- e OAK RIDGE NATIONAL LABORATORY operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION ORNL- TM- 328 R MASTER CORROSION BEHAVIOR OF REACTOR MATERIALS IN FLUORIDE SALT MIXTURES J. H. DeVan R. B. Evans, Il NOTICE This document contains information of a preliminary nature and was prepared primarily for internal use at the Oak Ridge National Laboratory. It is subject to revision or correction and therefore does not represent a final report, The information is not to be abstracted, reprinted or otherwise given public dis- semination without the approval of the ORNL patent branch, Legal and Infor- mation Control Department, LEGAL NOTICE This report was prepared os an account of Government sponsored work, Neither the Unitad Stctes, nor the Commission, ner any person acting on behalf of the Commissicn: A. Mckes any warronty or representation, expressed or implied, with respect to the accuracy, completeness, or usefulness of the information contained in this report, or that the use of any information, apparotus, methed, or process disclosed in this rapert may not infringe privately owned rights; or B Assumes any liabilities with respect to the use of, or for damages resulting from the use of any information, apparatus, method, or process disclosed in this report. As used in the above, '""person acting on behalf of the Commission'' includes dany employee or contractor of the Commission, or employee of such contracter, to the extent that such employee or contractor of the Commissien, ¢r employse of such contracter prepares, disseminates, or provides access to, any information pursuant to his employmant or contract with the Commission, or his employment with such contractor, ¥ ORNL-TM-328 Copy Contract No. W-7405-eng-26 METALS AND CERAMICS DIVISION CORROSION BEHAVIOR OF REACTOR MATERIALS IN FLUORIDE SALT MIXTURES J. H, DeVan Metals and Ceramics Division and R. B. Evans III Reactor Chemistry Division DATE ISSUED SEP 191962 OAK RIDGE NATIONAL IABORATORY Oak Ridge, Tennessee operated by UNION CARBIDE CORPORATION for the U. 5. ATOMIC ENERGY COMMISSION ABSTRACT - Molten fluoride salts, because of their radiation stability and ability to contain both thorium and uranium, offer important advantages as high-temperature fuel solutions for nuclear reactors and as media suitable for nuclear fuel processing. Both applications have stimulated experimental and theoretical studies of the corrosion processes by which molten-salt mixtures attack potential reactor materials. The subject report discusses (1) corrosion experiments dealing with fluoride salts which have been conducted in suppors of the Molten-Salt Reactor Experiment at the Oak Ridge National Iaboratory (ORNL) and (2) analytical methods employed to interpret corrosion and mass-transfer behavior in this reactor system. The products of corrosion of metals by fluoride melts are soluble in the molten salt; accordingly passivation is precluded and corrosion depends directly on the thermcdynamic driving force of the corrosion reactions. Compatibility of the container metal and molten salt, there- fore, demands the selection of salt constituents which are not appreciably reduced by useful structural alloys and the development of container materials whose components are in near thermodynamic equilibrium with the salt medium. Utilizing information gained in corrosion testing of commercial alloys and in fundamental interpretations of the corrosion process, an alloy development program was conducted at ORNL to provide a high- h temperature container material that combined corrosion resistance with : useful mechanical properties. The program culminated in the selection of a high-strength nickel-base alloy containing 17% Mo, 7% Cr, and 5% Fe. The results of several long-term corrosion loops and in-pile capsule tests ccmpleted with this alloy are reviewed to demonstrate the excellent corrosion resistance of this alloy composition to fluoride salt mixtures at high temperatures. Methods based on thermodynamic properties of the alloy container and fused salt are presented for predicting corrosion rates in these systems. The results of radiotracer studies conducted to demonstrate the proposed corrosion model also are discussed. CORROSION BEHAVIOR OF REACTOR MATERTALS IN FLUORTDE SALT MIXTURES J. H. DeVan, Speaker, and R. B. Evans TIII TNTRODUCTTON Molten fluoride salts exhibit exceptional irradiation