ORNL/TM-5503 Temperature Gradient Compatibility Tests of Some Refractory Metals and Alloys in Bismuth and Bismuth-Lithium Solutions | R. DiStefano B J. 0. B. Cavin Printed in the United States of America. Available from National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road, Springfield, Virginia 22161 Price: Printed Copy $4.00; Microfiche $2.26 This report was prepared as an account of work sponsored by the United States Government. Neither the United States nor the Energy Research and Development Administration/United States Nuclear Regulatory 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. Wy ‘ ORNL/TM-5503 Distribution Category UC-76 Contract No. W-7405-eng-26 METALS AND CERAMICS DIVISION TEMPERATURE GRADIENT COMPATIBILITY TESTS OF SOME REFRACTORY METALS AND ALLOYS IN BISMUTH AND BISMUTH-LITHIUM SOLUTIONS J. R. DiStefano and 0. B. Cavin Date Published: November' 1976 Sl NOTICE — E : eport was prepared as an account of work | 1 mmd by the Unitod States Government. Neither | ' i lthe United States mor the United States Energy i | mesearch and Development Administration, nor any of} i Jtheir employees, mor any of their contractors, | subcontractors, or their employees, makes any ; L'wasranty, sxpress or implied, or assuines sny legal * | dability or responsibitity for the sccuracy, completeness .| or uscfuiness of any informati tus, product or . | process disciosed, or represents that its wse would not i | infringe privately owned rights. OAK RIDGE NATIONAL LABORATORY MASM P ) Oak Ridge, Tennessee 37830 - operated by UNION CARBIDE CORPORATION for the ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION DISTRIBUTION OF THIS DOCUMENT IS UNLIMITED Qfi_ & Abstract .« « ¢ o« ¢ o o ¢ o o Introduction . « « ¢« ¢« ¢ « & Results o+ o o o ¢ ¢ o o ¢ o & Quartz Tests - Group 1 . Quartz Tests - Group 2 . Metal Loop Tests in Bi—2,5% Summary and Conclusions . . . Acknowledgments . . . . . . & REfETrenCeS « o o o o o o o o o o o o o o o o s CONTENTS - iii 11 11 15 30 32 33 33 TEMPERATURE GRADIENT COMPATIBILITY TESTS OF SOME REFRACTORY METALS AND ALLOYS IN BISMUTH AND BISMUTH-LITHIUM SOLUTIONS J. R. DiStefano and 0. B. Cavin ABSTRACT Quartz, T-111 and Mo thermal—convection 1oop tests were conducted at temperatures up to 700°C (100°C AT) to determine the compatibility of several refractory metals/alloys with bismuth and bismuth-lithium solutions for molten salt breeder reactor applications. Methods of evaluation included weight change measurements, metallographic examination, chemical and electron microprobe analysis, and mechanical properties tests. Molybdenum, T-111, and Ta—10% W appear to be the most promising containment materials, while niobium and iron-based alloys are unacceptable. INTRODUCTION A key feature in the conceptual design of the single-fluid molten salt breeder reactor (MSBR) is the connecting chemical processing plant to continuously remove protactinium and fission products from the fuel salt (Fig. 1). Protactinium, the intermediate element in the breeding chain between thorium and 233U, has a significant neutron capture cross section and must be kept out of the core to obtain a good breeding ratio. Rare-earth-element fission products are also neutron poisons and so must be stripped from the fuel. A promising approach for MSBR fuel processing uses liquid bismuth containing dissolved lithium and thorium as reductants to extract protactinium and rare earth elements from fuel salt containing both uranium and thorium (Fig. 2). 1In 1968 the chemical feasibility of this process was demonstrated. . ' One of the requirements for the development of the reductive extraction process is identifying materials that are compatible with both molten flouride salts and bismuth-containing reductants. Hastelloy N (Ni—?