ORNL-TM-4189 EVALUATION OF HASTELLOY N ALLOYS AFTER NINE YEARS EXPOSURE TO BOTH A MOLTEN FLUORIDE SALT AND AIR AT TEMPERATURES FROM 700 TO 560°C K J. W. Koger 0 7, VY S DOCUMENT CONFIRMED AS THIS® UNCLASSIFIED DIVISION OF CLASSIFICATION J% UPERAIED BY U'ON (ARBIDE (ORPORATION I FOR THE US ATOMI( ENERGY (OMMISSION'. | This report was prepared as an account of work sponsored 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. €« " "'_l # T 'c) ORNL-TM-4189 Contract No. W-7405-eng-26 METALS AND CERAMICS DIVISION EVALUATION OF HASTELLOY N ALLOYS AFTER NINE YEARS EXPOSURE TO BOTH A MOLTEN FLUORIDE SALT AND AIR AT TEMPERATURES FROM 700 TO 560°C J. W. Koger DECEMBER 1972 NOTICE - This report was prepared as an account of work sponsored 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, com- pleteness or usefulness of any information, apparatus, product or process disclosed, or represents that its use’ would not infringe privately owned rights, OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37830 operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION PISTRIBUTION OF THIS DOCUMENT IS UNLIMITED " a ¥ " » r & - » ¥y (3] /*\ CONTENTS ADSETaCT . ..ttt ittt eeiate ettt ettt e 1 Introduction . . .. ... i i i et ceie ettt eecaee et e 1 Experimental ProcedureandMaterials . ..... ... .. ... i ittt ittt 3 ReSUIS . .. i i i i e i et it ettt i e et 5 Loop Failure ........ ... i i i i ittt eeateeraeaanea s 5 MassTransfer .................... ... e et e et aee et aeeeieee et 11 AirOxidation .......... ... .. i, [N 19 Weld Corrosion Resistance ........ ettt e teaeeateae et ettt i, 20 Discussion ................ e e e e e e et e e et a e e enn 26 Temperature-Gradient Mass Transfer ... .......... it iirieirnnnnenrnnnnnns 26 VOid FOIMAtION . ..ottt ettt et e et e e e et e e e e et e e e e e e e e eeaeanan 29 Comparison of Mass Transfer in Loop 1255 and Another Hastelloy NLoop ................... 29 Comparison of Mass Transfer in Loop 1255 and a Type 304L Stainless Steel Loop .............. 31 Mass Transfer Calculations . . ....... .. ...ttt e et een s eonannamannennnn 31 Failure Analysis .. ... .ottt it i it it ettt i et i teeaeassnnseanannnnnn 34 AIrOxidation .. ... e e 34 Weld Corrosion Resistance . ... ... e e e e e 34 Coficlusions ................ ..................... . 35 iii * ™ n fe EVALUATION OF HASTELLOY N ALLOYS AFTER NINE YEARS EXPOSURE - TO BOTH A MOLTEN FLUORIDE SALT AND AIR AT TEMPERATURES FROM 700 TO 560°C J. W. Koger o e T ABSTRACT | A Hastelloy N thermal convection loop, some portions of which consisted of a Hastelloy N alloy modified with 2% Nb for improved weld properties and containing LiF—23 mole % BeF,-5 mole % - ZrF4—1 mole % UF4—1 mole % ThF4, was operated for 9.2 years at a maximum temperature of 700°C and a minimum of 560°C. Loop operation ended with the occurrence of a salt leak. The failure of the loop was attributed to a reaction between impurities in a ceramic bushing and the modified Hastelloy N—2% Nb tubing. Microscopic examination of the loop tubing revealed that mass transfer of material (material removal and deposition) did occur. The attack, which occurred in the hot section, was manifested in the formation of voids, in a zone of maximum depth of 4 mils. Deposition was noted on the colder portions. On the basis of salt analysis and microprobe analysis of the tubing, the " mass transfer appeared to be selective with respect to chromium, which is what would be predicted from thermodynamic considerations. The actual void formation and chromium depletion agrees favorably with that predicted from calculations. No difference in corrosion could be seen in the standard Hastelloy N and the modified Hastelloy N—2% Nb alloy. No change in mass transfer could be seen in-the welded areas. A two-layer oxide of 2 mils thickness was the maximum formed under the heaters in 9.2 years exposure to air. Hastelloy N is much more resistant to mass transfer than type 304L stainless steel exposed to the same salt under similar conditions. It was concluded from this _experiment that Hastelloy N is suitable for long-term use as a container material for the molten salt used in this test and has acceptable air oxidation resistance at the temperatures tested. o INTRODUCTION ‘Hastelloy N (initially known as INOR-8, nominal composition 72% Ni—16 % Mo—7 % Cr—5 % Fe) was developed at the Oak- Ridge National Laboratory (ORNL) in the Aircraft Nuclear Propulsion (ANP) Program and was viewed as the most promising container material for molten fluorides exposed to the severe (800°C) ANP conditions.! Actual use of Hastelloy N in the Molten Salt Reactor Experiment (MSRE) and in many experimental programs has shown that the alloy is an effective container material for molten fluoride salt under a variety of conditions. However, during routine qualification of welders on Hastelloy N material in 1961 (before widespread use of the alloy) the presence of a possible weld-cracking problem was detected in one heat of the alloy. Both bend tests and metallographic examination revealed the incidence of cracking, and a cursory examination of the welding procedures indicated no obvious remedy. Because of the importance of this problem Lan investigation was immediately