fl‘lh.‘:’ " - AT i . - lflli'h-ln'_i,' e T i‘ < Wy g ! e | i aawl i LT -_ W ;-‘I : L PR [ Bt W aWnlwy CLASBIFICATION CHANGED To: / — AL OF:. o BY AUTHORITY Ay AL ad G Lo omgn éc 3 4yy5b 0353224 7 CORROSION BY MOLTEN FLUORIDES OAK RIDGE NATIONAL LLABORATORY OPERATED BY CARBIDE AND CARBON CHEMICALS COMPANY A DIVISION OF UNION CARBIDE AND CARBON CORPORATION (<4 POST OFFICE BOX P OAK RIDGE. TENNESSEE ;1'! ORNL-1491 This document consists of 24 pages. Copy 6 of 157 copies. Series A. Contract No, W-7405-eng-26 METALLURGY DIVISION CORROSION BY MOLTEN FLUORIDES Interim Report — September 1952 L. S. Richardson D. C. Vreeland W. D. Manly Experimental work carried out by A. deS. Brasunas L. S. Richardson R. B. Day D. C. Vreeland E- Eb Hflffm&n w- DI Ma‘fll‘y DATE ISSUED WAR 17 1988 OAK RIDGE NATIONAL LABORATORY operated by CARBIDE AND CARBON CHEMICALS COMPANY A Division of Union Carbide and Carbon Corporation Post Office Box P 0ak Ridge, Tennessee aegumimndimran ||/ 111111] 3 445k D35322Y4 7 (4, S 6-7. 8-13. 14. 15. 16. 17. 18. 19. 20. 21. 22, 23. 24. ORNL-1491 Chemistry-General INTERNAL DISTRIBUTION (A} E. Center 25. F. L$Steahly BiNylogy Library 26. A. Snell HeaYth Physics Library 27. A. jHollaender MetaYlurgy Library 28. MJFT. Kelley React® Experimental 29. Z. Morgan Engin&ring Library 30. JJ. S. Felton Central Qesearch Library 31.0/A. S. Householder Central es 3 C. 8. Harrill C. E, Lars&yg 3. C. E. VWinters W. B. Humes $§K-25) 4. D. S. Billington L. B. Emlet (%-12) 5. D. W. Cardwell A. M. Weinberg y 36. E. M. King E. H. Taylor 37. A. J. Miller E. D. Shipley 38. D. D. Cowen F. C. VonderlLage 39. P. M. Reyling R. C. Briant 40. L. S. Richardson W. J. Fretague 41. D. C. Vreeland J. A. Swartout 42, W. D. Manly S. C. Lind 43. 44- 5.1- 52. 54. 55-59. 60. 61-63. 64-65. 66-67. 68-70. 71-74. 195, 16. 77 78 This EXTERNABRDISTRIBUTION CoWpany, Watertown LaboRatory scial W&gpons Project, Sandia pecial WeRgons Project, Washington American Cyanamj Argonne Nationg Armed Forces 3§ Armed Forces Army Chemicglf Center Atomic Ene Commission, Battelle MBmorial Institute Brookhavgi National Laborato Califogia Research and Devel@pgment Company Carbid® and Carbon Chemicals C&gpany (C-31 Plant) Carbgfle and Carbon Chemicals Comigany (K-25 Plant) Caghide and Carbon Chemicals Comp@py (Y-12 Plant) alytic Construction Company icago Patent Group ief of Naval Research Dow Chemical Company, Pittsburgh Dow Chemical Company, Rocky Flats ashington RESTRICTED DATA document contains Restricted Data as defined in the AtoWlic Energy Act of 1946. Ilts transmittal or the disclosure of its conte in any manner to an unauthorized person is prohibited. iii 80-84. duPont Company 85%87. General Electric Complny (ANPP) 88-93. General Electric Cofipany, Richland 94 Han ford Operationgf Of fice 95-98. daho OperationsfOffice 99. wa State Collgge 100-103. Knolls Atomic Fower Laboratory 104-106. Los amos Scgentific Laboratory 107. Mallin@krod emical Works 108. MassachuSetfs Institute of Technology (Kaufmann) 109-111. Mound Labdfatory 112. National AdWMisory Committee for Aeronautics, Cleveland 113, NationalBur&8gu of Standards 114. Nationgl Lead Eompany of Ohio 115. NavalgMedical Rgsearch Institute 116. Naval}f Besearch Daboratory 117. NewgBrunswick Labgratory 118-119. Ne@' York Operationg Office 120-121. th American Aviation, Inc. 1225 atent Branch, Washiggton 123. JRand Corporation 124.4 Savannah River Operat®pns Office, Augusta 12 Savannah River Operatidgns Office, Wilmington 1 Sylvania Electric Produ@ts, Inc. (. Tennessee Valley Authori 28. U. S. Naval Radiological Wefense Laboratory 29. UCLA Medical Research Labokatory (Warren) 130%133. University of California Radiation Laboratory 134-135. University of Rochester 136-137. Vitro Corporation of America 138. Western Reserve University (FRiedell) 139-140. Westinghouse Electric Corporatiion [41-142. Wright Air Development Center 143-157. Technical Information Service, Ofk Ridge This documenkesssRieine " e e iaiba Atomic Energy 946 he disclosure of it in any manneéf 10 | & - CORROSION BY MOLTEN FLUORIDES L. S. Richardson D. C. Vreeland W. D. Manly SUMMARY AND CONCLUSIONS Both static and dynamic corrosion tests of metals in molten mixtures of alkali metal fluorides have been made by the Metallurgy Division in the past year. Evaluation of these tests has led to some definite conclusions as to the nature of corrosion by molten fluorides under certain test conditions, The major formof attackby fluorides on Inconel results in the formation of