RESEARCH LIBRARY. Ak (e UNCLASSIFIED ORNL.-2048 L COPY NO. &7 Controct No. W-7405-eng-25 METALLURGY DIVISION CORROSION OF MATERIALS IN FUSED HYDROXIDES G. P. 5mith DATE ISSUED oo OAK RIDGE NATIONAL LABORATORY Operated by UNION CARBIDE MUCLEAR COMPANY A Division of Union Carbide and Carbon Corporation Post Office Box P Oak Ridge, Tennessee MARTIN MARIETTA FNSRGY SYSTEMS LIBRARIES uncLassiFiep TIRIREIRINED 3 4yyshL 0350099 5 UNGLASSIFIED ORNL-2048 Metallurgy and Ceramics INTERNAL DISTRIBUTION 1. C. E. Center 46, D. W, Cardwell 2. Biology Library 47. W. D. Manly 3. Health Physics Library 48. E. M. King 4. Metallurgy Library 49. A. L. Miller “56. Central Research Library 50. D. D. Cowen 7. Reactor Experimental 51. P. M. Reyling Engineering Library 52, G. C. Williams 8-20. Laboratory Records Department 53. R. A. Charpie 21. L.aboratory Records, ORNL R.C. 54, M, L. Picklesimer 22. A. M, Weinberg 55. G. E. Boyd 23. L. B, Emlet (K-25) 56. J. E. Cunningham 24. J. P. Murray (Y-12) 57. H. L. Yakel 25. J. A, Swartout 58. G, M. Adamson 26. E.H. Taylor 59. M. E. Steidlitz 27. E.D. Shipley 60. C. R. Boston 28. F. C, Vonderl.age 61. J. J. McBride 29, W. C. Jordan 62. G. F. Petersen 30. C.P. Keim 63. J. V. Cathcart 31. J. H, Frye, Jr, 64. W. H, Bridges 32. R. 5. Livingston 65. E. E. Hoffman 33. R. R. Dickison 66. W. H. Cook 34. S. C. Lind 67. W, R. Grimes 35. F. L, Culler 68, F. Kertesz 36. A. H. Sneli 69. F. A. Knox 37. A. Hollaender 70. F. F. Blankenship 38. M. T, Kelley 71-90. G, P, Smith 39. K. Z. Morgan 91. N. J. Grant (consultant) 40. J. A, Lane 92. E. Creutz {(consultant) 41. T. A. Lincoln 93. T. 3, Shevlin {consuitant) 42. A. S, Householder 94. E. E. Stansbury (consultant) 43. C. S. Harrill 95. ORNL - Y-12 Technical Library, 44. C. E. Winters Document Reference Section 45. D, S. Billington EXTERNAL DISTRIBUTION 96. R. F. Bacher, California Institute of Technology 97. Division of Research and Development, AEC, ORD 98. R. F. Kruk, University of Arkansas, Fayetteville, Arkansas 99. Douglas Hill, Duke University, Durham, North Carolina 100. L. D. Dyer, University of Virginia, Charlottesville 101. R. A. Lad, National Advisory Committee for Aeronautics, Cleveland - 102. L. F, Epstein, Knolls Atomic Power Laboratory 103. E. G. Brush, Knolls Atomic Power Laboratory UNCGLASSIFIED " 104. 105, 106. 107. 108. 109. 110-428. UNCLASSIFIED D. D. Williams, Naval Research Laboratory R. R. Miller, Naval Research L.aboratory E. M, Simons, Battelle Memorial Institute H. A. Pray, Battelle Memcrial Institute P. D. Miller, Bottelle Memorial Institute M. D. Banus, Metal Hydrides, Inc. Given distribution as shown in TID-4500 under Metallurgy and Ceramics category DISTRIBUTION PAGE TO BE REMOVED {F REPORT IS GIVEN PUBLIC DISTRIBUTION UNGLASSIFIED UNCLASSIFIED CONTENTS A S T R A T e e e ettt 1 1o INTRODUCTION Lot ettt ee et ee et ee e em e e s e e s et 1 2. ATOMIC NATURE OF FUSED SODIUM HYDROXIDE oo 2 3. CORROSION OF CERAMICS ot 2 Solubility Relations in Fused Hydroxides oo, 2 Oxide-lon Donor-Acceptor REACHIONS . oot 3 Ceramics with Saturated Cations and Saturated ARIONS .ooveeeoveeoe oo 3 Ceramics with Acceptor Cations and Saturated AnionS ..oooveeoeeeeeeeeee oo, 4 Ceramics with Saturated Cations and Acceptor ANTONS oo eev oo 6 Oxidation-Reduction Reactions oo, 7 Corrosion of Other Kinds of Ceramics oo 7 Secondary Corrosion PRENOMENT ..o iciiieeee ettt en e e 7 SUMIMIGEY ot ettt ettt et et et e e ee e e ee s 2t eaee e e et e emee et e e en e e e e s e e s e e te e ee et ens e e e e et es e eeess oo 8 4. CORROSION OF METALS Lo ettt evae et er e e e s e e et ees e e 8 Corrosion of Metals by Oxidation ...t 8 Corrosion by Hydroxyl Tons oottt ee e 8 Corrosion by Alkali Metal lons ..o, et sttt ner e ane s aben e et eeeane s 11 Corrosion by Oxidizing SolUtes . ..o es oot e oo eee e (A Corrosion of A0y S oo ettt e et sener et (R 5. MASS TRANSEER .ottt et ee e e ee e e er e e 16 TR | ettt bt e e et s et e e e e e e e et e et s oo es et n e e ee e rans 17 Other Elemental Metals ...ttt est e ee e ee s ers e 20 ALOY S ottt ettt et a et n e ee et e te e e et et e et ee s e e e e 20 Bimetal e B eets oottt 20 Mechanism of Mass Transfer oottt er e e e 21 Differential Solubility (Mechanism 1) ................................................................................... 21 Oxidation-Reduction Processes (Mechanisms V1) ..o 21 Local Cell Action (Mechanism Y o oo e 22 Reduction of Hydroxyl lohs(MechqniSm T e e e e 22 Reduction of Alkali Metal lons (Mechanism 1V) oo, 23 Reduction of Solutes (Mechanism V) e e e e 23 Disproportionation (Mechanism V1), ettt et as 23 Summary of MechaniSms ..ot er e nn. 24 SUMIMIAIY ettt e e asase s he s oaaa s s ase s e e et e ae s e et enesee et s e st e e estetsesaeseseebesseeeteesseses oo 25 REFERENCES ettt e et e ee e ee e s e e sannee s s et esens 27 UNGLASSIFIED UNGLASSIFIED CORRi)SION OF MATERIALS IN FUSED HYDROXIDES G. P. Smith ABSTRACT Some of the fused alkali-metsl hydroxides are of potential interest in reactor technology both as coolants and as moderators. The property. which most discourages the use of these substances is their corrosiveness, The corrosion of ceramics and ceramic-reloted substances by fused hydroxides occurs both by solution and by chemical reaction. A few ceramics rea¢t with hydroxides by oxidation- reduction reactions, but most have been found to be attacked by oxide-ion donor-acceptor re- actions, Magnesium oxide was the most corrosion-resistant of the ceramics which have been tested, although several other ceramics are sufficiently resistont to be useful. The corrosion of metc}ls and alleys by fused hydroxides takes place primarily by oxidation of the metal, accompanied by reduction of hydroxyl ions to form hydrogen and oxide iens. Three side reactions are known, but two of them are not important in most corrosion tests. The corrosion of alloys involves either the formation of subsurface veoids within the alloy or the formation of complex, twa-phase corrosion products at the surface of the alioy. Nickel is the most corrosion- resistant metal which has been studied in fused sodium hydroxide. Mass transfer, inducéd by temperature differenticls, has been found to be the primary factor limiting the use of metals which, under suitable conditions, do not corrode seriously. I. INTRODUCTION Some of the fused alkali-metal hydroxides are of potential interest in reactor technology as moder- ators, moderator-coolants, and, possibly, moderator- fuel vehicles.! The heat-transfer properties of these substances are sufficiently good for them to have been recommended as heat- transfer media.? For use as a high-temperature moderator without a cooling function, sodium hy- droxide possesses the virtues of being cheap and available as contrasted with its competitors, either liquid or solid. Furthermore, an essential sim- plicity in reactor design is achieved by combining in one material the two functions of slowing neutrons and cooling.? ' industrial There are only a few substances which will do both at high tempera- tures, Of the high-temperature moderator-coolants, the fused alkali-metal hydroxides are among the most stable with respect to both thermal dissociotion and radiation domage. |t has been known for some time that the alkali metal hydroxides should be stable toward thermal dissociation. However, not until recently have techniques been developed whereby this dissociation could be guantitatively Steidlitz and Smith? have shown that this dissociation is relatively small, For example, at 750°C the partial pressure of water in equi- librium with fused sodium hydroxide is of the order of magnitude of 0.1 mm Hg. The radiation stability of fused hydroxides' has been reported by Hochanadel® with regard to electron bombardment and by Keilholtz et ul. ® with regard to neutron bombardment. The radigtion levels reported were very substantial, and ne evidence was found for radiation damage. measured. The property of fused alkali-metal hydroxides which most discourages their application at high temperatures is their corrosiveness. It is the purpose of this discussion to review the current scientific status of studies of corrosion by fused hydroxides. , fi During the past five years, knowledge of the cotrosive properties of fused hydroxides at high temperatures has increased considerably. Never- theless, this is a new field of research and one in which there has been found a remarkable variety of corrosion phenomena, so that the areas of ignorance are more impressive than the areas of knowledge. For this reason, the writer has UNCLASSIFIED _ 1 chosen to ploce emphasis on some of the problems which are most in need of solution., This review will, accordingly, be concermned with the kinds of corrosion phenomena which have been observed rather than with the compilation of engineering reference data, Moreover, with a view toward stimulating further research, the writer has taken the liberty of presenting a number of quite specu- lative ideas on the origins of some of the corrosion phenomena, Only in this way con many importani possible accomplishments of further research be indicated. In this report, the term ‘‘fused hydroxides’ is taken to mean the fused alkali-metal hydroxides; the exception is lithium hydroxide, which, like the fused hydroxides of the alkaline-garth metals, is less stable with regard to thermal dissociotion than the other alkali metal hydroxides. Most of the available corrosion data are for fused sodium hydroxide. Information on corrosion in fused po- tassium hydroxide indicates that this substance behaves qualitatively like sodium hydroxide. There are very little dofo available on the hydroxides of rubidium and cesium. 2. ATOMIC NATURE OF FUSED 50DiUM HYDROXIDE Studies of electrical conductivity? and of the freezing-point depression by electrolyte solutes® indicate that fused sodium hydroxide is o typical fused electrolyte which obeys, at lzast approxi- mately, the model of ideal behavior proposed for such substances by Temkin.? Thus, sodium hy- droxide consists of soedium ions and hydroxyl ions held together by coulombic attraction such that no cotion may be considered bound to a particular anion, although, on the avercge, each cation has only anion neighbors and vice versa. When an ionic sodium compound is dissolved in fused sodium hydroxide, the cations of the solute become indistinguishable from those of the solvent. The hydroxyl anion is not a simple particle of unit negative chorge but has a dipole siructure and, under suitable conditions, u polypole structure.10 3. CORROSION OF CERAMICS Studies which have been made of the behavior of ceramic materials in fused hydroxides are limited in scope both as regords the variety of substances tested and the information obtained on any one substance. The use of petrography and x-ray diffraction, which have been largely ignored in the past, together with o proper regard for the reactivity of many corrosion products in contact with moisfure and carbon dioxide, is essential for further progress. The corrosion of ceramics and of ceramic-ralated substances by fused hydroxides has been observed to take place by solution and by chemical reaction, A few of the chemical reactions which have been observed were found to be oxidation-reduction re- actions, These will be described briefly near the end of this section. The majority of the known reactions of ceramics and ceramic-related sub- stances with fused hydroxides have been found to be oxide-ion donor-acceptor reactions. This kind of reaction in fused hydroxides may be treated in a quantitative although formal way by appli- cation of acid-base analog theory. Such a ireat- ment is beyond the scope of this report. However, some of the concepts derived from the theory of oxide-ion donor-acceptor reactions are very useful in describing the reactions between ceramics and hydroxides and will be considered. Solubility Relations in Fused Hydroxides, — A ceramic material which is chemically inert toward fused hydroxides might be limited in usefulness because of appreciable solubility in these media. There have besn very few quantitative measure- ments of solubility relations in fused hydroxides, and none of these measurements have been made for substances which are of importance as ceramics, However, in a discussion of the corrosion of