ORNL-TM-3884 THE MIGRATION OF A CLASS OF FISSION PRODUCTS (NOBLE METALS) IN THE MOLTEN-SALT REACTOR EXPERIMENT R. J. Kedl This report was prepared as an account of work sponscred by the United States Government. Neither the United States nor the United States Atomic Energy Commission, nor any of their employees, nor any of their contractors, subcontractors, or their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. . NOTICE This report was prepared as an sccount of work sponsored by the United States Government, Nejther the United States nor the United States Atomic Energy & } Commission, nor any of their employees, not any of their contractors, |ubcontrsctoril. 1; ;hek employees, | makes any warranty, express or implied, or assumes any | Jegal liability or responsibility for the aceuracy, com- | ORNL—TM—3881£ i‘:‘ pleteness or usefulness of any information, apparatus, | _ product or process disclosed, or represents that its use would not infringe privately owned rights, - - Contract No. W-T405-eng-26 Reactor Division THE MIGRATION OF A CLASS OF FISSION PRODUCTS (NOBLE METALS) IN THE MOLTEN-SALT REACTOR EXPERIMENT R. J. Kedl Molten-Salt Reactor Program December 1972 OAX RIDGE NATIONAL LABORATCRY Oak Ridge, Tennessee 37830 operated by UNION CARBIDE CORPORATION ap" ; ‘ j For the { _ U.5. ATOMIC ENERGY COMMISSION DISTRIBUTION OF THIS DOCUMENT IS UNUIMITED - @ 4 ( FOREWORD ABSTRACT 1. 2. N L, dy) INTRODUCTION DESCRIPTION OF THE MSRE iii TABLE OF CONTENTS 2.1 General Description 2.2 Description of the Reactor 2.3 Description of the Fuel Pump FISSION PRODUCT EXPERIENCE IN THE MSRE 3.1 General Fission Product Disposition 3.2 Fission Product Disposition Measurements 3.2.1 3.2.2 3.2.3 3.3 The Difference Between the Fuel Salt and Gas Phase Samples Gamma Spectrometry of the Primary Heat Exchanger Core Surveillance Samples ANALYTICAL MODEL 4.1 Physical Basis of Model 4.2 Analytical Model L.2.1 - h,2.2 L.2.3 .2k h.2.5 h.2.6 h.2.7 4,2.8 L.2.9 L.2.10 Generation Generation Decay Rate Deposition Deposition Deposition Deposition 235 233 U and U Runs from Fission from Decay of Precursor Rate Rate Rate Rate Equation for CS Noble Metals on Noble Metals on on Heat Exchanger on Graphite on Rest of Fuel Loop: on Liquid-Gas Interfaces Solid Surfaces Liquid-Gas Interfaces 1k 1L 17 17 19 21 23 29 29 31 32 32 33 33 33 33 3k 3L 35 36 iv. 5. RESULTS FROM THE MSRE 5.1 Introduction 5.2 (Comparison of Measured Deposition on Heat - Exchanger to Theoretical 5.3 Comparison of Measured Deposition on Core Surveillance Samples to Theoretlcal 5.4 Comparison of Fuel Salt Samples with c® 5.5 Comparison of Gas Samples with c® 5.6 Time Constant for Noble Metals on Interfaces S.Tr.Miscellaheous Noble Metal Observations 5.7.1 Leminar Flow Core Surveillance Sample 5.7.2 Noble Metal Distribution in the MSRE 6. CONCLUSIONS AND RECOMMENDATTIONS Conclusions Recommendations T. REFERENCES APPENDIX A AP?ENDIX B 43 18 61 65 68 68 - T3 T3 Th 76 79 &, r v ¥ " ) FOREWORD The behavior of the ™oble metal"” fission products in the fuel salt is a subject of major importance to the design of molten-salt reactors. A considerable amount of data bearing on the subject was obtained from operation of the Molten-Salt Reactor Experiment, All or part of the data has been studied with various degrees of thoroughness by many people with different backgrounds and different viewpoints. Unfortunately, the chemi- cal and physical situations in the reactor were complex and the data ob- tained are not very accurate or consistent. Consequently, no one has de- veloped an explanation of the detalled behavior of noble metals that is acceptable to a majority of the knowledgeable observers. It is generally agreed that most of the fission products from niobium through tellurium are reduced to metals in the fuel salt, that they migrate to metal and graphite surfaces and to salt-gas interfaces, and that they adhere to the surfacés with varying degrees of tenacity. The details of the processes involved and the manner in which the noble metal particles may be affected by other processes in the reactof are subjects of frequent debate in which opinions vary widely. This report describes the author's interpretation of the data and explanation of some aspects of the operation of the reactor. Although others would analyze the data differently and would reach differ- ent conclusions concerning some of the mechanics,.we believe that publica- tion of this report will provide information helpful to the design of molten-salt reactor systems and to the development of a better understand- ing of the behavior of fission products in those systems. vi ABSTRACT The Molten-Salt Reactor Experiment (MSRE) is a fluid-fueled experi- mental nuclear reactor; consequently, fission products are dispersed throughout the entire fuel circulation system. One group of fission products, referred to as noble metals, exists in the fuel salt in the reduced metallic state. They are insoluble and unwet by salt. They de- posit on surfaces exposed to salt such as the Hastelloy_N-piping,and the moderator graphite. They appérently accumulate in a stable form on the liquid-gas interface in the fuel pump. The amounts of noble metals on these surfaces and on other deposition sites have been measured. These ‘measurements have been analyzed within the framework of mass transfer theory. The analysis has been found to correlate the data from these gsources in a unified manner. It is therefore concluded that the noble ~ metals do migrate in accordance with mass transfer theory, although some parameters still remain unevaluated. A hypothesis is presented to ex- piain some of the dramatic differences in reactor operating characteris- tics between the ??°U and ?33U runs. It is recommended that additional noble metal deposition experiments be conducted in a circulating salt loop. Keywords: Fission Products + Noble Metals + Mass Transfer + MSRE + Experienée + Bubbles + Foaming + Mist + Physical Properties + Entrainment + Off-Gas System + Void Fractions + Corrosion Products + Fluid Flow ‘(-) ol " ny 1. INTRODUCTION The goal of the Molten-Salt Reactor Program at Oak Ridge National Laboratory is to develop the technology for an efficient power producing, thermal-breeding (based on the Th-233y cycle) nuclear reactor for commer- cial use., The fuel in a molten-salt breeder reactor is fluid, and con- - sists of UF, and ThF, dissolved in a carrier salt mixture of LiF and BeF>. The liquidus temperature ranges from 800 to 9406F, depending on the exact carrier salt composition, with the nominal reactor operating temperature being ~1200°F. The fuel salt is pumped through a graphite-moderated core and then through a heat exchanger where it transfers heat to a secondary salt system, which then'genérates steam in another heat exchanger. One of the unique features of this concept is that the fuel is in the liquid state, This gives rise to many advantages, the principal one being poten- tially very low fuel cycle cost. The fluid fuel also creates a few lia- bilities, one of which is that the fission products are spread throughout the entire fuel loop and other hydraulically connected regions (e.g., off- gas system, dump tanks, etc.). A 7.3-MW(t) experimental reactor based on this concept was built and operated at Oak Ridge National ILaboratory. This reactor, the Molten-Salt Reactor Experiment (hereafter called the MSRE), first went critical in June 1965, Nuclear operations wére terminated in December 1969. Being an experimental reactor, it was subjected to a good deal of testing and observation. One of the principal efforts was to determine the distribu- tion of the various fission products in the fuel loop and connected re- gions, This-information is critical in the design of large central power stations [1000 MW(e)] where the heat geherated by fission products is sub- stantial. _ , o Fission products in molten fuel salt'can be grouped intb three prin- cipal types where .the mechanics of migration is the diétinguishing fea- ture — (1) salt seekers, (2) noble gases, and (3) noble metals, The salt- seeking fission products (which include Sr, Y, Zr, I, Cs, Ba, and Ce) are - the best behaved. They are soluble in a fuel salt and remain with the fuel salt in inventory amounts. The nobie gases are Kr and Xe, A great 2 — deal of work has been done to understand ndblé gas migration; particu- larly 135Xe because of its thermal neutron cross section of over 106 barns. The third group, the so-called "noble metals " Nb, Mo, Ru, 5b, and Te, is the subject qf this report. The noble metals are reduced by the UF3 in the fuel salt, and therefore exist in salt in the metallic state, They are insoluble in fuel salt and are unwet by it. Because of their incompatibility with salt they migrate to various surfaces (graph- ite and Hastelloy N) and adhere to them, They apparently also migrate to _gas-liquid interfaces and adhere to these in a stable manner. Noble met- als have been found and measured in fuel salt samples and gas phase sam- .ples on the surfaces of Hastelloy N and graphite surveillance specimens in the core, and on the fuel loop and heat exchanger surfaces. In this vreport we shall present a theory of noble metal migration based on conven- tional mass transfer concepts. We shall then analyze data from the above mentioned samples and measurements in the framework of this theory, and show that noble metals apparently do migrate from the fuel salt to their various depositories in accordénce_with the theory. A major thesis of ¥ this analysis is that noble metals migrate and adhere to liquid-gaé inter- faces, As such, they apparently have properties similar to insoluble sur- face active agents. This idea will be used to explain many of the obser- vations on fission product behavior in the reactor. It will also be used to suggest an explanation for the rather dramatic difference in reactor operating characteristics between runs made with 235U and 233U fuels. 2. DESCRIPTION OF THE MSRE 2.1 General Description The purpose of the MSRE was to demonstrate, on a pilot plant scale, the safety, reliability, and maintainability of a molten-salt reactor. - The operating power level was 7.3 MW(t). Because of its small size and other considerations, it was not intended to be a breeder, and no thorifim was added to the fuel. The fuel consisted of UF,; and UF; dissolved in a mixture of LiF, BeF,, and ZrF,. Its composition and physical proPertiés - are given in Table 2.1. All fuel loop components are constructed from " ; X ) $ [ Table 2.1. MSRE Fuel Salt Composition and Physical Properties Composition — LiF, BeF;, ZrF,, UF, (65.0, 29.1, 5.0, 0.9 mole %) Liquid type — Newtonian English Units ' Metric Units Liquidus temperature g13°F 434°C Properties at 1200°F (650°C) Density - 141 1b/ft3 2.3 g/cm? Specific heat 0.47 Btu/1b«°F 2.0 x 103 J/kg+°C Thermal conductivity 0.83 Btu/hreft+°F 1.43 W/m+°C Viscosity 19 1b/ftehr 28 kg/h°m Vapor pressure <0.1 mm-Hg <1 X 10™% bar 80.3 mole % 25U and 0.6 mole % 238y, Hastelloy N, which is essentially unwet by fuel salt under normal operat- ing conditions. The'nominal operating temperature was 1200°F with a 40°F temperature change.across_the core and primary heat exchanger. Figure 2.1 ié a schematic flow diagram of the MSRE, which will be described briefly here. More detailed descriptions of the reactor and the concept are available in Refs. 1, 2, and 3, The fuel loop consisted essentially of a centrifugal pump, a heat exchanger, and the reactor vessel. The nominal flow rate was 1200 gpm. The heat exchanger was a conventional U-tube type with the fuel salt on the shell side. Heat was transferred to a secondary coolant salt that in turn dumped it to the atmosphere via a large radiator."Dfiring periods of shutdown, the fuel salt was drained into either of two drain tenks. In addition, a third “drain tank contained a load of flush sait for rinsing the fuel loop pridr to any maintenance that was required. Note the use of freeze flanges in the priméry and secondary salt loops for easy disconnection of main compo- nents should they need replacement, and also the use of freeze valves in the drain lines. Off-gas from the pump bowl passed through & volume holdup, " charcoal beds, and then absolute filters before it was discharged up the stack. Characteristics of the reactor core and fuel pump influenced the OFF-GAS HOLOUP ABSOLUTE FILTERS STACK FAN b et etagessas = w < w0 = m x SAMPLE ENRICHER PumP --------- i o M WA'I"ER ST FIGURE 2.1. SCHEMATIC FLOW DIAGRAM OF THE MSRE ORNL-DWG 65-H410R LEGEND S FUEL SALT wmrve— COOLANT SALY . ............. HELIWM COVER GAS mwem = RADIOACTIVE OFF -GAS SAMPLER TO ABSOLUTE FILTERS COOLANT ORAIN TANK .. r $ . SODtUM FLUORIDE BED nt ) ol > mechanics of noble metal fission product migration, so they will be des- cribed in somewhat more detail in the following sections. The MSRE went critical June 1, 1965, and nuclear operations were terminated December 12, 1969. The reactor was critical for a total of 17,655 hr. Figure 2.2 is a brief histérical outline of the MSRE's power operation. Note particularly that for the first 2 1/2 years the reactor was operated with 23°U fuel. The fuel was chemically.processed and the 235y replaced with 233y, Then for the last 1 1/2 years, the reactor was operated with 233U fuel, Most of the results presented in this report will be from data obtained during the 233U rums. _ 2.2 Description of the Reactor A detailed view of the MSRE core and reactor vessel is shown in Figure 2.3. Fuel éalt entered the reactor vessel through a flow distribu- tion volute near the top of the vessel. It then flowed down through a l—in.