MAATIN MARIETTA ENERGY SYSTEMS LIBRARIES “ | | i . ! i il 3 445k 03L0O97?L 7 33 3 Ly - DOCUMENT COLLECTION LIBRARY LOAN COPY DO NOT TRANSFER TO ANOTHER PERSON 3 e wisk semmene sive 10 sed o ogend inopame with dosument and S mpvange 8 lear | | ORNL- 1294 This document consists of 202 pages Ty Copy é@;&f0f1239» Series A. Contract No. W-7405-eng-26 AIRCRAFT NUCLEAR PROPULSION PROJECT QUARTERLY. PROGRESS REPORT for Period Ending June 10, 1952 R. C. Briant, Director - J. H. Buck, Associate Director A, J. Miller; Assistant Director . Edited by: W. B. Cottrell DATE 1SSUED OAK RIDGE NATIONAL LABORATORY operated by CARBIDE AND CARBON CHEMICALS COMPANY : A Division of Union Carbide and Carbon Corporation Post Office Box P " OAK RIDGE, TENNESSEE TR 3 445L D3ILO97?L 7 il INTERNAL DISTRIBUTION 1. G. M. Adamson 2. C. J. Barton 3. D. S. Billington 4. F. F. Blankenship 5@ FE. P. Blizard 6?%%$ Brasunas 7. R 8. F 9. gww - o = - o oQ o = ot J. H uck 10. A. D. 1lihan 11. D. W, dell 12. C. E. Centeg 13. J. M. Clsar-fl% 14. G. H. Clewett %, 15. C. E. Clifford %, 16. W. B. Cottrell & 17. D. D. Cowen 18. VW, K. Eister 19. L. B. Emlet (Y-12) 20. W. K. Ergen 21. G. T. Felbeck (C&CCC) 22. A. P. Fraas 23. W. R. Gall 24, C. B. Graham 25. W. W. Grigorieff (Consultant) 26. W. R. Grimes 27. L. F. Hewmphill 28. A. Hollaender 29. 30. 31. 32. A. S. Houspholderfif W. B. Humes (K- 255 R. J. Jones x G. W. Kellholi?“ fl?47 33. 34. 35. 36. 37. 38. 39. 40. 41. 42, 4:3 [ ;:,‘;-' 44:,--"' 43. 46. 48. 49. 50. 51. 52. 53. 54. 55. 56. “58. 60.%, 61. 62. 63. C. P. Keim M. T. Kelley F. Kertesz E. M. Kiag J. A. [_égéne C. E.fiiarson R. S& Livingston B.gfi. Lyon W,/ D. Manly KW; B. McDonald S J. L. Meem A, J. Miller K. Z. Morgan E. J. Murphy H. F. Poppendiek P. M. Reyling H. W. Savage R. W. Schroeder E. D. Shipley 0. Sisman I.. P. Smith (Consultant) A. H. Snell F. L. Steahly. R. W. Stoughton C. D. Susano J. A. Swartout E. H. Taylor F. C. Uffelman %fi, C. VonderLage fi%,M. Weinberg E. P, Wigner (Co% sultant) ORNL-1294 Progress » 64. C. E. Winters 83. Metallurgy Librg# . 65-Td. ANP Library B4. Physics Libra _f 75% Biology Library 85-89. Technical Ip#8rmation 76-79. Wgentral Files Departmer 80-81. Mgguistry Library 90-91. Traininggpchool Library 82. Hed¥gh Physics Library : B _ EXTERNAL DISTRIBUTION 92-102. National Laboratory 103. ces Special Weapons Proje 104-111. Avomic EHg 112. Battelle M 113-115. Brookhaven 116. Bureau of Aet 117. Bureau of Ship 118-119. Bevelopment Company 120-125. %Fcals Company (Y-12 Area) 126. Chicago Patent Gr 2 127. Chief of Naval Besg8gch 128-132. duPont Company & % 133-157. General Electrigt gy ( ANPP) 158-161. . Richland 162. 163. rters, Office’ Assistant for Atomic Energy 164-171. pations Office | 172, JIowa St College 173-176. #Fitomic Power Laberato 177-178. #nd Area Office 179-181. _ | 182, # achusetts Institute of TechnoBpgy (Benedict) 183. s sachusetts Institute of Technold®y (Kaufmann) 184-186. Mound Laboratory 187-190. gNational Advisory Committee for Aeron#ftics, Cleveland 191.4 National Advisory Committee for Aeronaifgcs, Washington 192—19£§‘ New York Operations Office 194-19§. North American Aviation, Inc. 199. Nuclear Development Associates (NDA) 19%. Patent Branch, Washington 111 1v "%% 198-199. 200. 201. 202-203. 204. 205-208. 209-224. 225-239. ..:...g‘.::.:_‘ U “Band Corporation Savvannah River Operations Office (Augusta) Savaniah River Operations Office (Wilmington) University of Califérnia Radiation Laboratory Vitro Corporapipflwof America Westinghouse ‘Electric Corporation Wright Ai# Development Center Techpni#dal Information Service, Oak Ridge e b A Reports previously issued ORNL-528 ORNL-629 ORNL-768 ORNL-858 ORNL-919 ANP - 60 ANP-65 ORNL-1154 ORNL-~1170 ORNL-1227 ORNL-1234 Period Period Period Period Perio& Perioa Period Period Period Period in this series are as follows: Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending November 30, 1949 February 28, 1950 May 31, 1950 August 31, 1950 December 10, 195Q March 16, 1951 1951 September 10, 1951 December 10, 1951 1952 June 10, March 10, Reactor Program of the Aircraft Nuclear Propulsion Project b e ] TABLE OF CONTENTS PAGE FOREWORD 1 PART I REACTOR THEORY AND DESIGN 3 SUMMARY AND INTRODUCTION , 5 1. CIRCULATING-¥FUEL AIRCRAFT REACTORS 6 Reflector-Moderated Circulating-Fuel Beactor : 6 - Spherical Heat Exchanger and Shield Design 7 2. CIRCULATING-FUEL AIRCRAFT REACTOR EXPERIMENT 9 Core and Pressure Shell 9 Stress analysis of the pressure shell 9 Fluid Circuit . . 10 Control 12 Operation | _ 12 Pretesting 12 Critical loading : 12 Power operation : 13 Shutdown 13 Building ' 13 Electrical power 13 Shielding by concrete pits 13 3. EXPERIMENTAL REACTOR ENGINEERING 15 Seals and Closures | 16 Seal tester 16 Stuffing-box seals 16 Bellows tests . _ . 17 Oval-ring seal | 17 viil 4» viliil Pumps ARE centrifugal pump Laboratory-size maintained-level gas-sealed pump Worthite frozen-sodium-sealed pump Durco convection-cooled frozen-sodium-sealed pump Durco frozen-fluoride-sealed pump Fluoride pumps with stuffing-box seals Valves Wide-open valve Closed valve Throttling valve Canned-rotor-driven valve Heat Exchangers NaK-to-NaK heat exchanger Tube-matrix spacer tests Sodium-~to-air radiator Fluoride-to-ligquid-metal heat transfer Instrumentation Pressure measurement Liquid-level indicators Fluid Dynamics of the ARE Technology of Fluoride Handling Fluoride production Fluoride removal Pretreatment of pipeline helium system Pickling procedures for fluoride containers REACTOR PHYSICS Oscillations in the Circulating-Fuel Aircraft Reactor Calculations for the Circulating-Fuel ARE Methods for Reactor Computation 17 17 17 18 18 18 19 19 20 20 20 20 20 22 22 24 26 26 26 26 28 29 29 29 29 30 31 31 32 33 A e R AR R A A a R a e a BR A AP A 5. CRITICAL EXPERIMENTS Direct-Cycle Reactor Shelf-type reactor assembly Box-type reactor assembly Circulating-Fuel Reactor Reactor materials Beactor assembly Graphlte Reactor Crlrlcal mass Control rod calibration Gap ef fect Flux and power distribution Danger coefficient measurements PART 11 SHIELDING RESEARCH SUMMARY AND INTRODUCTION 6. BULK SHIELDING REACTOR ‘ Mockup of the Divided Shield Air-Scattering Experiments Shield arrangement Radiation dosage at crew compartment Attenuation of gamma rays by lead Comparison of experiment with standard de51gn conditions Determination of the Albedo of Neutrons and Gamma Rays Irradiation of Animals Irradiation of Electronic Equipment 7. GAMMA-RAY ATTENUATION EXPEBIMENTS IN THE LID TANK 63% Iron-37% Water Thermal Sh1eld 34 34 34 36 38 38 38 39 39 39 39 39 40 41 43 43 44 47 47 47 49 49 52 54 54 54 54 1X 63% Iron—-37% Borated Water Thermal Shield Solid Iron Thermal Shield in Water Solid Lead Shadow Shields in Water 8. DUCT TESTS Duct Tests in the Thermal Column Straight ducts Ducts with bends Wall-scattering experiment Comparison with theory GE-ANP Annular Air Ducts 9. NUCLEAR MEASUBREMENTS Cross-Section Measurements with Van de Graaff Accelerator Time-of-Flight Neutron Spectrometer PART III MATERIALS RESEARCH SUMMARY AND INTRODUCTION 10. CHEMISTRY OF HIGH-TEMPERATURE LIQUIDS Low-Melting-Point Fluoride Fuel Systems NaF-KF-ZrF, -UF, KF-ZrF, -UF, NaF-ZrF, -UF, NaF-RbF-ZrF, -UF, Analyses of Fluoride Compounds X-ray examination of solid, complex fluorides Spectrographic analysis Study of solid phases in the NaF-BeF,-UF, system Petrographic examination of fluorides 35 55 55 69 69 69 71 71 72 72 76 76 76 79 81 82 83 83 85 85 85 85 85 86 86 88 Simulated Fuel Mixture for Cold Critical Experiment Preparation of Pure Hydroxides Sodium hydroxide Lithium hydroxide Potassium hydroxide Coolant Development NaF-ZrF, KF-ZrF, RbF-Zrk, Nal-KF-ZrF, NaF-RbF-Zrf, NaF-ZnF, KF-ZnF, RbF-Znt, NaF~-KF-ZnF, Fuel Preparation and Liquid.Handling Zirconium fluoride production Preparation of pure fuel mixtures Liguid handling equipment Solubility of Uranium in Sodium Cyanide 11. CORROSION RESEARCH Static Corrosion by Fluorides Effect of additives Temperature dependence Effect of plating metals Effect of cold-work Corrosion of ceramic materials Static Corrosion by Hydroxides Effect of additives Effect of plating metals 89 89 90 90 90 90 91 91 99 92 93 94 94 96 96 96 97 97 99 99 100 101 101 103 103 104 104 105 105 105 X1 ’llllll'llll‘lr"l.'l.l.l'iil'lll'llllfi Dynamic Corrosion by Liquid Metals 107 Dynamic Corrosion by Fluorides 107 Corrosion by fluorides in seesaw tests 107 Corrosion by fluorides in thermal convection loops 109 Corrosion in rotating dynamic test rig 116 Dynamic Corrosion by Hydroxides 116 Corrosion by hydroxides in seesaw tests 116 Standpipe tests of hydroxide corrosion 118 Hydroxide corrosion in cold-finger thermal convection apparatus 118 Modified thermal convection apparatus ~ TCA 118 Fundamental Corrosion Research 119 Interaction of fluorides and structural metals 119 Synthesis of complex fluorides 120 Examination of corrosion products 121 X-ray~-diffraction studies 123 EMF measurements in fused fluorides 124 Reactions in fused sodium hydroxide 124 Compounds resulting from hydroxide corrosion 125 Mechanism of fluoride corrosion 126 Free-energy relations of fluorides and structural metals 128 Solution of metals in molten halides 128 Fluoride corrosion phenomena 129 12. METALLURGY AND CERAMICS 131 Loose Powder Bonding 131 Sintering temperature 132 Sintering time 132 Fuel-componcnt particle size 132 Surface preparation 135 Sintering under load 135 Control Rod Fabrication 135 Safety rod slugs 135 Regulating rod slugs 135 xii SN PR, Mechanical Testing of Materials 136 Stress-rupture tests in argon 136 Stress~corrosion tests : 136 Cone-Arc Welding | | 138 "Effect of welding conditions : 138 Heat exchanger assembly ' 139 Tests of Brazing Alloys _ 140 Nicrobraz ‘ _ 140 Manganese-nickel 140 ‘Palladium-nickel ' ' 143 Ni-Cr-Si-Mn alloys 143 Silver-base alloys - 143 Ceramics 144 Ceramic coatings for metals ' 144 Ceramics laboratory program ' 146 13. HEAT TRANSFER AND PHYSICAL PROPERTIES RESEARCH 146 Viscosity of Fluoride Mixtures 146 Ber-bearing'fluorides 147 ZrF,-bearing fluorides - 147 Material from corrosion loops : 148 NaF-KF-LiF and NaF-KF-LiF-UF, mixtures : 148 Thermal Conductivity - | 149 Heat Capacity _ 149 Density : 150 Vapor Pressure of Fluorides . 150 Beryllium fluoride | 150 ZrF,-bearing fluoride mixtures 150 Physical Property Data | | 152 Heat Transfer in Sodium Hydroxide 152 X111 88O T R T 0,08 b A R R S e n s A s s s ) et Boiling Heat Transfer Entrance Region Heat Transfer Natural Convection in Confined Spaces with Internal Heat Generation Circulating-Fuel Heat Transfer Studies 14. RADIATION DAMAGE Irradiation of Fused Materials Pile irradiation of fuel Cyclotron irradiation of fuel Inpile Circulating Loops Creep Under Irradiation Radiation Effects on Thermal Conductivity PART IV APPENDIXES SUMMARY AND INTRODUCTION 15. ANALYTICAL CHEMISTRY Analyses for Components of Fluoride Mixtures Total alkalil metals Sodium, potassium, and lithium Beryllium Zirconium Analyses for Impurities in Fluoride Mixtures Iron Chromium Analyses for Impurities in Fluoride Compounds Sulfate and sulfide Zirconium oxide and oxyfluoride Water 156 156 158 159 159 160 160 163 163 164 164 167 169 170 170 170 170 170 170 171 171 171 171 171 172 172 Service Analysis 16. LIST OF REPORTS ISSUED 17. DIRECTORY OF ACTIVE ANP RESEARCH PROJECTS Reactor and Component Design Shielding Research Materials Research Technical Administration of Aircraft Nuclear Propulsion Project at Qak Ridge National Laboratory 172 173 176 176 178 179 184 ANP PROJECT QUARTERLY PROGRESS REPORT FOREWORD The Aircraft Nuclear Propulsion Project at the Qak Ridge National Laboratory i1s comprised of some 300 technical and scientific personnel engaged in many phases of research directed toward the nuclear propulsion of aircraft. A considerable portion of this research is carried out to provide support for other organizations participating in the national ANF effort. However, the purpose of the bulk of the ANP research at OBNL is the develdpment of a circu- lating-fuel type of reactor. The nucleus of this effort 15 now centered on the Aircraft Reactor Experiment mia 3-Mev, high-temperature prototype of a circulating-fuel reactor suitable for the propulsion of aircraft. 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 of the Laboratory under its contract, W-7405-eng-26. The report is divided into four parts; I. Reactor Theory and Design; II. Shielding Research; I11. Materials Research; and IV. Appendixes. " Each part has a separate "Summary and Introduction. SUMMARY AND Analysis of the circulating-fuel atrcraft reactor systems incorporating intermediate heat exchangers has led to the use of a spherical-shell type of heat exchanger and shield ar- rangement. This arrangement has resulted in an engineered reactor- shield assembly weighing about 8000 1b less than the best arrangement previ- ously examined. A most promisimg design study has been undertaken of. a circulating-fluoride-fuel ceactor core in which a thick reflector replaces the moderator. Such a design, which results in a homogeneous- fluoride reactor, has been shown teo have a low critical mass and a neutron lifetime more than adequate to permit control. This compact, spherical- core design is compatible with the low-weight spherical-shell type of heat exchanger and shield arrangements (sec. 1). The Aircraft Reactor Experiment to be constructed by the Oak Ridge National Laboratory was described in the last report. Only minor modifi- cations in the design have occurred during the past guarter; the most significant changes were the reduction in number of fuel passes through the core and the reversal of the inlet and outlet fuel headers. The design effort is concentrated on detailing drawings of components of the reactor and the procurement of these parts. The building for the ARE was released on June 6 to ORNL by the construction contractor, and installation of equipment 1is proceeding as rapidly as the receipt of parts permits (sec. 2). ' The emphasis of the research and development program for reactor plumbing and assocciated hardware continues to be placed on the tech- nology of high-temperature fluoride mixtures (sec. 3). Both pumps and valves have been operated successfully INTRODUCTION with molten fluoride mixtures for extended periods of time at 1500°F. In both instances the problem reduces to that of sealing a moving shafg; this has been satisfactorily effected by both frozen seals and stuffing-box seals. NaK-to-NaK and Na-to-a1ir heat exchangers have beenoperated — the former for over 3000 hours. - In addition considerable effort has been devoted to developing instruments for measuring flow and pressure and for liquid level indication and control as required both by the ABE and the experimental high-temperature-liguid systems. Satisfactory technigues have been developed for the manu- facture, transfer, and loading of fluoride mixtures without introducing contamination. . Studies of the kinetics of the circulating-fuel aircraft reactor have led to the conclusion that an spite of the loss of some delayed neutrons power oscillations i1n such reactors will be strongly damped by a mechanism associated with the actual circulation of the fuel (sec. 4). Physics research for the aircraft reactor experiment has involved numerous static calculations conse- guent to significant, but more or less minor, design changes. A critical assemblyof the box-type, direct-cycle reactor was investigated for the GE-ANP program. Fission rate, flux measurements, reflector effects, and temperature response were de- termined. Upon completion of these me asurements, a mockup of the circu- lating-fuel aircraft reactor will be assembled - the components are now on hand., ' The beryllium oxide machined for the ARE will be used 1in the critical assembly, tegether with a simulated fluoride mixture. A resume of the results from an earlier series of experiments with a graphite-~ moderated reactor is also presented (sec. 5). ANP PROJECT QUARTERLY PROGRESS REPORT 1. CIRCULATING-FUEL AIRCRAFT REACTORS A. P. Fraas, Several circulating-fuel aircraft reactor systems employing closely coupled i1ntermediate heat exchangers were described in the last report.¢1!) The extension of these design studies led to consideration of a spherical- shell type of heat exchanger and shield arrangement that resulted in a saving of about 8000 lb in engineered shield weight over the best arrangements previously examined.¢?? 1In addition, a modified core design having all of the moderator removed from the core and added to the reflector now appears to be an attractive possibility. With the modified design, the critical mass would not be large, the neutron lifetime would be longer than with previous designs {(which would ease the control problem), and the structure in the core would be eliminated. REFLECTOR-MODERATED CIRCULATING-FUEL REACTOR Calculations made on the basis of relations derived for thermal reactors had i1ndicated that circulating-fuel reactors employing a mixture of fluoride salts as a homogeneous working fluid would reguire a very high critical mass and would be difficult to control. Multigroup two-region IBM machine calculations made during the past quarter have shown that this 1s not the case. Removal of the moderator from the core of a circulating-fuel reactor does not result 1n excessive fuel 1f a thick reflector In a large measure this comes about because the removal of the moderator means the removal of the poisoning effect of a requirements 1s employed. also (1) Atrcreft Nuclear Propulsion Project Quarterly Progress Report for Period Ending March 10, 1952, ORNL-1227, p. 7. Dieia., . 12. ANF Division substantial amount of stainless steel. Critical mass calculations have been made on the IBM machines for two core diameters. With a 12-1a.-~thick beryllium oxide reflector, critical masses of 23 and 19 lb were obtained for 16- and 32-in.-dia spherical reactor cores. The structural sim- plicity and attendant advantages of this type of reactor are obvious. At first 1t also seemed that these unmoderated circulating-fuel reactors would have a high median energy for fission and hence might present a serious control problem. However, about one-~third of the fissions are caused by thermal or near thermal that have spent a large fraction of their lifetimes 1in the relatively poison-{free reflector. Thus the effective neutron lifetime from the control standpoint may actunally be greater than in beryllium oxide~moderated AMOUNT OF LEAD IN - CREW COMPARTMENT ~— - AS SPECIFIED BY ANP-53 Qc 0.0¢ N\ N 534cm \ N 0.0022—-—+\ ™~ AMOUNT OF LEAD INDICATED AS NECESSARY BY THESE EXPERIMENTS 0.001 0 1 0 3 4 5 6 THICKNESS OF LEAD (cm) Fig. 20, Air-Scattering Experiment Gamma-Ray Attenuvation by Lead at Crew Com-~ partment, 50 If ¢ = neutron dose, mrep/hr/w, as ., measured 1n the experiment, " = gamma dose, ¥/hr/w, as measured in the experiment, D, = airplane neutron dese, mrep/hr at 2 X 108 w, Dy = airplane gamma-ray dose, r/hr at 2 X 10% w, then p, - gpmie|le, O " Ay L, ®d, where Aa = ratio of effective source Ay areas exposed (explained below) = 16, La 2 = ratio of leakages per watt = 0.2, “b = correction factor for the leakage from a cylindrical shield surface instead of a plane surface = 1.8, P, = aircraftreactor power = 2% 108w, db 7 = correction, based on single a scattering i1n air, for the fact that the separation distance in the experiment ts 5 meters instead of 15 meters as in the airplane. A 1s the area of the cylindrical a side surface of the airplane reactor, and 4, 1is the area of the top of the Bulk Shielding Beactor. In the ex- periment the only radiation escaping upward through the surface of the FOR PERIOD ENDING JUNE 10, 1952 water 1s assumed to have originated on the top areca of the reactor. This is to be compared with the radiation emerging from the cylindrical side sur face of the airplane reactor (not including the front and rear surfaces). As discussed in more detail else- where, (%) the ratio of found to be 16. areas was In the unit shield report¢’? the ratio of the leakages per watt for the airplane reactor and the Bulk Shielding Reactor was found to be about 0.2. r_ 1s the aircraft reactor radius and r_ i1is the aircraft reactor shield radius. The factor vr /r_ is a correction to take 1nto account the fact that the airplane shield 1is cylindrical, whereas the Bulk Shielding Reactor shield is flat.(%> Applying all the corrections, db 8 e T 3.8 x 10° . da This should be an upper limit. The best estimates of the amount of scattering from the walls of the pool, reactor bridge structure, etc., in- dicate that wall-scattered radiation was 2 or 3 times the ailr-scattered radiation for the experiment for both neutrons and gamma rays. A lower limit on the scale-up factor is then 1.3 % 10® with a most probable value of around 2.1 % 10%, By using the latter value for the calculations: D, = (2.1 x 10%) ¢, and since ¢ was measured to be 8§ X 10°8 mrep/ hr/w, D = 16.8 mrep/hr. n (7)J. L. Meem and H. E. Hungerfard, The Unit Shield Experiments at the Balk Shielding Faecility, ORNL-1147 (April 390, 1952), ol ANP PROJECT QUARTERLY PROGRESS REPORT This compares reasonably well with the specified dose of 21.5 mrep/hr. In converting the experimental gamma-ray dose, [, to the airplane dose, Dy, an additional factor, f(Pb), must be introduced for the attenuation of the lead inside the crew shield: D -—rA“L“J—r—iP if(Pb) 7 Ab Lb " a ¢ a , D, =T (2.1 % 108) f(Pb). The experimental gamma-ray dose, I", was 5.4 X 10°7 r/hr/w through 17.8 cm of water, and 1f the specified value of 0.25 r/hr 1s used for Dy, the factor by which the lead should attenuate the gamma rays can be found; that 1s, D, 0.25 [ (2.1 % 10%) 5.4%x10°7 x 2,1%x108 1 f(Pb) 2.2 x 1073, i Inspection of Fig. 19 shows that 5.34 cm of lead 1s required to at- tenuate the gamma rays by this amount. The standard design specified 1.54 cm of lead, which would give an tenuation of 5.9 X 10°2%. The gamma dose 1n the standard design would therefore be too large by a factor of about 27. The additional lead required 1f 1t were placed inside the crew compartment is 3.8 em. This much lead would weigh approximately 25,000 pounds. Perhaps considerable savings 1n welght could be made by placing the lead 1in the reactor shield. The experiment gave no in- formation concerning such a dis- position of the lead. at - The approximations made in con- verting the experimental dosage rate to that expected 1n the airplane 52 configuration are admittedly crude. Furthermore, the effect of the walls of the pool and building in scattering the radiation back to the point of detection has not been determined. However, the disconcerting fact re- mains that the neutron data 1s 1n rough agreement with the calculations of the Shielding Board, whereas the gamma-ray dosage appears to be high by a factor, which, while less than 50, seems to be at least as highas 15. Further experiments, at 100 kw reactor power, are contemplated to verify or correct this high factor. DETERMINATION OF THE ALBEDO OF NEUTRONS AND GAMMA RAYS To aid in the calculations of backgrounds at the proposed Tower Shielding Facility several short experiments¢®’ were performed to determine the reflection of neutrons and gamma rays. DBy using a neutron source and the fast-neutron dosimeter shown i1n Fig. 21, measurements were made with the source and counter at varying heights above the concrete floor in the Bulk Shielding Reactor. By using the relation(®’ alN, F =, 107d? where F is the scattered flux, N; is the source strength, and d 1s the distance above the floor, the albedo, a, for neutrons was found to be 0.12 for concrete. A similar experiment in which an ion chamber and a Co®° source with lead shielding was used gave the reflection coefficient of gamma rays on concrete as 0.04. (S)H. E. Hungerford to J. L. Meem, Some Ground Scattering Experiments Performed at the Bulk Shielding Facility, ORNL CF-52.4-99 (April 16, 1952), g}A. Simon to E. P, Blizard, Estimate of Background at Tower Shielding Facility, ORNL CF- 51-12-185 (Dec. 17, 1951). 1 "fif' Ceen LHTA T ewe weeTia g fasaty, . * setr oy e a e . ’ ‘ol et Q. ¢ « . e . - FOR PERIOD ENDING JUNE 10, 1952 DWG. 14697 TO0 CRANE PARAFFIN BLOCK FAST-NEUTRON DOSIMETER Po-Be NEUTRON SOURCE “ b % M Eoaey e . . T eat i d et e st p e Sl e e LR e s A s a. :'.;" e P T T . : Lt Tttt e R sl e CONCRETE FLOOR . <" " o 1. .4 ., . .'. . . . . . . Lo . -. st . . . . . . . " T . - et "- . 0 = A i -~ ., - a . . . . ‘“ * . fi'. . . . : -~ s . - . - ¢ . . .« . . Fig. 21. Arrangement of Apparatus for Albedo Experiment. 53 ANP PROJECT QUARTERLY PROGRESS REPORT IRRADIATION OF ANIMALS In the 1fiterest of determining the threshold for eye cataract formation, time on the Cockcroft-Walton ac- celerator has been furnished to members of the Biology Division for exposures of rats, mice, and rabbits to source intensities of 10% neutrons/sec with 14-Mev energy. Approximately 60 hr of irradiation time on the machine has been used for these experiments during this quarter. HResults will be published by members of the Biology Division. This irradiation program will continue. IRRADIATION OF ELECTRONIC EQUIPMENT In cooperation with the Solid State Division and the Wraight Air Development Center, the Bulk Shielding Reactor is being used on week ends to irradiate electronic equipment. Reactor operators furnished by the Air Force have been trained by ORNL personnel. ment, posed for 100 hr to a fast-neutron flux of between 107 and 10® neutrons/sec with no appreciable effect. Detailed results will be reported by the Solid State Division. The first piece of equip- a radio compass, has been ex- 7. GAMMA-RAY ATTENUATION EXPERIMENTS IN THE LID TANK C. E. Clifford T. V. Blosser J. D. Flynn M. K. Hullings L. S. Abbott M. C. Marney Physics Division A series of short experiments on the gamma-ray attenuation of heavy materials (iron and lead) in regions close to the source have been completed in the Lid Tank. The experiments planned to aid in the design of both thermal and shadow shields for various aircraft reactors were necessary because of the difficulty of calculating the production of neutron-induced secondary gamma rays 1n these regions. 63% IRON-37% WATER THERMAL SHIELD Specific shield configurations requested by Goldstein of NDA for possible use in the supercritical-water reactor are shown as part of Figs. 22 through 26, The i1ron-water mixtures were placed at various distances from the source to simulate the varying water-reflector thicknesses present 1in the reactor design; although the reactor is cylindrical, 1t 1s sur- rounded by a spherical pressure shell and thermal shield. 24 The shield configurations were mocked up with one to six 2.22-cm iron slabs separated by 1.3 cm of water, and they were placed at various distances from the source. In each series the sixth slab was moved forward to touch the preceding slab to simulate a pressure shell, as shown in Fig. 26. The measurements were made with both an air-filled 1onization chamber and an anthracene scintillation counter. Figure 27 indicates the gamma-ray dosage distribution in water, and Figs. 22 through 26 give the gamma-ray measurements behind the i1ron-water thermal shields. A curve of the dose at a constant z (distance from the source) as a function of the water thickness between the source and the nearest iron slab (referred to as reflector thickness) is also shown. The tabulated "z measurements to back of last slab" (inset in each figure) are probe measurements of the distance from the source to the back of the last slab in each array. As would be expected, the data indicate that for thick thermal shields the secondary gamma-ray production is predominant up to rather large reflector thickness (i.e., 20 cm). 