started to determine its seriousness and to develop preventive methods. In one effort to overcome the weld-metal cracklng difficulties, an expenmental Hastelloy N weld-metal composmon containing 2 wt % Nb was investigated. This filler metal was developed in the course of the Hastelloy N welding program conducted by the Welding and Brazing Laboratory of ORNL and was considered exceptionally promising in view of its excellent elevated-temperature mechanical properties.? Welds made on the suspect heat of plate exhibited no evidence of weld-metal cracking, either in bend tests or in metallographic sections. Wire from two laboratory melts of this alloy was deposited with no evidence - of defects in the weld metal, indicating that use of this alloy would prevent weld-metal cracking. ! 1. W. D. Manly et al., “Metallurgical Problems in Molten Fluoride Systems,” Progress in Nuclear Energy, Series IV, vol. 2, Technology, Engineering and Safety, Pergamon Press, 1960, pp. 164—79. 2. MSR Program Quart. Progr. Rep. July 31, 1959, ORNL-2799, pp. 71-72. # ,} Fig. 1. Thermal convection loops in operation. Because of the possible widespread use of the modified alloy in a molten fluoride salt environment, corrosion tests were scheduled to evaluate the resistance to salt attack of Hastelloy N that had been modified by the addition of 2% Nb. In addition to corrosion data on Hastelloy N—2% Nb, it was planned that these tests would also supply data on the corrosion properties of various types of weld junctions listed below: 1. Hastelloy N welded with Hastelloy N weld rod, 2. Hastelloy N welded with Hastelloy N—2% Nb weld rod, " 3. Hastelloy N—2% Nb welded with Hastelloy N—2% Nb weld rod. Since molten salts in a reactor system would most likely be used as heat transfer fluids, the corrosion test was conducted in a loop system which would provide flow and a temperature gradient. The test system used was a Hastelloy N thermal convection loop (deSignated loop 1255) similar to the ones shown in Fig. 1. The temperature gradient is produced by heating a portion of the loop while insulating or exposing the remainder of the loop to ambient air as necessary to provide the desired temperature difference. The salt flow results from the difference in density of the salt in the hot and cold portions of the loop. The maximum temperature chosen for this test was 700°C and the minimum 560°C. The velocity of the salt T resulting from this temperature difference was approximately S fpm. L,/ oy 0. 3 The experiment was started April 11, 1962, and, with only a few minor interruptions, operated until July 20, 1971. On the day operation was ended, the amperage t6 the loop heaters decreased to one-half of normal. Investigation revealed that half the main heaters were open and, in addition, grounded (which often indicates the presence of salt); At this time, all power to the loop was shut down. After cool-down, the insulation and heaters were removed, revealing a small amount of salt on the outside of the tubing, During the last few years, we planned to operate the loop as long as possible to provide long-term data on mass transfer in a fluoride salt, air oxidation of Hastelloy N, and corrosion properties of various weld junctions. Thus, the failure provided an end to the experiment. In this report, we will discuss the loop failure, the compatibility of the Hastelloy N alloys and their welds with the salt, and the air oxidation of the alloy. The results of this experiment are still quite significant in that recent developments® have indicated that certain Hastelloy N alloys containing additions of Ti, Hf, Nb, and Zr have good mechanical properties (better than standard Hastelloy N) after being irradiated at 760°C. Thus, the possibility of using a modified alloy similar to the one tested in this experiment still exists. | EXPERIMENTAL PROCEDURE AND MATERIALS Figure 2 is a photograph of a typical thermal convection loop with heaters and thermocouples installed but without insulation. Thermocouples are located at the top of the hot and cold leg, at the bottom of the cold leg, at the insert specimen locations, and over the modified Hastelloy N tubing. A schematic view of the loop is shown in Fig. 3. The parts of the loop we will refer to"are the hot leg (heated vertical tubing), cold leg (unheated vertical tubing), upper crossover, and lower crossover. The major portion of the loop was constructed of ¥%-in. sched 10 standard Hastelloy N pipe (approx 0.675 in. OD) from heat No. Y-8460 (Superior Tube) and an experimental heat specified as Haynes SP-19. The top 9 in. of the hot leg contained two tube inserts. The top insert was Hastelloy N—2% Nb, and the bottom one was standard Hastelloy N. The inserts were 3-in.-long cylinders, 0.595 in. OD, 0.025 in. wall, which were fitted into machined portions of the loop tubing, thus retaining the same flow cross section throughout the loop.The next 6-in. section of the hot leg was Hastelloy N—2% Nb tubing, and the remainder of the loop was standard Hastelloy N. The Hastelloy N—2% Nb alloy was from experimental heat MP-13. The compositions of the alloys are given in Table 1. i In forming the modified Hastelloy N tubing, a ¥;-in. slab was first rolled and cross-rolled until the final thickness was approximately 0.040 in. The sheet was then formed into a tube and welded with Hastelloy N-2% Nb weld wire. The insert tubing was swaged to approximate size and then machined to final tolerance, while the loop tubing was used as fabricated. 