaborate gas train for removing the undesirable impurities from the cylinder gas used to provide the inert atmos- phere. This drybox will be described in detail in a future report, A study of the density and the viscosity of the fluoride fuel Nc:F-ZrF“-UF4 (53-43-4 mole %) is being made and preliminary data have been ob- tained. The viscosity was found to vary from about 13.5 ¢p ot 620°C to about 8.8 cp at 757°C. The density was found to be about 3.35 g/cm? at 653°C. These preliminary viscosity and density measure- ments are in good agreement with the corresponding values for the fluoride fuel NaF-ZrF, -UF, (50- 46-4 mole %), whose composition is very similar, Several glass mockups of possible capillary viscometer designs have been fabricated and tested. However, no final design for the device has been decided upon to date. The gas-bubble densitometer has been completed and tested. M will be used in conjunction with the plummet method to determine densities. HEAT CAPACITIES OF LIQUIDS W. D. Powers G. C. Blalock Reactor Experimental Engineering Division Investigations of the enthalpy and heat capacity of hydroxides and fluoride salt mixtures are being continued. The following measurements have been made:{1+2) for NaOH-LiOH eutectic (73-27 mole %) at 260 to 850°C, HT(quuid) Hooc(solid) 44 + 0.60T , c = » 0.60 = 0.02 , and for NaF-KF-LiF-UF , (10.9-43,5-44.5-1.1 mole %) at 500 to 1000°C, Ho(liquid) ~ Hgo (solid) = 21 + 0.44T , C = » 0.44 + 0.03 , where H is the enthalpy in cal/g, T is the tempera- ture in °C, and <, is the heat capacity in cal/g-°C. i “)W. D. Powers and G, C. Blalock, Heat Capacity of the Eutectic of Lithium Hydroxide and Seodium Hy- droxide, ORNL CF-53-5-103 (May 18, 1953}. (2w, D. Powers and G. C. Blalock, Heat Capacity of Composition No. 14, ORNL CF-53-5-113 (May 18, 1953). 88 SURFACE TENSIONS OF FLUORIDES S. I. Cohen T. N. Jones Reactor Experimental Engineering Division The surface tension of the fluoride fuel NaF- ZrF ,-UF, (50-46-4 mole %) was measured with a commercial tensiometer, which was modified for the experiment. A spring with a low spring constant was fabricated and used to give greater sensitivity to the instrument. The liquid- surface temperatures were obtained from tempera- somewhat ture profile measurements made with a traversing thermocouple probe. The surface tension was found to vary from 160 dynes/cm at 530°C to 115 DWG. |89f9A dynes/cm at 740°C (Fig. 9.2). 160 140 E 120 ° c - ol S 100 10 o | 500 600 700 800 TEMPERATURE (°C) Fig. 9.2. Surface Tension of i*lc|F-ZrF4-UF4 (50-46-4 mol %). VAPOR PRESSURES OF FLUORIDES R. E. Moore R. E. Traber Materials Chemistry Division Additional vapor pressure determinations of fluoride mixtures and components have been made by the method described by Rodebush and Dixon. (3) Data for the vapor pressure of NaF-ZrF -UF, (53-43-4 mole %), the composition presently con- sidered as the fuel for the ARE, range from 4.5 mm Hg at 790°C to 39 mm Hg at 958°C. The data are best represented by the equation -7160 =+ 7.37 , T(°K) from which the heat of vaporization of 33 kcal/mole is obtained. |°910 P(mm Hg) (S)W. H. Rodebush and A, L. Dixon, Phys. Rev. 26, 851 (1925). Work has begun on the mixture NaF-ZrF -UF (65-15-20 mole %), which is the new composition proposed for the addition of enriched uranium to bring the ARE to criticality., Preliminary results indicate that the vapor pressure of this composition at 900°C is approximately 8 mm Hg. Vapor pressure data for the mixture NaF-KF- ZeF -UF, (5-51-42-2 mole %) were given in @ previous repor’r.(‘” This composition was the first of the ZrF ,-bearing mixtures to be prepared, and it was the only mixture prepared before the hydro- fluorination treatment became the practice. Be- cause it had not been subjected to such treatment and because the zirconium tetrafluoride in use at the time was not so pure as that used in mixtures prepared more recently, vapor pressure measure- ments were repeated on a new sample. The recent data, which range from 8 mm Hg at 900°C to 64 mm Hg at 1123°C, fall considerably below the values obtained previously, A small amount of a more volatile impurity (possibly zir- conium tetrachloride) in the zirconium tetrafluoride used to prepare the first samples may account for the difference in pressures. The equation ~6789 o 4 6,743, T(°K) which was obtained from data on the new mixture over the range 900 to 1123°C, is believed to be more nearly correct than the equation obtained previously., The heat of vaporization derived from the equation is 32 kcal/mole. It is expected that mixtures containing beryllium fluoride will be subjected to investigation in the future. During the quarter, data were obtained for beryllium fluoride at higher temperatures than those reported previously.(4) The two sets of data are in satisfactory agreement, although the slope of the plot of log P vs. the reciprocal of the absolute temperature is slightly greater when values ob- tained af the higher temperatures are included. The equation 4 log10 'P(mm Hg) ~9236 ~ T(°K) was derived from data given before, as well as from recent dafa, over the temperature range 891 to 1073°C. The heat of vaporization (42.5 kcal/mole) long(mm Hg) + 9.136 (4R, E. Maore, ANP Quar. Prog. Rep. June 10, 1952, ORNL.-1294, p. 150-151. PERIGD ENDING JUNE 10, 1953 and the boiling point (1207°C) were derived from this equation. VELOCITY DISTRIBUTIONS IN THERMAL CONVECTION LOOPS D. C. Hamilton F. E. Lynch L. D. Palmer Reactor Experimental Engineering Division The objective of the investigation of velocity distributions in thermal convection loops is to determine the accuracy with which mean cirevlation velocities in thermal convection harps can be pre- dicted by the numerical solution of the heat transfer equation obtained by using experimental wall temperature data, The velocity distribution has been measured in the cold leg of a 17-mm-1D pyrex hamp similar to the one described previously,(5? Velocities were determined from observations of suspended particles illuminated by a collimated light beam which passed through the center of the tube. The internal wall temperature was measured at various positions in both the hot and the cold legs. These temperature data are being used to obtain a numerical solution to the heat conduction equation, When the numerical solutions are comr pleted, the resulting predicted velocities will be compared with the observed velocities to evaluate the method. Observed velocity distributions are presented in Figs. 9.3 and 9.4 for two values of the Reynolds’ modulus. At even greater vaolues of Reynolds’ modulus, negative velocities were observed in the center of the tube. The large deviation of the velocity distribution from the fully developed isothermal laminar distribution is apparent. Large- scale viscous eddies were observed to be super- imposed on the mean flow. The swirling or eddying increased in intensity and decreased in size as the temperafure difference (or Reynolds’ modulus) was increased. This eddying appeared to be sufficiently intense to contribute to the transport of momentum and heat. Since no data are available on ‘‘thermal turbulence’’ the accurate prediction of the hegt and momentum transfer in such o system is ex- tremely difficult, Blp, c. Mamilton, F. E, Lynch, and L. D. Palmer, ANP Quar. Prog. Rep. Dec. 10, 1952, ORNL-1439, p. 182, 89 ANP PROJECT QUARTERLY PROGRESS REPORYT UNCLASSIFIED DWG.1929114 2z ] 1 1 T 20 —h_-fi':;-.. DEVELOPED ISOTHERMAL LAMINAR DISTRIBUTION, g— 22 (1-R2) m 18— _\“\ o 16 e e LY ‘ at SY)y Fig. 9.3. Experimental Velocity Distribution in the Cold Leg of a Glass Convectian losp with « Reynolds Modulus (2r U /1) of 85. UNCLASSIFIED DWG. 19912 ! 2z l l DEVELOPED ISOTHERMAL LAMINAR 2.0 r———~—— U 2 — \ DISTRIBUTION, /[~ =2(1-R%) | 1 S L \\ 1.6 T e ‘-K-O' """""" (4t vd A <\ 12 1 P )= 06 foo — 04 | — 02 f—m - o L | o 0.2 o 0.6 0.8 10 R=F o Fig. 9.4, Experimental Velocity Distribution in the Cold Leg of a Glass Convection Loop with a Reynolds Modulus (2r U/ _/v) of 105, 90 FORCED-CONVECTION HEAT TRANSFER WITH NaoF-KF.LiF H. W. Hoffman J. Lones Reactor Experimental Engineering Division Additional heat tronsfer data have been obtained for the NaF-KF-LiF eutectic(11.5-42.0-46.5 mole %) flowing through a heated Inconel tube, The ex- perimental results in terms of Colburn’s ; function vs. Reynolds’ modulys are presented in Fig. 9.5, The previously reported data, which were obtained in a nickel tube, are also presented for comparison. The data obtained in the Inconel tube fall approxi- mately 50% below the curve obtained from the equation which correlates heat transfer in ordinary fluids, in sodium hydroxide,(®? and in NaF-KF-LiF in nickel tubes. One would expect NaF-KF-LiF to behave as the ordinary fluids do, as for as heat transfer is concemed, ond yet there is a difference between the data obtained in Inconel tubes and that obtained in nickel tubes. In order fo resclve this disparity, temperoture measurements and physical property data were reviewed. Checks on the thermo- couple calibrations showed no significant devia- tions from the original calibrations. Checks on the mixed mean fluid temperature measurement tech- nique also indicated that no errors appeared to be present, The heat capacity of the NaF-KF-LiF mixture was never in question, because good heat balances were obtained during the heat transfer experiments, The viscosity of the mixture had been checked previously and would have had to be (6)H. W. Hoffmen, Turbulent Forced Convection Heaf Transfer in Circular Tubes Containing Molten Sodium Hydroxide, ORNL-1370 (Oct. 3, 1952). UNCLASSIFIED DWG. 19913 0850 FTTT T T (7T ""“'J"T"L"“"'"'"“”T“*‘"”"“’ 0_ 9 e o 4 n ramna o - e e ——— .........