63% IRON-37% BORATED WATER THERMAL SHIELD A stfi%%fibf the effect of borating the water in the region of the thermal shield was made to determine the resulting reduction of secondary gamma- ray production (Figs. 28, 29, 30). The configurations were chosen to match the neutron and gamma-ray leakage in the Lid Tank with that of the reactor on the basis of a calculation by Goldstein. A scale-up factor was also calculated and used to determine the required Lid Tank dosage beyond the thermal shield. The 2.22-cm iron slabs were separated by the borated water, and since sufficient attenuation was obtained with a four-slab (27% borated water) thermal shield, attenu- ation of the pressure shell (two adjacent slabs) was then measured (Fig. 30). The measurements were made with the 10'% jonization chamber (air-filled). A comparison of attenuation of the borated-water thermal shield with that of the unborated-water shields previ- ously measured indicates that at this reflector thickness (20 cm) a reduction in intensity to 70% of the iron-water case (at z = 80 cm) was obtained. The reduction of intensity as the water reflector thickness was increased to 30 c¢cm indicates that secondary gamma- ray production 1s not entirely sup- pressed by the 1% boration of the thermal shield water. This 1s not surprising 1in view of the large volume percentage of i1ron in this region. There may also be some 1nelastic scattering of gamma rays, which would of course not be affected by the boron. FOR PERIOD ENDING JUNE 10, 1952 SOLID TIRON THERMAL SHIELD IN WATER As a continuation of the thermal shield investigation, the gamma-ray attenuation of a 9-cm iron thermal shield was measured for the GE-ANP design group. Four 2.2-cm iron slabs were clamped together and placed 1in the Lid Tank at various distances from the source, as indicated in Fig. 31, Since the iron slabs are not flat, water (approximately 0.3 cm) was present in the space between adjacent slabs., The measurements presented 1n Fig. 31 were obtained with the usual air-filled ionization chamber, which is, unfortunately, neutron-sensitive owing to the nitrogen (n,p) reaction, This invalidates the measurements close to the i1ron, but since the neutron relaxation length in water behind solid iron is short (4 to 6 cm), the neutron effect should be negligible beyond about 30 c¢cm of water. Measurements that indicate the magnitude of the neutron effect and confirm the fact that 1t 1s negligible beyond 30 cm have been made on similar mockups by using CO, in the i1onization chamber. SOLID LEAD SHADOW SHIELDS IN WATER Measurements were taken in the Lid Tank at the request of GE-ANP to determine the effect of various water- reflector thicknesses on the gamma attenuation of solid lead shadow shields, Since the experiment was performed in water, a large amount of secondary gamma-ray production reduced the effectiveness of the lead by a factor greater than 40 in the worst case. As the thickness of the water layer between the source and the lead was increased, the secondary gamma- ray production was, of course, reduced at a rate equivalent to the neutron attenuation of the water. 55 ANP PROJECT QUARTERLY PROGRESS REPORT The gamma-ray measurements for the shadow shields consisting of one, two, and three adjacent 3.8-cm lead slabs are given in Figs, 32, 33, and 34. The "z measurements to back of last sl ab"™ are probe measurements and include water gaps caused by warpage of the slabs. 1In each case the water- reflector thickness was held constant by wooden spacers 1inserted between the source and the first slab, The gamma measurements were made with the 10'° ionization chamber. During the course of the experiment CO, was substituted for air as the ion-chamber gas to eliminate the nitro- gen (n,p) reaction, since measurements were required in a high neutron flux. DWG. 14716R1 10 0.9-cm WATER REFLECTOR 5 GAMMA DQSE VS._. .. , REFLECTOR THICKNESS AT z=39 cm {mr/hr) N GAMMA DOSE o WATER Z MEASUREMENT REFLECTOR TO BACK OF LAST (cm) SLAB {cm) 2 fo . 3 4.7 7.2 ONE 2.22-cm IRON SLAB 2 4L S i b°°°° - SOURCE |-~ O 20 40 60 Fig. 22. Thermal Shield 56 WATER REFLECTOR (09 to 5.0 cm) 80 100 Z, DISTANCE FROM SOURCE (cm) Measurements Iron Slab with Various Water-Reflector Distances. 120 140 - Gamma-Ray Dose Beyond One 2. 22-cm FOR PERIOD ENDING JUNE 10, 1952 DWG. 1464 5R1 0.9-cm WATER REFLECTOR WATER Z MEASUREMENT REFL.LECTOR TO BACK OF (cm) LAST SLAB (cm) 6.6 @O NO NOOHNOW ! { 2 = > E L wn O O <1 = = IRON TANK, 0.32-cm WALLS | | = ~ ™~ v | { L5 SIX TO EIGHT 2.22-cm IRON SLABS SEPARATED BY— oo " 1.3 ¢cm OF BORATED WATER (1% BORON SOLUTION) e E e~ N - 5 z oo T L = DY b 2 _ 2 I soet 3 | SOURCE = - - ot oo |7 T BORATED WATER a0 o T Q ° o Q| - - —~10~! O 20 40 60 80 100 120 140 Z2, DISTANCE FROM SQURCE (cm) Fig. 28. Gamma-Ray Dose Beyond Iron-~Borated Water Thermal Shield with 20-cm Water Reflector. FOR PERIOD ENDING JUNE 16, 1952 DWG. 14905R1 FOUR 2.22-cm IRON SLABS SEPARATED BY 1.3 cm OF BORATED WATER 59 3 a3 & Ny ('_i) o M M S et == | T LT z . \_fi il — g:%fi% I NN ONAN —=f-BORATED WATER = |- SOURCEX \A’ WATER IN ALUMINUM TANKS “-IRON TANK, Q.3-cm WALLS — FOUR 2.22-cm IRON SLABS SEPARATED BY 1.3 cm OF BORATED WATER- TWO 2.22~-cm IRON SLABS (ADJACENT) §o S { o o Ny O aJ nl”) wn T UNONNE VN ol NUNA. . | \ /TONON o\ A - — IANANZANM—, — SOUF{CEl WATER IN ALUMINUM ~BORATED WATER TANKS IN IRON TANK “~IRON TANK,C.3-cm WALLS FOUR 2.22-¢m IRON SLLABS SEPARATED BY 1.3 cm OF BORATED WATER TWO 2.22-cm IRON SLABS — ( ADJACENT) 5o 0 3 5 5 N O -— 0 'R‘i‘ w ¢ o == o~ */g"-- —.#IT.—_ & "?/I;'""Q_,f:’, ¢ :_“‘“—1’— —_ — E T =oNONNE= = ol é§é§é i Sl 7z 1 %’“ w;\ 7/7 ? Z%fi% A SOURCE‘X ZWATEF.’ \Z—WATER’ IN ALUMINUM ZBORfl\TED WATER TANKS IN IRON TANK IRON TANK, 0.3-cm WALLS Fig. 29. Schematic Diagrams for Gamma-Ray Measurements Beyond Iron-Borated Water Thermal Shield. See Fig. 30 for plotted data. 63 ANP PROJECT QUARTERLY PROGRESS REPORT 4 Fe, 20.0-cm WATER REFLECTOR c o £ 1 w < Fe, 20.0-cm WATER REFLECTOR; < PRESSURE SHELL = = < o 6 Fe, 30.5-cm WATER REFECTOR, PRESSURE 0 20 40 60 80 100 z, DISTANCE FROM SOURCE {cm) Fig. 30. Gamma-Ray Dose Beyond Iron-Borated Water Thermal Solution). See Fig. 29 for schematic diagrams. 64 DWG. 14906R1 120 140 Shield (1% Boron o U {mr/hr) GAMMA DOSE N 10 10 FOR PERIOD ENDING JUNE 10, 1952 4 0-cm WATER REFLECTCR REFLECTOR (cm) DWG. 14694R1 Z MEASUREMENT TO BACK OF LAST SLLAB (cm) GAMMA DOSE VS. REFLECTOR THICKNESS AT 2=79.3 ¢m WATER REFLECTOR (4.0 TO 24.0 cm) FOUR 2.22-cm IRON SLABS (~ 0.3 cm OF WATER BETWEEN SLABS) 20 40 60 80 100 z, DISTANCE FROM SQURCE (cm) 13.7 18.¢ 23.7 33.8 120 140 31. Gamma-Ray Dose Beyond Solid TIron Thermal Shield. 65 ANP PROJECT QUARTERLY PROGRESS REPORT DWG. 14914R! - 10° N T T T e 5 ] 2 | - a ] 10 ] 5 [710.0-cm, AR Y7 e ] 18.0-cm, CO—=> NN N | . i 15.0-cm, AIR ——XQ\ 2 ~ 3 . .29:07em, Co——= "N\ N\ %rfem AR 10 RN NN N T T T = T . N _ AN g 5\\ """"""" . 34.8-cm, CO, \ O \ < 2 ‘H% e e z GAMMA DOSE_VS. REFLECTO I THICKNESS AT 7z =79.3 ¢m © 102 TR J\ . !_ | WATER z MEASUREMENT 7| REFLECTOR TO BACK OF 5 ——| {cm) LAST SLAB eeerenrensd {cm) 0.7 4.5 ) 5.0 8.8 - 10.0 13.8 15.0 8.8 25.0 28.8 10 34.8 386 —- ] T T ] T T o . | jWATER REFLECTOR (07 TO 35 cm)——— | / (ONE 3.8-cm LEAD SLAB | S . ] gfiat' :2 : ;7_77_.3 """""""""" 1 —J SOURCE |~~~ . © 060 ‘:O e — | 0 20 40 60 8O 100 120 140 Z, DISTANCE FROM SOURCE (cm) Fig. 32. Thermal Shield Measurements -~ Gamma-Bay Dose Beyond One 3.8-cm Lead Slab with Various Water-Reflector Thicknesses. 66 b'& {0 9 [5.0-cm, AIR FOR PERIOD ENDING JUNE 10, 1952 CWG. 14915R1 WATER Z MEASUREMENT REFLECTOR TO BACK OF {cm) LAST SLAR (cm) % 8.7 . 13.1 10, 18.1 15. 23.4 25.0 33.0 348 42.7 % ACCOUNTS FOR WATER GAPS 10.0~cm, AIR 0.7-cm WATER REFLECTOR, AIR-FILLED ION CHAMBER 15.0-cm, AIR WATER 15.0-cm 3 9 10 £ S~ £ 5 3)) 25.0-cm, CO, o 0 b 2 = 3 & 5 34.8-cm, CO 10 5 GAMMA DOSE VS, REFLECTOR THICKNESS AT 2=79.3 cm 2 10 5 WATER REFLECTOR (0.7 TO 34.8 cm ) SOURCE TWO 3.8-cm LEAD SLABS <@ 0 20 40 60 80 100 120 140 z, DISTANCE FROM SOURCE {cm) Fig., 33. Thermal Shield Measurements - Gamma-Ray Dose Beyound Two 3.8-cm Lead Slabs with Various Water-Reflector Thicknesses. 67 ANP PROJECT QUARTERLY PROGRESS REPORT 68 10 {mr/hr) GAMMA DOSE 10 t=——— THICKNESS AT 2= ?93cm ..... I ,,,,,, l e —TWATER REFLECTOR (O? TO 348cm) - > [ 7 THREE 38 cm LEAD SLABS _ Y 7 g ¥ g o — ——=— - oq°02 T T iAled SOQURCE| qao oo R R - 0 20 40 60 80 100 120 z, DISTANCE FROM SOURCE {(cm) Fig. 34. Thermal Shield Measurements - Gamma-Ray Dose Beyond T 411 Ay DWG. 149t6R1 X — O I s N—f—— S R — WATER 7 MEASUREMENT [ X ~—— REFLECTOR TO BACK OF \ ] (cm) LAST SLAB N S e (cm) * \ 10.0-cm WATER REFLECTOR, Y \« AIR-FILLED ION CHAMBER 0.7 131 . 5.0 17.6 A | 10.0 22.8 0.7-cm, AIR 15.0 28.4 25.0 38.2 oo X\ »\\ 1 .'50 C_T_. AR 34.8 47.3 X\ [ —F % ACCOUNTS FOR WATER GAPS Q‘?‘\ _______ - - —] GAMMA DOSE VS. REFLECTOR 140 Three 3.8-cm Lead Slabs with Various Water-Reflector Thicknesses. FOR PERIOD ENDING JUNE 10, 1952 8. DUCT TESTS F. J. A. Simon Muckenthaler C. E. Clifford M. K, Hullings Physics Division Duct work in the Thermal Column¢1) has continued to obtain further experi- mental corroboration of the simplified duct theory(?) of neutron transmission through cylindrical air-filled ducts in water. Measurements have been made on ducts with two and three bends to supplement the data previously reported on straight ducts and ducts with one bend. Some effort has been made to determine what portion of the flux measured by the counter 1s due to neutrons suffering at least one col- lision in the walls of a straight duct; the measured contribution appears to be somewhat higher than was predicted theoretically, Agreement with the theory has remained good; the value of the constant A, con- cerns the albedo at the corners and other factors assumed to be constant, had to be redetermined. ' however, which An extensive survey of the radiation transmitted by the annular air ducts designed by G.E. has been started. The 1id Tank will be used for this survey for approximately six weeks. DUCT TESTS IN THE THERMAL COLUMN Straight Ducts. Various lengths of a straight, 4 1/4-in.-d1a aluminum-wall (1/8 in. thick) duct were measured. A circular fission source (4 3/8 in. in diameter) was used for all the 4 1/4- in,-dia duct measurements. Typical curves for the neutron transmission measured along an extension of the duct center lines are givenin Fig. 35. (1) Progress Report for Period Ending March 10, ORNL- 1227, p. 79. ¢. E. Clifford and A. Simon, Simplified Theory of Neutron Traensmission Through Air- Filled Cylindrical Ducts in Water {in preparation). Air‘craft Nuclear Propulsion Quarterly 1952, o’ DUGT: 4 Yg—~in., , ALUMINUM WaLls {BENY DUCTS 5 HEVE TWD 7O FOUR £0~-in. STRAIGHT SECTVONSD) ¥ 2B-in. STRAIGHT SECTION @ 4g-in. STRAIGHT SECTION : + Bp-in STRAGHT SECTION 2 & 3-dog BEND ' 7 43-dey BEMD 08 0 66-deg BEND ¢ 9u-deq BEND S TWO 45-deq BENDS 5 ferfoens ¢ THREE as-deq BENDS ~ SOURCE: CIRCULAR, 4%~ in. DIA w04 o [ RELATIVE THERMAL NEUTRON FLUX N WATER{ 125‘2-in. BF COUNTER RESPONSE, counts /min) o2 0 © 20 30 z, COUNTER DISTANCE FROM END OF DUCT {cm) Fig. 35. Neutron Traosmission Through Water in Cylindrical Ducts with variable Bends. 69 ANP PROJECT QUARTERLY PROGRESS REPORT 7O i005C UNGCLASSIFIED PHO 70 Y ~ E | i 90w ~ 4" o ,wfiifiAOO SOURCE s —— om:||||||L and 45-deg Bends Straight Sectionmns, -in, Configurations of 4 1/4-in.-ID Air Ducts, 20 36. {Aluminum Walls). Fig. - Ducts with Bends. Neutron measure- ments were made on the 4 1/4-in.-~dia ducts with one to three bends formed by 20-1in. straight sections joined at 45 deg angles. Further experimental data for the angular correlation were obtained by using three ducts with single sharp bends of various angles (31, 65, and 90 deg). Typical duct configurations are shown in Fig. 36. To prevent neutrons from traveling the full length of the ducts in air without scattering at the bend, the smallest angle of bend afforded more than a one~-diameter displacement of the center of the duct., Measurements of the neutron transmission along the duct center lines are given in Fig. 37. Wall-Scattering Experiment. The duct transmission theory that has been used ignores all wall-scattering except at the bends, so the attenuation of each straight section 1s proportional to the square of its length. If single scattering in the walls of the 104 DWG. 15340 D N X W DUCT: B-in. I0 BY 36 in, PLASTIC WALL COUNTER: 8-in. BFy CHAMBER AT UPPER END OF DUGCT ¢ COUNTER RESPONSE {counis/mind A 103 ‘ 20 an 60 80 100 HEIGHT OF WATER ABOVE SQURCE PLATE (cm) Fig. 37. Measurement of Thermal- Neutron Flux as a Function of the Level of the Water Surrounding the Duct. ’ FOR PERIOD ENDING JUNE 10, 1952 straight sections is considered, a component attenuated: according to the cube of the length should be added. Therefore, an axperiment was carried out to investigate the magnitude of the inverse-cube component. A 6~in.-dia by 36-1n. duct (plastic wall) and a fission source {6 1in. in diameter) were mounted in the Thermal Column water tank so that the level of the water surrounding the vertical duct could be varied. A counter was mounted 10 cm from the upper end of the duct and surrounded by a small volume of water, as shown in Fig. 38. The neutron flux was then measured as a function of the depth of water around the duct, ' A second set of measurements was made with the upper two-thirds of the duct filled with water. This prevented FHOTS 10048 § Fig. 38. Duct Configuration for. Wall-Scattering Experiment. ANP PROJECT QUARTERLY PROGRESS REPORT the neutrons that would ordinarily travel the length of the duct 1n aar from reaching the counter. The daf- ference of the two curves should give the flux contributed by scattering in the walls. According to calculations®?) the ratio of the flux contributed to the counter by the neutrons that make one collision in the walls to those which travel the full length of the duct directly 1in air 1s approximately equal to 7r /2L, where r 1s the radius of the duct and L 1s the length. For a 6-in.- dia by 36-in. duct the ratio should be approximately 0.15, which 1s about a factor of 3 lower than the ratio found in the experiment. The curve in Fig. 37 shows that the "contribution 1s not linear but peaks at the end of the duct farthest from the source. Comparison with Theory. The results of the measurements with both the straight and bent aluminum-wall ducts indicate good agreement with the theory (Fig. 39). 1If agreement were perfect, all the curves 1n Fig. 39 would coincide. The spread indicates the errors in predictions of attenuations, which in some cases are as much as 10. A new value for A, the constant independent of duct geometry, has been obtained that gives a better prediction of the transmitted dose. The redetermination of A was necessary hecause of the two- and three-bend experimental results. of the wall-scattered components are higher than was predicted by the theory. The discrepancy is for the most part the result of degradation in energy of the Preliminary measurements Since thermal measurements the degraded neutrons were neutrons. were made, 72 counted efficiently. The position of the counter was approximately 10 cm from the end of the duct to corre- spond to the 10-cm point used in the. previous duct measurements., Calcu- lations of the scattered components for these points werealso high compared with the theory. I1f, however, the 20-cm points of the previous duct measurements are compared with the theory, the agreement 1s markedly improved, which is in accordance with the above reasoning. Except for the three points nearest the source, the shape of the curve shown in Fig. 37 1is about what would be expected. Failure of the initial points to agree with the theory is not understood, and plans are being made to repeat this experiment under different conditions, GE-ANP ANNULAR AIR DUCTS Mockups of the inlet and outlet air ducts that penetrate the shield in the G-E"direct-cycle design have been supplied by G.E. fora lLid Tank experi- ment. Measurements will be made to provide complete neutron and gamma-ray isodose plots in the water shield sur- rounding the duct. Such plots will indicate points of excessive leakage that may require additional shielding. A schematic diagram of the outlet air duct is shown in Fig. 40. The steel section between the source and the duct consists of alternate 1 1/2- in. air and water layers perpendicular to the source and simulates the transition region (perforated re- flector) in the reactor design. A photograph of the inlet air duct and transition region is shown in Fig, 41. FOR PERIOD ENDING JUNE 10, 1952 103 OWG, 15344 DUCT: 4% —in. 1D, ALUMINUM WALLS (BENT DUCTS HAVE TWO TO FOUR 20—in. STRAIGHT SECTIONS) 28-in. STRAIGHT SECTION 40-in. STRAIGHT SECTION 52—in. STRAIGHT SECTION 31—~deg BEND - 45~-deg BEND - 65-deg BEND 90-deg BEND TWO 45-deg BENDS THREE 45-deg BENDS SOURCE: GIRCULAR, 43%gin. DIA RELATIVE SOURCE STRENGTH (neutrons/cm? - sec) 4+ WX ood o 0 10 20 30 EQUIVALENT CENTIMETERS OF WATER FROM END Of DUCT Fig. 39. Comparison of Calculated Effective Source for Cylindrical pucts in Water, 73 ANP PROJECT QUARTERLY PROGRESS REPORT DWG. 15105 p | | MH ) . | ) TRANSITION \< SECTION Fig. 40. Schematic Diagram of GE-ANP Mockup of Outlet Air Duct. w7 TN 74 Sl Fig. 41. GE- ANP Mockup of Inlet Air Duct. PHOTO 10115 f [.n 30 in. ‘0T ANNL ONIANAI doIddd Hod ¢Sol ANP PROJECT QUARTERLY PROGRESS REPORT 9. NUCLEAR MEASUREMENTS A, H. Snell, Physics Division The total neutron cross section of Li® has been measured up to 4 Mev on the 5-Mev Van de Graaff. The measure- ments show only one resonance, which is at 270 kv, The time-of-flight neutron spectrometer has been operating satisfactorily with a full resolution width of less than 1.2 usec/meter. CROSS-SECTION MEASUREMENTS WITH VAN DE GRAAFF ACCELERATOR H. B. Willard, Physics Division The total neutron cross section of Li® has been measured on the 5-Mev Van de Graaff by a transmission experi- ment. In this experiment the lithium sample was 89% Li®, but the resulting data has been corrected for the 117. The cross-section curve up to 4 Mev is shown in Fig. 42. Only one resonance, which is at 270 kv, was observed in this energy range. Preliminary measurements of the fission cross section of U2%3* have 76 been made up to 4 Mev., In addition detailed information on the T(p,y)He* reaction, the B"(p,n)" angular distri- bution, the C!¥(p,n)N'? yield, and- the (p,n) thresholds in neon is reported in the Physics Division quarterly progress report, (1) TIME-OF-FLIGHT SPECTROMETER G. S. Pawlicka E. C. Smith Physics Division The neutron time-of-flight spec- trometer has been operating satis- factorily with a full resolution width of less than 1.2 psec/meter., Isotopic assignment of the levels of indium has been reported.¢!? Preliminary measurements indicate levels 1in copper at 700 £ 100 and 2200 600 ev. Measurements of the cross sections of the uranium isotopes are in progress. (I)Physics Division Quarterly Progress Report for Pertiod Ending March 20, 1952, ORNL-1289 (in press),. K FOR PERIOD ENDING JUNE 16, 1952 AENED OWG.15342 MEASURED BY A TRANSMISSION EXPERIMENT USING 89 % Li® AND CORRECTING FOR Li' { barns) o Gy | lll | 2 m i I !l’l,n,l!,mn' s ity i ™ .’"llfllufimfifi,'fi*"'“*" =L ”,m’mmMWMH'l lH'I'“l“[I“Il’l'l,IIIlllliiilldl!l'mi" In My I: 0 . e 0 | 2 3 a . NEUTRON ENERGY ( Mev) Fig. 42. Total Neutron Cross Section of Li6. 77 ......................................................................................... SUMMARY AND INTRODUCTION The research on high-temperature liquids has been devoted almostentirely to the development of a satisfactory fluoride. fuel for the aircraft reactor experiment {(sec. 10), The phase diagrams of numerous binary, ternary, and quaternary fluoride systems have been examined, and a number of compo- sitions in the NaF-ZrF -UF, and NaF-KF-ZrF,-UF, systems with as much as 4 mole % UF, have been shown to have melting points around 500°C. Although 1t 1s not certain that either of these fluoride mixtures can be used with the envisioned loading technique, preliminary data are encouraging. A substantial program of research and pilot~scale production of liguid fuels of haigh purity and a study of high- temperature transfer methods have recently been initiated. The corrosion research effort was divided among static and dynamic tests of fluorides, hydroxides, and liquid metals, with the fluoride corrosion tests predominating (sec. 11). Static tests and the modified dynamic tests (seesaw tests) that have been developed are useful for screeninga large number of samples but are inadequate for precisely predicting the corrosiveness of a circulating ligquid. However, the tests have shown that small additions of sodium, potassium, manganese, and calcium are beneficial in minimizing fluoride corrosion. Various fluoride mixtures were circulated in thermal convection loops during the past quarter. Stainless steel loops generally plug after short periods, whereas Inconel loops con- sistently operate for upto 1000 hours, Some reduction in dynamic fluoride corrosion has been obtained as a result of improved fluoride- and loop- preparation techniques. Fluoride attack on Inconel i1s believed to bhe the result of chromium diffusion out of the metal lattice. A considerable program of corrosion research om both hydroxides and fluorides has been undertaken. The metallurgical processes involved in the construction and assembly of a high-temperature reactor, including fabrication of control rods, welding and brazing, and fabrication of solid fuel elements, are being successfully developed (sec. 12). Brazing alloys were tested for flowability, corrosion resistance, and joint strength, and the 60% Pd-40% Nialloy proved superior in a fluoride environment. Powder mixtures of B,C with Fe and with Al 0O, are being hot pressed for the safety and regulating rods, respectively, of the ARE. Loose-powder sintering has been investigated as a technique for the fabrication of solid fuel elements. Heat transfer and physical property measurements on the various fluorides together with some measurements on hydroxides {(sec. 13). Determinationsof viscosity, thermal conductivity, density, heat capacity, and vapor pressure have been made on several fluoride mixtures. It as apparent that the physical properties of the ZrF -bearing fuels are compatible with reactor design requirements. In particular, the viscosity, which was so high ‘in the NaF-BeF, -UF, fuel mixture as to necessitate 1ts rejectiron, is less than 10 centipoises for the NabF-KF-Zr¥, -UF, fuel mixture at all reactor temperatures. Mathematical analyses pertaining to circulating fuel heat transfer systems have been developed. A measurement of the héat have continued, 81 ANP PROJECT QUARTERLY PROGRESS REPORT transfer coefficient of molten sodium hydroxide showed 1t to be similar to that of ordinary fluids rather than liquid metals. Pile irradiations of proposed fuel mixtures 1n Inconel capsules comprised the greater part of the effort on the radiation damage program, although measurements have continued on inpile creep and thermal conductivity of metals (sec. 14). Irradiation of Inconel capsules containing the NaF-KF-UF, fuel mixture at power densities considerably 1in excess of that to be found 1n the ABRE caused an attack on the similar capsules rate of However, increased container. containing BeF,-bearing fuels showed ne positive evidence of radiation- induced corrosion. An increase in the secondary creep rate of irradiated Inconel similar to that previously observed 1in nickel and stainless steel was found. No detectable change was observed in the thermal conductivity of irradiated specimens of specially heat-treated Inconel or nickel. 10. CHEMISTRY OF HIGH-TEMPERATURE LIQUIDS W. R. Grimes Materials Chemistry Division Researchon high-temperature liquids has been concerned almost entirely with their development for use as fuels for an aircraft reactor. In addition, some effort has been devoted to the purificationof alkali hydroxides and the determination of the high- temperature properties of these materials. The principal research work has been the phase equilibrium studies necessary to define the optimum concentration of the fuels, and, 1in conjunction with others, the de- termination of the physical properties and corrosiveness of these high-temper- ature systems. It has also been necessary to intensify the efforts to identify chemical species in the cooled melts by x-ray diffraction and other techniques. Research on liquid fuels is still directed toward development of low- melting-point solutions of UF, in mixtures of permissible fluorides. The experimental reactor (the ARE) will require about 16 1lb of U?3% per cubic foot of fuel solution, so 1t will be necessary to incorporate about 4 mole % of UF4 1n the fluoride fuel. 82 A number of compositions 1in the NaF-7ZrbF,-UF, and NaF-KF-ZrkF,-UF, systems with as much as 4 mole % Uk, have been shown to melt below 550°C. In addition, it has been demonstrated that the viscosity of typical mixtures of these materials 1s sufficiently low for successful operation (cf., 13). For easy startup, 1t 1s necessary to fill the reactor with a dilute (subcritical) fuel and add small increments of a concentrated solution to bring the system to criticality. This that solutions of widely varying uranium content and melting points considerably below the operating range be available and that no high-melting-point com- pounds be formed at intermediate concentrations. It is not yet certain, although preliminary data are en- couraging, that the NaF-KF-ZrF,-UF, system meets this reguirement. sec. requires A substantial program of research and pilot-scale production of liguid fuels of high purity and a study of high~temperature transfer methods have recently been 1nitiated. A smaller program for the study of .t chemical reactions of high-temperature fluorides and hydroxides is presently concerned with reactions that may help to explain the corrosion behavior of these liquids. Toward this end, a guenching apparatus for studying solid phases in the NaF-BeF,-UF, system has been placed in operation, and x~ray examination of complex mixtures in the NaF-KF-ZrF, system has yielded valuable information as to species present. Experimental preparation of simu- lated fuel for the cold critical experiment has been continued. Speci- fications for the powdered mixture can be completed as soon as the actual ARE fuel composition is fixed. LOW-MELTING-POINT FLUORIDE FUEL SYSTEMS I.. M. Bratcher R. E. Traber, Jr; C. J. Barton ‘ Materials Chemistry Division The high viscosity of the melts has virtually eliminated the BeF,- bearing salt mixtures from consideration as ARE fuels, and the emphasis 1s now being placed on systems containing ZrF4. In the previous report®!? it was indicated, on the basis of very limited data, that the addition of 1 to 5 mole % UF, to systems con- taining ZrF, affected the melting point only slightly. Further in- vestigation has shown that the ad- dition of 1 or 2 mole % UF, depresses the melting point of most alkali fluoride—-zirconium tetrafluoride mixtures slightly but that further additions produce an increase 1n melting point. (I)Aircraft Nucleaor Propulsion Project Quarterly Progress RBeport for Period Ending March 10, 1952, OBNL-1227, p. 101, FOR PERIOD ENDING JUNE 10, 1952 The first separation of a solid phase from these fused mixtures 1is frequently accompanied by a small thermal effect that makes detection of melting points by thermal analysis difficult. In several instances a second break, probably corresponding to a eutectic temperature, has been mistaken for the melting point. Some difficulty has been experienced with poor reproducibility of data. This may be due in part to sensitivity of the melting point to variances in the ZrF, concentration in the fused mixtures. Loss of ZrF, may result from both volatilization and con- version to zirconium oxide., These phenomena will be investigated in detail as soon as possible. NaF-KF-ZrF -UF,. Most of the effort on ZrF,-bearing fuels has been concentrated on the NaF-KF-ZrF, -UF, system, since preliminary studies indicated that it was most likely to produce a suitable fuel. Heating and cooling curves have been run on a large number of mixtures containing 4 mole% UF, and varying concentrations of the other three fluorides. The rather small thermal effect mnoted when the first solids crystallize from these systems has encouraged the use of larger samples (200 to 300 g) in these studies. Zirconium tetrafluoride purified by vacuum sublimation was used in all the ex- periments reported. The data in Table 3 show the ex- pected melting points when 4 mole % UF, is added to various mixtures 1in the NalF-KF-ZrF, system. The data are not sufficiently complete to state the optimum fuel composition at present, It 1s obvious, however, that fuels with melting points below 550°C can be obtained over fairly wide ranges of NaF-KF and ZrF, con- centration. These studies are being continued to define the optimum concentration. 