3. H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb. 28, 1969, ORNL-4396, pp. 235-40. Table 1. Composition of alloys Cr Mo Fe Nb Mn ~Si Ni . Hastelloy N—2% Nb 15 154 39 2.1 0.54 N.a. Bal Regular Hastelloy N 7.4 16.7 48 0.48 0.13 Bal (Haynes SP-19) : _ Regular Hastelloy N 7.3 15.9 24 0.31 0.15 Bal (Heat Y-8460) Fig. 2. Thermal convection loop with heaters and thermocouples installed. . The hot portion of each loop was heated by sets of clamshell heaters with the input power controlled | by silicon controlled rectifiers (SCR units) and the temperature controlled by a Leeds and Northrup Speedomax H series 60 type C.A.T. (current proportioning) controller. Ceramic bushings were used to space the heaters from the loop. The loop temperatures were measured by ChromelP vs Alumel thermocouples that were spot welded to the outside of the tubing, covered by a layer of quartz tape, and then covered with Inconel shim stock. | j"" - ORNL-DWG T2-1120R SURGE TANK HASTELLOY N WELDED WITH HASTELLOY N WELD ROD " 700°C 6-in. CLAMSHELL HEATERS HASTELLOY N—2% Nb o INSERT SPECIMEN ° 625°C REGULAR HASTELLOY N LAVA BUSHINGS INSERT SPECIMEN HASTELLOY N WELDED TO HASTELLOY N-2% Nb WITH HASTELLOY N-2% Nb WELD ROD HASTELLOY N—2 % Nb TUBING FAILURE HASTELLOY N—2 % Nb WELDED WITH HASTELLOY N -2 % Nb WELD ROD REGULAR HASTELLOY N TU BING\ o e aQ Q o o ~ 560°C Fig. 3. Schematic view of thermal convection loop. Prior to filling, the loop was heated under a vacuum of 10 u for leak-checking and bake-out. The loop was then filled with flush salt, which was dumped after 2 hr, and then was filled with the operating salt. The salt used was the pmposed MSRE fuel salt with a nominal composition of LiF—23 mole % BeF,—$ mole % Z1F4—1 mole % UF4—1 mole % ThF,. The operating temperatures are given in Fig. 3. ' - This loop had no facility in which to dump the salt at the end of operation, so the salt was frozen in place. To remove the salt, various portions of the loop were cut into small pieces, placed in a graphite-lined nickel container, and heated under argon for 4 hr at 800°C. - RESULTS Loop Failure The loop failure occurred in the middle of the hot leg under the ceramic bushing between two sets of heaters. The bushing material used was grade A Lava, which was manufactured by the American Lava Company. Grade A Lava is hydrous aluminum silicate fired after machining to dnve off the chemlcally bound water and develop a hard electric insulator. ' PHOTO 70514 Fig. 4. Corrosive attack on exterior of Hastelloy N tubing underneath ceramic bushings after §52 hr at 675 to 785°C. PHOTO 70513 Fig. 5. Close-up of deepest exterior corrosion pit on Hastelloy N tubing. Depth of pit, 0.037 in. Background. Metal-bushing compatibility problems have previously existed, and at least one case involving Hastelloy N at ORNL has been reported.*:5 In 1964, ORNL conducted a corrosion program in support of the SNAP-8 electrical-generating system. A “‘chromized” Hastelloy N was used as the fuel cladding and contained NaK. Several instances of excessive corrosion were noted on the exterior of the “chromized” Hastelloy N tubing underneath ceramic bushings similar to those used in loop 1255. The first instarice noted was on a loop after 552 hr of operation at design temperature, 760°C maximum, 593°C minimum. The corrosion was found during replacement of heaters. No NaK leakage had occurred. Exterior pitfing was found on the chromized %-in.-OD, 0.072-in.-wall Hastelloy. N tubing under the ceramic bushings (grade A Lava, unfired) used to space the electric heaters from the tubing (Fig. 4). The attack was noticeable on the tube wall at the location where the temperature was estimated to have reached 675°C, and the attack increased toward the high-temperature end of the tubing, where the temperature was calculated to have been approximately 785°C. The maximum attack found near the hottest end of the tube (Fig. 5) was 0.037 in. deep, as determined by external measurements and x rays of the area. While the exact 4. SNAP-8 Corrosion Program Quarterly Report, Nov. 30, 1964, ORNL-3784, pp. 20-21. 5. SNAP-8 Corrosion Program Summary Report, ORNL-3898, pp. 6062 (December 1965). -+ g1 o ) ~ -2 Fig. 6. Failed region of Hastelloy N hot-spot piping. Loop operated 1300 hr at 785°C. nature of this attack was not determined, it was attributed to oxidation-inducing volatile material driven out of the ceramic bushings. Subsequent to this, all ceramic bushinés of the grade A Lava type were fired at 1100°C prior to installation. | The second case of exterior corrosion on chromized Hastelloy N happened later in the same program on another loop.5:¢ The exterior corrosion occurred on the ¥%-in.-OD, 0.072-in.