stability and can be fused with fluorides of both thorium and uranium. Both properties have led to their utilization as fuel-bearing heat-transfer fluids [1] and as solvating agents for fuel reprocessing [2]. Intrinsic in these applications, however, is the need for experimental and theoretical studies of the corrosion processes by which molten salt mixtures attack potential reactor materials, Unlike the more conventional oxidizing media, the products of oxidation of metals by fluoride melts tend to be completely soluble in the corroding media [3]; hence passivation is precluded and corrosion depends directly on the thermodynamic driving force of the corrosion reactions, Design of a chemically stable system utilizing molten fluoride salts, therefore, demands the selection of salt constituents that are not appreciably reduced by available structural metals and the development of containers whose components are in near thermodynamic equilibrium with the salt medium. Following the initiation of design studies of molten fluoride fuel reactors, a corrosion program was begun at the Oak Ridge National Laboratory to investigate the compatibility of experimental fluoride salt mixtures [4,5] with commercially avallable high-temperature alloys [4,5]. As a result of these studies, the development of a preliminary reactor -2 - experiment was undertaken using a nickel-base alloy containing 15 Cr, 7 Fe,* and a fuel salt of the system NaF-ZrF,—UF,. This reactor experi- ment, although of intentionally short duration, successfully demonstrated the feasibility of the fluoride fuel concept [6,7]. The corrosive attack incurred by the Ni-Cr-Fe alloy was found to be selective toward chromium and was initiated through chromium oxidation at the metal surface by UF, and traces of impurities such as HF, NiF,, and FeF, [3]. The overall rate of attack was governed primarily by the diffusion rate of chromium within the alloy. Although suitable at low temperatures, corrosion rates of the alloy above 700°C were excegsive for leong-term use witnh most fluoride fuel systems. Utilizing information gained in corrosion testing of commercial alloys and in fTundamental Interpretations of the corrosion process, an alloy develcpment program was carried out to provide an advanced container material that combined corrcsion resistance with useful mechanical properties. The alloy system used as the basis for this program was composed of nickel with a primary strengthening addition of 15 to 20% Mo. Experimental evaluations of the effectis of other solid-sclution alloying additions to tnis basic composition culminated in the selection of a high=strength nickel-base alloy containing 17 Mo, 7 Cr, and 5 Fe. The purposc oi tThe present report 1s to summarize the corrosion properties of alloys based on tihe nickel-molybdenum system and to then discuss an analytical apvroach for precicting corrosion rates in these systems based on thermodynanmic properties of the alloy and fluoride salt *Compositions refer to percent by weight, except where otherwise noted, -3 o mixture. The report is divided into three major sections: (1) a presentation of experimental results showing the effects of alloying additions of Cr, Fe, Nb, V, W, Al, and Ti on the corrosion properties of nickel-molybdenum alloys; (2) a presentation of the experimentally determined corrosion properties of the 17 Mo—7 Cr—5 Fe—bal Ni composition (designated INOR-8); and (3) a discussion of the analytical model that has been employed to interpret the corrosion and mass transfer properties of these alloys. FFFECTS OF ALLOYING ADDITIONS ON THE CORROSION RESISTANCE OF NICKEL ALLOYS IN FLUCRIDE MIXTURES Experimental Several laboratory heats of experimental nickel-molybdenum alloy compositions were prepared by vacuum- and air-induction melting to afford a programmatic study of the effects of soclid-solution alloying additions on corrosion behavior in fluoride salts. The compositions of the alloys that were evaluated are shown in Table I, The cast alloys were formed into tubing and were subsequently fabricated into a thermal-convection loop for corrosion testing. Loops similar to the ones employed have been described elsewhere {3]. Loops were exposed to the salt mixture NaF-LiF-KF—UF, (11.2-45.3-41.0-2.5 mole %) for periods of 500 and 1000 hr and were operated at a hot-zone temperature of 815°C and a cold-zone temperature of 