ceramics, such solubility relations must inevitably be menticned if only in a qualitative way. There is a fundamental difference between solu- bility relations in nonelectrolytes such as water ond in fused electrolytes such as hydroxides. Solutions of a simple ionic substance in water may be described in terms of a binary system, while such solutions in fused hydroxides are reciprocal salt systems which require the specification of four components provided that the solute does not have an ion in common with the solvent, Under special conditions reciprocal salt systems may be treated as quasi-binary systems, but it is unwise to assume such behavior in the absence of confirmatory experimental evidence. Metathesis reactions in aqueous media are, of course, the manifestation of reciprocal salt solu- The corresponding type of re- action in a fused hydroxide solution is complicated by the fact that six composition variables would in general be needed to describe the possible solid phases which could precipitate from solution. bility relations, Oxide-lon Donor-Acceptor Reactions. — Many chemical substances have a significant affinity for oxide ions. Two well-known examples are carbon dioxide and water, which react with oxide ions to form, respectively, carbonate ions and hydroxyl ions. Such substances are referrad to as “‘oxide- ion acceptors.” Once an acceptor has reacted with an oxide ion, it becomes a potential oxide-ion donor and will give up its oxide ion to a stronger acceptor. Thus, the hydroxy! ion is an oxide-ion donor which is conjugate fo, or derived from, the oxide-ion acceptor, water. Carbon dioxide is « stronger acceptor than water and hence will react with hydroxyl ions to form carbonate ions and water, Such a reaction may be viewed as a compe- tition for oxide ions between the two acceptors, carbon dioxide and water: (3.1) CO, + 20H- it Co,~~ + H,0 acceptor | + donor Il = donor | + acceptor [} The reaction goes in that direction which produces the weaokest donor-acceptor pair, ' Many substances will behave toward hydroxyl ions like carbon dioxide in Eq. 3.1; that is, they will capture oxide ions and liberate water. This is not necessarily because the acceptor in question is intrinsically stronger than water. An acceptor which is stronger than water must successfully compete with water for oxide ions when water and the acceptor in question are at the same concen- tration. Frequently reactions like Eg. 3.1 take place because water is easily volatilized from fused hydroxides at high temperatures, and conse- quently its concentration is very low compared with the concentration of the competing oxide-ion acceptor. However, the water concentration in a fused hydroxide may be maintained constant by fixing the partial pressure of water over the melt, Under this condition o comparison may be made of the relative abilities of two oxide-ion acceptors to capture oxide ions from hydroxyl ions in the presence of the same fixed concentration of water. Thus it is meaningful to speck of the relative oxide-ion acceptor sirengths of two substances in fused hydroxide solutions. As was pointed out in the preceding section, when an ionic substance is dissolved in a fused hydroxide, dissociation, in the sense of separation of ions, occurs, and the cations and anions of the dissolved substance behave as chemically separate entities, This does not mean that complex ions may not form. It means that an ionic compound such as sodium chloride dissolves as sodium ions and chloride ions rather than as sodium chloride molecules. This concept of ionization is' the foundation of most of the modern chemistry of fused electrolytes. On the basis of this concept it is postulated that, when an ionizable solute is dis- solved to form a dilute solution in a fused hydroxide medium, separate oxide-ion affinities may be ascribed to the cations and to the anions of the solute., Although there are definite limi- tations to the application of this postulate, it is very useful for two reascns. First, it provides a satisfactory qualitative description of the known reactions between solutes and hydroxyl ions in fused hydroxides. Second, to the extent that this postulate is true, it allows inferences to be made about the reactivity of an untested ionic compound composed of cations C* and anions A~, provided that the behavior of C* and A~ is known separately trom the reactions of compounds which have as cations only C* and other compounds which have as anions only A—. Considerations of the origin of the oxide-ion affinity of cations, such as has been presented by Dietzel'! and by Flood and Férland,'? indicate that cations of very low ionic potential should have low oxide-ion affinities, The cations to be found in this category are the alkali-metal ond alkaline-earth These ions have ac- cordingly been found to show only o weak-to- negligible tendency to react with hydroxyl ions in tused hydroxide solution, _ Likewise, certain kinds of anions have such small oxide-ion affinities that they have been found to show only a weak-to-negligible tendency to capture oxide ions from hydroxyl ions in fused hydroxide solution. The onions in this category include the oxide ion, the halide ions, and some oxysalt anions such as carbonate and sulfate. Obviously, ortho-oxysalt anions belong in this category inasmuch as they are all saturated with oxide ions. lons which show little or no tendency to accept oxide ions from hydroxyl ions in fused hydroxides will be referred to, for convenience, os "‘saturated’’ ions. ; Ceramics with Saturoted Cotions and Soturated Anions, — lonic compounds with soturated cations and anions have thus far all proved to be too soluble for use as solid components in fused hydroxide media. The alkali metal oxides!3 gnd halides®:14 have been found to be quite soluble. cations. Barium chloride!3 showed significant solubility in sodium hydroxide ot 350°C. A sample of calcium oxide fired at 1900°C to give an apparent porosity of 3% was reported!® to be severely attacked by fused sodium hydroxide at 538°C, although other tests!3 indicate that its solubility is small at 350°C. Ceramics with Accepter Cotions and Soturated Anions, — A substantial proportion of the ceramics. and ceramic-related substances which have been tested in fused hydroxides is composed of cations with appreciable oxide-ion acceptor strengths and anions which are saturated. Cations which accept oxide ions from hydroxyl ions should do so by the formation of either oxides or oxysalt anions, The formation of an oxide by the reaction of an acceptor cation with a fused hydroxide has been observed in studies!3 of the reactions of mag- nesium chloride and nicke! chloride with fused sodium hydroxide. These compounds reacted very rapidly at 400°C to form insoluble magnesium oxide and insoluble nickel oxide. The reaction of mag- nesium chloride is given by the equation (3.2) MgClz(solid) + 20H~ (melt) = MgO(sclid) + 2CI~(melt) + HQO(gos) Considerakble data are available on the reaction of oxides with hydroxyl ions. Three oxides, those of magnesium, zinc, and thorium, have been tested up to temperatures without giving evidence of Four oxides, those of cerium(lV), nickel, zirconium, and aluminum, have been shown under some conditions to have a sig- nificant resistance toward reaction with hydroxyl ions, although they are all known to be capable of reacting completely. Two oxides, those of niobium(V) and titanium, have been found to react relatively rapidly at lower temperatures. substantial reaction, Further details on the above oxide-hydroxide reactions will be given below. The sction of fused sodium hydroxide on mag- nesium oxide was studied by D'Ans and Loffler!¢ at temperotures up to 800°C, but they were unable to detect any woter evolution, Steidlitz and Smith,? who made mass spectrometric determi- nations of the gases evolved on heating sodium hydroxide to 800°C in o vessel cut from a mag- nesium oxide single crystal, found water vapor, but its presence was accounted for quantitatively in terms of the thermal dissociation of hydroxyl ions in the presence of sodium ions. The chemical stability of magnesium oxide in fused sodium hydroxide is not surprising, since magnesium is not known to occur in an oxysalt anion. 7 Magnesium oxide is not only chemically stable in fused sodium hydroxide but is also quite in- soluble. In corrosion tests, Steidlitz and Smith4 found that single-crystal specimens of magnesium oxide in a large excess of anhydrous sodium hydroxide at 800°C. decreased in thickness by less than 0.001 in. in 117 hr. Water in fused sodium hydroxide has been found to attack magnesium oxide. Boston!® reported that anhydrous sodium hydroxide has no visible effect on cleavage planes and polished surfaces of mag- nesium oxide crystals up to 700°C but thut the presence of water vapor over the melt caused rapid etching of the crystal surfaces. This effect of water is enfirely consistent with the oxide-ion donor-acceptor concept. The equilibrium between magnesium oxide and water in the presence of sodium hydroxide may be expressed as (3.3) MgO(s) + H,O(melt) = 20H"(melt) + Mg++(me|f) The activity of solid magnesium oxide, MgQ(s), is a constant, the hydroxy! ions, OH=(melt), are present in great excess and should have approximately constant activity. Conse- quently, the application of the law of mass action to Eq. 3.3 shows that the activity of magnesium ions in the melt should be proportional to the activity of water in the melt; that is, an increase in the conceniration of water in the melt should increase the apparznt solubility of magnesium oxide. Furthermore, The practical application of magnesium oxide as a ceramic material for service in fused anhydrous hydroxides presents difficulties. Most sintered compacts of pure magnesium oxide ore sufficiently porous to absorb appreciable quantities of sodium hydroxide. l.arge magnesium oxide single crystals can be machined into fairly complicated shapes when a special need for a particularly corrosion- resistant part justifies the expense. At the Ook Ridge Nationa! Laboratory, reaction vessels have been machined from massive magnesium oxide crystals, and in every instance such vessels have proved to be very satisfactory containers for fused sodium hydroxide. Qualitatively, zinc oxide behaves much like magnesium oxide in the presence of fused sodium hydroxide. D'Ans and Loffler’® were unable to find evidence of a reaction between zinc oxide and anhydrous sodium hydroxide at 600°C. How- ever, oxysalt anions of zinc are known to exist, 17 and it is possible that a reaction occurs at temper- atures above 600°C. D’Ans and Léffler report that water greatly increased the apparent solubility of zinc oxide in sodium hydroxide. This behavior may be analogous to that of magnesium oxide, or it may result from the formation of complex zinc anions involving water.'? In a corrosion test!? compressed and sintered zinc oxide was found to have lost 8% in weight at 538°C. Chemical studies'® on thorium oxide indicate that it does not react with fused sodium hydroxide up to 1000°C. However, no corrosion or solubility data are available. As pointed out above, cerium(lV) oxide, nickel oxide, zirconium oxide, and aluminum oxide are all known to react completely with sodium hy- droxide under suitable conditions, However, they _all have been reported to show considerable re- sistance toward reaction under other conditions. D*Ans and L&ffler!® found that cerium(lV) oxide reacted very slowly with an excess of sodium hydroxide, the interaction beginning first between 950 and 1000°C, but, surprisingly enough, an excess of cerium(lV) oxide reacted more easily at 900°C to give N«JIZCeC)3 and water. ' Nickel oxide was found!3 to be unreactive in fused sodium hydroxide at 400°C, at least for short periods of time. Williams and Miller?? did not find any reaction even at 800°C for a period of 2 hr. However, Mathews, Nauman, and Kruh2? found that at B00°C nickel oxide reacted rapidly with sodium hydroxide to give water and Na,NiQ, according to the equation (3.4) NiO(s) + 2NaOH(l) : = quNioz(s) + Hzo(g) The difference between the results of these two groups of experimenters is not understood. Zirconium oxide was found by D’Ans and Loffler!