-thick wall of the hot-spot section, which was operating at 785°C. A NaK leak developed after 1300 hr of operation. Visual inspection of the failed area showed that the Hastelloy' N piping had been excessively oxidized in very localized areas under the fired ceramic insulator bushing in the heater section. Figure 6 shows both the hole through which the leak occurred and an adjacent area with accelerated oxidation that had not quite penetrated the wall. A magnified section taken through the area adjacent to the hole is shown in Fig. 7. The metallographic examination revealed a mixture of metal and oxides in the corrosmn products which x-ray-diffraction analysis proved to be mostly nickel and NiO. The circumstances surrounding this 1nc1dence of catastrophlc oxldatlon suggest that breakdown of the normally protective oxide layer on the pipe exterior surface was attributable either to contamination from some unknown element or compound in the ceramic bushing or to oxygen starvation in the stagnant area under the bushing. An Inconel shim was placed between the ceramic bushings and the tubing to eliminate these conditions in the remaining loops. ' Loop 1255. Figure 8 shows a macrograph and a micrograph of the failed region that ended the operation of loop 1255 (Fig. 3). The tubing in this section of the loop was modified Hastelloy N with the 2% Nb added. The longitudinal weld which joined the tubing is also shown in Fig. 8z. A micrograph of the weld wilt be seen in a later section of the report. Electron beam scanning images were made of the area just to the left of the portion of the tube wall that was attacked (circled area of Fig. 82). This area, shown in Fig. 8b, included the Inconel 600 shim stock {foil) which covered the thermocouple, a spongy oxide which 6. SNAP-8 Corrosion Program Quarterly Progress Report, Nov. 30, 1964, ORNL-3784, p. 34. Fig. 7. Cross section of oxidized region adjacent to hole in failed section of loop. As polished, about 14X. also included frozen salt that had leaked out, and a layered oxide found on both sides of the foil. The electron beam scanning images are shown in Fig. 9. The quantitative values are given in Table 2. The analysis of the spongy-oxide—salt mixture combination shows an agglomerate of oxides and metal which includes the elements zirconium, molybdequm, iron, nickel, and chromium. The composition shown in Table 2 reflects an average of three readings in different areas for each element. The zirconium probably comes from the salt, while the other elements are constituents of the alloy which have been removed from the Hastelloy N during the failure process. Table 2. Composition of foil and oxide in the failed region Weight percent? Ni Fe Cr Mo Zr Foil 900 08 9.1 <05 <05 Spongy-oxide—sait 19.8 1.6 1.9 17.3 11.3 mixture, average Adherent oxide 4.8 0.7 16.2 22.7 1.5 ®Data corrected for abs'orption, fluorescence, and atomic number effects using the theoretical alpha method. g 1 4 WA i i ! 1 i ! i i | i i i i 1 o Flg. 8. Failed region of loop 1255 @ Macrograph of modified Hastelloy N— 2% Nb tubmg and the fallure. Cu'cled area was analyzed with the electron beam mlcroanalyzer (b) Optical micrograph of circled area. 10 svonsr i ;,‘“ “r OXIpE E&cksccflerea Electrons . RM NiKa RM Zrla RM CrKa RM Mola ' . . RM FeKqg Fig. 9. Llectron beam scanning images of area near the failure. The foil, which analyzed as 90% Ni-9% Cr—1% Fe, was documented as Inconel 600, and its purpose was to cover the thermocouples. The nominal composition of Inconel 600 is 76% Ni—16% Cr—8% Fe; thus it was possible that iron and chromium were depleted from the Inconel 600 during the reaction that caused the failure. It is also possible that the foil was not Inconel 600. The adherent metal oxide on this alloy would normally contain only nickel and chromium, but the ahalysis also showed substantial quantities of elements that were present in the spongy-oxide—salt mixture. Thus, this analysis represented a mixture of reaction products. | Figure 10 shows the tubing at the point of failure, Note that the largest amount of material is removed at the outer surface, with the amount of material removed decreasing as the inner surface is approached. This mode of material removal establishes the outside-to-inside direction of the failure process. K by a) ) # LAY 11 b AR - T & P Y-109573 — 10.001 in. & 10.003 in. . 0.007 INCHES 10.005 in. 500X 10.007 in. Fig. 10. Micrograph of modified Hastelloy N—2% Nb at the point of failure. Location C, Fig. 11. Outside surface at - top; inside surface (exposed to salt) at bottom. As polished. Mass Transfer The salt ana}ysis before and after test is given in Table 3. Of importance here is the large amount of chromium after test, representing a large increase, with little if any change in the other constituents or impurities. . IR Figure 11 shows representative micrographs of the inside surface (eprséd to Sa}f) of specimens and tubing completely around the loop. Note the attack and void formation in the heated areas (hot leg) and the deposition in the cooler regions. Figures 12 and 13 show micrographs of the inside surfaces of the insert specimens. Figure 12 shows the modified Hastelloy N—2% Nb insert specimen from the top of the hot leg (A, Fig. 11). The salt temperature at this position was approximately 695°C. Voids extend 2 mils into the matrix. The etched specimen shows that the etching characteristics of the material near the surface are quite different from