83 ANP PROJECT QUARTERLY PROGRESS REPORT TABLE 3 Melting Points of NaF-KF-ZrF4-UF4 Mixtures Containing 4 mole 9 UF4 FIRST BREAK COMPOSITION (mole %) | rpvbre)s TURE NaF KF ZrF, (°c) 7.7 | 40.3 | 48.0 668 8.2 | 44.6 | 43.2 540 4.8 | 50.0 | 41.2 540 9.1 | 48.5 | 38.4 490 (?) 9.6 | 52.8 | 33.6 585 15.4 | 32.6 | 48.0 615 17.3 | 35.5 | 43.2 565 18.7 | 38.9 | 38.4 575 20.2 | 42.2 | 33.6 555 22.9 | 22.6 | 50.5 620 24.9 | 25.6 | 45.5 565 27.8 | 27.4 | 40.8 490 29.8 | 30.2 | 36.0 535 30.1 | 15.4 | 50.5 605 33.7 | 16.8 | 45.5 550 34.6 | 17.3 | 44.1 545 36.7 | 18.5 | 40.8 515 39.8 | 20.2 | 36.0 5490 36.5 | 13.5 | 46.0 540 52.8 9.6 | 33.6 545 The data in Table 4 show the effect of large additions of UF, to a single, ternary NaF-KF-ZrF, mixture. It appears from these data that the start-up operation could be safely conducted 1f the temperature were maintained at 700°C and if concen- trations of more than 25 mole % UF, were avoided at all times. In a similar study UF, was added as NaF-KF- UF, eutectic (46.5-26.0-27.5 mole %) to a NaF-KF-ZrF, mixture (33-16.1-50.9 mole %) in several proportions. No high melting points were observed at the intermediate compositions. These studies are encouraging, but they must be repeated when the final fuel composition 1s chosen. Should 1t become operate the ARE with a content than now seems the low-melting-point position previously reported(?) (4.9 mole % NaF, 51 mole % KF, 42.1 mole % ZrF,, and 2 mole % UF,) would be gquite i1mportant. It does not appear likely that compositions, of possible to lower uranium to be needed, (410°C) com- (2) Ibid., p. 102, TABLE ¢4 Effect of UF4 on the Melting Point of 2 NaF-KF-ZrF4 Mixture COMPOSITION {(mole %) BREAK TEMPERATURES* NaF KF ZrF, UF, (°c) 36 18 46 0 440, 425, 395 34.6 17.3 44.1 4.0 545, 493, 412 32.4 16,2 41.4 10.0 553, 505, 465, 405 30.6 15.3 39.1 15.0 585, 575, 500, 395 28.8 14. 4 36.8 20.0 626, 580, 480, 400 27.0 13.5 34.5 25.0 663, 575, 495 *Highest break temperature considered most reliable indication of melting point, melting in this aily uranium concentration, hbelow 400°C will be found system. KF=-2rF, -UF . Only a few com- positions in the KF-ZrF,-UF, system, all with 4 mole % UF,, have been tested. Since all these mixtures showed melting points at about 600°C (about 150°C higher than the corre- sponding KF-ZrF, binaries), it is not likely that a satisfactory fuel exists in this ternary system. NaF-ZrF,-UF,. The data obtained on the NaF-Zr[,-UF, system, shown 1in Table 5, indicate that fuels con- taining up to 4 mole % UF, and melting TABLE 5 Melting Points of NaF-ZrF, Mixtures Containing 4 mole % UF, FOR PERIOD ENDING JUNE 10, 1952 around 500°C can be obtained with this system. It may prove feasible to start the reactor with a suitable NaF-ZrF, binary and build the fuel to criti- cality by addition of the NaF-UF, binary eutectic (27 mole % UF,; melting point, 620°C). The difference in behavior of the Nal-ZrF -UF, and KF-ZxF,-UF, systems 1s certainly due to the strong tendency of KF to form stable, high-melting-point compounds with UF,. NaF-RbF-ZrF,-UF,. A few mixtures have been investigated in the NaF- RbF-ZrF,-UF, system to determine whether substitution of RbF for KF would result in lower melting points. Data obtained with such mixtures (Table 6) indicate that RbF has no decided advantage over KF as a com- ponent of a ZrF, fuel as far as COMPOSITION (mole %) MELTING POINT mMelting point is concerned. Na¥F ZrF, (°C) Y ~y D D 44.0 &9 0 597 ANALYSES OF FLUORIDE COMPOUNDS 50.0 46.0 513 X-Ray Examination of Solid, Com- 46.0 50.0 505 plex Fluorides (P. A. Agron, Materials Chemistry Division). To assist 1n . . 1 ) - . . 52.8 43 2 52 the 1dentification of the solid, 57.0 39.0 335 complex fluorides encountered in phase studies of fuel mixtures, the technique TABLE 6 Effect of UF, on Melting Point of NaF-RbF~ZrF, Mixtures COMPOSITION (mole %) FIRST BREAK TEMPERATURE NaF RbF ZrF, UF, (°C) 35.0 20.0 45,0 0 445 34.3 19,6 44,1 2.0 480 34.0 19.4 43.6 3.0 540 33.6 19,2 43.2 4.0 575 10.90 45,0 45.0 0 445 9.6 43,2 43.2 4,0 485 85 ANP PROJECT QUARTERLY PROGRESS REPORT of x-ray-diffraction examinatioen on a high-angle spectrometer was adepted. Analyses of the binary complexes occurring 1n cooled melts from the equilibrium phase studies indicate the presence of the compounds and polymorphic forms listed in Table 7. The crystal structures of the two polymorphic forms of K,ZrF, have not vyet been established. An examination of other concentrations in the KF-ZrF, and the NaF-ZrF, made. systems 1s being Spectrographic Analysis (Bussell Baldock, Stable Isotope Hesecarch and Production Division). Efforts to develop a mass spectrometer method for 1nvestigation of aircraft fuels has been delayed by the discovery that the molecular fragmentation patterns of the fuel i1ngredients were more conmplex than was anticipated. In particular, the UF, dissociation pattern was largely masked by the superposition of the dissociation fragments from UF ;. It is now believed that the UF; comes from UO,F,, which is a contaminant in the UlF,, and arrangements are being made to use some 1sotopically enriched ingredients to establish the origin and magnitude of the UF,. The presence of UF, as a contaminant in fuel mixtures could give rise to considerable corrosion. The 1nvestigations of elemental compounds are summarized 1n the following statements. Uranium tri- fluoride dissociates at temperatures above 700°C into UF, and uranium metal, 1n agreement with findings reported in the literature, but at lower temperatures than heretofore reported. No evidence of the subli- mation of UF;, as such, could be found at temperatures up to 750°C with the mass spectrometer or up to 1000°C in sublimation tests. Even UF, prepared by the reduction of 86 UF, with uranium metal shows evidence of higher valence compounds of uranium that give rise to UF, on heating. All unheat-treated UF, examined has shown pronounced evidence of UF5 contamination when examined 1in the mass spectrometer, some Study of Solid Phases in the NaF-BeF,-UF, System (A.G. H. Andersen, ANP Division, and C. J. Barton, Materials Chemistry Division). A study of the solid phases 1in the NaF-BeF,-UF, system was initiated at the time beryllium fuel was being considered for use 1in the ARE. Since a fuel composition in this system may be suitable for use 1n a reactor design not requiring a low viscosity fuel, this study has been continued. A furnace has been set up for the heat treatment of fluorides or fluoride mixtures sealed 1n quartz tubes under a vacuum. The furnace 1s arranged so that a number of samples can be heat-treated simultaneously and quenched by dropping into liquid nitrogen or some other quenching medium, The quartz capsules are heated to a definite temperature (£10°C) for intervals ranging from 16 to 60 hr and then quenched or cooled slowly. The samples are then examined with a petrographic micro- scope and by x-ray diffraction. In the UF,-BeF, system, three crystallinephases have been 1dentified: UF,, BelF,, and UO,. In addition, a BeF, glass 1s present in mixtures high in BeF,. This glass can be converted into crystalline BeF, by long heating at 300 to 600°C. The UO, present 1s produced either by hydrolysis or oxidation of the UF, during the original sample preparation. An unidentified phase appeared in a 90 mole % BeF,-10 mole % UF, mixture heated at 300°C for 60 hours. FOR PERIOD ENDING JUNE 10, 1952 TABLE 7 Phase Studies of Complex Fluoride Compounds . - Lattice Dimensions(?) A (b) Compound Crystal Form _ (A) D, (g/ce) a-K,UF, Cubic a = 9,21 4,12 a'-K,UF, Tetragonal ' a; = 9.20 4,13 ay, = 18.40 51*K2UF5 Hexagonal a, = 6,54 5.10 @, = 3.76 B,-K,UF, Hexagonal | a, = 6.53 4,77 _ ' a; = 4,04 | KUF, Rhombohedral _ a = 9,387 ' 5. 38 a = 107° 157 KU,F, Orthorhombic | a, = B8.68 ' 6.49 ' a, = 7,02 as = 11.4‘4‘ ‘ Na,UF, Tetragonal - a, = 5.448 4,49 | ' | a, = 10.896 | B,-Na,UF Hexagonal e, = 5.94 f 5.74 a, = 3.74 NaUF, Rhombohedral a = 9,08 | 5.81 ' a = ]107° 56 K ZcF, Cubic a = 8.951(¢) . {(d A~K22rF6( ) B-K,ZrF (4] (Q)W. H. Zachariasen, The Crystal Structure of Nn3UF7, AECD-1798 (Mar., 3, 1948); The Crystal Struc- ture of 7-NaZUF6 yAECD-2089 (June 29, 1948); New Crystal Structure Results., Part J., AECD-2093 (Jpne 28, 1948); The Crystal Structure of Alpha-Phase Coapounds ApXFg and AXF,, AECD-2162 (July 19, 1948); The Crystal Structure of Beta-Phase Compounds A XFg and AXF4, AECD-2163 (July 20, 1948). (b)Theoretical density calenlated from Xerayediffraction data. ()G, C. Hampson and L. Pauling, J. An. Chem. Soc. 60, 2702 (1038). (44 2ZrF6 is the Jow-temperature, stable polymorphic form, and B’Kzszfi is the high-temperature form. : ' 87 ANP PROJECT QUARTERLY PROGRESS REPORT In the NaF-BelF, binary system, only three samples have been prepared: one with 40 mole % NaF and two with 50 mole % NaF. An unidentified phase, NaBek,, appeared i1n these mixtures. 3 pp Only one composition in the NaF- BeF,-UF, system has been examined: 47 mole % NaF, 51 mole % BeF,, and 2 mole % UF, (composition 17). 1In each sample of this mixture, held at temperatures ranging from 250 to 500°C, either crystalline or glassy BeF, has appeared. Some UO, was present but UF, was not positively identified. An unidentified phase appeared 1n this mixture also. Petrographic Examination of Fluo- rides (1T. N. McVay, Consultant, Metallurgy Division). Several hundred examinations have been made of binary mixtures of KF-ZrF,, NaF-ZrF., NaF-UF,, Bel,-UF,, and K¥F-UF, as well as ternary and quaternary mixtures of a number of the fluoride components. In addition, numerous samples of fuels taken from capsules and loops have been examined. The optical properties of a number of fluoride compounds not found in the literature have been determined. All index of refraction determinations are probably accurate to 10.003. Li,CrF, Biaxial negative 2V = about 40 deg Alpha = 1,444 Gamma = 1.464 K,CrF, Cubic n = 1.422 Na,CrF, Cubic n = 1.411 88 KUng Orthorhombic Biaxial negative 2V = 10 deg Alpha = 1.544 Gamma = 1.588 UF, Monoclinic Biaxial negative 2V = 75 deg Alpha = 1.500 Beta = 1.585 Gamma = 1.598 Anomalons interference colors dark green z x = light green The following systems and indices were determined by previous 1n- vestligators: NazBeF4 Orthorhombic Y || C, low birefringence Average indices ~ 1.303 NaUF, Hexagonal Uniaxial negative 0 = 1.520 E = 1.512 KUF, Hexagonal Uniaxial negative 0 = 1.512 UF, Hexagonal Low birefringence n = 1,73 Anomalous interference colors Bel , Crystallized Anisotropic Low birefringence n o= 1,328 The optical properties of several other compounds have been determined, but the compounds have not as vet been identified. Jt was found in the examination of capsules containing fuel 21 that the oxygen present reacted with the ZrF4 to form Zr0,. The evidence to date is that the oxygen present will react with Zr¥, to form well-crystallized ZrQ,. Zirconium tetrafluoride appears to be a more potent getter for oxygen than UF,. SIMULATED FUEL MIXTURE FOR COLD CRITICAL EXPERIMENT D. R. Cuneo L. G. Overholser Materials Chemistry Division The lack of complete definition of the ARE fuel has prevented preparation of final specifications for the simulated fuel for the ARE critical experiment. Since the ARE fuel will probably contain ZrF,, NaF, KF, and UF,, experimental studies have been confined to mixtures containing these fluorides. Lack of an adequate supply of hafnium-free ZrF, has made necessary the substitution of Zr0, in the powder mixture. Carbon is also added to the mixture to compensate for the di fference in moderating power of the Zr(G, and ZrF,. Calcination of the hafnium-free Zr0, at 800°C is sufficient to reduce the water content to below 0.1%. Sodium fluoride has been shown to be sufficiently pure and dry as received, FOR PERIOD ENDING JUNE 10, 1952 but some samples of KF may require vacuum drying. By grinding UF,, Zr0,, NaF, KF, and activated carbon in a small ball mill and by packing the mixed powder imto stalnless steel tubes, 1t has been possible to ebtain a bulk density of about 1.8. The powder so packed can be made to show a uniform uranium density of the proper value, and no detectable segregation of the UF, occurs with normal handling. By use of reasonable precantiouns, such as drying of the Zr0, and KF and subsequent handling of the powders in a dry box and a sealed ball mill, the water content of the final mixture can be kept below 0.2%. The hydrogen- to-uranium ratio that may be tolerated corresponds roughly te 0.5% H,0 1in the mixture. Grinding, blending, and canning of the fuel and the preparation of the large quantity of uranium-free base material for the plenum chamber will be started soon after the final ARE fuel composition is chosen. PREPARATION OF PURE HYDROXIDES E. £. Ketchen .. G. Overholser Materials Chemistry Division The only experimental work on moderator-coolants during the past quarter has been concerned with purification of several of the alkala hydroxides. Sodium hydroxide with less than 0.1% Na,CO; has been con- sistently prepared, and a practical technique for the removal of carbonate and sodium i1ons from Li0OH has been developed. Pure KOH has been prepared both by removing the carbonate from and by the re- action of petassium and water. The latter technique is expected to ultimately produce the purest products; however, the best material prepared commercial material 89 ANP PROJECT QUARTERLY PROGRESS REPORT to date by either method has con- tained 0.10% K,O,. Sodium Hydroxide. Because of continued interest in purified NaOH, additional batches have been prepared by the method in which Na,CO,; is removed by filtration from a 50% aqueous solution of NaOH and the clear supernatant liquid is dehydrated under vacuum at 450°C. The material has assayed 100.0% NaOH, and the Na,CO; content has been less than 0.1% in all runs. Lithiusm Hydroxide. A limited amount of work has been devoted to the purification of LiOH by using the apparatus previously described(3) for the purification of Ba(OH),. Since the method proved to be ef- fective in removing the two major impurities, sodium and carbonate ions, it appears to be a practical procedure for the preparation of very pure LiOH when the demand warrants such production. Potassium Hydroxide. 'The major effort has again been devoted to the preparation of pure KOH by removing the carbonate as BaCO; in an aqueous medium and dehydrating the filtrate under vacuum at 475°C. Determination of the solubility of BaCO,; as a function of KOH concentration 1in- dicated that the minimum carbonate content could be obtained i1f a KOH concentration of 45 to 50% were used. Under these conditions 1t appeared possible to prepare a product con- taining carbonate equivalent to 0.07% K,CO,. To date, given a product corresponding to 0.10% K,CO; and containing 0.11% Ba. The low value for carbonate is due in part to a redesigned dehydration vessel that eliminates contamination only one run has (3) Aircraft Nuclear Propulsion Project Quarterly Progress Report for Period Ending December 10, 1951, ORNL-1170, p, 84, 90 of the KOH by atmospheric CO,; this preparation probably approaches the lowest value possible by this method. Substitution of Ca(OH), for Ba(OH), to precipitate the carbonate proved unsatisfactory. One run has been made in which pure potassium was reacted with excess water and the resultant solution of KOH was dehydrated by heating under vacuum at 500°C. The apparatus used allowed very slow addition of water and removal of both hydrogen and heat, thus the reaction occurred under carefully controlled conditions. The apparatus was design so that the dehydration could be accomplished 1n the same vessel in which the water reacted with the potassium. Material from this first runm contained 0.10% K,CO;. A vacuum dry box 1n which a very pure atmosphere can be attained will be used in future work, and the dehydration system will be modified to improve its vacuum performance. With these refinements it 1s expected that a purer product will be obtained. COOLANT DEVELOPMENT I.. M. Bratcher R. E. Traber, Jr. C. J. Barton Materials Chemistry Division The studies of uranium-free fluo- ride systems to determine thear possibilities as coolants as well as fuel solvents have been confined to systems containing ZnF, and ZrF, for which preliminary data have been presented previously.(*%) It now appears that the previous data on systems containing ZrF, was influenced by the presence of Zr0, or ZrOF, 1in the samples studied. Sublimed ZrF, of high purity became available 1in quantity during the past qguarter and has been used in all the studies 4)op. cie., ORNL-1227, p. 104, reported here. The simpler thermal data obtained with the highly pure material have helped to clarify the phase relationships to some extent. It should be emphasized, however, that the applicationof the thermal analysis technique to the study of alkali fluoride-zirconium fluoride systems has not yet produced completely satisfactory phase equilibrium dia- grams. The diagrams presented here should be regarded as tentative and subject to revision after further study. Thermal studies are being supplemented by examination of solid phases by means of the petrographic microscope and x-ray-diffraction equipment. The results of these studies are given elsewhere in this report. ' FOR PERIOD ENDING JUNE 10, 1952 The two- and three-componentsystems containing ZnF, have not shown melting points sufficiently low to be of interest at present; however, they may be of value as constituents of more complex mixtures. : NaF-2ZrF,. Cooling curves for the NaF-ZrF, system have been obtained from a large number of mixtures con- taining from 5 te 60 mole % ZrF,, but the data obtained do not give a satisfactory phase diagram. The compound NaZrF, melts at 510 * 10°C and thereis a eutectic at approximately 43 mole% ZrF, that melts at 480 t 10°C, KF-ZrF,. Thermal breaks for KF-ZrF, mixtures containing from 5 to 65 mole % ZrF, are shown in Fig. 43. OWG. 14623 l l | ] T 1 : KstF? 200 e 800 -— © 2. w 7Q0 . 4ol > — < 1 i = = 600 — - 500 — 4 00— . s00 L | | | L KF 10 20 30 40 . 50 60 70 80 30 .ZrF4 ZrF (mole %) Fig. 43. The System KF-ZrF4. 91 ANP PROJECT QUARTERLY PROGRESS REPFORT The melting point of the eutectic at 13 mole % ZrF, (765°C) and of the compound K§ZrF7 (915°C) were reported earlier. () The data indicate ad- ditional eutectics at 42 and 55 mole % ZrF, with melting points at 430 and 440°C, respectively. The compound KZrF, appears to melt congruently at 445°C. The compound K,Zrk, probably exists, but efforts to determine 1its melting point with fused mixtures of KF and ZrF, and with commercially obtained material have been un- snccessful. Data on mixtures con- taining more than 65 mole % ZrF, are unreliable because of the volatility of ZrF,. S rhia., p. 105, RbF-Zr¥,. The phase equilibrium diagram for the RbF-ZrF, system 1s shown in Fig. 44. The melting point of the eutectic at 6 mole % ZrF, 1is 725°C and that of the compound Rb;ZrF, is 880°C. The eutectic at about 40 mole % ZrF, appears to melt at 400+ 10°C. There 1s probably a compound RbZrF, melting at 425 + 10°C. The data at 55 and 60 mole % ZrF, do not show whether the compound melts congruently. There 1is some evidence for the ex- istence of the Rb,ZrF. compound, but, as 1n the case of K,ZrF., 1t was not possible to determine its melting point by the conventional thermal analysis techniques with mixtures of RbF and ZrF4. NaF-KF-ZrF,. Tentative contour lines for the NaF-KF-ZrF, system - DWG.14624 1000 [ ] S00 700 TEMPERATURE {°C) 500 400 | 200 | | | | RbF 10 20 30 40 50 60 70 80 90 ZrFy Zrfg {mole To) rig. 44. 62 The System RbF-ZrF4. are shown in Fig. 45. There appears to be a low-melting region between 40 and 50 mole % ZrF, that extends almost all the way across the diagram. The lowest melting region lies close to the KF-ZrF, eutectic. The data give no definite indication of ternary compound formation. ‘No attempt has been made to obtain data on mixtures containing more than 55 mole % ZrF, because of the high volatility of ZrF,. FOR PERIOD ENDING JUNE 10, 1952 NaF-RbF-ZrF,. The tentative contours for the NaF-RbF-ZrF, system, shown in Fig. 46, differ to a con- siderable extent from those for the NaF-KF-ZrF, system. The lowest melting region appears to occur at higher ZrF, concentrations except for the region very close to the RbF-ZrF, eutectic. On the basis of the data available at this time, this system shows no definite advantage over the DWG. 14622R1 Na¥F 995°C Fig. 45. KF 850°C Tentative Contours for the System NaF-KF—ZrF4. - 93 ANP PROJECT QUARTERLY PROGRESS REPORT NaF-KF-ZrF, system, and no further work on it 1s contemplated at present. Na¥-ZnF,. The equilibrium diagram for the NaF-ZnkF, system is shown in Fig. 47. Only one compound is formed in this system: NaZnF;, which has a melting point of 748 + 10°C. The two eutectics are at approximately 32.5 and 69 mole % ZnF, and melt at 640 and 685°C, respectively. Zrky NoZrFS INDETERMINATE KF-ZnF,. The data for the KF-ZnF, binary system, as shown in Faig. 48 give some evidence for the formation of K,ZnF, and Kink,. The KZnF, melts congruently at 850 + 10°C, and the K,ZnF, appears to melt (with at 720 + 10°C. The two eutectics at about 21 and 80 mole % ZnF, melt at 670 and 740°C, respectively. decomposition) DWG. 15346 510°C \ u ' » e, — ‘ = 600 N NaF \\\\\Qk\J \ RbF 995°C Fig. 46. 94 795°C Tentative Contours for the System NaF-RhF-ZrF4. FOR PERIOD ENDING JUNE 10, 1952 Owe. 14627 1000 1 T I [ l I ® 300 +— o 06 BOOF— G — NoZnFB |31 | . & : %3 700 @ @ ® = @ Lt b o @ .. ® & ® @ © ® © 6 00— _ ~ 500/ . _‘ soo | { | | | L 1 | | NaF 10 20 30 40 50 &0 70 80 20 ZnFy ZnfF, (mole %) Fig. 47. The System NaF-Zan. OWG. 14626 £4000 { ~ T T | 1 200 800 S N 700 x D & o i . £ 600 |- ' . ] - 500 |— ' _ — 400 | : ' — | 300 l I ! J + l L 1 KF 10 20 30 40 50 60 ° 70 80 90 ZnFy _ . ZnFy (mole %) Fig. 48. The System KF~ZnF2. 95 ANP PROJECT QUARTERLY PROGRESS REPORT RbF-Zn¥F,. Strong evidence for the existence of the Rb,ZnF, compound 1is found in the equilibrium diagram for the RbF-ZnF, system, as shown in Fig. 49. This compound melts 1in- congruently at 620 + 10°C, whereas RbZnF, melts congruentlyat 730 * 10°C. The lowest melting eutectic in this system 1s at approximately 20 mole % ZnF, and it melts at 595 %+ 10°C, The other eutectic, which 1s near 70 mole % ZnF,, melts at 650 & 10 °C. NaF-KF-ZnF,. Not enough datahave been obtained for the NaF-KF-ZnF, ternary system to define the location of contour lines accurately. The lowest melting point found was 626°C for the mixture containing 25 mole % IZnF,, 30 mole % NalF, and 45 mole?% KF. The cooling curves do not a1ndicate that any lower melting compositions exist in the system, so no further work on it is contemplated at this time . FUEL PREPARATION AND LIQUID HANDLING F. F. Blankenship Materials Chemistry Division Because of the need for liguad fuel samples for chemical examination, physical property evaluation, and corrosion testing, a program ifor preparation of the molten liquids on research and pilot-plant scales has been started. The need for materials of high purity for the final reactor fuel, as well as the indications that corrosion by the liquids 1s strongly dependent on purity of the materials, has directed the research program to a study of feasible methods for SR DWG. 15349 1000 1 200 (°c) = = 700 MPERATUR TE 600 500 (— | 400 L RbF 10 20 30 40 50 60 - [4¢] 80 InF, (male %) Fig. 49. 