® to react with excess sodium hydroxide to form Na,ZrO,. However, corrosion tests on stabilized zirconium oxide have shown this substance to be resistant to attack. Craighead, Smith, Phillips, and Jaffee'’ found that stabilized zirconium oxide, fired at 1700°C in either air or argon to give an apparent density of less than 0.5%, was unaffected by exposure to fused sodium hydroxide for 25 hr at 538°C. Stabilized zirconium oxide differs from the pure substance in having o different crystal structure at the temperatures of interest here and in having a few weight per cent of calcium or magnesium oxide in solid solution. However, it is difficult to see how these differences could significantly alter the thermochemistry of the zir- conium oxide—sodium hydroxide reaction, although they may have an effect on the rate of reaction. Aluminum oxide was found by D'Ans and Léffler!® to react with sodium hydroxide to give the oxysalt NuzAlzod. Craighead, Smith, Phillips, and Jaffee!? exposed a single crystal of aluminum oxide to sodium hydroxide at 538°C and reported that the crystal decreased in weight by 0.40% after 24 hr. Simons, Stang, and Logedrost?? used hot-pressed A|203 as a pump bearing submerged in fused sodium hydroxide ot 538°C. After 8 hr of pump operation they were unable to detect any signs of bearing wear or other damage. Tests have also been reported!® for hot-pressed and for sintered aluminum oxide. Weight changes which were observed were less than those for single-crystal material under the same conditions. This sur- prising result may have been caused by abserption of hydroxide into pores of the hot-pressed sample, which compensated for a weight loss. : The relatively good corrosion resistance of aluminum oxide found in these corrosion tests is confirmed by day-to-day experience at ORNL. Vessels made of pure, binder-free, recrystallized aluminum oxide without poresity have been found to be excellent containers for fused sodium hy- droxide under a wide variety of service conditions where a small aluminate contamination could be tolerated. It should be remembered, however, that most commercial alumina and some so-called ““corundum ware’’ have a silica or fluoride binder which is readily ottacked by fused hydroxides. Niobium(V) oxide has been found to react very readily with fused sodium hydroxide. Spitsyn and Lapitskii?3 heated niobium(V) oxide with fused sodium hydroxide for 1-hr periods at temperatures of 350 to 650°C and found the reaction : (3.5) Nb205 + T0NaCH = QNGSNE)OS + 5!’120 Titanium dioxide reacts readily with fused sodium hydroxide at 800°C. D’Ans and Loffler'® studied this reaction and concluded that when an excess of hydroxide was present the reaction was not quanti- tative but that an equilibrium existed, probably between sodium orthotitanaie and a metatitanate. Such an equilibrium might be as follows: (3.6) Na,Ti, 0, + 2NaOH = 2Na,TiO; + H,O In corrosion tests,!? titanium dioxide ceramics sintered at 760°C were found to be very severely attacked by sodium hydroxide at 538°C. Inasmuch os sodium aluminate appears to be stable in fused sodium hydroxide, it might be supposed that the aluminate anion is of itself inert. It has been shown!3 that ceramics with compo- sitions in the neighborhood of MgAl, O, are quite corrosion~resistant in fused sodium hydroxide. This result is not surprising, bacause sodium aluminate has a low solubility and the magnesium ion would tend fo react fo give magnesium oxide, which is very corrosion-resistant, Ceramics with Soturcted Caticns and Accepter Anions. — The saturated cations, as pointed out above, are those of the alkali and alkaline-earth metals. Only a few tests have been made on compounds with cations which are saturated and anions which will accept oxide ions from hydroxyl ions, and they have been limited to the sodium oxysalts with high oxide-ion affinities. Obviously, such substances should not be corresion-resistant. However, an exomination of their behavior will serve to illustrate the kinds of reactions which anions undergo. The chemical composition of a sodium oxysalt can alwsys be expressed in terms of the number of equivalent weights of sodium oxide per equivalent weight of the other oxide of which the oxysalt is staichiometrically composed. Feor convenience, this quantity will be designated as 3. In general, the greater the 2 value of an oxysalt, the less will be the oxide-ion affinity of its anion as compared with other cnions composed of the same elements. Therefore, if an oxide is found to react with fused sodium hydroxide to yield at equilibrium an oxysalt with a value of 2 equal to o and if there are other oxysalts of the oxide in question with values of X less than o, it may be presumed that the other oxysalts will also react with fused sodium hydroxide. This well-known postulate, herein called the 'Y postulate,’’ under- lies all phase-diagram work in which an oxide-ion donor such as sodium carbonate is used as a source of oxide ions. In agreement with this postulate Spitsyn and Lapitskii?® showed that sodium metaniobate(V), like niobium(V) oxide, reacts with sodium hy- droxide to give sodium orthoniobate(VY). This reaction is as follows: (3.7) NaNbO, + 4NaOH = NagNbQ, + 2H,0 Some information is available on two nonmetallic systems of sodium oxysalts, [t has been known for a long time that silicon dioxide, boron oxide, the silicates with smoll & values, the borates of boron{lil) with small 2 values, and the glasses based on these substances nearly all react readily with fused hydroxides to form water and the more alkaline silicates and borates. This reactivity has been utilized for many years in the solubilization of minerals by a process known as alkali fusion, On the other hand, very little information is available on the action of fused hydroxides on the more alkaline silicates and borates, many of which should be relatively unreactive toward fused hydroxides provided that they contain unreactive cations. From a knowledge of the end products of alkali fusions it would be possible to estimate the reactivity of the silicates and borates with high 3 values by application of the 3 postulate. Un- fortunately, there are very few data of this kind. Usually the products of alkali fusions are not examined until ofter they have been hydrolyzed. The results of two experiments are available, however. First, Morey and MerwinZ4 mention, without giving experimental conditions, that sodium pyroborate is stoble in fused sodium hydroxide. Second, D'Ans and Loffler1® found that silicon dioxide reacts with an excess of sodium hydroxide to yield an equilibrium between the pyrosilicate and the orthosilicate as follows: (3.8) Na $i,0, + 2NaOH = 2Na,5i0, + H,0 From these dota it is concluded thot borates with Y less than 2 and silicates with 2 less than 1.5 will react in a similar fashion. This rather terse conclusion can be interpreted in terms of the structural chemistry of the borates and the sili- cates. Oxygen atoms in simple borates and silicates are of twe kinds: those which are bound to two boron or silicon atoms and form so-called ‘‘oxygen bridges" and those which are bound to only cne boron or silicon atom and carry o negative chorge. Thus, the pyroborate ion has one oxygen bridge and may be structurally represented as “O\B/O\B/O“ 1L When an oxide ion reacts with a borate or sili- cate, an oxygen bridge is broken and is replaced by two nonbridge oxygen atoms. Thus, the pyro- borate ion shown above can react with one oxide ion to form two orthoborate ions each with the following structure: -0 o- N I The X postulate implies that the greater the number of oxygen bridges attached to a given boron or silicon atom, the greater the oxide-ion affinity of that atem. Since simple borates and silicates with Y greater than 2 and 1.5, respectively, react with sodium hydroxide, it is concluded that for these compounds a boron or silicon atom with more than one oxygen bridge has sufficient oxide-ion affinity to accept an oxide ion from a hydroxyl ion. | There are ‘other alternatives to the structural interpretation of silicate and borate reactions in fused hydroxides. For example, equilibrium 3.8 is thermodynamically consistent with the as- sumption that, rather than the oxygen bridge in the melt being a physical entity, the pyrosilicate ions (known to exist in solids) dissociafe into orthosilicate and SiG_ 2~ ions. Another alternative is that there is an equilibrium in the melt between 5i,0 6~ on the one hand and :SiOA‘*" and Si03:2“ on the other without either one or the other pre- dominating. The present technique of studying fused electrolytes does not make it possible to determine which reaction takes place, ‘ Oxidation-Reduction Reactions. — Only a few examples have been found of oxidation-reduction reactions of ceramics and ceramic-related sub- stances in fused anhydrous hydroxides. Further- more, all oxides and oxysalts which contain metals capable of existing in more than one valence state in fused hydroxides should undergo oxidation or reduction reqctions, There are very few studies of oxidationreduction reactions of oxides or oxy- salts in fused hydroxides., In the former category the behavior of silicon and carbon, two elements which might be considered ceramics, has been studied. : As shown by LeBlanc and Wey!,?% silicon reacts violently when displacing hydrogen from fused potassium hydroxide at 400°C. The over-all re- action was not determined, but an obvious guess, on the basis of the information in the preceding section, is that the orthosilicote is formed ac- cording to the equation (3.9) Si + 4NoOH = Na,Si0, + 2H, Carbon in the form of massive graphite shows very fittle tendency tfo react :o‘r temperatures up to 500°C but does absorb some hydroxide because of its porosity. However, it has been demonstrated that oxidation of graphite does occur, although net rapidly, at temperatures as low as 500°C. Ketchen and Overholser?® made a careful study of: the action of a large excess of sodium hydroxide on graphite (possible oxidants other than the hy- droxide were excluded) and found that, above 500°C, graphite is oxidized to form sodium car- bonate. At 815°C, graphite has been shown to react quite rapidly.!3 ' D'Ans and L&ffler'6 report that under an atmos- phere of nitrogen, iron(lil) oxide reacts with fused sodium hydroxide to give mostly sodium ferrate(ll1), with a small amount of an oxysalt in a higher valence state, which they suggest may have been sodium ferrate(V1). Corrosion of Other Kinds of Ceramics. — The action of sodium hydroxide on a number of other kinds of ceramics such as carbides has been studied13 in corrosion tests, However, no experi- mental information was obtained which would identify the chemical reactions. Secondary Corrosion Phenomena. — The three causes of the deterioration of ceramic bodies, which will be considered secondary in the sense that they do not reflect the infrinsic chemical reactivity of the primary ceramic constituents, are spalling, binder attack, and pore penetration. It is well known that many ceramics spall when subjected to severe thermal shock. Occasionally, spalling which resulted from the im- mersion of a cold specimen into a hot liquid has been confused with corrosion, however, Many ceramic bodies contain binders which have different chemical properties from those of the principal constituent. Some of the more common binders are among the most reactive materials in fused hydroxides, so that aitack on the binder cavses disintegration of the ceramic body. An example of this is the ottack of fused hydroxides on magnesia ond olumina. The pure oxides MgO and Al,O, are resistont to corrosion by fused hydroxides. However, these substances are fre- quently fabricated with the aid of a high-silica binder, which is strongly attacked. Fused hydroxides show a pronounced capillary activity, ond they very readily penetrate small There is no evidence that such pore penetration will of itself cause disintegration of the ceramic, pores and cracks in ceramic wmaterials, How- ever, ceramic spscimens ars frequently washed with waoter after test to remove adhering hydroxide Under these conditions ahydroxide-impregnated specimen and are exposed to air during examination. will disintegrate, since the hydroxide rapidly forms hydrates or hydrated carbonates with appreciable volume expansien. Such behavior has