voids which are either scattered uniformly throughout the area near the surface or appear preferentially at grain boundaries leading from the surface inward. It is believed that these voids are caused by depletion of one constituent (usually chromium), which leaves vacancies and precipi- tation of these vacancies (cf., Appendix). A thin, adherent film of U0, forms at temperatures above 1200°C (and occasionally lower) and apparently inhibits the depletion of chromium,. The rate of attack (as measured by depth of subsurface voids) decreases with time and levels off to a very slow rate after 500 hours. Prior heat treatment and the presence of small amounts of impurities in the metal apparently do not change the rate of attack appreciably. However, in some cases in which a very large grain size was induced by heat treat- ment, void formation was found to be localized at the grain boundaries and caused deeper but less extensive damage. Cold working the metal seems to have no effect. Addition of nickel and iron fluorides to the corroding medium was found to increase the corrosion, probably because of the reduction of these fluorides by chromium that entered the system from the container walls. The ad- dition of small amounts of certain & materials to the molten fluoride mixture is, in some cases, effective in reducing corrosion. In general, metals high in the electromotive series (such as calcium, beryllium, and the alkali metals) reduce or entirely stop void formation, whereas metals that are relatively more noble (such as iron and silver) have no effect. The more active metals were so effective in minimizing corrosion that they are being used in the larger-scale testing program at the Y-12 site. Additions of zirconium, sodium, and titanium have been found to minimize corrosion in the thermal convection loops used in these tests, It is thought that the active metals minimize corrosion by reducing the availability of fluorine and/or oxygen to the structural metals. Mass transfer effects are not so serious a problem in fluorides as in liquid metals or molten hydroxides. Metallic layers or crystals are noted with pure metals, and nonmetallic (sometimes partially UO,) films are often (though not always) noted with alloys. Metallic layers have been noted in Inconel in both thermal convection loops and seesaw tests, INTRODUCTION One of the better waysof introducing fuel into a high-temperature reactor is through the use of uranium-bearing fused fluorides. The various metal fluoride systems are being investigated by the Materials Chemistry Division, and mixtures of the fluoride salts are being used in corrosion tests by the Materials Chemistry Division, the Experimental Engineering Group of the ANP Division, and the Metallurgy Division. This report contains the results of tests made to date by the Metallurgy CORROSION BY Division on the corrosion of metals by some of the fluoride mixtures. The tests were performed at 816°C, and two different types of tests were used. The major portion of the tests were run under static conditions in evacuated metal capsules. Other tests (seesaw tests) were run in tilting furnaces in which the ligquid is cycled continually from the hot zone to the colder zone of evacuated capsules; a complete cycle 1s made every 13 seconds.(!) Al]l tests were run with either fluoride mixture No. 2 (46.5 mole % NaF, 26.0 mole % KF, 27.5 mole % UF,) or No. 14 (43.5 mole% KF, 44.5 mole% LiF, 10.9 mole % NaF, 1.1 mole % UF,). EXPERIMENTAL PROCEDURE AND RESULTS Static Test Loading Method. The most recent technique evolved at the Oak Ridge National Laboratory for the preparation of static tests is shown in Fig. 1. Tubing is loaded, with specimen and corrodant, in a dry box with a purified helium atmosphere and sealed with vacuum tape, as 1s seen in Part 3 of Fig. 1. After the tube is removed from the dry beox, a vacuum is applied, as in Part 4, to break the vacuum tape; thus the material in the tube is not exposed to air. The tubing is then crimped twice (also shownin Part 4 of Fig. 1), and the bottom crimp is held in a vise while the top of the tubing is being welded, as shown in Part 5 of Fig. 1. It was found that a vacuum could not be held in the tubing unless the crimped portion was gripped in a vise during welding. Apparently the crimps will relax enough, unless gripped, to allow air to seep into the tube. If the specimen tube is connected into a manifold when the vacuum tape 1s broken prior to welding, any type of atmosphere