CHAPTER 21 MATERIALS OF CONSTRUCTION—METALLURGY* 21-1. LMFR MATERIALS 21-1.1 Metals. Alloy steel. For maximum power production, it is de- sirable to operate an LMFR at the highest possible temperature consistent with the mechanical properties and corrosion resistance of the materials of construction. A maximum temperature of 500°C or higher is deemed desirable for economically attractive operation of the reactor. No ma- terials have yet been found that are mechanically strong at these tempera- tures, readily fabricable, and also completely resistant to corrosion by the {-Bi fuel. This does not mean that there is no hope for obtaining a good material for holding bismuth fuel. On the contrary, very significant advances have been made in the past few years. It must be realized that before work was started on liquid metal fuel reactors, very little was known about the solubility and corrosion characteristics of liquid bismuth with reference to containing materials. There is general optimism that continuing research and development will lead to suitable materials for containing the U-Bi fuel system. , The low-alloy steels offer a good compromise for use in the heat ex- changer, piping, and reactor vessel, particularly sinee their corrosion re- sistance ean be greatly improved by the addition to the fuel of Zr + Mg as corrosion inhibitors. Nickel-containing stainless steels cannot be used, despite their good high-temperature mechanical properties, because of the high solubility of Niin Bi, and the greatly lowered U solubility in the presence of this dissolved Ni. Extensive engineering and fundamental studies have been made on the corrosion of the low alloy steels by inhibited U-DBi, as well as the mechanism of corrosion inhibition. Radiation effects are currently being investigated. Of course, besides steels, there are other materials, notably the rarer metals, which have characteristics making them suitable for certain uses in a liquid-metal system. However, unless the cost and ability to fabricate these materials can be improved significantly, heavy dependence will have to be placed upon alloy steels for the main containment problem. *Based on contributions by D. H. Gurinsky, D. G. Schweitzer, J. R. Weeks, J. =0 Bryner, M. B. Brodsky, C. J. Klamut, J. G. Y. Chow, R. A. Meyver, R. Bour- deau, O. I'. Kammerer, all of Brookhaven National Laboratory; L. Green, United IEngineers & Constructors, Ine., Philadelphia, Pa.; and W. P. Eatherly, M. Janes, and R. L. Mansfield, National Carbon Company, Cleveland, Ohio. AN 7H MATERIALS OF CONSTRUCTION—METALLURGY [cHap. 21 21-1.2 Graphite. In the LMFR, graphite is considered as the principal choice for the moderating material because of its availability, cost, and knowledge of its characteristics under radiation. However, there are addi- tional special requirements for the graphite in the LMFR system. It not only 1s the moderator, but is also the container material for the U-Bi solution in the reactor. Hence it should be impervious to the liquid metal and mechanically strong. Experimental work at BNL has shown that graphite can be used directly in contact with the fuel stream without danger of corrosion. By preferen- tially reacting to form ZrC at the fuel-graphite interface, the Zr corrosion inhibitor also prevents reaction of the U and fission products with the graphite. Special grades of graphite are being developed that appear to have the desired mechanical strength and low porosity required for use as moderator and reflector in the reactor. Reactions of graphite with the fuel, and the possible effects of pile radiation on these reactions, are de- scribed in the following sections. 21-2. STEELS 21-2.1 Static tests. In order to attack the steel corrosion problem in a basic manner, solubilities of the various components and combinations have been determined. Most of these solubilities are given in Chapter 20. However, more solubility work, important from a corrosion point of view, is discussed here. Solubility of steel components and inhibiting additives in liqued Bi. [ron. The solubilities of iron in Bi, Bi+0.19%:Mg, Bi+40.29%U + 0.1 Mg, and Bi+ 0.19 Mg+saturation Zr are given in Fig. 21-1. Uranium and Mg, in the quantities added, have no effect on the iron solubility over the temperature range 400 to 700°C. Zirconium increases iron solubility slightly at temperatures above 500°C. Titanium (which might be present as a corrosion inhibitor) has been found to decrease the iron solubility at temperatures above 450°C, the extent of this decrease being proportional (but not linearly) to the amount of Ti in the liquid. Below 400°C, there appears to be a considerable increase in the iron solubility. For example, Bi containing 1600 ppm Ti dissolved only 309 as much iron as pure Bi at 690°C, while Bi containing 300 ppm Ti (saturation) at 350°C dissolved more than ten times as much iron ag pure Bi. Zirconium. The solubility of Zr in Bi is given in Fig. 20-5. This appears to be unaffected by the presence of Mg, Cr, or Fe in the liquid metal. Chromium. The solubility of Cr in Bi is given in Fig. 20-6. This also appears to be unaffected by the presence of Mg, Zr or Fe in the liquid bismuth. However, the presence of Cr in Bi causes 2 marked reduetion in the iron solubihity. 21-2] STEELS 715 T°C 727 637 561 497 442 395 353 500 i | | l I — — fe in Bi+Mg+U 100 |- — 50— — — Fe in Bi+Mg+Zr —] £ — ] a a — .f ] 0= — S5+ : ” & g - — 1 1 | 1.0 1.1 1.2 1.3 1.4 1.5 1.6 103/T°K Fig. 21-1. Solubility of Fe in Bi alloys. Miscellaneous data. The Fe—Zr intermetallic compound ZrFes appears :0 decompose when added to Bi, Zr dissolving approximately to its normal saturation and Fe somewhat in excess of its normal solubility in the presence of Zr. The amount of excess Fe present in the liquid metal ean possibly be attributed to a finite solubility of the undissociated intermetallic com- pound ZrFes;. The solubility of Ta in Bi is estimated to be less than 0.01 ppm (detec- tion limit) at 500°C. The solubility of Ni in Bi is close to 5% at 500°C and probably greater than 19, at 400°C. The =olubility of Mg in Bi is close to 4% at 500°C and 2% at 400°C. Surface reactions. Experimental evidence has shown that the corrosion resistance of steels in Bi is in part due to the formation of insoluble films on the steel surfaces. The effect of these films on the corrosion behavior of different steels 1s not readily determined by thermal convection loop experi- ments because of the relatively low temperatures (400 to 550°C) and long times assoclated with such tests. The comparative behavior of different 746 MATERIALS OF CONSTRUCTION—METALLURGY [cHAP. 21 stecls and different films is more easily obtained from high-temperature (600 to 850°C), short-time, static contact tests. Steel specimens approximately 1/2 in. wide, 2 in. long, and 1/8 in. thick are cleaned and given various surface treatments, such as sandblasting, chemical etches, polishes, ete. Six to ten different materials are then placed in a vacuum