—thick annular passage bounded by the reactor vessel and reactor core can and into the lower vessel plenum. The fuel then passed up through the graphite moderator region and out the top outlet pipe. The moderator assembly was composed of graphite stringers about 5 ft long and 2 in. square. The stringers had grooves cut longitudinally in the four faces, so that when the stringers were stacked together vertically, the grooves formed the fuel channels. The graphite was grade CGB (trade name of Union Carbide Corporation). It was unclad, in intimate contact with the fuel, and unwet by fuel salt under normsl MSRE operating conditions. In the bulk of the fuel channels (95 percent of them), the fuel salt velocity was about 0.7 ft/sec,_yielding a Reynolds number of about 1000. . The entrafice_to-the fuel channels through the moderator'support grid structure was rather tortuous and turbulence was generated that persisted for some distance up into the fuel channels. Nevertheless, the flcw‘is thought to have béen eésenfiially laminar ifi most of the length of the fuel channel. ' | | | | L Located near thercenter line of the core in a square array was an arrangement of three control rods_afid 6ne surveillance specimen holder. Details of the specimen holder are shown in Figure 2.k, It fias positioned vertically in the reactor and extended the entire height of the moderator region. At times when the reactor was shut down and drained, the FUEL POWER (Mw) REPLACE VALVES AND FILTERS 6 l RAISE POWER REPAIR SAMPLER ATTAN FLL POWER CHECK CONTAINMENT ,5 FULL - POWER RUN TEST CONTAINMENT n .....9 ROD QUT OFFGAS LINE CHECK CONTAINMENT i‘ 10 Aragu.d:yPguwNER }RE?LACEAIRLNE ll SUSTAINED OPERATION AT HIGH POWER } TEST CONTAINMENT FLUSHI':] ' =— MAIN BLOWER FALURE REPLACE MAIN BLOWER MELT SALT FROM GAS LINES REPLACE CORE SAMPLES f 8 RUN WITH ONE BLOWER ‘ > WNSTALL SECOND BLOWER REPLACE CORE SAMPLES 0 2 46 8 ¥ POWER (Mw)} ORNL-DWG 69 - 7293R2 XENON STRIPPING EXPERIMENTS 1] WSPECTION AND MAINTENANCE RE{‘LACE CORE SAMPLES TEST AND MODIFY FLUORINE DISPOSAL SYSTEM PROCESS FLUSH SALT PROCESS FUEL SALY LOAD URANIUM =233 REMOVE LOADING DEVICE 233y ZER0 - POWER PHYSICS EXPERIMENTS INVESTIGATE FUEL SALT BEMAVIOR CLEAR OFFGAS LINES REPAIR SAMPLER AND CONTROL ROD DRIVE 233); DYNAMICS TESTS INVESTIGATE GAS N FUEL LOOP HIGH-POWER OPERATION TO MEASURE 233y o, /g ANVESTIGATE COVER GAS, XENON, AND FISSION PRODUCT BEHAVIOR ADD PLUTONIUM IRRADIATE ENCAPSULATED U MAP F.P. DEPOSITION WITH GAMMA SPECTROMETER MEASURE TRITIUM, SAMPLE FUEL REMOVE CORE ARRAY PUT REACTOR IN STANDEBY FIGURE 2.2, HISTORICA.L OUTLINE OF MSRE'S OPERATION L ¥ 2 ORNL-LR-DWG GI097R1A - ¢ _ FLEXIBLE CONDUIT TO GRAPHITE SAMPLE ACCESS PORT /; sy CONTROL ROD DRIVES ’ COOLING AIR LINES . ACCESS PORT COOLING JACKETS FUEL OUTLET R it REACTOR ACCESS PORT CORE ROD THIMBLES A »-!‘«, figfuaED%mmggg SAMPLES LARGE GRAPHITE SAMPLES il |'15 OUTLET STRAINER CORE CENTERING GRID A ] FLOW DISTRIBUTOR VOLUTE s GRAPHITE ~ MODERATOR STRINGER R - FUEL INLET -/ : ,_ ~— CORE WALL COOLING ANNULUS REACTOR CORE CAN ] ' REACTOR VESSEL — TR ) fl"\“’l-ll . ‘\ Q .511 r" $ \ > \f‘ . ANTI-SWIRL VANES 4 e _ DERATOR VESSEL DRAIN LINE ggpFEORRT GRID FIGURE 2.3. MSRE CORE AND REACTOR VESSEL S ORNL-DWS 68-12146R TOP GUIDE ~HASTELLOY N TENSILE SPECIMENS ROD _ENTIRE SPECIMEN ASSEMBLY IS ENCLOSED IN A PERFORATED 4 BASKET AS SHOWN IN PLAN ™ VIEW A-A FLUX MONITOR TUBE - ///1/81m 0.D. 7”7 PERFORATED BASKET 3/8 din. DIAMETER HOLES ON 9/16 in. CENTERS TRIANGULAR PITCH 2 in. TYPICAL OUTLINE OF GRAPHITE CHANNEL CONTAINING CORE SURVEILLANCE SAMPLER LOWER GUIDE G-GRAPHITE SPECIMENS N-HASTELLOY N TENSILE SPECIMENS ROD SECTION A-A INCLUDING BASKET AND SHOWING FLOW AREA FIGU?E 2.4. MSRE CORE SURVEILLANCE SPECIMENS AND CROSS SECTION OF FLOW GEOMETRY A, & & " / o} o1 9 surveillance specimen holder was removed from the core and taken to a hot cell where the samples could be removed. Samples consisted of graphite specimens, Hastelloy-N rods for tensile strength tests, and Hastelloy-N flux monitor tubes. The specimen assembly was made so that only part of the samples need to be removed, and the remaining samples along with new replacements could be reinserted into the reactor for further irradiation. Fission product deposition measurements were then made on the graphite and Hastelloy-N specimens, which had been removed from the sample holder. 2.3 Descriptioh of the Fuel Pump The fuel pump turned out to be a very important component in under- standing noble metal migfation in the MSRE. A detail drawing of the pump is shown in Figure 2.5. Its rated capacity is 1200 gpm of fuel salt at 418.5 ft of head. The volute is completely enclosed in another vessel referred to as the pump bowl, which served a variety of purposes. It contained the only free liquid surface in the system and therefore served as an expansion volume for salt; It also contained the 135%e stripping system, a salt level indica@or, and‘thé‘fuel salt sampling facility. The fuel expansion capacity of the pump bowl was more than adequate for the normal operating rangés of the MSRE, but was not adequate for some postulated accident conditions. Thérefore, the pump bowl was provided with an overflow tank that would fi;l if the salt level in the pump bowl reached the level of'the.ovérflcw pipe. The normal operating helium pressure in the pump bcfil, which was dléo the pump suction pres- sure, was about 5 psig. [ - B | The xenon stripper was a gas-liqui& contacting device. A toroidal spray ring containing many small holes'(1h6 - 1/8 in. holes, and 145 - 1/16 in. holes), sPr&yed:the salt through the gés-phase.' The salt flow, which resulted in a mesn jet velocity of T.2 ft/sec, was estimated (not measured) to bé at a rate of about 50 gpm. The xenon rich cover gas was purged from the pump bowl by a continuous flow of clean helium from (1) the pump shaft purge, (2) two bubbler level indicators, and (3) the reference pressure line purge for the bubblers. The purge gas then went to the off-gas system described earlier. The overflow tank also had a bubbler level indicator and its purge also went to the off-gas system. 10 . ORNL-OWG €9-10172R gHAFT URGE aTs OFF-GAS ' . LINE 1 BUBBLER LEVEL : O inorcaToR ["U Xe STRIPPER 100 (B W SPRAY RiNG v Q] A [T5 RO wof|l__ " , [ _\\g":” 40 N o ’ - I e 2 W 3O | -;").l A 3 0 o x 7 SAMPLE = LEVEL Sl L2 DISCHARGE CAPSULE SCALE PUMP A _ CAGE (%) BOWL O < SALT L, <— RADIOACTIVE GAS SUCTION <= CLZAN GAS ) f 7\ PUMP SUCTION PIPE OVERFLOW TANK JL DIMPLE_A \UJ__ \_/ FIGURE 2.5. MSRE FUEL PUMP AND OVERFLOW TANK 1 a0 i ‘were measured in this region. 11 The spray jets impinging on the salt surface generated large amounts of helium bubbles and fluid turbulence in the pump bowl., These bubbles manifested themselves in several ways in the operation of the MSRE. 1In - one sense their présence was unfortunate because it complicated the under- standing of many observations in the reactor, but in another sense it was fortunate because it led to the suggestion of efficient ways of removing fission products (particularly noble gases, but possibly also noble ‘metals) from future molten-salt reactor systems. Let us examine the behavior of gas bubbles in the salt in the pump bowl by first considering the carryunder of bubbles. Note from Figure 2.5 that there are two bubblers and they are at different depths. By differ- ence in reading between the bubblers, one can deduce information on the average fluid density between them. Void fractions as high as 18 percent (&) and would rise to the surface, but some of them were small and would be Certainly most of the bubbles were large carried down into the pump suction to circulate with the fuel salt. Estimates of the size of the small bubbles indicate they were less than 0.010 in. in diameter. The amounts of bubbles circulating with fuel salt could be estimated by analysis of (1) level changes in the pump bowl, (2) nuclear reactivity balances in the core, (3) sudden pressure release tests, (4) small induced pressure perturbations, and (5) other less direct observations. The general conclusion from these analyses is that the volume fraction of circulating bubbles in the fuel loop during the 233U runs was 0.0002 to 0;000h5,(h?5)and the void fraction during the 233y runs was 0.005 and 0.006;(h)A hypothesis is presented in Section 3.3 to explain this rather large change in circulating void fraction between ‘the 235U and 233U runs. The same hypothesis is used to explaln the dif- ference in overflow rates discussed next. Now consider the bubbles that rise to the surface in the pump bowl. There are strong indications that they produced a froth with a high liquid . content on - the salt surface. For instance, there was a constent flow of fuel salt from the pump bowl to the overflow tank even though the indi~ cated salt level in the pump bowl was well below the overflow pipe. | Periodically this salt was forced back into the pump bowl by pressuriz- ing the overflow tank with helium. This transfer must have been due to 12 froth flowing over the lip of the overflow pipe, or to a mist of salt drifting into the overflow pipe, or to both. Overflow rates during the 235U runs ranged from 0.4 to 1.5 1b/hr. During the 233U rums they ranged - from L to 10 1b/hr with excursions up to TO 1b/hr.(h) A simplé comparison u (W) indicates that a rain of "more than the hardest torrential downpour would have been required to match the lowest overflow rate during the 233y runs. A reasonable conclusion is that froth with a high liquid con- tent flowing down the overflow pipe accounted for the high overflow rates. There may also have been mist in the pump bowl. It is well known that the mechanical action of a high velocity jet impinging on a liquid surface - will generate a mist. Bursting bubbles are also known to produce mist. There are physical indications from the reactor that a mist was present in the pump bowl. PFor example, the off-gas line plugged periodically vhich was probably due to freezing of salt from a mist. Strips-bf metal suspended in the gas phase of the pump bowl were covered with small -drop- lets when retrieved (see Figure I of Ref. 4). Later when we discuss the analysis 6f gas samples taken from the pump bowl, we will explain the results by theorizing that a salt mist was probably being sampled. . ‘A facility was provided for taking salt samples from the bowl of the fuel pump. This sampling facility was used for a variety of purposes including (1) taking fuel salt samples, (2) teking gas phase samples, - (3) adding uranium to the fuel salt, (4) adding chemicals to the fuel ‘salt to control its oxidation-reduction state, and (5) exposing materials to fuel salt for short periods of time. The sampling facility was quite complex,'&nd included.a dry box, isolation valves, remote handling gear, - shielding, instrumentation, etc. A detailed description of the entire 'facility is not necessary for the purposes of this report,'and only the sample station in the pump bowl is shown in Figure 2.6. Note that the sample station is enclosed by an overlapping shield arrangement. It was desired to take the salt and gas samples under rather quiescent con- ditions, and this shield was intended to prevent the mist, foam, and fluid turbulence generated by the 5pray ring from penétrating too aggres- sively into the sampling region. The shield does not, however, prevent free movement of salt through the sample region. The overlapping portiofi of the shield is not closed, and also, the shield is elevated.off the s ] wh ») NORMAL OPERATING SALT LEVEL - CAGE FOR SAMPLE CAPSULE—— FIGURE 2.6. '¢?:=flr-- Di Taa. ' % Y ~3 I~ SHIELD 13 ORNL-DWG 67-10766R SAMPLE TRANSPORT CABLE TRANSPORT PIPE LATCH ASSEMBLY NORMAL POSITION TOP OF FUEL PUMP //SAMPLE W CAPSULE == A CanLE fi¢i;;;£;;5§§:»-CAPSULE it ! 