can be placed in the “)D. C. Vreeland, R. B, Day, E. E. Hoffman, and L. D. Dyer, ANP Quar. Prog. Rep. March 10, 1952, ORNL-1227, p. 120. tube by metering the proper gases through the manifold system. Unless otherwise mentioned in the text, all the tests reported were run with a vacuum in the capsule. The specimen tubes used in the seesaw test were prepared in the same manner as those for the static tests. Screening Tests., A series of screening tests of various materials in molten fluoride mixture No. 14 were made, These tests were made with dehydrated, unpretreated fluoride mixture for 100 hr at 816°C under static vacuum. The results of these tests are givenin Table 1. Molybdenum, columbium, Monel, Stellite No. 25, and nickel plus 1/4% zirconium alloy were apparently unattacked. All of the 300 series stainless steels tested (304 ELC; 309, 310,317, 321, 347) were attacked to a depth of 1 mil or less, except types 304 ELC and 309, which were attacked to a depth of 2 mils. Four-hundred series stainless steels (only two types) were not attacked over 1 mil. Z nickel was the most severely attacked material of those tested, being affected to a depth of 5 mils. Figures 2 and 3 show typical samples of the types of attack observed. A seesaw test run with nickel-1/4% zirconium alloy showed severe mass transfer effects. Figure 4 shows this test. In Table 2 are shown the results of a few screening tests with fluoride mixture No. 2. A comparison of these results with those in Table 1 reveals that corrosive attack to approximately the same extent can be expected with both fluoride mixtures No. 2 and No. 14 1in static tests, Effect of Temperature om Corrosion. Tests on the effect of temperature have been runin both fluoride mixtures No. 2 and No. 14. Tables 3, 4, and § summarize the results of these tests. At temperatures below 1100°C, no significant effects can be noted that could not be caused by normal test UNCL ASSIFIED " e WACULM TAPE | - . i TLBING A5 Lk AP LD REFCEVED 2 TUBING WELDED 3 COMBODING MEDILN ? ::rp:l SREARE - 5 TUBING HELD 1N VigE TURING CRUMPED TWICE wWHILE BEING WELDED ! F » " ) I | o L] - O SPECIMEN EXANMINED, LEANED AND wWEIGHED & WELDED TuBE INSERTED T CAPSULE EVACUATED 4ND AOSULE OPENED WiTh 9 TUEELE“; “:};'fl' 11, SPECIMEN AND SECTION OF N FROTECTIVE CAPSULE WELDED - READY fOR TEST PIRE CUTTER ;;umw FOR ANALYSIS TUBE MOUNTED FOR METALLOGRAPHIC EXAMINATION Fig. 1. Loading and Testing Technigque Employed for Static Corrosion Tests. SHATHONTd NALTOW CORROSION BY TABLE 1. CORROSION OF VARIOUS MATERIALS TESTED IN FLUORIDE MIXTURE NO. 14 AT 1500°F FOR 100 hr IN VACUUM DEPTH OF METAL MATERIAL AFFECTED METALLOGRAPHIC NOTES (mils) Type 304 ELC stainless steel 2 Subsurface voids, some following grain bound- aries, attack somewhat irregular Type 309 stainless steel 2 Subsurface voids Type 310 stainless steel 1 Subsurface voids Type 317 stainless steel 1 Intergranular penetration Type 321 stainless steel 0 to 1/2 Intergranular penetration Type 347 stainless steel 1 Subsurface voids Type 430 stainless steel 1 Slight intergranular penetration and de- carburization Type 446 stainless steel 1 Subsurface voids Nickel + 1/4% zirconium 0 No visible attack Hastelloy B 0 to 1/4 Subsurface voids Hastelloy C 2 Subsurface voids Z nickel 5 Voids along grain boundaries Stellite No. 25 0 No visible attack Nichrome V 3 Subsurface voids, some following grain boundaries Monel 0 No visible attack Inconel Subsurface voids, some following grain boundaries Inconel X 1 Subsurface voids Tantalum 1 Surface of specimen roughened Columbium 0 Surface somewhat roughened Globe iron 2 Surface of specimen very rough, thickness decrease indicates solution type of attack Vanadium 1 Voids and intergranular penetration Molybdenum 0 Surface somewhat roughened TABLE 2. CORROSION OF VARIOUS MATERIALS TESTED IN FLUORIDE MIXTURE NO. 2 AT 1500°F FOR 100 hr IN VACUUM MATERIAL DEETHQF :T:?:;AFFECTED METALLOGRAPHIC NOTES 26% Mo-T74% Ni 1/2 Subsurface voids Molybdenum 0 No attack Inconel 2 Subsurface voids Type 310 stainless steel L 2 Subsurface voids Type 316 stainless steel 1 Subsurface voids URCL ASSIFIED ¥t e iy g o oy o TYPE 304LC TYPE M7 UNCL ARMIED oy LR CL ARKIFICD T il UMICL ABMFVED Foamd¥ N -Ri e B e e ~wiam &q :'1]1::. n{*;ymé}-» o £t - ! : 70 L SO M TR S AT Y . A { =y e w L' § ¥’ i - #%wwnfiflu$gfl?kfifl%+ A e & ba = A2 T & -.‘.‘.".T.J' '!r..i'\.“‘“"{":rJ "~ A.—;va'.' 5 q"...