NEW DEVELOPMENTS IN MATERIALS FOR MOLTEN-SALT REACTORS H. E. McCOY, R. L. BEATTY, W. H. COOK, R. E. GEHLBACH C. R. KENNEDY, J. W. KOGER, A. P. LITMAN, C. E. SESSIONS and J. R. WEIR Metals and Cervamics Division, Oak Ridge National Laboratory Received August 4, 1969 Revised October 10, 1969 Opevating experience with the Molten-Salt Re- actor Experiment (MSRE) has demonstrated the excellent compatibility of the gvaphite-Hastelloy- N-fluoride salt system at 650°C. Sewveral im- provements in wmaterials are needed for a molten-salt breeder reactor with a basic plant life of 30 years; specifically: Hastelloy-N with im- proved vesistance to embrittlement by thermal neutvons; graphite with better dimensional stq- bility in a fast neutron flux; gvaphite that is sealed to obtain a surface permeability of <10~ cm?/sec; and a secondary coolant that is inexpensive and has a melting point of ~400°C. A brief description is given of the materials work in progress to satisfy each of these requivements. INTRODUCTION Our present concept of a molten-salt breeder reactor' utilizes graphite as moderator and re- flector, Hastelloy-N for the containment vessel and other metallic parts of the system, and a liquid fluoride salt containing LiF, BeF,, UF,, and ThF, as the fertile-fissile medium. The fertile- fissile salt will leave the reactor vessel at a temperature of ~700°C and energy will be trans- ferred to a coolant salt which in turn is used to produce supercritical steam. Experience with the Molten-Salt Reactor Ex- periment (MSRE) has demonstrated the basic compatibility of the graphite-Hastelloy-N2 -fluo- ride salt (LiF-BeF,-ZrF,-UF,) system at 650°C. However, a breeder reactor will impose more stringent material requirements; namely: the de- sign life of the basic plant of a breeder is 30 years at a maximum operating temperature of 156 NUCLEAR APPLICATIONS & TECHNOLOGY Oak Ridge, Tennessee 37830 KEYWORDS: radiation effects, molten-salt reactors, breeder reactors, power reactors, reactor core, fused salts, coolants, chromium alloys, molybdenum alloys, nickel al- loys, corrosion, brittleness, thermal neutrons, sodium fluo- rides, sodium borides, mixing, graphite, fast neutrons, sta- bility, expansion, porosity, Hastelloy, embrittlement, mix- tures, surfaces 700°C; the power density will be higher in a breeder and will require the core graphite to sustain higher damaging neutron flux and fluence; and neutron economy is of utmost importance in the breeder and the retention of fission products (particularly **°Xe) by the core graphite must be minimized. Each of these factors requires a specific improvement in the behavior of materials. Experience has shown that the mechanical properties of Hastelloy-N deteriorate as a result of thermal-neutron exposure and a method must be found of improving the mechanical properties of this material to ensure the desired 30-year plant life. Similarly, graphite is damaged by irradiation. Although the core graphite can be replaced, the allowable fast neutron fluence for the graphite has an important influence on the economics of molten-salt breeder reactors. Thus, a program has been undertaken to learn more about irradi- ation damage in graphite and to develop graphites with improved resistance to damage. A big factor in neutron economy is reducing the quantity of '*Xe that resides in the core. This gas can be removed by continuously sparging the system with helium bubbles, but the transfer by this method probably will not be rapid enough to prevent excessive quantities of '*Xe from being absorbed by the graphite. This can be prevented by reducing the surface diffusivity to <102 cm?/ sec, and we feel that this is best accomplished by carbon impregnation by internal decomposition of a hydrocarbon. “Hastelloy-N is the trade name of UCC for a nickel- base alloy containing 169 Mo, "% Cr, 5% Fe, 0.05% C. This alloy was originally developed at ORNL specifi- cally for use in molten-salt systems, It has been ap- proved by the ASME for use in pressure vessels under code cases 1315 and 1345. VOL. 8 FEBRUARY 1970 A new secondary coolant is also needed that will allow us greater latitude in operating temper- ature. Sodium fluoroborate has reasonable phys- ical properties for this application, and the compatibility of Hastelloy-N with this salt is being evaluated. Our work in each of these areas will be de- scribed in some detail. EXPERIENCE WITH THE MSRE Other papers in this series have elaborated on the information gained from the MSRE regarding operating experience, physics, chemistry, and fission-product behavior. Additionally, valuable information has been gained about the materials involved.?