‘.A....<|'..... 0008 | ] H 1L L —- Q007 - - + NoF-KF-LiF IN NICKEL TUBE | 0006 f———- ~4- a NaF-KF-LiF IN {NCONEL TUBE — 0005 b— 2 7 MODULUS (St-Pr Q Q o Q S o I o | / L o T | 0001 1000 5000 10,000 REYNOLDS MODULUS 20,000 Fig. 9.5. The j-Modulus vs. Reynolds Modulus for NaF-KF-LiF Systems, an unreasonably low value to account for the dis- crepancy. A several-fold change in thermal con- ductivity would account for the difference in heat transfer, but the checks on conductivity, which are currently being made, indicate that the original values are correct. : The explanation of the difference in heat transfer between the nickel and the Inconel system is ap- parenily that on interface resistance, such as a layer of corrosion products, exists between the tube wall and the NaF-KF-LiF in the Inconel system. For example, it would only take a l-mil filmhaving athermal conductivity of 0.5 Btu/hr-ft-°F to account for the reduced j modulus observed; if the conductivity of such a film were lower, it could be much thinner than 1 mil. Examination of the in- side surface of one of the Inconel tubes used in the experiments showed the presence of a thin, green film, which has recently been identified as K3CrF6, with some Li,CrF,. These corrosion products are typical of those found when KF- and LiF-bearing fluoride mixtures are contained in Inconel. It is significant, however, that no such film appears when the NaF-ZrF4-UF4 system (o which the ARE fuel belongs) is contained in Inconel, The thickness and, possibly, the thermal conductivity of the green film will be examined. HIGH-TEMPERATURE REACTOR COOLANT STUDIES M. F. Poppendiek J. I Lang Reactor Experimental Engineering Division The effectiveness of a reactor coolant is a func- tion of coolant duct dimensions and spacing, the amount of heat to be removed, the total coolant tem- perature rise, and the physical properties of the coolant. An analysis of the effectiveness ofreactor coolants from a heat transfer standpoint has been completed. Four simultaneous heat transfer equo- tions were reduced fo a single equation which con- tains several dimensionless moduli. This equation was evaluated, with the aid of {BM machines, over a wide range of the various parameters involved, and the results have been plotted in chart form. The pressure-drop and pumping-power equations can readily be evaluated next, In particular, the pump- ing power per heat extracted (a dimensionless pumping-power modulus) can be obtained. Recently the pumping-power moduli of several reactor coolants were calculated for heat removel from & hypothetical solid-fuel-element aircraft PERIOD ENDING JUNE 10, 1953 reactor with the following characteristics: total heat flux, 200 megawatts; length of the cylindrical core, 2.75 ft; cylindrical diameter, 3 ft; coolant volume, 30%; total coolant temperature rise, 200°F, The coolants lithium, bismuth, sedium, and NaF- KF-LiF were considered, and the pumping-power modulus for each coolant was evaluated over g range of coolant tube diameters, The results indicated that therewas a minimum in each pumping- power modulus vs, tube-diameter curve and that of the four coclants lithium was the best and bismuth was the worst, (NaF-KF-LiF and sodium were, in general, about the same for the particular case considered.) It is felt that the charts described above can be used quickly to make effectiveness optimizations of coolant systems for solid-fuel-element reactors from a heat transfer — momentum transfer standpoint, TURBULENT CONVECTION IN ANNULI J. O. Bradfute J. 1. Lang Reactor Experimental Engineering Division The design of the previously described annulus flow system for measuring velocity profiles has been completed, and fabrication of the machined part is under way. The structural steel bracket which will rigidly support the annulus has been installed. A direct-drive blower has been obtained and modified to a belt-driven device so that the rotor speed can be increased until an adequate head is developed. ' A small flow system has been constructed so that the photographic techniques can be con- veniently developed and refined while the equip- ment is being fubricated. A few preliminary photo- graphs have been made which revealed the need for increasing the light intensity in the beam. Several photographs of dust particles that were so small as to be invisible in diffuse light have been taken by using the light scattered from a Tyndall beam. The resolution or grain size difficulties anticipated have not materialized; the particles appear in the photographs as well-defined streaks. The grid, which is simply d 1/]6-in.-thick sheet of lucite with scratches ruled on one side to form a cartesian coordinate system, has been photographed in light emitted from the scratches by illuminating one edge. This procedure exposes the negative only on the image of the grid lines and leaves the remainder of the film unexposed and, hence, trans- parent after development. 91 ANP PROJECT QUARTERLY PROGRESS REPORT CIRCULATING-FUEL HEAT TRANSFER H. F. Poppendiek G. Winn N. D. Greene Reactor Experimental Engineering Division The apparatus which simulates heat tronsfer in a circulating-fuel pipe system by passing an electric current through o flowing electrolyte{”} has been successfully operated. Preliminary wall and mixed mean tluid temperature measurements were obtained over the Reynolds’ modulus range of about 5,000 to 10,000 and at a power level of about 0.1 kw/em3, The limited wall-to-fluid temperature differences measured were in the range predicted from the previously developed theory. Some gassing was observed at the stainless steel electrodes at high current flows. Experiments have indicated that if the electrode surfoces which are in contact with the electrolyte are made of platinum, the gassing can practically be eliminated. A brief study of strong electrolytes(®! for use in volume heat source experiments was made to find an electrolyte having a low electrical resistivity and, aiso, flat resistivity-temperature character- istics. The study indicated that phosphoric acid satisfies these requirements, [t was also found that this electrolyte would be satisfoctory from the corrosion standpoint. When the platinum- surfaced electrodes ond a centrifugal pump with a somewhat higher head have been installed, the volume heat source experiments are to be resumed, BIFLUID HEAT TRANSFER EXPERIMENTS D. F. Salmon, ANP Division Heat transfer data from the previously de- scribed!?) bifluid heat transfer system containing NoK and the fluoride mixture NaF-ZrF -UF, (50-46-4 mole %) have been analyzed. Fifteen runs were made which yielded about 80 data points, The only data points that were considered valid were those obtained before a gradual reduction of the inside diameter of the tube occurred; the reduction was evidenced by increased pressure-drop meas- urements, Subsequent examination showed that the reduction of the tube diameter was the result (7)H. F. Pappendiek and G, Winn, ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 162, (B)N. D. Greene, A Preliminary Study of the Electrical Conductivity of Strong Electrolytes for Possible Appli- cation in Volume Heat Source Experiments, ORNL CF-53-5-149 (May 19, 1953). (9)9. D, Whitman and D. F. Salmon, ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 30, 92 of build-up of mass-transferred material in the heat exchange element of this trimetallic system: nickel heat exchange tube, type 316 stainless steel pump, and Incone! piping. The heat transfer occurred in a concentric tube section; the center tube contained the fluoride and the NaK was in the annulus. The center tube was made of nickel, 0.269 in, ID with o 0.030-in. wall. The L/D ratio of the tube was 40, while the 1./D ratio for the annulus was 22, The fluoride Reynolds’ numbers ranged from 5,000 to 20,000, and the NaK Reynolds' numbers ranged from 20,000 to 100,000, Inlet and outlet temperatures of the two liquid streams were measured with chromel-alumel probes welded into the piping at the center line of the streams. Inlet fluoride temperatures varied from 1200 to 1500°F. Fluoride radial temperature dif- ferences varied from 50 to 230°F, The over-all heat transfer coefficient, based on the outside tube wall surface, was calculated from the equation q AO U, = 0 Dt; where Dt is the log-mean temperature difference. The fluoride heat transfer coefficient was obtained from the over-all heat transfer coefficient in con- junction with a Wilson Plot.{1?) The fluoride data are compared to the Colburn’s(11) equation, Nu = 0,023 Re0:8 p,!/3 in Fig. 9.6, the viscosity of the fluoride was 10y, H, McAdams, Heot Transmission, p. 272, 2d ed., McGraw-Hill, New York, 1942. Uibid., p. 168, Eq. 4D, UNCLASSIFIED DWG. 19915 bomerd 100 10,000 REYNOLDS MODULUS Fig. 9.6. Comparison of Fluoride Heat Transfer Data with Colburn’s Equation, evaluated of the so-called ‘'film temperature,”’ 1 T --2-..(1.?