- J:'D‘f‘..f :'fl:i;ri.tzfir' TYPE 347 TYPE 430 TYPE 448& Fig. 2. Static Corrosion Tests of Stainless Steels in Fluoride Mixture No. 14 for 100 hr at 1500°F. Original magnification 250X, reduced 58%. SAATHONTA NALTOW INCONEL NICHROME Vv Z NICKEL UNCL ARSIFITD , : / ‘ . Pei & fl T . ‘-J-a'._ > ;-I “?"} AR A ..'-fl"?-‘}n' b7 NICKEL + 'fi"a ZIRCONIUM STELLITE NO. 25 MOLYBDENUM Fig. 3. Static Corrosion Tests in Fluoride Mixture No. 14 for 100 hr at 1500°F. Original magnifi- cation 250X, reduced 58%. A9 NOISOHMOD variations. However, at higher temper- corrosion is appreciably Analysis of the fluoride and atures, diminished. bath was made after these tests, Fig. 4. Cold End of Nickel-1/4% Zirconium Alloy Seesaw Test Specimen Run for 216 hr at aHot-End Temperature of 798°C and a Cold-End Temperature of 610°cC. TABLE 3. MAXIMUM PENETRATION OF INCONEL TESTED IN FLUORIDE MIXTURE NO. 14 AT ELEVATED TEMPERATURES .- FOR 200 HOURS MOLTEN FLUORIDES the results are in good agreement with the metallographic data. After testing at 800°C, 1800 ppm'6f chromium was detected in the bath; at 1300°C only 65 ppm of chromium was found. Figure 5 shows results obtained at different temperatures. Figure 6 shows the type of layer formed at temperatures of 1200°C and above. It is believed that this film acts as a barrier to the diffusion of chromium and thus prevents the formation of voids, X-ray-diffraction patterns of this film show it to be UQ,. This test was repeated but the sample was first heated in fluoride mixture No. 14 at 1250°C for a long period of time and then tested at 816°C for 100 hr; at this temperature also, the layer inhibited corrosion. It might be mentioned that this is a very unreliable method of minimizing corrosion because the film might rupture and localize the corrosion in one region. The high-temperature heat treatment is also undesirable because it results in considerable grain coarsening that is detrimental from a corrosion standpoint and to the ability to form the metal. The TEMPERATURE MAXIMUM PENETRATION formation of this film was dependent (“¢) (mils) on the fluoride mixture, and such a film appeared only in the tests in 800 3 which the fluoride mixture contained 900 1 1/2 Na, Li, K, and U, 1000 1 Effect of Time on Corrosion. A 1100 1 series of seesaw tests with Inconel 1200 0.5 was made over time intervals from 1300 <0.5 66 to 3000 hours. The results are summarized in Table 6. After about TABLE 4. MAXIMUM PENETRATION IN STATIC TESTS OF MATERIALS IN FLUORIDE MIXTURE NO. 2 AT ELEVATED TEMPERATURES FOR 100 HOURS MAXIMUM PENETRATION (mils) MATERIAL At B816°C At B850°C At 900°C At 1000°C Inconel 2 4 5 5 Type 310 stainless steel 1 172 2 4 3 Type 317 stainless steel 0 3 4 3 CORROSION BY UNCLASSIFIED Y-7183 —— S ——— e = (o) . © UNCLASSIFIED (56) _ . (d) Fig. 5. Attack on Inconel by Fluoride Mixture No. 14 at (a) 900, (b) 1000, (c) 1100, and (d) 1200°C in Static Test for 200 Hours. 250X. TABLE 3. MOLTEN FLUORIDES MAXIMUM PENETRATION IN STATIC TESTS OF MATERIALS IN FLUORIDE MIXTURE NO. 14 AT ELEVATED TEMPERATURES FOR 100 HOURS MAXIMUM PENETRATION (wils) MATERIAL At 538°cC At T04°C At 816°C At l000°C Specimen Tube Spacimen Tube Specimen Tube Speacimen Tube Inconel Slight Slight 4 1 1/2 1 1/2 3 3 3 roughening roughening Type 430 stainless steal <1 <1 1/2 1/2 1/4 1/4 Ne attack Type 304 stainless stesl < 1 <] 2 1 1/2 2 2 1 1/2 1 Type 321 stainless steel <1 <1 <1 1 1/2 /2 11/2 11/2 Fig. 6. No. 14 in Static Test for 200 Hours, 500 hr, the amount of additional attack is quite small. This is possibly due to saturation of the fluoride with chromium or the removal UNCLASSIF IED Y7074 Layer Formed on Inconel at 1200°C by Exposure to Fluoride Mixture 1000x%. of impurities in the fluorides. The test at 3000 hr had one small area of dense corrosion, that was 13 mils in depth. CORROSION BY Effect of Cold Work on Corrosion. A series of cold-worked Inconel specimens was prepared with from 0 to 70% reduction in thickness by rolling, These specimens were tested in fluoride mixture No. 14 for 100 hr at 815°C. No significant differences in behavior were noted. Another series of Inconel specimens was prepared with from 0 to 90% cold work. No differences were noted in this steel and Inconel were tested in fluoride mixture No. 2 in both the as-received and 20% cold-worked condition. No appreciable differences were noted in these tests. Data from these tests are presented in Tables 7 and 8. Addi tion of Inhibitors. A large number of both static and seesaw tests have been run with the addition of small percentages of other materials series. Types 316 