furnace, heat-treated as desired, and immersed in a Bi alloy containing the desired additives. The crucible used to contain the liquid metal is either a material inert to Bi, such as Mo or graphite, or the same material as the specimen. After contacting, the samples are removed from the solution at temperature and allowed to cool in He or in vacuum. The adherent Bi is removed from the steel by immersing in Hg at 200°C in a vacuum or inert atmosphere. After rinsing, the residual adherent Hg is completely removed by vacuum distillation at 100 to 200°C. The cleaned surfaces are examined by x-ray reflection techniques, utilizing a North American Phillips High Angle Diffractometer. Surface reaction of zirconium, titanium, and magnesium. When pure iron was contacted with bismuth containing radioactive zirconium tracer for 1 hr at 450°C, a Langmuir type adsorption of the zirconium on the iron crucible surface was obtained. Increasing the temperature to 520°C and the contact time as much as 24 hr showed an increased amount of reaction. The structure of this deposit is not known. On the other hand, when pure iron 18 contacted in saturated solutions of zirconium in bismuth for times ranging from 100 to 300 hours at 500 to 750°C neither corrosion nor x-ray detectable surface deposits occur. At concentrations of zirconium below saturation value, pure iron is extensively attacked. A tightly adherent, thick, uniform, metallic deposit was found on the surfaces of pure I'e dipsticks contacted with liquid Bi saturated with Ti at 650 to 790°C. In all cases the x-ray patterns were the same but could not be identified. The 15- to 25-micron layers were carefully scraped off and chemically analyzed. The results corresponded to a compound having the composition FeTi4Bis. Pure Fe and 239, Cr-19, Mo steel samples contacted with 2.5 w/0 Mg in Bi at 700°C for 250 hr showed no deposit detectable by x-ray diffraction. Slight uniform intergranular attack was observed on all the samples. Pure I'e samples contacted with Bi solutions containing 0.569, Mg + 170 ppm Zr, and 0.239, Mg + 325 ppm Zr at 700°C were not attacked and did not have detectable surface films. These solutions acted similarly to those saturated with Zr. Reactions of steels with UB7 solutions. Uranium nitride (UN) deposits have been identified on the surfaces of 59, Cr-1,/29, Mo, 219, Cr-19, Mo, Bessemer, and mild steels, after these samples were contacted with Bi solutions containing U or U 4 Mg. Extensive attack always accompanied UN formation, indicating that this film is not protective. Nitrogen analyses 21-2] STEELS 747 made on these contacted specimens show that depletion of the N in the steel is much more rapid than it is when the same steels are contacted with solutions containing Zr. Reactions of steels with Bv solutions conlaining combinations of Zr, Mg, U, Th, and Ti. Deposits of ZrN, ZrC, and mixtures of the two have been identified on many different steels contacted with Bi solutions containing Zr with or without combination of Mg, U, and Th. No corrosion has ever becn observed on such samples contacted at 600 to 850°C for 20 to 550 hr, nor have films other than ZrN or ZrC been found. When a mild steel was contacted with Bl containing 1000 ppm Zr and 200 ppm Ti at 650°C, x-ray examination showed strong lines for TiN and a less intense pattern of TiC. Considerable difficulty was experienced 1n establishing the correct unit cell dimension for the nitrides and carbides of Zr and Ti. Many different values may be found i the literature. The inconsistency in the data probably can be attributed to the existence of varying amounts of C, O, or N 1n the samples. Table 21-1 gives the parameters determined by a nunber of mvestigators. The values of ap used in this research were those given by Duwez and Odell [1]. These compared favorahly with the values found on test specimens, powdered compact samples, and ZrN prepared by heating Zr in purified N2 at 1000°C for 20 hr. A nondestructive x-ray method of measuring film thickness has been de- veloped for this research [2]. The x-rays pass through the film and are diffracted by the substrate back to a counter. The intensity is reduced by the absorption of the film. Unknown conditions of the substrate are eliminated by measuring the intensity of two orders of reflection or by nmeasuring the intensity of a reflection using two different radiations. The method 1= accurate to about 209. TaBLE 21-1 PuBLisHED X-rAY PARAMETERS FOR THE UNIT CELLS OF ZrC, TiC, ZrN, axp TIN (CuBic, NACl-TYrE) Becker and | Van Arkel | Kovalski and | Dawihl and | Duwez and ! Ehert [20] [21] Umanskii [22] | Rix [23] Odell [24] ; | IaC 4.76 4.73 46734 4685 ‘ TiC 4.60 4.26 4,4442 4.31 4.32 N 4 63 4.61 4 567 - TiX 440 4.23 4,234 4936 4.237 | 748 MATERIALS OF CONSTRUCTION—METALLURGY [caAp. 21 TaBLE 21-2 ORIGINAL ANALYSES AND Frovms FORMED ON SPECIAL STEELS UseEp 1N StaTic TESTS ‘ , o7, Al N % N as Film Material Sol) | (Toty | EENT L UUBHEN | formed 5Cr-+Mo 0.016 (0.023 0.0002 1.0 ZrN 2:Cr-1Mo 0.003 0.042 0.0001 0 » 21Cr-1Mo 0.055 | 0.050 — — ” 21Cr-1Mo 0.003 0.01 — — ” 24Cr-1Mo 0.06 0.047 0.0003 1.0 ? 2+Cr-1Mo (.009 0.013 0.0001 1.0 " Bessemer 0.003 0.009 0.0002 2.0 ” Carbon 0.007 0.005 0.0001 2.0 ” 2+Cr-1Mo 0.015 0.015 100 ZrC 23Cr-1Mo 0.44 0.054 0.025 50 ” 22Cr-1Mo ¢.014 0.013 0.009 70 ? 24Cr-1Mo 0.022 0.015 0.010 70 ” 24Cr-1Mo 0.02 0.015 0.011 75 ” 11Cr-1Mo 0.02 0.014 | 0.010. 70 & RH 1081 (0.3 Ti) " *EHN: Ister-halogen insoluble nitrogen. This is believed to be an indication of the nitrogen combined as AIN or TiN in steels [26]. Effect of steel composition and heat treatment. It has been found experi- mentally that some steels with very similar over-all compositions behave quite differently in the same static corrosion tests. Films that form on these materials range from pure ZrN to pure ZrC. Table 21-2 gives typical analyses selected from the more than 100 steels run in static corrosion fests, and identifies the surface films. After contacting, the only changes in analyses were found in the total nitrogen remaining and the amount of ester halogen insoluble nitrogen (ISIIN) present in the steels. The only significant difference 1n analyses between nitride-formers and carbide- formers in Table 21-2 is found in the relative amounts of IX'HN. The carbide-formers have more than 509 of the total nitrogen combined as EHXN, while the nitride-formers have only a few percent of the total nitro- gen combined. At present, the relationship between the N, Al Cr, and the Mo contents of the steels and their film-forming properties is not obvious. Some excellent nitride-formers have very low nitrogen content, while some carbide-formers have high nitrogen content. The same holds true for the 21-2] STELLS 710 Al Cr, and Mo contents of the steels. The TN content of a steel can be readily changed by short-time heat treatment @t 700°C and higher [3], so that this variable is controllable within limits. To a first approximation, the corrosion resistance of a particular steel is enhanced by high “inhibitor’” concentrations and/or the presence of in- soluble adherent films formed on the steel surface. The first of these con- ditions 1s neither desirable nor practical in a solution-type fuel reactor be- cause of the adverse effect of Zr on the U solubility. At present, work is being done to measure quantitatively the etfects of different alloying con- stituents on the activities of N and C in steels. Consider the following reactions: ZI‘(Bi) + N(stccl) O~ ZrN (Alm), (2]*1) Zr(Bi) + C(Hteel) - Zr(-\j(fllm)- (21—2) Assuming that the films are insoluble in Bi, then at equilibrium 1 v and Koy = (@ (a0 . 