% Cr-167% 'Mo—5% Fe) has excellent compatibility with molten salts at 500~700°C;? however, it does not have good compatibility with bismuth. Of ten elements tested, graphite, tungsten, and molybdenum appear most promising (Table 1) and they are also compatible with molten flouride mixtures, Extensive investigations of bismuth as a reactor coolant have shown that inhibitors - such as magnesium, titanium, and zirconium are often required to reduce the high corrosion rates of conventional materials in bismuth.?! However, these types of inhibitors would not be readily acceptable for use in the molten salt system because they would be eliminated from the bismuth stream by several of the proposed steps in processing. - ORNL-DWG 68-1185ER PRIMARY SALT PUMP SECONDARY NoBFy-NaF SALT PUMP COOLANT SALT PURIFIED SALT +— GRAPHITE - MODERATOR REACTOR HEAT EXCHANGER 566°C . - | -CHEMICAL PROCESSING PLANT TLiF ~BeF, - Thiy - UF, FUEL SALT e STEAM GENERATOR TURBO- GENERATOR STEAM Fig. 1. 8ingle-Fluid, Two-Region Molten-Salt Breeder Reactor. N ORNL-DWG 71-9004R FUEL SALT RECONSTITUTION URANIUM PROTACTINIUM RARE EARTH ELEMENTS REMOVAL REACTOR REMOVAL REMOVAL FISSION PRODUCTS TO WASTE Fig;'Z. - Simplified Flowsheet for Processing the Fuel in a Molten- - Salt Breeder Reactor. . - ' , . o : v} Table 1. Solubilities of Some Elements in Bismuth at 600°C Element Solubility, ppm -~ Reference W -~ Not detectable 3 C | - <1 > , 4 Mo | <1 S 536 Be 6 - 6 v 45;110 . 11 Fei o =10 | - 5 Cr : 145 5 Nb o 100200 8 Co ' 590 9 Ni | 65,750 | 10 The low solubility of graphite, tungsten, and molybdenum in bismuth has been confirmed in compatibility tests by other investigators, and, in addition, rhenium, tantalum, tantalum alloys, and certain ceramic oxides and carbides also appeared to be promising for bismuth containment.12_17 In a high temperature isothermal system thermodynamic equilibrium among components is generally reached relatively quickly; the compatibility of the system can often be predicted if phase diagrams of the components are known. When there are thermal fluctuations ‘and/or gradients within a system and the equilibrium constant for a corrosion reaction is temperature dependent, continuing corrosion can occur. For example, 1if iron dissolves in bismuth, and the solubility increases with temperature, simultaneous dissolution and deposition of iron will occur in the system. Under steady state conditions the solution or deposition rate equation for a given location can be expressed. J K( Cear = ) where J is the flux of material entering or leaving the liquid metal, X is a rate constant, C,,. 15 the equilibrium solubility of the solute, and C is the actual concentration of the solute. The driving force for either dissolution or deposition is directly proportional to the concentration differential that exists in the system at a particular location and the appropriate rate constant for the rocess. Our data (Table 1) agree with previous compatibility experiments'? that suggest that C is very low for graphite and the refractory metals tungsten, rhenium, moiybdenum, and tantalum. Niobium and iron are somewhat less promising, but still show relatively low solubility in bismuth. The reductive extraction process for protactinium removal uses bismuth containing small concentrations of lithium, and the metal transfer process for removing rare earth element fission products from the fuel ‘salt uses bismuth containing 2-3 wt %Z Li. Since the solubility of materials in bismuth-lithium solutions could be appreciably different from their solubility in pure bismuth (Table 1), tests had to be conducted in the appropriate chemical processing solutions. | Thermal convection loops used in these tests were fabricated from quartz, T-111 (Ta—87% W-2%Z Hf), and molybdenum. The quartz loops were satisfactory for testing in pure bismuth and in bismuth containing up to 0.01 wt Z lithium. No quartz loops with