been ob- served for porous specimens made of pure mag- nesium oxide and pure aluminum oxide. The pronounced capillary activity of fused hy- droxides lenods to other difficulties in corrosion tests, Fused hydroxides wet o very wide variety of substonces, both metallic and nonmetallic, and show a strong tendency to creep. This property has not been observed to be serious near the melting point of the hydroxide, bui at higher temperatures it presenis significant experimental difficulties. Summory, — Studies which have been made of the behavior of ceramic materials in fused hydroxides are limited in scope both as regards the variety of substances tested and the information abtained on any one substance. The corresion of ceramics and ceramic-related materials hos been found to occur both by solution and by chemical reaction. A few substances, among them graphite and silicon, were found to undergo oxidation-reduction reactions, but for the majority of compounds the attack proceeded by an oxide-ion donor-acceptor reaction, Compounds of the alkali and alkaline-earth metals such as oxides, halides, and saturated oxysalts have not been found to react with fused sodium hydroxide but have usually shown ap- preciable solubility, Compounds of other metals with the above anions were found to react by an oxide-ion acceptor-donor mechonism to form an oxide of the metal or to form an oxysalt. Oxides of magnesium, zinc, and thorium have been tested at substantial temperatures without giving evidence of chemical reaction. oxide Magnesium in anhydrous sodium hydroxide but is susceptible to mild attack is very corrosion-resistant by small amounts of water. Zinc oxide is not so corrosion-rgsistant as magnesium oxide. No cor- rosion data are available on thorium oxide, Oxides of cerium(lV), nickel, zirconium, and aluminum have been shown under some conditions to have a significant resistance toward reaction with hydroxyl ions, although they are all known to be capable of reacting by an oxide-ien donor- acceptor mechanism. Under many conditions alu- minum oxide is sufficiently inert to serve as a useful container material for fused hydroxides. Oxides of niobium({V) and titanium react fairly readily with fused sodium hydroxide by an oxide- ion donor-acceptor mechanism at relatively low temperatures., There have been very few studies of compounds containing reactive anions but unreactive cations, In the few examples known, the reactions all proceeded by oxide-ion donor-acceptor mechanisms. lndications are that the oxide-ion affinity of simple borates in fused hydroxides is not satisfied until the pyroborate anion is formed., Simple silicates react by a mechanism like that of the borates, but the end product seems to be an equilibrium between the pyrosilicate and orthosilicate anions. The corrosion resistance of a fabricated ceramic body is dependent on the intrinsic solubility and chemical stability of the primary constituents and is also determined by porosity and the resistance to attack of any binder material it may contain, 4, CORROSION OF METALS Studies have been made of the action of fused hydroxides on at least 31 elemental metals and 65 alloy compositions. The results of these studies show a wide variety of corrosion phe- In the following discussion, only the more common types of corrosion will be treated. These types will be illustrated by the better known examples. Comrosion of Metols by Oxidation. — The oxi- nomend. dation of metal ctoms to form ions is a basic step in the corrosion of metals by fused hydroxides. Oxidation may be coused by the action of hydroxyl ions, alkali metal ions, or foreign substances dissolved in the hydroxide. Corresien by Hydroxyl lons, - Oxidation by hydroxy! ions is the most common form of corrosion of metals in fused hydroxides. the hydroxyl ions are reduced to hydrogen and oxide ions, while the meta! is oxidized to form metal ions. These metal ions may dissolve as In such reactions such or may act as oxide-ion acceptors and form on oxide or an oxysalt, Oxide-ion acceptor re- actions were discussed in the preceding section on ceramics, For a metal M of valence v, the basic step in corrosion by hydroxyl ions may be represented as (4.1) M + OH- = M & LO-= 4 li H, It has been demonstrated that this reaction occurs for a wide variety of metals from the most reactive to the most inert. In a few instances this hydroxy! ion reaction is accompanied by a side reaction which produces a hydride of the metal. If the hydride is volatile, it may escape; thus arsine and stibine are evolved from the reactions of arsenic and antimony, re- spectively, with fused sodium hydroxide.?” If the hydride is not volatife, it may react with the hydroxide as do the alkali metal hydrides. This side reaction does not seem to be important in the corrosion of most metals. Water is formed in a second side reaction which is sometimes observed, The possible processes involved here will be discussed later. Attempts to account for the relative corrosion resistance of different metals foward fused hy- droxides have had only limited success. There is some correlation between the standard free energies of formation of the oxides and corrosion resistance, as has been pointed out by Brasunas,?8 The metals whose oxides are most stable are frequently the least corrosion-resistant and vice versa, Thus calcium with a very stable oxide reacts vigorously with fused hydroxides, while silver with a relatively unstable oxide is fairly corrosion-resistant. This correlation, however, is no more than a rough, general guide. For example, nickel is more corrosion-resistant to fused hy- droxides than gold is, although gold oxide is mueh less stable than nickel oxide. For present purposes, elemental metals will be classified in two groups according to corrosion resistance toward fused sodium hydroxide. The first group consists of those metals which seem to be strongly and irrepressibly attocked by hy- droxyl ions at 500°C under all conditions which have been studied. This group includes alkaline- earth metals, antimony, arsenic, beryllium, cerium, niobium, magnesium, manganese, molybdenum, tan- talum, titanium, tungsten, and vanadium. The second group consists of those metals which show a measure of corrosion resistance toward fused sodium hydroxide under suitable conditions at 500°C. This group includes aluminum, bismuth, chromium, cobolt, copper, gold, indium, iron, lead, nickel, palladium, platinum, silver, and zirconium, The second group is, of course, of more interest, Some of the metals appear to be corrosion-resistant because of protective film formation, while others appear to be corrosion-resistant under suitable conditions because the thermodynamic driving force is small. Aluminum will serve as an example of a metal which seems to be protected by film formation, and nickel is an example of a metal which can be maintained almost in thermodynamic equilibrium with the melt under isothermal con- ditions. Film formation will be considered first. There is no thermodynamic barrier to the reaction with sodium hydroxide no matter whether the reacfion product is aluminum oxide or sodium aluminate or whether the product occurs as a solid phase or is dissolved in the melt. Free- energy chonges which accompany the possible reactions between aluminum and sodium hydroxide have been estimated and were shown to be quite negative. The hydrogen pressure required to reverse these reactions was likewise shown to be much greater than any hydrogen pressures thus far encounfered in corrosion tests. Mevertheless, at [ower temperatures the reaction between aluminum and fused hydroxides proceeds slowly. Craighead, Smith, and Jaffee?? tested 2S5 and high-purity aluminum in fused sodium hydroxide for 24 hr at 538°C. The specimen of 25 aluminum lost 0.337 mg/em? in weight and pitted slightly, while the high-purity. aluminum lost 1.68 mg/cm? but showed no other signs of corrosion on metallographic examination. The origin of the corrosion re- sistance of aluminum is not known, but in view of the corrosion resistance of aluminum oxide (see Sec, 3, ‘‘Corrosion of Ceramics’’) it seems probable that the reaction of hydroxyl ions with aluminum produces at first a thin, pré’fective film of oxide, which then very slowly reacts to form the slightly soluble aluminate. of aluminum Nickel represents a metal which can be held, even at high temperatures, almost at thermodynamic equilibrium with the melt. The corrosion rate of nickel at high temperotures has been found to be dependent on the partial pressure of hydrogen over the melt. ‘Under 1 atm of hydrogen, nickel shows only negligible signs of corresion in fused sodium hydroxide for 100 hr up to 815°C, provided that thermal gradients are absent, as shown by Smith, Steidlitz, and Hoffman.?? However, if the hydrogen atmosphere is not maintained, the corrosion rate of nickel in fused sodium hydroxide may become very rapid at high temperatures. Williams and Miller?® gnd Petersen and Smith!'3 have studied this reaction undeir such conditions that the hy- drogen which formed was removed by high-speed vocuum pumping. It was found that below 400°C the reaction is exceedingly slow but that above 800°C it is exceedingly rapid. The only gaseous product found was hydrogen. The metal formed a sodium nickelate(ll) of uncertain composition but was probably Na,NiO,. This reaction might be represented by (4.2) Ni(s) + 2NaOH(I) = Na,NiO,(s) + H,(g) Williams and Miller2? suggested that the product of the above reaction might be either a compound or a mixture of sodium oxide and nickel oxide. Petersen and Smith'? isolated single crystals of the reaction product and were able to obtain the unit ce!! dimensions by x-ray methods. The results left no doubt that the reaction product which they obtained was a sodium nickelate(l!) compound. In other studies at 950°C the hydrogen which formed was not rapidly evacuated from the system but was removed slowly so that a significant partia! pressure existed over the melt, Under these conditions, and Hannan®! and Mathews, Nauman, and Kruh?! report that con- Peoples, Miller, siderable water vapor was evolved. Mathews et al. found that water and hydrogen were evolved ap- proximately in an equal molar ratio and suggested that the corrosion process consisted in twe steps: the formation of nickel oxide with the evolution of hydrogen, (4.3) Ni + 2NaOH = NiO + Na,0 + H, followed by (4.4) NiO + 2NaOH = Na,NiQ, + H,0 The latter reaction was discussed in Sec. 3. The over-all reaction would be (4.5) Ni + 4NaOH = N\:12Ni02 + NaZO + HZO + H2 Peoples, Miller, and Hannan3! did not check the hydrogen-water correlation, but they, also, sug- gested equations similar to those above as a source of water. An additiona! source of woter was also postu- lated by both Williams and Miller20 and by Peoples, 10 Miller, and Hannan,®! namely, the reduciion of oxide ions by hydrogen when the latter is present Such a reduction of oxide ions would have to be accompanied by a reduciion of positive ions such as nicke! ions either in NiO or in Na2Ni02. There are several in sufficient concentration. kinds of evidence which support the plausibility of this second postulate. It is known from the work of Pray and Miller3? that the sodium nickelate(!!) which is formed from the hydroxide-nickel reaction can be reduced to metallic nickel by a partial pressure of hydrogen of 0.1 atm ot 950°C. In reactions between nickel and sodium hydroxide, particulate nickel is fre- quently found in the hydroxide phase, Mathews, Nauman, and Kruh?! found particulate nickel in the experiments from which they deduced Egs. 4.3 and 4.4, Williams and Miller?® reported that in their experiments there was a definite correlation between the occurrence of particulate nickel in the melt and the occurrence of water in the gas phase and that neither particulate nicke! nor water occurred if the hydrogen evolved was removed rapidly enough for its partial pressure over the melt to remain low, Data on the action of hydroxy! ions on the metals bismuth, chromium, cobalt, copper, gold, indium, iron, lead, palladium, platinum, and silver are very fragmentary, and no attempt at a review will be made here except to point out two facts which are important in evaluating corrosion generally. mechanisms First, Lad3?