650°C. The corrosion susceptibility of alloying additions was determined from analyses of the concentrations of corrosion products in after-test salt samples and from metallographic examinations of the loop walls. Toble I. Compositions of Experimental Nickel-Mclybdenum Alloys Used for Corrosion Studies Composition (wt %) Heat No,* Ni Mo Cr Fe Ti Al Nb W v Series I: OR 30-1 8C.12 16.93 2.83 OR 30-2 78.55 16.65 4.62 OR 30-4 73.65 16.37 9.21 OR 30-6 78.50 15.11 6.40 OR 37A-1 77.0 20.39 2.62 OR 43A-3 73.30 20.34 6.34 Series [: OR 30-7 82.10 15.93 1.88 OR 30-8 80.30 17.80 1.89 OR 30-9 81.10 16.8 2.09 OR 30-10 81.10 16.60 2.23 OR 30-11% 79.80 16.53 3.68 OR 30-12 80.00 16.80 3.22 CR 30-19 79.00 16.90 4.10 OR 30-20 79.20 16.60 4.18 OR 30-21 78.90 16.40 4.71 ST 23012 82.00 17.42 0.53 OR 1491 86.58 11.23 2.19 Series HI: OR 30-13 79.93 17.56 1.56 0.95 OR 30-14 79.53 16.50 1.52 2.45 OR 30-16 77.74 16.00 3.65 1.49 1.12 OR 30-22 77.65 15.90 5.69 1.16 0.60 OR 30-33 74,07 15.15 5.01 5.07 0.70 B 2897 76.13 20.50 1.25 1.32 B 2898 76.30 20.50 2.44 1.31 B 3276 69.19 2110 7.58 2.16 B 3277 66.95 21.60 7.82 1.31 2.32 ST 23011 71.50 15.06 3.84 0.53 4.17 4,90 ST 23013 74.42 15.20 0.58 4,57 5.23 ST 23014 80.86 16.70 2.10 0.57 **QR’’ denotes heats furnished by the ORNL Metals and Ceramics Division. “ST"" denotes heats furnished by Superior Tube Company. *'B’' denotes heats furnished by Battelle Memorial Institute. -5 - Salt mixtures utilized for this investigation were prepared from reagent-grade materials and were purified to obtain a total impurity content below 500 ppm. Results Chemical analyses of salts tested with alloys containing a single alloying addition, i.e., with ternary alloys, are plotted as a Tunction of alloy content in Fig. 1. It is seen that the corrosion susceptibility of alloying elements, based on the concentration of the elements in the after-test salt, tended to increase in the order: e, Nb, V, Cr, W, Ti, and Al., Table IT indicates a similar trend in the corrosion-product concentrations of salts tested with alloys containing multiple additions. However, in comparing the values in Table II with Fig. 1, it is evident that the corrosion-product concentrations assoclated with either iron, niobium, or tungsten alloying additions were substantially lower when these elements were present in multicomponent alloys than in simple ternary alloys. In contrast, corrosion-product concentrations assoclated with chromium, aluminum, or titanium were effectively unchanged by the presence of other alloying constituents. Because of beneficial effects on oxidation resistance and mechanical properties imparted by chromium additions, a relatively large number of alloy compositions containing this element were evaluated. The extent of reaction between chromium and fluoride constituents, as indicated by the chromium ion concentration of the salt, varied markedly with the amount of chromium in the alloy (or the element). This variation is illustrated graphically in Fig. 2, where the data are compared with data for a 15 Cr—7 Fe—bal Ni alloy [4] and for pure chromium [8]. 1In all UNCL ASSIFIED ORNL — LR— DWG 46943 500 / 400 l /, y 300 / AI// 200 I - / ) o 5 [+ H] ° E — 100 J < 90 < 80 | o i w 70 — S / Ho60 i/ 50 g A | = / T o Ti/r 5 p / x 40 , ’ e 7/ 7 c Y o * 4 S Wy l /7 & 7 / / 7 / A4k 20 , -//V/. 7/ Nb/ / / }W@ F,/’ ’ 10 1 2 3 4 5 6 7 8 9 | 2 ALLLOY CONTENT (at. %) Fig. 1. Corrosion-Product Concentrations of Salts Tested with Experimental Nickel-Molybdenum Alloys Containing Single Alloying Ad- Salt mixture: NaF-TiFP-KF-UF, (11.2-45.341.0-2.5 mole %). ditionse. Table Il. Corrosion-Product Concentrations of Salts Tested with Experimental Nickel-Molybdenum Alloys Containing Multiple Alloy Additions Salt Mixture: NaF-LiF-K F—UF4 {11.2—45.3-41.0~2.5 mole %) Concentration of Element in Salt (Mole %) Heat No.* Alloy Composition {atomic %) Cr Al Ti Nb Fe W Test Duration: 500 hr OR 30-13 2.18 Al, 2.02 Ti, 11.35 Mo, bal Ni 0.040 CR 30.