higher lithium concentrations were operated because of the reaction between Si02 and lithium. Loops fabricated from quartz were relatively inexpensive and could be operated in an alr environment. Samples of the materials being tested were suspended in the vertical hot- and cold-leg sections (Fig. 3). To prevent the samples from floating, they were attached to a quartz rod that was held at the top of each leg. Several types of sample geometries were used, including flat tabs, cylindrical tubes, and cylindrical and sheet tensile specimens. - External heaters were placed on each leg to preheat the loop above the melting point of bismuth or the Bi-Li solution. A pot containing the - melt to be circulated was attached by a mechanical connector to the bottom of the cold leg (Fig. 3.). Before filling, the loop was evacuated so a pressure differential would force the liquid into the loop. During operation, the metal line below the quartz was unheated to allow formation of a solid plug to serve as a freeze valve. When the test was completed, this section of the line was heated to allow the fluid to drain back into the attached pot. The design of the metal thermal convection loops was similar to that of the quartz loops (Fig. 4). Two sections of 7/8-imn. OD X 0.050 in. tubing were bent to form one vertical-horizontal segment (Fig, 4). The segments were joined by making gas-tungsten-arc saddle welds at the top and bottom of the vertical legs. Tubular sheet and tensile specimens were suspended in each of the vertical legs. Wires linked the specimens together to ridigly attach the specimen chain to the vertical legs at the top and bottom. The metal loops were used to test a bismuth-lithium alloy containing 2,5 wt Z Li. This alloy was prepared and purified in a molybdenum-lined stainless steel container (labeled "transfer pot" in Fig. 4)., The lines extending from the top of the loop surge tank were made of tantalum for ease of fabrication and corrosion resistance. A commercially-produced Nb—-1% Zr-type 316 stainless steel dissimilar metal joint connected the transfer pot to the loop. The Nb—1% Zr end was welded to the tantalum tube; a mechanical fitting connected the stainless steel end to the , transfer pot lines. Two types of heaters were used: The main hot-leg heater contained a radiating tantalum element while standard Calrod heaters were strapped to other portions of the loop (Fig. 4). The molybdenum loop is shown (Fig. 5) prior to being installed in a vacuum chamber where the test was runm. _ vl * Fig. 3. and Cold Legs. Y-93581 Quértz Thermal Convection LooP'w:Lth.Metal Samples 1in Hot " ORNL-DWG 74-12762 TO VACUUM/ ' Ly THERMOCOUPLE WELL THERMOCOUPLE WELL = TO vACUUM CLOSED FILL LINE — SURGE TANK MOUNTING FRAME ~ VERTICAL LEGS —___ || P L N } Lt HOT LEG HEATER b Ly i i | TEST SPECIMENS - L TYPICAL 2 L b t CALROD HEATERS {3 LEGS) A ,s POST TEST h_, DRAIN LINE - OUMP TANK @m Fig. 4. Design of Metal Thermal Convection Loop Apparatus. (i TR L5k kA ¥ ER T Blish LR Vo Molybdenum Thermal Convection Loop Prior to Test in Fig. 5. Vacuum Chamber. After evacuation and baking out of the vacuum chamber, the entire loop was heated to 500°C, since the melting point of Bi—2.5 wt % Li is slightly less than 500°C (Fig. 6). As the transfer pot was heated to 600°C (900°C in the case of the molybdenum loop), the loop was evacuated, and differential pressure was used to transfer the bismuth-lithium solution into the loop. During operation the tantalum heater heated the hot leg, but the required temperature differential was maintained with little or no heat to the rest of the loop from the Calrod heaters. To stop the test, the heaters were turned off and the bismuth- lithjum solution was allowed to solidify. When the system had cooled, the vacuum chamber was opened and the bottom of the cold leg cut off. A drain tank