- zw) . In Fig. 9.7, the fluoride data are compared with the Hausen's equation,{1?) which account the L/D effect: takes into PERIOD ENDING JUNE 10, 1953 The bifluid system is now operating with a con- centric tube exchanger constructed of Inconel to operate ot fluoride velocities of 8 fps and oxial temperature differences of 350°F. The current tests are to supply information on dynamic corrosion and mass transfer. | ' 2/3 0,14 Nu = 0.116 (Re2/3 - 125) prl/3 [1 +<—D-> } (i) . i In this equation, p is evaluated at the bulk tempera- ture and p_ is evaluated at the wall temperature, The fluoride center line temperature rather than the bulk temperature was used in Fig. 9.6. Theo- retical investigations have shown thot for the Reynolds’ and Prandtl’s numbers involved the ratio t - f Bl v x oo . t, -1 c u The heat balances were in error by a maximum of 28%, with an overage deviation of 11%. The fluoride data fell within approximately 11% of Colburn’s equation and about 27% below Housen's equation, The bifluid loop was sectioned at the conclusion of the test ond examination of the nickel heat exchange tube showed that a metallic fayer had built up on the inside to a thickness of approxi- mately 30 mils at the exit end. This large deposit probably was built up after the represented heat transfer data were obtained and during the 300 hr of operation with a fluoride velocity of 8 fps. Spectrographic and x-ray diffraction analyses both identified this 30-mil layer as iron. The iron was undoubtedly transferred from the type 316 stain- less steel pump to the nickel heat exchange tube by the fluoride. This mass transfer, which re- sulted from the presence of more than one metal in the same fluoride system, would mask such corrosion as existed., [t is significant that this mass transfer phenomenon has not been observed in any monometallic system; the ARE fuel sysfem is monometallic. Dynamic tests of the fluoride corrosion of Inconel in all-lnconel systems are reported in sec. 7, ‘‘Corrosion Research."’ ' Mo Uz)E. Eckert, Introduction to the Tronsfer of Heat and Mass, p. 115, McGraw-Hill, New York, 1950, UNCLASSIFIED DWG. 19314 100 o Q n < 1 - W { Py 73 (O 0 LD 1000 4000 10,000 REYNOLDS MODULUS 30,000 Fig. 9.7. Comparison of Flueride Heat Transfer Data with Hausen’s Equation, 93 ANP PROJECT QUARTERLY PROGRESS REPORT 10, RADIATION DAMAGE D. S. Billington, Solid State Division A. J. Miller, ANP Division Additional radiction damage studies of reactor fuels and structural materials were carried out in the LITR and the MTR. Efforts are being made to determine and explain the apparent uneven uranium distribution in the fuel irradiated in Inconel cap- sules. Since the state of the fuel in the capsules with regard to turbulence and temperature gradients is quite different from that found under service conditions, the design and construction of an in- pile loop for circulating fluoride fuel are being carried out as rapidly as possible, The cantilever type of creep measurements made on Inconel in a helium atmosphere indicated no serious change in creep properties upon irradiation, and a speci- men of Inconel irradiated in the MTR showed no change in thermal conductivity to within the ac- curacy of the measurements, IRRADIATION OF FUSED MATERIALS G. W. Keilholtz P. R. Kiein J. G, Moargan M. T. Robinson H. E. Robertson A. Richt C. C. Webster W. R. Willis M, J. Feldman Solid State Division K. J. Kelly, Pratt and Whitney Aircraft Division Exaominations have been made of the Inconel capsules containing small amounts of the ARE type of fuel that were irradiated in the LITR and in the MTR for various periods of up to several hundreds of hours. In the LITR, 230 watts were dissipated in each cubic centimeter of fuel, while in the MTR, approximately 3400 watts were dissi- pated by increasing the uranium fluoride content of the fuel, In general, the samples irradiated for long periods of time showed an intergranular type of corrosion which does not occur in out-of-pile control samples; the penetration occurred occa- sionally to a depth of 3 mils. Numerous chemical analyses indicated, as in the case reported previously for two capsules irradiated in the MTR, (") that the uranium content of the fuel which could be readily melted from irradiated capsules was lower than would be expected after allowance for burnup. On the other hand, several samples of fuel bored from various (1)G. W. Keilholtz et al., ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL.-1515, p. 166. 94 sections of the capsules were analyzed ond found to be deficient in uranium, while in other such samples, an excess of uranium was indicated. Capsules irradiated in the LITR and in the MTR were sectioned and examined visually at low magnification, No evidence could be detected of changes in the fuel or of segregation in this ex- amination; thersfore odequate facilities for petro- graphic examination are now being constructed in one of the hot cells, An investigation of the validity of the methods usad for the chemical analyses of the irradiated materials is continuing. it is possible that thermal gradients in the fue! contained in the capsules are causing concen- tration gradients in the fuel, ond an inpile pump loop would be a much better method of simulating service conditions than the static copsule. A few, simple, preliminary experiments were carried out to see whether it would be possible to readily obtain information concerning phencmena caused by temperature gradients or overheating of the capsules. Qut-of-pile control samples were heated to 1800 and 2280°F for 140 hours, Very little intergranular corrosion occurred; subsurface voids to a depth of about 1 mil were observed; and the uranium concentration of the central portion of Other control capsules were heated in a manner that provided a tempera- ture of 2240°F at the top and 1475°F at the bottom, approximately 3 in. away. No significant change occurred in the uranium concentration between the hot end and the cold end. the fuel was unaffected. INPILE CIRCULATING LOOPS 0. Sisman R. M. Carroll W, W. Parkinson A. S. Olson W. E. Brundage C. Ellis C. D. Baumann F. M. Blacksher Solid State Division Developmental work continued on a pump for an inpile fused-fluoride-fuel loop to be operated in the LITR and in the MTR., Design and fabrication of other portions of the loop are in progress. Beryllium oxide in contact with sodium at 1500°F was exposed in the LITR to determine the effect of radiation on its stability. Specimens from both the high- and low-density regions of ARE moderator blocks were sealed in capsules containing about 2 em? of sodium to give a surfoce-to-volume ratio of 3to 1. Eight such copsules were enclosed in a steel can that provided a helium atmosphere and had heaters attached to maintain the temperature at about 1500°F. The irradiation was carried out for three weeks, thaot is, until an exposure of greater than 10'8 fast neutrons/em? was obtained. Inspection of the beryllium oxide and sodium is under way. The preparation of samples for out-of- pile control runs is almost completed. CREEP UNDER IRRADIATION J. C. Wilson J. C. Zukas W. W. Davis Solid State Division The inpile cantilever creep apparatus used previ- ously has been modified to permit fests in an inert atmosphere. Figure 10,1 shows that previous re- sults obtained in the Metallurgy Laborafory(z) are confirmed by the bench tests; that is, creep of ), Potriarca and G. M. Slaughter, ANP Quar. Prog. Rep. Sept. 10, 1952, ORNL.-1375, p. 135. PERIOD ENDING JUNE 10, 1953 Inconel appears to proceed more ropidly in an inert atmosphere than in air. The creep curve in the only inpile inert-atmosphere test to date lies above that resulting from a test in air but below the corresponding inert-otmosphere bench test. Ap- parently, neutron irradiction slightly reduces creep in an inert atmosphere. The results are not yet conclusive, inasmuch as the purity of the inert atmosphere is not yet high enough to prevent oxidation and some difficulties in temperoture control have been encountered because of the parasitic emf’s generated in the seals through which the thermocouple wires enter the sealed experimental can. The main effort at present is directed toward eliminating these difficulties. In addition, an inpile test (in air) is running at the conditions of stress and temperature expected in the ARE pressure shell. Two constant-strength cantilever specimens of 0.065-in.-thick Inconel sheet were tested ofter irradiation in the LITR at 100°F to an integraied S8D-B-756 o_ 30 e A m e e Am e e mm Mk e e e f e S e e s e ki ANNEALED 2hr AT 1700°F PRIOR TO TESTS o CANTILEVER BEAM EXTENSION (%) 010 200 “HOLE MB-3 LITR, IN AIR, FAST FLUX, > 10'2 __ HOLE HB-3 LITR, IN HELIUM, FAST FLUX, >10'% "fl.\‘ - - - - """ - - ::..""