® has shown thai, when chromium reacts with fused sodium hydroxide, it is oxidized first to a lower valence state and is followed by a very slow oxidation to a higher valence state. Similar data are not available for other metals which have more than one oxidation state in fused hydroxides, but the work of Lad suggests that the reaction kingtics of such metals in fused hy- droxides moy be quite complicated. Second, although the reaction between nickel and sodium hydroxide can be inhibited by hydrogen, Williams and Miller?® have shown that hydrogen has no inhibiting effect on the reaction of iron with sodium hydroxide. Furthermore, they were unable to find any metallic iron resulting from attempts ai hydrogen reduction of iron corrosion products in the presence of fused sodium hydroxide. This result is important in a consideration of the mechanisms of mass transfer in fused hydroxides and will be discussed in Sec. 5. Corrosion by Alkeli Metal lons, -~ The ability of metals such as iron to reduce alkali meta! ions in fused alkali-metal hydroxides has been known for more than a century and was af one time used in the laboratory preparation of the alkali metals. LeBlanc and Weyi” reported that chromium, mo- lybdenum, tungsten, magnesium, and sodium were found to displace potassium metal, as well as hydrogen, from fused potassium hydroxide. Villard®4 found that fused sedium hydroxide was reduced to give sodium or sodium hydride by magnesium, chromium, tungsten, iron, cobalt, nickel, and ferro- manganese. Brasunas?® found that zirconium reduced sodium hydroxide to give sodium metgl and hydrogen. ' Williams and Miller?? reported that nickel acts on fused sodium hydroxide in a two-step process by which hydroxy! ions are reduced and, atter they are consumed, sodium ions are reduced. This sequence of events corresponds to the reaction in which sodium nickelate(ll) is formed and then the sodium ions are reduced in the nickelate, From the standpoint of over-all reactions it might be concluded that (1) mild reducing agents such as nickel act first on hydroxyl ions and when they are reduced to a low cencentration, sedium ions are attacked; (2) very strong reducing agents like zirconium act on both kinds of ions at the same time, However, the experimental findings are also consistent with the postulate that hydroxyl ond sodium ions are simultaneously reduced by all metals. Sodium-ion reduction may be represented as (4.6) Ni o+ 2Ma® = Ni*" 4+ 2Na The sodium thus generated con react with hydroxyl ions as follows: ‘ (4.7) 2Na + 20H- Na® + 207~ 4+ H, It reaction 4.7 proceeds as rapidly as sodium is generafed, the net result is as though only hydroxyl ions hod been reduced. As the concentration of hydroxy! ions decreases, reaction 4.7 will proceed more slowly, untii finally sodium is generated faster than it is consumed. In the case of the very active metals such as zirconium, the rate of gener- ation of sodium may be much more rapid than the rate of reaction 4.7 even for a high concentration of hydroxyl ions. Petersen and Smith'? studied the reduction of sodium hydroxide by sodium and found thot it was not by any meons so rapid as reduction by such active metals as calcium, The action of weok reducing agents such as nickel on alkali metal ions in fused hydroxides would be brought te equilibrivm by a very small concenfration of the alkali metal. Hence, the generation of appreciable quantities of alkali metal is only possible under conditions in which the alkali metal can escape from the reacting mixture. Gold forms an alloy with sodium, so that in the sodiuvm hydroxide—geld reaction some sodium may be removed from the field of reaction by alloy formation, as was pointed out by Williams and Milier.20 |n corrosion tests at temperatures above 800°C smoll amounts of sodium metal have some- times been found on cooler parts of the apparatus above the liguid., Presumably, this metal had distiilled out of the melt. However, under the conditions of most corrosion tests on relatively corrosionresistant metals in fused hydroxides below 800°C, appreciable quantities of alkali metal are never found, Therefore, from the standpoint of the thermodynamics of corrosion, over-all reactions in which alkali metals are produced would seem to be of little importance. Mevertheless, such reactions may be quite significant in the kinetics of corrosion. _ Corrosion by Oxidizing Solutes. — Dissolved oxygen and the peroxide ion are known to be powerful oxidizing agents in fused hydroxides. By using such solutions, Dyer, Borie, and Smith33 were able to cause very rapid corrosion of nickel with the production of oxysalts in a valence state greater than 2, Water at sufficient concentrations seems to act as a weak oxidizing agent. Peoples, Miller, and Hannan®! reported that at 950°C small additions of water to the blanketing atmosphere over fused sodium hydroxide had little effect on the corrosion rate of nickel but that substantial amounts of water made the corrosion much more severe. Comosion of Alloys. ~ Although nickel hos o superior corrosion resistance in fused hydroxides at high temperaiures, it has o low mechanical strength. Consequentiy, studies have been made of the effect of alloying constituents on corrosion behavior with the ultimate aim of obtaining a ma- terial which is both corrosion-resistant and use- fully strong. There is no evidence as yvet that the corrosion of alieys by fused hydroxides involves any kind of chemical reaction which is different from that en- countered with elemental metals. However, an examination of microstructures shows two corrosion i1 phenomena distinctive of alloys: (1) the formation of pores or voids benegath the surfuce of the cor- roded alloy and (2) the fonmation of complex, iwo- phase, corrosion products at the surface and along the grain boundaries of the corroded alloy. Ex- amples of each of these forms of coirosion will be cited below from the work of Smith, Steidlitz, and Hofman, 30 The formation of pores or subsurface voids was found to be characteristic of the attack of fused sodium hydroxide on high-purity nickel-iron alloys. Nickel exposed to fused sodium hydroxide at 815°C for 100 hr in an evecuated capsule was found to be slightly etched, with an average loss in thickness of 4.5 x 1073 in. as computed from weight-change data. lvon under the same conditions was found to lose about 100 times as much in thickness. A high-purity nickel-iron alloy containing 20 wt % iron not only suffered atiack by removal of surface Fig. . metal but also developed subsurface voids to a depth £ 310 5x 1073 in, Subsurface void formation is not an uncommon phenomenon in high-temperature corrosion. and Gront36 Manly proposed a mechanism for this form of coirasion based on the well-established theory of void formation during interdiffusion of metals. Brasunas®’ has discussed this theory and has presented some pertinent experimental data. Quali. tatively, the thecry con be applied to hydroxide corrosion of nickel-iron alloys as follows. During the eorly stages of corrosion, the iron atoms at the surface of the metal specimen are oxidized much more rapidly thon the nickel atoms. As a result, the surface becomes depleted in iron. This deple- tion establishes a concentration gradient in the proper direction to couse a diffusion of iron from the interior of the specimen to the surface, where reaction continues. The diffusion of iron is pre- L Y.sgo1 B Complex Corrosion Product Layer Formed on Surface of Corpenter Compensator 30 Alley (Nomina! Composition 30% Ni, 70% Fe) by Action of Fused NoaOH for 100 hr ot 815°C. 250X. (From the sfudies of Smith, Steidlitz, and Hoffman, ORNL.)} 12 sumed fo occur by a vacani-laitice-site mechanism so that there is a flux of vacancies diffusing from the surface into the interior of the specimen. Since the crystal lattice cannot maintain more than a certain concentration of vacant lattice sites, the excess vacancies, which eccumulate from inward diffusion, “‘precipitate’’ as wvoids. These voids continue to grow as long as the inward Hux is maintained. | Subsurface void formation has been found not only in high-purity nickel-iron alloys but also in high-purity nickel-molybdenum and nickel-iron- moly bdenum alloys with o high nickel content. Pure molybdenum under the some test conditions was severely attacked. As an example of the second kind of cormosion phenomenon characteristic of the corrosion of alloys by fused hydroxides, Fig. 1 shows in cross section a corrosion product layer which formed on a specimen of Carpenter Compensator 30 golloy {nominal composition 30% Ni, 70% Fe) exposed to fused sodium hydroxide for 100 hr at 815°C. The corrosion product loyer consisted of three zanes: (1) an irregular, gray layer of nonmetallic material in contuct with the fused hydroxide; (2) beneath this layer, a zone consisting of a mixture of me- tallic ond nonmetaliic phases; {3) lost, a zone of intergranular penetration. The formation of such complex layers of corrosion products was charace teristic of the corrosion of a number of alloys, in- cluding the commercial nickel-iron-chromium afloys. Very few alloys show all three zones seen in Fig. 1. Usually the zone of massive nonmetallic material is absent, and, under suitable conditi;ons, one of the ather two zenes may be absent. Figure 2 shows the corrasion preduct formed on Timken 35-15 alley {neminal composition 35% Ni, 15% Cr, 50% Fe) after exposure to fused sodium Fig. 2. Corrosion Product Layer Formed on Timken 35-15 Alloy (Mominal Composition 35% Ni, 15% Cr, 50% Fe) by the Action of Fused NaOH for 100 hr ot 815°C. 100X. (From the studies of Smith, Steidlitz, and Hoffman, ORNL) ' 13 hydroxide for 100 hr at 815°C. Here, only the zone consisting of a metallic and nonmetallic phase is to be seen, with a rudimentary grein boundary at- tack appearing at the comosion product-base alloy interface. Of the various metals which form complex cor- rosion-product layers in fused hydroxides, only Incone! has been extensively studied. Results with this metal illusirate the structural complexity of the layers. Most of the tests have been con- ducied for various times up to 100 hi, at various temperctures up to 800°C, and under blanketing atmospheres of purified helium and of purified hy- drogen at 1 atm. The results obtained for tests under helium were quite different from those ob- tained for tests under hydrogen with regard to both corrosion rate ond microstructure of the corrosion product, Under a blonketing atmosphere of hydrogen, lncone! showed a slight amount of corrosion ofter 100 hr at 600°C. From 600 to 800°C the corosion increased very ropidly. At all temperaivies cor- rosion begon as grain boundory attack, an example of which is shown in Fig., 3. When thess grain boundary regions were examined at a high magnifi- cation, it was found that a comrosion prodyct had formed in these regions and that this corrosion product consisted of a metallic matrix within which were acicular particles of a nonmetallic phase. After groin boundary attack had advonced a few thousandths of an inch into the alloy, a massive, twe-phase, corrosion product layer had begun to form on the surface and to thicken with time, Fige- ure 4 shows the microstructure of this two-phase layer at a high magnification. The corrosion rate of Inconel at a given tempera- ture was significantly greater under o blanketing atmosphere of helium than under hydrogen. A mas- Fig. 3. Grain Boundary Attack at the Surfoce of an Incone! Specimen Exposed to Fused NoOH for 100 hr ot §00°C. 250X. (From the studies of Smith, Steidlitz, and Hoffman, ORNL) 14 Fig. 4. Microstructure Typi;ul of the Corrosion Produet Layer Formed on lncanellby Exposure to Fused NaDH Under ¢ Blanketing Atmosphere of Hydrogen for Periods of up to 100 hr ot Temperatures above 650°C. 1000X. (From the studies. of Smith, Steidlitz, and Hoffman, ORNL) sive, two-phase zone also formed under helium, but the microstructure was quite different, as can be seen in Fig. 5. Other, quite different, two-phase, corrosion prod- uvct microstructures were observed when Inconel was corroded by fused sodium hydroxide to which various inorgonic compounds had been added. - In oll cases the layer of corrosion product consisted of various geometric arrangements of nonmemlhc particles within a metallic matrix. A study is now being made to determine the chemical nature of the phases which occur in the corrosion product layers on Inconel. It is as yet too early for any firm conclusions to be drawn other than those that the metallic matrix is depleted in chromium and that the nonmetallic particles have o relatively high content of a sodium oxysalt. _ Summary. — Studies have been made of the action of fused hydroxides on at least 31 elemental metals and 65 alloy compositions. Most of the results can be interpreted in terms of the oxidation of the metal accompanied by the reduction of hydroxyl ions to form hydrogen and oxide ions. The oxidized metal may occur either as an oxide or ¢s an oxy- salt. Three complicating side reactions are some- times found: the formation of a metal hydride, the reduction of alkali metal ions, and the produchon of water. Film formation is apporently important in the corrosion resistance of niummum in fused sodium hydroxide. _ The corrosion of nickel by fused sodium hydrox- ide under isothermal conditions con be stopped almost altogether at temperatures as high as 815°C by the application of a pressure of hydrogen of 1 gtm over the melt, 15 o B A Fig. 5. Micrestructure Typical of the Corrosion Product Layer Formed on Incone! by Exposure to Fused NMaOH Under a Blanketing Atmosphere of Helium for Periods of up to 100 hr ot Temperatures above 650°C, The Cu plate was applied after test to protect the edge of the specimen during metallographic polishing. 