and 310 stainless to the fluoride mixtures. In general, TABLE 6. EFFECT OF TIME ON THE CORROSION OF INCONEL BY FLUORIDE MIXTURE NO. 14 IN SEESAW TESTS Frir op TEMPERATURE (°C) METALLOGRAPHIC OBSERVATIONS Hot Zone TEST (hr) Hot Zone Cold Zone (depth of voids) Cold Zone 66 810 620 1/2 mil, average No evidence of 1 mil, maximum reaction 115 780 600 1 mil, average No evidence of 2 mils, maximum reaction 210 780 630 2 mils, average No evidence of 5 mils, maximum reaction 500 750 610 3 mils, average Metallic deposit 1/2 7 mils, maximum mil thick 750 815 620 3 1/2 mils, average | Metallic deposit 1/2 6 mils, maximum mil thick 3000 800 560 4 mils, average Metallic deposit 1/2 13 mils, maximum mil thick TABLE 7. EFFECT OF COLD WORK ON THE CORROSION OF INCONEL TESTED IN FLUORIDE MIXTURE NO. 14 FOR 100 hr AT 815°C REDUCTION BY COLD-ROLLING WEIGHT LOSS DEPTH OF SUBSURFACE THICKNESS (%) (g) VOIDS (mils) CHANGE 0 0.0020 1 0 D 0.0043 2 0 11.5 0.0036 13,2 0 26.0 0.0036 1 1/2 0 50.0 0.0023 1 0 71.0 0.0028 1 0 10 MOLTEN FLUORIDES TABLE 8. EFFECT OF COLD WORK ON CORROSION OF STAINLESS AND INCONEL TESTED IN FLUORIDE MIXTURE NO. 2 FOR 100 hr AT 815°C MATERIAL COLD WORK (%) PENETRATION (mils) Type 316 stainless steel As-received 1 20 2 12 Type 310 stainless steel As-received 1 tol 1/2 21 /2 te 1 V)2 Inconel As-received 1/2 to 2 26 1/2 active metals are effective in reducing TABLE 9. RESULTS OF ADDITIONS TO corrosion and in many tests they stopped FLUORIDE MIXTURE NO. 14 IN corrosion completely; other metal CORROSION TESTS* additions have no effect on the corrosion; the addition of chemical ADDITION MAXIMUM DEPTH OF ATTACK compounds often increases corrosion; (0.25 wet %) (mils) however, many exceptions are noted. Tables 9 and 10 summarize the results Na 0 of the seesaw and static tests. Be 0 In both types of tests, Na, Li, Ti, Mg 0 and Ca reduced corrosion. G. M. Al 0 Adamson has found that additions of Ti 0 zirconium and titanium are quite v 0 beneficial in minimizing corrosion Cr (2 tests) 0 by fluoride fuels in Inconel thermal Li (2 tests) 0 convection loops. Contradictory Cu 0 results were obtained on Zr, Mn, and NaH 0 Mg. The following tabulation classifies Ca 0 materials as those that have been Cc 1 beneficial and those that have not. Si 2 It is thought that the Zr and Mn used Mo (3 tests) 1/2 in the seesaw tests were not thoroughly Zn 4 dried. W 3 Fe 3 BENEFICIAL NOT BENEFICIAL Zr 4 Mn 4 Na Si Ag 4 Be Zn Nal 8 Al W KCl 10 ¥ Fe KBr 10 Cr Ag KI 8 Li Nal NaCl 8 Cu KCl LilO 10 NaH KBr 2 Ca Mnoz *Seesaw tests run for 200 hr at BHBOC; material C Nin tested was Inconel. 11 CORROSION BY TABLE 10. EFFECT OF VARIOUS ADDITIONS TO FLUORIDE NO. 14 IN STATIC CORROSION TESTS AT 816°C FOR 100 HOURS DEPTH OF METAL MATERIAL ADDITION AFFECTED METALLOGRAPHIC NOTES (mils) Inconel 2% Mg 5 Large voids in both specimen and tube 2% Zr 1 Surface layer 1 1/2 to 2 mils thick, 1/2 to 1 mil of attack beneath surface layer 10% Zr 1 1 1{2-ni1 surface layer on specimen and tube, subsurface voids to 1 mil under surface layer. 10% Na 0 No attack 1% Na 1/2 Specimen and tube have a few subsurface voids 10% U 1 l1- to 2-mil surface layer 5% Li 1/2 l-mil surface layer 5% K 0 No attack 5% Ca 0 1/2-mil surface layer, no attack be- neath surface layer 5% Ti 0 1/2-mil surface layer, no attack be- neath it 5% Mn 0 No attack Type 321 stainless steel 10% Zr 0 1/2-mil surface layer on specimen and tube, no attack beneath layer 10% Na 0 No attack 1% Na 0 No attack 10% U 0 2 to 3-mil surface layer, no attack under layer 5% Li 0 No attack 5% K 0 No attack 5% Ca 1/2 1/2-mil surface layer, attack to 1/2 mil beneath surface layer 5% Ti 0 No attack 5% Mn 1 Subsurface voids 1/2 to 1 mil A-nickel 10% Zr 0 2 1/2-mil surface layer, no attack under layer 10% Na 0 1/2-mil surface layer, no attack be- neath layer Type 309 stainless steel 2% Mg 5 Subsurface voids, attack irregular in occurrence 2% Ir 0 No attack on specimen or tube, specimen has 1/2-mil surface layer 10% MnO, 4 Subsurface voids and intergranular penetration to 4 mils Inconel 10% MnO, 9 Subsurface voids and intergranular penetration 10% CrF, 4 Subsurface voids 10% NiF, 7 Subsurface voids 10% FeF, 9 Subsurface voids 31/3% Cer. 2 Subsurface voids 3 1/3% NiFI. 3 1/3% Fer 12 Mo FelF, Ti CrF, Zr NaCl LiIO, It is thought possible that the beneficial materials added may be acting to reduce the available amount of fluorine in the system, which would cause the atoms of the structural metals to go into solution. It is also possible that the beneficial