00 i 1 CAPSULEQAQXQ;;L' Y ~SALT INTAKE CAGE" S TN O 1 SHIELD SECTION A-A T‘SAMPLE CAPSULE (LADLE TYPE) 3/16 in. MAX | ;{ 1001 2 3 4 - INCHES- PUMP BOWL SAMPLE.STATION, 1L bottom of the pump bowl. The sample capsule shown is typical of thg first kind of capsulés used in the MSRE. It is referred to as a "ladle capsule." It is simply a container with open ports in the side. During normal éampling5 it was lowered with the sample transport cable until it was well below thé salt surface. After filling, it was lifted out of the salt and suspended for & period of time a short distance up in the trans- port pipe in order to freeze the salt. It was then withdrawn, isolated from the fuel system, and taken to a hot cell for analeis. During the sampling period, a small purge of helium was maintained down the trans- port pipe. Fig. 2.7 shows other sample capsules used which will be des- cribed lsater. 3. FISSION PRODUCT EXPERIENCE IN THE MSRE 3.1 General Fission Product Disposition In this section we will discuss in a qualitative way the general . disposition of fission products in the MSRE, and the techniques used to measure this disposition. The emphasis®will be on noble metals. We will also discuss the rather dramatic differences in the reactor operating characteristics when fueled with 235U end 233U, and propose a reason for this difference, A systematic way of classifying fission products in the MSRE, based on their migrational characteristics, is as follows: (1) salt seekers, (2) noble gases, and (3) noble metals. As a group and under normal MSRE operating conditions, the salt seekers are the best behaved of all fission products. Examples of salt ' seekers are Sr, Y, Zr, I, Cs, Ba, Ce and Nd. Unless affected by migrational characteristics of their precursors, they remain dissolved in the fuel salt in inventory quantities. Consider the following gen- eralized beta decay scheme of fission pioducts: 2) 1) @ @ @ & (3 *r - Rb - S > Y »> Zr > Nb = Mo = (3) (3) (3) (3) (3) (&) (1) Te - Ru - Rh - Pd~» Ag - C4d > In -~ -l b " & VOLUME 20cc-—'_\_i_. N w o Li, BeF, 15 ORNL~-DWG 67-4784 ' ‘ ORNL-DWG 68-6079 | .——STAINLESS STEEL CABLE AL ERLRRRY /3/4-in.-oo NICKEL TUBE 0.020 -in. WALL Tl | _.—Y4~in. OD NICKEL T N T S T T T T T T YT TITLTTLTST TTOTTTTINTS TSI eI T T TTLITT | _— Yg-in. Ni TUBING LT ITTRIY / NICKEL CAPILLARY 46 ~in. WELD ROD S —— FLARED CUP {(a) Two Modifications of Freeze Valve Capsules ORNL-DWG 70-6758 CUT FOR , SAMPLE REMOVAL - he—— NICKEL OUTER TUBE s ' > VACUUM - COPPER INNER CAPSULE = | BALL RETAINER -BALL WITH SOFT SOLDER SEAL \ l// “|| ——COPPER NOZZLE TUBE ABRADE FOR ) SAMPLE REMOVAL: {b) Double-Wall Freeze Valve Capsule . FIGURE 2.7. FREEZE VALVE SAMPLE CAPSULES 16 3y (3) (3 @) (&) (1) (1) "Sn > Sb > Te > I » Xe -+ Cs -+ Ba = (1) (1) (1) (1) la » Ce -+ Pr - NdA - where (1) - salt seeker, (2) (3) (L) noble gas, noble metal, and noble metal or salt seeker depending on oxidation-reduction state of fuel salt. Note that Kr (a noble gas) is a precursor of Rb, Sr, Y and Zr (salt seekers) and also that Xe (a noble gas) is a precursor of Cs, Ba, La and Ce (salt seekers). Also note that Sb and Te (noble metals)-ére precursors of I (a salt seeker). The behavior of precursors of?éalt seekers may dramatically affect the ultimate disposition of the salt seekers them- ' selves. TFor instance 2%Zr has a yield of sbout 6.2 percent but its pre- cursor, 95Kr, has a yield of only 0.007 percent and is very short lived. Therefore any migrational tendency of the 95Kr will have little effect on the ?°Zr. Indeed the complete 35zr inventory was found in the fuel salt.(6) As opposed to this, consider the salt seeker 13705 with a cumulative yield of 6.15 percent. Most of its yield comes from the decay of 137Xe that has a yield of sbout 6.0 percent and a half life of 3.9 min. Xenon can be transferred to the off-gas system and also can diffuse into the porous structure of the graphite. Accordingly, only 80 to 90 percent of the 137Cs inventory was found in the fuel salt( and significant quantities were found deep inside the graphite where.it (7,8) deposited upon decay of its precursor.¥ In conclusion, the salt seeking fission products are well behaved and remain dissolved'in the circulating fuel salt in inventory quantities except when influenced by the behavior of their precursors. | ' There are only two noble gas fissicfi products, Kr and Xe. They appear, however, in over 30 mass number decay chains and therefore significantly affect the general fission product disposition. Notable among these fission products is 135%e with its large thermal neutron ¥The amownts and concentration profiles of salt seekers inside the graphite have been correlated with theory quite well in Ref. (9) for the case where the noble gas precursor is short lived. n wcl - o K9 17 cross section of 3 x 106 barns. Because of 135Xe, a great deal of wofk has gone into quantitgtively understanding noble gas migration in the MSRE. This work has been reported in the semiannual reports and other documents. The noble gases, particularly xenon, are very insoluble in fuel salt; therefore, they are readily transported into any available gas phase such as the circulating bubbles, the gas space in the pfifip bowl, and the pores of the moderator graphite. The third group of fission products is the so-called "noble metals." They are reduced by UF3 in the fuel salt and exist in the reactor environment in the metallic state, hence the name. Examples are Mo, Ru, Ag, Sb, Te and sometimes Nb. If the fuel salt is in a well reduced state, then the Nb exists as a noble metal, but if the fuel is more oxi- dized, it exists as a salt seeker. Noble metals have been found throughé out the entire reactor fuel salt loop. They have been found in large gquantities in fuel salt and gas phase samples from the pump bowl. 'They have been found in large quantities on the Hastelloy-N and graphite~core surveillance samples, and on the primary'heat exchanger tube surfaces and - loop piping surfaces. They have been found at various locations in the off-gas system. In this report we will look quantitatively at the noble metals in these depositories, within the framework of mass transfer theory, and try to developla unified‘model of noble metal migration in the MSRE. 3.2 TFission Product Disposition Measurements —3;2.1 Fuel Salt and Gas Phase Sampies Three typés of capsules were fised to obtain fuel salt samples from the MSRE pump bowl sample station, and these are shown in Figures.216 and 2.T. Numerous other sampling devices were used for special tests but only thoée_illustrated were used on a routine basis. The ladle capsule is the simplest and was the fifst used for sampling fuel salt. This capsule is illustrated in Fiéufie 2.6 in the process of taking a salt sample. It is simply a small container with open ports on the side to allow salt to enter. During the course of fuel salt sampling from the reactor, it was found that s mist of salt, which was heavily contaminated with noble metals, existed over the salt pool. Presumably this mist was generated by 18 the mechanical action of the spray impinging on the salt surface and by bursting bubbles. When the ladles were lowered to take a salt sample, the mist adhered to the capsule end grossly contaminated the sample. "Freeze vg1ve“ capsules, which had previously been developed for taking gas samples, were therefore used to take salt samples in the hope of allevigting this problem. Two modifications of these freeze valve capsuies are shown in Figure 2.Ta. Each capsule contains a vacuum which is held by a frozen salt sesl around the capillary entrance tube, The selt seal is designed to melt after a delay time long enough to allow the capsule to be completely submerged in fuel salt before 1t opens. The inside surface of the capsule is therefore not contamlnated W1th the salt -mist when taking a fuel salt sample. Freeze valve capsules for taking salt samples were used first in run 14. The noble metal concentration measured in salt from these capsules was generally'abofit two orders of magnitude less than in salt from the ladle capsules.. Note that in both modifications of the freeze valve capsule, the sealing salt remains inside the capsule, and puddles over the capillary nozzle. When taking gas phase samples, the capsule is lowered into the pump bowl until the valve thaws and the sample is taken, then it is withdrawn to & cooler region so the salt puddle may again freeze and contain the sample. These freeze valve capsules did not exclude mist and scum on the salt surface from adhering to the outside surfaces of the capsule. During capsule processing the outside surfaces were always well leached to remove this material. Never- theless, the question of transfer of contamination from the outside sur- face to the inside materisls during chemical processing always remained. This worry led to the development of the double walled, freeze valve capsule.shown in Figure 2.Tb. It is basically a freeze valve capsule but it is doubly contained, and can be used for salt or gas phase sampling. During processing, the nozzle and top of the outer container are cut through. The inner capsule containing the sample then falls away from the outer container and is free from mist and scum., There was little difference in the measured noble metal concentration between samples taken = with freeze valve and with double-walled freeze valve capsules. 1§ ] Xy ) 19 These’three kinds of capsules were routinely used for taking fuel salt and gas phase samplee. In the analytical work presented in this report, only data from the freeze valve and double-walled freeze valve capsules will be used, and they will be used with equal weight. .Ladle samples will not be used because of the problem of mist and scum contami- nation. On all plots that follow, the data points will specify whether the capsule was a freeze valve or a double-walled freeze valve type. Another point to note is the difference in volume of the two capsule types. The freeze valve capsule when full holds 50 to 60 grams of fuel salt, but the double-walled capsule when full holds only 14 to 15 grams. During the sampling procedure, a purge of helium was maintained down the sample transport pipe. This should make little difference to the salt samples but could be a significant parameter for the gas samples. The helium purge varied from 575 to T5 standard cm3 /min depending on the sample. All samples were taken with the higher purge rate before sample 19-16; after that the salt samples had the higher purge rate and the gas. eamples the lower purge rate. - Only four freeze valve samples of fuel salt were taken during the 235y yuns (during run 14) but meny freeze valve and double-walled freeze valve samples were taken during the 233y puns. The 233U runs will there- fore be discussed most extensively. Many ladle samples of salt were taken during the 233y runs because they continued to give good results on salt- seeking fission products. Both the salt and gas samples were analyzed radiochemically. This technique was used to determine both the identifi- cation and the amount of the isotopes present 3.2.2 Gamma Spectrometry of the Prlmary Heat Exchanger ‘The amounts of certain noble metal fission products dep051ted on the pr;mary.heat_exchanger tube surfaces were measured by gamme ray spec- trometry. The technique consisted of looking at & spot on the heat ex- changer with a colllmated detector and measuring the gamma energy spectrum | emitted from that spot. The resultlng spectra were used to 1dent1fy and measure quantitatively the fission products present. ~ The first gamma scans were made after run 1k (last 23%U run) and were, to an extent, exploratory.in nature; i.e., an attempt was made to see if 20 specific fission prodfict information could be extracted quantitatively from the tremendous amount of background gemma radiation emitted by an operating reactor system. Briefly, the equipment consisted of a highly collimated lithium-drifted germanium diode detector that was coupled to a 400 channel analyzer to determine the energy Speétra. All scans were takenrwith‘the reactor shut down and either drained 6r filled with flush salt. The equipment was mounted ofi a portable maintenance shield, so gammé scans‘could be taken at many locations on the heat exchanger. All together ébout 100 spectra were determined. Most of these were from the heat exchanger but a few were from other components. The equipment was calibrated so that the measured count rate could be reduced to'atqms of fission,product per unit area of tube surface. Data processing was done by hand and'was quite tedious. It was possiblé to isolate quantitatively four noble metals from a typical spectrum (2%Mo, !03Ru, 132Te, and 95xD). More details of the equipfient and procedures éan be obtained frbm Ref. 10. At the end of the experiment it was concluded that the quality of the data and'its potential applications were sufficiently promising to warrant improving the system and repeating the experiments during later runs. After considerable equipment and calibration procedure improvement, the gamma spectrometry measurements were repeated after run 19 (233U). An improved Ge(li) detector and a 4096 channel analyzer were used. Data processing was done with a computer program developed for this purpose. Precise alignment was achieved by use of a laser beam and surveyor's transit.