1000X. (From the studies of Smith, Steidlitz, and Hoffman, ORNL) The chemistry of the reaction between nickel and sodium hydroxide has been extensively investi- gated but is still not well understood. For all metals other than nicke!, details on the chemistry of corrosion reactions in fused hydroxides are very fragmentary. The corrosion of alloys by fused sodium hydrox- ide does not seem to involve any kind of reaciion which is different from the reactions encountered with elemental metals. However, the microstiuc- tures of commoded alloys show the distinctive cor- rosion phenomena of the formation of subsurface voids and the formation of complex, two-phase cor- rosion products at the surface of the alloy and along grain boundaries near the surface. 16 5. MASS TRANSFER When liquids are circulated through metal plumb- ing systems at high temperatures, it is often found that the hot parts of the plumbing are corroded and that the metal thus removed is deposited in the cool parts of the system. This corrosion-related phenomencn, usually called “mass transfer,”” is a particularly serious problem with fused hydroxides. Mass transfer has been observed in sodium hydrox- ide for nickel, iron, copper, silver, gold, and a number of alloys. Moreover, nickel has been ob- served to masseiransfer in all the fused alkali- metal hydroxides and, it might be mentioned, in all the fused alkaline-earth hydroxides.®® Mass transfer is the primary factor limiting the use of metals which, under suitable conditions, do not corrode seriously. The mechanisms of mass transfer in fused hy- droxides are not known. However, several pos- sibilities exist, and they will be discussed at the end of this section. First, however, some of the empirical facts of mass transfer will be presented. Nickel. « Some of the characteristics of mass transfer of nickel in fused hydroxides are illys- trated with the results of thermal-convection loop tests.3837 Figure 6 is a photograph of a nickel thermal-convection loop which was cut open to show the appearance of the inside wall surfaces after test. During test the loop was filled with fused sodium hydroxide and placed in an upright position. One side of the loop was heated and the other cooled to maintain the temperature distribuy- tion shown in Fig. 6. The highest temperature measured was 825°C, near the bottom of the hot leg (left side in Fig. 6), and the lowest tempera- ture measured was 535°C, near the bottom of the cold leg (right side in Fig. 6). This temperature distribution caused a flow of fused hydroxide around the loop because of thermal convection. The numbers in Fig. 6 from 5 to 45, at infervals of 5, identify positions around the loop. Before the loop was exposed to the hydroxide, the inside walls were somewhat rough and had a dull appearance. After test, as seen in Fig. 6, the hot leg from position 25 to position 45 became polished, whereas the cold leg from position 5 to position 20 became encrusted with a deposit of dendritic nickel crystals. Grain boundary grooving occurred in the polished zone, but the groove angle was wide and there was no serious grain boundary attack. The polishing observed in the hot leg is characteristic of the corrasion step of mass trans- fer at high temperatures in a number of different kinds of systems, including some containing liquid metals. : Figure 7 shows a deposit of dendritic nickel crystals from a cold-leg section of a loop with the solidified hydroxide intact. In the initial stages of deposition it was found that the crystals grew as a dense deposit, forming an almost continuous plate over the cold-leg surface. The crystal grains in this plate frequently continued the orientation of the grains in the base metal,: which necessitated the use of special techniques for metallographic detemminations of the place at which the base metal ended and the deposit began. When this dense plate became several thousandths of an inch thick, LNCLASSIFIEDY 8 Y.5623 CHaN LA | Fig. 6. Nickel Thermol-Convection Loop Cout Open to Show Appearance of inside Wall Surfaces After Test. The numbers 5 through 45 identify positions around the loop. * (From the studies of Smith, Cathcart, and Bridges, ORNL) very long dendrites began to appear. Because of the dendritic form of the thicker deposits, a rela- tively small amount of nickel was found to have a large effect in restricting flow through a pipe. When the flow of liquid was sufficiently rapid, dendrites become dislodged. and they collected in irregular masses which formed plugs that effec. tively stopped fluid flow altogether. Figure 8 shows a section of pipe from a large loop which contained several such plugs, and Fig. 2 shows a 17 UNCL ASSIFIED Y.5543 Fig. 7. Section of Tubing Removed from the Coo! Parts of a Nickel Thermal-Convection Loop and Cut Open After Test with the Frozen NaOH Intact, Dendritic nickel crystals may be seen attached to the inside walls of the tube. (From the studies of Smith, Cathcart, and Bridges, ORNL) [ UNCLASSIFIEDR Ev.4383 | i h Fig. 8 Section of Nickel Pipe Cu? from the Cocl Portions of a Thermal-Convection Loop. Irregular masses of nickel dendrites which stopped the flow of fused NaOH in the pipe may be seen. (From the studies of G. M. Adamson, ORNL) 18 Fig. 9. Plug Formed of Nickel Dendrites Taken from the Section of Nickel Pipe Shown in Fig. 8, The plug was about 1 in. across. (From the studies of G. M. Adamson, GRNL) single plug removed from the loop. Because the corrosion accompanying mass frans- fer tended to occur uniformly, while deposition produced long dendrites, the consiriction of fHlow in the cool parts of the system usually reached seri- ous proportions long before the corrosion in the hot parts of the system became troublesome, Nearly all studies of mass transfer in fused hy- droxides have been carried out in thermal-convec- tion systems, although some forced-circulation work has been reported.?® Several different de- signs of thermal-convection apparatus and several different methods of measuring mass transfer have been used. The only extensive studies in which numerical valves of the amount of mass transfer were obtained were the work of Lod and Simon49 on the mass tronsfer of nickel in sodium hydroxide and the work of Smith, Steidlitz, and Hoffman3? on several metals and alloys in sodium hydroxide. These two groups used not only different designs of thermal-convection apparatus but also different means of measuring mass transfer, Lad and Simon measured mass transfer by determining the weight loss of a specimen in the hottest part of the sys- tem, while Smith, Steidlitz, and Hoffman micro- scopically measured the thickness of deposits formed in the coldest parts of the system. Never- theless, confirmatory resuits were obtained when the two studies overlapped. Nickel, because of its excellent corrosion re- sistance, has received more attention than other metals in mass-transfer research. Studies have been made of the effect of several variables on the rate of mass transfer of nickel in fused sodium hydroxide. These variables were temperature level of the system, temperature differential, time of op- eration, composition of the atmosphere over the melt, and composition of the melt itself. _ Lad and Simon?9 ond Smith, Steidlitz, and Hoff- man3® found that the rate of mass transfer accel- erated with linearly increasing temperature level, Smith, Steidlitz, and Hoffman found that mass transfer of nickel in fused sodium hydroxide under helium could be detected by their techniques after 100 hr at a maximum system temperature of 600°C and a temperature differential of 100°C but not at a maximum system temperature of 550°C, other conditions being constant, The lowest maximum system temperature for which Lad and Simon re- ported data was 538°C, at which temperature under helium with a temperature differential of 46°C they found a very small amount of mass transfer after about 100 hr. , The studies by Lad and Simon showed that, under helium, mass troansfer of nickel was de- pendent upon the time of operation and the tempera- ture differential. They found that the rate of trans- fer decreased slightly over the first 50 hr and thereafter remained constant, The initial tronsient in the rote agreed with the change of nickel con- centration ' in the melt; both rates built up to a steady-state value after 100 hr. The rate of trans- fer increased rapidly with temperature differential and then became independent of the differential at higher values. Lad ond Simon suggested that at higher temperature differentials the rote of flow in their apparatus was too great to permit complete safuration of the fluid with nickel compounds. Because of the ability of hydrogen to suppress corrosion it might be expected that this element would be effective in inhibiting mass transfer, This inhibiting effect was demonstrated by Smith, Cathcart, and Bridges?? in 1951, but only recently has it been subjected to extensive quantitative test. Lad and Simon found that at 815°C the inhibiting action of hydrogen on the mass transfer 19 of nicke! was considerable, Smith, Steidlitz, ond Hoffman confirmed this result for temperciures below 815°C and found that the maximum system temperature at which no mass transfer was meas. urable was about 75°C above that found under helium in comparable tests. In addition, Lod and Simon found that 3% water vapor in the blanketing atmosphere is beneficial with and without hydrogen at 815°C, L.ad and Simon?? studied the effect on mass transfer at 815°C of additions to the melt of up to 5 wt % of 33 substances. Many of the materials which they odded react with sodium hydroxide at this temperature; some are soluble, while some are insoluble ond relatively inert. Sodium car- bonate, commercial sodium hydroxide, was found to have no effect for additions up to 1 wt %. Above this concentration, sodium carbonate was found to accelerate mass transfer. Sodium chloride and sodium orthophos- phate, both soluble, had no effect. Sodium and lithium hydrides had very detrimental effects. Lad a common copntaminantd in and Simoen suggested that these compounds react to form sodium oxide which is ‘‘very corrosive.” FHowever, tests made by adding sodium oxide, as such, caused only about half as much mass iransfer as was caused by the addition of amounts of hy- dride which would give roughly the some oxide concentration after reaction. Other Elemental Metals. = Data on elemental metals other than nickel are meager. The moss transfer of iron in fused sodium hydroxide hes been found to be wore severe than that of nickel under comparable conditions. [or example, Smith, Steid- litz, and Hoffman3? found that under hydrogen at 55%°C the noss transfer of iron was worse than thai of nickel at B00°C, other conditions being constant. However, iron was not