additions act as oxygen getters., The metal fluorides were added with the thought that perhaps a reaction of the following type might be the corrosion mechanism: ALkF + Me T— MeF + Alk where Alk could represent Na, Li, or K and Me could represent Fe, Ni, or Cr. If this were the case, then additions of iron, nickel, or chromium fluorides and also sodium, lithium, or potassium should tend to drive the reaction to the left and help reduce corrosion. However, additions of iron, nickel, and chromium fluorides had no such effect in reducing corrosion. The addition of iron and nickel fluorides actually increased the corrosion; this is probably due to the reduction of the nickel and iron fluorides by the chromium that was removed from the container wall, Metallic sodium, lithium, and potassium did reduce corrosion; but, as mentioned above, their action is probably that of getters of oxygen and fluorine rather than that of taking part in the reaction. The MnO additions were made to see whether oxygen added to these tests in this form would increase corrosion, and 1t did. The alkali halides were added to see whether the presence of other negative ions than fluoride would have an inhibiting effect, and they did not. In many of the tests that were run with additions, surface layers were apparent on the specimens after test. The following tabulation gives the MOLTEN FLUORIDES combinations of containing metals and additions that resulted in surface layers. It is possible, of course, that in some cases in which surface layers were not observed, the layers may have been inadvertently cracked off during stripping. ADDITION TO FLUORIDE BATH CONTAINING METAL Zr Inconel Zr Type 309 stainless steel Zr Type 321 stainless steel Zr A Nickel U Inconel U Type 321 stainless steel U A Nickel Ca Type 321 stainless steel Ca Inconel Ti Inconel Li Inconel The i1dentity of the surface layers is not definitely established. 1In the case of zirconium additions in type 309 stainless steel, the surface layer was reported by x-ray to contain UO, and ZrO,. In other checks on these surface layers, UOzis.usually reported and, occasionally, unknown constituents are detected. Effect of Heat Treatment and Carbon Content. Five tests with different heat treatments were made with Inconel, and no significant differences were noted. However, in tests in which the Inconel had been hydrogen-fired at temperatures up to 1200°C, which created a large grain size, attack has been observed to be more localized. The depth of attack was greater, but the total volume was less. Table 11 summarizes the tests run to determine the effect of heat treatment, Mass Transfer in Molten Fluorides. Metallic deposits have been noted in seesaw tests with fluoride mixture No. 14 in Globe iron, A nickel, and in a nickel alloy containing 1/4% zir- conium. Yery slight amounts of metallic deposits have been noted in a few Inconel tests. Thin (less than 0.5 mil) layers have been seen in the cold 13 CORROSION BY TABLE 11. EFFECT OF HEAT TREATMENT OF INCONEL ON CORROSION BY FLUORIDE MIXTURE NO. 2 AT 815°C FOR 100 hr IN STATIC TEST HEAT TREATMENT DEPTH OF SUBSURFACE VOIDS (miis) Water-quenched from 2100°F 1/2 to 2 Furnace-cooled from 2100°F 142 0] 1/2 Water-quenched from 2100°F, reheated to 1600°F for 24 hr, and water-quenched 1/2 ta 3 1/2 Furnace-cooled from 2100 to 1600°F, held for 24 hr, and water-quenched 1/2 to 1 Heated to 1600°F, held for 24 hr, and water- quenched 1/2 to 3 1/2 legs of thermal convection loops with both Inconel and stainless steels, Deposition of a nonmetallic layer occurs in the cold zone of many seesaw tests, thermal convection loops, and in some static tests. This coating is quite thin, being less than 1 mil in all cases except with the additions of zirconium and uranium. The film 14 apparently has an inhibiting effect on mass transfer, and has been found by x-ray-diffraction studies of several tests to be composed of UO, and an unknown constituent or constituents, The small amount of mass transfer occurring in these tests is not believed to be a problem except in the case of pure metals. MOLTEN FLUORIDES APPENDIX SUBSURFACE VOID FORMATION IN METALS DURING HIGH-TEMPERATURE CORROSION TESTS A. deS. Brasunas Inconel and similar alloys have been observed to be susceptible to subsur face void formation during processes involving chromium depletion of the surface zone by any of the following methods: 1. leaching with certain molten salts, 2. high-temperature vacuum treatment, 3. high-temperature