™* There are surveillance facilities exposed to the salt in the core of the reactor and outside the reactor vessel, where the environment is nitrogen plus ~2% O,. Hastelloy-N tensile rods and sam- ples of the grade CGB graphite® used in the core of the MSRE are exposed in the core facility. The components are assembled so that portions can be removed in a hot cell, new samples added, and the assembly returned to the reactor. Sam- ples were removed after 1100, 4400, and 9000 h of full-power (~8 MW) operation at 650°C. As shown in Fig. 1, the physical condition of the graphite and metal samples was excellent; identification numbers and machining marks were clearly visi- ble. The peak fast fluence received by the graph- ite has been 4.8 x 10*° n/cm® (>50 keV) and the dimensional changes are <0.1%. Pieces of graph- ite from the MSRE have been sectioned and most of the fission products were found to be located on the surface and within 10 mils below the surface. However, a few of the fission products have gaseous precursors and penetrated the graphite to greater depths. The microstructure of the Has- telloy-N near the surface was modified to a depth of ~1 mil, but a similar modification was found in samples exposed to static nonfissioning salt for an equivalent time. The near-surface modification has not been positively identified, but its presence is likely of no consequence. The very small changes in the amounts of chromium and iron in the fuel salt also indicate very low corrosion rates and support our metallographic observa- tions. The observed low corrosion rate of Hastelloy-N in the MSRE does not come as a surprise, since thousands of hours of corrosion tests preceded brrade name of Union Carbide Corporation for the needle-coke graphite used in the MSRE. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 McCoy et al. MATERIALS Fig. 1. Graphite and Hastelloy-N surveillance assembly removed from the core of the MSRE after 72 400 MWh of operation. salt for 15 300 h at 650°C. Exposed to flowing construction of the reactor. Hastelloy-N is nickel based and contains about 16% Mo, 7% Cr, and 5% Fe. Under normal operating conditions, the fuel salt cannot oxidize (form fluorides) any of these elements except Cr. Since Cr is present in very small concentrations in the alloy, the corrosion is limited by the diffusion of Cr to the metal surface. Corrosion can be reduced even further by con- trolling the oxidation state of the salt, thus reducing the rate of the corrosion reaction at the metal-salt interface. The oxidation state of the salt in the MSRE is controlled by the addition of beryllium metal. Hastelloy-N samples were removed from the surveillance facility outside the reactor vessel after 4400 and 9000 h of full-power operation. This environment is oxidizing, and an oxide film ~92 mils thick was formed on the surface after the longer exposure. There was no evidence of ni- triding, and the mechanical propertiies of these samples were not affected adversely by the pres- ence of the thin oxide film. Thus, experience with the MSRE has proved in service the excellent compatibility of the Hastel- loy-N-graphite-fluoride salt system. DEVELOPMENT OF A MODIFIED HASTELLOY-N WITH IMPROVED RESISTANCE TO IRRADIATION DAMAGE Since the MSRE was constructed, Hastelloy-N, as well as most other iron- and nickel-base alloys, was found to be subject to a type of high- temperature irradiation damage that reduces the FEBRUARY 1970 157 McCoy et al. MATERIALS stress rupture life and the fraction strain.’-'° This effect is characterized in Figs. 2 and 3 for a test temperature of 650°c. The rupture lives for irradiated and unirradiated materials differ most at high stress levels and are about the same for below 20 000 psi. The property change of most concern in reactor design and operation is the reduction in fracture strain. The postirradiation fracture strain is shown in Fig. 3 as a function of strain rate; the scatter band is based upon test results for three different heats of metal., The plot includes results from both tensile and creep tests. In tensile tests the strain rate is a controlled parameter and the test results are plotted directly. In creep tests the stress is controlled and the strain measured as a function of time. The minimum strain rate was used in constructing Fig. 3. The data are characterized by a curve with a minumum at a strain rate of ~0.1%/h, with rapidly increasing fracture strain as the strain rate is increased, and slowly in- creasing fracture strain as the strain rate is decreased. Thus, under normal operating con- ditions for a reactor where the stress levels (and the strain rates) are low, the rupture life will not be affected significantly (Fig. 2), but the fracture strain will be only 2 to 4% (Fig. 3). However, transient conditions that would impose higher stresses or Trequire that the material absorb thermally induced strains could cause failure of the material, Therefore, a material is desired ‘that has improved properties in the irradiated condition and a program with this as its goal has been embarked upon.© The changes in high-temperature properties of iron- and nickel-base alloys during irradiation in thermal reactors have been shown rather con- clusively to be related to the thermal fluence and more specifically related to the quantity of helium produced in the metal from the thermal '°B(n,0)"Li transmutation.”