observed to form the long dendrites thot werz found for nickel. Rather, the crystals in ihe iron deposits remained somewhat equiaxed in habit. Copper in sodium hydroxide under hydrogen was found®® to mass-iransfer appreciably in 100 hr at 600°C with a temperature differential of 100°C. Thus, copper mass-transfers more readily thon nickel under these conditions but not so severely as iroi. The mass transfer of silver and of gold has not been quantitatively studied, but it has been ob- served repeotedly. Alloys. — Hastelloy B ond the stainless steels have been observed3? to undergo mass transfer in 20 fused sodium hydroxide under hydrogen, but at elevaied temperatures coifosion is o serious prob- lem with these metals whethar mass transfer occurs or not. Monel also showed3? mass transfer under both hydrogen and helivm. The most extensive studies of mass transfer of an alloy have been those made on Inconel,30 |t was f{ound that Incone! showed corrosion in fused sodium hydroxide at roughly the same rate whether mass transfer took pluce or not and that mass transfer took place at roughly the same rate as for Mass- transfer deposits formed in Inconel systems con- taining fused sodium hydroxide under on atmosphere pure nickel under the same conditions. of helium were virtually deveid of chromium. Hy- drogen suppressed mass transfer of Inconel with about the same effectiveness as it did for pure nickel. Bimetallic EHects, — When two different solid metals are immersed in o liquid — for example, a liquid metal — under isothermal conditions at high temperatures, it is frequently observed that one of the solid metals contominates the other metal or that the metals contaminaote each other. This process of metal transport without a temperature differential is usually referred to os “‘isothermal mass transfer,”*' |n the case of liguid-metal systems, the driving force for isothermal mass transfer seems to be the tendency of the two solid metals to form alloys, A number of experiments have been conducted in which two different metallic elements were im- mersed in the some hydroxide melt under pre- sumably isothermal conditions. LeBlanc and Bergmann4? performed o series of experiments in which different metals were immersed in fused sodium hydroxide contained in a gold crucible at 700°C wunder nitrogen., When copper was immersed in the melt, they found that the gold crucible be- came alloyed with copper; when silver waz im- mersed, the silver specimen betcome alloyed with gold; but when nicke!l was immersed, nothing was observed to happen. Brasunas?® heated sodium hydroxide in an evacuated nickel capsule together with a copper specimen at 815°C, He reported that nickel plated out on the copper. [t would be tempting to conclude that the in- stances of metal transport cited above are anale- gous to isothermal mass transfer in liquid metal systems if it were not for some studies by Williams and Miller.?® They immersed strips of nickel in fused sodium hydroxide contained in a gold vessel under a hydrogen atmosphere and found that the nickel became gold-plated. However, they gave convincing evidence that this plating occurred anly during cooling. This observation shows that the transfer of gold to nickel which was observed was not an example of isothermal mass transfer, al. though Williams and Milles's techniques would not have detected small amounts of alloying which might have token place in addition to the much grosser plating effect, Craighead, Smith, and Jaffee?? report the trans- fer of nickel from a container vessel onto o number of metals immersed in sodium hydroxide at 677 and 815°C., However, they point out that thermal gradi- ents, which were known to exist, could have ex- plained their results, Mechanism of Mass Transfer. ~ The mechanism of mass transfer of metals in fused hydroxides is not known nor do the proper kinds of data exist which are necessary to distinguish omong several possible mechanisms., The pumose of the foilows. ing discussion isto point ouf the various processes by which a chemically reactive fused electrolyte such as sodium hydroxide might transport metals under the influence of a temperature differential. These mechanisms are all possible in the sense that they do not violate known principles. ever, they are all quite speculative. It is assumed that the mode of metal transport involves either solution of the container material as metal atoms or its solution as metal ions plus electrons. How- If the solute consisis of metal atoms, mass tronsfer could be considered in tenns of dif- ferential solubility. If the solute consists of metal ions plus electrons, several possibilities exist, depending on the mechonism of transport ot the electrons. These various possibilities will be discussed in the following sections, Difterential Solubility (Mechanism ). - In a superficial way the mass transfer of metals in tused hydroxides appears to obey a simple solu- bility-temperature relation. Mass transfer in liquid- metal systems usually follows such a relation, and it can also, in principle, be applied to fused hy- droxide systems. Every metal must have a finite solubility in a fused hydroxide, just as every metadl must have a finite vapor pressure at all tempera. tures, although this solubility may be so small as to have only a statistical meaning., Symbolically this process may be represented as (5.1) M(s) = M{melt) It is postulated here, however, that metals have a negligible solubility as metal atoms in fused hydroxides, This postulate is difficult to justify Nevertheless, the differences which exist in the intemal pressures and in the binding forces for metals and tused hydroxides strongly suggest that the solubility of the metal as atoms in the tused hydroxide will be exceed- ingly small. by formal argument, There are no direct experimental studies of the solubility of any metals in fused hydroxides, There is indirect evidence!3+18 that sodium metal has a significent solubility in fused sodium hy- droxide and that it dissolves before reacting. This phenomenon appears to be analogous to the solu- tion of a metal in its fused halide.¥? This process, however, is not simply o case of the solution of metal otoms but is dependent on some intferaction between the valence electrons of the metal atoms ond the ions of the same metal in the melt. The dissolving of sodium in a sodium halide and possibly in sodium hydroxide can be schematically represented by the equation (5.2) Nalg) = Nat(melt) + O(melt) where the electrons B(melt) may be thought of as 8 ¥ &8 sk " excess’’ or "'solvated’’ electrons which are more or less associated with the cations of the melt. In some instances the dissolution of a metal in its fused halide seems fo involve the formation of o Cd2++ is formed when cadmium is dissolved in cadmium chioride. diatemic cation; for example, It is possible, in principle, for any metal to dis- solve to some extent in a fused electrolyte as metal ions plus excess electrons which may be con- sidered as more or less associated with the cations of the electrolyte. This process will be discussed further in the section “Mechanism IV."” Oxidation-Reduction Processes {Mechanisms {i- V¥1). ~ The mechanisms considered most probable for mass transfer in fused hydroxides all involve oxidation-reduction processes. The two basicsteps for all such processes may be schematically repre- sented for a metal M of valence v by the following equations., At the hot metal-melt interfaces there occurs the oxidation h (5.3) M(s) s M¥(melt) + 10 where 0 indicates an electron and where M¥(melt) represents a metal in whatever form in which it may be stable in the melt, that is, as an uncomplexed 21 cation or as a suitable oxysalt anion. At the cool metal-melt interfaces thers accurs the reduction ol (5.4) M¥(melt) + v0 ——> M(s) which is the reverse of Eq, 5.3, For all kinds of oxidation-reduction mechanisms the M¥{(melt) ions will be transported by diffusion through boundary layers at the interfaces and by flyid flow within the bulk of the melt, move by two quite different routes. However, the electrons may They may be conducted through the metallic ports of the system, or they may be transported through the melt by some chemical species which is reduced at the hot metal surfaces and oxidized at the cool surfaces. The first method of electron transport represents The second method might be effected by any of a number of chemical species. local cell action. It is convenient to classify these species into four groups according to the oxidized form of the elec- tron carrier: hydroxyl ions, sodium ions, foreign solutes, and a higher valence state of the metal undergoing transport. These five modes of electron transport may be considered as representing five additional mass-transfer mechanisms. Local Celi Action {(Mechanism ), ~ !f the elec- frons are transported fhrough the metallic parts of the system, Fq. 5.3 represents the anodic dissolu- tion of M, and £q. 5.4 the cathodic deposition of M. The driving force for this process con be thought of as o thermoelectric potential. Such potentials have been measured in fused electrolytes.?? order for such a process to tcke place at all, the transported ions M*{(melt) must be already present, However, any oxidizing substonce, including the hydroxide itself, could serve this purpose. lnasmuch as Pray and Miller?2 have induced In nickel to deposit preferentially on on Alundum ir @ sodium hydroxide melt, local cell action connot be the exclusive mode of mass ring immerse transfer. However, there is no reason why local cell action might not be the primary mechanism under selected conditions. Reduction of Hydroxyl lons {(Mechanism ). -~ In Sec. 4, *‘Corrosion of Metals,”’ it was pointed out that the most common over-c!l corrosion reac- tion was the reduction of hydroxyl ions represented by £q. 5.5 (5.5) M(s) + vOH~{ms!lt) = MY(melt) + 107" (mel) 4 = Hyfme) / e . — P e e 22 and that, at least in the case of. nicke!, MY{melt) ions may be reduced to the metal again by hydro- gen. It would not be surprising, therefore, to find that hydrogen molecules serve to transport elec- trons, the hydrogen acting together with an oxide ion as an electron donor conjugate to a hydroxyl ion.” This mechanism was proposed independently by Skinner?> in 1951 and by Williams and Miller in 1952. Since then it has received wide ccceptance as the only mechanism for the mass transfer of, nickel, | There is no doubt that £q. 5,5 represents a likely mechanism of mass transfer under many conditions. However, the evidence in favor of this mechonism is ambiguous, and mass transfer has been observed to occur under conditions where the operation of this mechanisn seems doubtful. The evidence most frequently cited in its support is the effectiveness of a blanketing atmos phere of hydrogen in suppress- However, regardless of the nature of the electron carriers, hydrogen should, to ing mass transfer. a greater or lesser extent, suppress all oxidation processes in fused hydroxides and thereby depress the concentration of nickel ions moving through the fluid. Furthermore, Smith, Catheart, and Bridges3? have conducted severa! kinds of experiments in which nickel preferentially deposited on a surface at which the hydroxide was saturated with atmos- pheric oxygen. In these experiments mass transfer cannot be accounted for by Eq. 5.5, According to this equation, hydrogen must be present ot the interface of deposition ot a concentration greater thon the minimum necessary to reduce M¥(melt) ions. In the experiments of Smith, Cathcart, ond Bridges, this condition could not have heen ful- filled. only according to Eq. 5.5, the absesnce of hydrogen at the hot surfaces would occelerate the solution step, but an absence of hydiogen at the cool sur- faces would likewise prevent the deposition step. The quontitative specification of the hydrogen concentrations or, via Henry's law, of the hydragen In other words, if mass transfer occurred pressures necessary to cause mass fransfer under given circumstances is dependent, omong other things, on the temperatures and nickel concentra- tions at both hot and cold interfaces, Data are not yet available with which to make such specifica- tions, but outstanding progress is being made toward this end for the case of nickel by Kertesz, Knox, and Grimes. 