oxidation. Porosity inmetals has been observed previously in diffusion experi- ments(1+2+3) and is attributed to the uneven diffusion rates of different metal atoms through a metal lattice. Although similar observations have not previously been reported in corrosion studies, their occurrence should not be too surprising. (Voids have been reported(** %) in the oxide layers of metals and alloys undergoing oxidation in air. Under these conditions, metal ions diffuse outward more rapidly than oxygen ions can diffuse inward through the oxide layer.) Alloys are known to change in composition in the surface regions because of the higher reactivity of certain alloy constituents at the metal inter face. Such depletion without adequate replacement of removed atoms causes vacancies, which may “precipitate’ to form visible voids. Al though undesirable, this form of corrosion is not as damaging as some and could be tolerated in certain instances. (1)), Smigelskas and E. Kirkeandall, Tranas. Am, Inst. Mining Wet. Engrs. 171, 130 (1947). U w6 daBlive snd B B Mokl i Wetala 191, 153 (1951). 3)H. W. Balluff and B. H. Alexander, Sylvania Electrie Produets, Inc., SEP-83 (Feb. 1952). 6D L Preil, J. Insts Watdla, 119 561 (1929). ISIA. deS. Brasunas and N. J. Grant, Trans. An. Soc. Metals 44, 1117 (1952). Corrosion in Molten Fluoride Salts. Corrosion tests in which fused alkala metal fluoride salts (fluoride mixture No. 14) come into contact with chromium- bearing alloys such as Inconel at temperatures above 1300°F have resulted in the formation of a porous surface layer, as illustrated in Figs. Al and A2. Although it is fully recognized that in many cases voids have been observed metallographically because (1) particles have been physically removed during polishing, (2) certain phases may react with water and dissolve, or (3) etch pits can result, it seemed nevertheless probable that these may be true voids. Careful polishing by using nonaqueous techniques has failed to show any difference in the appearance of the porous surface region of the metal specimens in the etched or un- etched conditions. Chemical analyses of the metal surface before and after testing have indicated only one significant change; there was an appreciable drop in the chromium con- tent from about 15% to about 5%. The analysis of the fluoride salt bath showed correspondingly higher chromium- to-iron and chromium-to-nickel ratios than would result from uniform attack of the alloy. This clearly establishes the selective removal of chromium. Supplementary Data. The saturation of the fluoride salt with chromium prior to testing was effective in suppressing void formation in tests that would otherwise have resulted in voids to a depth of several mils. Furthermore, the measured area of voids observed agrees very closely with the amount calculated on the basis of the change in chemistry of the metal surface mentioned above. 15 CORROSION BY UNCLASSIFIED Y-7014 Fig. Al. Subsurface Voids Usually Encountered After Inconel or Stainless Steels Contact Molten Alkali Fluoride at Temperatures of About 816°C for 100 Hours. Inconel, unetched, 250X. UNCLASSIFIED : - Y6794 S e Fig. AZ. Appearance of Subsurface Voids at High Magnification. Etched, 2000x. 16 This computation assumes no over-all volume change of the specimen, If voids, are, caused by chromium depletion, as these experiments indi- cate, then voids should also be observed when loss of chromium is effected by other means. Therefore two methods were tried; the first technique in- volved high-temperature vacuum treat- ment and the second involved high- temperature oxidation. Both attempts were successful and will be described briefly. High-Temperature Vacuum Treatment. The high vapor pressure of chromium relative to that of iron or nickel suggests that high-temperature vacuum treatment would also be a suitable means of producing a chromium gradient, Specimens of Inconel and an 80% Ni-20% Cr alloy were exposed to a vacuum of 0.1 mm Hg for 42 hr at 1375°C and then furnace-cooled. Both samples showed many subsurface voids; the voids in Inconel appeared to be spherical, whereas those in the 80% Ni-20% Cr alloy were angular. These are shown in Fig. A3. High-Temperature Oxidation. It 1is well known that chromium-bearing alloys have good high-temperature oxidation resistance that can be attributed to the formation of a chromium-rich, diffusion-resistant oxide phase on the surface of the alloy. The chromium in the oxide phase is supplied