™** The mechanical properties are only affected under test conditions that pro- duce intergranular fracturing of the material. Under these conditions both creep and tensile curves for irradiated and unirradiated materials are identical up to some strain where the irradi- ated material fractures and the unirradiated ma- terial continues to deform. Thus, the main in- fluence of irradiation is to enhance intergranular fracture. “Although the fast-neutron flux will be quite high in the core of our propased breeder, the neutrons reaching the Hastelloy-N vessel will be reduced in energy by the graphite present. Thus, the fast fluence seen by the vessel during 30 years will be <1 x 10%2'n/cm?, and we do not feel that fast-neutron displacement damage is of concern. Experiments will be run to confirm this point. 158 NUCLEAR APPLICATIONS & TECHNOLOGY 70 \ TTTI HEAT NO. 60 NG a 5067 N o 5085 ] il N \\ >0 'KTYPICAL UNIRRADIATED —~ N a \\\~ { \\ S 40 T N\ S S~ 3 * ° Iy ™ Th 2 50 P~ al N g) ?“‘u& \o\ F i\ M 20 Ehas PRETEST ANNEAL-1h AT 1177°C 10 SOLID POINTS-IRRADIATED < 150°C OP‘EN POllNT‘S—IITRADIATEDl 500-6|50°(': 100 10t 102 103 104 RUPTURE TIME (h) Fig. 2. Creep-rupture properties of Hastelloy-N at 650°C after irradiation to a thermal fluence of ~ 5% 10%°n/cm?, ©27.3 022.2 a22.2 14 021.2 13 HEAT NO. © 5065 12 s 5067 0 5085 PRETEST ANNEAL-1 h AT 4177°C 10 - SOLID POINTS-IRRADIATED < 150°C OPEN POINTS-IRRADIATED 500- 650°C FRACTURE STRAIN (%) 10~ 100 To} 102 103 STRAIN RATE (% h) Fig. 3. Fracture strain of Hastelloy-N at 650°C after irradiation to a thermal fluence of ~5 x 10%° n/cm?, VOL. 8 FEBRUARY 1970 A logical solution to this problem would be to remove the boron from the alloy. However, boron is present as an impurity in most refractories used for melting, and the lowest boron concen- trations obtainable by commercial melting prac- tice are in the range of 1 to 5 ppm. The low ‘helium levels that have caused the creep-rupture properties to deteriorate in Hastelloy-N make this approach very unattractive. For example, in- reactor tube burst tests at 760°C showed that the rupture life was reduced by an order of magnitude and that the fracture strain was only a few tenths of a per cent when the computed helium levels were in the parts-per-billion range.'’ Thus, we have concluded that the properties of Hastelloy-N cannot be improved solely by reducing the boron level. Another very important observation has been that the properties are altered by irradiation at elevated temperatures only when the test temper - ature is high enough for grain boundary deforma- tion to occur (above about half the absolute melting temperature for many materials). Thus, the role of helium must be to alter the properties of the grain boundaries so that they fracture more easily. The size of boron lies intermediate between the sizes of small atoms such as carbon that occupy interstitial lattice positions and the larger metal atoms, such as nickel and iron, that occupy the normal lattice positions. For this reason boron concentrates in the grain boundary regions where the atomic disorder provides holes large enough to accommodate the boron atoms. Thus, the transmuted helium will be generated near the grain boundaries where it will have its most devastating effects. We reasoned that the addition of an element that formed stable borides would result in the boron being concentrated in discrete precipitates rather than being distributed uni- formly along the grain boundaries. The trans- muted helium would likely remain associated with the precipitate and be less detrimental. Addition- ally, certain precipitate morphologies and alloying elements are beneficial in improving the resis- tance of alloys to intergranular fracture. Following this reasoning, small additions of Ti, Hf, and Zr have been made to Hastelloy-N and the postirradiation properties were improved mark- edly.'*’*® The titanium-modified alloy was chosen for development as a structural material for a molten-salt breeder experiment (MSBE). A fur- ther modification made in the composition was reducing the molybdenum content from 16 to 12%. This change was prompted by the observation that McCoy et al. MATERIALS tice was adopted to reduce the concentrations of other residual elements thought to be deleterious. The stress-rupture properties at 650°C of several heats of the titanium-modified alloy (0.5% Ti) are summarized in Fig. 4. The properties in the absence of irradiation are improved over those of standard Hastelloy-N and the rupture life of the modified alloy is not reduced more than ~10% by irradiation at 650°C to a thermal fluence of 5% 10%° n/cm® The postirradiation fracture strains of the titanium-modified alloy are also improved over those of the standard alloy (Fig. 5). The modified alloy has a very well-defined duc- tility minimum as a function of strain rate, but the minimum strain is ~3% compared with 0.5% for the standard alloy. Electron microscopy has shown that the titanium-modified alloy forms very fine-grain boundary and matrix precipitates of the MC type when annealed at 650°C. These precipitates are only a few tenths of a micron in size and those in the grain boundaries are spaced at ~2-u intervals. They have a face-centered cubic crystal structure with a lattice parameter of ~4,24 A and are likely complexes involving Mo, Cr, Ti, C, N, and B. These complex compounds also form precipitates in the matrix. This microstructure should lead to trapping of some of the helium as proposed earlier and should also inhibit fracture along the grain boundaries. However, further studies have shown that the precipitates which form during long exposure at 760°C are relatively coarse MoxC carbides and that the postirradiation properties of 70 STD. UNIRRADIATED \< HEAT NOS. /7 . o -66-548 (0.5% Ti) r 60 /\ /;:7\ o -21545 (0.5% Ti) N PRETEST ANNEAL-I h AT 1177°C ??