-14 5.51 Al, 1.92 Ti, 10.42 mo, bal Ni 0.43 0.045 OR 30-16 2.54 Al, 1.90 Ti, 4.30 Cr, 0.023 0.33 0.038 10.20 Mo, bal Ni OR 30-22 2.61 Al, 0.39 Nb, 6.64 Cr, 0.055 0.38 0.0018 10.05 Mo, bal Ni B 2898 3.24 Ti, 0.90 Nb, 13.20 Mo, bal Ni 0.030 <0.0005 B 3276 1.48 Nb, 9.32 Cr, 14.00 Mo, bal Ni 0.061 0.0007 B 3277 3.06 Al, 1.57 Nb, 9.47 Cr, 0.067 0.50 <0.0005 14.20 Mo, bal Ni ST 23011 1.27 Al, 291 Nb, 4.79 Cr, 1.72 W, 0.049 0.007 10.20 Mo, bal Ni ST 23013 1.40 Al, 3.23 Nb, 1.85 W, 10,36 Mo, <0.001 <0.0005 0.002 bal Ni ST 23014 1.30 Al, 2.71 Ti, 10.76 Mo, bal Ni 0.060 0.019 Test Duration: 1000 hr OR 30-14 5.51 Al, 1.92 Ti, 10.42 Mo, bal Ni 0.25 0.043 OR 30-22 261 Al, 0.39 Nb, 6.64 Cr, 0.041 0.25 0.0010 10.05 Mo, bal Ni OR 30-33 1.59 Al, 5.56 Fe, 5.90 Cr, 0.069 0.082 0.0051 Q.66 Mo, bal Ni B 2897 1.68 Ti, 0.92 Nb, 13.77 Mo, bal Ni 0.038 0.0012 B 3277 3.06 Al, 1.57 Nb, 9.47 Cr, 0.073 0.47 0.0020 14.20 Mo, bal Ni *"OR'" denotes heats furnished by the ORNL Metals and Ceramics Division “*ST'' denotes heats furnished by Superior Tube Company. **B'' denotes heats furnished by Battelle Memorial Institute. UNCLASSIFIED ORNL-LR-DWG 68376 3000 T T T T TTPURE Cr=800°¢ | 1~ 1 17777 - 2000 — 200 & 1600 F = 130 & = 1200 PURE Cr—600°C < Y cue S chie s - . —— . P S G S G — e caver opre o jEn wEn GEn| . NS GG AL Gl mwy [ S SSE-w— 100 a 1000 M > g0 @ & 800 - fi a‘é——i} Ni-Cr—Fe ALLOY ] < & — Je0 & © 600 § o © @ £ B — O O £ _ 40 S 400 s ¥ N S o —l ] RANGE OF 3 SALT SAMPLES — T 1 O SINGLE ALLOY ADDITION =500 hr TESTS - O ® SINGLE ALLOY ADDITION —1000 hr TESTS _5 § O MULTIPLE ALLOY ADDITIONS =500 hr TESTS _{ 29 200 H& ® MULTIPLE ALLOY ADDITIONS —1000 hr TESTS — 0.03 004 (006 008 040 0.2 0.4 06 0.8 10 ATOM FRACTION OF CHROMIUM IN ALLOY Fig. 2. Chromium Concentration of Fluoride Salt Circulated in Thermal-Convection Loops as a Function of Chromium Content of the Loop. Salt mixture: NaP-LiF—KF-UF, (11.2—45.3=41.0~2.5 mole %). ILoop tem- perature: hot leg, 815°C; cold leg, 650°C. cases the corrosion-product concentrations in the experimental alloy loops (which contained up to 11.0 at. % Cr) were less than corresponding concentrations in nickel-chromium-iron loops under similar temperature conditions or in pure chromium capsules exposed isothermally at 600 and 800°C., The latter observation indicates that the observed chromium ion concentrations were below those required for the formation ol pure chromium crystals in the cold-lecg region of the loops (650°C). Metallographic examinations of alloys investigated under this program showed little evidence of corrosion except for systems containing combined additions of aluminum and titanium or aluminum and chromium, Significant alloy depletion and attendant subsurface void formatlion in the latter alloys occurred to depths of from 0,003 tc 0,004 in., In all other systems, attack was manifested by shallow surface pits less than 0,00l in. in depth., Figure 3 illustrates the typlcal appearance of attack in alloy systems containing chromium at levels of 3,2 and 11.0 at. %, respectively. Although the depth of pitting was comparable in both alloys, the intensity of pitting increased slightly with chromium concentration. In the case of the majority of alloys tested, the rate of attack between O and 500 hr was substantially greater than the rate occurring between 500 and 1000 hr. This finding is illustrated in Fig. 4, which compares the surface appearance of a ternmary alloy containing 5.55 at. % Cr after 500- and 1000-hr exposures. This result 1s in agreement with the observed corrosion-product concentrations, which increased only slightly between 500 and 1000 hr, and suggests that nearly steady-state conditions were established within the first 500 hr of test opecration, « 10 - | Unclassified T-12037 1 ¥ INCHES 1 A g ; ] » 280X Composition: 3.2 Cr—13.5 Mo-bal Ni (at. %) A Unclassified - T=11298 | " INCHES ‘l ] E Composition: 11.0 Cr—10.6 Mo-bal Ni (at. %) Fig. 3. Hot-Leg Sections of Nickel—Molybdenum-Chromilfin Thermal- Convection Loops Following 500-hr Exposure to Fluoride Fuel. ©Salt mixture: NaF-TiF-KF-UF, (11.2-45.3-41.0-2.5 mole %). 250X. Reduced 5. Unelassified T-11327 s oo 7 /\#’\ \\\ e e ~ 3 O 250X E After 5 hr Unclassified - £-12031 - After 1000 hr . Pig. 4. Appearance of Hot-Leg Surface of a Ternary Nickel-Molyb- denum Alloy Containing 5.55 at. % Cr Following Exposure to Fluoride Fuel. Heat No. OR30-2. Salt mixture: NaP-TiF-KF-UF; (11.2-45.3-41.0- 2.5 mole %). 