1 Ky = Tam)(an) (21-3) azy (aN) If the products of the Zr activity in the Bi with the activities of the N and (' in the steel are not sufficient to satisfy the respective equilibrium con- stants, the reactions will not occur, and the steel will not form ZrC or ZrN films. If the activity products are greater than the constants, Ko or Az, the reactions will proceed until the activities are lowered to these values. Thus, for a fixed Zr activity, the activities of N and C in the steel determine whether the carbide and nitride film-producing reactions should occur. The excess of N or C above these equilibrium values should be a meazure of the driving force of reactions (21-1) and (21-2) to the right. Solution rate tests. The solution rates of Fe into Bi, and Bi+ Zr and Mg, were measured in erueibles of a carbon steel, a 239% Cr-1% Mo, a 5% Cr-1 29 Mo, and an AIST type—410 steel. The crucible, Bi, and additives were equilibrated at 400 to 425°C, the temperature rapidly raised to 600°C, and the concentration of Fe in solution measured as a function of time. Results are shown in Fig. 21-2. In the presence of Zr-+ Mg, the 5% Cr—1 27 Mo and the AISI type—410 steels dissolved at approximately the =ame rate, while the 219 Cr-1% Mo steel dissolved more slowly. No detectable dissolution of Fe from the carbon steel was measured in 44 hr at 610°C. These results are parallel to the thermal convection loop results, and consistent with the film-formation studies in that the measured solu- tion rates are inversely proportional to the ability and rate at which the steels form ZrN films. At present no data are available on rates of solution for ZrC-forming steels. 750 MATERIALS OF CONSTRUCTION—METALLURGY [cHAP. 21 ] | 70 r ] | ] T I /Sofubflify {Temp Cycle} 60 — /5% Cr Steel into Pyre 8i | | l I | 12% Cr Steel into / — Bi + 0.01% Mg . 5% Cr Steefl into ................ graneer BI + 0‘10/0 Zr + 0.]0/0 Mg L /‘?\__ o . . 12% Cr Steel into Bi + 0.05% Zr - / - + 0.04% Mg " _______ |- _________ T——2.1/4% Cr Steel into B + 0.05% Zr — e D=7 7 21020 Steel into Bi + 0.04% Z + 0.12% Mg J','/ / eel Into Bi r i }_ i E l ! ! 1 i I} I | | | 0 1 2 3 4 3 6 7 g 20 40 60 80 Time in Hours F1a. 21-2. Dissolution of Fe into Bi (plus additives) at 600°C from steel crucible. Rates of precipitation. The rate of precipitation of iron from bismuth in a pure iron steel crucible is very rapid. Iron precipitated from bismuth, saturated at 615°C, as rapidly as the temperature could be lowered to 425°C. The addition of Zr plus Mg to liquid metal did not change the rapid precipitation of most of the iron from the bismuth under these same condi- tions, but produced a marked delay in the precipitation of the last amount of iron in excess of equilibrium solubility. An apparently stable super- saturation ratio of 2.0 was observed for more than 7 hr at 425°C in a pure iron crucible containing Bi 4 1000 ppm Mg <+ 500 ppm Zr, and 1.7 for more than 48 hr at 450°C. In a 59 Cr steel crucible, a supersatura- tion ratio of iron in Bi+ Mg+ Zr of 2.9 was observed after 24 hr at 425°C. This phenomenon may be due to the ability of the formed surface deposits to poison the effectiveness of the iron surface as a nucleation promotor or catalyst, the different supersaturations observed being due to the relative abilities of a Zr-Fe intermetallic compound or of ZrN to promote nuclea- tion of iron. This observed supersaturation suggests that mass transfer should be nearly eliminated in a circulating system in which the solu- bility ratio due to the temperature gradient does not exceed the meas- ured “‘stable supersaturation’” at the cold-leg temperature. Precipitation rate experiments made in AISI type-410 steel crucibles show that Zr 4 Mg stabilize Cr supersaturations of 2.0 to 3.0 for more than 24 hr. However, no Cr supersaturation was found during precipita- tion rate experiments made in pure Cr crucibles when Zr+4 Mg were present in the melt [4]. The measured supersaturations should therefore be due to the films present on the steel surfaces. 21-21 STEELS 751 21-2.2 Corrosion testing on steels. The research effort on materials for containment of the LMIER has been concerned mainly with low-alloy steels having constituents which have low solubilities in 131, such as C, Cr, and Mo. Although the solubilities of Fe and Cr are only 28 and 80 ppm respectively at the intended maximum temperature of operation, severe corrosion and mass transfer are encountered when pure Bi or a U-Ii solu- tion 1s circulated through a temperature differential in a steel loop. This results {rom the continuous solution of the pipe material in the hot portion of the system and subsequent precipitation from the supersaturated solu- tion in the colder portions. Zirconium additions to U-Bi greatly reduce this corrosion and mass transfer. The behavior of steels in U-Bi is studied in three types of tests. Thermal convection loops are used to test materials under dynamic conditions. In these, the fuel solution is continuously circulated through a temperature differential in a closed loop of pipe. Variables such as material composi- tion, maximum temperature, temperature differential, and additive con- centrations are studied in this test. More than sixty such loops have now been ran at BNL. The prineipal imitation in these tests is that the veloc- ities obtained hy thermal pumping are extremely low when compared with the LAIFR design conditions. Foreed circulation loops are used to study materials under environments more closely approximating LMER conditions. Three such loops are now i operation at BNL and two more are under construction. A very large loop (+ in. ID) which will circulate U-Bi at 360 U.S. gpm and transfer about 2% X 10% watts of heat, is now under eonstruction and is expected to g0 nto operation late this year. Statie tests, as discussed previously, in which steels are isothermally im- mersed in high-temperature U-Bi containing various additives, are used to ~tudy their corrosion resistance and the inhibition process as a function of additive concentration and steel composition. Most of the tests have been performed on a 239 Cr-19 Mo steel (Table 21-3). However, some testx have also been made with higher Cr steels, 139 Cr-1/2% Mo, 1200 Cr=1 29 Mo, and carbon steels. 