-’—-’/ T TP BENGH TEST IN AR e HOLE 20 ORNL GRAPHITE REACTOR, IN AIR, FAST FLUX, 4 x10'C 500 800 TIME (hr} Fig. 10.1. In-Reactor Cantilever Creep Tests on Inconel at 1500°F and 3000 psi in Air and Helium, 95 ANP PROJECT QUARTERLY PROGRESS REPORT fastneutron flux (greater than 1 Mev) of 2.5 x 1017, The postirradiation test was carried out ot 1500°F and at a stress of 3000 psi. In Fig. 10.2 it can be seen that the irradiated specimens showed less creep in the earlier stages than did the unirradi- After several hundred hours, the creep rates of both the irradiated specimens and ated controls. the one control sample were the same within the margin of experimental error. This result indicates that at least part of the effect of irradiation under these conditions is to reduce the creep of Inconel, because the radiation-induced hardening effect apparently persists over tens of hours at 1500°F, In similar tests (at 1500°F) on type 347 stainless steel, the creep strength did not appear to be influenced by prior irradiation. The difference in the behavior of stainless steel and of Inconel was expected from the results of hardness data reported previously.3) Several additional samples are now being irradiated. A satisfactory extensometer design has been evolved for the MTR tensile-creep apparatus, and I, W. Davis, J. C. Wilson, and J. C. Zukas, ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 167. 0.4 [ _[ | BENCH TEST 03— ———— 0.2 CANTILEVER BEAM EXTENSION (%) 0 50 100 ~= HOLE HB-3 LITR, FAST FLUX,=10 ANNEALED 2hr AT {TOO®F —_ means for mounting and connecting the apparatus to the specimen have been worked out. The be- havior of the thermocouples under the unusual furnace conditions is now understood sufficiently well to assure reliable results. A heat transfer and fluid-flow mockup of the experimental appa- ratus is being made to test the ability of the unit to dissipate the 15 kw of induced gamma heating. RADIATION EFFECTS ON THERMAL CONDUCTIVITY A. Foner Cohen .. C. Templeton Solid State Division A small Inconel sample was irradiated in the MTR at approximately 600°F for 400 hr in a flux greater than 1013 fast neutrons/cm2.sec. After removal from the reactor, the thermal conductivity of the sample was measured at 300°F and com- pared with the thermal conductivities of unirradi- ated specimens. The irradiation coused no change in the thermal conductivity within the accuracy of the measurements, which was 10%. S5D-B-757 DWG. 194694 /__ Tt 12 150 200 250 - 300 TIME {hr} Fig. 10.2. Post-lrradiation Creep of Inconel in Air at 1500°F and 3000 psi. 96 PERIOD ENDING JUNE 10, 1953 11, ANALYTICAL STUDIES OF REACTOR MATERIALS C. D. Susano, Analytical Chemistry Division J. M. Warde, Metallurgy Division Studies were continued on the use of bromine trifluoride as a reagent for the determination of oxygen in reactor fuels., A procedure was stand- ardized in which the oxygen liberated by the fluorination of uranium trioxide is measured, A calibration curve was established by plotting the increase in terminal pressure as ¢ function of the concentration of oxygen added in the form of uranium frioxide, Efforts to fluorinate zirconium oxide quantitatively were continued. The reaction between zirconium tetra p-bromo- mandelate and sodium methylate was shown to be stoichiometric, and therefore p-bromomandelic acid was established as a reagent for the volumetric determination of zirconium, The concentrations of uranium trifluoride and metallic zirconium in NcF-ZrF4-UF4 {50-46-4 mole %) were determined by the evolution of hydrogen upon treatment with hydrochloric and hydrofluoric acids, Preliminary studies were conducted on the hydrolysis of uranium frifluoride as a function of temperature and time, After 48 hr, approximately 22% of the vuranium originally present as uranium trifluoride was in solution, ";Ditferential colorimetry’’ was applied to the colorimetric determination of zirconium as zir- conium alizarin sulfonate, The optimum concen- fration of the reference solution of the alizarin complex, which represents the maximum precision theoretically possible, was found to be 1.0 mg of zirconium per 10 ml of solution. The method will be applied to the determination of zirconium in reactor fuels when a sufficient order of precision i s obtained, ““Tiron,” disodium-1,2-dihydroxybenzenedisulfo- nate, was shown to be a suitable reagent for the determination of uranium. The composition and the stability of the complex were determined, in addition to the sensitivity of the reagent for the determination of uranium, Petrographic examinations of over 700 samples of fluoride mixtures were completed, Optical data are reported for NaUF,, Na,UF ., NaZrF,, Nuzsz6, Na,ZrF_, Li3CrF6, K3CrF6, N03CrF6, FeF,, K2NaFeF6, NaMF,, LEZZrFé, and Li4ZrF8, The activities of the service laboratory during this quarter included the analysis of 872 samples involving 9335 determinations, CHEMICAL ANALYSES OF FLUORIDES AND THEIR CONTAMINANTS J. C, White, Analytical Chemistry Division Determination of Oxygen in Metallic Oxides with Bromine Trifluoride (J. E. Lee, Jr.,, Analytical Chemistry Division). The use of bromine tri- fluoride as a reagent for the determination of the oxygen which probably exists in reactor fuels as oxides of uranium ond zirconium is being studied.{" In order to conduct the fluorination at the temperatures and pressures which have been indicated by Emeleus(? as necessary for quanti- tative reaction with zirconium oxide and to resist the high corrosivity of the reagent, an apparatus was fabricated from nickel ond stainless steel. The fluorination step is accompanied by quanti- tative liberation of oxygen from the metallic oxide, and the oxygen is collected ond measured in the glass portion of the apparatus, Attempts to apply simple gas law relationships for calculating the oxygen by measuring the increase in pressure of the system were unsuccessful because of the ex- treme temperature gradients encountered in the system, An empirical procedure was established whereby conditions of the fluorination were stand- ardized, and the increase in pressure was plotted as a function of oxygen added in the form of uranium trioxide. The relationship over a pressure range of 0 to 1100 p and 0 to 5 mg of oxygen is shown in the calibration curve (Fig. 11.1). Pre- vious experiments showed that a maximum of 85% of the oxygen was released from zirconium oxide. When uranium trioxide was mixed with zirconium oxide and the mixture was fluorinated for 30 min at 300°C, the results, in some cases, were high by as much as 50%. Subsequent investigation revealed that the stopcocks were not adequate for continbous use under high vacuum, The faulty (D), E. Lee, ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 173. (2)H. J. Emeleus and A. A, Woolf, J. Chem.. Soc., p. 164-168 (1950). 97 ANP PROJECT QUARTERLY PROGRESS REPORT UNCLASSIFIED OWG. 19946 T 1 ( N s ! S OXYGEN {mg) N Q 400 BOO 1200 TERMINAL PRESSURE (u) Fig. 11.1. Calibration Curve for Determination of Oxygen by Reaction with BrF ., stopcacks have been replaced by stopcocks with longer barrels which are better suited for high- vacuum work, Further work to produce quantitative liberation of oxygen from zirconium oxide and to apply the reaction to reactor fuels is under way, Volumetric Determination of Zirconium (T. R. Phillips, Analytical Chemistry Division). The advantages of halomandelic acids as reagents for the determination of zirconium were listed in « previous report.t®) The application of a volumetric method in which p-chloromandelic acid is used was also described previously. A method of syn- thesis of p-bromomandelic acid was obtained (private communication from R, B. Hahn to J. C, White, April 3, 1953), and suitable quantities of the acid were prepared fo investigate its applica- bility to the volumetric technique. In generdl, the bromo derivative is equally as good as the chloro acid. The reaction Zr(p-BrC H ,CHOHCOO), + 4Na(OCH,) —> 4Na(p-BrC H ,CHOHCOO) + Zr(OCH,), proceeds stoichiometricolly, Complete precipi- tation of zirconium takes place within 30 min when a 40-fold excess of reagent is used, The standard deviation of the method is approximately 1%. The compilation of the data on zirconium in the ARE fuel mixture is presently under way in preparation for the writing of a final report on this project, Determination of Uranium Trifluoride and Me- tallic Zirconium (W. J. Ross, Analytical Chemistry Division), The uranium trifluoride and metallic (e, K. Talbott and J. M. McCown, ANP Quar. Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 172. 98 zirconium contents of the fluoride NaF-ZrF -UF (50-46-4 mole %) to which 1 wt % of zirconium hydride was added were determined by the two-step oxidation method described in the previous re- port,(4) Zirconium was calculated from the volume of hydrogen liberated by treatment with 0.2 M hydrofluoric acid, and the trivalent uranium was calculated from the volume of hydrogen liberated by treatment with 9.6 M hydrochloric acid. In connection with this problem, preliminary studies were conducted on the hydrolysis of ura- nium trifluoride as a function of temperature and time, Uranium trifluoride (124 mg) was placed in Erlenmeyer flasks containing 100 ml of water and stirred at 25 + 1°C for various periods of time, An dliquot of the supemnatant liquid was with- drawn, and the concentration of the uranium was determined. The results are shown in Table 11,1, The dota are inadequate ot present for any con- clusions to be drawn, After 48 hr, however, ap- proximately 22% of the uranium originally present in the form of the trifluoride was in solution, The oxidation state of the uranium in solution has not been definitely established, Tests to determine the extent of hydrolysis at 100°C are also planned, TABLE 11.1. CONCENTRATION OF URANIUM IN WATER AFTER CONTACT WITH URANIUM TRIFLUORIDE AT 25°C CONTACT TIME URANIUM (hr) (mmol es) 1 0.016 3 0.046 5 0.056 24 0.061 48 0.096 Colorimetric Determination of Zirconium (D. L. Manning, Analytical Chemistry Division). A