the matrix. This generally ‘represents depletion of an alloy constituent — in this case, chromium. Figure 13 shows the standard Hastelloy N insert specimen located 6—9 in. below the top of the hot leg (B, Fig. 11). The salt temperature at this location was apprdximately 675°C. These voids extend about 3 mils into the matrix. Again, the etching delineates the depleted area. Table 3. Analysis of salt circulated in Hastelloy N loop ' for nine years before and after test ' Weight percent - Ppm Li Be U Th Zr F Ni Cr I}'ci Mo Nb Béfore 100 560 597 573 994 62.6 40 100 120 After 108 499 6.15 6.28 8.77 629 101 1800 69 23 87 WELD FAWLURE -——E HASTELLOY N WELDED WITH HASTELLOY N ROD HASTELLOY N—-2% Nb INSERT SPECIMEN REGULAR HASTELLOY N INSERT SPECIMEN HASTELLOY N WELDED TO HASTELLOY HASTELLOY N-2% Nb WELD ROD - 1 4 —HASTELLOY N~2% Nb TUBING HASTELLOY N—2 % Nb WELDED WITH HASTELLOY N—2 % Nb WELD ROD Y 112590 ORNL~OWG T2-1120 ! N-2 %Nb WITH Fig. 11. Micrographs of tubing and specimens from loop 1255 exposed to LiF—23 mole % BeF,—5 mole % ZrF4—1 mole % ThF4—1 mole % UF 4 molten salt at 560—700°C for 9.2 years. As polished. 500X. Reduced 15%. o C - 27 ¥Y-111351 Fig. 33. Weld joining two pieces of Hastelloy N tubing. Left side exposed to salt at 580° C right side exposed to air. Etched with glyceria regia. 20X, 3. Due to constituents in the fuel, particularly, , ‘ Cr + 2UF, = 2UF; +CiF, . - | - 3) If pure salt contaihing UF4. (and no corrosion products) is added to a Hastelloy N loop operating - polythermally, all points of the loop initially experience a loss of chromium in accordance with the Cr-UF, reaction, Eq. (3), and by reaction with impurities in the salt (such as HF, NiF,, or FeF,). Impurity reactions go rapidly to completion at all temperature points and are important only in terms of short-range corrosion effects. The UF, reaction, however which is temperature-sensmve provides a mechanism by which the alloy at high temperature is continuously, depleted and the alloy at low temperature is continuously enriched in * chromium. As the corrosion -product concentration of salt is increased by the impurity and UF, reactions, the lowest temperature point of the loop eventua]ly achieves equitibrium with respect to the UF, reaction. At regions of lugher temperature, because of the temperature depen ce for this reaction, a driving force still exists for chromium to react with UF,. ‘Thus, the corrosmn-product concentration will continue to increase, and the temperature points at equilibrium will begin to move away from the coldest temperature point. At this stage, chromium is returned to the walls of the coldest pénnt of the system. The rise in 0.035 INCHES N 100X Fig. 34. Microstructure of weld joining two pieces of Hastelloy N tubing, exposed to salt at §80°C. Etched with glyceria regia. HOT SECTION \\\\\\\\\\\\\\\\\\\\\\\\\‘l\\\ ORNL-DWG 67-6800R DIFFUSION TO SURFACE SOLUTE ESCAPE THROUGH NEAR-SURFACE LIQUID LAYER — ~ T DIFFUSION INTO BULK LIQUID : TRANSPORT TO COLD PORTION OF SYSTEM COLD SECTION —— — — — | x SUPERSATURATION — — — 1 * NUCLEATION - T T o GROWTH TO STABLE CRYSTAL. SIZE - —OR _____ _ - - SUPERSATURATION AND DIFFUSION THROUGH LIQUID ON METALLIC WALL — —— - — OR- DIFFUSION INTO WALL - — —— — Fig. 35. Temperature-gradient mass transfer. rd | i iy 29 corrosion-product concentration in the circulating salt continues until the amount of chromium returning to the walls exactly balances the amount of chromium entering the system in the hot-leg regions. Under these conditions, the two positions of the loop at equilibrium with the salt, termed the “balance point,” do not shift measurably with time. Thus, a quasi-steady-state situation is eventually achieved whereby chromium is transported at very low rates and under conditions of a fixed chromium surface concentration at any given loop position. A schematic of this mass transfer process is shown in Fig. 35. Our results in this experiment show material loss in the hot portions of the loop and material deposition in the cold portion. Void Formation The formation of subsurface voids as seen in loop 1255 is initiated by the oxidation of chromium along exposed surfaces through oxidation-reduction reactions with impurities or constituents of the molten fluoride mixture. As the surface is depleted in chromium, chromium from the interior diffuses down the concentration gradient to the surface. Since diffusion occurs by a vacancy process and in this particular situation is essentially unidirectional, it is possible to build up an excess number of vacancies in the metal. These precipitate in areas of disregistry, principally at grairi boundaries and impurities, to form voids. These ~voids tend to agglomerate and grow in size with increasing time and/or temperature. Examinations have demonstrated that the subsurface voids are not interconnected with each other or with the surface. Voids of this same type have been developed in Inconel 600 by high-temperature oxidation tests and high-temperature vacuum tests in which chromium is selectively removed.? Voids similar to lthese have also been developed in copper-brass diffusion couples and by the dezincification of brass.1® All of these phenomena arise from the so-called