250X. Reduced 4%. ' - 12 - Discussion In view of the uniform compositions of salt mixtures employed for these alloy evaluations, it follows that the mixtures afforded comparable oxidation potentials at the start of each test. Accordingly, if passi- vation did not occur, one can readily show that the extent of reaction resulting from equilibration of the salt mixture with given alloying elements should be governed simply by the activity of the element in the metal and by the stability (or standard free energy of formation) of the fluoride compound involving the element. Consider, for example, the component* chromium and the oxidation reaction Cr + U, = CrFy + 2UF43 (1) for which 2 Yorr, ¥ UF, Q_I_'_' UF, At the very dilute concentrations of CrF, and UF3, which are realized under the test conditions, the activities of these products may be approximated by thelr mole fractions in accordance with Henry's law. Thus, for a salt system of fixed UF, concentrations, assuming the refer- ence states for salt components to be the infinitely dilute solution, 2 - NCI‘Fg'N UF3 = — 3 Ky S (3) Cre™ UF, and 2 _ %’ NCrF2,N UF, " Koo, - (4) *Solid-solution alloying elements are underlined. - 13 - If reaction products are initially absent, a mass balance exists between h & -1 . the products formed such that NCTF2 5 NfiF3 and Eq. (4) reduces to - 1/3 Ny, = K 0191: , (5) where K y? - 1/3 [ K" - a'QfiFg J 4 Since o o o RT In K = AF° o+ 2(8F = 0F ) it follows that and O N - 6(ag—l:-) AF (b) NCng Cng) : In Table IIT are listed the standard free energies of formation, per gram-atom of fluorine, of Fluoride compounds at 8C0 and 600°C associated with ecach of the alloying elements investigated [9]. Valucs are given for the most stable compounds (i.e., those with most ncgative free energies) and are listed in order of decreasing stabilities. Tne resultant order suggests thatl corrosion-product concentrations associated with each element (at a given activity) should have increased in the following order: W, INb, Fe, Cr, V, Ti, and Al. Comparison with IFig. 1 shows that, with the exception of niobium and tungsten, the corrosion-product concentrations per atomic percent of alloy addition did increase in the exact order predicted. Only tungsten noticeably deviates from the predicted pattern, although the tests made on this addition were statistically limited. - 14 - Table III. Relative Thermodynamic Stabilities of Fluoride Compounds Formed by Elements Employed as Alloying Additions Standard Free Energy of Formation per Gram Atom of Fluorine kcal/g~-atom of F Most Stable (keal/e ) Element Fluoride Compound at 800°C at 600°C Al AlF, -87 92 Ti TiF, -85 -90 v VF -80 -84 Cr Cr¥F, -72 -'77 Fe FeF 5 -66 -69 Ni NiF, ~59 ~63 Nb NbF 5 -58 -60 Mo MoF 5 -57 -58 W WE 5 ~46 438 - 15 - When tested in the presence of other alloylng elements, the corrosion- product concentrations of iron, niobium,or tungsten were noticeably lower than the values attained for ternary alloys. The reason for this behavior undoubtedly relates to the presence of the more reactive alloying additions in the multicomponent alloys. If one considers, for example, an alloy containing comparable additions of chromium and iron, for which the corro- sion reactions can be written il I o Cr + 2UF, = CrF, + 2UF;5: AF Fe + 2UF, ~D I Il FeF, + 2UF5: AF where a] > [v] the equilibrium UF, concentration produced for the first reaction is higher than that which would be produced by the second reaction. Accord- ingly, in the presence of chromium, the FeF, concentration at equilibrium will be reduced compared to the system containing iron only. The results of these alloy evaluations provided further evidence that corrosion in molten fluoride systems involved essentially the attainment of thermodynamic equilibrium between the fluoride melt and container metal. The results also implied that, over the alloy compositions studled, the activities of the alloying additions could be reasonably approximated by their atom fractions, i.e., that activity coefficients were nearly the same for all of the alloying additions tested. The favorable corrosion properties of the majority of alloys tested permitted wide latitude in the selection of an optimum alloy composition for fluoride fuel containment. Only titanium and aluminum were felt to afford potential corrosion problems, particularly if used as combined additions or in - 16 - combination with chromium. Because chromium had proved an extremely effective alloying agent in regard to both strength and oxidation properties, this alloying addition was utilized in the selected alloy composition., The level of chromium in this alloy was fixed at 7%, which is the mimimum amount required to impart oxidation resistance to the Wi-17% Mo system [10]. The addition