96 The System RbF-Zan. preparation of very pure liquids and methods for handling the liquids to minimize contamination. Preparation of pure ZrF, has been accomplished by vacuum sublimation of impure commercial preparations and incompletely hydrofluorinated materi- als available in this laboratory. The preparation of this material will be handled in the future by the Y-12 Production Division; a supply of up to 50 lb per week of pure, sublimed ZrF, will be available after about June 15. The preparation of pure liguid fuels of various types has, apparently, been accomplished on a research (5-1b batch) scale. Installation of similar e quipment to turn out 10- to 100-1b batches for larger scale testing should be complete by the time the ZrF, production permits their use. Zirconium Fluoride Production (C. M. Blood, J. E. Eorgan, G. J. Nessle, Materials Chemistry Division). The sudden large demand for ZrF,, which developed when mixtures con- taining BeF, were shown to be un- satisfactory, could not be met without delay. Delivery from commercial suppliers was slow, and their products contained up to 20% of ZrO, and ZrOF,. The commercial material has been purified satisfactorily by vacuum sublimation in nickel equipment capable of handling 1 to 1.5 kg of material per charge. The sublimers are 4-1in. cylinders, 18 inches 1in length, containing disk-shaped baffles arranged to provide a tortuous path for the vapor. An air-cooled finger suspended from the gasketed lid serves to condense the pure ZrF, - Thermocouples and inert gas and vacuum connections are introduced through the laid. The output of the sublimers has increased, along with the supply of FOR PERIOD ENDING JUNE 10, 1952 raw material, to about 10 kg per week. The best index of purity of the material has been examination of the product with the petrographic microscope to ascertain uniformity and the presence of not more rhan trace quantities of ZrD With the equipment and charge material available at present 1t appears that the sublimation of 1500-g portions is best conducted at 800°C for about 4 hr at a pressure of 450 microus. Under these conditions virtually all the ZrF, is volatilized and about 80% of the material 1s deposited on the cold finger. The rest is deposited on the upper portion of the cylinder walls and 1s added to another batch to be sublimed. Under optimum conditiens the product is snow-white. However, a slightly green coloration appears, which 1in- creases with time, temperature, and decreasing pressure. This discoloration seems to be from nickel fluoraide formed by attack by HF, which 1is formed from the hydrolysis of ZrF, by the trace of water it contaains. Arrangements have been completed for the Y-12 Production Division to supply 50:1b per week of pure, sublimed ZrF, beginning in mid-June. The plant for hydrofluorination of hydrous zirconium oxide and sublimation of the product is scheduled to be finished June 1. This plant, which should be capable of double the design figure 1f necessary, is also adaptable to the use of hafnium-free ZrO, when this becomes desirable. PREPARATION OF PURE FUEL MIXTURES C. M. Blood A. J. Weinberger: F. P. Boody G. J. Nessle Materials Chemistry Division Several impurities are known to be present in the materials that are 97 ANP PROJECT QUARTERLY PROGRESS REPORT incorporated in the fuel preparations. The sodium and potassium fluorides are hygroscopic and contain up to 0.4% H,0 and about 200 to 300 ppm of sulfur, mostly as sulfate. The ZrF, even after sublimation contains traces of ZrO, and probably ZrOF,. The commercial UF, contains small amounts of water, up to 1% of hexa- valent uranium probably as UO,F,, and traces of UO,. Sulfur compounds are known to be quite corrosive expecially to nickel and its alloys and hexu- valent uranium is a strong oxidant. Water reacts at elevated temperatures with UF, and ZrF, to yield U0, (and Zr0,) and HF. Accordingly, a purifi- cation technique has been developed for the preparation of small batches of fuel for laboratory research purposes. This technique has been adapted to equipment capable of the production of 10- to 100-1b batches (cf., "Fluoride Production™in sec. 3). It appeared that a sequence of operations that would reduce SO4= and U02++ to lower valence states followed by high-temperature hydrofluerination of the liquid would be desirable. The sulfur, which had been reduced to S°, would be eliminated as H,5, and the oxides and oxyfluorides of uranium and zirconium formed by hydrolysis would be reconverted to fluorides. The apparatus used for this purpose is shown schematically 1n Fig. 50. The heated copper and titanium traps for removing oxygen from the 1inert gas and the cold trap assembly for removing water from the inert gas and hydrogen are not shown. The fuel constituents are mixed as powders and charged to the vessel. The air is removed and an HF atmosphere 1s introduced by repeated evacuation and flushing. The melting operation UNCLASSIFIED DWG.15350 GAGE GAGE . S ;%6 g fl O:i"é 2 QlL VACUUM TRAP PUMP b U X5 2} SODA-LIME TRAPS H,0 AND LIQUID HF ( Dot REACTOR RECEIVER & & T 5 < = T Fig. 50. Apparatus for Fuel Hydrofluorimation. 98 is then conducted under an HF atmosphere to minimize the hydrolysis produced from the small amounts of water adsorbed on the powders. Hydrogen is bubbled through the molten liquid for 2 hrat 600°C to reduce the uranium to the quadrivalent state. Hydro- fluorination of the liquid 1s carried out for at least 1 hr at 600°C or higher to insure conversion of the oxides. The HF is& removed by stripping with an inert gas at 800°C and the fuel is forced through a sintered- nickel filter to the recetiver. After the sample has cooled to room tempera- ture under an inert gas the container can be detached and sealed without exposure of the material to an un- controlled atmosphere. The apparatus and all associated lines exposedto HF at high temperature are made of nickel. Lines carrying HF at low temperatures are made of copper, and copper 1s also used as the gasket material. The valves seats are Monel or Fluorothene. The major difficulties at present are those associated with maintaining a high-temperature system (600 to 800°C) that will contain HF without leaks. The receiver must be connected and disconnected for each run, and the connections are somewhat vulnerable to the extreme conditions of exposure. Approximately twenty 2-kg batches of fuel have been prepared during the quarter. The present production rate is 3 batchés per week. The hydrogenation and hydrofluor- ination treatment seemsto be effective in removing sulfur compounds. It is significant that hydrolysis is mini- mized, and re-hydrofluorination seems quite successful, since no UD, or Zr(Q, has appeared in or on the nickel filter. Liquid Handling Equipment (C. M. Blood, F. P. Boody, A. J. Weinberger, FOR PERIOD ENDING JUNE 10, 1952 G. J. Nessle, Materials Chemistry Division). Purification procedures have been perfected to the point that the handling of the fuel materials after purification can be the major source of contamination. However, equipment has been developed to ac- complish transfer of fuels without exposure to contaminating atmospheres. In this equipment, molten fluoride from a large reservoir is transferred by inert-gas pressure to a small container equipped with an overflow so that a sample of constant size 1is caught. This sample 1isthen transferred to the capsule or other receiver by pressure of inert gas. There are no valves in the ligquid lines and transfers from the container to the metering volume and thence to the receiver are controlled by valves in the inert gas—vacuum manifold. Except for the gas and vacuum lines, the whole system must be maintained at about 600°C while transfers are 1n progress. Access to vacuwnm, HF, and H, manifolds is provided; the receiving container can be hydrogenated, hydrofluorinated, or otherwise treated as required. Trials to date have shown that the method is feasible and indicate that the apparatus can be operated on a routine basis after a few modifications to eliminate some of the minor daffi- culties. Tt 1s worthy of note that apparatus of this type should be quite useful for bringing the reactor slowly to criticality by injection of a concentrated solution of UF, in the alkali fluorides. : SOLUBILITY OF URANIUM IN SODIUM CYANIDE(®) Sodium cyanide has been proposed for use as a fuel carrier because of its stability at high temperatures (6)op. cit., ORNL-1170, p. 103 99 ANP PROJECT QUARTERLY PROGRESS REPORT and its nonoxidizing characteristics. Accordingly, several tests have been run 1n an effort to determine the solubility of uranium in this cyanide. It was considered necessary to filter the melten NalCN after a soaking period at high temperature to eliminate any particles of uranium that might possibly be present. For this reason the test was conductedin the following manner. Sodlum cyanide contalning uranium turnings was sealed into the lower end of the test container, which consisted of a piece of 1/2-in.- OD type~-1035 steel tubing with an iron filter welded in at the center. After holding for 100 hr at 816°C, the tube was inverted and held foT 24 hr at 816°C to allow the molten NaCN to filter through into the other It 1is believed that the relatively hi1gh percentage of uranium in the first test might be due to a leak between the filter that was welded 1nto the tube and the tube wall. A small amount of free sodium was found in the upper part of the tubes at the conclusion of the tests. Since NaCN 1s hygroscopic, 1t is possible that much of the uranium may have united with the water that was present instead of dissolving in the NaCN. Therefore it is planned to run similar tests with dehydrated NaCN to determine whether the solubility of uranium can be 1ncreased i1n this manner. TABLE 8 Solubility of Uraniom in Sodium Cyanide end of the tube. When the filtering period had ended, the tube was allowed TEST URANTUM (%) IRON (%) to cool and the NaCN was removed and analyzed for uranium. The analyses 1 0.026 0.01 that were reFeived from the;e tests 9 0.001 0.096 are listed in Table 8. Since the test was run 1n a type-1035 steel 3 0.001 0.038 tube, an 1iron analysis was also made. 11. CORROSION RESEARCH W. D. Manly, Metallurgy Division W. R. Grimes, Materials Chemistry Division H. W. Dynamic corrosion tests in thermal convection loops have been conducted with uranium-bearing fluoride mixtures and nonuranium-bearing fluoride mixtures in an effort to find a suitable container for the fluoride fuels to be used in the aircraft reactor experiment, Thirty loops have been run with fluoride as a circulating medium, The container materials tested have included the 300- and 400-series stainless steels, Nimonic, and Inconel. The stainless steel loops operated only a short period of time before pPlugging, whereas the Inconel loops 100 Savage, ANP Division operated for 500 and 1000 hr success- fully. However, the Inconel suffered severe corrosion to a depth of 10 to 15 mils with the formation of voids. The formation of these subsurface holes at the hot zone was found to be the result of chromium diffusion to the molten bath., Lattice vacancies resulting from such diffusion merely "precipitate"” to form the observed voids, Static and modified dynamic tests are being made in a concentrated effort to ascertain possible inhibitors for fluoride corrusion. Additions of sodium, potassium, manganese, and calcium metals have been shown to be guite beneficral in minimizing the corrosive action of the fluorides on the container material. Several static tests on the compatibility of various fluoride mixtures with beryllium oxide and other ceramics have been performed. Other static and modified dynamic tests were conducted in an effort to find a possible corrosion and/or mass-transfer inhibitor for the metal- hydroxide systems., Preliminary experi- ments have shown that the corrosion and mass transfer usually associated with the metal-hydroxide systems can be appreciably reduced by the removal of oxygen from the system with purified hydrogen. In the study of the reaction products of container materials and the hydroxides, a new compound has been found that is believed to be NaNiO,. This is the first time that NaNi0O, has been shown to exist. A similar compound, NaFeOZ, has also been found as a reaction product of iron with sodium hydroxide, but this compound has been identified by previous investigators. Two simplified methods for studying dynamic corrosion have been developed. Preliminary tests using these methods for studying the metal-hydroxide mass-transfer phenomenon have shown that the methods operate successfully. In the belief that a better under- standing of the hydroxide and fluoride corrosion mechanism will ultimately lead to the reduction 1f not the elimination of corrosion, considerable effort is being devoted to fundamental corrosion studies. The work includes not only the synthesis and determi- nation of corrosion reaction products, but also such problems as emf measure- ments of mass-transportfphenomena and determination of the free energy of postulated reactions. FOR PERIOD ENDING JUNE 10, 1952 STATIC CORROSION BY FLUORIDES D. E. Vreeland R. B. Day E. E. Hoffman .. D. Dyer ‘Metallurgy Division Static corrosion tests, while not conclusively indicative of the cor- rosion resistance of a material under dynamic conditions, are useful for the preliminary screening of a Jarge number of specimens. In particular, the effect on corrosion of various additives in a fluoride mixture and the effect of fluoride corrosion of plated metals have been examined by the static- capsule technique. Neither the additives nor the plating was signifi- cantly beneficial, and when an imper- fection existed 1n the plating the resulting attack of the base metal was quite severe. Measurements of the static corrosion resistance of Inconel and stainless steel have shown little or no correlation with temperature between 538 and 1000°C. Effect of Additives. Static cor- rosion tests of several materials in fluoride mixture NaF-KF-LiF-UF, (10.9-43.5-44.5-1.1 mole %) with approximately 10 wt % additions of Zr, Na, and U and also approximately 5wt % additions of L1, K, Ca, Li, aund Mn have been run for 100 hr at 816°C. These additions appeared to have some effect i1in inhibiting corrosion 1in these tests. Inconel, type-321 stainless steel, and A nickel were the metals tested., Specimens and tubes from the tests with Zr and U additions had developed surface layers. A nickel tested with the 5% Na additive also showed some evidence of developing a surface layer. Inconel developed surface layvers in the Li, Ca, and Ti addition tests. Type-321 stainless steel developed surface layers in the Na and Ca addition tests. In those 101 ANP PROJECT QUARTERLY PROGRESS REPORT tests in which no surface layers were observed, i1t 1s, of course, possible that a surface layer may have been inadvertently removed during stripping. test with type-309 the same fluoride In a previous stainless steel 1in a surface This layer mixture with a Zr addition, layer was also developed. was examined by x-ray diffraction and reported to be composed of UO, and Zr0,. The resnlts of the most recent tests are summarized in Table 9. TABLE 9 Effect of vVvarious Additives on the Static Corrosion of Several Metals by NaF-KF-LiF-UF, After 100 hr at 816°C MATERI AL ADDITION METALLOGRAPHIC NOTES Type-321 stainless steel 10% Zr 1/2-mil surface layer on specimen and tube, no attack beneath layer Inconel 10% &r 1 1/2-mi] surface layer on specimen and tube, sub- surface voids to 1 mil under surface layer A nickel 10% Zr 2 1/2-mil surface laysr, no attack under layer Type-321 stainless steel i0% Na No attack Incenel 10% Na No attack A nickel 10% Na 1/2-mi] surface layer on tube only, no attack Type-321 stainless steel 10% U 2- to 3-mil surface layer, no attack under layer Inconel 10% U 1- to 3-mil surface layer, voids to 1 wmil under layer A nickel 10% U 2-mi]l surface layer, mno attack under layer Inconel 5% Li 1-mi] surface layer on specimen, tube has traces of surface layer; specimen attacked less than 1/2 mil Type-321 stainless steel 5% 1.1 No attack Inconel 5% K No attack Type-321 stainless steel 5% K No attack Inconel 5% Ca 1/2-m1l surface layer on tube, no attack beneath tube; specimen lost Type-321 stainless steel 5% Ca Tube has surface layer 1/2 to 1 mil thick; specimen was attacked and has a surface }layer abont 1/2 mil in depth Inconel 5% Ti Surface layer of 1/2 mi]l present, no attack beneath Yayer Type-321 stainless steel 5% Ti No attack Inconel 5% Mn No attack Type-321 stainless steel 5% Mn Subsurface voids to 1/2 mil on specimen and 1 mil on tube 1062 Temperature Pependence. Temperature dependence tests 1in NaF-KF—LiF»UF4 (10,9-43.5-44.5-1.1 mole %) have been completed, Inconel and types-430, -304, and -321 stainless steel have been tested for 100 hr at 538, 704, and 1000°C, The maximum penetrations of both containers and specimens are listed in Table 10. The results of previous tests of these materials at 816°C are also included for comparison. It would appear from these results that in statlc corrosion tests with these materials there is little or no effect on the extent of corrosioen that can be traced to varying the tempera- ture within the range mentioned above. Apparently the phenomenon of sensiti- zation occurring in type-304 stainless steel and Inconel has little if any effect on corrosion by these fluoride mixtures under the testing conditions employed. The temperature range for sensitization is usually accepted as 400 to 850°C, The first three test temperatures in Table 10 are within the sensitization range (538, 704, and 816°C). The final test temperature of 1000°C is well above the sensitization range, and FOR PERIOD ENDING JUNE 10, 1952 yet no significant reduction of cor- rosion can be noted. Effect of Plating Metals. A speci- men of type-304 stainless steel that had been electroplated with approxi- mately 8 mils of nickel was lightly cut along one face with a hacksaw in order to:expose the base metal, This specimen was then tested inNaF-KF-LiF- UF, (10.9-43.5-44.5-1.1 mole %) for 100 hr at 816°C under vacuum. The nitckel plating was apparently un- attacked, although many voids, which may have occurred during plating, could be noted, At the sawcut the type-304 stainless steel was more severely attacked than any other steel corrosioen specimen that had been tested in the static fluoride mixtures. There was a layer approximately 12 mils thick of either corrosion product or unremoved fluoride, beneath which was a region of severe intergranular attack 15 mils deep (Fig. 51). : The results of this test might be interpreted as giving some indication that an electrochemicdl type of attack had taken place. Tt 1s well known in aqueous corrosion, in which it has been TABLE 10 Depth of Attack of Metal Specimens in NaF-KF—LiF-UF4 for 100 hr at Various Temperatures DEPTH OF ATTACK (mils) METAL 538°C 704°C 816°C 1000°C Specimen Container | Specimen | Container | Specimen | Container |Specimen | Container Inconel Slight Slight 4 1 1/2 1 1/2 3 ' 3 3 roughening{ roughening Type-430 stain- <1 <1 1/2 1/2 /4 /4 Specimen | No attack less steel lost Type-304 stain- less steel <1 <1 2 11/2 2 2 11/2 1 Type-321 stain- : _ less steel <1 <1 <1 <1 1/2 1/2 11/2 11/2 103 ANP PROJECT QUARTERLY Fig. Stainless Steel, 51, Nickel-Plated Type-304 with Exposed Base Metal, Static Corrosion Tested in NaF-KF-LiF-UF4 for 160 hr at 816°C. established that electrochemical attack takes place, that if a metal is coated with a more noble metal and then the coating is brcken and exposes a relatively small area of the anodic material beneath the cathodic coating, the small exposed area will suffer from accelerated corrosion attack. It would appear that a similar process had taken place in this test with molten fluoride. Effect of Cold Work. Previous static corrosion tests in NaF-KF-LiF- UF, (10.9-43.5-44.5-1.1 mole %) on Inconel that had been cold worked approximately 20% prior to testing showed no i1ncrease 1n corrosion over as-received material, Tests have now been completed on Inconel with 52 and also 74% cold working prior to test,. There appeared to be no increase 1in corrosion even after this severe cold working. In these tests for 100 hr at 816°C both the specimens and tubes were attacked to a depth of approxi- mately 1 mil -~ the tubing was not cold worked, Corrosion of Ceramic Materials (C. R, Croft, N. V, Swmith, R, Meadows, H. J. Buttram, Materials Chemistry 104 PROGRESS REPORT A number of refractory were tested 1in Division). ceramic materials various fluoride mixtures. The main purpose of this study was to find container materials suitable for electrochemical studies. In addition, the behavior of beryllium oxide in contact with several fluoride mixtures was especially investigated because of the importance of this material as the moderator for the ARE, The time of exposure for these exploratery tests was reduced to 25 hr from the usual 100 hours. Comparison of results in NaF-KF-LiF-UF_ fuel and in the corresponding nonuranium-bearing mixture at 800°C showed that the attack is heavier in the fuel since the UF, appears to react with the oxides to form a coating of UO,. In both mixtures, vitrified beryllium oxide was found to be the most satisfactory material; 1t showed weight gains of 7% in the uranium-bearing mixtures and 1% in the uranium-free mixtures but The materials various grades no dimensional changes. tested included zircon, of aluminum oxide, wagnesium oxide, and sapphire, Also, a number of hot- pressed refractory carbides were tested in the uranium-bearing mixture. Silicon carbide was penetrated, but the carbides of titanium, tantalum, and columbium did not show visible signs of attack., The penetration of silicon carbide might have been due to the porosity of the sample. A special recrystallized aluminum oxide, Morganite, satisfactorily resisted attack by the NaF-KF-LiF mixture but gained over 90% when UF, was present because of the formation of a heavy layer of UO, on the surface. A similar deposit but a much lower weight gain (18%) occurred when a NaF-Ber-UF4 fuel was used, The resistance of vitrified beryllium oxide to uranium-free ZrF,-bearing mixtures was followed in a series of experiments using exposure times up to 500 hours, The weight gains ranged from 0.3% after 100 hr to over 11% after 500 hours. X-ray-diffraction studies showed a layer containing considerable amounts of ZrQ, on the surface of the specimen, whereas the zirconium content of the liguid decreased., It 1s possible that the ZrO2 layer will protect the body of material from further attack. Beryllium oxide specimens tested in a NaF-BeF, mixture at 800°C showed no apparent attack aon the surface; the specimens increased in weight by 0.7% after 100 hr, 1.5% after 250 hr, and 8.7% after 500 hours. STATIC CORROSION BY HYDROXIDES D. C. Vreeland ‘R. B. Day E. E. Hoffman L. D. Dyer Metallurgy Division Static screening tests have also been conducted on the effect of various additives on hydroxide corrosion. Although there was some indication that manganese addition had an in- hibiting effect on corrosive attack, the corrosion was still gquite severe, The nickel plating of hydroxide cor- rosion speclmens is not yet an entirely reliable technique of reducing hy- droxide corrosion, but it 1s most effective when the plating lavyer is thick, that is, much greater than 4 mils. * Effect of Additives. There was some indication in several tests that when Inconel is tested inNaOH additions of manganese may be beneficial inm inhibiting attack. TInconel, when tested 1n regular static corrosion tests with NaOH, is severely attacked; the entire thickness of a 35 mil FOR PERIOD ENDING JUNE 10, 1952 specimen is often affected. In order to check the possible inhibiting charactéristics of manganese, static corrosion tests were run with approxi- mately 4% and also 1 1/2% additions of manganese powder and a 1 1/2% addition of electrolytic manganese. The manganese additions did seem to have some inhibiting effect on corrosive attack, with the manganese powder additions being the most effective; however, corrosion was still quite severe even with these additions, A nickel-zirconium alloy (1/4% zirconium) was also tested in NaQH. This alloy has attracted attention because of its better physical proper- ties at high temperatures as compared with pure nickel, The results of the static corrosion test {110 hr, 816°C, vacuum) were encouraging. The only attack that could be noted was 1/2 mil of light intergranular penetration. Effect of Plating Metals. A series of tests in NaOH of some specimens of Inconel and Inconel X that were plated with different thicknesses of nickel by the International Nickel Company has been completed. All the specimens with platings under 4 mils were severely attacked, but several of the specimens with thicker platings were unattacked even though the nickel plating appeared to have many voids (Fig. 52). Nickel plating as a means of protecting materials against attack by hydroxides does not appear to he entirely dependable., When protection 1s afforded, 1t is complete and ex- cellent; but when this protection 1s not complete, attack 1s very severe. The plated specimens used were vacuum- diffusion heat treated before testing for 100 hr at 816°C under vacuum. A summary of metallographic observations 15 presented in Table 11. 105 ANP PROJECT QUARTERLY PROGRESS REPORT t UMCLASSIFIED § g Y-5743 (a) 2 mils of nickel plate on Inconel X. 75X, | UNCLASSI F{ED | v-.s747 L] :1_9* ‘l* (b) 12 mils of nickel plate on Inconel, 100X. rig. 52. Effect of Thickness of Nickel Plate on Static Corrosion by Sodium Hydrexide after 100 hr at 816°C. 106 FOR PERIOD ENDING JUNE 10, 1952 TABLE 11 gorrosion of Nickel-Plated Materials Tested in NaOH at 816°C for 100 Hours Base metal: Inconel THICKNESS OF NICKEL PLATE (mils) METALLOGRAPHIC NOTES 1 45 mils of oxide bet%een creased from 241 to 180 2 Most of specimen covered no attack on plating or plating and specimen; unattacked material de- mils with 10 to 35 mils of oxide on base material; base material along part of specimen 3 Corrosion product up to 50 mils between plating and specimen 4 Base material. unattacked; no thickness change; some voids apparent in nickel plating 8 No apparent attack on plating or base material; no thickness change 12 No apparent attack; some voids apparent in nickel plating 26 51 mils of oxide on base material; unattacked base material decreased frofi 241 to 182 mils; many voids in plating DYNAMIC CORROSION BY LIQUID METALS A. D. Brasunas L. S. Richardson Metallurgy Division Two tests on the dynamic corrosion of liquid metals were performed with the seesaw corrosion apparatus previ- ously described, (!} The tests were made at a 1500°F hot-zone temperature and a 950°F cold-zone temperature; the specimens were made of type-310 stain- less steel. One specimen contained lead and the other contained lead plus 3% sodium. After 187 hr (50,000 cycles) the specimens were sectioned for examination., Metallographic exami-~ nation of the specimens showed inter- granular corrosion at the hot zone and metal-crystal deposition at the cold (I)AircraftfluclearPropulsifin Project Quarterly Progress Report for Period Ending March 10, 1952, ORNL-1227, p. 120. zone (Fig. 53). Both corrosion 'and mass transfer were less severe in the test specimen containing 3% sodium. A similar test is being made with an appreciably greater addition of sodium. DYNAMIC CORROSION BY FLUORIDES Corrosion by Fluorides in Seesaw Tests (C. R. Croft, N. V. Smith, R. Meadows, H. J. Buttram, Materials Chemistry Division; A. D. Brasunas, .. S. Richardson, Metallurgy Division). All studies performed to date have justified the former conclusion that when carefully prepared fluoride preparations are tested in sealed capsules the corrosion on Inconel and stainless steel can be tolerated at temperatures above 1500°F. Recent experiments have also verified that when the fuel contains less than about 10 mole % UF, pretreatment of the 107 ANP PROJECT QUARTERLY PROGRESS REPORT N 1 TR, TR s N - D P UNCLASSIFIE (a) Lead with no sodium addition. 150X, . : 3 A UNCLASSIFIED E Y-6592 UNCLASS | F I Y.6639 . s o (b) Lead with 3% sodium addition. 250X. Cold End (950°F) Hot End (1500°F) Fig. 53. Effect of Sodium Additive on the Corrosion of Type-310 Stainless Steel by Lead After 187 hr at 1500°F in a Seesaw Test. 108 stainless steel or Inconel does not appreciably improve the corrosion behavior. : Addition of UO, to various fuel preparations has not resulted in in- creased corrosion by the mixtures; addition of hexavalent uranium as U,0, or as UOiFg, however, results 1in considerably increased corrosion. 1t still appears likely, therefore, that oxidation of the fuel components during preparation 1s more important than hydrolysis in so far as the corrosion behavior is concerned. Addition of sulfate, as Na2804, to various fuel samples has resulted in large increases The increase ln penetration and weight loss of the specimens seems to be pro- portional to the sulfate added. Tt does not appear likely, however, that the small gquantities of sulfate present in the commercial fluorides used can be responsible for the cor- rosion routinely observed. 