method for the colorimetric determination of microamounts of zirconium as zirconium alizarin sulfonate was reported by Green,!5) The method is of insuf- ficient accuracy to permit application to the de- termination of zirconium in reactor fuels, However, (4)W. J. Ross, ANP Quar, Prog. Rep. Mar. 10, 1953, ORNL-1515, p. 172. Glp, g, Green, Anal. Chem. 20, 370 (1948). ‘*difterential colorimetry,”” in which the absorb- ancy of reference solutions of known concen- trations rather than the absorbancy of distilled water is used, has been shown to yield an order of precision comparable to that obtained by gravi- metric techniques. By using Hiskey’s method,(® the optimum concentration of the reference solution of zirconium dlizarin sulfonate was found to be 1.0 mg of zirconium per 10 ml of solution; this reference solution yields the maximum precision., The relative accuracy of the method, as calculated from the formula derived by Bastian, ") was calcu- {oted to be 14: Relative accuracy = s¢ x 2.7 , 0.434 where § = slope of the differential Beet’'s law curve, ¢ = concentration of the reference stond- ard, 0.434 = mathematical optimum absorbancy, 2.7 = instrument factor, The differential colorimetry technigque has been applied to the determination of zirconium in fuel mixtures, and a standord deviation of opproxi- mately 2% was found, This standard deviation is arger than thot of either the gravimetric or volu- metric methods, |t is believed that greater pre- cision can be obtained by using extremely careful techniques. The method is of great potential value because of its extreme simplicity and ropidity, “Tiron'' As a Reagent for the Determination of Uranium (D. L. Manning, Analytical Chemistry Di- vision), A new colorimetric reagent for uranium has been found, Disodium-1,2 -dihydroxybenzene- disulfonate (tiron) forms o reddish-brown complex in basic solution with uranyl ion, The complex is unimolecular, it forms immediately, and it is stable for several hours. The sensitivity of the reagent for uronium is equal to that of ammonium thiocyanate and ascorbic acid, Tests on anionic that sulfate, chloride, and Phosphate in concen- interference reveol nitrate do not interfere. trations greater than 20 times that of vranium does interfere. lron, titanium, and vanadium are the major cationic interferences. [t is planned to test the feasibility of applying differential colorimetry (8)c, F. Hiskey, Anal. Chem. 21, 1440 (1949). ()R, Bastian, R. Weberling, and F. Palilla, Anal. Chem. 22, 164 (1950). PERIOCD ENDING JUNE 10, 1953 to the determination of uranium with this reagent in reactor fuels in which the interference by iron is not serious. PETROGRAPHIC EXAMINATION OF FLUORIDES G. D. White, Metallurgy Division T. N, McVay, Consultant Metallurgy Division Petrographic examinations of about 700 samples of fluoride mixtures were carried out. The optical data collected for various new fluoride compounds are given below, NaUF . Color: green Interference figure: uniaxial negative Birefringence: crystals show first-order gray and yellow between crossed nicols Refractive indices: ¢ = 1.510 E = 1,500 Na,UF Color: green Interference figure: uniaxial negative Birefringence: interference colors are first-order gray Refractive indices: O E = 1,490 NalrF Color; colorless Interference figure: uniaxial negative; some crystals produce a biaxial negative figure with a small optic angle Birefringence: interference colors are first-order white to blue Refractive indices: 0 = 1,508 E = 1.500 Forms solid solution with NaUF Na,ZrF Color: colorless Interference figure: biaxial positive with 2V =75 deg Birefringence: interference colors up to first- order blue Refractive indices: « = 1,419 y = 1.430 Na Z:F Color: colorless Interference figure: unioxial negative Birefringence: interference colors to first-order white 99 ANP PROJECT QUARTERLY PROGRESS REPORT Refractive indices: 0 = 1.386 £ = 1,381 LiCrFy Color: pale green Interference figure: biaxial negative with 2V = 40 deg Birefringence: interference colors through the second order Refractive indices: a = 1.444 y = 1,464 K3ch6 Color: pale green Interference figure: isotropic (no figure) Refractive index: 1,422 Na,CrF Color: pale green Interference figure: isotropic Refractive index: 1,411 FefF Cofor: Colorless Interference figure: uniaxial positive Birefringence: interference colors through second order Refractive indices: 0 = 1,524 E = 1,540 K NaFef Color: colorless Interference figure: isotropic Refractive index: 1,414 NaHF Color: colorless Interference figure: uniaxial positive Birefringence: interference colors through the third order Refractive indices: 0 = 1.261 E = 1,328 LJZZrF6 Color: colorless Interference figure: biaxial positive with 2V =20 deg Birefringence: interference colors through the first order Refractive indices: o« = 1.462 y = 1,482 Li4ZrF8 Color: colarless Interference figure: biaxial negative with 2v = 30 deg Birefringence: interference colors through the first order 100 Refractive indices: a = 1.445 y 1,465 SUMMARY OF SERVICE CHEMICAL ANALYSES J. C, White A. F. Roemer, Jr. Analytical Chemistry Division The work of the Analytical Chemistry Laboratory during the quarter has continued to consist chiefly of the analysis of fluoride fuel mixtures, alkali metal fluorides, ond NaK. Zirconium is being determined gravimetrically by precipitation with mandelic acid and ignitien to the oxide. The ignited oxide has been much purer with respect to contamination by iron and silica than the oxide obtained from ignition of the phenyl- arsonic acid salt. Precision ond accuracy are somewhat improved also, Uranium in these mix- tures is being determined by reduction with the Jones reductor and oxidation with standard ceric sulfate solution with the use of ferroin as the indicator. This method is used because of its simplicity and rapidity. A number of Incone!l specimens, which had been used as containers in corrosion tests of fluoride fuels, was analyzed for concentration of fluorides. The pyrohydrolysis method was used with satis- factory results, During the quarter the laboratory reported 872 samples involving a total of 9335 determinations. These analyses were mode for various groups in the ANP project, as indicated in Table 11,2, TABLE 11.2. SUMMARY OF SERVICE ANALYSES NUMBER OF NUMBER OF REQUESTOR SAMPLES |DETERMINATIONS REPORTED REPORTED Reactor Chemistry Group 267 2274 Corrosion Group 309 3762 Experimental Engi- neering Group 284 3240 Heat Transfer and Physical Prop- erties Dept. 12 59 Total 872 9335 PERIOD ENDING JUNE 10, 1953 12. FLUORIDE FUEL REPROCESSING D. E, Ferguson G. . Cathers O. K. Tallent Chemical Technology Division Development of a method for chemical processing of ARE fuel was initiated to ensure recovery and U235 prior to its allocation and use. The processing method now proposed was adapted from that in use by the Y-12 Chemical Production Division for the processing of uranium fluorides., It was desirable that the method be adaptable to existing facilities at ORNL without much additional equipment and cost, since the fuel decontamination of the moy be further modified in future developments of the ANP program. It was considered in this study that the fuel to be processed was NoF-ZrF -UF, (50-46-4 mole %), which has a melting point of 504°C and o density of 3.0 g/cm?. (Although the ARE fuel is now the 53-43-4 mole % mixture, the considerable research on the 50-46-4 mole % mixture is appli- cable, in general.) The amount of U233 to be processed is 70 kg per batch, that is, 16 to 18 3 of fused salt containing equal parts by weight of NaF-ZrF -UF , and flush material consisting of ZrF , and NaF (50-50 mole %). DISSOLUTION OF ARE FUEL The initial work on the preparation of an aqueous solution of ARE fuel was carried out on material containing 4 mole % UF ,, that is, undiluted ARE fuel. The fused salt was used in the form of very coarse lumps to simulate probable conditions in ARE fuel processing. Exploratory solution experiments showed ‘that the fuel mixture could be dissolved in an alumiaum nitrate~nitric acid solution at reflux temperature. In effect, this leads to oxidation of guadrivalent uranium to uranyl ion, complexing of the fluoride ion with aluminum, and conversion of insoluble NaF and ZrF, to soluble compounds. Although in this preliminary work, solutions containing some insoluble precipitate were obtained, the method promised to be of value with further development. Additional tests with different proportions of aluminum nitrate nonahydrate (ANN), concentroted nitric acid, and water were run to see whether complete solution was possible and to observe dissolution rates. The results showed that fast and complete solution was possible with various material proportions. The optimum proportions seemed to be an ANN-to-fuel ratio of 4 to .5 with 20 to 30 ml of acid and 40 to 50 ml of water for a 5-g batch of fuel. Less acid and less ANN led to a decrease in rote of dissolution and some residual precipitate. Best results were also obtained when an optimum amount of water was present to dilute the ANN-nitric acid solution. The actual composition of the fuel used in the above tests was 4.25 47.9, and 47.9 mole %, re- spectively, of UF ,, NaF, ond ZrF,. An ANN-to- fuel ratio of 5 therefore corresponded to a tlueride- to-aluminum mole ratio of 1.7, while the lower limit ratio of 3 corresponded to