by the surface and the underlying regions of the speci- men, which would tend to be partially depleted of chromium. An Inconel specimen was therefore exposed toair at 1250°C for 200 hours. After oxidation, metallographic exami- nation revealed the usual oxide layer on the surface and in the adjacent grain-boundary areas. In addition, subsurface voids were observed as anticipated; these are shown in Fig. A4, Theoretical Comnsiderations. The tests just described strongly indicate that void formation is caused by MOLTEN FLUORIDES chromium depletion. The formation of cavities large enough to be resolved " by an ordinary microscope is visualized to occur as illustrated schematically in Fig. AS5. As chromium is leached from the surface of the metal, a con- centration (activity) gradient results that causes chromium atoms from the underlying region to diffuse toward the surface and leave behind a zone enriched with vacancies, as shown in the center sketch of Fig. A5. These vacancies can agglomerate at suitable sites and become visible as voids. A certain number of vacancies can be tolerated in crystal lattices of solid metals, and such imperfections are believed to exist in practically UMCLASSIFIED ¥-4825 Fig. A3. Section of 80% Ni-20% Cr Alloy After 42-hr Exposure to Vacuum at 1375°C. Void shape bears some relationship to twin lines; note simi- larity to Fig. A2, Etched, Original magnification 250X, reduced 32%. 17 CORROSION BY UNCLASSIFIED 7072 Fig. A4. Section of Inconel Specimen After 200-hr Exposure to Air at 1250°C. Note void formation beneath oxidized region. Unetched, 250X. C CHROMIUM ATOM UNCLASSIFIED F IRON ATOM DWG. 16322 N NICKEL ATOM BEFORE LEACHING VACANCIES CAUSED BY LEACHING VACANCIES PRECIPITATED Fig. A5. Mechanism of Subsurface Void Formation. 18 all solid metals, the amount increasing with increasing temperature. Further- more, they can also be generated at sur faces, grain boundaries, inclusions, dislocations, and other possible crystal defect areas. One of the mechanisms of diffusion is therefore visualized as taking place by the movement of these vacancies; that 1is, neighboring atoms move to occupy vacant sites which give the 1llusion of vacancy movement in the opposite direction, For the sake of clarity none of the existing “tolerated’ vacancies are shown in Fig. A5, although they must be assumed to be present at all times. The left sketch shows the Fe, Ni, and Cr atoms arranged at random as one may expect to find them on the 111 plane of an Inconel specimen. After selective diffusion of chromium has occurred by any of the test conditions described (from about 15 to 5%), vacancies must result if these vacant lattice sites are not filled by diffusion of nickel or iron atoms from the surface proper. If the lattice had been “saturated” with vacancies, then these additional ones are not ‘“‘tolerated’ and must “precipitate” as shown in the sketch at the right in Fig. A5. For ease of illustration, only six atoms are shown to have precipitated to form a void; the number of missing atoms in voids actually observed in these experiments is of the order of 102, e S - — — — — 5 MOLTEN FLUORIDES \\ The voids shown in Fig, A3 are of two sizes, The larger ones are believed to have occurred at temperature in the usual manner, whereas in the case of the smaller ones, it is interesting to visualize them as having occurred on cooling. Rough approximations of vacancy densities at several tempera- tures based on the formula of Mott and Gurney‘®’ indicate that such a theory is not improbable. Hence, it may be assumed that there is a temperature coefficient of “solubility” for vacant atoms such as there i1s for foreign atoms. Conclusions. In reactions in which the neteffect of the diffusion involved is essentially monodirectional, there is a movement of mass that results in a change in density and/or shape of that portion of the specimen. If the concentration of vacancies left behind is greater than some critical value, they will tend to collect at suitable locations and appear as visible voids. Such diffusion phenomena have been observed on a number of occasions when bimetallic (i.e., metal-metal) dif- fusion couples were studied,(1:2.3) It has been demonstrated that identical effects can be obtained in metal- liquid and metal-gas systems in which similar diffusion phenomena occur. {511'!. F. Mott and R. W, Gurney, Electronic Processes in Jonie Crystals, p. 31 Oxferd Clarendon Press, 1940. 19 S | ION losure of its contents - If - TION