21-2.3 Thermal convection loop tests at BNL. A typical thermal con- veetion loop that has been used at BNL is shown in Fig. 21-3. The loop 1= provided with a double-valve air lock at the top of the vertical section which permits taking liquid metal samples while the loop is running without contaminating the protective atmosphere. The hot leg is insulated and heat 1= = the applied load (Ib), r = the radius of sleeve (ft), and T = the torque (ft-lb). In general, the hard-to-hard material combinations have shown good wear resistance except for some scoring. The best hard material tested thus far is AloO3 flame-coated on AISI 4130 steel. When this material was contacted against itself, no wear or scoring could be detected. This ma- 21-3] SALT CORROSION 773 terial will be thermally cyeled and exposed to inhibited U-Bi for long times to evaluate its utility. Stellite 90 and Rex AA also behaved well. Contacts made with common die steels and low-alloy steels have exhibited severe scoring and wear. Corrosion has also been detected on these samples. Of the cemented carbides, only TiC with either a mild steel or 239, Cr-19, Mo binder has heen tested. This material did not show good wear properties and also exhibited some pitting corrosion. Graphitar versus tool steel and Mo versus Rex AA or Stellite 90 have shown the best results of the hard versus soft combinations tested. The results have been good in that the wear has been very smooth; however, the wear has heen excessive. The use of these combinations would be limited to very low-load applications. 21-5. SaLT CORROSION In earlier chapters, it was pointed out that one of the chief advantages of the LMFER lies in the possibility of easy chemical processing. Several processing techniques have been studied, most of which are based on pyrometallurgical processes. The two chief pyrometallurgical methods under consideration are the chloride process, in which the bismuth fuel is contacted with a ternary mixture of molten chloride salts, and the fluoride proces=s, in which the bismuth fuel is contacted with molten fluoride salts contaming hydrogen fluoride. As may be imagined, the construction material problem for these plants is very difficult. A corrosion test program is actively under way at BNL and Argonne National Laboratory on the chloride and fluoride processes respectively. At BNL, these tests have consisted principally of rocking furnace and tab exposure tests. In the rocking furnace test, a piece of tubing approximately 12 in. long and 1 2 1in. ID, containing a charge of either salt or a mixture of salt and bhi=muth, is placed on a rocking rack in a furnace. This rack alternately tilt< to one end for a period of 1 min and then to the other end for a like amount of time. The two ends of the furnace are kept at 450 and 500°C in order to give a temperature differential and thus induce mass-transter corro=ion. The standard test period has been 1000 hr. These tests are part of the initial sereening program. When they are completed, the metals which have given the best performance will be further evaluated in test loop=s and pilot-plant equipment. At present, only molybdenum has been satisfactorily tested against a mixture of salt and bismuth fuel. However, the results are definitely en- couraging, It has been found that the ternary salt, MgCle-NaCl-KCl, with or without zirconium and uranium chlorides, can be contained fairly well 11 austenitic stainless steels, particularly 347 stainless steel. When a 774 MATERIALS OF CONSTRUCTION—METALLURGY [cHAP, 21 mixture of bismuth fuel and the ternary salt containing less than 1% BiCls was tested, the ferritic stainless steels were the best materials. These include 410, 430, and 446 stainless steels. Probably the best of the ferritics is the 229 Cr-19% Mo stainless steel. During one step in the chloride chemical process, it is necessary to have the ternary salt, containing more than 1% BiCls, in contact with bismuth fuel. For this mixture only molybdenum has been satisfactory. However, considerably more testing is required before this can be considered a satisfactory material. The experience in handling salt with larger-sized equipment is quite limited. A small loop built of 347 stainless steel has been operated satis- factorily for a fairly short time. A much larger loop, loop “N,” is now being constructed at BNL. This will contact the chloride salt and the bismuth fuel. The salt part of the loop is constructed of 347 stainless steel. The bismuth fuel section of the unit is constructed of 21% Cr-19% Mo steel. The actual eontacting units are constructed of both 347 and the low-chrome steels. This pilot plant, when placed in operation, should furnish considerable information on the corrosion characteristics of the molten chloride salt. The fluoride process also presents difficulties with materials of construc- tion. The mixture of the molten fluoride salts, containing HF, with the bismuth fuel is extremely corrosive. Pure nickel has been found to stand up fairly well to the molten fluoride salts alone. However, the combination of the three materials has proved to be very corrosive even to nickel. The extensive development program to investigate the materials of construction for the fluoride process is continuing. 21-6. (GRAPHITE 21-6.1 Mechanical properties. In the proposed LMFR system, the moderator, graphite, is also employed as the container material. Therefore, the graphite should have good physical properties such as strength, hard- ness, and resistance to shock. Since graphite is to be the container material for the bismuth solution, it should theoretically be completely impervious to the solution. For this reason, special graphites, more impervious than the usual reactor grades, have been developed and are under development. Physical properties of typical examples of these graphites are given in Table 21-6. In comparison with the usual reactor grade, AGOT, having a compressive strength of 6000 psi, these impervious grades have a strength of 6500 to 9700 psi. Another special requirement for the graphite is that it withstand erosion or pitting by the flowing fuel. Test sections of accurately bored graphite were placed in test loops where the flow velocity of bismuth was 6 to 8 fps. No observable effect was noted after 1000 hr of test at 550°C. 21 6] GRAPHITI 775 Although tests so far have been on rather small samples, the mechanical properties of these improved graphites appear sufficiently good for use in LEMFR systems. These new graphites must be manufactured in large sizes in order to conveniently make up the core of an LMFR. "The graphite industry in the United States is now developing manufacturing techniques for making such large sizes. 