a mole ratio of 2.8. The value of 1.7 appears preferable, because it ensures more complete complexing of fluoride ion and thereby less corrosion to processing equipment. It also aveids approaching too closely to the mole ratio of 3 represented by the insoluble compound AIF,. The use of more aluminum would lead to faster dissolution and less corrosion, but it would also lead to a lower uranium content in the feed for solvent extraction processing. On the basis of 1.7 as the fluoride-to-aluminum mele ratio, the following weight proportion was selected for feed preparation in solvent extraction work: Weight Proportion ARE fuel (210 4 mole % UF4) 1 ANN 5 Ceoncentrated nitric acid 7 Water 11 This weight proportion leads to uranium concen- trations of 2.2 and 4.3 mg/mi, respectively, for fuel plus flush material (2 mole % UFA) and for fuel alone (4 mole % UF,). Other concentrations used are, approximately, 4 M HNO, 0.2 M Zr, 0.2 MNa, 1.2 M F, and 0.67 M Al Loter work showed that if the femperature was raised too rapidly in heating the inhomogeneous mixture of all sofution components, the ARE fuel began to dissolve in the lower layer that was rich 101 ANP PROJECT QUARTERLY PROGRESS REFPORT in aluminum nitrate before thorough thermal mixing had occurred. This led to formation of a yellowish precipitate which was not redissolvable. [t is preferable, therefore, that a homogeneous solution of ANN, nitric acid, and water be obtained before addition of the ARE fuel. SOLVENT EXTRACTION Feed solution made from 2 mole % UF& material (fuel plus flush) was used in batch countercurrent runs to test the feasibility of a TBP solvent exiraction process. 1he low uranium concentration of 2.2 mg/ml made operation of such a process at the usual 60 to 70% uranium saturation of the TBP impossible. Too low a concentration of TBP in Amsco to obtain an opprecioble saturation would result in low exiraction factors and o large uranium loss in the hot wasie, A compromise was there- fore tried in three batch countercurrent runs with 3, 5, and 7% TBP in Amsco 123-15 with cold feed salution. A nonocid scrub of 0.67 M ANN was vsad to maintain high salting strength in the scrub secticn and to secure an acid-free product suitable for concentration and isolation by ion exchange. The distribution coefficients in both the extrac- tion and the scrub sections showed that 5 and 7% TBP was very satisfactory and that the uranium losses in the waste were less than 0.01%, The 3% TBP run, however, gave a uranium distribution co- efficient of less than 2 in the scrub section, and the vranium loss was 0.2%. Zirconium carryover with the uranium varied over the range of 0.02 to 0.06 mg/ml for the three cases; thus, zirconium carryover does not appear to be dependent on TBP concentration. Fluoride and acid decontaminations were likewise satisfactory in the 5% TBP run. Two runs with feed spiked with trace fission products have shown that adequate first cycle radicactive decontomination is obtainable with both 5 and 7% TBP in Amsco. Gross beta decon- tamination was better than 104 in both cases; thus, the TBP concentration is not critical. The wranium saturation of the TBP, 10 and 7%, respectively, was far below that for which TBP concentration is critical, A rough schematic flowsheet can be drawn on the assumption thot decontamination will be suf- ficient in one solvent extraction cycle for coupling A second cycle will probably be necessory only for material of very high burnup. to an ion exchange column (Fig. 12.1). CORROSION IN DISSOLUTION OF ARE FUEL Corrosion tests have been carried out on types 309 and 347 stainless steel with a solution of ARE fuel (2.2 mg of uranium per m!) operated with reflux at aboutr 110°C. desirable, since it was expected that a 4 M nitric acid—1.2 M fluoride ion solution would be very corrosive, despite the high concentration of aluminum, and since most equipment available for scaling up the process would be made of either type 309 or type 347 stainless steel. The tests were made by suspending square plates of the materials at the liquid-vapor interface, in the liquid, and in the vopor of the test solution. Glass supparis were used to avoid metallic junc- tion potentials. The total time of the tests was 33.5 hr, extended over a five-day period, with weighings being made at the end of each day’s run. The data are summarized in Table 12,1. The re- sults clearly indicate a very large corrosion rate, Type 309 stainless steel was somewhat better than type 347, presumably because of its higher The highest cor- rosion rates were observed duringthe initial period, but they decrease to a third or less during later Information on corrosion was nickel and chromium contents. periods. The lower rates will probably prevail in TABLFE 12.1, CORROSIOM TESTS ON TYPES 309 AND 347 STAINLESS STEEL IN ARE FUEL DISSOLVER AT 110°C CORROSION RATE (mpy) Type 309 Stainless Steel Type 347 Stainless Stes! At Interface In Liquid in Vapar At Interface in LLiquid In Vapor First 3.5-hr period 153 131 117 238 166 129 Last 7.5-hr period 31 24 19 57 62 38 Over-all 33.5-hr period 53 44 35 g1 67 59 102 PERIOD ENDING JUNE 10, 1953 DWG {97084 SCRUB 9.67 M ANN WATER (AsNBng) { VO*L‘UME 1.25 VO*LUMES ko) /// - \\w__._._. SOLVENT wasTE WATER (128 kg) / \ [ 1 \ //% \\\\ SECOND SOLVENT EXTRACTION CYCLE FEED PREPARATION - voLumes ///gé \\\ié\ OR ION EXCHANGE COLUMN arerue==l 2z m0aag i N\ T 6 kg) OF U PER LITER ?’U¢ &8\% /S/ \xi“z"’\ PRODUCT ézé \\@% B.8 TO17.2 g ///i/ \\\E\ OF U PER LITER % NEX EXTRACTANT / § o Top 5 VOLUMES /A hk\\ - IN AMSCO * HOT WASTE Fig. 12.1. Flow Sheet for ARE Fuel Processing. Basis: 1kg of U235, equipment which has already suffered from nitric acid corrasion. The corrosion was highest at the liquid-vapor interface and lowest in the vapor. PLANT PROCESSING It appears, on the basis of present knowledge, that batches of ARE fuel plus flush material con- taining 70 kg of U?35 can be decontaminated and recovered by using two solvent extraction cycles coupled with an ion exchange column, The use of Metal Recovery Plant equipment for this process is possible without major edditions provided the following precautions are observed to avoid criti- cality. 1. Proceeding with processing of ARE material on a crash basis will require batch operation and a high order of supervision to ensure safety with respect to criticality. Frequent sampling will be necessary and carewill have to be taken to prevent any heel buildup in the dissolution step. The process should be scheduled on the basis of 1-kg batches of U235, 2. One-kilogram batches of U235, that is, about 1 gal of ARE fuel or 2 gal of fuel plus flush, con be transported in cylindrical aluminum cans of noncritical geometry with adequate shielding. 3. Dissolution of the material can be carried out in a 500-gal tank in the Metal Recovery Plant. The charge would be 60 gal for ARE fuel or 120 gal for fuel diluted with an equal weight of flush material. Criticality is to be avoided partly by slab geometry, 4. Two batches of feed may be stored in the 2500-gal tank of the Metal Recovery Plant that has a concave bottom. Noncriticality is favored by the ring geometry. 5. The 1A pulse column available for this work unfortunately extends into a sump well in which neutron reflection must be considered. A safe concentration limit, however, would be 12.5 mg of U235 per ml, or a concentration 3- to 6-fold above that of the feed being considered. Several coincident errors, such as the use of water in place of 0.67 M ANN solution for the scrub and reflux and buildup of uranium at the feed plate with subsequent leakage from the column into the sump well, would be necessary to approach crificality. 103 INTRODUCTION AND SUMMARY E. P. Blizard J. L. Meem, Associate Physics Division The measurements of neutron and gamma spectra in the divided-shield mockup in the Bulk Shielding Facility constitute an important advance in shield- ing research (sec. 13). These spectra give the energy and angular distribution of neutrons from a fission source after attenuation in water and are directly comparable to some calculations being carried out under subcontract by Nuclear Develop- ment Associates. [t is by the extension of this approach that a real understanding of shield design will be achieved. The Lid Tank Facility is being used for the engi- neering type of studies of unit shields for which it is especidlly applicable (sec. 14). Many combina- tions of materials and of thicknesses of materials are being tried in an effort fo obtain the lightest possible unit shield for the reflector-moderated liquid-fuel reactor. The majoruncerfainty remaining is the possibility that the delayed neutrons and gamma rays from the circuldting fuel in the heat exchanger cannot be measured in the Lid Tank Facility, Efforts have been initiated, however, to calculate these components in order to evaluate their possible importance. If their importance appears fo be overriding, which might well be the case, some special experiments will have to be devised., With the use of this facility, on effective removal cross section of 0.99 b was obtained for oxygen. ' : Construction of the Tower Shielding Facility, which was delayed by strike, has now been resumed and should be completed by the end of 1953 (sec, 15). The report to the Reactor Sofeguard Committee revealed no untoward hazards, A Shielding