was placed immediately under this portion of the cold leg, the chamber re-evacuated, and the loop heated to approximately 600°C to allow it to drain. Samples generally_required cleaning before meaningful weight and | dimensional measurements could be obtained. Several different techniques removed the bismuth or Bi-Li that adhered to the samples. If the amount were small, mechanically cleaning the samples was sufficient. A more satisfactory technique was to amalgamate the bismuth by dipping the sample into hot mercury (approximately 100—-150°C) and then removing the amalgam mechanically. Bismuth used in these studies was Grade 69 obtained from Cominco American Corporation. Except for Quartz loops 1 and 2, the as~-received material was first purified by bubbling hydrogen through the molten metal for two hours at 350°C, in a molybdenum-lined container. Initially the exit gas burned intermittently and had a reddish hue. Subsequently the gas burned continuously and was almost colorless indicating that reaction with hydrogen had ceased. If an alloy were required, solid lithium was added to solid, purified bismuth, and the mixture was heated above the melting temperature of the alloy. The phase diagram shows that the melting temperature of the alloy varies with lithium concentration (Fig. 6). Additionally, in the Group 1 series (see results section) the bismuth or bismuth-1ithium solution was filtered prior to introduction into the loop through a type 316 stainless steel filter having openings of 10 um. Determinations of various impurities before and after the various treat- ments (Table 2) showed no significant improvement in the purity of the bismuth; however, visual observations of the exit gas flame and the appearance of the bismuth indicated that some purification had occurred during hydrogen firing. The addition of more lithium to bismuth as required for the two metal thermal convection loop tests was much more difficult because (1) the melting temperature of the alloy is much higher, (2) the bismuth-lithium reaction is strongly exothermic, and (3) during alloying there is a ‘tendency to form LizBi (Bi—9% Li), which melts at 1150°C. A special apparatus was used to prepare Bi—2.5% Li for these tests (Fig. 7). Lithium was first purified by hot gettering with zirconium foil at 800°C, and the required amount transferred into a type 304 stainless steel container. Sequential steps were then as follows: 1. Solid bismuth was loaded into a molybdenum-lined stainless steel container and the container was sealed under argon. ORNL-DWG 75-9963 WEIGHT PERCENT BISMUTH 60 80 9092 94 9% 97 28 99 1200 g MEFT T 7 | 1100 /M R W1/ /1N 8 \\ LiyBi / / LiBi TEMPERATURE (°C) 3 Q 8 'l 415° by, 400 ' 300° _ \ | LiBi(H) ! 300 | 27 ! 195.5) 200 1R 15) ] T 475 (180% > _j, 5 100 T o . 0O © 20 30 4 S € 70 8 %0 100 ) ATOMIC PERCENT ISMUTH Bi ‘Fig. 6. Lithium-Bismuth Phase Diagram. Table 2. Concentration of Impurities in Bismuth Used in Quartz Loop Tests, ppnm c:::‘;:fl‘n o er Fe Mn M Mo c Hy 0z As-received <5 <5 <5 <5 <1 <20 <1 <6 | After hydrogefi ireatment 2 8 1l 3 | 0.1 - NDa. 5 9 Filtered through stainless 20 30 3 0.2 w* s 7 steel screen aND = Not determined. TO OIL BUBBLER-VENT T ORNL-DWG 74-12579 304SS LITHIUM FILL LINE Tovi Ay TO JRGON AND BLOW BACK LINE Vi 304SS THERMOCOUPLE WELL 30455~ 3 Va4 TO VACUUM <1 N\ — 304584V, 1o ] ~"ARGON T ’ ' : ml A ot : / { o } } - E ; HYDROGEN § / TO ARGON 1~ MOLYBDENUM TUBES | YN et 4—304SS 2. : | THERMOCOUPLE ; MOLYBDENUM LINER | WELL | ? i=-304LSS : PISMUTH / 4| | LITHIUM 4 iy i< 304s5S e 4 in,———— 25 in—» Fig. 7. System Used to Prepare and Purify Bi-Li Alloy. ot 11 2. The bismuth container was evacuated and heated at 650°C. 3. Hydrogen was bubbled through the molten bismuth for approximately 24 hr. 4, The temperature of the bismuth was raised to 700°C; the lithium was heated to 250°C in a separate container. 