of chromium as an iron-chromium alloy also imparted approximately 5% Fe to the system. While this finalized alloy composition, designated INOR-8, was not tested as part of the initial alloy study, its corrosion properties can be considered equivalent to those of the ternary chromium-containing alloys discussed above, CORROSION PROPERTIES OF INOR-8 Experimental The materialization of INOR-& as a container material for fluoride fuels led to an extensive investigation of the corrosion properties of this specific alloy composition under simulated reactor conditions. Studies were conducted in forced-convection loops of the type shown in Fig. 5. Tubular inserts contalned within the heated sections of the loops provided an analysis of weight losses occurring during the tests, and salt samplers located above the pump bowls provided a semicontinuous indication of corrosion-product concentrations in the circulating salt. All parts of the loops were constructed of commercially supplied INCR-8 material, The salt compositions initially utilized for these studies were of the type LiF-BeF UF, but, in later tests, the compositions were changed to LiF-BeF ,—ThF,UF,, Corrosion rates were determined at a series of Unclassip; ed ORNL Photg 34693 LT - - 18 - hot-leg operating temperatures ranging from 700 to 815°C. At each tempera- ture level, the mixtures experienced a temperature change of approximately 200°C between heater and cooler sections. Loop operating times were generally in the range 15,000 to 20,000 hr, although loop operation was interrupted at shorter time intervals to allow the removal of corrosion inserts. Results The weight losses of corrosion inserts contained in the hot legs of INOR-8 forced-convection loops at 700, 760, and 815°C are shown in Table IV. 1In two test loops operated at 700°C, weight losses were in the range 2 to 5 mg/cm2 and showed little change after 5000 hr of loop operation., Weight losses at 760 and 815°C were slightly higher, being in the range 8 to 10 mg/cm2, and agaln were essentially unchanged after 50C0 hr of operation., Comparisons of before- and after-test dimensions of the inserts revealed no measurable changes in wall thickness. However, if uniform attack is assumed, weight-loss measurements indicate wall reductions on the order of 2 to 12 u (Table IV). Chemical analyses of salt samples that were periodically witndrawn from the loops showed a slight upward shift in chromium concentration, while concentrations of other metallic components remained unchanged. During a typical run, as pictured in Fig. &, the chromium concentration reached an asymptotic limit after about 5000 hr of operation. This limit was between 300 and 500 ppm at 700°C and between 600 and 800 ppm at 760 and 815°C. Metallographic examinations of loops operated at 700°C for time periods up to 5000 hr showed no evidence of surface attack. When test - 19 - Table IV. Corrosion Rates of Inserts Located in the Hot Legs of INOR-8 Forced-Convection Loops as a Function of Operating Temperature Loop temperature gradient: 200°C Flow rate: approximately 2.0 gal/min Reynolds number: approximately 3000 Insert Weight Loss Equivalent Loss Loop Salt a Temperature Time per Unit Ares in Wall Thickness Number Mixture (°c)P (hr) (mg/cm?) (p) 9354-4 130 700 5,000 1.8 2.0 10, GO0 2.1 2.3 15,140 1.8 2.0 MSRP-14 Bu-l4 700 2,200 0.7 0.8 8,460 3.8 4,3 10,570 5.1 5.8 MSRP-15 Bu-l4 760 8,770 11.2 12.7 10,880 10.0° 11.2 MSRP-16 Bu-1l4 815 5,250 9.6 10.9 7,240 9.0° 9.1 %521t Compositiocns: 130 LiF-BeF2-UF,; (62-37-1 mole %) Bu-14 LiF-BeF ~ThF,—UF, (67-18.5-14-0.5 mole %). b . Same as maximum wall temperature. C . Average of two inserts. UNCLASSIFIED ORNL-LR-DWG 58961R CCRROSION PRODUCT CONCENTRATION (ppm) T T ! ; i | | f | ! | | SALT COMPOSITION: Lif -BeFp-ThF4-UF4 (67-18.5-14-0.5 mole %) | - MAXIMUM METAL INTERFACE TEMPERATURE: 760°C e ~ LOOP AT: 200°F ‘ ! | o | | | | i | | | | ! : o Ni | F l I 1 l 2310 hr,LOOP e e . . 8770 hr,LOOP ——— 290 hr,LOOP ISOTHERMAL . 4 Cr | DOWN 195 hr O LOOP TERMINATED | ISOTHERMAL FOR 29 hr ' m Fe | REMOVE INSERT ~ AFTER 10,878 hr FOR 14 hr I | | | | OPERATION 800 I i e o ot e b Il ‘L iy | P - ,.‘,_.‘.H..‘..._.___.._____JI._.... .o + II e = e ] | 12400 hr,LOOP ; s ! | 4 A | l|s THERMAL | B | | 1l 3650n,L00P f | | 1 ISOTHERMAL , i : A 1L FORItthr S S | o 400 lug i " | | | " I | L | | | A | ‘ | | | | | ' 200 _Jf, ‘ 1 || _ | | ~ . ‘ B } A — || ] i | f I . | " i " | | | | | I | l ot s | |l 4 e | o P ? 