- Altogether_some 1000 spectra were measured, many of which were teken with the reactor at different power levels (a few watts to full _ §0wer); Another 400 spectra were taken for calibration purposes. Details of the equipment, calibration procedures, and data analysis can be obtained from Ref. 11. Gamme spectra were obtained from other fuel loop components besides the heat exchanger, such as the pump bowl, off-gas line, and main loop piping. The principal intent of the experiment, how- ever, was to measure fission product deposition quantitatively in the primary heat exchangef. This was the only component for which an absolute calibration of the detector was made. Therefore, these are the only data that will bé analyzed in this report. The principal sPectfa of the primary heat exchanger were obtained after run 19, although some »% &) 21 prelimihary data were taken after run 18. The data were taken with the reactor shut down and drained. Altogether, quantitative information was obtained for 10 noble metal isotopes from these data. Figure 3.1 is typical of the kind of data obtained from this experiment. It shows the amount of 132Te per square centimeter of tube surface plotted against a longitudinal representation of the heat exchanger. The total range in concentration is from 0.3 x 1012 t0 0.75 x 1012 disintegrations/min—cmz. The higher value seems to be associated with the baffle plates and their windows, whereas the lower value is associated with the crossflow part of the heat exchanger. In this report the lower value was considered to be most representative of the heat exchanger and, as an example, 0.3 x 1012 disintegrations/min-0m2 was chosen for !32Te., The quality of data from the gamma scens after run 14 were not nearly as good as shown in Figure 3.1, and a sort of weighted average of all the data was used. 3.2.3 Core Surveillance Samples The core surveillance specimens were briefly described (see Sect. 2.2) and illustrated (see Figure 2.4) earlier. Periodically, when the reactor was shut down and drained, all graphite and Hastelloy-N specimens were removed and the amounts of fission products deposited on the surfaées and the interior of some were measured. Radipchemical techniques were used to determine the identity and amount of the isotopes present. Unfortunate- ly, the fluid dynamic conditions in the surveillance specimen holder are not well known because the flow passages are so complicated. These fluid dynamic difficulties are discussed in detail in Appendix B where we estimate the mass transfer coefficient. Briefly, let us just say that the surveillance specimens feature inside corners, outside corners, fluid entrance and éxit_regions, and possible stagnation areas. Superimposed on these is a flow that is only marginally turbulent (Re ¥ 3000), These difficulties were & result of the_many different kinds and geometries of specimens that had to be incbrporated into a-vefy confined space. The measured noble metal deposition'data on graéhite thatiwill be anal&zed in this report came from the.surfacegfexposed'to fuei salt, although some noble metals were also found on the inside surfacés that presumably weré ORNL-DWG 70-7365 SPACEFI?' '_ . 5 .LIJ ‘ 1 T T T 3 0.800E12 |- | S S % = ° — ,_i [ .:.I . . * . Jr_ o » . “. & e ® ° * o & 0600ER |— —- e e s e et o & —— %W . - s . [ - . ° T e I © $ 040012 |- -Ly—peul } ol A e b T e } § £ s | — SR e | wu - } - ! T 1 ' | : | HTR. PLUG I-IITR. PLUG I HTR. PLUG ! l SPACER FUEL SALT FLOW —» Ie— - —— | ! FIGURE 3.1. F1SSION PRODUCT DEPOSITION; POSITION HOLE IN SHIELD PLUGS MSRE HEAT EXCHANGER 1 32¢e01) cc =y 2y o a) 23 not exposed to salt. The measured noble metal deposition data on Hastelloy N came from the dosimetry wires. Noble metal deposition measure- ments were also made dfi‘the perforated Haséelloy N cage but these will not be analyzed because of the almost impossible job of estimating th%;mass transfer coefficient. In Appendix B, fihe mass transfer coefficient to both the graphite and the Héstelloy N is estimated to be about 0.25 ft/hr. There is a Significantrunéertainty associated with this value for the reasons described above, The mass transfer coefficient from the fuel salt to the heat exchanger surfaces is a much better known number than to the surveillance specimens. Figures 3.2 and 3.3 show typical data obtained from these specimens. 3.3 The Difference Between the 235U and 233U Runs Following run 1k the 235U fuel was removed from the carrier salt (and flush salt) and the reactor was refueled with 233U, Chemical processing of the salt was done at the MSRE site, and the process is discussed in detail in Ref. 12. The basic process was fluorination of the fuel salt and removal of the uranium as a volatile fluoride. The corrosion rate on the process tank during processing was high and has been estimated to be sbout 0.1 mil/hr. The tank was constructed of Hastelloy N and the corro- sion produéts were-Nin, FeF,, and CrFy. Following the fluorination step, it was therefore necessary to remove the corrosion products from the salt. This was done by reduction with hydrogen and zirconium powder and subse- quent filtration. Removal of 23°U from the salt was essenfiially complete. ~ The carrier salt was then returned to the rea¢tor system and loaded with 233y, BRwn 15, the first 233y run, was concerned with the ZEerO-power physics experiments’with thié new fuel, During run 15 a significant change in 0perating'characteristics of the reactor occurred and persisted.until the reactor was pérmanently shut down."During the 239U runs, the volume fraction of circulating bubbles in the fuel loop was determined to be between 0.0002 and 0.00045 (see Ref. 5), end the overflow rate from the pump bowl to the overflow tank ranged from 0.4 to 1.5 Ib/hr, with essentially no overflow excursions following beryllium additions. Beryllium was periodically added to the fuel salt, primarily to reduce some UF, to UF3; and control the oxidation state of the fuel. During the 233U runs the volume fraction of circulating bubbles DEPOSITION (dpm/cm2) 1012 o1t 1010 ]010 10° 108 2L ORNL-DWG 70-15013 CGB IMPREGNATED CGB POCO PYROLYTIC PARALLEL TO PLANES PYROLYTIC PERPENDICULAR TO PLANES WIDE FACE NARROW FACE DOUBLY EXPOSED O CWax PYil o —3 O PY] 10 - 20 30 40 - 50 60 4 70 DISTANCE FROM BOTTOM OF CORE (in.) 99 29 FIGURE 3.2. DEPOSITION OF 2°MO AND '2oTE ON GRAPHITE CORE SURVE!LLANCE SPECIMENS &} -t &) » Y] DEPOSITION (dpm/cm?2) 25 i ORNL-DWG 70-15014 11 10 ]010 10° o w0 2 30 40 50 . 60 70 - DISTANCE FROM BOTTOM OF CORE (in.) | FIGURE 3.3. DEPOSITION OF ~omo, '*%7e, '2%¢, 103y, 196y anp Pzr- ON HASTELLOY N CORE SURVEILLANCE SPECIMENS AFTER 32,000 MW HRS 26 in the fuel loop was 0.005 to 0.006, and the overflow rate ranged from b to 10 1b/hr with excursions up to 70 1b/hr following a beryllium addition. The mean cireulating void fraction had gone up by & factor of about 20 “and the overflow rate had gone up by a factor of about 10 in the 233U runs as compared to the 235y runs. Reference 4 provides a detalled discussion of these'operatihg variables and others. The composition of the fuel salt remalned nominally the same for both fuels, although the uranium concentra-. tlon was reduced from 0 .9 mole percent to about 0.2 mole percent prlmarlly because the 233U was not diluted with 238U, This resulted in a lowering of salt density by 3 to 4 percent. Other physical properties (viscosity, surface tension, etc.) might be expected to change an equivalently small amount. The rate of bubble injection from the pump bowl to the loop, and the overflow rate are certainly functions of these varisbles. In my opifiibn,.the change in reactor operational parameters is much too great to be explained by such small changes in physical pfoperties. Actually, there is evidence that the bubble ingestion phenomenon was near a thresh- - old region. This was indicated by a steep change in void fraction when the pump speed was changed s small amount.* It has been speculated that the small changes in physical properties were coupled in some way to the 1ngest10n threshold to yield the high void fractions during the 233y runsg ) This suggestion may account for part of the increased void frac- tion, however, I believe that it accounts for only a sfiall amount of the increase, The question then is - Why the difference in the above parameters when fueled with 23°U and 233U. A clue is given during the initial history of the 233y runs. Prior to the start of run 15, flush salt was circulated in the fuel loop for about 40 hr. There was no sbnormal behavior during this period. Both the circulating void fraction and the overflow rate . were consistent with what had been expected from past opérating history. The flush salt was drained end the fuel salt was added. Circulation was. started and again no abnormal behavior was noted. After about 14 hr of *The fuel pump was powered by a variable frequency unit for a period of time during the 233U runs to 1nvest1gate the effects of circulating voids on !3%Xe behavior. h " e 4+ €) 3 27 circulation, a beryllium rod was added for the purpose of reducing part of the U to U3*, About 2 hr later, the salt level in the pump'bofil began to rise indicating increased bubble ingestion into the fuel loop. When it peaked out, the circulating void fraction was about 0.5 percent. A short time later the overflow rate spparently went up to a highér than normal value (v L 1b/hr). Detailed records are not available because the data logger was not fully operable during this period. When the beryllium rod was removed after 12 hr of exposure to salt, 10.1 grams of beryllium had been dissolved. The circulating void fraction and overflow rate- remained high for 10 hr of circulation until the fuel salt was drained into the dump tanks. The above occurred before the reactor went critical with 233y, A suggested explanation for this behavior is as follows. During the early operational history of run 15, the fuel salt was in a more oxidized state than expected.