5. Lithium was transferred into the bismuth container by pressure differential. Molten lithium was injected below the liquid bismuth surface (Fig. 7). After alloying a sample was taken and a typical analysis was lithium-2.4%, hydrogen-l4 ppm, and oxygen-90 ppm. Nominal operating conditions for all of the loop tests was 700°C maximum hot leg temperature and a minimum cold leg temperature of 590— 625°C. For the quartz loop tests Cr-Al thermocouples were used; they were located in wells protruding through the quartz into the liquid stream. Thermocouples made of Pt, Pt—10%Z Rh were similarly inserted into wells through the T-111 tubing wall to measure temperature in test CPML-1. In test CPML-2 Pt, Pt—10% Rh thermocouples were again used, but they were strapped to the outside surface of the molybdenum tubing. All quartz loops were operated in an external air environment. Refractory metal loogs were operated in a vacuum chamber at pressures of 107" to 10' Pa (10~° to 10~7 torr). RESULTS Compatibility results from these experiments can be grouped according to the materials tested and the lithium concentration of the bismuth, Bismuth and Bi—0.01% Li solutions (Groups 1 and 2) were circulated in quartz loops containing samples of various metals (Table 3). A solution of Bi—2.5% Li (Group 3) was circulated in a molybdenum loop (CPML-1) for 3000 hr and in a T-111 loop (CPML-2) for 8700 hr. 1In each case the loops contalned samples of the same material as the loop itself. Alloys investigated were based on molybdenum, tantalum, niobium, and iron. When Nb or Fe-based alloys were included, the tests operated for relatively short times before the flow stopped (Quartz tests — Group 1). All tests using only molybdenum alloy or tantalum alloy samples completed their scheduled operating period (up to 10,000 hr) (Quartz tests — Group 2). Quartz Tests - Group 1 , Quartz loops 1 and 2 contained samples of TZM, tantalum, Nb—1 Zr and niobium and operated for 18 and 23 hr respectively before flow stopped. In loop 1 flow was stopped by a plug which formed at the top of the cold leg. The quartz fractured at this point, exposing the bismuth to air. The samples remianed immersed in bismuth, but the weight changes (Table 4) may reflect exposure to air as well as mass~transfer effects, Loop 2 stopped flowing after 23 hr. Although a solid plug had again formed, it did not fracture the quartz. Samples of the plug were analyzed spectro- graphically and niobium was found to be the major constituent of the plug. 12 Table 3. Summary of Quartz Loop Tests (700°C max, ~100°C AT) Time Loop Samples Fluid * (hr) , : : Group 1 1 TZM2/Ta/Nb/Nb~1% Zr Bi 185 2 TZM/Ta/Nb/Nb—1%Z Zr Bi 23 4 Nb/Nb—1% Zr - Bi 115, 9 Fe—5% Mo Bi 423 Group 2 _ 3 Mo/T2M® Bi 3,000° 6 Ta/T-111 Bi 3,000° 10 Mo brazed with Bi 2,1002 . Fe-based alloys 3,000 17 Ta—10% W Bi 3,000° 7 Mo/TZM Bi—0.01% Li 3,000° 11 Mo Bi—0.01% Li 3,0002 | 12 T-111 Bi—0.01% Li 3,000° 13 Ta Bi—0.01% Li 3,000, 18 Ta—10% W Bi—0.01% Li 3,000° 15 Mo Bi—0.01% Li 10,000° o] TZM is Mo—0.5%Z Ti—0.1% Zr. bFlow stopped. cTerminated, samples dissolved. d -111 is Ta—8%7 W—-2% Hf. eCom.pleted. Table 4. Weight Changes in Samples From Quartz Loop 12 Material Location We%iz;cgginge C?;;?i;gfiyzite Tantalum Hot leg 0.4 —=195 | TZM Hot leg -15.9 —-7,738 Nb*i% Zr Hot leg —54.3 '26’42fi;f}gn Niobium Hot leg —31.0 15,084, ; Tantalum Cold leg +0:3 !