0 1000 2000 3ooo 4000 5000 6000 7000 8000 9000 10,000 11,000 12,000 TIME (hr) Fig. 6. Concentration of Corrosion Products in Fluoride Salt Circulated in an INOR-8 Forced- Convection Loop. _OZ— - 21 - times exceeded 5000 hr, however, an extremely thin continuous surface layer was evident along exposed surfaces, as shown in Fig, 7, Similar results were observed in 760°C tests. No transition or diffusion zone was apparent between the layer and the base metal, and analyses of the layer showed it to be composed predominantly of nickel with smaller amounts of iron, chromium, and molybdenumn., At 815°C surface attack was manifested by the appearance of sub- surface voids, shown in Fig. 8, and the surface layer was not observed. Discussion The corrosion rates of INOR-8 measured in pumped loops cperating with maximum temperatures ranging from 700 toc 815°C indicate that corro- sion reactions with fluoride salts are essentially completed within the first 5000 hr of loop operation and that weight losses thereafter remain essentially constant. When Jjudged from the standpoint of total weight loss, the temperature dependence of the corrosion rates was relatively small over the ranges studied, However, the metallographic appearance of specimen surfaces was noticeably influenced by the test temperature. At the highest test temperature (815°C), exposed surfaces underwent a noticeable depletion of chromium, as indicated by the appearance of sub- surface voids, At lower temperatures, surfaces were slightly pitted and were lined with a thin surface film which became apparent only after 5000 hr of operation. The composition of this layer indicates a higher nickel-to-molybdenum ratio than originally present in the base metal and suggests that the layer may constitute an intermetallic transformation product of the nickel-molybdenum system. Unclassified [, W T-19555 }¥- S z : 006 007 . A ':* - ""' \\,\ . . el - -~ . * o - - . - : TN ! Rt . L.Q02 - ~ ‘o . - " . ;,._—\‘h"‘ ) . T- Vl:' :- * .- ST ‘ T BT Y 010 Lo s -, o . - -t -' T ; - ~ \ "‘ * . .\. ¥ .- D a Ot 1 - ot Lt “ LRt i tain . h - 7 ‘. - - t . / 2 Y o * s ’ * *_,-...l ¥ L 1 - * .. . o . . Ol4 * . - * * 018 ) - - - .‘V x . - o Toe o e . - o LU L | - ' s LT TR Fig. 7. Appearance of Metallographic Specimen from Point of Maximum Wall Temperature (700°C) of INOR-8 Forced-Convection Loop 9354-4. Operating time: 15,140 hr. Salt mixture: LiF-BeFo—~UF, (62371 mole %). 250X. Reduced 7%. o - e Dy e R LT B - . . \ : W - S XNRS Fig. 8. Appearance of metallographic Specimen from Point of Maximum Wall Temperature (815°C) of INOR-8 Forced-Convection Loop MSRP-16. Operating time: 7,240 hr. Salt mixture: ILiF-BeF,~ThF,~UF, (6718.5-14-0.5 mole %). 250X. Reduced 7%. , - 23 . ANATLYSTS OF CORROSION PROCESSES The above-mentioned tests of nickel-molybdenum alloys in fluoride mixtures have given considerable insight into the mechanisms of corrosion in these systems, In turn, the conclusions drawn have enabled the develop- ment of procedures for predicting corrosion rates on the basis of the operating conditions and chemical properties of the system. The method of approach and assumptions used for these calculations are discussed below. For purposes of illustration, the discussion deals specifically with the NaF-ZrF,—UF, salt system contained in INOR-&. The corrosion resistance of metals to fluoride fuels has been found to vary directly with the "nobility" of the metal — that is, inversely with the magnitude of free energy of formation of fluorides involving the metal. Accordingly, corrosion of multicomponent alloys tends to be manifested by the selective oxidation and removal of the least noble component., In the case of INOR-8, corrosion is selective with respect to chromium (Figs. 2 and 8). If pure salt containing UF, (and no corrosion products) is added to an INOR-8 loop operating polythermally, all points of the loop initially experience a loss of chromium in accordance with the Cr-UF, reaction,Eq. (1), 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-sensitive, 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 - 24 - and UF, reactions, the lowest temperature point of the loop eventually achieves egquilibrium with respect to the UF, reaction. At regions of higher temperature, because of the temperature depencence for this reaction, a