1n COrTOS10N. The evidence to date justifies the conclusion that the ZrF,-bearing fuel mixtures are somewhat less corrosive to Inconel and stainless steel than any of the mixtures previously tested. Static testing has indicated that negligible corrosion results when these materials are tested in stainless steel and Inconel. Studies with the tilting furnace seem to indicate the superiority of Inconel to stainless steel as a container material for the zirconium-bearing fuels. These experi- ments are being continued and extended to recently developed fuel compositions as raptdly as possible, Seesaw tests with NaF-KF-LiF-UF, (10.9-43.5-44.5-1.1 mole %) fuel have been made with type-304 and type-304 ELC stainless steel for 190 hr with a hot-zone temperature of 715°C and a cold-zone temperature of 575°C. The tests revealed no difference 1in FOR PERIOD ENDING JUNE 10, 1952 behavior between these two specimens, and confirmed the theory that the carbon content plays an exceedingly small role, i1f any, in fluoride cor- rosion under these conditions. The hot zone revealed voids to a depth of 4 mils; whereas a film slightly less than 1 mil thick was observed on the cold-zone surface. Similar tests with Inconel have repeatedly shown about 5 mils of voids in the hot zone and a small amount of crystal deposition at the cold zone. Various additions are being made to the fuel in an effort to minimize or eliminate this attack. Thus far, three additions, namely, chromium, magnesium, and sodium, have met with success i1n single tests. These results should be confirmed by either static or additional dynamic tests. However, 1t 1s felt that chromium additions should be expected to be effective in suppressing void formation, since it has been demon- strated that the voids are caused by chromium depletion. Corrosion by Fluorides in Thermal Convection Loops. (G. M. Adamson, K. W. Reber, Metallurgy Division). The programof dynamic testing with thermal convection loops has received greater during the past quarter. zirconium fluoride emphasis Tests with new fuel have produced optimistic results. Although it appears that the fluoride fuel NaF-KF-LiF-UF, (10.9-43.5-44.5- 1.1 mole %) will not be used as the coolant, 1t is the only possible one now available in large quantities, so the preliminary work with it has been continued, Considerable corrosion has been measured with all container materials in which this fluoride was circulated. The same combination of fluorides, but without the uranium, 1is not nearly so corrosive, and the addition of small amounts of NaK reduces coOrrosion even more, The attack, at least in Inconel, appears to be a leaching out of the chromium 109 ANP PROJECT QUARTERLY PROGRESS REPORT in the metal and a gathering of the voilds, Two Inconel loops and one type-316 stainless steel loopwere operated with NabF-KF-ZrF,-UF, (4.8-50.1-41.3-3.8 mole %) at 1500°F. One Inconel loop and the type-316 stainless steel loop (No. 124) were cleaned by passing dry hydrogen through them at 1950°F, whereas the other Inconel loop was degreased. In the degreased loop the attack was very similar to that found with the NaF-KF-LiF-UF, fuel but only about half the number of pits was present (Fig. 54). The depth of pitting varied from 3 to 7 mils. In the hydrogen-cleaned loops the number of pits was reduced further and the attack was more nearly confined to a few grain boundaries; however, 1in a few cases the attack extended ‘to a depth of 13 mils (Fig. 54). The 1in- crease in depth of attack 1s thought to have been caused by the large grain size produced by the hydrogen faring, which would not be present with other cleaning methods. Very thin deposits were found i1in the cold legs of both loops. The other special coolant tested was NaF-BeF,-UF, (47.0-51.0-2.0 mole %)}). This mixture was circulated in two Inconel loops at 1500°F. Intergranular pitting that varied from heavy to moderate was found 1in the hot leg. The attack varied in depth from 6 to 13 mils. Thin deposits were present in the cold legs of both loops. Coolant samples from both loops showed slightly lower uranium concentration in all sections than reported for the original batches. This may be an indication that some uranium precipitated from the system, As mentioned above, since the NafF-KF-LiF-UF, mixture was the only fuel available 1in quantities, most of the work has been done with 1t. The data from the Jloops in which this 110 mixture has been circulated are tabu- lated in the following tabkles. The results from a few of these loops were listed in the last report, but they are repeated here for comparison. Table 12 summarizes the results from the Inconel loops, Table 13 those from the 300-series stainless steel loops, and Table 14 those from the remaining loops. A study of the data presented in these tables leads to the following conclusions: 1. Tf NaF-KF-LiF-UF, 1s used as the circulating media, all stainless steel loops plug in a relatively short time. The 400-series steel loops plug in a shorter time than the 300 series loops. 2. Inconel loops have not plugged when any of the fluorides have been con- circulated in them; however, siderable corrosion has been evident. 3. By comparing either Inconel or stainless steel loops in which NaF-KF- LiF has been circulated with those 1n which NaF-KF-LiF-UF, has been circu- lated, it is evident that the presently available UF, accelerates both plugging and corrosion. 4, When NaF-KF-LiF-UF, 1s circu- lated in a loop in which chromium 1s one of the alloying elements, the chromium concentration of the mixture increases and the iron concentration decreases., Little, 1f any, change takes place in the nickel and uranium concentrations., This 1s discussed further below. 5. When NaF-KI-LiF 1s circulated, no large changes are found 1in the concentration of the 1mpuraities. 6. A thin metallic layer 1s de- posited on the walls of the cold leg. This deposit is usually thickest 1n the top, or hotter, part of the cold leg. FOR PERIOD ENDING JUNE 10, 1952 .t ~ "" . . : o o . a-" ‘.‘.' -~ N l' . 7? * k "~ a C . » ~‘ o . e e e ¥ ‘ T et . T a o S « - - . . o e y - - . - » :: :: » . . ’q\ “-u‘__“ ' B,‘ . ' N o ‘_r.'f"' . /,r/ ‘ - - + ® » 1 G - « . \’*\:Jf;’,,’%‘. # 5 . . " : % ’ vt ¥ I 2 . * ) - e . " N -~ < ey - (b) Hot leg of hydrogen-flred loop. Fig. 54. corrosion of Inconel Thermal Convection Loop by NaF-KF-ZrF —UF After 100 hr at 1500°F, Showing Effect of Hydrogen Flring. Etched with aqua regia, 250X, 111 Il TABLE 12 Corrosion Data from Inconel Thermal Convection Loops Containing various Fluoride Mixtures TIME OF HOT {EG METALLOGRAPHIC NOTES 1 - . ,{EXO)p COOLANT* CIBCULATION RLASOE\‘ FOH TEMPERATURE CHEMTCAI. NOTES NG, (hr) TEPMINATION {°F) Hot Leg Cold Leg 73 | NaF-KF-1if 1000 Scheduled 1500 fayer of pits, 5 mils; intergranular No attack; possibly a very thin deposit Cr increased, Fe decreased; Cr slightly attack, 10 mils; considerable grain higher in cold leg growth 214 | NaF-KF-LiF + 500 Scheduled 1500 Slight intergranular attack, ! to 3 mils Penetration, | mil; no metal deposition All impurities low; Ni slightly higher Nak than normal 219 § NaF-KF-LiF-UF, 500 Scheduled 1500 layer of pits, 10 mils, with maximem of Slight roughening Large increase in Cr, Fe decreased, Ni 15 mils; possible surface layer, 2 mils variable; Cr higher in cold leg 21} | NaF-KF-LiF-UF, 524 Scheduled 1500 Intergranular pitting attack, 4 to 8 Metallic deposit, 0.5 mil; no attack Cr increased, Fe decreased; even distri- mils; some pitting in grains bution 212 NaF-KF-LiF—UF‘ 37 leak 1500 Intermittent layer of intergranular No atrack Cr, Fe, and Ni all show large increases pits, 5 mils in both hot and cold leg 213 | NaF-Kf-LiF-UF, 500 Scheduled 1500 Intergranular pitting, 2 to 11 mils; Thin, continuous metallic iayer with an Large increase in Cr, Fe decreased; no some pitting in grains intermittent naonmetallic layer on it systematic distribution; H, fired, H, atmosphere 218 | NaF-KF-LiF-UF, 500 Scheduled 1309 Intergranuiar pitting, 6 to 10 mils Metailic deposit, 0.1 to 1 mil; appeared Cr increased, Fe decreased, %I decreased, to be two layers; the second layer was ne systematic distributinn not continucus and contained inclusions 219 | NaF-KF-Lif-UF, 480 Heater 1560 Heavy pitting, 5 to 13 mils Wall rough with thin layer; bottom layer | lLarge increase in Cr, Fe decreased; cold failure metallic with discontinuous nonmetallic leg high in Fe and Cr layer (particles) on top 216 NaF-BeF2-UF‘ 574 Scheduled 1500 Heavy to moderate intergranular attack, No uttack; some nonmetallic particles on Large increase in Cr, Fe decreased, siight 6 to 13 mils; an hot herizonta! leg wall decrease in U there was attack only on one side 217 NBF-BEF!-UF‘ 500 Scheduled 15490 Heavy to moderate intergranular attack No attack; light metal deposit as layer Increase in Cr, decrease in Fe and U up to 13 mils and crystals 220 | NaF-KF-LiF-UF, 2 Scheduled 1500 Light pitting, 2 1o 4.5 mils Scattered crystal deposit Cr decreased slightly in cold leg and considerably in fiot leg, Fe decreased slightly in both legs 221 } NaF-KF-ZrF -UfF 500 Scheduled 1500 f.ight to moderate intergranular pitting, Rough surface; thin nonmetallic layer 4 4 f g P 3 te 7 mils 222 | Ma¥-KF-Lif-UF, 500 Scheduled 1650 228 NnF—KF-ZrF.-UF‘ 500 Scheduled 1500 Light intergranular pitting, 3 to 13 Rough surface; thin deposited layer mils + - - Composition of coolants: mole %. NaF-KF-LiF - 11.5-42.0-46.5 mole %; NaF-KF-LiF-UF, - 10.9-43.5-44.5-1.1 mole %; Naf-Bef,-UF, - 47.0-51.0-2,0 mole % NaF-KF-Zr¥ -UF, - 4,6-50,1-41.3-3.8 LHOdIY SSHYI0Ud X THALHUYNO I2af0odd dNV €11 Corrosion Data TABLE 13 from Stainless Steel Thermal Convection Loops Containing vVarious Fluoride Mixtures TYPE OF TIME OF | peasow pop | HOT LEG METALLOGRAPHIC NOTES LOOF | STAINLESS coDLANT(2) | CIRCULATION! pppuinarion | TRNPERATURE CHEMICAL NOTES * STEEL (hr) {°F} Hot Leg Cold Leg 111 316 NaF-KF-LifF i74 Leak 1508 Intergranuiar atteck, 10 mils; some wall No attack; continuous layer, | mil; some Large increase in (r, Fe variable; gas reduction; tremendous grain growth grain growch leak around spark plug 116 316 NaF-KF-LiF 500 Schedulegd 13906 Some intergrenular pitting, 2 to 4 mils; Thin, continuons metaliic deposit with in- No impurities increased; no =zystematic surface layer shows grain growth clusions variations 119 316 NaF-KF-LiF + NaK 500 Scheduled 1500 Surface rough with some deprassions, 2 No attack or deposit All impurities very low - much lower o . wily; no pitting or intergranulsr attack than at start 112 318 NaF-KF-LiF-UF, 82 Plug 1500 Intergranular attack, 8 wils; grain growth Intermittent thin layer; grain growth Large increcase in Cr, Fe decreased 113 316 NaF-KF—LiF'UF4 123 Plug 1500(6) lntergranular attack, 8 mils; some pitiing | Thin deposited layer Large increass in Cr, decrease ia Fe but irregular, {§ higher in cold leg 118 316 MaF-KF-LiF-UF, 147 Plug 1509 Intergranuiar attack, 5 te 12 mils; sowe Rough surface with 1 mil layver Cr increased, Fe decreesed, highest in grains removed cold horizontal leg; Hz ficed, #, atmosphere 121 3t NaF-KF-LiF*UF‘ 153 Plog 1360 Heavy intergranular attack, # to 10 mils; Duposited laver, 0.5 to 1.5 ail; irregular Increase in Cr, decrease in Fe; no some grains removed in thickness systematic distribuvion 275 347 MNaF-KF-LiF-UF, 39 Leak 1500 Ynnergranuiar attack, 8 to 13 fiils;. Metallic dcfiosit, 0.5 to 1 mil; contains " 'Cr increased with no large varisvioh; tremendous grain growth inclusionsg Fe decreased irregunlarly 274 347 NuF-KF-LiF-UF, 123 Plug 15008 Intergranular attack, 2 to 4 mils Bough surface with = thin depesited layer Cr increassd, Fe decrcased; bath slightly higher in cold leg 251 316 NaF-KF-LiF-UF, 75 Plug 1500 Yery heavy intergranular pitting, 8 to 15 General nonmetallic deposit with thin Cr increased, Fe decrsased; cold leg mils; surface rough, trewendous grain metallic layer; some crystals attached to siightly higher grawth surface 252 310 NaF-KF-LiF-UF, 368 Plug and 1548 Very heavy intergranular pitting, up to Thin deposited laver; large crvstals Chemical snalyses irregular and no instrument 20 mils ettached to surface definite trends were discernabla; failure - - : . metnllic plug found in lower cald lag 127 116 NaF-KF-LiF-UF‘ 87 Teak 17490 Heavy intergranular attack, 3 to 7 mile; Metallic layer 2 mils thick sSome grains removed 1232 316 NaF-KF-LiF-UF, 500 Scheduled 1650 40 418 Raf-KF-Lif-UF, 9 Plog 1500 No pitting or intergranular attmck; up to Meraltic deposit with inclusions; slight Cr increased, Fe decreased; batk 10 mils removed; considerable martensitic attack; Erain growth higher in cold leg structure 43 410 NaF-KF-LiF-UF, 12 Plug 156G No pitting or usual intergranular attack; Metallic deposit with inclusions; grain Cr increased in cold leg but only however, many grains are }ooee; con- growth slightly in hov leg, Fe decrsused but siderable martemsitic structure varivd; wetullic crystals found in cold leg a3 430 Nab-KF-LiF-UF, 8 Plag 1500 Heavily pitted, 2 mils; general rTemoval of § Merallic deposit with inclusisns; surface Cr increased, Fe decressed; both higher about 2 mils rough under the deposit in hot leg 14 43¢ NaF-BF-LiF-UF, 9 Plug and 1500 lrak (G)CompositiOhs of conlants: NaF-KF-LiF — 11.5-42.0-46.5 mole %; NaF-K¥-LiF-UF, - 10.9-43.5-44,5-1.1 mole %. {b}Operaced far 72 hr at 1500°C and at 1600°C for remaining time, ‘0T ANNT ONIANA AoXdiad 404 2861 PIT TABLE 14 Corrosion Data From Miscellaneous Thermal Convection Loops Containing NaF-KF-LiF-UF4 Composition of NaF-LiF-UFq: 100 9‘43- 5'44- 5'1| l mOle % TIME OF REASON HOT LEG METALLOGRAPHIC NOTES LOOP | MATERIAL | CIRCULATION FOR TEMPERATURE CHEMICAL NOTES 0. {hr) TERMINATION (°F) Hot Leg Cold Leg 104 Nickel 560 Scheduled 1500 Surface polished: no No attack; layer of Fe and Cr unchanged, Ni pitting or inter- crystals, 1 to 3 mils higher in cold leg; some granular attack; up to evidence of plugging; 9 mils removed; some crystals found in all Erain growth sections 107 Nickel 1690 Scheduled 1500 Surface polished; no Crystal deposit intergranular attack or pitting; from § to 10 mils removed; large gTains 340 Monel 117 Leak 1500 No pitting or jpter- No attack; no deposit No appreciable change in granular attack impurities 365 Nimonic 500 Scheduled 1500 Intergranular pitting, | No attack; thin de- Cr increased, Fe decreased, B to 13 mils, a few j posited layer siight increase in Ni | spots up to 15 mils ; | | 341 Monel 31 Leak J 1500 Ne intergranular attackj No attack; no deposit or pitting; up tc 14 Q ! mils removed : | | 44 Iron 25 Plug ; 1350 Rough surface with no g Rough surface with no E i i large attack evident E attack or deposit 5 | | | | : | ‘ 375 | Hastelloy 80 Plug ‘ 1500 | | | B | I i | LYOdIY SSIY908d X THALYVNO 1DAL0¥d JNV 7. The addition of NaK to NaF-KF- LiF reduces corrosion in both Inconel and type-316 stainless steel and suppresses formation of the metallic- layer deposit. 8. The mechanism of plugging in the loops has not been determined. Definite crystal masses were found during sectioning and melting out of the fluorides in only three loops (Nos. 43, 252, end 104; see Tables 13 and 14). In the other loops either no metallic crystals or only a few scattered ones were found. The chemistry group has checked melting points and viscosities of both hot- and cold-leg material without finding changes from the original values. FOR PERTOD ENDING JUNE 10, 1952 9, The attack is about the same at 1300°F as it 1s at 1500°F. As an additional step in the under- standing of corrosion in Inconel loops, successive samples were drilled from the inside of a section of hot-leg pipe and analyzed. The results are reported in Table 15. Since the pipe from which these samples were drilled was not perfectly round, the figures in Table 15 should only be used for discerning trends. The poor sampling is one reason for the apparent presence of fluorides at considerable depths. In spite of the discrepancies 1t 1is obvious that chromiumwas being removed from the wall and that iron and nickel were remaining essentially unchanged. TABLE 15 Chemical Analyses of Subsurface Layers from an Inconel Thermal (onvection Loop in Which NaF-KF-LiF-UF, Was Circulated CEPTH OF CUT COMPOSITION (%) | (in.) Fe Ni cr | Fe/Ni | KF(® | Liv(®) | Nar(®) | UF,(?) | Total(® 0.002 7.33 80.84 8.74|0.091 | 0.45 0.55 0.18 NDCE) 98.09 0.005 7. 42 80. 42 8.73 | 0.092 | 1.50 0.55 0.36 ND 98.98: 0.010 7.46 79.60 9. 39 0.094 | 1.20 0.55 0.36 ND 93,56: 0.015 7.60 79.98 | 10.63| 0.095 | ©.30 0.30 0.18 ND 98.99 0.020 T.48 79.88 | 11.08| 0.095 | 0.30 0. 30 0.18 ND 99.22: 0.025 7.15 77.57 | 13.18| 0.092 | 0.15 0.15 0.07 ND 98. 27 0.030 7.20 76.90 | 14.15 | 0.094 | 0.15 0.15 0.07 'ND 98.62‘ Exterior 7.08 | 76.46 | 15.850.093 99. 39 Inconel specifications| 7 77 15 0.091 99 (a) that all the wetal was present as the fluoride. (b) would add am ‘additiomal 0.6% to 0.75% to the total. (G)ND = none detected. These values were calculated from the spectrographic analyses for the metal with the essumption In addition to the items repofted, the percentages of cobalt, copper, silicen, and manganese 115 ANP PROJECT QUARTERLY PROGRESS REPORT The chromium was being removed from depths that appear by metallographic examination to be unattacked. As a further study of this attack, the metallographic section was asked to determine whether the apparently unconnected spherical holes ohserved actually connected to the surface. The section was examined and photographed while successive 1-mil layers were being removed. By examining the photographs and trying to follow the holes, 1t was determined that they did not connect to the surface, Corrosion in Rotating Dynamic Test gig (W. C. Tunnell, ANP Division). As mentioned in the last quarterly re- port, {1 the rotating dynamic test rig 1s an attempt to determine corrosion effects more conveniently than 1m the thermal convection loops. The apparatus was calibrated for temperature measure- ments with lead and appeared to be accurate within 5°F. The rig was charged with NaF-KF-LiF (11.5-42.0- A6 .5 mole %)} contalned 1in nickel liners. The mixture was held below the melting temperature (550°F) and under S5-micron pressure for several days to dehydrate the fluoride, Following this pretreatment the tem- perature was raised to 1500°F and held for one day. Raising the temper- ature did not affect the vacuum, so 1t was assumed that the material was dry, The rigs were then allowed to cool and the stuffing-box seal assembly was installed with the Inconel specimens. The temperature was then raised to 1500°F and the specimens were rotated for 100 hours. The Inconel specimens were examined metal lographically and the fuel mixture was subjected to chemical analysis, The fluoride mixture before rotation showed 1340 ppm of Ni, 35 ppm of Cr, and 170 ppm of Fe, but after rotation 116 one flvoride sample showed less than 20 ppm of Ni, 160 ppm of Cr, and 75 ppm of Fe and the other sample showed less than 20 ppm of Ni, 45 ppm of Cr, and 80 ppm of Fe, Photomicrographs of the Inconel have not yet been made, but preliminary examination on the metallograph does not show an appreci- able change. DYNAMIC CORROSION BY HYDROXIDES Hydroxides in Seesaw L. S. Richardson, Corrosion bhy Tests (A.D. Brasunas, Metallurgy Division). Since oxygen, as Ni0 or as 0,, added to NaOH has resulted in greatly accelerated mass transfer of nickel in nickel tubes, a series of oxygen getters have been added to seesaw tests. These have 1n- cluded Na, NaH, Be, Mg, Al, and Zr, None has shown beneficial results with the possible exception of Zr, which has given very erratic results, A summary of the seesaw test results 1s given in Table 16, The temperatures listed in Table 16 are not the liquid temperatures but merely the tempera- tures measured with a thermocouple spot welded to the metal tube. Pre- liminary experiments show that the temperature gradient is probably an error by 100°C. The data in Table 16, 1in con- junction with previous experimental work, lead to several general con- clusions, 1. Mass transfer is accelerated by the presence of relatively large guantities of oxygen. 2. Apparently there 1s a low temperature limit belcw which little or no mass transfer occurs. 3. Additions to the NaOH have little or no beneficial effect on mass transfer. L11 TABLE 16 Results of various Test Conditions on Mass Transfer of Nickel in Nickel Tubes Containing NaOH TEST CONDITIONS AMOUNT OF Ni DEPOSITED : Hot-Zone Cold-Zone Time No. of Environment Temperature Temperature AT COLD ZONE (hr) Cycles (°C) {(°C) 117 14,000 Vacuum 797 505 Light to moderate 100 12,000 Yacuum 600 371 Practically none 125 45,000 Vacuum 645 475 Light 72 19,600 Oxygen atmosphere 750 540 Very heavy 147 39,500 Hydrogen atmosphere 740 520 Moderate 147 39, 500 Hydrogen atmosphere 765 515 Moderate 259 69,000 Argon 715 560 Light to moderate 8 2, 000 Vacuum + 5% Na 715 5G0 Heavy 8 2,000 YVacuum + 10% NaH 750 500 Heavy 0.2 56 Vacuum + 30% NaH 750 300 Failed immediately 102 27,000 Vacuum + 2% Nafi 700 555 None 142 38,000 Vacuum + 1/4% Zr02 720 540 Very light 104 28,000 Vacuum + 1/4% Al 750 660 Light to moderate 104 28,000 Vacuum + 1/4% B 750 665 Light 104 28, 000 Vecuum + 1/4% C 170 660 Moderate 104 28, 000 Vacuum + 1/4% Mrn 770 650 Very heavy 168 45,000 Vacuum + 1/4% Be 730 575 Heavy 114 31,000 Vacuum + 1/4% Ag 710 360 VYery heavy 114 31,000 Vacuum + 1/4% Li 790 635 Moderate ‘0T aNAl INIGNA aolddd 404 cS61l ANP PROJECT QUARTERLY PROGRESS REPORT Standpipe Tests of Hydroxide Corrosion (A. D. Brasunas, L. S. Richardson, Metallurgy Division). A standpipe test run with A nickel and NaOH vyielded deposition at a 700°C zone surrounded by higher temperature zones and solution at another 700°C zone surrounded by lower temperature zones. These results indicate that, as observed many times, mass transfer can occur 1n a so-called "static'™ system 1f a temperature gradient or temperature cvycle 1s present. Hydroxide Corrosion in Cold-Finger Thermal Convection Apparatus, (J. V. Cathcart, W, H. Bridges, G. P. Smith, Metallurgy Division), A new type of thermal convection apparatus, desig- nated as theicold -finger" apparatus, beern developed for use 1n the corrosion testing of hydroxides. In principle, the cold-finger apparatus consists of a small-diameter tube that 1s sealed off at one end and inserted has into a pot containing molten hydroxide, The inside of the small tube is cooled by a jet of air to create a tempera- ture differential between the inner tube, or cold-finger, and the walls of the hydroxide pot. As a result of the temperature difference convection currents are set up in the hydroxide. This new apparatus has several ad- vantages over the thermal convection loops commonly used research: in corrosion (1) few welds are required 1n i1ts comnstruction; (2) it easily operated and repaired; and (3) it has a relatively large area of contact between hydroxide and blanket- ing atmosphere. It is probable that the effective rate of flow in a thermal convection loop 1s greater than that in the new apparatus, but the rate of flow in the cold-finger apparatus is sufficiently large to cause copious mass transfer under conditions that were knownm to produce mass transfer in a loop, 1S more 118 With the apparatus completely assembled except for the gquartz jacket, the hydroxide pot was filled with c.p.- grade NaOH pellets, which were then dehydrated at 500°C under a vacuum of approximately 0.04 wm Hg for 48 hours. The temperature was then increased until the outside of the hydroxide pot reached the desired temperature. At the same time the air jet was turned on and regulated so that the cold finger remained at a temperature of approximately 500°C. In the first of two runs in the apparatus the hydroxide was blanketed with a hydrogen atmosphere. The pot walls and the cold finger were main- tained at temperatures of 625 and 525°C, respectively, for a period of 117 hours., Neither loose dendritic crystals nor mass transfer of any sort was noted during this run, The second run was carried out under a vacuum of about 0,04 mm Hg, and the temperature was adjusted so that the pot walls and the cold finger were at 725 and 500°C, respectively. Heavy mass transfer occurred between the walls of the dehydration pot and the cold finger. Large quantities of dentritic crystals were observed on the surface of the cold finger. It was shown that at least under the conditions specified for the first run, a blanketing atmosphere of hy- drogen effectively stopped mass transfer in a nickel-sodium hydroxade system, Modified Thermal Comvection Appa- ratus - TCA (J. V. Cathcart, W. H. Bridges, G. P. Smith, Metallurgy Division). A second, modified, thermal convection apparatus, representing a simplification of both the thermal convection loop and the cold-finger apparatus, has been constructed and given preliminary testing. Because of its simplicity, the modified apparatus is easy to erect, requires a small amount of laboratory space, and can be put into operation in a short period of time. The apparatus 1is designated as the TCA, or thermal convection apparatus. Jt consists of a 1-ft length of l-in.-dia nickel tubing with a 1/8-in. plate welded on the bottom. A Kovar seal is attached to the top end. A nickel vane about 3 in. long 1s posationed inside the tube near the bottom. There i1s suf- ficient space below the vane to permit circulation. The nickel tube 1is placed between 4-in., split-core, semicircular heaters and the system is connected to hydrogen and vacuum lines. The desired blanketing atmosphere is then introduced, and one of the semi- circular heaters 1s removed to expose one side of the apparatus to the atmosphere. The power in the other heater may be increased until the wall temperature on the hot side of the apparatus reaches 800°C. This procedure results in the formation of a temperature differential between opposite sides of the vtube and sets up convection currents in the hydroxide. In the two runs to date with sodium hydroxide the walls in the hot zone in the vicinity of the heater were highly polished whereas those in the cold zone had a roughened appearance. Massive deposits of crystals were found only at the hydroxide-vacuum interface. On the basis of these results it is felt that the TCA represents an apparatus, which, because of its simplicity and ease of operation, might be well suited to preliminary corrosion testing in new systems. 