(l3) The direct evidence that these fluorides were there and that the fuel oxidized is as follows: (1) The corrosion rate on the fuel loop was high during run 15. (2) During the 23°U runs, 2°Nb behaved as & noble metal. During the initial 233U runs, it behaved as a salt seeking fission product, indi- cating the fuel was much more oxidized;(lh) (3) When the beryllium capsule was removed from the salt, a thick crust was found on the nickel cage enclosing the capsule. The crust was predominantly salt, but the residue after extracting the salt was preé (13) Similar dominantly iron with small amounts of nickel and chromium. crusts were found on other beryllium capsules added during run 15. The perforated capsule for beryllium additions during the 235y yruns usually came out of the pump bowl relstively clean and did not have iron deposits. The beryllium then, rather than reducing the uranium ifi the fuel, apparently reduced the iron and hickei corrosion'product fluorides to the metallic state and they formed a scum on the surface of the fuel salt in ' the pump bowl. A hypothesis of this analysis is that the floating scum possesses many properties of insoluble surface active agents. One characteristic of such surface active materials is that they enhance’ froth stebility, and an effect would be produced in the pump bowl much like in a froth floatation chamber. It was suggested earlier that a 28 mechanism involving heavy froth must be resorted to to explain the high 'overflow rates experienced dfiring’the 2337 runs. Another effect of sur- face active agents is in the size of the bubbles generated. Development (15,16) | i work with bubble generating devices n the form of a venturi or jet pump has shown that cbnsiderably smaller bubbles are generated when a surface active material is present than.when absent. It is unclear whether they are attually generated smaller or whether the presence of surface active agents prevents their coalescence in the immediate vicinity of their generation. The result, however, is that they are considerably smaller., Now to extrapolate'to the MSRE, one wofild say that with reduced corrosibn products acting as surface active materials in the pump bowl, more small Bubbles were generated during the 233U runs than the.235U | rus. The smaller the'bubbles, the better their chance of being swept into the fuel loop by the undér flow. This then would be.a mechanistic hypothe- sis to explain the higher circulating void fractions and pump bowl overflow - rates during the 233y runs. Of Course, it must be shown that surface active materials existed in the pump bowl and fuel loop for all the 233y runs. Although regular salt sampling capsules came out of the feactor fairly clean, the beryllium addition capsules continued to show deposité_ of Fe and Ni in varying amounts during the remainder of the 233y runs. In addition, magnets were periodically lowered into the pump bowl in an effdrt to recover free metallic particles. They did recover these u materials although the gquantities were small, less than a gram; however, it doesn't take a large amount of surface active materials to have a dramatic effect on surface behavior. o The physical state or stability of thé froth in the pump bowl will be referred to several times in this report, particularly with regard to the differences in its--\sta.t_e between the 235U and 233U runs. In common usage a froth or foam, such as the foam on a_glass of beer, often implies a high degree of stability. In terms of bubble 1ifetime; the mean life of a bubble in a head of beer is orders of magnitude greater than in pure water. I do not intend to suggest an increase in bubble stability in the pump bowl anything like this, rather that the increase in bubble stability between the 235U and 233U runs was relatively minor. Let us. see if we can extract something meaningful regarding bubble lifetimes in [ " ) i » 29 the pump bowl. At steady state the bubble generation rate must equal the bubble destruction rate. The bubble generation rate was determined by the rate of gas carryunder from the xenon stripper salt spray. One might expect that the carryunder was more or léss constant for the 23U and 233y runs; at least we will assume this to be the case. We will- also assume that the bubble bursting rate is proportional to the total volume of gas bubbles in the salt. This neglects many variables which are certainly important, such as bubble size, depth of ffoth, salt drainage from the'bubble swarm, etc., nevertheless, in the interests of continuing with this discussion, the assumption probably isn't too bad. If it is true then at steady state, we have Bubble Generstion Rate = AV where V is the total vqlume of gas entrained and A is the "bursting constant" for the bubbles. The bubble half life would be 0.693/x. Note that within these assumptions at a constant bubble generation rate, if the bubble half life is doubled, then the total volume of gas entrained by the salt would also be doubled. The froth height in the pump bowl would then increase a corresponding amount. A mechanism like this, with an increase in bubble lifetime of this magnitude, could easily account for the higher overflow rates experienced during the 233y yuns., Note par- ticularly that doubling or even tripling bubble lifetimes represents a rather mild increase in bubble stability compared to more common notions. This, then is a description of the differences in operating char- acteristics between the 23°U and ?33U ruhs in the MSRE, and a suggested hypothesis to explain the differehces. The essentials df this hypdthesis, that noble metals and reduced corrosion products will adhere to liquid- | gas interfaées, will be referred to meny times in the following sectionms. 4. ANALYTICAL MODEL 4.1 Phyéical Basis of Model TFor reasons already discussed, and for.others which will bé@ome apparent in the section on RESULTS FROM THE MSRE, one might expeét that the transport of noble metals from the fuel salt to the various surfaces 30 where they deposit is controlled by the laws of mass transfer. Let us review briefly the behavior of noble metal fission products and fuel | sglt in the MSRE as follows: 1. Noble metals are born as ions from fission and decay of their precursor, but become atoms very quickly. They are homogeneously dis- persed in the salt. 2. Noble metals are unstable in fuel salt; i.e., they are present _in the reduced metallic state and are quite insoluble. They may even be unwet by salt. Massive metal objects display contact angles in the renge of 90-150°. (17} 3. Noble metals deposit on Hastelloy N and graphite surfaces, and large amounts are found there. _ 4.. They also deposit on liquid-gas ihterfaces, and we infer that they display some of the properties of surface active agents. This con- cept has been used to explain the differences in 23U and 233y operation of the MSRE, and will be used to explain many of the results observed iater. 5. Fuel salt is known to behave as a conventional Newtonian fluid. For instance, the primary heat exchanger was designed using conventional heat transfer correlations and the measured overall heat transfer coef- ficient was in good agreement with the design value. One would expect the same degree of success in estimating mass transfer coefficients since the transport phenomenon is the same in both cases. One must élWays_be wary, of course, because sometimes physical and chemical phenomena come info play that complicate the simple approach. | The above points constitute the essential requirements for a material to transport through the fluid boundary layer and deposit on the surfaces according to the laws of mass transfer. Development work from other reactor systems has shown that, if the chemistry of the fission products permits, they will be transported according to these laws. See, for ~example, Refs. 18 through 21. Tt would therefore seem fruitful to attack the noble metal migration question in the MSRE within the framework of mass transfer theory in its simplest form and let the results speak for themselves. ; i | i | | | ] ] ” ) w [ » 31 4.2 Analytical Model First, we must write a rate balance on the noble metals in fuel salt where they are born. The conditions of this rate balance are as follows. Note from the generalized beta decay chain from Sect. 3.l-that noble metals appear.in groups and we have a situation where noble metals can decay into noble metals. ‘Note also from the generalized decay scheme that each noble metal grouping may start‘off with a sait seeking precursor. The analytical model will therefore consider the last salt seeking precursor of the noble metal chain. We must do this primarily because of 95Nb whose precursor is 95Zr with a half life of 65 days. The analytical model must be for the unsteady state. This is because the MSRE, being an experimental reactor, had a quite erratic power history. Seldom did any of the longer lived isotopes reach steady state. The model will consider the entire fuel loop to be a "well stirred pot." One might expect this assumption to be adequate because the fuel circuit time around the loop is about 25 seconds and the isotopes we will be dealing with have half lives rang- ing from hours to years. A better indication of the adequacy of this assumption would be to compare the fuel circuit time to the cdmputed residence time of a noble metal atom in salt before it deposits on a surface. In Appendix B, it is shown that the longest residence time (expressed as the half life of ncble metals in fuel'salt) is twice the loop circuit time. Lastly, it is assumed that all noble metals migrate independently of each other, even.when of the same chemical species. Consistent with all the considerations discussed above, we can now write a rate balance on the noble metals associated with fuel salt where " they are born. The equation in words is as follows where the units of each term is atoms/time. S v dcC dt generstion rate from fission + generation rate from decay of precursor - decay rete - deposition rate on graphite ~ deposition rate on heat exchanger | | (1) - deposition rate on rest of fuel loop - deposition rate on 1iquid—gas interfaces (bubbles). 32 ‘See Appendix A for the Nomenclature. The two generation terms are func- tions of the reactor power history and yields of the specific isotopes involved. The decay rate is a function of the concentration of each specific noble metal ahd its decay constafit. A1l the other terms are deposition‘rates and are functions of the surface area, mass trfinsfer coefficient and thé concentration potential. The deposition rates on the graphite in the core and on the heat exchanger have been listed separately. The rest of the loop is lumped into one term. Deposition on bubbles has also been listed as a separate term because of its importance. Each term will now be evaluated separafely and the equation will be integrated over 4.,2.1 Generation from Fission ' The generation rate direct from fission is simply as follows: Generation from fission = yP | (2) 4.2.2 Generation from Decay of Precursor A rate balence on the last soluble precursor before decaying into a noble metal will be pS | dC _ P p _ 3PcPS y - o (3) V& =Y Integrating and evaluating at the boundary condition ¢ =cPatt=0 (L) we get o - P | | ¢S - LR, (o PS_¥By (5) APy © APy The generation rate from decay of the precursor will be Generation from decay of presursor = }PVCPS‘ (6) P 2Pt = PP + (WPve P° - yPp) e o/ * . " &) » & 33 fi.2.3 Decay Rate The decay rate of each noble metal isotope is simply as follows: Decay rate =7)\VCs - o | (7) 4.2.4 Deposition Rate on Heat Exchanger Written in the framework of mass transfer theory, the deposition rate of noble metals on the heat exchanger surfaces can be expressed as “follows: Deposition rate on heat exchanger = n® a"e (c® - ct) (8) At this point it will be necessary to make an assumption concerning c®t (the concentration of noble metal in the fuel salt at the fluid-solid interface). We will dssume that all solid surfaces involved (Hastelloy N and graphite) behave as an infinite sink; i.e., if_a noble metal atom migrates through the fluid boundary layer and contacts the surface, it will stick there forever. This is, of course, one of the unknowns in this analysis. The result of this assumption is that CSi effectively becomes zero, and the equation reduces to Deposition rate on heat exchanger = pe aPe ¢S (9) h.2.5 Deposition Rate on Graphite | By reasoning similar to that above we can arrive at the equation for deposition of noble metals on the core graphite as follows: gra ,gre cs (10) Deposition rate on graphite = h A We have again assumed a sticking fraction of 1.0, 4h.2.6 Dep051t1an Rate on Rest of Fuel Loop Again by reasonlng similar to that above, we arrive at the equatlon for the deposition of noble metals on the rest of the fuel loop (all Hastelloy N). | Deposition rate on rest of fuel loop ==2:(hA)rESt of fuel 1oopaS(yy)- J 3 4.2.7 Deposition Rate on Liquid-Gas Interfaces As noted before, there are small bubbles c1rcu1at1ng with the fuel salt. Their amount is small but the product of their surface area and 'mass transfer coefficient is higher than the same product for all the rest of the solid loop surfaces combined. Even if the sticking fraction to a bubble is low, their effects will still be quite strong. If we assume that noble metals deposit on liquid-gas interfaces in accordance with the mass transfer theory and that the sticking fraction 1s 1. 