driving force still exists for chromium to react with UF,. Thus, the corrosion-product concentratiorn will continue to increase and the tempera- ture points at equilibrium will begin to move away from the coldest tem- perature point., At this stage, chromium is returned toc the walls of the coldest point of the system. The rise in 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 lecop at equilitrium with the galt that are termed the "balance points" do not shi’lt meastrably with time. Thus, a quasi-steady~state situation is eventually acnieved whereby chromium 1s transported at very low rates and under concditions of a [ixed chromium surface concentration at any given loop positior., This Zdea is supported by the fact that concentrations of CrF,, UF,, and UFs; achieve steady-state values even though attack cslowly increases with time, During the initial or unsieacy-state period of corrosion, the mathematical interpretatior of chromium migration rates in polythermal systems 1s extremely comnlex, because of rapid compositional changes which occur continuously in both the salt mixture and at the surface of the container wall. A manifestation of this complexity is apparent in cold- leg regions, where chromium concentration gradients are reversed &as corrosicn products build up within the salt. Unfortunately, the effects of unsteady-siate operation serve also to complicate the interpretation - 25 - of steady-state operation., However, an idealized approach to the study of chromium migration under steady-state conditions i1s afforded by assuming the salt mixture to be preequilibrated so as to contaln amounts of CrF, and UF5; which establish a steady-state condition at the beginning of loop Operation.* Under this assumption the following conditions may be specified for the resulting migration of chromium: 1. The overall rate of transfer is controlled by the diffusion rate of chromium in the container material, since this diffusiocon rate can be shown to be considerably lower than the rates of reaction and cf mixing which are involved in the transfer process. 2. Chemical equilibrium with respect to chromium and the salt mixture exists at every surface point within the loop system. 3. At any time the cumulative amount of chromium removed from the hot zone equals the cumulative amount deposited in the cold zone. Accord- ingly, the concentrations of UF;, UF,, and CrF,, assuming the salt to be rapidly circulated, do not vary with loop position or time. The "equilibrium" concentrations of UF,, UF;, and CrF, can be estimated for steady~state loop operation by applying these three condi- tions to a given set of loop parameters and operating conditions., As a means for demonstrating these calculations, it is convenient to apply them to a preequilibrated polythermal INOR-8 loop in which the salt system *It is evident that preequilibration of the salt mixture would afford an expedient means for reducing corrosive attack in INOR-8 systems, since it would eliminate the initial or unsteady-state period of corrosion during which relatively high corrosion rates are sustained. However, to achieve criticality in reactor systems which are presently under study, it is planned to add UF, incrementally to the sall mixture until a desired reactivity is achieved. Under these conditions, the program required to maintain a preequilibrated salt with each UF, level becomes extremely complex. - 26 - NaF-ZrF ,—UF,; (50-46—4 mole %) is circulated between the temperature limits of 600 and 800°C. It is assumed that the balance point of the system is at 600°C. Under this condition, the amount of chromium depletion and hence the corrosion rate at 800°C are the highest attainable under steady- state conditions. In zccordance with the second condition given above, the surface concentration, <#9£>s’ at the 800°C temperature point can be expressed in terms of the equilibrium constant, Ka, determined for the UE4-Cr reaction in accordance with Eq. (3). - N\ a? O KN@/J T [ s (7) 1 800° ¢ & 800°¢ UF, but a2 Q _ UF3' CrF, KN )J : ; (8) 2 - ‘Cr o a UF, Z=/8 600°C 600°C so that [/ } - K eooec g ‘) o)) o Ve | ( — 5 goce0 a’ 800°¢C 600°C The valance points ol the systerm, wnich are at 600°C, have been defined BZNCKJ J - (FQ; in INOR-a) ' (10) — 57600°C such tnat Therefore, since the concentration of fluorides involved do not change with position or time, [ N‘_M - ((I;a)jeomc