1t is planned to continue the development of the equipment, FUNDAMENTAL CORROSION RESEARCH In addition to the program of empirically testing corrosion of structural metals by fused salts, considerable effort has been devoted FOR PERIOD ENDING JUNE 10, 1952 to a number of studies designed to assist in the discovery of the cor- rosion mechanism. These studies have included careful examination by physical and chemical means of cor- rosion products formed during large- scale cyclic corrosion tests, exami- nation of reactions of structural metals with high-temperature liquids in sample systems, and studies of possible reactions of molten fluorides and hydroxides under applied po- tentials., A considerable program of preparation of complex fluorides of the structural metals has been con- ducted to assist in the identification of corrosion products. Preliminary results have been obtained by several independent groups, but the corrosion phenomena and mechanisms proposed are not vet entirely consistent. Interaction of Fluorides and Structural Metals (H. Powers, J. D. Redman, .. G. Overholser, Materials Chemistry Division). The interaction of fluorides and structural metals at high temperatures is being studied 1in an attempt to define the mechanism of attack on containers and to assist 1n development of measures to control the corrosion observed. In general the studies have been concerned with the NaF-KF-LiF eutectice, with and without UF,. This fluoride mixture is heated in contact with type-316 stainless steel at 800°C for about 6 hr under an inert atmosphere and then cooled te room temperature. After representative samples are taken for analysis, the remainder 1s re- heated under an 1inert atmosphere, and samples are removed by application of vacuum to the filter in which the porous diaphragm 1s sintered nickel or a thin membrane of graphite, As the data in Table 17 indicate, the reaction of the NaF-KF-LiF mixture under these conditions tends to solubilize some iron but virtually no 119 ANP PROJECT QUARTERLY PROGRESS REPORT TABLE 17 Reaction of LiF-NaF-KF-UF4 Mixtures with Stainless Steel at 800°C STRUCTURAL METAL IN LIQUID (ppm) UF, CONTENT Before Filtration © After Filtration (mole %) T T Re T Cr Ni Fe cr | N 0 4600 1600 600 3500 20 20 0 6 400 1200 470 2600 20 30 0 6300 1500 500 2400 20 30 0 8500 2200 1300 2300 20 30 2 8300 2000 800 7400 1700 20 2 9200 2400 1600 7100 500 700 2 8100 2400 500 6000 1400 100 2 8600 2000 1100 7000 1600 180 chromium or nickel, Although the iron passing the filters shows con- siderable variation from test to test, at least 2000 ppm are solubilized by this technique. Tt is significant to note that the use of nickel filters (Micro-Metallic Corporation), which are considerably more porous than the graphite, does not change the values for soluble iron and chromium com- pounds; also, when nickel filters are used, the nickel content of the filtrate rises to about 300 ppm., In general the agreement of results from the two filters indicates that the material found in the filtrate is dis- solved rather than suspended and that reaction with the filter is probably not significant in so far as and chromium are concerned. iron When UF, 1s a constituent of the mixture, however, significant changes are apparent. The iron content of the filtrates rises considerably (2 to 3 fold) and the chromium obviously becomes solubilized to a large extent. The nickel content of the filtrates 120 also rose significantly, except in one case (as yet unexplained) in which the soluble nickel low level. remained at a These studies have not yet yielded sufficient concentrations of the soluble corrosion products for identi- fication by x-ray diffraction or other means, However, 1t 1s believed, in view of data presented below, that the soluble materials are complex fluorides, Similar and somewhat more detailed studies are planned for the more complex NaF-ZrF,-UF, systems in the near future. Synthesis of Complex Fluorides (B. J. Sturm, L. G. Overholser, Materials Chemistry Division). Analyses of the various fluoride fuels after static or dynamic corrosion testing have shown that significant amounts of the container (chiefly iron, nickel, and chromium from Inconel or stainless steels) 1nteract with the fused fluorides. Tdentification of the chemical combinations in which these g structural elements occur 1is an important part of the study of the reaction mechanism. It was assumed that the iron, nickel, and chromium are present as simple or complex flvorides 1in the mixture of alkali fluorides; consequently, the prepa- ration of various fluorides containing iron, nickel, or chromium was under- takeno The dehydration of most fluorides can be accomplished without hydrolysis only if an atmosphere of HF is main- tained over the material. Dehydration of KF, NaF, and LiF by'heating under vacuum or inert gas results in hy- drolysis of the alkali fluorides; the production of a free base may easily be detected by measuring the pH of a solution of the dehydrated alkali fluoride. Melting of hydrated nickelous and chromic fluorides results 1in extensive hydrolysis unless the ma- terials are kept under an atmosphere of HF while heating, Ammonium bi- fluoride may be added as an alternate to provide the protective atmosphere, Anhyvdrous nickelous, chromic, and manganic fluorides have been prepared from the commercially available fluorides., Ferrous fluoride has been prepared by the decomposition of NH,FeF, at 800°C; the ferric fluoride results from decowposition at a lower temperature, The NH,FeF, may be prepared by reacting anhydrous FeCl with NH,HF, at 350°C. Compounds corresponding to K3CrF6, Na,CrF,, and (NH,),CrF, have been prepared by heating the proper molar gquantities of the corresponding acid fluorides with chromic fluoride. Samples of Li CrF, were prepared by heating LiF with (NH4)3CrF6. NaK, CrF, has been prepared by fusing a mixture of fluorides corresponding to this composition, Attempts to prepare NaQKCrF6 by a similar method yielded NaK,CrF, contaminated with excess NaF FOR PERIOD ENDING JUNE 10, 1952 and some unidentified chromium com- pound, Preparations of complex fluorides of iron and nickel with the alkali fluorides are under way. Examination of Corrosion Products (D. C. Hoffman, F. F. Blankenship, Materials Chemistry Division). Loop tests in which ZrF, -bearing fuel mixtures have been circulated by thermal convection have not shown any restriction to flow in tests up to 500 hr, however, stainless steel loops in which NaF-KF-LiF-UF, mixtures have been circulated nearly always plugged after intervals of about 100 hours. During the past quarter a concerted effort has been made to determine the nature and location of these plugs. The loop sections to be examined were selected, 1n general, to contain the portion indicated by radiographs to be the most dense. The apparent density differences, however, seem to be a result of contraction and re- arrangement of the melt on cooling and do not necessarily reveal the presence of a metallic phase, The presence of large metallic deposits could be deduced from the radiogram in the few cases where they have been shown to exist, but this technique was com- pletely unreliable for the more finely dispersed metallic erystals. Exami- nation of representative sections throughout the loops demonstrated that the trap and the vertical cold leg contained most of the deposited metal if there were any,. One section from the hot leg, two sections from the cold leg, the trap, and any additional sections showing radiographic deposits were customarily examined., Material was removed from the sections, examined under the microscope, and separated by means of a magnet before and after selective leaching with aluminum nitrate or ammonium oxalate solutions, The magnetic fractions and any others that 121 ANP PROJECT QUARTERLY PROGRESS REPORT could be separated by the selective leaching procedures were submitted for examination by chemical microscopy, x-ray diffraction, and spectrographic and chemical analyses. Locops in which NaF-KF-LiF was circulated in type-316 stainless steel and loops in which NaF-KF-LiF eutectic plus 2 mole % UF, was circulated in types-310, -316, -347, -410, and -430 stainless steel, nickel, Inconel, and Nimonic were examined, In addition, one Inconel loop in which NaF-KF-Zr¥F, - UF, was circulated has been examined. Metal "plugs" in quantity sufficient to stop the flow were observed 1in loops of types-410, -430, and possibly, -310 stainless steel. In all other cases, regardless of whether the loops had plugged or not, only scattered metal crystals, if any, were found, therefore it does not seem possible that physical plugging of the tubes by these small quantities of material could have occurred. In the loops of types-316, -347, -410, and -430 stainless steel the metallic phase was highly magnetic and all examinations indicated that the material was nearly pure iron. The deposit in the type-310 stainless steel loop was only weakly magnetic and showed a high chromium content: the x-ray-diffraction pattern indi- cated the presence of a noncubic crystal i1n addition to a material that was presumably a chrome-iron with a shifted iron lattice. No evidence of deposits of metallic nickel has been found in loops of any of these alloys, although in nickel loops copious quantities of nickel crystals were obtained. Unidentified sulfides often seem to concentrate in the vicinity of the metallic crystals, as revealed by the evolution of H,S during leaching with acidified aluminum nitrate solution, 122 A green, crystalline substance that was sparingly soluble in aluminum nitrate solutions was found associated with UO, and the metallic crystals in many of the loops examined. In addition, layers of the same material were observed 1in containers in which NaF-KF-LiF-UF, fuels were prepared and stored. Spectrographiec analysis of samples from these sources indicated the presence of Na, K, and Cr but virtually no iron. X-ray-diffraction studies by Agron and microscopic examination by McVay indicated the material to be a cubic erystal with an index of refraction close to 1.422. By comparison with complex fluorides prepared by Sturm and Overholser, the material has been positively identified as K,NaCrF, . This material, which appears quite soluble in the molten fuels at temperatures above 1000°F, seems to be the major product of attack on chromium by the fluorides, Since the chromium in this compound 1s trivalent, whereas Cr*? would be expected from the reaction Cr® + 2U%*3 —— 9u*3 + (Crt? it appears likely that reduction of U*? does not explain the occurrence of this compound. Examination of the Inconel loop, which circulated NaF-KF¥-ZrF,-UF, for 500 hr with a hot-leg temperature of 1500°F and showed no signs of plugging, revealed only very slight guantities of metal deposition in the trap. These metal crystals, which have not vet been identified, grew to a length of about 1/8 in. in a band perhaps 3/16 in, wide, No other metallic deposits or crystals were observed. No H,S odor was detectable on leaching the material from this loop with acidified aluminum nitrate solution. Since the fuel mixture used was treated with H, and HF prior to transfer to the loop, the sulfur should have been at a low level prior to the run, ot Chromium has been shown by other groups on the project to be selectively removed from Inconel and stainless steel containers by the molten fluo- rides; however, chromium was rarely found in metallic deposits from these experiments. Quantitative analysis generally revealed the presence of large amounts (about 3000 ppm) of chromium in the fluoride melts from the loop tests; the chromium was generally uniformly distributed in both hot and cold legs of the loops in a manner such as to suggest that it was dissolved in some fashion. Chromium removal 1is still being studied, but it 1s suggested that it may be explained by the reaction 3UF, + 2Cr,0,——> 300, + 4CrF, . If thais then all active oxygen 1in the system would eventually be converted to UO, at the expense of solubilizatioen of the chromium. The fact that U0, 1is a constant component of material from corrosion tests, whereas UF3 has not is the reaction, been positively identified, would seemto lend support to this hypothesis, It 1s certain that difficulty from the reaction 2UF, + Cr —> 2UF, + CrF, would be experienced. It seems, how- ever, that a large portion of the cor- rosion presently observed may be due to the other mechanism and that elimination of oxides and active oxygen will remove a major portion of this trouble. X-Ray-Diffraction Studies (P. Agron, Materials Chemistry Division)., Stand- ard, x-ray-spectrometer, diffraction patterns were obtained for Zr¥F,, ZrOF,, ZrOz; UF,, UF,, and UD, to facilitate their identification in FOR PERTIOD ENDING JUNE 10, 1952 fuels before and after corrosion tests., The patterns of FeF,, FelF,, FeO, Cr¥F,, and Cr203 were also con- sidered i1in the examination of test samples. The indexed lines of known com- plexes, such as K3FeF6, Na, FeF_, and KNiF,, were compared with unknown lines found in corrosion studies. None of the latter materials were identified. However, synthesis of the two new compounds K,CrF, and K,NaCrF, (prepared by B. J. Sturm, Materials Chemistry Division) showed that the latter double-alkali complex was a residue commonly found in corrosion tests with the NaF-KF-LiF-UF, mixture. Table 18 lists the crystal structure of the complexes discussed above. Examinations of synthesized com- pounds of iron and chromium complexed with sodium and lithium fluorides are in progress. An attempt is being made to synthesize K,NaFeF. X-ray-diffraction patterns have been obtained for several fuel com- positions and two uranium-free eutectic mixtures. Further study of these TABLE 18 Complex Fluoride Compounds Formed by Fluoride-Container Reactions _ LATTICE compounp{ @) DIMENSION, | DENSITY (g/ce) a (&) D, D, KNiF, 4.02 K FeF_ 8.62 2.96 K,CrF ) 8.53 2.99 (b K, NaCrF (%) 8.27 Na FeF, 7.95 (2)Fach of these compounds had a cubiccrystal form. (b)Isomorphous with KaFeFG. 123 ANP PROJECT QUARTERLY PROGRESS REPORT materials will be required to establish whether solid complexes and/or solid solutions occur, EMF Measurements in Fused Fluorides (L. E. Topol, L. G. Overholser, Ma- terials Chemistry Division). Research on the electrochemistry of fused salts has been undertaken in order to learn more regarding the fundamental aspects of corrosion. The current approach to the problem involves the decomposition-potential measurement of various fluorides (pure and 1n tures). From these decomposition voltages and the egquation mix- E; = Egy ~——1Ina, some knowledge should be obtained about the activities of the components in these mixtures. In addition, a relative, chemical-activity series of metals 1n fused fluorides can be formed analogous to that in agqueous solutions, This series can be compared to one determined from emf measurements if a suitable reference electrode can be found. A cell assembly employing nickel as the reference electrode, of the type TABLE 13 Reversible Decomposition Voltages of various Fluorides DECOMPOSITION VOLTAGE (v) AT FLUORIDE VARIOUS TEMPERATURES 298, 1°K | 500°K | 1000°K | 1500°K NaF 5.60 5.39 4.86 4.45 KF 5.52 5.28 4.72 4.31 LiF 6.05 5.88 5. 40 4,95 Several electrolyses have been made with KF at 885°C by using nickel electrodes and a helium atmosphere, In all cases there was a definite break in the voltage vs. current curve at about 1.2 volts, Two different crucibles have been used as contalners. Morganite, a special recrystallized aluminum oxide, did not hold up at all and virtually disintegrated after a few hours of contact with the molten salt. How- ever, the Norton alumina RA 7232 crucible, which had a much greater wall thickness, survived contact with the molten salt fairly well, although Ni | Nin(cl) in a fluoride, or fluoride mixture | MF,(c,) in same fluoride composition | M, where the concentrations ¢, and c, are in mole % and are approximately egual and M = Cr, Fe, etc., would vield the desired results. Finally, from the temperature coefficients of the measured potentials, i1f reversible, useful thermodynamic information may be obtained. (The accuracy of these data depends on the current efficiency and nature of the electrodes used.) From known free energy data (Quill’s Thermodynamics), the reversible de- composition voltages of NaF, KF, and LiF have been calculated for various temperatures (Table 19). 124 1t 1s doubtful whether 1t can be used again, Additional work is planned with KF with and without added NiF, in which crucibles of BeO, MgO, and Ni and electrodes of Ni, and graphite will be used. 2 graphite, Pt, Reactions in Fused Sodium HYy- droxide¢?) (A. R. Nichols, Jr., Ma- terials Chemistry Division). Morganite has been found to withstand fused NaOH at T00°C sufficiently well to permit (2 rpid, p. 135. 1ts use as a vessel for potentiometric and electrolytic experiments. Because of difficulties in obtaining repro- ducible results from the concentration cells previously described, efforts have been made to use the measurements of decomposition potentials as an approach to the evaluation cfelectrode potentials; however, attention is stil]l being given to the relationship of temperature to potential. Pre- liminary measurements indicate that- a nickel electrode placed i1in a hot region of NaOH melt becomes negative with respect to one in acooler region. This is consistent with present 1ideas of the processes occurring during mass transport. | Voltage-current curves have been obtained for the electrolysis of fused NaOH with and without added NiO by using nickel electrodes. For the sizes and spacings of electrodes used, currents ranged up to 2 ampat voltages of up to 2 volts. With only NaOH, extrapolation of the linear part of the curve gave an apparent decompo- sition potential between 1.2 and 1.4 volts, whereas in the melts containing added Ni0O, an additional electrode process having an apparent decompo- sition potential of 0.7 to 0.8 volt was indicated. The melts containing Ni0O gave a large deposit of metallic nickel at the cathode but little or no attack on the anode. Further experi- ments will be necessary to obtain more quantitative information about these processes. Various phenomena observed during the electrochemical experiments de- scribed have led to experiments con- cerned with reactions occurring in fused NaOH in the absence of applied electric fields. In each case 1imn which NaQOH and NiO were 1in contact at temperatures of 700°C or higher, a fine, crystalline deposit of metallic nickel was observed when the melt was cooled and leached with water. This FOR PERIOD ENDING JUNE 10, 1952 has been observed whether the container used was nickel, recrystallized alumina, or corrundum, The nickel: may be presumed to be formed by the same process as that taking place as the second step of the mass transport phenomenon. ' : When N10O and NaOH are held at 700°C in a nickel vessel for several hours and then slowly cooled, the soliditfied melt is found te contain wiry, black needles or fibers. These fibers are from 0.001 to 0.003 in. in diameter and up to 1/4 in, in length. Their chemical analysis, 44% nickel and 29% sodium, does not correspond well with any simple formula. X-ray-diffraction studies of this material indicate that it is crystalline but not identical with any compound for which data are available. When this material 1is allowed to stand 1in water, 1t slowly changes from black to dark green in color without changing noticeably 1in crystalline form. These washed needles contain no sodium and have a nickel content corresponding fairly closely to the formula Ni(OH)Z'ZHZO. X-ray diffraction shows lines characteristic of Ni(OH)z. However, its index of refraction of approximately 1.652 1s well below that of Ni(OH)z. Studies of the nature of these products are continuing. Compounds Resultiting from Hydroxide corrosion (G. P. Smith, L. D. Dyer, B. Borie, Metallurgy Division)., Studies are being made of reactions that can occur in fused hydroxide media. In making these studies, compounds have been prepared that are believed to be NaNiQ, and NaFeO,. This is the first time that NaNi0O, has been shown to exist. In preparing the NaNi10,, c.p.- grade NaOH was heated in a nickel tube at 700°C and oxygen was bubbled through the melt for 48 hours. The tube of hydroxide was then quenched in o1l and the top portions of the melt were immediately placed in absolute ethanol. 125 ANP PROJECT QUARTERLY PROGRESS REPORT When the hydroxide had dissolved, black crystals were obtained. These crystals were washed with absolute ethanol until the washings failed to turn a phenolphthalein solution red. The product was vacuum dried over Drierite and then washed with water. The compound obtained by washing with water was i1dentified by x-ray dif- fraction as Ni,0,;°H,0, which is a known compound. The compound obtained by washing with alcohol gave an x-ray- diffraction pattern that could not be identified; ,however, this compound reacted with water to form Ni O,°H,O. The product obtained by washing with absolute alcohol was analyzed for nickel, sodium, and carbon., The results of the analysis are given 1in Table 20. TABLE 28 Analysis of Hydroxide Corrosion Product FOUND THEORETICAL FOR ELEMENT (%) NaNiQ, (%) Nickel 51.02 51.6 Sodium 21.2 20.2 Oxygen 28.2 Carbon 0.1 0.0 In a2 similar experiment using a Globe iron tube as the container for the molten hydroxide, NaFeO, crystals were formed. This compound was identi- fied by x-ray diffraction. Mechanism of Fluoride Corrosion (G. M. Adamson, A. D. Brasunas, L. S. Richardson, Metallurgy Division). As mentioned in the last report, (3) magnetic changes were noted in stain- less steels after exposure to fluo- rides. Similar changes have also been noted in the hot legs of the Inconel loops. As an additional check on the (3)roid. p. 136. 126 corrosion mechanism, a microscopic, magnetic method for determining the presence of these transformed layers has been developed. Figure 55 shows that the transformation in an Inconel loop follows a fairly straight-line interface and extends to a depth of approximately two-thirds the depth of the holes. This change of Inconel to a magnetic material may be explained by the removal of chromium, which leaves an iron-nickel alloy. The fact that complete removal of chromium 1s not necessary for this transformation, explains the observation that the area of apparent attack extends deeper than the transformed layer. A study of these facts leads to the conclusion that the corrosion mechanism of fluorides in Inconel is a leaching of chromium from the surface of the Inconel followed by diffusion of new chromium from the interior of the pipe to the surface from which 1t in turn is also leached. As the chromium diffuses to the surface, very small voids are left behind. These voids seem to concentrate in grain boundaries and at similar areas of weakness faster than nickel or iron can diffuse inward to fill them, A number of other experiments have been made that strongly indicate that these are true voids caused simply by chromium depletion of the regions. It has already been es- tablished that the chromium content of fluoride 1s high after corrosion test- ing. Such preferential solution would necessarily result in reduced chromium at the surface of the metal. In one test, metal crystals precipitated from the fluoride contained 4.9% Cr.(%? surface Calculations of void volume based on such chromium depletion are in good agreement with the density of voids M rpia, p. 124, FOR PERIOD ENDING JUNE 10, 1952 T e ; A P s sl 4 . - . o - - : T o i T .ot s sl il L : L\ Lo s -~ * ] S . 4 T, e . ., . . . Loy # ‘o , e e £ ~;\ . i i ’g cf & 3 9 "fl N L R Fe . § ..-\ B . Low L e . ba . ST e, . - - o i ; . o N . £ . » . . P . e rm , . - . \ :,Aff‘; - [ o P . . L S \ - e . , - - T -k .. . " . ¥ o TH . “{‘T S L C s . : LR ;s . s [ magnetic test, P ad - Tty oY A o . "“- - C e kS S : o T - . Foa (6) Transformed layer of hot-leg section. Magnetic Susceptibility of Inconel Thermal Convection Loop After . % Sy At e e Fig. 55. 166 hr at 1500°F with NaF-KF-LiF~UF4. 127 ANP PROJECT QUARTERLY PROGRESS REPORT actwally observed; hence, such voids may be expected. Similar vacancies have been reported¢®'%) jin diffusion experiments by other investigators. It would be reasonable to expect the formation of voids if chromium depletion were accomplished by other means. Therefore two tests were made. (1) Inconel was oxidized at 1200°C for 200 hours. Chromium was selectively oxidized and left a chromium-depleted area., (2) Inconel and an 80% Ni1-20% Cr alloy were heated to 1375°C for 42 hours. The vapor pressure of chromium was appreciably greater than that of iron or nickel. both cases The voids formed in appeared to be 1dentical with those formed in fluoride media. As a further verification of the theory that holes are caused merely by chromium depletion, a test was made using NaF-—KF-LiF-UF4 (10.9-43.5-44,5- 1.1 mole %) in Inconel at 850°C for 100 hours. (This test has always resulted in hole formation.) Abont 10% of -300 mesh chromium powder was added to saturate the fluoride with chromium, and, as expected, no voids were formed. These data and observations are in complete agreement with the theory that the holes observed in fluoride-attacked alloys are true voids caused by selective solution of an alloy component. The mechanisms discussed above apply to Inconel, but further work is necessary to determine whether they also hold for fluoride corrosion of stainless steels. Free-Energy Relations of gFluorides and Structural Metals (R. C. Briant, M. E. Lee, L. A. Mann, ANP Division). Considerable effort is being devoted to determining the causes and mechanisms (S)H. Buckle and J, Blin, J. Inst. Metals 88, Part 7, p. 385 (19052). {6)A. Smigelskas and E. Kirkendall, Trans, A=, Inst. Kining Met. Engrs. 171, 130 (1947). 128 of corrosion of structural materials by ligquid fluoride salts, Free energies of formation of a considerable number of fluorides, oxides, and miscel laneous pertinent compounds have been calculated and plotted against temperature as basic data for de- termining the likelihood of proposed chemical mechanisms as well as the concentrations of resulting products. Solutionof Metals in Moltenm Halides A. Bredig, J. W. Johnson, H. F. Chemistry Division). A (M. Bronstein, general survey of the subject of so- lutions of metals 1n their molten halides and some preliminary experi- mentation were carried on during 1951(7) following a suggestion made in 1950(8) that melts containing "subhalides" or, in other words, "halogen (fluorine) deficient" melts may behave more favorably in regard to corrosion than halides of normal stoichiometry, Although the general aspects of the subject, i1ncluding phase eqguilibria, thermodynamics, and atomic and electronic structure of such melts, continue to be studied on systems contalning a varlety of melts and halogens, a more specific investigation of fluoride melts of particular 1interest tothe ANP work has been started during the past quarter. The results obtained may be summarized as follows: (1) 3 mole % potassium metal dissolved at 860°C in the molten binary eutectac of KFand NaF; (2) 1.6 mole % potassium dissolved at 860°C in the molten ternary eutectic KF-NaF-LiF; (3) sodium and lithiuwm metal added to these eutectics would reduce considerable amounts of potassium fluoride and some sodium fluoride and thus change the composition and melting points of the (T}ChEIistry Division Quarterly Progress Reports for Periods Ending June 30, 1951, ORNL- 1116, p. 65-67; September 30, 1951, OBNL-1153, p- 81; December 34, 1551, ORNL-1260, p. 167-108. (S)M. A. Bredig to A. M. communication, QOctober 1950, Weinberg, private molten mixtures; (4) the solubility of potassium metal at lower tempera- tures is expected to be considerably smaller, perhaps by a factor of 10 at 500°C, and is now being measured. It appears as a result of thermo- dynamic considerations that even small concentrations of potassium metal added and dissolved in the fluoride melt, of the order of 0.1%, can be expected to have a considerable beneficial effect on corrosion, For details the Chemistry Division quarterly progress report for the period ending March 31, 1952 (ORNL-~ 1285) should be consulted. Fluoride corrosion Phenomena (M. A. Bredig, J.W. Johnson, H. F. Bronstein, Chemistry Division). The observation that alkali metals dissolve 1n appreci-~ able quantities in the melts of their halides is of significance in several respects. In the absence of other corrosive agents such as oxides, cor- rosive mass transfer under "dynamic" conditions and the absence for practical purposes of corrosion under "static" conditions may be explained. The equilibrium in corrosion reactions of the type Fe + 2KFT=== Fe¥, + 2K may be favorably influenced by the additioen of alkali metal to the fluoride mixture, ‘ Thermodynamic estimates of equili- brium constants for reactions of this type, based on the most recent data available in the literature, (%) indi- cate concentrations of Cer, Fer, or NiF, at 1500°F to be of the order of 10-4-8 , 1()-5 . 