Kirkendall effect, whereby solute atoms of a given type diffuse out at a faster rate than other atoms comprising the crystal lattice can diffuse in to fill vacancies which result from the outward diffusion. | ' - | The time dependénce of void formation in Inconel observed both in thermal- and forced-convection systems indicates that attack is initially -quite rapid, but then decreases until a straight-line relationship exists between depth of void formation and time.! This effect, which is illustrated in Fig. 36 for the salt mixture NaF—46 mole % ZrF;—4 mole % UF,, can be explained in terms of the corrosion reactions discussed above. The initial rapid attack shown for both types of loops stems from the reaction of chromium with impurities in the melt (reactions 1 and 2) and with the UF, constituent of the salt (reaction 3) to establish a quasi-equilibrivm amount of CsF, in the salt. At this point, attack proceeds linearly with time and occurs by a mass transfer mechanism discussed earlier. During this latter stage of attack the chromium content of the salt remains at essentially a constant value. This can be seen by referring to Table S, where chromium concentrations are shown for fluoride salts at a series of operating times. Note that between the 50 hr and 1000 hr operating times, no significant increase in CrF,; content has occurred. Comparison of Mass Transfer in Loop 1255 and Another Hastelloy N Loop - Figure 37 shows quantitative mass transfer data during 9000 hr of ‘operation for a stahdard Hastelloy " N-LiF-34.0 mole % BeF,—0.5 mole % UF, thermal convection loop system (NCL-16) which operated at a2 maximum temperature of 700°C and a minimum temperature of 540°C. This loop operated an additional 9. A. DeS. Brasunas, “Sub-surface Porosity Developed in Sound Metals during High-temperature Corrosion,” Metals Progr. 62(6), 88 (1952). 10. R. W. Balluffi and B. H. Alexander, “Development of Porosity by Unequal Diffusion in Substitutional Solutions,” SEP-83, Sylvania Electric Products (February 1952). 30 35 ORNL-LR-DWG 21487 30 és ' | ' ' | /'/ 1 / | - 20 5 2 G.‘\ow\/‘po?/ THERMAL CONVECTION LOOPS £ e BATCHNO. SYMBOL o ’ o g | e 28 o < 45 | 5 0 T ‘ | - _ 35 ¢ a * W o . / "3 A o e / » Y 160 ¢ - - . 73 z )‘B/ o _— 203 " X L - ® /0- ) / 232 ) | 188 5 ‘E’ v : PUMP LOOPS R SERIES SYMBOL 4 v 5 v 0 1000 2000 2000 4000 5000 TIME (hrs) Fig. 36. Variation in depth of fluoride corrosion in Inconel thermal convection and forced-circulation loops as a function of operating time. From W. D. Manly, J. H. Coobs, J. H. DeVan, D. A. Douglas, H. Inouye, P. Patriarca, T.K. Roche, and J. L. Scott, “Metallurgical Problems in Molten Fluoride Systems,” Progress in Nuclear Energy, Series IV, vol. 2, pp. 164—79, 1960. i ORNIL-DWG 68—11779RA | , 55 558°C COL SPECIM £ 676°C 704°C °C H0|TTEST SPECIMENS WEIGHT CHANGE (mg/cm?) Q o o Ti-MODIFIED HASTELLOY N s STANDARD HASTELLOY N 5000 6000 7000 8000 2000 o 1000 2000 3000 4000 ‘ TIME (hr) Fig. 37. Weight change vs time for standard and titanium-modified specimens in loop NCL-16, exposed to fuel salt (LiF—34.0 mole % BeF,~0.5 mole % UF,) at various temperatures. . : 'X e oy 31 Table 5. Analyses of fluoride mixtures before and after circulation in Inconel forced-clrculatnon loops at 815°C : Table 6. Chromium concentration in salt Operating | Impurities (ppm) of Hastelloy N loop NCL-16 as 2 time When sampled ———— . s (hn) Ni Cr Fe function of time 10 During filling 15 20 45 Time (hr) Cr concentration in salt (ppm) ' After termination 25 635 30 3 25 S0 During filling 50 65 30 600 15 _ After termination 10 800 25 1,631 125 100 During filling 15 35 20 2,979 162 After termination : 8 725 50 4,970 205 1000 During filling 25 60 60 6,637 242 After termination 10 765 45 - 9,520 279 : 11,381 300 Source: W. D. Manly, J. H. Coobs, J. H. DeVan, D. A. ' 16,546 403 ' Douglas, H. Inouye, P. Patriarca, T. K. Roche, and J. L. 20,458 421 Scott, “Metallurgical Problems in Molten Fluoride Systems,” : 26,646 556 Progress in Nuclear Energy, Series IV, vol. 2, pp. 164-79, 30,508 571 1960. 21,000 hr, and the only change in salt chemistry was a chromium increase to 570 ppm. Table 6 gives the chromium concentration in the salt for various times. These data indicate a_ gradual decrease in mass transfer with time. No voids were seen in any Hastelloy N specimens from NCL-16. In comparison to this loop there appears to be more mass transfer in loop 1255. The increase in chromium concentration of the salt is about the same in both cases if you assume a linear increase with time. However; this .is generally not the case, and we would expect very little additional increase of chromium in the salt of NCL-16. Thus, the chromium concentration in the salt of loop 1255 would probably have been much higher than that of NCL-16 after 30,000 hr. Also, as mentioned above, no void formation was seen in any specimens from NCL-16. Com'parison of Mass Transfer in Loop 1255and a Type 304L Stainless Steel Loop A thermal convection loop of type 304L stainless steel has contamed salt from the same batch as loop 1255 for over eight years at a maximum temperature of 688°C and a minimum temperature of 588°C. A plot of the weight change of specimens in the loop as a function