\? IRRADIATED AT 500-650°C _ LR w ~ ‘\ 0 -t o ~ O N Q40 \\\ N w ™1 ¥ é 1 N [2] ~. q w g N 3 ™~ N | g . N n 30 - . ~ \ STD. |RRADlATED] TN . 20 o L 1 1 10 100 1000 RUPTURE LIFE, h 10,000 the additional molybdenum was used in forming Fig, 4. Creep-rupture properties of several heats of large carbide particles that made it difficult to modified Hastelloy-N at 650°C. Samples irra- control the grain size. The vacuum-melting prac- diated to a thermal fluence of ~5x 10%° n/cm? NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 1970 159 McCoy et al. MATERIALS HE AT IRRADIATED TEST TEMPERATURE °C TEMPERATURE °C © 6252 650 650 ® 6252 760 760 © 5911 650 650 ® 5911 760 760 a 24545 500-650 650 A 21545 500-650 760 0 66-548 500-650 650 ® 66-548 500-650 760 ANNEAL 41 h AT 41477°C BEFORE TESTING 18.11a 516.45 19.8 6 MODIFIED (05% Ti) FRACTURE STRAIN (%) TENSILE 2 STANDARD TESTS 10~ 100 10! 102 STRAIN RATE (%/h) 0 10-3 10~2 Variation of fracture strain with rate for sev- eral Hastelloy-N type alloys. Samples irra- diated to a fluence of ~5 x10?® n/cm? prior to testing. Fig. 5. material irradiated at 760°C are very poor. Since some designs require the vessel of the MSBR to operate at ~700°C, this difference in precipitate morphology and subsequent deterioration of prop- erties was of concern. Further work has shown that the desired MC - type carbide can be stabilized at higher service temperatures. This carbide is favored by in- creasing amounts of Ti, Hf, and Nb and decreasing concentrations of Si. Several alloys have been 160 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 prepared that retain good postirradiation prop- erties after irradiation at 760°C. Several of the alloy additions that appear satisfactory are 1.2% Ti, 1% Hf, 1% Hf plus 1% Ti, 0.5% Ti plus 2% Nb, and 2% Hf. The results of tests on these alloys give encouragement that a commercial alloy can be developed that has properties at least as good as those shown in Figs. 4 and 5. IRRADIATION DAMAGE IN GRAPHITE Neutron irradiation alters the physical prop- erties of graphite, but the dimensional changes that occur'®'” are of major concern. These dimensional changes are illustrated in Fig. 6 where the data of Henson et al.’ are presented for isotropic graphite. With increasing fluence the graphite first contracts and then begins to expand at a very high rate. Several potential problems arise as a result of these dimensional changes. First, the initial contraction will change the volume occupied by fuel salt and change the nuclear characteristics of the reactor. These dimensional changes seem small enough for most isotropic graphites that the nuclear effects may be accommodated by design. A second problem is stress generation due to flux gradients across a piece of graphite. Graphite creeps under irra- diation™ and this creep is large enough to reduce the stress intensities to quite acceptable values. The third and most serious problem is that the rapid growth rate represents a rapid decrease in density with potential crack and void formation. At some fluence this will cause the mechanical properties to deteriorate and the permeability to salt and fission products to increase. We feel that properties will be acceptable, at least until the material returns to its original volume, and have defined this fluence as the lifetime. A fourth 10 ;\f 5 W 550 - 600° C 2 I o O o / @ L_J N . / = \ / 3-5 B — 400-440° C- 50 kev) Fig. 6. Volume change in isotropic graphite Dounreay fast reactor irradiations. FEBRUARY 1970 problem is that the dimensional changes are dependent on temperature and the curve in Fig. 6 is shifted up and to the left for increasing temperature. Thus, stresses develop in a part having a temperature gradient since segments of the part are seeking different dimensions. Again, this stress is relieved by the irradiation-induced creep in graphite geometries of interest. There- fore, we consider the onset of rapid growth to be the primary problem and the initial dimensional changes of secondary importance. Graphite temperatures between 550 and 750°C are anticipated in MSBR’s, and operation with a fast flux (>50 keV) as high as 1 x 10" n/(cm? sec) is desired. Data in Fig. 6 indicate that this flux will cause this particular graphite to expand rapidly after a fluence of ~3x 10% n/cm® is reached (~1 year of operation). The flux can be reduced by decreasing the power density, but this usually increases the fuel inventory and doubling time. Hence, it is quite desirable that graphite be used with better resistance to irradiation damage than the graphite shown in Fig. 6. The data available on current reactor graphites irradiated to high fluences were examined and the results described a fairly consistent picture. The flu- ences required for graphite to reach its minimum volume were strongly temperature dependent (de- creased with increasing temperature), but were not appreciably different for