*Ifi61 TZM Cold leg —11.0 ;:i%% Nb—1% Zr Cold leg 4.9 —2,385; Niobium Cold leg -1.3 —633 @After 18 hr in bismuth at 700°C maximum temperature differential. temperature and 100°C 13 X-ray diffraction also indicated the presence of small amounts of Bi203-Nb20s. To eliminate the possibility that oxygen contamination of bismuth might be responsible for these results, bismuth used in subsequent group 1 tests was purified by bubbling hydrogen through the molten metal for 2 hr at 350°C and then filtering as described previously. Loop 4 contained niobium and Nb—1Z%Z Zr samples and was operated at a maximum temperature of 705°C with a temperature differential of 75°C for a period of 115 hr before flow stopped. ZX-ray diffraction analysis of a sample taken from the plug that had formed in the cold leg showed the presence of several phases: bismuth, alpha zirconium, and several unidentified phases. ' Niobium was not detected by x-ray diffraction analysis, but the atomic radius of niobium is the same as bismuth and it could occupy substitutional lattice sites in the bismuth unit cell without affecting the lattice parameter of the bismuth. Spectrographic analysis of a sample of the plug materials indicated that it contained 0.5% Nb, 100 ppm Zr, 300 ppm Ni, 500 ppm Fe, with the remainder bismuth. The iron, nickel, and chromium probably dissolved in the bismuth during transfer through stainless steel lines and filter during loading!of the loop. The samples from this test are shown (Fig. 8). The Nb—1% Zr sample from the hot leg was almost completely dissolved. Bismuth remained on the samples after draining and was not successfully removed; therefore, weight change data were not recorded. Fig. 8. Niobium and Nb—1% Zr Samples From Quartz Thermal Convection Loop 4. K i 14 Photomicrographs of selected samples are shown (Fig. 9).. The niobium -sample from the hot leg is thicker than it was originally; however, the unattacked center section is less than 0.020 in. compared to an original sample thickness of 0.027 in. In electron probe scanning images of this sample (Fig. 10) the layer adhering to the niobium appears to be primarily bismuth, but the particles within the bismuth are niobium. In niobium—17% zirconium samples from hot- and cold-leg regions of ~loop 4 (Fig. 9) the cold-leg sample has a greater unaffected center section (about 0,027 in.) than the hot-leg sample, and has greater total thickness (about 0.040 in. compared with 0.030 in.). This suggests that there has been considerable mass transfer of material from hot- to cold-leg sectioms. Loop 9, which contained Fe—5% Mo samples, operated for 423 hr before a power failure stopped the test. After 216 hr one of the samples from the hot leg came loose from its wire holder and floated to the top of the bismuth. When the loop operation was stopped after 423 hr this was the only sample recovered; all other samples apparently had dissolved. Analysis of a sample of the bismuth from the top of the cold leg determined that it contained 13% iron and 0.018%7 molybdenum. The unheated area of the cold leg above the samples was the coldest portion of the loop and is probably where deposition occurred. If so, bismuth flowing past the cold- _Y-132369 s C.035 INCHES — 030 m T wox 1000 m 1 - -Iz‘ig. 9. Niobium and Nb—1Z Zr Samples From Quartz Thermal Convection Loop 4. ' ‘ N 15 Y -1002%0 Back-Scattered Elecirons Fig. 10. Electron Probe Scanning Images of Niobium Bot Leg Sample From Quartz Thermal Convection Loop 4. leg samples could continue to have a strong dissolution effect. Static capsule tests of this alloy for 600 hr at 650°C in bismuth did not show detectable dissolution, thus indicating a strong influence of the rate constant X (Eq. 1) on mass transfer of iron in temperature gradient systems involving bismuth. Quartz Tests - Group 2 Ten quartz loops were tested with samples of tantalum (alloys) and/or molybdenum (alloys).. Loops 3, 6, 10, and 16 circulated bismuth while the remaining six loops circulated Bi—0.01% Li. All of these loops were operated for 3000 hr except for loop 10, which was operated 2100 hr, and loop 15, which was operated 10,000 hr. Loop 10 contained molybdenum samples that had been brazed together with iron-based alloys. Four different compositions were used: Braze Alloy