21-6.2 Graphite-to-metal seals. I.caktight joints of steel to graphite are required at several places in the core of an LMFR. These seals must with- stand an average of 125 psi at approximately 5530°C. This is done by joining finely machined steel and graphite surfaces under sufficient spring loading to prevent bismuth leaking across the seal. Tests were run by Markert at the Babeock & Wileox Research Center to evaluate =uch pressure secals. Three-inch and six-inch steel pipes (21, Cr-19. Mo) with machined ends were pressed against a flat surface of a block of MH4LM graphite (Great Lakes Carbon Co., density 1.9 g/ce). The graphite surface had been prepared by sanding and polishing with No. 000 emery paper. A seal was effected against Biat 438° with a pressure differential across the seal of 100 psi, and with 1500 psi stress between the egraphite and the steel. The minimum stress that may be used without vizible Bi leakage at this pressure differential was found to be as low as 600 psi. It was not necessary to resort to complicated interface configura- tions to obtain a seal. Thesc Initial results are very encouraging, and turther development work is being directed toward more complicated seals. 21-6.3 Graphite reactions. If graphite is to be in direct contact with the 1'=B31 fuel, it should be inert to the various fuel constituents and also to fi=sion products and corrosion products. Work has heen done at various locations on these reactions. Thermodynamic data on chemical equilib- rium, when available, have proved to be extremely valuable in guiding the experiments. U raniim-graphite reactions. The reaction between uranium and graph- ite i~ probably the most important one to consider in the LMEIR. Mallett, Grerds, and Nelson [11] reported that uranium forms three stable carbides: UC, UCs, and UaCy. Further work on this subject [7,12] indicates that when less than 197, U 1s present in bismuth, it does not react with graphite to form carbides at temperatures below 1200°C. However, the nitride of uranium, UN, has been identified on graphite contacted with 0.059, U n Bi at 850°C for 28 hr. This nitrogen was un- doubtedly adsorbed on the surface of the graphite and had not been dis- lodged by outgassing at high temperatures and vacuum. When zirconium and magnesium are present with uranium in the bis- muth, zirconium reacts preferentiaily with the graphite to form ZrC. This TaBLE 21-6 GENERAL PHYSICAL ProOPERTIES OoF GraruiTe AT 20°C Grade Base grades R-0015 R-0018 ATJ R 0025 R-0020 Tmpregnated grades ATJ-82 MI4LLA-90 Manufacturer Max. production size™ National Carbon Company 40" dia. 407 dia. 207 x 247 x 8”7 National Carbon Company 407 dia. 40" dia. 207 X 24" x 8”7 (rreat Lakes ("arbon Co. 48 (lia. Dengity Electrical resistivity Thermal conductivity Flexural strength Compressive strength Coeflicient of thermal expansion Helium flow at < p> =27 atm Ap=1atm W a W & W L W a W W d .85 16 .40 0.28 0.23 3250 3000 7500 7500 — 1.85 1.53 1.77 0. 21 0.17 3400 3200 8700 8700 3 U Lo b it 1.73 I.16 1.43 (.30 0.22 3300 3300 8400 8500 =2.7atm w 5.7 1.7 0.21 300 ml/min through Ap=1atm 0.0427 0.0557 0.0217 1 em cube a 0. 00040 *Ag gpecificd at present time by the individual manufacturers. TProcessed and measured as small samples. [9-12 HALIHAVHD =1 =1 778 MATERIALS OF CONSTRUCTION—METALLURGY [cHaPp. 21 is predictable from the chemical thermodynamic data. An experiment in which graphite was contacted with 1000 ppm U, 50 ppm Zr, and 300 ppm Mg in Bi at 1000°C for 8 hr showed only a single intense x-ray diffraction line corresponding to the most intense line for ZrC. The x-ray analysis was carried out after the adherent bismuth was removed from the samples by mercury rinsing. These experiments indicate that the reaction of uranium with graphite is not likely to' occur under the LM IR operational conditions and can be prevented by the addition of zirconium to the bismuth. Zircontum and titanium reactions with graphite. Since zirconium and pos- sibly some titanium will be present in the bismuth, reactions of these materials with graphite have been investigated. As described above, zir- conium reacts to form the carbide with a strong negative free energy. At temperatures around 550°C, ZrC and solid solutions of ZrC and ZrN have been 1dentified on graphite surfaces contacted with bismuth solutions con- taming 130 ppm Zr. On the other hand, no reaction between graphite and 1600 ppm Ti in Bi solutions has been observed up to 800°C for contact times up to 170 hr. A strong TiC x-ray pattern and a less intense ZrC—ZrN solution pattern were observed on graphite contacted with approximately 0.29;, Ti and 0.29, Zr in Bi at 1250°C for 44 hr. When steel samples are reacted with U-Bi fuels containing zirconium and magnesium, the x-ray patterns of the surface are those for pure or very nearly pure nitrides or carbides. When graphite is contacted with the fuel, however, solid solutions of the carbide and nitride are often found. The unit cells vary from 4.567 Kx* to 4.685 Kx for the zirconium com- pounds and from 4.237 Kx to 4.320 Kx for the titanium compounds. These parameters are low for complete carbon carbide structures. Parameters for carbon-deficient carbide structures have been reported in the literature [13]. Yor ZrC the reported ao varied from 4.376 Kx at 20 atomic percent C to 4.67 Kx at 50 atomic percent C. However, up to the present time no evidence of carbon deficient structures has been observed in studying the graphite-fuel experiments. The low parameters are instead believed due to nitrogen replacing the carbon atoms in the carbide lattice (NaCl-type). Parameters for such solid carbide-nitride solutions are de- scribed by Duwez and Odell [1]. Fission product-graphite reactions. The products of uranium fission may also react chemically with graphite to form carbides, A series of experi- ments have shown that materials such as cerium will definitely react with graphite. When 25 ppm Ce in bismuth was placed in contaet with graph- ite at 700°C for 110 hr, CeCs was identified as a film on the graphite. Graphite contacted with 140 ppm Sm in bismuth at 800°C for 140 hr, on *Kx = 1000x units = 1.00202 4 0.00003A.. 21-6] GRAPHITE 779 the other hand, gave an x-ray diffraction pattern of the graphite surface which could not be identified or indexed. Under similar reaction conditions, neodymium, barium, or beryllium in bismuth solutions have no reaction product. However, Miller [12] has shown by an autoradiograph technique that 180 ppm irradiated Nd in bismuth reacts with graphite at 1100°C in 100 hr, concentrating the radioactive Nd at the graphite-liquid metal in- terface. When the same experiment was repeated with the addition of 100 ppm Zr to the solution, no Nd was identified at the graphite surface. This experiment indicates that zirconium will probably form the car- bide and nitride preferentially to most fission products. However, further research is required to determine whether a zirconium carbide-nitride layer on the graphite can be depended upon to prevent the adhesion of the fission products to the graphite surface. This point is not only important from the chemical and graphite surface point of view, but is also important from the neutron economy point of view. To obtain the highest breeding ratio, the fission products, which are all fairly good neutron adsorbers, must be removed from the graphite core soon after their formation. This is espe- ciallv true of samarium, which has a very high neutron absorption cross section. Therefore, the experiments reported above on zirconium are quite encouraging in that there is no trace of a samarium film on the graphite. 