any of the graphites studied to date. Although this observation is McCoy et al. MATERIALS discouraging, current experiments show that bet- ter graphites already exist and that others can probably be developed with only small changes in present materials and processing. Let us look briefly at a simple description of the origin of the dimensional changes and then return to our spe- cific observations. Graphite, after being well graphitized at tem- peratures above 2000°C, has a hexagonal close- packed crystal structure consisting of close-packed layers, (basal planes) of carbon atoms with very strong covalent bonds within the basal planes (a direction) and very weak Van der Waals’ forces between atoms in adjacent basal planes (¢ direc- tion). This anistropy in atomic density and bond strength is reflected by very anisotropic prop- erties. The changes that take place in a single crystal of graphite during irradiation are shown schemat- ically in Fig. 7. A neutron having an energy above ~0.18 eV can displace a carbon atom from a close-packed basal plane with a reasonable prob- ability of creating a vacancy in a basal plane and an interstitial carbon atom between the basal planes. Repetition of this process and diffusion at elevated temperatures can result in the formation of defect clusters, specifically partial planes of atoms between the basal planes and vacancy clusters within the original planes. This leads to an expansion perpendicular to the basal planes (c direction) and a contraction within the layer PERFECT SINGLE CRYSTAL i C DEFECT —_— * * PLANES 50 keV) is 1 X 10" n/(cm® sec). Thus, samples can be irradiated to fluences of 1 x 10** n/em? in ~4 months. GRAPHITE Pitch coke crystallites Fig. 8. 162 NUCLEAR APPLICATIONS & TECHNOLOGY Petroleum coke crystallites Petroleum / coke particle Proposed arrangement of crystallites in graphitized stock. VOL. 8 FEBRUARY 1970 A summary of the results obtained to date is shown in Fig. 9. Several materials show a significant deviation from the ‘‘typical’’ behavior illustrated in Fig. 6. The POCOd graphites show excellent resistance to irradiation with very small dimensional changes out to fluences of 2.5 X 10%2 n/em®. Thus, these data support the suppo- sition that the typical behavior of graphite can be improved markedly, but test results have not been extended to fluences high enough to deter- mine the exact magnitude of this improvement. SEALING GRAPHITE Entry of fuel salt into graphite can be pre- vented by keeping the entrance diameter of the accessible porosity smaller than 1 p. Although this does require some extra care during pro- cessing it can be accomplished routinely on large shapes. In fact, the grade CGB graphite obtained 5 years ago for the MSRE satisfies this require- ment.? However, the graphite structure must be much more restrictive to prevent gaseous fission products, particularly '*Xe, from diffusing into the graphite. We presently propose to strip '*Xe from the fuel salt by purging with helium. Helium bubbles will be injected and later removed in a gas-liquid separator. The efficiency of this purging depends very heavily on the size of bubbles that can be injected and circulated and the mass transfer of ®*Xe from the salt to the helium bubbles. Both of these factors are presently uncertain, and we must anticipate that large quantities of '*Xe will be available to the graphite surfaces and that excessive (>0.5%) retention of 135x%e will result if this gas can enter the graphite surface at a high rate. Present calculations show that the accessibility of '*Xe to the graphite surfaces will be impeded by a laminar salt film and that the graphite offers an additional resis- tance to gas flow only when its diffusivity to '®Xe is <~10-% cm?/sec. The best grades of commercial graphites pres- ently available have bulk diffusivities in the range of 107! to 10~* cm?/sec, and it is unreason- able to expect that techniques can be developed for making massive shapes with such a restrictive structure. The techniques used for reducing the porosity of graphite involve multiple impreg- nations of the material with liquid hydrocarbons and then firing to graphitize this material. As the bulk diffusivity decreases, it becomes progres- sively more difficult for the gases released by the decomposing impregnants to diffuse out of the material and the times required to reach the dpOCO Graphite, Inc., Garland, Texas. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 McCoy et al. MATERIALS RY—12-297// 6 / /\RY-12-31 a / BY-12 ‘/ / / / —_ o Q // / NAT. FLAKE ATJ-SG A H-315A / / AGOT (BNWL) - 7/ / _6 CSF (BNWL) 4425—>\ - N +1) POCO GRADES N / AV Vo {00In( S > = o ow AGOT I \ 0 5 10 15 20 25 30 21 FLUENCE (n/cm2) (£>50keV) (X10°7) Fig. 9. Volume changes in graphite irradiated at 705°C. graphitizing temperature become excessive. Thus, it is more reasonable to reduce the surface diffusivity by a postfabrication surface-sealing process involving gaseous impregnation. Since the pyrolytic carbon, that would be deposited, and the graphite substrate will change dimensions differently under irradiation, it is imperative that the pyrolytic carbon be linked with the substrate structure and not deposited as a