0 we agaln arrive at 8 similar equation for the rate of migration to bubbles. Deposition rate on liquid-gas interfaces (bubbles) = pPUPPUbS (12) Later we will deduce that the effective sticking fraction to bubbles is "less than unity. 4,2.8 Equation for C° Now the individual generation rate terms ahd decay and deposition rate terms are substituted into the original rate balance around the fuel salt; Eq. (1), and the equation is integrated. The constant of integration is evaluated at the boundary condition ¢® = coS at t = 0 | (13) This will yield the equation for the concentration (Cs) of a noble metal in the fuel salt at any time (t). This equation is as follows: _BGE 4 y) (14%) S P PS _D P P AT CT =y P/ At s _Ply" +y) (L _o "' s ¢ = =g+ | ) e + [ Co VX X - AP AP cgs -yPe/v ] -xt - (————) ] e X - AP where S ' gra ,gra he . he 'Z (hA)reSt of loop bub ,bub | Loy s DETT ABTE "€ AR L h A (15) : } vV vV vV Vv This equation can be carried through the power history of the reactor. If the power level is changed, then the value of Cs preceding the power change will be the initial condition (Ci) at the new power level. Also note that - [ [ .oab i »”» 35 X has units of time_l, end is the theoretical rate constant for migration of noble metals from the fuel salt to their sinks. 4.2.9 Noble Metals on Solid Surfaces Nofi'that we have an expression for the concentration of noble metals in fuel salt as a function of time and the power history of the reactor, we can compute the amount of noble metals deposited on any surface in the reactor. To do this we must again set up an unsteady state rate balance for that surface as we did for the fuel salt. The same assumptions and considerations hold true here as they did for the fuel salt. In addition, we will assume there is no interaction between the bubbles and solid surfaces; i.e., the only source of noble metals for the surfaces is direct- ly from the salt. The rate balance is as follows where the units of each term is atoms/time-unit area. 0 | g%—-= Deposition Rate - Decay Rate (16) The individual term are as follows: Deposition Rate = h" C° (17) Decay Rate = ACT (18) Substituting we have | m dC” _ .m.s m == hC -ACc (19) Now, substituting the value of c® determined in the previous section (Eq. 1b4), integrating over time, and evalusting the constant of integration at m_ m : , ' ¢ =C, stt=0 | - (20) we get : ' o W (PgPE _ P ) -\" e (v2 + v) ‘h (A"C" -y P/V)e m ot D BBy, o - ol M) (o (x-2®) (- AP) | a PePS _ D © . ymp.ps _ P APcPS _ yPpyy | - | PoPS _ - /V:]e " +{Cm_ ey +y) B (A0 - yR/V) AVX X - AP ,, (X-2F) (A-2F) At m P~ PS p _h s _pGPey) MO -pR/V A-X'| 7o VX T X i p 36 This then is the equation for the concentration of a noble metal isotope on & surface at any time as a function of power., It can be carried through the power history of the reactor just as the equation for c®. The equation is applicable to any noble metal isotope and to any solid surface ~in the reactor (Heat Exchanger, Graphite in the Core, Core Surveillance Specimens,'étc.) by proper choice of the mass transfer coefficient. 4.2.10 Noble Metals on Liquid-Gas Interfaces The principal difference befween.deposition on & solid surface treated above and the ligquid-gas surface trested in this sectibn is the sink terms once the noble metal has reached the surface. In the case of the solid'surfaée,‘thé only sink term considered was decay.- In the case of a liquid—gaé interface, migration of noble metals to the off-gas system repre- sents another sink term that must be considered. | Later on in the analysis of reSulté from the MSRE we will want to con- sider the entire gas phase in the fuel loop as a well mixed pot. The entire gas phase will consist of bubbles circulating in the lodp, the gas phase in the pump bowl and also part of the gas phase in the overflow tank. The gas phase is defined to include the liquid-gas interface, so that noble netals attached to the interface will be considered as part of the gas phaéé; The reasons will become more apparent later on, but for now let us justlsay that it will be necessary to find a process by which noble metals are removed from the reactor system gas phase as defined above with = rather low rate constant. For example, noble metals transported to the of f-gas system or the drain tanks would be considered as removed from the gas phase as defined above. At any rate we will again write an unsteadyf state rate balance around the gas phase. The same assumptions and con~- siderations are still applicable as in the rate balance around the fuél salt. This time, however,instead of using concentration units (atoms/vol), we wiil use total inventory units (atoms) in the gas phase. The rate belence is as follows: al i - Migration rate to bubbles - decay rate (22) - migration rate to off-gas system. it " -} " » 37 The individual terms are as follows. Migration rate to bubbles - pPuP 4bub 8 (23) Decay rate = AI 7 (2k) Migration rate to off-gas system - ZI (25) The transfer of noble metals from the pump bowl to the off-gas system is dealt with in a very generaly way. We simply defined a rate constant (Z) which says that the rate of transfer is proportional to the total amount of a noble metal isotope. Now, substituting the individual terms into the rate balance, then substituting the value of ¢® from the fuel salt analysis (Eq. 1L), integrating over time and evaluating the constant of integration at I =1 att=0, (26) we get . D P~PS _ P -A RPA%p(y® + y) I AC -y P/V e I ( ) (27) v (i+e) (A + 2 - AP) X - AP pP-ps _ D n° 8 fps _BGPay) Xl TVEN | xe M B - X o VX x - 2P PsPS _ P o PP (y® 4+ v) ~ LP b e -y P/V) VX (3 + ) (A+2-2F) x-P L o P.PS _ D __w AP os RGP ry) MO “YENIL (e O+ 28 - X) o vV X x - 2P This then is the equation for the total amount of a noble metal isotope in the gas phase of the fuel loop as a function of time and the power history of the réactor,=andrthe rate constant (Z) for noble metal transfer to the off-gas system or dump tanks. 38 5. RESULTS FROM THE MSRE 5.1 Introduction. The general approach in the asnalysis of noble metal migration in the - MSRE will be as follows. First, we will compare the measured to theoreti- cal concentration of noble metals found on the primary heat exchanger by gamma spectrometry. This measurement tedhnique.gafie quantitative results for the greatest number of noble metal isotopes. The measurements were made in situ, so there can be no question df contamination or problems associated with hot cell processing. Also the fluid dynamic conditions in this conventional U tube heat exchanger are fairly well ihown; This com- parison then is quite good and seems to put the analyticael model on a firm foundatibn. We will then look at the core surveillance sambies:and show how they generally confirm the results from the heat exchanger. Then we will look at the fuel saltkand gas phase samples. After each one of these comparisons, observations will be made on the nature"offnoble'metai migra-~ tion. We will then mske a rather crude attempt to quantify the hypothesis ' that noble metals accumulate in the pump bowl and determine rate constants for removal of noble metals from the pump bowl to the off-gas system to see if they are physically reasonable. As noted previbusly, most of the. 233 analysis will be for measurements made during the U operation (runs 15- 235 20), although some data are available for the U operation (run 1-1%4). 5.2 Comparison of Measured Deposition on Heat Exchanger to Theoretical | The theoretical amount of each noble metal isotope deposited on the- surface of the heat exchanger tubes was computed with Eq. 21, and compared to the measured amount as determined by gesmma spectrometry. The measured values following Run 14 were obtained from Ref. 10 and after runs 18 and 19 from Ref. 11. Recall that the equipment and techniques used>to ‘ determine fission product deposition after run 14 were less sophisticated than used after runs 18 and 19, therefore, the data following run 14 are less certein than the newer data. The parameters used to calculate the theoretical amount (e.g., mass transfer coefficients, isotope parameters, ete.) are evaluated and tabulated in Appendix B. The computed concentra- tion on the heat exchanger takes into account the entire power history of the reactor. The only sink term for noble metals attached to the heat ” » o) 39 exchanger is decay. It should be noted that periodically the fuel loop is cleaned out by circulating flush salt for a short period of time. The computed concentration assumes that this process does not leach noble metals from the surfaces. Some of the gamma spectra following run 1l were made with and without flush salt in the loop and they indicate this aésump- tion to be true. The theoretical amounts of noble metals on the héat exchanger were computed on the basis that they do not adhere to liquid-gas interfaces. Then, by comparison of the computed amounts with the measured amounts, conclusions of a qualitative nature will be drawn to the effect that noble metals apparently adhere to liquid-gas intérfaces. The results of this comparison are shown in Figure 5.1 where the ratio of measured to computed amounts on the heat éxchanger is plotted against , the noble metal half life. First let us look at the curve measured during 233U operation. These measurements were made with the improved gamma spectrometry equipment and.are Judged to be the best. The following observations can be made. 1. The curve is madé up from 10 isotopes; 3.; rutheniums, 3 - telluriums, 2 - antimonies, 1 - molybdenum and 1 - niobium, some of which are duplicated measurements. They seem to be rather tightly grouped 132Te and l03Ru. One would conclude around the line, except perhaps therefore that each noble metal isotope migrates as a function of its own concentration in salt and is not ififluenced by other elemental species of noble metals or even isotopic species of the same element. 2. Note that the curve is & straight line and has a slope very close to zero;'i.e., ndble-métai migration is not an unaccounted-for function of its own half life. It is important to note at this time, because later we will see that this is not true for the fuel salt end gas phese samples. 3. Because the sldpe.pf-therline is almost zero and bécause_of the tight gfouping of data around this liné, some credence must be given to the hypothesis that noble metals migrate according to the simplest form of mass transfer theory. The good correlation 81s0 speaks well fbr the quality of the gamma spectrometry deta. Something is still missing, however, because the measured to theoretical concentration ratio is con- siderably less than unit for the 233U run determinations, and this will be discussed shortly. MEASURED AMOUNT ON HEAT EXCHANGER (dpm/cmfi) THEORETICAL AMOUNT ON HEAT EXCHANGER {dpm/cm?) Lo ORNL-DWG 70-15021 10 THEORETICAL AMOUNT COMPUTED FOR CASE WHEN NOBLE METALS DO 5.0 DEPOSIT ON LIQUID-GAS INTERFACE O MEASURED AFTER RUN 14 (235u) A MEASURED AFTER RUN 18 (233y) @ MEASURED AFTER RUN 19 (233y) 2.0 oy kgt 1.0 ' lloflnu MEASURED DURING 235U OPERATION 95Nb 0.5 DURING 233y OPERATION 0.2 106Ry v 1 0.1 o » o o 0.02 - 0.01 1 2 5 10 2 5 102 2 5 103 2 5 10 2 5 105 NOBLE METAL HALF LIFE (hr) FIGURE 5.1. COMPARISON OF THE MEASURED TO THEORETICAL AMOUNTS OF NOBLE METALS ON THE PRIMARY HEAT EXCHANGER o «f . n Y yuns than the L1 L. Note the good agreement between 95Nb and the other noble metals. 95 It was pointed out previously that ““Nb sometimes behaves as 8 salt seeking fission product and sometimes as a noble metal, depending on the oxidation- reduction state of the fuel salt. The fuel was in a reduced state at the end of runs 18 and 19 so this 95Nb behavior is as expected. 35U operation. The same Now consider the data measured during the observations and conclusions can, in general, be made for this curve, but on & lesser scale because there are fewer isotopes involved. The magnitude of the measured to calculated ratio is different and is in the vieinity of 95Nb and the other noble metals. 235 1.0. Again note the good agreement between The fuel salt was generally in a more reduced state during the U oper- ation and Nb behaved as a noble metal. Now - why is the magnitude of the measured to theoretical concentra- tion ratio on the tube surface close to unlty when measured after run 1k (last 235 (233U.runs)? As noted in section 3.3, there was a dramatic difference in 235U and 233U. U run) and 0.15 = 0.20 when measured during runs 18 and 19 operational characteristics of the MSRE when operating with The difference was most apparent in the amount of bubbles;circulating with the fuel salt. As noted earlier the void fraction of bubbles during the 235y runs was 0.02 - 0.045%, while during the 233y runs, it was 0.5 to . 