21-6.4 Radiation effects on graphite. The graphite core, in order to serve as moderator and container for the flowing fuel, must be stable to radiation. Iission recoils may cause spalling and reduction in the thermal conductivity that might increase the thermal stresses within the core struc- ture. The graphite must not adsorb large quantities of U or fission products. A capsule test has been developed in which samples are irradiated in a highly enriched U-Bi solution containing Mg and Zr inhibitors for study of radiation effects on materials, The test has the advantage of attaining high temperatures (700°C) and a high fission recoil density. Ciraphite samples have been exposed in these capsules under conditions given in Table 21-7. Metallographic examination of these graphite sam- ples indicates that there is no excessive spalling or corrosion. However, some samples were treated with Bi containing Zr at 1300°C to obtain 10- to 30-micron ZrN-ZrC layers on the graphite prior to irradia- tion. and postirradiation examination indicated some change of this layer. The cause of these effects i1s still being determined. The effect of neutrons on the growth and thermal conductivity proper- ties of special low-permeability grades of graphite has been measured. Three sets of samples have been irradiated to exposures as great as 5 X 10%° thermal neutrons/em? in the temperature range 400 to 475°C. Results of these tests are listed in Table 21-8. The change in physical length of the graphite is primarily contraction and should not present a major engineer- g problem. TaBLE 21-7 UranNiuM SoLuTioN CapsurLE CoNDITIONS ON GRAPHITE Sol. concentration, ppm Capsule Za}rii)l(:a f Exp;):ture, R::;;rréd. Observations U-235 Zr Mg Y 6 Graphite (AGHT) 4000 200 350 12.8 X 108 392 Zrn layer not evident after irradiation 8 » ” 3250 200 350 6.36 320 No spalling 10 7 ” 3750 180 700 12.3 320 No spalling 15 ” ” 4450 1500 1900 13.1 420 No spalling STVISHLVIN 082 NOILLDAYISNOD 40 ADHNTIVILHW 1Z "dVHD] [9-12 TapLe 21-8 Puysicar ProreErTYy CHANGES IN Low PrrMmeapiLiTy GrapHITES INDUCED BY NEUTRON IRBRADIATION AT ELEvATED TEMPERATURES* Thermal conductivity at Electrical resistivity, Graphite Kxposure Irrad., 50°C, cal/{em)(°C)(sec) Gross ohm-em X 1075 type nut 7°C growth, 9 Preirrad. Postirrad. Preirrad. Postirrad. At 1.8% 1029 475 0.290 (0.198 0.01 9.51 18.21 % Gt 1.8 475 0.393 0.286 0.002 7.18 16.31 o ATI-R82% 1.8 475 0.253 0.163 0.02 11.9 24 43 S R-0025% 1.8 475 0.217 0.127 0.02 13.3 25.84 -3¢ 10 475 0.308 0.137 — 11.2 19.4 G--77 10 475 (.329 0.131 — 8.2 18.4 G-121 10 475 0.347 0.141 — 7.4 17.8 *Irradiations and measurements were made at the Hanford Atomic Products Operation of the General Electric Company. FGraphite Specialties Company. iNational Carbon Company. 8L 782 MATERIALS OF CONSTRUCTION——METALLURGY [cHaP. 21 On the other hand, the neutrons reduced the thermal conductivity by 25 to 359, of the preirradiation value. Moreover, some of these graphites had a preirradiation conductivity only 659, that of the usual reactor-grade graphite. The combination of low permeability and neutron irradiation therefore reduces the thermal conductivity to 5097 that of the usual reactor- grade graphites. Bismuth penetration into graphite permits the diffusion of some uranium into the graphite. The resulting fission recoil particles may {urther reduce the thermal conductivity of graphite. Tests are now under way to deter- mine the degree of damage produced. 21-6.5 Bismuth permeation and diffusion into graphite. Iarly in the work on the Liquid Metal Fuel Reactor it was recognized that a special type of graphite was required if it were to be both a moderator and a con- tainer for the reactor fluid. Such an impermeable graphite would have not only the usual advantages of being both structural material and moderator, but would also not hold up quantities of coolant or fuel, with resulting decreased neutron efficiency and control. Besides these characteristics, because the design fluxes are of the order of 10! n/(cm?)(sec) it is necessary that the graphite have a high degree of resistance to radiation damage, specifically to physical growth and reduction in thermal conductivity. In considering impermeable graphite, two characteristics of the graphite are concerned: liquid pickup and permeability. Liquid pickup refers to the amount of fluid which is held in the interior pore volume of the graphite in the manner of a sponge. Permeability is the rate at which a fluid can be made to flow through the graphite. Both these properties depend primarily on the accessible void volume and the pore spectrum. In research work on graphite, it is customary to divide the size of pores into two categories: macropores larger than 1 micron and averaging 2.5 microns in radius, and micropores with radii less than 1 micron and pre- dominantly below 0.5 micron. The development of new types of “impermeable” graphite has necessi- tated an examination and improvement of manufacturing processes con- current with experiments on bismuth uptake [14]. The classic process of graphite manufacture is based on cokes and pitch binders, baking and pitch reimpregnations, and finally graphitization. Around this scheme has evolved a complex technology involving careful particle and flour sizing to obtain optimum compaction, elaborate baking and graphitizing schedules, and extremes of pressure vacuum treatments. The production of the new relatively impermeable grades was made pos- sible by three significant advances over the older technology: (1) the ability to use raw materials of more advanced form, including graphites and blacks of various types, (2) impregnation techniques now span a variety of resins 21-6] GRAPHITE 783 with various viscosity and wetting properties, and (3) new forming tech- niques permit much more uniform and finer-grained artificial graphites. This last includes, in particular, the development of pressure baking, where heat is applied by passing electric current through the carbon in the mold and under pressure. The conventional pitch-type impregnation has the effect of increasing density without markedly reducing permeability. This is apparently due to the tendency of the pitch to coke out only in the voids of large effective pore radius. Conversely, the newer impregnating materials, with their lower viscosities and increased wetting, tend to block the pores themselves as well as the larger open volumes. Consequently, there is no general rela- tionship between density and permeability. Several conclusions may be drawn: (1) The newer impregnates primarily attack the macropore distri- bution and shift it from the 1- to 5-micron range into the submicron range. (2) Optimum particle packing in the original