Session is scheduled for the month of June in Oak Ridge. The pumose of the session will be to stondardize methods of shield design, and members of a number of interested organiza- tions will porticipate. In order to ensure that the problems studied will be generic to the basic question of shield weight, actual shield designs will be studied for the several cycles which are of most interest. A report will be issued, probably late in July. 107 ANP PROJECT QUARTERLY PROGRESS REPORT 13. BULK SHIELDING FACILITY J. L. Meem R. G. Cochran M. P. Haydon K. M. Henry H. E. Hungerford E. B. Johnson J. K. Leslie T. A. Love F. C. Mgienschein G. M. McCammon Physics Division During this quarter, the major effort at the Bulk Shielding Facility has been devoted to measure- ments on the top-plug mockup for the SIR shield. This work has been very useful in that it offorded an opportunity to study the perturbations to a shield introduced by engineering features. A description of these experiments will be given in the next Physics Division semionnual progress report. Some results of the neutron spectral meas- urements, as well as results of air-scattering cal- culations on gomma rays for the divided aircraft shield, are presented here; the final results on the energy-per-fission experiment are also given, NEUTRON SPECTRA FOR THE DIVIDED SHIELD The fast-neutron spectrometer(” at the Bulk Shielding Facility has been used to measure the reactor spectra as attenuated by the beryllium oxide reflector and various amounts of water,{?) The experimental setup used is shown in Fig. 13.1. An odluminum collimator tube 6 ft long (0D, 1 in,; 1D, 7/8 in.) was used for the measure- ments close to the reactor (0 and 12.5 em). The 0-cm distance was the position in which the collimator was pushed through a hole in o beryl- lium oxide reflector can so that the end of the collimator tube touched a fuel element. For the 27,5~ ond the 37.5-¢m distances, the collimator length was reduced to 4 ft to increase the neutron intensity, Previous experiments at this facility have indicated that as long as the ratio of length to diameter is large, the collimator does not distort the neutron spectrum, The principal purpose of the lead spectrometer housing (shown in Fig. 13.1} was to ottenuate the (”R. G. Cochran and K. M, Henry, A Proton Recoil Tgpe) FastNeutron Spectrometer, ORNL-1479 (Apr, 2, 1953). (Q)R. G. Cochran and K, M. Henry, Neutron Spectra of the BSF Reactor, ORNL CF«53.5-105 (in press). 108 infense gamma rays encountered in making measure- ments close to the reactor. In addition, it served as a waterprocf container for the spectrometer components. lo ensure water exclusion, the box was pressurized with 2 to 3 psi of dry air. A logarithmic plot of the data obtgined out to 37.5 cm from the reactor is given in Fig. 13.2 The data indicate that the spectrum hardens ap- preciably as the neutrons penetrate largerdistances of water, In addition to the proton-recoil spectrometer data, threshold detector measurements(®! have been made in a fuel element of the BSF reactor. Figures 13.3 and 13.4 show the spectrum obtained by this method. ENERGY PER FISSION AND POWER OF THE BULK SHIELDING REACTOR The energy per fission in the BSR has been measured as 193 Mev.*) In addition, the graphite sigma pile, which is the stondard for thermal- neutron flux measurements ot ORNL, hos been recalibrated. ) Since all shielding dato at the BSF have been reported as so much radiation per watt of reactor power, a correction should be applied to all previous data. The fast-neutron and gamma data should be multiplied by o factor of 0.8306 and the thermal-neutron data should be multiplied by a factor of 0.9250. A list of reports to which these corrections apply has been pub- lished.(®) (B)J. B. Trice, Neufron Threshold Measurements in the BSR, ORNL CF-53-5-13%9 (in press). (4)J. .. Meem, L. B. Holland, ond G. M, McCammon, Determination of the Power of the Bulk Shielding Re- actor. Part {{l. Measurement of the Energy Released per Fission, DRNILL-1537 {to be issued). (S)E. D, Klema, R, H. Ritchie, and G. M, McCammon, Recalibration of the X-10 Standord Graphite Pile, ORNL- 1398 (Oct, 17, 1952). (6)J. l.. Meem, E. B, Johnson, and H. E, Hungerford, Energy per Fission and Power of the BSR, ORNL CF- 53.5-21 (May 12, 1953). =7 HOLE IN REFLECTOR CAN Fig. 13.1, merged in water, GAMMA-RAY AIR-SCATTERING CALCULATIONS Measurements have been reported”) on the spec- tra of gamma radiation emerging from a mockup of the reactor portion of the aircraft divided shield.(® Calculations have been made for the gamma radie- tion arriving at the outside of the aircraft crew shield, and further calculations are plonned on the penetration of the gamma radiation through the (7)F. C. Mdienschein, Gamma-Ray Spectral Measure- ments with the Divided Shield Mockup, ORNL CF-52-3-1 (Mar. 3, 1952); Part |l, ORNL CF-52-7-71 (July 8, 1952); Part 111, ORNL CF-52-8-38 {Aug. 8, 1952). (S)Reporf of the ANP Shielding Board for the Aircraft Nuclear Propulsion Program, ANP-53 (Oct. 16, 1950). \ FUEL ELEMENTS . // P '/hBERYL_LlUM OXIDE REFLECTOR CANS -COLLIMATOR Experimental Arrangement for Hewtron Spectral Measurements, PERIOD ENDING JUNE 10, 1953 OWG. 19887 - LEAD SPECTROMETER HOUSING Apparatus operates sub- crew shield, The data collected recently by the Bureau of Standards(®) will be used, in part, for these calculations, The gamma-ray spectral measurements were recorded?) as a function of energy and as a function of the angles, as shown in Fig, 13,5, Data obtained through the lead shadow shields shown in Fig. 13.5 and through various thicknesses of water were available, The data included the dependence of I" (the emerging gamma-ray flux in (?)F.. S. Kirn, R, J. Kennedy, and H. 0. Wyckoff, Oblique Atftenuation of Gamma Rays from Cobalt-60 ond Cesium 137 and Polyethylene, Concrete and [ead, NBS- 2125 (Dec, 23, 1952). 109 ANP PROJECT QUARTERLY PROGRESS REPORT DWG. 193692A CCRRECTED TO I-WATT REACTOR POWER O COLLIMATOR AGAINST FUEL ELEMENT AScm H20+?.50m BeO ¢ 20cm Hy0+7.5¢m BeO V 30cm H;0+7.5¢cm BeO ¢ (neutrons /cm? /sec/Nev/Q /watt } 6o 1 2 3 4 5 & 7 8 9 10 NEUTRON ENERGY (Mev) Fig. 13.2. BSF Reactor Neutron Spectra, photons/cm?/sec/Mev/watt/steradian) on the ele- vation angle 9 and also on the azimuth angle ¢ by virtue of the existing symmetry along the aircraft axis, Gamma-ray dose measurements indicated that data token behind the lead shadow shield would %(Eifi) mf;faffefqé be applicable for « less than 40 deg, whereas the open-water data would be applicable for a greater than 70 deg, where a is the polar angle indicating the point of departure of the photons from the re- actor shield. The intermediate region at the edge 110 5SD-A-709 D'wG. 184004 T - ] e nfemiem - S !7,7 P - - - " A|27 A!ZB”" BULK SHIELDING FACILITY A (7. REACTOR POWER: 100 kw 12 V0" e e N ! ] o L S | R | - —t R g e } T < ! $(£), DIFFERENTIAL FLUX (neutrons/cm?/sec/Mev} 11 i | 10 | !| R — 0.0001 0.001 0.01 0.05 ENERGY {Mev)} Fig. 13,3, Neutron Energy Spectrum for the Epithermal Energy Region (0.5 kev < E < 10 kev) by Threshold Detector Measurements, of the shadow shield had to be investigated in further detail. Measurements were made at a = 50 deg for all values of 6 and at a = 60 deg as a function of both @ and ¢, The measurements then afforded an approximate knowledge of 1'(E,q,0,0) over the entire reactor shield, where E is the energy (Mev) of the emerging gamma ray. The gamma radiation arriving at the crew shield was then obtained by integrating over the energy and the surface of the reactor shield and by applying the differential scattering cross section and inte- grating over all space. Single scattering and no air attenuation were assumed in the calculations, which should be adequate with the reactor-crew separation distances being considered. 1(E,a,0,¢) KR™2 dE dA 4V WI'I ere ¢ ~— = gommaeray flux arriving at the crew shield P (photons/em?/sec/watt), E’ = degraded gamma-ray energy {(Mev), receiver angle, = energy of the emerging gamma ray (Mev), T | a = polar angle at the source, r = distance from the point of emergence from the reactor shield to the point of scatter (em), ¢ = elevation angle, ¢ = azimuth angle, K = Klein-Nishina differential scattering cross section,“o) R = distance from the point of scatter to the center of the crew shield (em), dA = surface element at the reactor shield, dV = volume element of scattering space. These calculations were simplified considerably because, for the most part, the gomma radiation emerged radially in the shield regions away from the edge of the shadow disks., Calculations for a typical element of volume were carried out by using the measured dependence on @ ond ¢ and by assuming that all the radiation was emitted radial ly, Since the results agreed fo within 30%, the radial assumption was used for all regions away from the shadow-disk edges. This component was called the radial beam, while the contribution from the edge of the shaodow shield was called the skew beam. The direct (unscattered) beam calculation was given simply by ¢ , 2 — (EB) = IT'R™ dE du . E EF "a The actual integrations were performed numeri- cally, in steps, as shown in Table 13.1. The (]D)C. M. DPavisson and R. P. Evans, Revs. Mod. Phys., 24, 79 (1952). TABLE 13,1, PERIOD ENDING JUNE 10, 1953 computations were made on a UNIVAC after they had been coded by the Mathematics Panel. Differential Results, The specirc of the scattered gamma-ray flux arriving at the crew compartment are shown in Fig. 13.6 for the front region of the SED-A=-T10 DWG. 1B40IA 7x1c" n 7 7 Aia {r, ,c:v)Mg2 A FLUX {neutrons /om?