0.6%, up by a factor of about 20. The theoretical amount in the denomina- tor of the ordinate of Figure 5.1 is computed assuming that noble metals do not adhere to liquid-gas interfaces (circulating bubbles). It has been hypothesized that noble metals do deposit on these interfaces, so the bubbles will compete with the heat exchanger as a noble metal defiository. If this concept is included in the‘calculation, then the computed amount on the fieat exchahger will diminiéh and the value of the ordinate will increase. It is also assuméd'that there is no interaction between the bubbles and the heat exchanger surfaces, so if a noble mefal atom migrates to a bubble, it is no longer available to deposit on the heat exchanger. Since the fuel salt contained many times more bubbles during the.?33 35U runs , 1nclud1ng bubbles in the calculation, 35U curve. Qualitatively the 233U curve will move up much more than the then this will explain the difference in elevation of the two curves. Quantitetively, it is a bit more difficult, because uSing the best ho _‘estimated values of bubble surface area and mass transfer coefficient (seé Appendix B), both curves are pushed upward parallel to themselves to a value considerably greater than unity. If the empirical but nevertheless mechanistic "sticking fraction" is brought into the calculation, we can agéin‘bring both curves down to a value very close to unity. The sticking fraction is defined as that fraction of atoms that contact thg_interface and adhere to it. If the sticking fraction of noble metals to solid - surfaces is maintained at unity, then a sticking fraction to iiquid—gas interfaces of 0.1 - 0.2 is required to bring the value at the ordinate of both curves close to unity. A sticking fraction of noble metals to a liguid-gas interface con- (22) siderably less than 1.0 was not expected. For instance ,.noble metals were found in significant quantities in helium passed over the surface of 'quiescent molten fuel salt samples in a hot cell experiment. - The vapor pressure of noble metals is diminishingly small at these temperatures and cannot possibly account for the chcentrations observed. The indivi- dual noble\mgtal atoms or very small clusters of them, are apparently spontaneously expelled from the surface. Considerations of the interfacial (23) Therefore, one would energies involved indicate this to be possible. expect a sticking fraction to bubbles of unity. A rationalization of this apparent paradox is as follows. An observation from the reactor is that many of the smaller bubbles circulating with the fuel salt completely 135 dissolve in the higher pressure part of the fuel loop. observation is a result of an analysis to explain the Xe poisoning effects observed in the reactor. Note that the surface tension of molten salt is quite high, about 200 dynes/em. When a bubble is pressurized by the pump and begins to dissolve, its diameter decreases. Thé internal pressure sbove ambient, as generated by the surface tension (4o/d) becomes quite high (0.8 psi for 0.00S in. bubble)rand further enhances the dis- solution rate. This continues to the limit and the bubble dissolves. The process is very rapid ‘and analysis indicates only a few seconds are required. If this bubble contained some noble metalS'before it dissolved, we would end up with a noble metal cluster associated with the salt again, rather than a bubble. Qualitatively then, here is & mechanism where the i O ) L 1] L3 the sticking fraction of noble metal atoms to a liquid-gas interface can be unity but where it actually appears to be less than unity. Possibly other mechanisms cofild also be devised. Conclusions. The conclusion from examination of noble metal deposi- tion on the primary heat exchanger as determined by gamma spectrometry, is that noble metals apparently do migrate and deposit on these surfaces in accordance with the laws of mass transfer in the simplest form. Deposi- tion on liquid-gas interfaces (bubbles) must be included in the calculation to‘force the calculated concentration to equal the measured concentration on the tubes. The exact mechanism of noble metal deposition on liquid- gas interfaces is not known and had to be handled in a semiquantitative ‘way involving apparent sticking fractions to bubbles, Essentially these same conclusions will be reached after each analysis section in this -report. 5.3 Comparison of Measured Deposition on Core Surveillance Samples to Theoretical The theoretical amount of each noble metal isotope deposited on the surface of'the graphite and the Hastelloy-N core surveillance samples was computed and compared to the measured amount. The measured values were obtained from semiannual reports.(26’ 2T, 28 and 29) The same introduc- tory remarks are applicable here as in the first paragraph of the last section. Most important, recali'that the theoretical amount will be com- puted on the basis that noble metals do not deposit on liquid-gas inter- faces. The mass transfer coefficient was estimated in Appendix B to be abofit 0.25 ft/hr for both the gfaphité and Hastelloy-N. The uncertainty in this mumber is considerable. | The results of this comparison are shown in Figure 5.2 for the graphite specimens and Figfire 5.3 for the Hastelloy-N specimens. Again, the ratio of the measured to theoretical amount is plotted against the noble metal half life. In drawing lines through the data, no weight 95 was given to the ““Nb points for reasons to be discussed later. The 'following observations can be made : Ly ORNL-DWG 70-15015 | THEORETICAL AMOUNT COMPUTED FOR CASE WHERE NOBLE METALS NOT DEPOSIT ON LIQUID-GAS INTERFACE 10 SAMPLES REMOVED FROM INTEGRATED CORE AFTER : EXPOSURE . FUEL 2 RUN 7 7,800 Mw hrs 235y RUN 11 24,000 Mw hrs 235y RUN 14 32,000 Mw hrs 233y RUN 18 20,000 Mw hrs 233y RUN 18 76,000 Mw hrs 235 AND 233y 129m Te 103Ry — o o » wn IOGRU o Y SPECIMENS EXPOSED TO FUEL SALT DURING: o — 233) OPERATION o o o 235 AND 233y TION MEASURED AMOUNT OF GRAPHITE SPECIMEN (dpm/cm2 THEORETICAL AMQUNT ON GRAPHITE SPECIMEN {dpm/cm 0.02 33y OPERATI 0.01 1 2 5 10 2 5 102 2 5 103 2 5 10 NOBLE METAL HALF LIFE (hr) FIGURE 5.2. COMPARISON OF THE MEASURED TO THEORETICAL AMOUNTS OF NOBLE METALS ON THE GRAPHITE CORE SURVEILLANCE SPECIMENS ¥ ”) 0.2 0.1 0.05 MEASURED AMOUNT ON HASTELLOY-N SPECIMEN (dpm/cm? THEORETICAL AMOUNT OF HASTELLOY-N SPECIMEN 45 ORNL-DWG 70-15016 THEORETICAL AMOUNT COMPUTED FQR CASE WHERE NOBLE METALS DO NOT DEPOSIT ON LIQUID-GAS .INTERFACE SAMPLES REMOVED FROM INTEGRATED CORE AFTER EXPOSURE FUEL RUN 14 32,000 Mw hrs 235y 2 RUN 18 20,000 Mw hrs 233)) L] g _ Te 1.0, Mo fE L32Te ISNbHILO3RU 0.5 SPECIMENS . , EXPOSED TO SALT DURING: 33y OPERATION 235y OPERATION 1 2 5 0 2 - 5 102 2 5 103 2 5 104 NOBLE METAL HALF LIFE (hr) FIGURE 5.3. COMPARISON OF THE MEASURED TO THEORET | CAL AMOUNTS OF NOBLE METALS ON THE HASTELLOY-N CORE SURVE!LLANCE SPECIMENS | (DOSIMETER TUBES) L6 7~ - 1. Three different sets of graphite samples (Figure 5.2) were ex- 235 posed to fuel salt during only the U operations. All three are con- sistent with each other and the data points are rather tightly grouped around a single line. An exception is 95Nb which lies well above the | 235 line. Recall that the fuel salt was in a reduced state during the U 95 operations and ““Nb behaved as a noble metal and would therefore be ex- pected to fall on the line. More will be said about this below. 2. One set of graphite samples was exposed to fuel salt during only 233 ' ' U opérations. This curve falls below and parallel to the previous 235, curve for U operation. This is consistent with observations of noble metals in the heat exchanger in the last section and the same conclusions can be drawn. 3. The curves for graphite may have a significant slope at the low half life end which is not consistent with the gamma spectrometry data from the heat exchanger. The slope seems to be zero however for half lives over about 500 hours. This will be discussed further below. L, The relative elevation of the curves thréugh the deposition data on the Hastelloy-N surveillasnce samples (Figure 5.3) do not confirm the previous observations from the heat exchanger and graphite. There is however more scatter in the data. | 5. Concerning the absolute value of the measured to theoretical ratio of the noble metal concentration. An overall average of the 233U line for Hastelloy-~N (Figure 5fl3) seems to agree with the previous value from the heat exchanger, however, the data from the 235U runs do not agree. 6. The measured to theoretical concentration ratio for all the graphite samples seems to fall somewhat below the data from the heat ex- changer‘and Hastelloy-N samples. There are two explanations for this. First, inadequate knowledge of the mass transfer coefficient, although this probably isn't enough to account for the entire discrepancy. Second, the more plausible explanation is that'the sticking fraction for graphite is less than unity. More will be said about this below. Let us consider some of the above discrepancies with graphite'ahd see if we can bring them into line with previous observations and conclu- sions from the heat exchanger analysis. First let us hypothesize that the ‘noble metal sticking fraction to graphite is less than unity. We have of *} ) k7 noted that this is 1likely in observation number 6 above. This will be con~ firmed more directly in Section 5.7 when we look at the results of a special 1aMinar flow core surveillance test during the last MSRE operation- al period. Nofi, there is evidence(BO) that Nb forms a stable carbide with graphite under MSRE operating conditions. The graphite could therefore act as an infinite sink for Nb, i.e., its sticking fraction could be unity. 95 An effective sticking fraction of ““Nb to graphite of unity, and a sticking fraction of the other noble metals of less than unity, would explain the 95Nb points being so much higher than the other noble metals in Figure 5.2. In fact, the ratio between the curves and the‘95Nb data cluster could be a direct-measure_of'the sticking fraction of the other noble metals relastive to 7°Nb. We can carry this discussion one step further. w5 (half life = 2.4 min) is a precursor of 9 Mo. Its half life is high enough so that it is a slgnlflcant mlgratlng species in this decay chain. If then, amounts 99Nb migrate to the graphite with a sticking fraction of unity and decay 99 -~ 99 into ““"Mo, the measured amounts of ““Mo would also be relatively higher then the other noble metals. Therefore, the left side of the curves in Figure 5.2, which are controlled to an extent by 99Mo might tend to be higher than the other noble metals. Another effect fihat must be considered is that during the final week of run 18, the fuel pump was operated at a slightiy reduced speed. The principal effect was that during this period 235 the c1rcu1ating void fraction was more equlvalent to the U runs than the 233 during this period the nobie metal deposition rate on solid surfaces would . U runs. Since there were fewer bubbles,.one might expect that be greater. Since the time perlod was. relatlvely short it would elevate only the low half life end of the curve. Thls last point may also help explain why the short hsalf life end of the 33U curve for dep051tion on | Hastelloy-N (Figure 5. 3) is higher than expected. Conclusion. The conclusions then, from examination of noble metal dep051tion data from the core surveill&nce samples,’ is that they generally confirm the conclusions arrived at after examination of the heat exchanger in Sectlon 5.2. It appears ‘that the sticking fraction of noble metals to grephite is less than‘finity, We have had to form a hypothesis to explain the behavior of Nb. | ~is sensitive and a small error in the 48 5.4 Comparison of Fuel Salt Samples with C° The theoretical amount of each noble metal isotope contained in the ffiel salt (Eq. (14) in Section 4.2) was computed and compared to the measured amount in the fuel salt samples. Again the théorétical amount Was-computed on the basis that noble metals do not deposit'on liquid- gas interfaces. The measured and computed‘concentration differed by 1-L4 orders of magnitude, the measured concentration always being higher. In an attempt to resolve this discrepancy, the measured-to-theoretical concentration ratio was plotted against the fraction'of its total capacity thét each sample capsule.was filled. These plots seem to be unique and are shown in Figures 5.4 through 5.9 for 103Ru, 106Ru, 99Mo, 129mTe, l32'I'e' and 95Nb, respectively. The kind of sample freeze valve or double- ‘wall freeze valve,is distinguished on the plots. Recall that a freeze valve capsule contains 50-60 grams of salt then full and a double wali freeze valve capsuie holds about 15 grams of salt when full. Ladle sample data are not included on these plots for reasons pointed out in Section 3.2. All reported freeze valve capsule data are on these plots exgept those taken at zero power. It is a characteristic of the analytical model that the noble metal concentration ifi salt goes to zero shortiy after the reactor power goes to zero. The measured to theoretical con- centration ratio wduld therefore approach infinity if any noble metals at all were measured in the sample. The following observations can bé made about these curves. 