base material is, in general, not an advantage upon reimpregnation. (3) It is essential to use base ma- terials in which the long tail at high pore radii is missing. The relatively minor effect of the present impregnates upon the micropore distribution demonstrates that the present materials, markedly improved as they are, do not as yet represent the achievable ultimate. The amount of bismuth uptake in graphite is probably the most im- portant property concerned in evaluating the graphite, and was one of the first investigated. For this purpose, a simple pot arrangement is used to hold samples of graphite in molten bismuth at pressures from vacuum to 530 psi and at temperatures from 550°C. The graphite samples (0.5-in. OD and 1.75-in. long) are outgassed in a vacuum at 550°C and then sub- merged in the bismuth. Helium pressure is then applied to the molten bis- muth. The amount of bismuth uptake into the graphite is determined by the difference in the density of the sample before and after submersion. The accuracy of this measurement is within 0.01 g bismuth per ce graphite. The degree of bismuth uptake by graphite as a function of time, de- termined at a pressure of 250 psi at 550°C, is shown in Fig, 21-13. As was mentioned above, there is no correlation between uptake and graphite denzity. The densities of these graphites range from 1.73 to 1.92 g/cc. The total percent of void volumes in the impregnated grades varies from 16.0 to 19.0¢¢ of the bulk volume. Of these totals, the inaccessible volumes range from 6 to 109%. Although samples EY-9 and ATL-82 have nearly the xame density, the bismuth absorption differs by as much as a factor of 3 because of the difference in pore spectrum. As can be seen from the figure, the amount of bismuth uptake varies as a function of time. This behavior was obtained using separate samples for each point on a curve and also by measuring the same sample at various time intervals. In making determinations, considerable oscillation about 784 MATERIALS OF CONSTRUCTION—METALLURGY [crAP. 21 1.3 T ] l [ ! | Right Hand Scale Bismuth Uptake, grams/cc graphite Immersion, Hundreds of Hours Fia. 21-13. Bi penetration; successive immersion of same specimen. Tempera- ture = 550°C, pressure = 250 psi, outgassed 550°C for 20 hr. a mean value is found for investigations extended to as much as 3500 hr. Evidently these variances are caused by outgassing by the graphite over the time interval of the experiment. Outgassing the graphite at 900°C in- stead of at 550°C reduced the amplitude of the excursions, but the mean value remained from 0.425 to 0.525 g bismuth per cc graphite for most of the types investigated. When the outgassing temperature was increased to 900°C, the saturation or maximum value of uptake was reached in some cases within 2.5 hr. The rate of bismuth penetration into graphite was determined in order to estimate the effect of an unexpected pressure excursion in the reactor. Samples were subjected to 250 psi for times varying from 5 see to 5 min, as shown in Fig. 21-14. This time span far exceeds that expected for a reactor pressure surge. The test conditions were 250 psi at 550°C after an outgassing period of 20 hr at 550°C. The data indicate that the practical maximum uptake is reached in about 10 sec for all graphites except types A and G. These graphites, which are essentially coatings instead of bulk impregnations, have their uptake increased continuously with time. In a long-term test their equilibrium values were not reached until after some 800 hr of submersion. Since the core of the reactor will be subjected to various pressures, a study was made of the effect of pressure on the absorption of bismuth by graphite. Long-term tests covering hundreds of hours were conducted at Bismuth Uptake, grams/cc grophite 21-6] GRAPHITE 785 IIEIII T T lll|[ level | Bismuth Penetration, in Grams Bi/cc Graphite 0 | L 1 Pt | o e ey ! 10 100 500 Time of Immersion, in Seconds Fig. 21-14. Short-time Bi penetration. Temperature = 550°C, He pressure = 250 psi, outgassed 550°C for 20 hr. 125 ps1 to duplicate the test discussed above. It was found that the bismuth uptake is approximately the same as for the 250-psi pressure and the rela- tive absorption remain the same between the different grades of graphite. In another series of tests, samples were immersed for 20 hr at 550°C at varying pressures, as shown in Fig. 21-15. The samples for each curve were first evacuated for 20 hr at 550°C before being immersed in bismuth. With each type of graphite, the bismuth uptake at 450 psi corresponds ap- proximately to the values attained at 250 psi for longer periods of sub- mersion. Type R, CCN, HLM, and ATI~-82 are insensitive to pressure imereases beyond 200 psi. The remaining three types, A, G, EY-9, do in- crease continuously In bismuth uptake and furthermore show a threshold pressure below which no bismuth penetrates the graphite for the 20-hr duration of the test. However, results of long term tests at 125 psi showed that graphites hiving a threshold pressure at 20 hr do absorb bismuth after several hundred hours. After a pressure surge in the reactor core, the operating pressure will return to approximately 120 psi, and the amount of bismuth in the graphite nmight decrease. To investigate this, samples impregnated at 450 psi were resubmerged in bismuth at 25 and at 100 psi to determine what quantity of bismuth might leave the graphite. The dotted lines in Fig. 21-15 con- nect these points. It can be seen that there is no significant reduction of the bismuth contained in each type of sample. 786 MATERIALS OF CONSTRUCTION—METALLURGY [cHAP. 21 81— e EYQ 50 Add. Hrs. T ¢ e 20 - Ach. o ,/////////////// Hfi_ G - 140 R Hrs. J . s 4 a-" ——— v— N nlllllfin\ . ,,rl////// Bismuth Penetration, groms Bi/cc graphite | | l [ 0 100 200 300 400 500 Helium Pressure in psi Fic. 21-15. Effect of pressure on Bi penetration; successive immersions of 20 hr. Temperature = 550°C, outgassed 550°C for 20 hr. Dashed lines for reduced pressure after 450-psi impregnation. Caleulations of the percent of voids filled with bismuth were made for the maximum bismuth uptake obtained in the experiments shown in Fig. 21-15. Table 21-9 gives these calculated values for the approximate saturation level reached. In this table, the last graphite, AGOT, is the conventional reactor graphite. The 1009, filling of the voids is obviously a good check of the assumption that it is quite permeable. All the other graphites are impermeable grades under development by various comi- panies. The percent voids filled for these graphites do not represent total saturation of the accessible voids. Rather, these values show that about 1/3 to 1/2 of the accessible void volumes have been filled in these experi- ments. In studying threshold penetration effect, surface tensions of bismuth on various surfaces of graphite were measured (Table 21-10). Although dif- fering from the accepted values, these determinations probably represent more closely