surface layer that can be sheared easily. The task of sealing the graphite is illustrated by the photomicrograph in Fig. 10. The pro- cessing parameters must be adjusted so that the voids are filled internally in preference to closing the voids near the surface and forming a coating. This can be accomplished by using a flowing stream of hydrocarbon at low partial pressure and temperature appropriate to maintain very low deposition kinetics, but this requires long pro- cessing times. We have used a different method FEBRUARY 1970 163 McCoy et al. MATERIALS — i~ 0.005 INCHES 750X fo 1 Fig. 10. Photomicrograph of the edge of graphite showing a pore that has been partially coated and then sealed over with pyrocarbon, to accomplish penetration which involves pulsing the sample environment between a rich hydro- carbon environment and vacuum. The vacuum cycle removes the reaction products (primarily hydrogen) and allows more hydrocarbon gas to enter the void. Specifically, we have used 1, 3 butadiene at 20 psig, deposition temperatures of 800 and 1000°C, and cycle times of ~1 min for the vacuum and a fraction of a second for the hydro- carbon. Butadiene was chosen because it is a gas at room temperature and because of its high carbon yield per molecule. The temperature range is restricted to 800 and 1000°C because higher temperatures result in a surface coating not penetrating the pore structure and lower temperatures yield intolerably low deposition rates. The lengths of the vacuum and pressure periods are very important because they not only influence the processing rate, but also the depth of penetration of the impregnant. The time re- quired for the process will be important in determining the cost. Two commercial graphites have been used in this work—-viz., AXF made by POCO and ATJ-SG 164 NUCLEAR APPLICATIONS & TECHNOLOGY made by UCC. These materials had widely dif- ferent accessible pore spectra; nearly all the pores in the AXF material were <0.8 p in diameter while the ATJ-SG grade had pores in all size ranges up to 17 p. Thus, the sealing charac- teristics of the two materials were widely dif- ferent. The results of some parameter studies are shown in Fig. 11 where the vacuum-hydro- carbon cycle times were varied at 850°C. The initial slopes are proportional to the surface area being coated and the slope is much steeper for the AXF graphite than for the ATJ-SG material. The sharp break in the curves for the AXF graphite indicates that the pores have been filled or closed off and that the surface area being coated is reduced. The sharpness of this break attests to the uniform pore size of the AXF graphite. The horizontal portions of the curves represent, essen- tially, surface coating, and the data suggest that some finite amount of surface coating is neces- sary to attain the MSBR permeability specifi- cation of <10-° cm%sec for '*Xe at 700°C. (A ¥°Xe permeability at 700°C of 4 x 10~ cm?/sec is approximately equal to a helium permeability at VOL. 8 FEBRUARY 1970 0.10 0.08 0.06 004 MATERIAL: POCO AXF GRAPHITE 0.02 010 0.08 TOTAL WEIGHT GAINED (q) 006 004 McCoy et al. MATERIALS VACUUM (sec) HYDROCARBON (sec) a 60 Yo s 60 o 30 e 30 o 15 v 10x10°8cm?2/sec (HELIUM) v <10°9%m2/sec (HELIUM) 5x mm Hg (AIR 2.5x mm 2.5x1072 . Hg (Al Hg (AIR 75x1072 Hg (AIR) MATERIAL: ATJ-SG GRAPHITE 0.02 0 4 8 12 16 20 24 28 32 36 40 46 TOTAL PROCESSING TIME (h) Fig. 11, Impregnation rate of graphite using 1, 3 butadiene at 850°C. 25°C of 1 x 1078 cm?/sec.) Helium permeability measurements were made and the goal was a helium permeability of <1 x 10~% cm?/sec (denoted on these curves by a ‘v’ mark). The ATJ-SG graphite was not sealed to the desired level under the conditions shown in Fig. 11 and the slope changes very gradually due to the wide variation in the pore sizes. | The data indicate that the processing time could be reduced by shortening the length of the vacuum cycle. Another interesting feature of the process for the AXF graphite was that the final total weight of carbon deposited was increased by shortening the vacuum pulse. This indicates that the depth of penetration of carbon into the ma- terial was increased. Thus, shortening the vac- uum pulse accelerated the process and improved the product, both very desirable characteristics. These studies are not yet extensive enough to optimize the deposition conditions but are suffi- cient to make us optimistic about being able to reduce the diffusivity of graphite to the desired level. The remaining question of prime impor- NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 tance is the integrity of the seal after exposure to high neutron fluences. CORROSION IN FLUORIDE SALT SYSTEMS Two decades of corrosion testing”?