95 1. All curves, except that for ““Nb, behave in a similar manner. For samples that were mostly empty, the measured to theoretical concentra- tion ratio is orders of magnitude higher than for those samples that were mostly full. 5 95 Nb is being carried along for comparison, since it can behave either as a noble metal or a salt seeking fission product, depending on the oxidation-reduction state of the salt. Many of the reported o concentrations in fuel salt are negative. This is because the reported 95 Nb concentration was extrapolated back from the time of measurement to 95 the time of sampling. 95 Zr with a 65-day half life is a precursor of Nb, and therefore it must also be counted. The back calculation in time 95 Zr concentration measurement = | )] MEASURED CONCENTRATION IN SALT (dpm/g—salt) THEORETICAL CONCENTRATION IN SALT (dpm/g—salt) 100 L9 ORNL-DWG 71-1876A 104 — —— THEORETICAL CONCENTRATION COMPUTED FOR CASE WHERE NOBLE | METALS DO NOT DEPOSIT ON LIQUID—GAS INTERFACE 5 }— | | ® 17-31 FREEZE VALVE CAPSULE L O 17-32 DOUBLE-WALL FREEZE VALVE CAPSULE L—SAMPLE NUMBER TAKEN DURING 2 — L—THIS RUN NUMBER - SAMPLE GROUPING NO. 1 (SEE TEXT) 0O 19-44 017-32 102 \\ 5 N\, A’ @177 N 2 - \\ @17-10 __ ®17_2 o l \ "14-20 18219 L 101 18-44 *18—12 ::o. 17-31 ~—~=19-36 5 : ®17-22 — ®14-66 ‘@14—63 2 > SAMPLE GROUPING NO. 2 0 01 02 03 04 05 06 07 08 09 10 FUEL SALT IN CAPSULE AS FRACTION OF TOTAL CAPSULE CAPACITY FIGURE 5.4. COMPARISON OF THE MEASUREb I03RU CONCENTRAT I ON IN FUEL SALT TO THE THEORET!ICAL CONCENTRATION MEASURED CONCENTRATION IN SALT {dpm/g—salt) THEORETICAL CONCENTRATION IN SALT (dpm/g—salt) 50 ORNL-DWG 71-1877A THEORETICAL CONCENTRATION COMPUTED FOR CASE WHERE NOBLE METALS DO NOT DEPOSIT ON LIQUID--GAS INTERFACE ©® 17-31 FREEZE VALVE CAPSULE O 17-32 DOUBLE-WALL FREEZE VALVE CAPSULE SAMPLE NUMBER TAKEN DURING THIS RUN NUMBER 122 SAMPLE GROUPING NO. 2 ® © 146 14-20 \ 0 0.1 02 03 04 05 06 0.7 08 09 1.0 | FUEL SALT IN CAPSULE AS FRACTION OF TOTAL CAPSULE CAPACITY FIGURE 5.5. COMPARISON OF MEASURED lO6RU CONCENTRAT I ON IN FUEL SALT TO THE THEORETICAL CONCENTRATION pat » ORNL-DWG 71-1878A THEORETICAL CONCENTRATION COMPUTED FOR CASE WHERE NOBLE METALS DO NOT DEPOSIT ON LIQUID-GAS INTERFACE ® 17-31 FREEZE VALVE CAPSULE O 17-32 DOUBLE-WALL FREEZE VALVE CAPSULE SAMPLE NUMBER TAKEN DURING THIS RUN NUMBER SAMPLE GROUPING NO. 1 (SEE TEXT) [ 13 MEASURED CONCENTRATION IN SALT (dpm/g-salt) THEORETICAL CONCENTRATION IN SALT (dpm/g-salt) 7-2 l9- C]}?'36 018-2 - 14-20 "17-31 ' < O1s- 18-4 18-19 = : ®:1.30 SAMPLE GROUPING NO. 2 017-32 7 | | ’ . ]4-66 0 0.1 0.2 0.3 0.4 05 0.6 0.7 0.8 0.9 1.0 FUEL SALT IN CAPSULE AS FRACTION OF TOTAL CAPSULE CAPACITY FIGURE 5.6. COMPARISON OF THE MEASURED °°MO CONCENTRATION _RIN FUEL SALT TO THE THEORET ICAL CONCENTRATION L] MEASURED CONCENTRATION IN SALT (dpm/g-salt) THEORETICAL CONCENTRATION IN SALT (dpm/g-sa #1103 52 ORNL-DWG 71-1879A THEORECTICAL CONCENTRATION COMPUTED FOR CASE WHERE NOBLE METALS DO NOT DEPQSIT ON LIQUID-GAS INTERFACE @ 17-31 FREEZE VALVE CAPSULE O 17-32 DOUBLE-WALL FREEZE VALVE CAPSULE b SAMPLE NUMBER TAKEN DURING THIS RUN NUMBER SAMPLE GROUPING NO. 1 (SEE TEXT) O 20-1 102 10 1.0 o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 FUEL SALT IN CAPSULE AS FRACTION OF TOTAL CAPSULE CAPACITY FIGURE 5.7. COMPARISON OF THE MEASURED |29mTE CONCENTRATION IN FUEL SALT TO THE THEORETICAL CONCENTRATION " " " i + MEASURED CONCENTRATION IN SALT (dpm/g-salt) 2 104 5 2 ’E . w103 & ~— E o T 5 — - °U Runs 5.0 3Ls5 1725 During 25U Runs 5.0 5581 27,900 Mass Transfer Parameters to the Core Surveillance Specimens The complexity of the fluid dynamic conditions in this region has been discussed briefly in Section 3.2. As the cross sectional view in Figure "2.4 shows, the rectangular graphite specimens are packed together to form a rather complicated fluid dyanmic arrangement featuring both inside and outside corners. Adjaéent to each major graphite surface and almost touching it are the Hastelloy-N tensile specimens and dosimeter tube. The tensile specimens are particularly difficult to analyze because they form a large number of fluid entrance and exit regions for themselves and the adjacent graphite. This effect would be expected to increase the mass transfer coefficient. On the other hand,lthe close proximity of the wide part of the tensile specmens to the graphite (about 0.040 in.) might be expected to locally stagnate the fuel salt. This effect would tend to decrease the mass transfer éoefficient-in the affected region. Finally, the entire sample assembly is enclosed in a perforated basket. The intent of the basket is to allow for cross flow, but'it also contributes to the uncertainty of fuel salt velocity past the sample. The perforations also generate an unknown amount of turbulence in the salt. All these fluid dynamic complications were a necessary result of inéorporating a large _number of different kinds of sampleSf(bothlgraphite and metal) for differ- ent purposes into a rather confined volume. I should aiso point out that when the sample station was built, the necessity of being able to accurate- ly estimate mass transfer coefficients.was-nof fully realized. The salt -fielocity in the sample station has been estimated to be about 2 ft/sec, partly by indirect measurements and partly by estimate. The uncertdinty could be as much as 25 percent. The equivalent diameter of the annular region between the samples and perforated basket, and based on all sample [ atems e e e et e e e % a3 e e n ¢ m . e 8l components exposed to salt is about 0.62 in. The Reynolds Number then, based on these values, is about 3000. This implies that the flow is only marginally turbulent. This being the case, one might expect that the confining nature of the inside corners formed by the graphite, and the zones between the Hastelloy-N samples and graphite surfaces would tend to make the flow laminar. Using the usual relationships for heat transfer coef- ficients and applying the heat-mass transfer analogy, we would compute an overall mass transfer coefficient of 0.31 ft/hr for turbulent flow and 0.077 ft/hr for laminar flow in this region. In this analysis then, we will pick an intermediate value of 0.25 £t /hr for the mass transfer coef- ficient to both the graphite and Hastelloy-N specimens. The uncertainty in this nfimbér is high, probably more so for the Hastelloy-N than for the graphite. Noble Metal Fission Product Parameters ‘Table B-3 lists the noble metal (and precursor) yield and half life quantities during the 233U runs, and Table B-L4 1lists those parameters during the 235U runs, used in this analysis. Table B-3 Noble Metal Fission Product Parameters (233U Runs ) Cumulative Yield¥ Direct Yield Noble of Precursor of Noble Metal Half Life Half Life Metal A % of Precursor of Noble Metal Mo 4.89 0.0 2.4 min 66.5 hrs 103Ry 2.00 0.0 1.2 min ~ 39.7 days 105gy 0.706 0.0 9 min 4.45 hrs 106Ry 0.438 0.0 <1 min 1.01 yrs 125gy 0.0839 0.0 G days 2.0 yrs 127gp = 0.58 0.0 1.9 hrs 91 hrs 129mpe 0.71 0.0 4.6 nrs - 37 days 131mqe 0.441 0.0 23 min | 30 hrs 132mpg b.h3 0.0 2.1 min - 7T hrs 25Nb 6.00 0.0 65 days 35 days ¥Based on the following Fission Distribution - Component 233U 235y 239py Percent of Fissions 93.2 2.3 I 85 Table B-L Noble Metal Fission Product Parameters (235U Runs ) Cumulative Yield Direct Yield Noble of Precursor of Noble Metal Half Life Half Life Metal % % of Precursor of Noble Metal 39Mo ' 6.06 , 0.0 2.4 min 66.5 hrs 103gy 3.00 . 0.0 1.2 min 39.7 days 106Ry 0.39 0.0 <1 min. 1.01 yrs 1291mpg 0.71 0.0 4.6 hrs 37 days 1327e L. 71 0.0 2.1 min 7T hrs 35Nb 6.22 - 0.0 65 days 35 days Time Constant for Noble Metals in Fuel Salt In Section 4.2 we derived an expression for C° (the noble metal con- centration in fuel salt) where the fuel loop is considered to be a well stirred pot. For this assumption to be adequate, the reside: ;e time of noble metals in fuel salt must be greater than the circuit time of salt around the fuel loop (25 seconds). The rate constant is defined as X = ) + %_(hgr& A8Y8 | hhe Ahe + (hA)rest of loop + hbub Apub) With the parameters listed in Tables B-1 and B-2, we can compute the time constants involved and they are iisted in Table B-5. The value of X in all cases is negligible so the value of X is the same for all noble metal isotopes. In.the case of fuel salt with no bubbles and fuel salt during the 235 V apparently not adequate during the 233U runs. However, valves in this U runs, the well stirred pot assumpfiion_is adequate. It is table assume the sticking fraction of noble metals to bubbles of unity. In Section 5.2 we deduced that the effective sticking fraction to bubbles is considerably less than unity and =a value‘of 0.1 - 0.2 is'suggested.' If a sticking fraction to bubbles of 0.1 is aséumed, then the noble metal half life in fuel salt during the 233U runs becomes 54 seconds. This indicates the well stirred pot assumption is adequate in all cases. 86 Table B-5 Time Constants of Noble Metals in Fuel Salt X Half Life in Salt (hrs-1) (sec) ~ Fuel Salt with no Bubbles 6.7 370 Fuel Salt During the 235y Runs 31.2 . 80 Fuel Salt During the 233) Runs 402 ‘ 6.2 "y £rF 10'—2. O o= A\ £~ W 10. 12. l3l 1k, 16. 17. 18. 1G. 20. 21. 22, 23. 2%, 25. 26. 27. 28, 29. 30. 31l. 32. 33. 3k, 35. 36. 37, 38. 39. Lo, k1. L2, 43. bk, L5, 6. b7, L48. k9. 50. g ;:?au:?-zzg«?:cah-r*p:c4S:m-haufcag‘wjg+pflcafa - » - . * -* - > - » - L - - L4 - * » - » zoa:-mcumomm*dgtq:u:dmbuwzt.—*t.-war*zuwv.—' 51. 52, 53. 54, 55. 56. 5T. 58, 59. 60.-6L, 65. 66. 67. 68. 69. 70. T1. 72. 73. Th, 5. 76. T7. 78. 79. 80. 81. 82. 83. 8k, 85.. 86. 87, 88J 891 90. 91. 92. 93. gly, 95. 96. 97. 98. - 99. 100. 101. 102, 103. 104, * f—qfibmibfi'dibs:“ltdm'::t'.m()!flm:GEAEU*—:’:SPQEHQQO‘;U*UE*UUKGWML«'JU ORNL-TM-3884 Helms Hightower Hise Hitch Hoffman Holmes Holz Huntley .. Kasten Kedl . Kee Kennedy Keyes Kirslis Koger Krakoviak Kress Lindauer . Lundin . Lyon MacPherson MacPherson Matthews, AEC-OSR McCoy « MeDuffie A, McLain McNabb E. McNeese P. Malinauskas L. Manning Mamantov Meyer Moore Moorehead Nichols Nicholson Noggle Ohr . . Perry Pickel Pollock Prince Ragan Ragan | Redman Richardson Robbins Robertson Ross Sanders - - HEEFEHQE HunHsS gy D R = =G Wk . MOOURUOHEWERROE YSH® 105. 106. 107. 108, 109. 110. . 111. 112, 113. 11k, 115, 116, 117. 118. 119. 120. 121, 122, ‘H. C. Savage "W. F. Schaffer Dunlap Scott J. H. Shaffer M. J. Skinner G. M, Slaughter 0. L. Smith - I. Spiewak R. A. Strehlow J. R. Tallackson 0. K. Tallent R. E. Thoma L. M. Toth D. B. Trauger H. L. Watts C. F. Weaver A. M, Weinberg J. R. Weir 1k0. 141.-142. 143, 14k, 145, 146. 1L, 148, 1k9. 150. 151. - 152. 153. 154, 155-156. 88 INTERNAL DISTRIBUTION Contd. 123. M. E. Whatley 12k, J. C. White 125. R. P. Wichner 126. M. K. Silkinson 127. W. M. Woods 128, Gale Young 129, H. C. Young 130. J. P. Young 131. E. L. Youngblood 132. MSRP Program Manager, AEC, Wash. 133.-134. Central Research Library 135. Document Reference Section 136.-138. Laboratory Records 139. Laboratory Records (LRD-RC) EXTERNAL DISTRIBUTION M. Shaw, AEC, Washington N. Haberman, AEC, Washington D. F. Cope, AEC-OSR. Kermit Laughon, AEC-OSR A. R. DeGrazia, AEC, Washington David Elias, AEC, Washington J. E. Fox, AEC, Washington W. B. Hannum, AEC, Weshington E. C. Kovacic, AEC, Washington D. R. Riley, AEC, Washington M. A. Rosen, AEC, Washington F. N. Watson, ABEC, Weshington "M, J. Whitman, AEC, Washington Research & Technical Support Division, AEC Technical Information Center, Oak Ridge