of time and temperature is given in Fig. 38. These specimens were placed in the loop after the loop had operated three years. Figure 39 shows the voids formed in the specimen exposed at 688°C for 5700 hr. Microprobe analyses of this specunen disclosed an appreciable chxormum gradient for 1 2 mils. ' ' ‘Comparisons between loop 1255 and the type 304L stamless steel 100p show that the void formation in the stainless steel after 5700 hr equaled that of the Hastelloy N after 9.2 years Thus, exposed to identical fluoride salts under similar conditions, it appears that Hastelloy N is more res1stant to mass transfer. ‘Mass Transfer Calculations The schematic temperature profile of loop 1255 as a function of loop position is shown in Fig. 40. The points are actual measured temperatures at certain positions. On the right-hand ordinate scale the probable regions of weight loss and weight gain based on mass transfer theory and quantitative data from loops such WEIGHT LOSS (mg/cm? ) - o o 8 10 3 €0 ORNL-DWG 68-608TBR2 LOOP 1258 ® EQUIVALENT TO 1 mil/year 0 EQUIVALENT TO 1.5 mil/year : WEIGHT CONTINUED e 2 4 6 8 10 12 14 16 8 20 22 24 26 28 30 32 SPECIMEN TIME IN SYSTEM (1000 hr) Fig. 38. Weight loss of type 304L stainless steel specimens as a function of operation time at various temperatures in LiF~23 mole % BeF,—5 mole % Z1F 4—1 mole % ThF4—1 mole % UF, salt. ' -y 33 | 0.035 INCHES N 100% - o Fig. 39. Microstructure of type 304L stainless steel specimen in loop 1258, exposed to fuel salt for 5700 hr at 688°C. 700 650 r(°C) €00 550 ORNL-DWG 72— 1122 | | T /e 25 S € —y -] 9 o . 7/ °©2 o 2§ |4 E L I . ‘ o I ] I 20 40 60 80 DISTANCE AROUND LOOP (in.) Fig. 40. Temperature profile and mass transfer eche_matic.__ as those discussed previously are noted Posmons (temperatures) above the zero line will probably lose material, while those below will gain material. : ' . A simple calculation based on the increase of 1700 ppm chromium in the salt durmg operation gives us some insight on the amount of material removed. The amount of material that would have to be removed to equal the number of grams of chromium found in the salt was 2.2 mils of Hastelloy N over one-half the length of the loop. It is assumed that half the loop gained material while half lost material. Thus, the average depth of complete chromium depletion in the hot leg would be 2.2 mils. Based on past experience 34 \. and results schematically represented in Fig. 40, this means we could expect a maxunum depletion of 4.4 mils at the hottest position at the top of the hot leg, 700°C, decreasmg to zero at 630°C. We have assumed that all positions at temperatures below 630°C gain weight. Micrographs of the inside surfaces of insert specimens and loop tubing that were exposed to the salt disclose regions of both material loss and gain. In most cases the micrographs agree with the calculations and with Fig. 40. Fewer voids were seen in the regulaf Hastelloy N than the Hastelloy N modified with 2% Nb; however, the temperatures were different, so no definite statement can be made comparing the compatibility of these two alloys. The concentration of chromium corrosion product in the salt was never large enough to cause precipitation in the cold leg, so no plugging trends were ever noted. - Failure Analysis In Fig: 8, the shim stock (foil) that covered the thermocouple can still be seen. This shim stock was located between the ceramic bushings that separated the heaters in the center of the hot leg. The failure was adjacent.to the shim stock and was directly under a ceramic bushing. The failure in loop 1255 looked very much like the loop failure referenced in an earlier portion of this report. Thus, on the basis of the earlier work and our findings, we - attribute the failure to a reaction between the bushing and the modified Hastelloy N—2% Nb tubing. It would appear, although it cannot be proved that the bushmg in question was perhaps unfired or at least contained impurities responsible for the failure. With time and temperature, the moisture and other impurities, normally removed before installation, reacted with the outside surface, moving inward until the tubing was finally penetrated. At this time the molten salt leaked outward and caused the noticeable failure of the heaters. It is worth while to mention that MoO; melts at 795°C and the eutectic between MoO, and M003 is at 778°C. Thus one might also consider the possibility of a low-melting combination between MoO; and an oxide associated with the bushing that could lead to a failure of the type mentioned in this report. | Air Oxidation Our observations of the air oxidation of the Hastelloy N during the 9.2 years of operation disclosed about a 2-mil-thick layer-of mixed oxide as the worst condition. As the temperature decreased (Figs. 23—28), the oxide layer thickness decreased, along with a decrease in intergranular penetration and chromium depletion. As part of the molten salt corrosion program, the outside of Hastelloy N tubing containing fluoride salt is routinely examined to determine its behavior under air oxidation conditions.!! Table 7 gives some typical results. The only evidence of intergranular penetration was seen in the material exposed for 