48 Iron has been shown3? to undergo mass transfer in sodium hydroxide much more rapidly than nicke! did under the same conditions. MHowever, Williams and Miller?® have pointed out that the reaction be- tween iron and sodium hydroxide is not reversed or even inhibited by hydrogen at pressures up to 1atm. It is, therefore, doubtful that hydrogen is an im- portant electron carrier in this instonce, Reduction of Alkali Metal fons (Mechanism iV), ~ |t does not seem likely thor aikaii metal ions would be effective agents in mass transfer under most conditions, because they are thermo- dynamically inferior oxidizing agents compared with other possible species such as hydroxyl ions. On the other hand, alkali metal jons and alkali metal atoms could, et least in principle, act os conjugate electron acceptor-donor pairs for slectron transport in fused hydroxides. Therefore, this possible mechanism will be discussed briefly, Every metai must be capable of eflecting the reduction of a finite number of sodiym ions in fused sodium hydroxide. Hence, if is possible for mass transfer to occur by the following process: (5.6) M(s) + vNa*(melt) = M¥(melt) + E/Na(melt) where Na(melt) is to be censidered as sodium ions plus excess electrons, as discussed under Mecha- nism |. However, sedium dissolved in fused sodium hydroxide will react with hydroxyl ions: (57) Nalmelt) + OH™(melt) = Na*(melt) + O~ (melt) + ]i.ij, Hz(g) The reverse of this reaction is also known,?’ so that Eq. 5.7 may be tdaken as representing an equilibrium. (For reasons of simplicity, the inter- mediate equilibrium involving hydride ions is omitted.}) Therefore, it local thermodynamic equili- brium exists in the melt, beth Eqgs. 5.6 and 5.7 must be simultaneously satisfied, and the over-all equilibrium {the summation of Egs. 5,8 and 57) becomes identical with that represented by £q. 5.5. Therefore, if {ocal thermodynamic equilibrium exists, Mechanism IV is the same as Mechanism lii; that is, the same end result is achieved if it is assumed that either the sodium ions or the hydroxyl ions ure the electron acceptors. On the cther hond, the reaction between sodivm metal and sodium hy- droxide has been found to be surprisingly slow,*® and it is pessible that Na(melt) is not in local thermodynamic equilibrium according to Eq. 5.7. Under this latter circumstance, the kinetics of Mechanism 1V could be quite ditferent from the kinetics of Mechanism [, Reduction of dolutes {Mechanism V). -« In some instonces dissolved substances which are more easily reduced than the fused hydroxide may serve as electron goceptors in the mass-transfer mecha- nism, Such a substonce mighi be the ions of anothier metal, A, with two avoilable valence states yand y — 1. Then a reaction such as ' (5.8) M(s) + vAY(melt) = MY(melt) + vAY" melt) would be expected to ovcur, Displacement of this equilibrium could couse mass transfer. This mech- anism will not account for the moss transfer ob- served in most of the systems investigoted, since suitably high concentrations of anions A” do not usunlly exist, However, if such ions are present, it is possible that mass transfer would be acceler- ated. One possible source of ions of A is from impuri- ties in the original sample of metal undergoing mass fransfer. Willioms ond Miller?? have shown that iron is selectively lenched from the surfaces of samples of commercial nickel which originally corntained 0.09 to 0.41 % iron as an impurity. Pisproportionation {(Mechonism V1), - If the metal M exists in more than one valence state, it is pos- sible for disproportionation te cause mass transfer, For example, with M having the valence states v and v + 1, ‘;thez} . (5.9) M(s) + ;/gfij;flfmeh) - (v + a}M':"(mem For this reoction to occur, ions of M must be pres- ent in the melt, os wos true for the local cell mechanism, Any of the oxidation mechanisms just described could generate them. The possibility that disproportionation might serve as a mechanism for moss tronsfer in fused hydroxides was first nointed out by Grimes and Hill,*? It was noted above thot hydrogen is an unlikely electron carrier in the mass transfer of iron. How- ever, 1 cool metal) 2. Electrons transported by the reduced form of some species in the melt a. Reduced form of hydroxy! ions — Mechanism |1} 24 10 + YOH {melt) ';’f:voflm(meh) ’}'"21{' H2(me|t) Over-all reaction v M(s) + POH ™ (melt) T== M*(melt) + O™ "(melr) + 5 Hylmelt) Reduced form of sodium ions - Mechanism IV 10 + vNa'(melt) =" vNa(me!1) Over-all reaction M(s) + ¥Na (melt) T==M*(melt) + vNa(melt) Reduced form of a solute A”(melt) — Mechanism V v + vAY (melt) == vA” *(melt) Over-all reaction M(s) + vAY (melt) == M"(melt) + vAY " Nmel) Disproportionation — Mechanism VI 10 + M7 melt) == vM¥(meln) Over-a!l reaction M(s) + M7 melt) == (v + 1) M¥(melt) (See Eq. (See Eq. (Sze Eq. (See Eq. 5.5) 5.6) 5.8) 5.9 Mechanism 1, transport of metal atoms as such, was discounted inasmuch as the solubility of metals as atoms in fused hydroxides would be ex- pected to be very small. The remaining mechanisms, {I through VI, are all oxidation-reduction processes. No single cne of these mechanisms will aceount for all observations of metal transport in fused hydroxide media. It seems that no one process is always the exclusive mode of transport but rather that different mecha- nisms predominate under different conditions. Much further research will be needed before specific modes of metal transport can be proposed as the probable mechanisms of mass fransfer under stipu- lated conditions. Two substances, hydregen and water, have been noted to suppress mass transfer of nickel in fused This cited in support of any particular one of the oxida- tion-reduction mass-transfer mechanisms. Hydrogen should suppressthe corresion step for all oxidation- reduction mechanisms, sodium hydroxide. information cannot be The role of water in moss transfer is unknown, It is possible that water functions to suppress the oxide ion concentration in the melt and thereby causes the corrosion steps to produce a film of relatively insoluble nickel oxide. The action of water might be expressed by an equilibrium such as (5.10) NaNiO,(meit) + H,O(melt) = NiO(s) + ZNaOH{melt) This film could function as o diffusion barrier to slow down the corrosion step. If water were fo act in some way such os this, it would be effective in suppressing mass transfer by olmost any mecha- nism, _ ' Symmary. - Moss fronsfer has been observed in fused sedium hydroxide for nicksel, iron, copper, silver, and a number of alloys. Moreover, nickel has been chserved fo mass-transfer in all the fused alkali-metal ond olkaline-earth hydroxides. In the majority of investigations undertaken, mass transfer wos induced by a temperature differential in the system, The rate of the mass transfer which takes place under the influence of a temperature differential was found to be affected by the following variobles: nature of the metal undergoing transfer, composition of the melt, composition of the atmosphere over the melt, temperoture level of the system, temperature differential within the system, and geometry of the Undoubtedly, the rate of fluid flow is @ very important factor, but existing dota are in- adequate for evaluation of this variable. system. Under ¢ particular set of comparable conditions, was found faster than nickel, and iron to mass-iransfer faster than copoer. copoper to mass-transter With alloy systems, mass transfer and corrosion were found to occur simultaneously. ' The mechonisms of mass transfer in fused hy- droxides are not known. Some of the possible modes of mass fransfer were exomined, and five mechanisms were proposed as being possible. It was suggested that no one process is always the exclusive ‘mode of transport but that different mechanisms predominate under different conditions. 25 REFERENCES 1. W. R. Grimes, D. R. Cuneo, and F. ¥, Blankenship, The Reactor Handbook {ed. by J. F. Hogerton and R. C, Gross), Vol. 2, Sec. 6, AECD-3646 (May 1955). 2. J. Spangler, Z. Ver deut. Ing. 87, 356 (1943). 3. A. M. Weinberg, Sci. American 191(6), 33 (1954). 4. M. E. Steidlitz and G. P. Smith, unpubli shed research, Quk Ridge National L aboratory (1955). 5. C. Hochanadel, [. Am. Chem. Soc. 76, 2675 (1954}. 6. G. W, Keilholtz et al, ORNL-1261 (Sept. 25, 1952); cited in ref 1, p 850, 7. K. Amdt and G. Ploetz, Z. physik. Chem. 110, 237 (1924). B. R. P, Seward, J. Am. Chem. Soc. 64, 1053 (1942); 77, 5507 {1955). 9. M. Temkin, Acta Physicochim. U.R.5.5, 20, 411 (1945). 10. A. F. Wells, Structural Inorganic Chemistry, p 350 ff, Clarendon Press, Oxford, 1945, 11. A. Dietzel, Glastech. Ber. 22, 41{1948); ond subsequent papers. 12. H. Flood and T. Férland, Acta Chem. Scand. 1, 592 (1947); ond subsequent papers. 13. G. F. Petersen and G. P. Saith, unpublished research, Ook Ridge MNational Laboratory (1955). 14. G. Scarpa, Atti accad. nazl, Lincei, Rend. 24(1), 738, 955 (1915); 24(2}, 476 (1913). 15. C. M. Craighead, L. A. Smith, E. C. Phillips, and R. |. Jaffee, Continued Studies of Corro- sion by Fused Caustic, BMI-794 (Dec. 1952) (declussified). 16. J. D’Ans and J. Léffler, Ber deut. chem. Ges. 63B, 1446 (1930). 17. G. Woltersdorf, Z. anorg. Chem. 252, 126 (1943). 18. C. R. Boston, private communication, Qak Ridge National Lokoratory (1955). 19. R. Scholder and H. Weber, Z. wnorg. allgem. Chem. 215, 355 (1933). 20. D. D. Williams and R. R. Miller, Thermal and Related Physical Properties of Molten Ma- terials, Part I, High Tempemiture Reactions of Sodwm Hydroxide, WADC TR-54-185 (Feb. 1955). 21. D. M. Mathews, R. V. Nauman, and R. F, Kruh, private communication, University of Arkan- sas {1955). 22. E. M. Simons, J. H. Stang, and J. . Lagedrost, cited in ref 15, 23. V. L Spitsyn and A, V. Lapitskii, Zbur. Prikiad. Khim. 26, 117 {(1953). 24. G. W. Morey and H. E. Merwin, J. Am, Chem., Soc. 58, 2248 (1936). The compound which these authors designated as “orthoborate” is now known to be the pyroborate. 25, M. LeBlanc and O. Weyl, Ber. deut. chem. Ges. 45, 2300 (1912). 26. E. E. Ketchen and L. G. Ovetholser, private communication, Oak Ridge National Laboratory {1954). 27. B A. Nekleevich, Khim. Referat. Zhur 4, 26, 27 {1941). 28. A. deS. Brasunas, pri vate communication, Qak Ridge National Laboratory (1951). 29. C. M. Craighead, L. A, Smith, and R. |. Joffee, Screening Tests on Metals and Alloys in Contact with Sodium Hydroxide at 1000-and 1500°F, BMI-706 (Nov. 1951) (declassified). 30. G. P. Smith, M. E. Steidiitz, and E. E. Hoffman, unpublished research, Oak Ridge National L.aboratory (1952-1955). ' 27 31. 32 33, 34. 35. 36. 37. 38. 39. 40. 41. 42, 43. 44. 45. 46. 47, 48. 49, R. S. Peoples, P. D. Miller, and H. D. Hannan, Reactions of Nickel in Molten Sodium Hy- droxide, BMI-1041 {Sept. 1955). H. A. Pray and P. D. Miller, privote communication, Battelle Memorial Institute (1952). R. A. Lad, private communication, Lewis Flight Propulsion Laboratory {1955). P. Villard, Compt. rend. 193, 681 (1931). .. D. Dyer, B. 5. Borie, and G. P. Smith, . Am. Chem. Soc. 76, 1499 (1954); and unpub- lished research, Oak Ridge National L aboratory (1953). W. D. Manly and N. ). Grant, private communication, Ock Ridge National Laboratery (1951), A. ded. Brasunas, Metal Progr. 62(6), 88 (1952), G. M. Adamson, private communication, Oak Ridge Naticnal Laboratery (1951), G. P. Smith, J. V. Cathcart and W. H. Bridges, unpublished research, Ouk Ridge National L aboratory (1951). R. A. Lad and S. L. Simon, Corrosion 10, 435 (1954). W. D. Manly, ""Fundamentals of Liquid Metal Corresion,”’ Corrosion (in press). M. LeBlanc and L. Bergmann, Ber, deut. chem. Ges. 42, 4728 {1909). M. A. Bredig, J. W. Johnson, and W. T. Smith, J. Am. Chem. Soc. 77, 307 (1955); K. Grjotheim, F. Gronvold, and J. Krogh-Moe, j. Am. Chem. Soc. 77, 5824 (1955); E. Heymann, R. J. L. Martin, and M. F. R, Mulcalhy, J. Phys. Chem, 47, 473 (1943); 48, 159 (1944). H. Reinhold, Z. anorg. allgem. Chem, 171, 181 (1928}, E. N. Skinner, private communication, International Nickel Co. (1951). F. Kertesz, F. A. Knox, and W. H., Grimes, private communication, Oak Ridge National L aboratory (1955). H. N. Gilbert, U. S, Patent 2,377,874, June 1945, N. D. Scott, U. S, Patent 2,202,270, May 1940. W. R. Grimes and D. Hili, private communication, Qck Ridge National L aboratory (1951). 29