- sec) BULK SHIELDING FACILITY 5 REACTOR POWER: 100 kw 0 2 4 6 8 10 12 THRESHOLD ENERGY (Mev) Fig. 13.4. Neutron Flux Above Threshold Energy as a Function of Threshold Energy. : STEPS CHOSEN FOR NUMERICAL INTEGRATION CALCULATION VARIABLE Radial Beam Skew Beam Direct Beam 7 6 steps of 1.2 to 2.5 Mev 4 steps of 1.2 to 2.5 Mev 6 steps of 1.2 to 2.5 Mev a 18 steps of 10 deg 2 steps of 10 deg 9 steps of 10 deg fi 18 steps of 10 deg - r 40 steps of 5 to 500 + 111 ANP PROJECT QUARTERLY PROGRESS REPORT N 90° ) 90° . ‘ \\“ POINT OF SCATTERING y LEAD DWG. 19374 SHADOW SHIELDS pa REACTOR—~ e - 'REACTOR SHIELD CREW GHIELD MEASURE: T'(E,a, 8, ¢) CALCULATE: T"(E’, B) Fig. 13.5. Definition of Angles for Gamma-Ray Air-Scottering Calculations, DWG. 19868 0.012 r ‘ —_ S AREA UNDER CURVE = &/p 2 soo Lo SOURCE ANGLE (@)=0 TO 65 deg ) > (SHADOW-SHIEL.D REGION) o = ~ D >~ 0008 | RECEIVER ANGLE, B (deg) CODE £ 0 TO 180 —A— T = 0 TO 30 B 2 30 TO 80 — G o 60 TO 90 meme Do o 0008 — 90 TO 120 e = 5 120 TO 150 e & 150TO 180 e Grenne- o E 0.004 |- AT ] E s e —— o B m > - ~D--1 A 2 e - g 0002 f— |c oA - Q ~TB ~ o _B__ - "“"%”"‘“ ------- ¢ F--p--- ] . | L 0 ¢ oG ] A I 0 1 2 3 4 E/, POST-SCATTERING ENERGY (Mev) Fig. 13.6. S3pectra at the Crew Compartment of the Scottered Gamma Rays from the Front of the Re- actor Shield, 112 DWG. 19869 48 | ] g SOURCE ANGLE (@) =65 TO 125 deg £ 40— r—-a (SIDES OF REACTOR SHIELD) = o 5 o S RECEIVER ANGLE , B{deg) CODE < O TO 180 —A— £ 0 70 30 ——B-— 3 0 10 80 - o 070 90 wmnDee £ 24— 90 TO 120 . 2 L 8 & E 15— AREA UNDER CURVE =Dsp [ey) o = C——] = - =B a - D--- o 8 . <) b ene] 0 Ay I I 0. 0.5 1.0 1.5 20 £7, POST-SCATTERING ENERGY (Mev) Fig. 13.7. Spectra at the Crew Compariment of the Scottered Gamma Rays from the Side of the Reactor Shield, shield. from this region. All the high-energy flux arrives The side of the reactor shield contributes much more low-energy radiation as shown in Fig. 13.7 and, finally, Fig. 13.8 shows the small contribution of low-energy radiation from the rear of the shield. A graphical representation of the contribution from various volume elements in space is shown in Figs. 13,9 and 13.10. Figure 13,9 gives the flux contribution in photons/cm?/sec/watt, while Fig. 13.10 is shaded to demonstrate the same informa- Thus it may be seen that regions far from the aircraft contribute a substantial portion of the gamma radiation arriving ot the outside of the crew compariment. reactor tion for larger volume elements. However, since this radiation has to be scattered through a large angle, it is of low energy and it loses some of its importance after penetrating the crew shield. All gamma rays arriving at the crew shield with an PERIOD ENDING JUNE 10, 1953 . DWG. 19B70 32 3 l & | @ < SOURCE ANGLE (@)=125 TO 180 deg B (REAR OF REACTOR SHIELD) = 5 el - 3 RECEIVER ANGLE, S{deg) CODE e 0 TO 180 —p < 0 TO 30 —— - Q 2 30 TO 60 - 8 18— o o £ S S AREA UNDER CURVE = $/p £ E Q & o 2, L L a 0 0.5 1.0 1.5 2.0 % £/, POST-SCATTERING ENERGY (Mev) Fig. 13.B. Spectra at the Crew Compartment of the Scattered Gamma Rays from the Rear of the Reactor Shield, energy greater than 0.5 Mev, for example, must have a+ 3 < 90 degrees. This condition implies scattering from a point closer thon B.6 meters from the aircraft axis. The direct-beam radiation orriving at the crew shield is shown in Fig. 13.11. Since this contri- bution is ploited as a function of the angle g, rather than as that of the solid angle, the dose will tend to increase with a. Counteracting this, how- ever, is the fact that @ increases with increasing a and that the radial nature of the emerging radia- tion causes the contribution to rapidly decrease. The marked increase at the edge of the shadow shield indicates that the shadow shield should be made larger and that it should be tapered in thick. ness toward the edges. Integral Results, The integral results are shown in Table 13.2, for which the gamma flux was converted into dose units.!17? The results are compared with the calculations of ANP-53,¢8) and they agree to within a factor of about 3. This is (”)E. P. Blizard, Introduction to Shield Design, (]:);?Ng_ CF-51-10-70, Part | Revised, p. 41 {Jan. 30, 52}. 113 DWG, 194834 Noooose " < o o o o e o w g -— o T S S : : S60000" el o 00 QMOOOO,O — ~ © 3 0.000048 g g ——— o o el o o \\\u\\\n\n 0 OOOO.mU S ~ ; e 15 L~ g & S S e W o Qo & o o o Q o < Q.20002Q N, Q.000024 0.000053 0.000070 0.000063 0.000058 0.000074 0.0000%90 0.000023 0.00007% 0.00C08 0.00015 0.00022 0.00032 0.0003049 0.085%4 \ 0.000065 \ \0.000048 | \0.000038 0.000034 i 0.000027 // 0.000019 0.000024’ / 0.000018 0,000022 0.000048' 0,000048 00000208 VOo {10 Jgo -O) 5 s /qé‘o 20° & “y O 3 155 165° SEE INSERT ¢80° 2 VALUES ARE IN photons/em™/sec/ watt 0.000035 0.0000179 ./ REACTOR 10° 0.0859y 180° CREW COMPARTMENT Fig. 13.9. Gamma-Ray Scottering for an Aircraft Divided Shield, 115 DWG. 19470A e REACTOR CREW COMPARTMENT 0.0005 TC 00002 0.0002 TO 0.0004 VALUES ARE IN pho‘rons/cmd/sec/wa’r*r Fig. 13.10. Gammo-Ray Scatfering for an Aircroft Divided Shield, TABLE 13.2, TOTAL GAMMA DOSE AT THE OUTSIDE OF THE CREW SHIELD SCATTERED BEAM N -6 DIRECT BEAM METHOD (+/hy/watt x 1077 {r/he/ watt x 10'6) Radia! Skew This report 2.3 . 0.034 29 ANP-53 3.0* Experiment 50 *Corrected for a leckage ratio of 5. 117 ANP PROJECT QUARTERLY PROGRESS REPORT S DWG. 194774 0.8 TO 2 Mev % {gammos/cmzfsec/wafl/io deg} o 2 TC 3.5 Mev 1072 3.5 TO 5 Mev -5 5 15 25 35 45 55 65 75 85 a{deg) Fig. 13.11. Direct-Beam Gamma Rays Arriving at the Crew Shield as a Function of a. 118 surprisingly good agreement, since for the calcula- tions in ANP-53 it was assumed that all the gamma rays had an energy of 2.5 Mev and that they emerged radially. A comparison was also made with the air-scatter- ing experiment{12) which was carried out at the Bulk Shielding Facility. After corrections had been made for the difference in reactor-crew sepa- ration distance, the calculations just described indicated o dose about five times less than that measured in the air-scattering experiment. This disagreement is not to be interpreted as fundo- mental, since the experimantal arrangement was but a poor replica of the shield which was calcu- lated. {IZ)J. L. Meem and H. E. Hungerford, Air Scattering Experiments at the Bulk Shielding Facility {Preliminary [ssue), ORNL CF-52.7-37 (July 8, 1952); H, E. Hunger- ford, private communication, PERIOD ENDING JUNE 10, 1953 14. LID TANK FACILITY J. D. Flynn G. T. Chapman J. N. Miller F. N. Watsen Physics Division M. E. LaVerne F. H. Abernathy ANP Division Experiments on a unit shield for the circulating- fuel reflector-moderated reactor, as described in the previous quarterly report, are in progress. Fifty-four configurations have been tested to dafe. Effective removal cross sections have been ob- tained for oxygen, nickel, ond Inconel. REFLECTOR-MODERATED REACTOR SHIELD TESTS A preliminary series of shielding tests has been run in the Lid Tank Facility on the basic type of shield for the reflector-moderated re- actor.V Three basic refl ector—intermediate-heat- exchanger—pressure-shell configurations have been tested, Various thicknesses and spacings of lead layers were tested with each configuration, A typical assembly was then used to determine the effects of such factors as boron concentration, replacement of iron by Inconel, water by trans- former oil, lead by tungsten carbide, etc, The only berylliium available for this first series of tests consisted of one tank of beryllium pellets approximately 1 ft thick with an average beryllium density of 1.23 g/em® and two beryllium slabs, 21 by 42 by 3.6 in., with a density of 1.84 g/em®, By placing the two beryllium slabs together to form a 42-in. square and placing this square in series with the tank of beryllium pellets, an equivalent thickness of 11.3 in. of solid beryllium was ob- tained. The heat exchanger was simulated by using thin, slab-shaped steel tanks that were 3.5 cm thick and filled with NaF. The sheet-steel tank walls of the tanks served to simulate the matrix of tubing that would be present in a full- scale heat exchanger. Provision was made for the insertion of capsules of NaF in each tank to permit measurements of sodium activation at wvarious points in the heat exchanger. A typical shield configuration is shown in Fig. 14.1. “)A. P. Fraas and J. B. Trice, ANP Quar. Prog. Rep. Mar, 10, 1953, ORNL-1515, p. 74. The capsules consisted of 0,250-in.-0OD aluminum tubing attached to the lower end of 0.250-in.-dia aluminum rod. Even though the heat exchanger tanks were loaded with NaF in as dry a condition as possible and were closed off, it was feared that the moisture content might change during the course of the test and have an effect on the amount of neutron moderation in the simulated heat ex- changer. Therefore the central position of each heat exchanger tank was filled with cans of sodium rather than with NaF powder. The cans of sodium were borrowed from the Critical Experiment Fa- cility. DWG.19871 /VOID T // - N, S e —— T - - "l e % 1 - ] ) — e ‘-;‘_\ E ™ - -/, _Cfl;"’ e W8 hE! ol el T ) / ™ ~ -0 . /, __B\. v N ~ N, AN B — Lt '/Cr‘&