11,300 hr, and in this case the penetration was quite small. Thus, we conclude that Hastelloy N in service at temperatures below 700°C has shown good resistance to air oxidation, with penetration not exceeding 2 mils/year. Weld Corrosion Resistance Part of the purpose of this experiment was to determine the corrosion resistance of various types of weld junctions: 1. Hastelloy N welded with Hastelloy N weld rod, 11. J. W. Koger (unpublished results). o - 35 Table 7. Thickness of oxide layer formed on the outside of Hastelloy N tubing exposed to air as a function of time and temperature Inside of tubing was exposed to molten fluoride salt Temperature Time Thickness of oxide layer 0 (hr) (mils) ' 550 70 | 0 550 1,400 <0.05 550 2,700 -0l 550 4,800 0.1 550 11,300 ©0.25 610 4,700 0.25 700 4,800 0.2 Source: J. W. Koger, unpublished results, 2. Hastelloy N welded with Hastelloy N—2% Nb weld rod, 3. Hastelloy N—2% Nb welded with Hastelloy N—2% Nb weld rod. We found no difference in the corrosion (either in air or salt) of the weld and the adjacent alloy. The corrosion that occurred at the base of the Hastelloy N—2% Nb tubing weld (Fig. 32) was interesting as it showed much greater corrosion at the tip; the voids extended much deeper into the alloy at this point. The weld corrosion resistance was as good as that of the alloys. CONCLUSIONS 1. We saw, from microscopic examination of the loop tubing, that mass transfer of material (material removal and material deposition) did occur during the 9.2-year exposure of the Hastelloy N alloys to the LiF—23 mole % BeF; —5 mole % Z1F4 —1 mole % UF,—1 mole % ThF salt. | 2. The attack, which occurred in the hot section, was manifested in the formation of voids. The “maximum depth of the void zone was 4 mils. Deposition was noted on the colder portions. 3. On the basis of salt analysis and microprobe analysis of the tubing, the mass transfer appeared to be selective with respect to chromium, which is what would be predicted. 4, The actual vond formation and chromium depletion agree favorably with that predicted from calculations. 5. No mass transfer difference could be seen between the standard Hastelloy N and the modified Hastelloy N—2% Nb alloy. - ' - 6. No increase or decrease in mass transfer could be seen in the welded areas. | 7. A two-layer oxide of 2 mils thickness was the maximum formed in 9.2 years exposure to air. 8. The failure of the loop was tentatively attributed to a reaction between the impurities in a ceramic bushing and the modified Hastelloy N—2% Nb tubing, | | 9. In comparison with type 304L stainless steel exposed to salt from the same batch and under similar * conditions, Hastelloy N is much more resistant to mass transfer. 10. Hastelloy N is suitable for long-term use as a container material for a molten salt of the type used in this test and has acceptable air oxidation resistance at the temperatures used. = oo e e by i L e - L e e o/ | ‘ ORNL-TM-4189 INTERNAL DISTRIBUTION - {79 copies) ; _ (3) Central Research Library (5) J. w. Koger ORNL — Y-12 Technical Library- : E. J. Lawrence Document Reference Section A. L. Lotts (10) Laboratory Records Department T. S. Lundy Laboratory Records, ORNL RC R.N. Lyon ORNL Patent Office - H. G. MacPherson G. M. Adamson, Jr. R. E. MacPherson C. F. Baes W. R. Martin C. E. Bamberger R. W. McClung S. E. Beall H. E. McCoy E. G. Bohlmann C. J. McHargue R. B. Briggs H. A. McLain S. Cantor B. McNabb E. L. Compere L. E. McNeese W. H. Cook A. S. Meyer F. L. Culler R. B. Parker J. E. Cunningham P. Patriarca J.M. Dale A M. Perry J. H. DeVan M. W. Rosenthal J. R. DiStefano H. C. Savage J. R. Engel J. L. Scott D. E. Ferguson J. H. Shaffer J. H. Frye, Jr. G. M. Slaughter L. O. Gilpatrick G. P. Smith W. R. Grimes R. A. Strehlow A. G. Grindell R. E. Thoma W. O. Harms D. B. Trauger - P. N. Haubenreich A. M. Weinberg (3) M. R.Hill S J. R. Weir - W. R. Huntley ' , J. C. White H. Inouye | L. V. Wilson P.R. Kasten EX TERNAL DISTRIBUTION ' (24 copies) BABCOCK& WILCOX COMPANY P. O. Box 1260 Lynchburg, VA 24505 Mong BLACK AND VEATCH, P. O. Box 8405, Kansas Cxty, MO 64114 & C.B. Deenng ‘ BRYON JACKSON PUMP, P. O. Box 2017, Los Angeles, CA 90054 . G.C.Clasby _ u CABOT CORPORATION, STELLITE DIVISION, 1020 Park Ave., Kokomo, IN 46901 T. K. Roche 37 38 CONTINENTAL OIL COMPANY, Ponca City, OK 74601 J. A. Acciami EBASCO SERVICES, INC., 2 Rector Street, New York, NY 10006 o D. R. deBoisblanc ' T. A. Flynn THE INTERNATIONAL NICKEL COMPANY, Huntington, WV 25720 : J. M. Martin UNION CARBIDE CORPORATION, CARBON PRODUCTS DIVISION, 12900 Snow Road, Parma, OH 44130 R. M. Bushong USAEC, DIVISION OF REACTOR DEVELOPMENT AND TECHNOLOGY, Washington, DC 20545 David Elias J.E.Fox Norton Haberman C. E. Johnson ' T. C. Reuther S. Rosen Milton Shaw J. M. Simmons USAEC DIVISION OF REGULATIONS Washington, DC 20545 A. Giambusso USAEC, RDT SITE REPRESENTATIVES, Oak Ridge National Laboratory, P. O. Box X, Oak Ridge, TN 37830 D. F. Cope Kermit Laughon C. L. Matthews USAEC, OAK RIDGE OPERATIONS, P. O. Box E, Oak Ridge, TN 37830 €« A Research and Technical Support Division | v USAEC, TECHNICAL INFORMATION CENTER, P. O. Box 62, Oak Ridge, TN 37830 ' o (2) - L3 » W Uu.S. GOVERNMENT PRINTING OFFICE: 1973-747-302/33