the actual circumstances in a reactor core. In none of the four cases was wetting of the graphite obtained by the bismuth or bismuth solution. Uranium diffusion into bismuth in graphite pores. Since a certain amount of fuel absorption will have to be tolerated with the graphites now avail- able, it is essential to measure the diffusion of uranium into graphite by 21-6G] GRAPHITE TasLE 21-9 Voips FiLLep at 450 psi AFTER 20 HR. 787 Graphite type Voids filled with Bi, 97 100 37 Y-9 44 A 17 G 37 HLM 32 M 23 R 32 AGOT 100 TaBLE 21-10 SURFACE TENSIONS _ : Time af Wetti : i Ciraphite surface Constituents 1me 4 te_r erting Surface tension, contact, min | properties dynes/cm Smooth Bi 1 None 276 29 257 78 241 Rough and loose | Bi D None 153 particles 60 142 Polished Bi+ 350 ppm Mg 15 None 66 90 66 Polished Bi+4 350 ppm Zr 15 None 285 : 120 275 240 282 360 283 788 MATERIALS OF CONSTRUCTION—METALLURGY [cuap. 21 means of the bismuth solution. FExperiments to measure this effect have been made. This was done by first impregnating graphite with bismuth so- lution containing magnesium and/or zirconium. After bismuth impregna- tion at a given pressure, uranium was added to the solution and the graph- ite allowed to soak in the bismuth solution for a period of time. The amount of uranium which diffused into the graphite was measured by sectioning the graphite and analyzing for uranium concentration as a function of distance from the surface of the sample. These experiments were run at 550°C with a pressure of 200 psi. The graphite was first allowed to soak in the bismuth solution for 90 hr; then the uranium was added and the conditions were maintained for the duration of the experiment. Results of two experiments are given in Table 21-11. In the first experiment, the bismuth contained 390 ppm Mg and 1000 ppm U. The uranium concen- tration in the graphite specimen was found to be less than in the melt solution and decreased from the sample face inwards. The second experiment was performed exactly like the first except that no magnesium was present. The graphite specimen (Great Lakes Type HLM) absorbs less bismuth than the EY-9 graphite used in the first ex- periment. However, the uranium concentration near the surface of the specimen built up to an amount considerably greater than that initially in the solution, and the concentration gradient is much steeper than was found when magnesium was present in the solution. This high value for the uranium-to-bismuth ratio near the interface may be explained by assuming that uranium reacted with impurities present on the graphite surfaces. Apparently when magnesium is present in the solution it reacts preferentially with these impurities. These experiments definitely show that uranium and other solutes present in the bismuth can be expected to diffuse into the graphite as far as the bismuth has penetrated. Ior the graphites now at hand, this means diffusion through the entire thickness of the graphite for the long-term exposures contemplated in a reactor core. Of course, since the diffusion of uranium itself takes considerable time, fission will convert it to other prod- ucts before it has an opportunity to diffuse many inches into the graphite. The effect of diffusion of the various solutes and fuel into graphite on neutron economy and reactor operational characteristics is recognized, and studies have to be made in large-scale experiments. In general, it is believed that the graphites at hand will meet the require- ments for the first experiment of an LMFR reactor. It i1s already possible to produce some of these in sizes as large as 40 to 60 in. in diameter. As this development progresses, graphites of greater impermeability will be produced. Improvements in graphite have taken place steadily, and markedly improved materials are anticipated in the future. 21-6] GRAPHITE 789 TaBLE 21-11 URANIUM DIFFUSION INTO (GRAPHITE Specimen no. Distance, in. ‘ Bi, 9, 1 Mg, ppm ‘ U, ppm A. EY-9 Graphite 1 0 0312 32.2 1150 900 2 0.0937 30.5 1050 840 3 0.1562 29.6 750 810 4 0.250 28.0 800 760 5 0.500 | 26.0 820 740 B. HLM Graphite 1 0.0312 16.35 5600 2 0.0937 16.87 2630 3 0.1562 16.38 460 4 (.250 16.78 70 . 5 0.500 17.78 30 | 790 MATERIALS OF CONSTRUCTION—METALLURGY [cHAP. 21 REFERENCES 1. P. Duwez and F. OprLy, Phase Relationships in the Binary Systems of Nitrides and Carbides of Zirconium, Columbium, Titanium, and Vanadium, J. Electrochem. Soc. 97, 299-304 (1950). 2. D, T. Kzating and O. F. Kammerer, Film Thickness Determination from Substrate X-ray Reflections, Rev. Sct. Instr. 29, 34 (1958). 3. L. 8. DarkeN et al., Solubility of Nitrogen in Gamma Iron and the Effect of Alloying Constituents—Aluminum Nitride Precipitation, J. Metals 3, 1174~ 1179 (1951). 4. J. R. Wreks and D. H. Gurinsxy, Solid Metal-Liquid Metal Reactions in Bismuth and Sodium, in ASM Symposium on Liquid Metals and Solidification, ed. by B. Chalmers. Cleveland, Ohio: The American Society for Metals, 1958. 5. 0. T. KaMMERER et al., Zirconium and Titanium Inhibit Corrosion and Mass Transfer of Stecls by Liquid Heavy Metals, Trans. Met. Soc. AIME 212, 20-25 (1938). 6. G. W. Horsrey and J. T. Maskruy, The Corrosion of 2149 Cr—1% Mo Steel by Liquid Bismuth, Report AERE M/R-2343, Great DBritain, Atomic Energy Research Establishment, 1957. 7. J. R. Weeks ¢t al., Corrosion Problems with Bismuth-Uranium Fuels, in Proceedings of the International Conference on the Peaceful Uses of Alomic Energy, Vol. 9. New York: United Nations, 1956 (P/118, pp. 341--355); D. H. GURINSKY and G. J. Dienes (Eds.), Nuclear Fuels. Princeton, N. J.: D. Van Nostrand Co., Ine., 1956. (Chap. XIII); J. R. Werks, Metallurgical Studies on Liquid Bismuth and Bismuth Alloys for Reactor Fuels or Coolants, in Progress in Nuclear Energy, Series IV, Technology and FEngineering, Vol. 1. New York: Pergamon Press, 1956. (pp. 378-408) 8. W. C. Lesuie and M. G. Fontana, Mechanism of the Rapid Oxidation of High Temperature, High Strength Alloys Containing Molybdenum, Trans. Am. Soc. Metals 41, 1213 (1949). 9. L. S. Marks (Ed.), Mechanical Engineers Handbook. 4th ed. New York: McGraw-Hill Book Company, Ine., 1941. (p. 232) 10. W. E. MARKERT, JR., personal communication to J. R. Weeks, Mar. 20, 1958. 11. M. W. Marrerr et al., The Uranium-Carbon System, USAEC Report AECD-3226, Battelle Memorial Institute, 1951; The Reactor Handbook, Vol. 3, General Properties of Materials, USAEC Report AECD-3647, 1955. (p. 316) 12. W. E. Mituer and J. R. Wekeks, Reaclions between LIMFR Fuel and Its Container Materials, USAEC Report BNL-2913, Brookhaven National Labora- tory, 1956. 13. G. V. Samsonov and N. 8. RoziNova, Some Physicochemical Properties of Zirconium-Carbon Alloys, Izvest. Sektora Fiz-~Khim. Anal. Inst. Obshcher. Neorg. Khim. Akad. Nauk. 8.8.8.R. 27, 126-132 (1956). 14. W. P. EATHERLY ct al., Physical Properties of New Graphite Materals for Special Nuclear Applications, paper prepared for the Second International Conference on the Peaceful Uses of Atomic Energy, Geneva, 1958.