Designation Nominal Composition (wt %) 16 M Fe—0% Mo—4% C-1% B 35 M Fe—~15% Mo—4% C—1% B 36 M Fe—25%Z Mo—4%Z C—1%Z B 42 M Fe—15% Mo—4% C1% B—5Z Ge 16 The loop operation was stopped after 2100 hr and weight changes of the samples were measured (Table 5). They were very small weight changes, and analysis of the bismuth gave little evidence that dissolution had occurred. Iron and molybdenum concentrations were less than 3 ppm, and carbon was &-13 ppm. However, metallographic examination showed that each of the braze alloys contained an outer surface layer (Fig. 11). A more detailed examination was made of the samples brazed with alloys 16 M and 35 M. For the alloys before test no outer surface layer was visible (Fig. 12). An electron beam microprobe analysis determined the distribution of iron, molybdenum, and bismuth in the braze after test. In alloy 16 M the outer surface layer was rich in iron; the white particles contained 88% bismuth, 5% molybdepum, and less than 0.57 iron (Fig. 13). An area adjacent to the white particles contained 41% bismuth, 23% molybdenum, and 4% ironm, while the dark spots were predominantly iron (about 70%Z). 1In alloy 35 M the surface layer appears to be subdivided into two segments (Fig. 14) with the outermost segment rich in molybdenum and the inner segment predominantly iron. Bismuth has completely penetrated the braze (Fig. 15). Thus, although there was little evidence of mass transfer, bismuth did react with the iron-based braze alloys. - The effect of lithium in bismuth on the corrosion rates of Mo/TZM, Ta/T-111, and Ta—10% W is shown in Tables 6, 7, and 8. Maximum corrosion rate (weight loss) of molybdenum in Bi—0. 01% Li was 16.9 mg/cm®-yr compared with 1.64 mg/cm?eyr in pure bismuth. Tantalum and T-111 also showed higher rates of mass transfer in Bi~0.01%Z Li, but Ta—10%Z W showed the opposite effect — a higher corrosion rate in bismuth compared with Bi-0. 01/ Li, where all the samples exhibited small weight gains. Table 5. Weight Cfi;nges in Molybdenum—~Braze Samples Exposed to Bismuth in Quartz Thermal Convection Loop Tests Conditions: 700°C Max Hot Leg Temperature; 95°C Temperature Differential Weight Change (mg) Braze Alloy Hot Leg Cold Leg 16 M (Fe—4%Z C-1%Z B) | - -1.6 Sample not included 35 M (Fe4% C-1% B—15% Mo) —4,2 +12.2 36 M (Fe—4% C—1% B—25% Mo) ~16.6 +5.0 42 M (Fe4% C-1%Z B—-15% Mo—5%Z Ge) 0.2 +2.2 Y-106373 Alloy 16 M (Fe-4C ~18) i Alloy 36M (Fe-25Mo-4C ~18) . NS ~ Alloy 35M (Fe-15Mo0-4C -18) Atloy 42M (Fe-15Mo~4C -1B-5Ge) Fig. 11. Brazed Molybdenum Samples After Exposure to Bismuth in Quartz Thermal Convection Loop Test 10, . 18 Y=-132372 ith Alloy 16M Brazed wi ith Alloy 35M Mo Brazed wi Brazed Molybdenum Samples Before Test in Quartz Loop 10. Fig. 12. ’ t Lighter Ke | Fe B 19 Electron Beam Scanning Images of Molybdenum Sample Brazed Backscattered Electrons 13. Fig. regions indicate higher concentration of listed element. o —t o o o - N 4 : o = i = 3 g 0 o K o + 5 o 12 G &l H Q 4 & ~— B o 2 5 = 20 Mola ' FeKa Fig. 14. Electron Beam Scanning Images of Molybdenum Sample Brazed With Alloy 35M After Exposure to Bismuth in Quartz Loop 10. Lighter regions indicate higher concentration of listed element. 21 -Y=-106001 Molybdenum Base Metal Blsmufh la X—Rays ‘Fig. 15. Electron Beam Sg,anning Image of Molybdenum Sample Brazed With Alloy 35M After Exposure to Bismuth in Quartz Loop 10, Bismuth has completely penetrated the b¥aze alloy. e 22 Table 6. Comparison of Corrosion Rates of Molybdenum and TZM Samples Exposed to Bismuth or Bi—0.01% Li in Quartz Thermal Convection Loop Tests for 3000 hr Corrosion Rate (mg/cmZeyr) Location - Temperature — ' (°C) Loop 3 Loop 7 Loop 11 Bismuth B1-0.01% Li Bi—0.01% Li Hot Leg 630 . +0.53 . 