8 , and 1()—6 . 3 , réfi— spectively, At 1200°F the figures are 1076, 10°7, and 10”8, respectively, that 1s, for each metal the concen- trations are less by more thana factor (9)yg, Brewer, L. A. Bromley, P. W. Gilles, and N. L. Lofgren, The Chemistry and Hetallurgy of Miscellaneous Matertials: Thermodynamics, N.N.E.S., Div. 1Y, Vol. 19B, p. 76 {19590), : FOR PERIOD ENDING JUNE 10, 1952 of 10 than at 1500°F. Thus the con- stituents of the structural metal dissolved as fluorides at 1500°F can crystallize out as metallic elements at 1200°F as a result of almost com- plete reversal of the reaction occurring at 1500°F. Assuming rapid establish- ment of equilibrium for 2000 g of molten NaF-KF-LiF (11.5~-42.0-46.5 mole %) in both legs of the thermal loops and a flow rate of 1 cycle per minute, a quantity of 6000 »x 2000 x 10°%8 or approximately 15 g of metallic iron would deposit in the cold leg for 6000 cyeles in 100 hours., For chromium and nickel, these quantities are approximately 100 and 6 g, respectively. These are very crude estimates that. are probably correct to only within a a factor of 10 except for the relations between Fe, Cr, and Ni, which appear qualitatively of the right order. ' The effect of the formation of complexes suchzu;K3CrF5, for which thermodynamic data are not available at present, remains to be considered 1n more accurate calculations. It is clear that the establishment of equilibrium in a static test, that is, without thermal gradients, would not, with a mole fraction of the erder of 107% to 107° of CrF,, FeF,, or NiF,, produce any considerable cor- rosion, : In fuel mixtures containing UF,, the free energy of reactions of UF,, such as Fe + 2UF,z=——= 2UF; + FeF, (AF o ~6 kg/cal), is much less positive than for the reactions dis- cussed thus far for KF (AF = ~80 kg/cal)., The comparatively small increase in cerrosion and mass transfer observed with NaF-KF-LiF-UF, compared with NaF-KF-LiF, in spite of the large disparity of the free-energy changes of the reactions, is explained by the much lower concentration of the UF4. It has been possible to rather dramatically demonstrate the oc- currence, under suitable conditions, of the reaction Fe + 2KF =====FeF, + 2K, 129 ANP PROJECT QUARTERLY PROGRESS REPORT A mixture of iron powder and potassium fluoride was heated for several hours at 920°C in an evacuated stainless steel tube that had an air-cooled section and a water-cooled section, After the test it was found that most of the potassium fluoride had evaporated and redeposited at the air-cooled portion., A very small guantity of black particles were found that were insoluble on dissolution of the fluo- ride in water, The particles, which were metallic 1ron, were probably produced by the partial back reaction FEFz(g) + 2K(g)€“‘"::"::}' Fe(s) + QKF(S)- At the water-cooled part, approximately 1 g of potassium metal was collected, which was the portion that could not react backwards with FeF, because of rapid precipitation of the latter at the air-cooled part of the walls, It 15 impossible to state at present with certainty that the cor- rosive mass transfer observed 1in thermal loops must be entirely or partly due to fluoride reactions of this sort rather than to oxide reactions as long as the possibility of the presence of oxide i1n some form cannot be ruled out. Eguilibria of the type or Fe + UO,F, w===FeF, + UQ, obvicusly may also produce corrosive mass transfer of metal. It seems that recent observations on the decrease or absence of corrosive mass transfer in fuel mixtures containing ZrF, instead of LiF or BeF, have been attributable ta improved purification and decreased oxygen content. Although the removal of oxygen appears to be a mnecessary condition for a low corrosion rate, 130 the possibility of a fluoride reaction cannot be neglected., Sufficiently accurate thermodynamic data for Zr¥F, are not available., The heat of for- mation of ZrF,, AF = 445 t 30 kg/cal, is 10 times that of KF, and the equilibrium concentration of FelF, at 1100°K, which is from 105 to 1079, is one-thousandth that of KF. If the figure 105 were correct, it would be necessary to find an explanation for the low rate of corrosive mass trans- fer actually observed. Observations on the peculiar behavior {(the apparent self-welding) of nickel in fluoride melts containing 7ZrF,, which might perhaps be connected with some alloying of nickel with zirconium produced by the reaction ZrF, + 2Ni === 2NiF, + Zr, may indicate a possible way of reconciling the large figure of 107° for the concentration of FeF, (zf 1t is correct) with the low corrosion rate observed in the thermal loop. Zirconium from the above reaction might be thought to plate out on, or form an alloy with, the steel or the Inconel, This could reduce the cor- rosion rate by at least two independent mechanisms: (1) formation of a pro- tective surface film containing zir- conium and (2) absence of transport of zirconium to the cold leg for reversal of the equilibrium there, which reversal alone would make new attack by KF in the hot leg possible. It is obvious that much more work is required to completely elucidate the mechanism of corrosive mass trans- fer, However, the possible occurrence of zirconium plating or alloying may lead to the development of methods other than the use of fluorine-deficient melts that may permit general control of corrosion and mass transfer 1in fluoride melts, ,l AR D b T A AT R 000, P N P e e e FOR PERIOD ENDING JUNE 10, 1952 12. METALLURGY AND CERAMICS W. D. Manly J. M. Warde Metallurgy Division High-temperature brazing alloys have been studied for possible use in the heat exchanger and reactor as- semblies. The tests conducted on brazing alloys include flowability of the braze, determination of the physical properties of the resulting joint, and static corrosion tests of the brazed joint in reactor fluids. The 60% Pd-40% Ni alloy was found to be the best brazing material to use in contact with fluoride mixtures. Additional experiments have been con- ducted on the cone-arc techniques to obtain an understanding of the variables of the process. Some work has been completed on the practical application of cone-arc welding to fabrication of the tube-to-header joints of heat exchangers, l.oose powder sintering has been studied as a method for the production of solid fuel elements. This technique involves sintering a loose powder of the uranium-bearing mixture to a solid backing plate. The variables studied have included sintering temperature, sintering time, fuel-component particle srze, cold working and resintering, surface preparation, and sintering under load. Powder mixtures of boron carbide and 20% iron by volume have been hot pressed to make slugs for the safety rod of the ARE. Hot-pressed mixtures of aluminum oxide and boron carbide are being studied for use in the regulating rod of the ARE. An oxidation-resistant ceramic coating has been successfully applied to stainless steel, and similar coatings are being developed for nickel and for a stainless steel radiator intended for use with the experimental reactor. LOOSE POWDER BONDING E. S. Bomar - J. H. Coobs Metallurgy Division : In the development of fuel elements utilizing a solid fuel, a variety of possible configurations has been suggested for incorporating the fuel with a structural material. One method of fabrication would be to bond a fuel-bearing powder to an appropriate backing material by sintering the powder either with or without appli- cation of pressure., Preliminary experiments to prepare flat sheets covered on one side with a sintered fuel-bearing powder were carried out in a furnace at the Micro-Metallic Corporation in New York. This work has been reported elsewhere,(!'+?) A furnace with a suitable reducing atmosphere has been made available at ORNL and used for further work on this problem. Several variables of the bonding technique, including sintering temperature, sintering time, fuel- component particle size, surface preparation, and sintering under load, have been investigated. Samples processed to include these variables were also cold worked and resintered, and they were examined metallographi- cally for evaluation of results. Only the sintering temperature and the fuel-component particle size had appreciable effect on the bond. Both the higher temperatures and larger particle sizes tested were shown to be beneficial., ' ‘ (l)MetalZurgy Division:Quarteriy Progress Report for Period Ending April 30, 1951, ORNL- 1033, p. 54. Metallurgy Division Quarterly Progress feport for Period Ending July 31, 1954, ORNL-1108, p. 37 _ 131 ANP PROJECT QUARTERLY PROGRESS REPORT The materials used i1n these bonding experiments included type-302 stain- less steel (-325 mesh) as the metallic component in the powder mixture and UO, fired to 2100°C in a hydrogen atmosphere as the fuel component, Standard U,S., screens were used for sizing particles. A Globar heated, ceramic-tube furnace was used for all the runs excepting those requiring loading of the samples. These were carried out in a larger, molybdenum- wound furnace equipped with an Inconel tube. The furnace atmosphere in every instance was hydrogen that had been passed over a platinum catalyst and through an activated-alumina drying column, The type-316 stainless steel stock was annealed i1n the hydrogen atmosphere prior to coating with the powder mixture. Sintering Temperature. The trend was toward improved strength of bond between the backing sheet and powder with 1ncreasing sintering tempera- tures, After sintering at 1300°C and lower, bonding of the 30% by volume (-200, +325 mesh) U0,~70% (-325 mesh) type-302 stainless steel mixture used was very poor, whereas the samples sintered at 1360°C or higher showed considerable tenacity even on bending. The four samples shown in Fig. 56 were sintered at 1300, 1325, 1360, and 1385°C, respectively, The powder mixture was placed on the backing and leveled to approximately 0,030 in. in thickness, Retention of the powder became successively poorer as the sintering temperature was lowered. In an effort to consolidate the powder layer, the composites were rolled to a 40% reduction in area and heated to 1250°C for 30 minutes, These samples are also 1llustrated in Fig. 56, Sintering at high temperatures caused transformation of some of the austenite to delta ferrite, which appeared as needle-like crystals in 132 the type-302 stainless steel matrix and as a band about individual grains of the type-316 stainless steel. The latter distribution of the ferrite might be expected to lead to a pref- ferential grain boundary attack 1f the backing were subsequently exposed to one of the coolants of current interest. The presence of the magnetic phase 1n the type-316 stainless steel was made evident by placing adrop of an aqueous suspension of a colloidal 1ron com- pound on the surface of the metallo- graphic specimens and then exposing the samples to a magnetic field. Sintering Time. The sintering-time study indicated there was no appreci- able increase 1n the quality of the bond obtained with samples heated for times up to 2 hr over that for a sample held at temperature for 30 minutes. Duplicate samples from this group were also given a 40% cold reduction and reheated to 1250°C for 30 minutes. In every case the rolled and sintered samples showed well-consolidated structures with relatively small amounts of porosity. The U0, retention in several instances was poor and the UO, could be removed by polishing. The poor retention is thought to be caused by partial crushing of the oxide particles during rolling. Fuel-Component Particle Size. The quality of bond between the metallic particles in the powder layer and backing sheet showed general improve- ment as the particle size of the UQ, was 1ncreased, Particle sizes ranging from -100,+140 mesh to =325 mesh were used. The better continuity of metallic particles with the larger sizes is shown in Fig. 57. Both the material sintered at 1300°C for 45 min and its the rolled and resintered counterpart are illustrated in the figure. Cold rolling followed by reheating at 1250°C resulted in considerable compacting, as shown in Fig. 57. tel _Fig. 56. sintered. O - A d’jvusfi&@! %* I 260°C < (250 X). #* ,§.' fi385°c A4__ (200 X} g Effect of Sintering Temperature on Loose Powder Bonding. Sintered 45 minutes. (b} Cold rolled to 40% reduction in area and resintered 30 min at 1250°C. (a) As ¢S6T ‘0T 3NNT ONIGNH doTIvdd HOA VET Y-5617 4 (=100, + 140 MESH) ’ SRR ‘ WY 5508 . (-140,+200 MESH) (- 200, + 325 MESH) (—325 MESH) (g),,,. | I - (b)A Fig. 57. Effect of Fuel Component Particle Size on Loose Powder Bonding. Sintered 45 min at 1300°C. {a) As sintered, {b) Cold rolled to 40% reduction in area and resintered 30 min at 1250°C. 200X. LHOdAY SSHYI0Hd ATHALYVNO LOAT0Ud dNV w Surface Preparation. Contrary to the results obtained with samples sintered at the Micro-Metallic Corpo- ration plant, surface preparation of the type-316 stainless steel sheet had little effect on the degree of bonding to the powder laver. Some small benefit may have been obtained in roughening by sand blasting or by using annealed as-received stonck as opposed to electropolished sheet. Sintering Under Load. Samples were loaded during sintering with steel weights to obtain unit pressures of 1/2 to 5 psi. Cross sections of these samples i1ndicated that no gross improvements were effected by the maximum loading used. CONTROL ROD FABRICATION Several important decigiocuns have been reached concerning materials to be used in these rods, and fabrication of the components has begun. After considering the availability of stock for the cans and the time scheduls, 1t was decided to make the cans from the types-316 and -304 stainless steel tubing on hand. These steels are admittedly not the best materials for compatibility with boren carbide and perhaps not the best from the self- welding viewpoint; however, the delay necessary to obtain the more desirable type-430 stainless steel tubing is prohibitive. The cans are being fabricated by the Y-12 Shops, and they will be loaded with slugs and brazed with Colomonoy Nicrobraz by the Welding Group. ' ‘ Safety Rod Slugs. The safety rod slugs are being fabricated by hot pressiog an iron-boron carbide mixture containing 56% by weight (80% by volume) boron carbide. The mixture was selected as an alternate to pure boron carbide because the latter offers speciral fabricatiocu preoblems and is A A e 8RB B b R R S s n A mm mmm e FOR PERIOD ENDING JUNE 1¢, 1952 considerably more exvensive. The slugs are prepared from boron carbide that has been milied 16 hr in a steel mill with steel balls and blended 8 hr with the requisite amount of -325 mesh iron powder, The mixture i1s then fabricated by hot pressing at 1520°C and 2500 psi. Enough slugs to prepare one safety rod have been pressed with little difficulty., Cracking of the samples because of the difference in thermal expansion of the iron~-boron carbide compact and the graphite mandrel ‘'was overcome by ejecting the mandrel at temperature, The pressed parts have densities of 2.80 g/cc {about 80% of theoretical) and will be brought within acceptable dimen~ sional tolerances by grinding. : Four slugs less than full size have been pressed, canned, and brazed: two in type-304 and two in type-316 stain- less steel, One can of each material was sectioned and examined after the brazing cycle., There appeared to be no appreciable interaction between the cermet insert and the containecr. The other two samples were held for 100 hr at 815°C before cpening for exami-~ nation. In each of these a thin layer of a white, crystalline compound that was slightly soluble in hot water was found on the inside surface of the container, The white, crystalline compound was present 1n greater cone- centration at points wet by the Nicrobraz. Efforts are being made to establish the identity of the compound by x-ray-diffraction technigques. Regulating Rod Slugs. The require- ments for the regulating (shim) rod are somewhat different., For this purpose a small amount of beron (3 to 12.5 g) wust be uniformly dispersed in a material relatively transparent to neutrons, such as Al,0,., A l-in. by 1/2-in,-dia slug was made by hot pressing a mixture composed of 99% grade 38-500 Al,0, plus 0.17% B,C at at 1750°C in a graphite die under a ANP PROJECT QUARTERLY PROGRESS REPORT pressure of 2500 psi. Metallographic examlination revealed that the two materials are compatible. A second slug containing 0,74% by weight of -325 mesh B,C, prepared by the same technique, had a density of 3.45 g/cc (86% of theoretical) and possessed satisfactory physical properties. The composition of this compact corresponds closely to that of material to be used in the "high-B,C content" regulating rad, MECHANICAL TESTING OF MATERIALS R. B. Oliver C. W. Weaver D. A. Douglas J. W. Woods Metallurgy Division The installation of facilities for stress-rupture testing in liqguid metals and fused salts is complete. The testing machines were calibrated with a standard-sheet test specimen on which SR-4 strain gages had been mounted, and this specimen i1n turn was calibrated agailnst a standard proving ring. The testing chambers were originally designed to contain sodium or other low-melting-point materials. With the change in the program and the need for data from tests using the higher melting fused fluorides, a few minor design changes had to be made. Containers to charge and empty the testing chambers have been designed and are being constructed of Inconel, These containers are designed to also fit the apparatus used to produce the fluoride mixtures., Testing will be started shortly after the first lot of fluoride mixture 1s received, The primary purpose of these tests will be to study the effect of stress- corrosion on the creep rate and stress- rupture life of these materials. Both Inconel and type-316 stainless steel are being tested in the new machines in an argon atmosphere in the same stress ranges as they were previously 136 tested in the old machines. A com- parison of these two sets of data will establish the effects of the inherent variables of the new testing systems, and thus the results of the corrosive action can be studied. Stress-Rupture Tests im Argon. Both fine- and coarse-grained Inconel specimens are being tested at stresses from 1500 to 4500 psi in argon at 815°C., The expected test duration in this stress range 1s from 800 to 3000 hours. Upon completion of these tests, revised design curves will be compiled. A short series of tests at 815°C in air are being initiated to obtain data for a comparison of the effects of oxidation and fluoride attack on the creep rate and rupture 1ife. Type-316 stainless steel specimens are being tested 1in argon in both the old and the modified test chambers at 5300, 5800, 6300, 6800, and 7300 psi. These duplicate sets of data will serve to correlate the work 1n the two types of testing chambers and will furnish data for a design curve for type-316 stainless steel tested 1n argon. Stress-Corrosion Tests (G. M, Adamson, K. W. Reber, Metallurgy Division)s. Preliminary experiments to determine the effect of fluoride corrosion on the physical properties of structural metals have continued in the crude apparatus currently available pending the installation of more sensitive equipment. Both stress- rupture and tube-burst tests were carried out during this period. The tube-burst tests have been described in previous reports; however, it should be noted that these tests are designed to yield hoop stress- rupture data. Table 21 is a summary of the data obtained by immersing the specimen under the desired stress 1n NaF-KF-LiF-UF, (10.9-43.5-44.5-1.1 mole %) at 1500°F. The stress-rupture apparatus trans- mits a dead-load stress to the specimen by means of a lever arm and a bellows arrangement, A thin flat specimen is subjected to a tensile stress while immersed in a bath at the desired temperature. For all these tests a bath of NaF-KF-LiF-UF, held at 1500°F was used. The data obtained are tabulated in Table 22, The third TABLE FOR PERIOD ENDING JUNE 10, 1952 column i1n this table presents data obtained 1n argon atmospheres. It 15 readily apparent that a contradiction is present in these data. At high stresses the two sets of data check, whereas at stresses below 10,000 psi there is no corre- lation. In such testing, as the stresses are lowered the relative effects of errors become larger. It is apparent that the corrosion-stress- rupture apparatus i1s not sufficiently sensitive at the very low stresses. 21 Tube-Burst Data for Structural Metals Tested in NaF-KF-LiF-UF, STRESS HOURS TQ RUPTURE INCREASE IN (psi} In Inconel In Type-316 Stainless Steel DIAMETER (%) 1200 1000* L7 1600 - 1000+ 2135 6588 . 2030 - 880 24990 128 2.2 *Did not rupture. TABLE 22 Inconel Sttess-aupture Data , HOURS TO RUPTURE STRESS (psi) In NaF-KF-LiF-UF, In Argon Atmosphere 12,500 6.5 6 10,000 20 19 7,500 31 80 21.5 14.5 5,000 57504) 270 4,000 661¢8) 430 (a)Test apparatus fajiled before rupture. (b Specimen ruptured in weld area of grip rather than in test section, 137 ANP PROJECT QUARTERLY PROGRESS REPORT CONE-ARC WELDING P, Patriarca G. M., Slaughter Metallurgy Division It was found desirable to determine the applicability of cone-arc welding techniques to fabrication of tube-to- header joints in heat exchangers. This study was given precedence over the long-range approach of determining the individual and combined effects of the various cone-~arc welding process varliables. However, the variables, such as arc time, arc current, and electrode gap distance, have been extensively tested on materials and geometries applicable to heat exchanger designs. One heat exchanger was fabricated, and the fabrication of test assemblies by cone-arc welding will continue. However, studies to evaluate the fundamental effects of the cone-arc welding variables will receive con- current attention in future work. Eguipment used for cone-arc welding experiments was described previ- ously, (3 Effect of Welding Conditiouns. A series of headers were fabricated to conform to a close-packed tube and hole arrangement in a circular header. The header material used was type-304 stainless steel sheet 1/8 in. thick. Nineteen tube holes were drilled to receive type-316 stainless steel tubing, 0,100 in. OD with 0.010-1in. wall thickness. The basic pattern was an equilateral triangle with a tube hole at each apex. Each tube was 0,100 in., or one diameter, from its nearest neighbor. Tube-hole to header-edge distances were initially chosen as 0.050 in. but were increased in subsequent experiments to 0.100 (3) Airceraft Nuclear Propulsion Project Quarterly Progress Reports for Periods Ending December 10, 1951, ORNL-1170, p. 132; March 10, 1952, ORNL- 1227, p. 142. 138 inch. Suitable welding conditions were chosen individually by experiment be fore being applied to a 19-hole header assembly for a consistency evaluation, Figure 58 illustrates a series of typical tube-to-header consistency determinations., Each header was fabricated by using adifferent welding condition, as set forth in Table 23, to illustrate the flexibility of the cone-arc process. Cone-arc welds within the limits of the material and joint design of this investigation may be made over a range of welding conditions, Although the welds 1llustrated by Fig. 58 and Table 23 are not representative of a systematic study, theilr examination reveals trends to be verified by future work, It appears from comparison of these welds that the primary effect of shortening the arc distance 1s to recess the cone-arc weld. Although welds a, b, and ¢ illustrate the com- parable net effect of increasing . R Y. 6667 848 S hee o gy 2pH » Q'.‘;j 2L $a 9 Qfifi@@iii&-@g ¢ o990y &!l!}fi et e L 2 bee ‘ ‘yl .‘:‘m * ' ’ . \ ‘ ‘ ‘ o) (&) {¢) ! i l i { Fig. 58. Typical Cone-Arc Welds. For description of the welding con- ditions for each of these samples see Table 23. FOR PERIOD ENDING JUNE 10, 1952 TABLE 23 Cone-Arc Welding Conditions for Tube-to~Header Joints HEADER DESIGNATION | PISTANCE FROM ELECTRODE | ppc cURRENT | ARC TIME (Ref. toFig. 58) | THPTOPLANE OF WORK (amp) (sec) a 0.020 64 1.5 b 0.050 60 1.8 ¢ 0,040 70 1.2 d 0,100 70 ' 2.0 e 0.100 84 1.4 f 0.050 60 1.8 current with a corresponding decrease 1n arc time, 1t 1s not believed possible to determine these effects on variables such as weld penetration without thorough metallographic examination. Welds d and e illustrate the combined effects of further increasing arc length, arc current, and arc time - a condition is reached that indicates superfluous welding, This is apparent by observing the relatively large welds and the tendency to form a rosette pattern on the header edges. The heat flow pattern, the uniformity of which seems to be an important variable in cone-arc welding, appears to have been distorted on prolonged heating. It may be noted that the outer holes of headers e and f do not contain welded tubing. The tube-periphery to header-edge distance in this case was 0,050 inch, This distance was found to upset the uniformity of the heat flow pattern and welding conditions were not found which would give satisfactory welds. There 18 a tendency for an are that is terminated instantaneously without a current taper to leave a crater. Examination of the periphery of cone- arc welds reveals a small spot that is characterized by a small amount of scale. The extent of these craters indicates that porosity due to cratering 1s not a serious problem. However, the presence of a single pronounced crater on the cone-arc weld periphery tends to confirm the belief that the cone-arc may be a single, rotating-arc beam. : * Heat Exchanger Assembly. Cone-arc welding as applied to a heat exchanger assembly is illustrated in Fig. 59. This photograph shows one end of an Inconel counterflow heat exchanger consisting of 52 tubes, 0.150 1in. OD with 0.,025-in, wall thickness, cone-arc welded into a cylindrical Inconel header, 3 in, ID with 1/8-in. wall thickness. A total of 104 welds (the other end of the heat exchanger is identical to the one illustrated), were made by using the following conditions: : Arec current 102 amp Are time 2 gec Electrode 3/3%-in,-dia thoriated- tungsten, pointed Distance from elec- trode to plane of work : 0.165 in. Shielding gas - argon 30 e¢fh 139 ANP PROJECT QUARTERLY PROGRESS REPORT Fig. 59. One End of A cone-Arc-~ Welded Inconel Heat Exchanger As- sembly. TESTS OF BRAZING ALLOYS P, Patriarca G. M. Slaughter Metallurgy Division The primary objective of the high- temperature brazing alloy investigation during the past few months has been to screen brazing alloys for aircraft reactor applications by using flow- ability and static corrosion tests 1m sodium hydroxide and NaF-KF-LiF-UF, fuel mixture. After these pre- liminary investigations, promising alloys can be further studied in physical property tests such as butt- tensile tests (at both room and elevated temperatures), elevated-temperature creep tests, dynamic corrosion tests, and room- and elevated-temperature ductility tests. Such screening procedures seem appropriate 1n view of the large number of alloys now under consideration, In Table 24 the results of pre- liminary investigations of eight high- temperature brazling alloys are re- corded. Statlc corrosion tests on 140 joints brazed with some of these alloys are being made, and the results will be reported later. The melting points of the various alloys are only approxi- mate and will vary somewhat since chemical analyses of various heats of the same alloy differ slightly, Although most of the brazing alloys tested flowed satisfactorily, none exhibited satisfactory corrosion resistance 1in sodium hydroxide. Nicrobraz, 60% Pd-40% Ni (with and without a 3% Si additien), and 64% Ag- 33% Pd-3%Mn were the only alloys not severely attacked by the fluoride fuel NaF-KF-LiF-UF, (10,9-43,5-44.5-1.1 mo le %). At the present time, static tests on Anickel joints brazed withpromising alloys are being conducted. It 1is hoped that better resistance of the brazing alloys to corrosion by sodium hydroxide can be obtained, since nickel 1s attacked only slightly by NaOH ~ this may not hold true with the introduction of a bimetallic system, Nicrobraz, Static corrosion tests of Nicrobrazed joints on Inconel and type-316 stainless steel 1in both NaF-KF-LiF-UF, fuel mixture and NaOH for 100 hr at 1500°F have been com- pleted. Photomicrographs of corrosion- tested, Nicrobrazed, Inconel joints were presented in aprevious report. (%) Nicrobrazed specimens of type-316 stainless steel tested in both the fluoride mixture, NaF-KF-LiF-UF,, and NaOH are shown in Fig, 60, The fluoride bath had a relatively minor corrosive effect onthis joint, whereas the sodium hydroxide was much more severely corrosive. Manganese-Nickel Alloy. The 60% Mn- 40% N1 alloy was found to be extremely brittle and many brazed joints made with it were subject to severe cracking, This brazing alloy was found to have op. cit., ORNL-1227, p, 145-146. 