™®° and ex- perience with the MSRE®* have demonstrated the excellent compatibility of Hastelloy-N and graphite with fluoride salts containing LiF, BeF,, ThF,, and UF,. The fertile-fissile salt for an MSBR will contain these same fluorides, so only proof- testing will be required for the primary reactor circuit. However, a coolant salt is needed with a lower melting temperature than the LiF-BeF; salt presently used in the MSRE; a sodium fluoro- borate salt (NaBF,-8 mole% NaF) has been chosen as a potential coolant salt for further study. This salt is inexpensive (<$0.50/1b) and has a low melting point of 385°C. A significant characteris- tic of this salt is that is has an appreciable equilibrium overpressure of BF; gas (e.g., 180 mm at 600°C). FEBRUARY 1970 165 McCoy et al. MATERIALS Much of our present corrosion work is con- cerned with the compatibility of Hastelloy-N with sodium fluoroborate. Some earlier thermal con- vection loop studies involving a relatively impure salt of composition NaBF,-4 mole% NaF-6 mole% KBF4 showed that a Croloy-9M loop plugged after 1440 h at a maximum temperature of 607°C and a temperature difference of 145°C, and that a Has- telloy-N loop was partially plugged after 8765 h of operation under the same temperature condi- tions.® The plug in the Croloy was comprised primarily of pure iron crystals and the partial plug in the Hastelloy-N loop was made up of a compact mass of green single crystals of NazCrFe. The salt charge from the Hastelloy-N contained large amounts of Cr, Fe, Ni, and Mo, all major alloying elements in Hastelloy-N. In more recent tests a purer fluoroborate salt of composition NaBF,-8 mole% NaF has been used.””** A thermal convection loop is being used from which we can remove salt samples for chemical analysis and metal samples for weighing without interrupting operation of the loop. The weight changes for the hottest and coldest samples are shown in Fig. 12 for the two loops presently in operation. The loops are constructed of identical materials, but the removable samples in one loop (NCL-13) are standard Hastelloy-N and those in the other loop (NCL-14) are a modified Hastelloy- N containing 0.5% Ti and only 0.1% Fe (standard Hastelloy-N contains 4% Fe). As shown in Fig. 12, the weight changes of the modified Hastelloy-N are smaller, and this is later shown to be due primarily to the lower iron content of the modified material. The rate of weight change was steady except for a small perturbation after 1500 h of operation and a large variation after 4200 h of operation. These times corresponded to times when moist air inadvertently came in contact with the salt. The changes in chemistry shown in Fig. 13 also reflect the admission of air at these times since the oxygen and water levels in the salt increased. The iron and chronium concentrations have continued to increase at a rate proportional to (time)7?, indicating that the process is con- trolled by diffusion in the metal. The nickel and molybdenum concentrations in the salt have re- mained very low except for times when moist air was inadvertently contacted with the salt. These results show qualitatively that the corrosion rates increased when the oxygen and water levels in- creased. Capsule tests in which sodium fluoro- borate containing 1400 ppm O, and 400 ppm H,O was contacted with Hastelloy-N for 6800 h at 607°C exhibited very low corrosion rates (<0.1 mil/year). Future work will be directed toward defining the oxygen and water levels that result in acceptable corrosion rates. 166 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 6 ] : COLDEST SPECIMENS 4 —— . 460°C — ® A A—A—— /“ 2 e o — A/“/ __.-——.""“/./ N : ’_.-——"’.____— £ o d=2" ~ a-el | o \A\ .\ £ \o\ b 2 < o g \A\A\\.\ S ~a AN _4 I o ] }_ I Z N\ W A = -6 HOTTEST SPECIMENS 607 °C -8 [— @ TITANIUM MODIFIED HASTELLOY \ N-NCL-14 ° A STANDARD HASTELLOY \ -0 — N-NCL-13 | T 2 J 0 { 2 3 4 5 6 (x103) TIME (h) Fig. 12, Comparison of the weight changes of Hastelloy- N specimens inserted in NaBF4NaF (92-8 mole%) thermal convection loops. (Assuming uniform corrosion, a weight change of 20 mg/ cm? in 8750 h is equivalent to a corrosion rate of ~1 mil/year.) The information obtained on the changes in salt composition and the weight change of samples located at various points (and temperatures) around the loop is sufficient to attempt a mass balance for the system. The weight of metal lost must equal the weight of metal deposited plus the weight of metal in the salt-i.e., AI/Vloss-—'AI/Vdeposited-*-AI/Vsa.lt ° (1) A weight change vs temperature profile is con- structed based on the removable samples and the assumption is made that each segment of the loop wall follows this same curve. This procedure results in mass balances, Eq. (1), that close within 10%. Diffusion theory can be used for further anal- ysis. As mentioned earlier, chemical analyses (Fig. 13) indicate that the iron and chromium concentrations in the salt are increasing, so it is assumed that the salt selectively removes these elements from the alloy. The modified Hastelloy- N tested here is relatively free of iron, so the FEBRUARY 1970 W o o o € - Hy0 S 2000 i- | % oL 2 | | ; ' \\ £ AL 4 1000 ¥ \\ o i \ S ° \ 0_/1 \ N 2 < \ P | - 0 — 400 T Fe o= . A /’1 350 ‘ & fi 2 C'/ a 300 " 250 ——t N e 200 CONCENTRATION (ppm) 100 ——— Mo - Ni 50 | o\rw‘r:r O O O [ O [ o 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 TIME (h) Fig. 13, Variation of impurities with time in NaBFs- NaF (92-8 mole%), thermal convection loop NCL-14, weight loss of this sample should be due primarily to the removal of chromium. The quantity of material removed by diffusion under conditions where the surface concentration of the diffusing element is zero is given by: AM =Co VDt (2) where A M = material removed, g/cm’ Co = bulk concentration of diffusing species, g/cm’ D = diffusivity, cm®/sec t = time, sec. Using the data of Grimes et al.