48.65 +3.10 640 +1.72% -0.36% —1.14 660 ~1.64 ~16.9 —3.80 690 +1.58 | -1.74 -9.70 Cold Leg 680 +3.80 +3,56 —0.03 660 -1.02 . ~0.58 +0.26 640 +0.90% +0.59% +0.44 615 +1.43 0o +1.34 VA sample. Table 7. Comparison of Corrosion Rates of Tantalum and T-111 Exposed to Bismuth and Bi—0.01% Li for 3000 hr in Quartz Thermal Convection Loops Corrosion Rate (mg/cm?¢yr) Temperature Location (°C) Loop 6% Loop 13b Loop 12° (Ta/T-111, Bismuth) (Ta, Bi—0.01% Li) (T-111, Bi—0.01% Li) Hot Leg 630 +13.73¢ - -16.8 +0.82 660 #0.759 418 —2.28 660 ' —47.16 ' —6.5 +1.08 690 ~79.79 ~137 -13.6 Cold Leg 680 ~17.50 —68.9 —0.85 660 . —24.40 —47.7 +0.70 640 +0.15¢ —46.1 | —1.46 615 +0.29¢ ~112.4 +0.44 220 ppk tantalum in bismuth after test; less than 2 ppm before test. b30—-150 ppm tantalum in bismuth after test; less than 1 ppm before test. €20-800 ppm tantalum in bismuth after test; 3 ppm before test. T-111 samples, others were tantalum samples. 23 Table 8. Comparison of Coxrrosion Rates of Ta—10% W Samples Exposed to Bismuth or Bi—0.01% Li in Quartz Loop Tests for 3000 Hr Corrosion Rate (mg/cm?eyr) | Location : Te???;?ture Loop 17 Loop 18 J (Bismuth) (Bi—0.01% Li) Hot Leg 630 —1.40 : +0.22 640 ~3.65 +0,20 660 a +0.26 | 690 - a —0.37 Cold Leg 680 | ~2.86 +0.46 ' 660 —1.43 40,13 640 —2.48 - 40.35 615 —1.23 +0.13 aSample damaged during bismuth removal. - The metallographic appearance of Mo and TZM samples from loops 3 and 7 is shown (Figs. 16 and 17)., Hot-leg samples of molybdenum from both of these tests show grain boundary corrosion to about 0.0005 in. Little or no attack on TZM was noted. There was a layer of extremely fine grains at ; the surface of the molybdenum which were attacked. intergranularly. | grains were probably caused by residual cold work from machining and | subsequent recrystallization at the test temperature. If these grain boundaries picked up iron during machining, preferential attack might be expected. These Room temperature mechanical property tests of the molybdenum samples showed their properties to be changed only slightly from those of the as-received samples (Table 9). Metallographic examination of the tensile - samples showed there was no effect of the intergranular attack of the fine surface grains. Chemical analysis of the melts (Table 10) showed slightly higher molybdenum in Bi—0.01% Li after testing. In general, the results indicate that the addition of 0.01% Li to bismuth had little effect on its compatibility with molybdenum. The alloy Ta—10% W showed similar resitance to both ‘bismuth and Bi—0.01% Li (Fig. 18). A hot-leg sample (from loop 17) exposed to - bismuth was slightly corroded to a depth of approximately 0.0005 in., but there was no visible corrosion of any of the specimens exposed to ! Bi—0.01% Li. Chemical analyses of the melts after test were similar to : : those before test, indicating that Ta—10% W has good corrosion resistance to both fluids. 24 Y-99457 Control Hot Leg Cold Leg Mo Fig. 16. Molybdenum and TZM Samples From Quartz Loop 3. Fig. 17. Molybdenum and TZM Samples From Quartz Loop 7. 25 Table 9. Effectvof:E2posure to Bi—0.01%Z Li in Quartz Thermal Convection Loops on the Room Temperature Tensile Properties of Refractory Metals 7 Material Sample location - Molybdenum Tantalum | CT-111 Ta—10Z2 W ‘ ’ uTs® Elong. urs? Elong. - yrs? Elong. urs? Elong. (10? psi) (2) (10% psi) (1) (10% psi) ) (10% psi) ) As-received 110 - 8.7 &4 24 94 10 84 14 Hot leg 690°C 11 11.3 47 23 52 0 87 8 Hot leg 660°C 111 8.0 47 26 61 0 92 10 Cold leg 660°C 112 11.3 50 26 88 1 88 7 Cold leg 640°C 112 11.3 47 25 47 0 89 7 aUltimate tensile stress. Table 10. Composition of Bismuth and Bi—0.01% Li Melts Before and After Exposure to Molybdenum and TZM in Quartz Loops 3, 7, and 11 Concentration (ppm) Loop Time Sampled - ‘ : _ c H N 0 ‘Li Mo -~ Ni si Ta Zr '3 Beforetest 10 5 2 8 0.1 <0.5 30 8 3 0.3 After test: 9 <1 <1 2 0.1 <0.5 30 8 3 0.3 7 Before test 9 1 2 100 0.5 70 8 0.3 After test 30 1 77 60 4 70 10 11 Before test 22