1¥T TABLE 24 - nC A Summary of Flowability and Corrosion Tests of Various Brazing Alloys STATIC CORROSION RESULTS APPROXIMATE STATIC CORROSION RESULLTS BRAZING ON BRAZED INCONEL ON BRAZED BRAZING ALLOY “fi;:fi:s TEMPERATURE J::Qg:ifi;gii TYPE-316 STAINLESS STEEL (°F) (°F) In Fluoride In Fluoride No. 14 In NaOH No. 14 In NaOH Nicrobraz 1850 2050 Excellent Slight at- Extremely Moderate Severe (70.17% Ni-13.95% Cr-5.86% tack with severe leaching Fe-~4,59% Si~4.92% R) of boron 60% Mn-40% Ni 1859 1960 Poor flow, Severe Severe . L ..cracking . |. o prevalent 60% Pd-40% Ni 2260 2320 Excellent No apparent Severe attack {60% Pd—40% Ni) + 3% Si L2150 2200 -Excellent Moderate: - Severe Moderate - Severe 16.5% Cr-10.0% Si-73.5% Ni 2100 2200 Moderate 16.5% Cr-10.0% Si-2.5% 2100 22060 Moderate Severe Severe Moderate Severe ‘Mn—71‘0% 8@ 64% Ag-33% Pd-3% Mn 2130 2200 Excellent Moderate Severe VYery slight Severe 5% Ag-20% Pd-5% Mn 2100 2200 Excellent Severe Severe cset ‘0T ANNL ONIaNd d0I¥dd 404 ANP PROJECT QUARTERLY PROGRESS REPORT (a) Tested in NaF-KF-LiF-UF, #2< =@#@.'”’ *‘@' ' vy J‘ /—*V\w 'Z - . e (b) Tested in NaQH Fig. 60. Type-316 Stainless Steel Joints Nicrobrazed and Static Corrosion Tested for 100 hr at 1500°F. FEtched with agua regia. 200X. 142 low resistance to corrosion ian the media in which 1t was tested; 1t was severely attacked by the fluoride mixture, NaF-KF-LiF-UF,, and NaOH (as shown in the previous quarterly report(3))., The poor corrosion re- sistance coupled with relatively poor flow properties seem to make this alloy unsuitable for c1rculat1ng~fue1 reactor applications., ~ Palladium-Nickel Alloy. The 60% Pd- 40% Ni brazing alloy has excellent flowability properties, and the rather limited experimental data obtained to date indicate that 1t has better than average resistance to corrosion by the fluorides, The insignificant attack of the fluoride mixture, NaF-KF-LiF-UF, on an Inconel joint brazed with this alloy 1is shown in Fig. 61. It can be seen that i1ntimate bonding with the base metal 1s obtained during brazing, since Lhe grain boundaries of the ....... as the parent Inconel grains. The 60% Pd-40% Ni alloy, with 3% silicon added for 1ts effect upon melting-point lowering, was also in- vestigated. The s1licon addition caused phenomenal flow that traversed the entire: length of a 6-in. joint and even flowed up the scratches., However, the room-temperature tensile strength of type-316 stainless steel joints butt-brazed with the alloy was low, and the brazed joint was very brittle. In addition, the resistance to corrosion in the fluorides of brazed Inconel joints seems to have been lowered somewhat by the addition of the silicon to the 60% Pd-40% Ni alloy. Further attempts to lower the melting point of the 60% Pd-40% Ni alley by the addition of a third component are being made. A 1% beryl- lium addition and a 10% manganese G rbea., p. 147, FOR PERIOD ENDING JUNE 10, 1932 addition are being studied, but more work has to be done before the desir- ability of such additions can be determined. Ni-Cr-Si-Mn Alloys. Two brazing alloys of the Ni-Cr-51-Mn type have been investigated in preliminary experiments., The 16.5% Cr~10.0% Si~73.5% Ni and the 16.5% Cr~10,0% S1-2.5% Mn~-71.0% Ni alloys have desirable flowability characteristics, but they are apparently attacked to some extent by the fluorides and rather severely by sodium hydroxide. Static corrosion tests for 100 hr at 1500°F on Inconel joints brazed with the alloy containing a small amount of manganese have been completed. Typical sections of joilnts show moderate attack in the fluoride, NaF- KF—LiF—UF‘, and rather severe attack in sodium hydrox1de‘ Silver~-Base Alloys. Two silver-base ~brazing alloys of the compositions 64% Ag-33% Pd-3% Mn and 75% Ag-20% Pd- 5% Mn, which melt at approximately 2100°F, have also been investigated. The flowability of these alloys 1s very good, but the static corrosion results look promising in only a few instances, Inconel and type-316 stain- less steel joints brazed with the 64% Ag-33% Pd-3% Mn alloy were static corrosion tested for 100 hr at 1500°F in both NaF-KF-LiF-UF, and NaOH. The fluorides were relatively noncorrosive (Fig. 62), particularly on the stain- less steel, whereas the sodium hy- droxide was severely corrosive. Inconel joints brazed with the 75% Ag-20% Pd-5% Mn alloy have been tested in NaF-KF-LiF-UF4 and NaOH for 100 hr at 1500°F. The resistance to corrosion of both joints in these media was poor, as evidenced by severe attack, particularly in sodium hy- droxide. ' 143 ANP PROJECT QUARTERLY PROGRESS REPORT CERAMICS J. M. Warde, Metallurgy Division National Bureau of Standards ceramic coating A-418 has been successfully applied to samples of stainless steel, and 1t showed good resistance to oxidation at 900°C in a 100-hr test. Work is in progress to apply a ceramic coating to nickel and to develop a technique for coating a complete assembly of a radiator intended for the ARE. A study of the effect of ceramic coatings on the thermal effi- ciency of a radiator is under con- sideration. Moderator blocks for use in the ARE have been fabricated from beryllium oxide at the U. S. Bureau of Mines Station at Norris, Ceramic Coatings for Metals. Work was continued during the past quarter on the application of ceramic coatings to metals for possible ln a radiator. National Bureau of Standards coating A-418 applied to about 2 mils thickness to three stainless steels (types-302, -316, -347) successfully retarded oxidation of the metals in a 100-hr test at 900°C. The frit com- position of this coating has been described in a previous report. (% An attempt to apply a ceramic coating to nickel is now in progress and a technique for coating the complete radiator assembly 1s under con- sideration. It is proposed to construct a thermal conductivity apparatus for the purpose of determining the effect of ceramic coatings on the thermal efficiency of the radiator, use Fig. 61. Inconel Joint Brazed with a 60% Pd-40% Ni Alloy and Static Corrosion Tested in NaF-KF-LiF-UF4 for 100 hr at 1I500°F, 144 Etched with agqua regia. 100X, FOR PERIOD ENDING JUNE 106, 1952 (b) Type-316 stainless steel joint. Unetched. Fig. 62, Joints Brazed with a 64% Ag-33% Pd-3% Mn Alloy and Static Corrosion Tested in NaF-KF-LiF-UF, for 100 hr at 1500°F. 100x. 145 ANP PROJECT QUARTERLY PROGRESS REPORT Ceramics Laboratory Program. Among the projects planned for the Ceramics Laboratory that are of particular inter- est to the ANP are the investigation of ceramic metal combinations (cermets) for reactor components, radiation damage studies of ceramic materials, and a program of measurement of physical properties of ceramic materials for use in reactor design, These programs are developing as rapidly as equipment can be completed and installed. 13. HEAT TRANSFER AND PHYSICAL PROPERTIES RESEARCH H. F. Poppendiek Reactor Experimental Engineering Division The viscosities of NaF-KF-LiF-UF, (10.9-43.5-44.5-1.1 mole %) and two BeF,-bearing fluorides have been determined with a rotational vis- cometer. The viscosities of NaF-KF- LiF-UF, range from about 9 centipoises at 530°C to about 3 centipoises at 750°C. The viscosities of both BeF,- bearing fluoride mixtures were so high, greater than 12 centipoises at 725°C and greater than 25 centi- poises at 600°C, that BelF, was eliminated from*further consideration as a potential fuel constituent for the ARE. Preliminary viscosity measure- ments of the NaF-KF-ZrF, -UF, fuel mixture (5-51-42-2 mole %) have been obtained with the efflux viscometer. The viscosities, which were from about 10 centipoises at 570°C to about 4 centipoises at (50°C, appear satis- factory for the ARE. Thermal conductivity measurements on NaF-KF-LiF (eutectic) have in- dicated a mean value of about 2.6 Btu/hr-ft? (°F/ft) over the tempera- ture range 500 < t < 750°C. Heat capacity data on barium and potassium hydroxide have been determined. The density-temperature relations of several of the zirconium fluoride salt mixtures have been determined,. In addition, the vapor pressures of BeF,- and a ZrF,-bearing fuels have been measured. A table summarizing some of the physical properties of materials of interest to the ANP Froject is included. 146 Experimental determinations were made of the heat transfer coefficients of molten sodium hydroxide flowing turbulently in a heated tube. Results indicated that sodium hydroxide heat transfer was similar to convective heat transfer of ordinary fluids. Fundamental experimental information on heat transfer of turbulently flowing mercury in thermal entrance regions has been obtained; the experimental results are 1n agreement with previous theoretical determinations. Mathe- matical analyses pertaining to circu- lating-fuel heat transfer systems have been developed. In one analysis a study of the thermal structure 1in the entrance region was made, and another anmalysis considered non- isothermal heat transfer. Some ex- perimental velocity information on natural convection in an idealized liguid fuel has been obtained. VISCOSITY OF FLUORIDE MIXTURES J. M. Cisar, ANP Division F. A. Knox F. Kertesz Materials Chemistry Division R. F. Hedmond T. N. Jones Reactor Experimental Engineering Division Study of the viscosity of fluoride mixtures has been continued with both the rotational (Brookfield) and efflux viscometers that were modified tosebtain control over the melt. In the case of the efflux viscometer, the whole system has been housed in a small dry box so that determinations can be made in inert atmospheres. The efflux viscometer and the can- tainer for the test liquid are instru- mented with several thermocouples so that a reasonably accurate value of the liquid temperature can be obtained while making determinations. Tempera- ture errors due to thermal radiation losses have been reduced in the system by reducing the radiation- configuration factor. The dry box is divided inteo two sections; one contains the furnace, viscometer, and balance, and the other can be used to load containers with test materials. The Brockfield apparatus was similarly enclosed. With the modified apparatus, the viscosities of several BeF,-bearing fluorides, one ZrF, -bearing fluoride, and several other fluoride mixtures were determined. The two BeF ,-bearing fluorides had viscosities greater than 12 centipoises at 725°C and both approached 25 centipoises at 650°C. These viscosities were too high toe permit satisfactory heat transfer in the experimental reactor, and as a result a search is being made for a lower viscosity fuel. The one ZrF,-bearing fluoride measured has a satisfactory viscosity, that is, 8 centipoises at 600°C, which de- creases to 4 centipoises at 750°C. BeF,-Bearing Fluoride Mixtures, The viscosities of two Bel,-bearing mixtures were determined with the Brookfield viscometer in the range 600 to 800°C. For these measurements 1t was necessary to minimize the formation of beryllium oxide by bubbling HF gas through the molten liquid prior to the run and maintain- ing an atmosphere of HF over the melt during the measurement. The first mixture contained 31 mole % BeF,, FOR PERIOD ENDING JUNE 10, 1952 47 mole % NaF, and 2 mole % UF, and showed viscosities from 12 centipoises at 800°C up to about 25 centipoises at 650°C. Another preparation of considerably lower Bel, concentration contained 12 mole % BeF,, 76 mole % NaF, and 12 mole % UF, and was con- siderably less viscous at the high- temperature end of the range. Iowever, the viscosity (5 centipoises at 810°C) rose rapidly as the temperature was lowered, reached 13 centipoises at 725°C, and approached 25 centipoises at 650°C, : These findings are in agreement with the fact that BeF, is a glass formed with directionally bonded atoms; 1ts viscosity increases with decreasing temperatire considerably more rapidly than the viscosity of other ionically bonded salt mixtures. These high viscosities have virtually eliminated BeF, from the list of potential compounds for the ARE fuel mixture. ' ZrF4~Bearing Fluoride Mixtures,. Preliminary viscosity measurements have been obtained for a ZrF -bearing fuel mixture, Nal' ~KF-ZrF, -UF, (5-51- 42-2 mole %), on the efflux viscometer. The results are shewn in Fig. 63. Although the viscosity is satisfactory for the present ARE, this mixture does not contain sufficient uranium for criticality. However, the in- crease in the uranium concentration that would be required should not have a significant effect on the viscositvy. Jn an attempt to determine possible variations of viscosity with various compositions of the mixture, measure- ments have been made with the rotational viscometer on a mixture of 55 mole % ZrF -45 mole % KF after additions of varying guantities of NaF, The viscosity of the ZrF, -KF binary was guite low: 3 centipoises at B00°C and 8.5 centipoises at 600°C, Four successive additions, which brought 147 ANP PROJECT QUARTERLY PROGRESS REPORT R DWG. 15351 17¢ (oK) 0.0009 0.0010 0.0011 0.0012 0.0013 0.0014 0.0015 10 , | | T 1 | - P g8 | // /4 NaF - KF —Zrfg ~UFg — s } (5.0 ~51.0-42.0-2.0 mole % ); ~ NaF- KF — LiF —UF, 6 (ClsAR) 1 (109-435-44.5-1.1 mole %)l [ 7 / (REDMOND AND JONES) 2 T Ly ~§ 4 f/ ////// > v |-»,_ & / o / > ——— NoF - KF - LiF H (11.5 -42.0-46.5 mole % ) g 2 . (KAPLAN AND TOBIAS) 7)) m < 1 L | | | | 800 700 600 500 400 7 (°C) Fig. 63. Absalute Viscosity of Several Fluoride Mixtures as a Function of Temperature. the NaF content up to 20% by weight, resulted in no detectable change in the viscosity. Material from Corrosion Loops. A number of dynamic corrosion loops of type-316 stainless steel have shown signs of plugging during circu- lation of NaF-KF-LiF-UF, mixtures. The melting point of the fluoride mixture remained unchanged during the runs and, i1n some cases, no physical evidence of a plug was found. Repeated determinations have shown that the viscosityof the mixture 148 after circulation was identical, within experimental error, with that before circulation. Measurements made immediately after melting the mixture were also identical with those made after keeping 1t molten under a blanket of inert gas for 24 hours. NaF-KF-LiF and NaF-KF-LiF-UF, Mixtures. The falling-ball apparatus has been used to obtain the viscosity of NaF-KF-LiF (11.5-42.0-46.5 mole %) at several temperatures in order to check earlier results by other methods and to study behavior of the apparatus ~ the results checked satisfactorily. The modified Brookfield viscometer has been used to measure viscosities in the 10-centipoise range at high temperatures. Data on the viscosity of NabF~KF-LiF-UF, (10.9-43.5-44.5-1.1 mole %) have been obtained with this instrument. The results are shown in Fig. 63. THERMAL CONDUCTIVITY W. D. Powers B. M. Burnett: L. Cooper S. J. Claiborne Reactor Experimental Engineering Division Three separate experimental de- terminations of the thermal conductivity of NaF-KF-LiF (eutectic) are presented in Table 25. These determinations were made with a modified, Deem-type apparatus that had previously been checked by making measurements of fluids of known conductivity. An error analysis of these thermal con- ductivity measurements indicated that no abnormal deviations existed. [t 1s next planned to measure the con- ductivity of NaF-KF-LiF-UF, and the ZrF,-bearing fuel mixtures. : TABLE 25 ThQrmal Conductivity of NaF-KF-LiF TEMPERATURE RANGE THERMAL CONDUCTIVITY (°C) [Beu/hr £t (°F/ft)] 634 to T47 2.9 490 to 590 2.6 567 to 670 2.5 The longitudinal heat flow ap- paratus has been adapted to measure the thermal conductivity of ligquid FOR PERIOD ENDING JUNE 10, 1952 metals contained in thin-walled capsules. A numerical analysis of the heat flow in the apparatus in- dicated unidimensional heat flow 1in the region where measurements are made. The apparatus is designed to minimize convective heat flow. Measure- ments on liquid sodium agree with the values given in the literature to within about 10%. HEAT CAPACITY W. D. Powers G. C. Blaleck Reactor Experimental Engineering Division The enthalpies and heat capacities of the following hydroxides have been determined by the use of Bunsen 1ce calorimeters: Barium hydroxide (470 to 900°C) Hy (liguid) - H o, (solid) = 0,20 T + 7 €y = $,20 £ 0,01 Potassium hydroxide (400 to 950°C) HT (liQUid) - Hooc {s0lid) = 0.35 T + 58 °, = 0,35+ 0.02 where H 1s in cal/g and T 1imn °C. At present the heat capacities of LiOH, Sr(OH),, a NaF-KF-LiF mixture with and without UF,, and a NakF-KF- ZrF,-UF, mixture are being determined. The following empirical heat capacity relation has been found to approximate the fused fluoride and hydroxide data: ¥ Cp“;“z 9, where ¢, 1s the specific heat 1in cal/g*°C, M is the average molecular weight, and m i1s the average number of species present {the metal and 149 ANP PROJECT QUARTERLY PROGRESS REPORT fluorine atoms and the hydroxyl radical being considered as separate species). This relation follows from Kopp’s rule and holds to within 20% for the liguids investigated. DENSITY J. M. Cisar, ANP Division The densities of several fuel mixtures have been measured 1n the new dry-box density system designed to obtain control of the atmosphere over the melt. The results for several fuel compositions are given in Table 26. Further density measure- ments on the ZrF,-bearing fuels are being made 1n order to obtain the uranium concentration vs. density information necessary to make the final fuel selection. VAPQOR PRESSURE 0F FLUORIDES R. E. Moore Materials Chemistry Division The vapor pressures of pure Bef, and of a typical Zrf,-bearing fuel mixture have been determined during the past qgquarter by using the ap- paratus and procedure previously described, (12 Beryllium Fluoride. Since Bel, had earlier been considered as a possible fuel compomnent, nation of 1ts vapor pressure was made. Considerable scatter of data in the low-temperature region was observed. At high temperatures (above 920°C), however, reasonable reproducibility was obtained. The vapor pressure data are shown 1in Table 27 and may be represented by the equation a determi-~ 8476.6 T(°C) log P (mm Hg) + 8.50. The heat of vaporizationwas calculated to be 39 kg-cal/mole, and the ex- trapolated boiling point 1s 1240°C. LTrF,-Bearing Fluoride Mixtures. Since fluoride mixtures containing ZrF, show considerable promise as fuels, vapor pressure measurements (l)AircraftNuclearPropulsion Project Quarterly Progress Report for Period Ending September 10, 1951, ORNL- 1154, p. 136, (2)Aircrafth’uclearPropulsion Project Quarterly Progress Report for Period Ernding December 10, 1951, ORNL-1170, p. 126. TABLE 26 Densities of Sowme NaF-KF«ZrF4nUF4 Mixtures MOLE % (o - g/DcECN_SItTY:—_ ocy TEMPEHA(TOUCH)E RANGE 5-51-42-2 p = 3.78 - 1,09 x 103 ¢ 430 < ¢t < 600 4.8-50,1-41.3-3.8 o= 4,27 - 1.63 x 10°3% ¢ 750 < t < 900 34,7-17.4-44,4-3.5 p = 3.78 - 0.909 x 1073 ¢ 580 < t < 900 35.0-17.6-49.9-2.5 P = 3.65 ~ 0.800 x 103 ¢ 580 < t < 900 150 26, 851 (1925). TABLE 27 Vapor Pressure of Beryllium Fluoride FOR PERIOD ENDING JUNE 10, 1952 TABLE 28 Vapor Pressure of NaB‘~KF~ZrF4-UF4. TEMPERATURE OBSERVED PRESSURE TEMPERATURE OBSERVED PRESSURE (°C) (mm Hg) (°¢) (mm Hg) 828 8.7 768 5.0 838 10.6 772 5.8 846 - 10.1 866 13.1 851 12,4 866 13.4 889 13.8 866 13.8 891 13.6 866" 14.5 893 14.5 901 15.7 893 17.0 901 19.3 896 17.1 306 _ 16.9 899 17.8 911 18.1 901 22.5 916 19.3 901 19.5 926 22.7 903 22,5 929 23,0 920 24. 4 931 23.0 925 . 28.5 954 - _ 31.4 935 30.5 956 ' 30.8 966 44,5 961 31.8 966 44.0 984 59.5 987 . 62.5 zation is 24.6 kg-cal/mole, and the 1019 - 89.5 extrapolated boiling point is 1530°C. 1019 88.5 : for a typical composition have been - performed over the temperature range 768 to 961°C. The mixture contained - 42 mole % ZrF,, 51 mole % KF, 5 mole % NaF, and 2 mole % UF,. The data - are given in Table 28, and may be - represented by the equation 5352. 1 70°C) + 5.842. log P (mm Hg) = - ~In determining the best straight line on the plot, the higher values were favored because it is believed that . some of the low values resulted from unsaturation. (3) (3)w The heat of vapori- Ro&ebush and A, L. Dixon, Phys. Hcv; Analyses of material distilled out of the apparatus during the measure- ments showed that the vapor above the liquid at the temperatures employed was virtually pure ZrF,. At 927°C, the estimated boiling point of ZrF4,(4) the vapor pressure of the mixture is only 24 mm Hg as compared with the 320 mm Hg to be expected from Raocult’s law. The considerable negative deviation from ideality i1s not unexpected since compounds such as Kj;ZrF,, K,ZrF., and KZrF, are known to be quite stable. - : (4)L. Brewer, The Cheaistry and Metallurgy of Miscellaneous Materials: Thersodynomics, N.N.E.S., Div, IV, Vol. 19B, p. 193 (1930). : 151 ANP PROJECT QUARTERLY PROGRESS PHYSICAL PROPERTY DATA Physical Properties Group Reactor Experimental Engineering Division Summaries of pertinent available data on the physical properties of fluoride salts and hydroxides presented in Tables 29 and 30. Most of the data were obtained by the Physical Properties Group at ORNL; however, pertinent physical property measurements obtained by other organi- zations are also included. The ORNL physical property measurements are being made to quickly supply the ANP Project with data of reasonable accuracy. The general accuracies assigned to the data presented in the tables are listed below. are 1. Melting point: within about +10°C. 2. Heat capacity: within about 110%. 3. Thermal conductivity: checks with the thermal con- ductivity device for liguids (made by using molten lead) dicated that the recent 1n- data fell within +15% of the known values, and an error analysis of the system (containing platinum, platinum- rhodium thermocouples) also suggested that the data are probably within 115% of the true values. results so far are so a definite ac- not yet be assigned. 4. Viscosity: preliminary, curacy can 5. Density: within about *5%. Some preliminary physical property data are presented; other preliminary data are merely referred to and are to be presented in the future. 152 REPORT HEAT TRANSFER IN SODIUM HYDROXIDE H. W. Hoffman Reactor Experimental Engineering Division J. Lones Experimental determinations were made of the heat transfer coefficients for molten sodium hydroxide flowing by forced convection through a heated tube. Measurements were made 1n the Reynolds modulus range of 5,000 to 12,000. Results show that molten sodium hydroxide may be considered an ordinary fluid - that is, any fluid other than the ligquid metals - so far as heat transfer 1s concerned. The system used to make the study consistedof two heated tanks separated by a long tube, part of which was the heated test section. The test section, a 24-in. length of the A nickel tube (0.1875 in. OD, 0.1175 in. ID), was heated by passing a high-amperage electric current through the tube wall. Chromel-alumel thermocouples (36 gage) were welded to the outside of the tube wall. The test fluid was moved by inert-gas pressure from one tank to the other through the test section and the fluid flow rate was determined by weighing the fluid as it left the test section. The heat transfer coefficients were calculated from the experi- mentally obtained tube-surface and fluid temperatures and the heat 1input by the equation: 9 A t -t s m 3 where g/Ais the heat fluxin Btu/hr-ft? through the tube wall based on the heat gained by the fluid in passing through the test section, t_ 1s the €ST TABLE 29 Physical Properties of Fluoride Salt Mixtures MELTING RANGE °C) HEAY CAPACITY {caljoC-g) THERMAL CONDUCTIVITY [Brushe-fe2 (°F/fc}] VISCOSITY {centipoise) NaF-KF-ZrF4-UF4 5-51-42.2 mole 2 4.8-50,1-41.3-3.8 mole % 34.7-17.4-44.4-3.5 nole % 35.0-17.6-44.9-2.5 mole % NaF-KF-LiF 11.5-42-46.5 mole % NeF«KF-LiF-UF4 10.5.43.5-44.5-1,1 mole 3 450 to 510(0? 450 vo 510(8) 255(¢) a55ic) | Prelimipary 0.46 at 500°C o = M= 2000 2000 2000 2000 2000 Gerity Mich. Gerity Mich, 2000 3012, 2000 Gerity Mich, Gerity Mich, 2000 2000 2000 9766 9766 OSU LABORATORY R, C. Briant* H. Buck* A, J. Miller* L. M. Cook W. B. Cottrell Bomar, Coobs Bomar, Coobs Bomar, Coobs Bomar, Coobs Bomar, Coobs Graaf Graaf Bomar Bomar, Cunningham, Leonard Graaf Graaf Bomar, Coobs Bomar, (Coobs Bomar, Coobs Johnson, Shevlin White Shevlin TECHNICAL ADMINISTRATION OF AIRCRAFT NUCLEAR PROPULSION PROJECT PROJECT DIRECTORY SECTION NO. K. Ergen* P, Blizard 11 A,B,C,D,E K. Ergen* I E,F D. Callihan I G H. Snell IT A STAFF ASSISTANT FOR BRADIATION DAMAGE Radiation Damage STAFF ASSISTANT FOR GENERAL DESIGN General Design STAFF ASSISTANT FOR ARE ARE Operations ARE Design STAFF ASSISTANT FOR ENGINEERING RESEARCH Heat Transfer and Physical Properties Ceramics Experimental Engineering STAFF ASSISTANT FOR METALLURGY Metallurgy STAFF ASSISTANT FOR CHEMISTRY Chemistry Chemical Analyses FOR PERIOD ENDING JUNE 10, o > = R. H. J H. JI = = =} R. Bl ° DI Miller* . Billington . Fraas* . Fraas* Buck* Bettis Schroeder PBriant* Poppendiek . Warde . Savage .. Manly#* . Manly Grimes* Grimes*® Susano 11y 111 111 111 111 I11 111 1952 L A C,D B,D F,G Q H,1,J,K D,H,1,J,K C,D,E,J,K,M, N,0,P,Q A,B,D,E,F,H K 185