®* for diffusion of chromium in Hastelloy-N, this analysis showed that the quantity of chromium removed by dif- fusion cannot account for the total weight lost. This discrepancy can be accounted for by short-circuit diffusion mechanisms enhancing the rate of chromium removal at these low tempera- tures or by impurities (likely HF formed by water ingestion) that lead to some general attack of the NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 McCoy et al. MATERIALS metal that is not diffusion controlled. Two obser- vations argue against the latter possibility. First, microprobe analyses have not shown any transfer of nickel or molybdenum to the colder surfaces of the loop except during a brief period after 4200 h of operation in which we knew that large amounts of impurities were present. A second and more convincing argument is based on the relative behavior of standard and modified Hastelloy-N during the first 4000 h of operation. Rewriting of Eq. (2) in terms of a reaction rate constant, K, instead of the diffusion coefficient yields AM = CoVKI . (3) This equation predicts the material transported by diffusion to be too low for the modified alloy when K equal D, but let K take on a value so that the predicted and observed weight losses for a given time of operation agree with Co equal to ™% Cr. Now consider the standard alloy in which Co corresponds to 7% Cr + 4% Fe. The same K chosen for the modified alloy predicts the ob- served weight change for the standard alloy. Thus, the difference in the weight losses between the modified and standard alloy appears due principally to the iron content. Had much general corrosion occurred, this adjustment in Co should not have worked. In fact, this analytical proce- dure was entirely unsatisfactory for the short time period after 4200 h when the water and oxygen levels were high and nickel and molybdenum were being removed (Figs.. 12 and 13). A further possible role of impurities is to provide the oxidizing potential necessary to keep the surface concentration of iron and chromium at zero. Thus, even though the process remains diffusion controlled, the rate can be increased by im- purities. Although some of the curves in Figs. 12 and 13 have quite large slopes, the corrosion rates are not very high. Using the rather rough method of converting the weight losses to corrosion depths indicates that the average rate during 8000 h of operation has been 0.7 mil/year. The rate has decreased to ~0.3 mil/year for long time periods in which operation was not disturbed. In scaled- up MSBR systems, a cold trapping technique will likely be used to remove some of the corrosion products so that their solubilities are not ex- ceeded. We presently feel that the sodium fluoro- porate salt will provide a satisfactory and economical secondary coolant for molten-salt re- actors. SUMMARY Experience with the MSRE has proven the basic compatibility of the graphite—Hastelloy-N-fluoride FEBRUARY 1970 167 McCoy et al. MATERIALS salt system at elevated temperatures. However, -a molten-salt breeder reactor will impose more stringent operating conditions, and some improve- ments in the graphite and Hastelloy-N for this system are needed. The mechanical properties of Hastelloy-N deteriorate under thermal-neutron irradiation, but the addition of titanium in combi- nation with strong carbide formers such as nio- bium and hafnium makes the alloy more resistant to this type of irradiation damage. Graphite undergoes dimensional changes due to exposure to fast neutrons, and the possible loss of structural integrity due to these dimensional changes pres- ently limits the lifetime of the core graphite. Although the core graphite can be replaced as often as necessary, these replacements influence the economics of the reactor, and a program to find a better graphite has been initiated. Studies to date indicate that graphites can be developed that have better resistance to irradiation damage than conventional nuclear graphites. The graphite used in the core will be sealed with pyrocarbon to reduce the amount of '*Xe that is absorbed. Techniques have been developed for this sealing, and studies are in progress to determine whether the low permeability is retained after irradiation. Corrosion studies indicate that the corrosion rate of Hastelloy-N in sodium fluoroborate is accep- table as long as the salt does not contain large amounts of impurities, such as HF and H,O. ACKNOWLEDGMENTS This research was sponsored by the U.S. Atomic Energy Commission under contract with the Union Carbide Corporation. REFERENCES 1. E. S. BETTIS and R. C. ROBERTSON, “The Design and Performance Features of a Single-Fluid Molten-Salt Breeder Reactor,” Nucl. Appl. Tech., 8, 190 (1970). 2. W. H. COOK, ‘‘Molten-Salt Reactor Program Semi- annual Progress Report, August 31, 1965, ORNL-3872, pp. 87-92, Oak Ridge National Laboratory., 3. H. E. McCOY, ‘““An Evaluation of the Molten-Salt Reactor Experiment Hastelloy N Surveillance Speci- mens—First Group,”” ORNL-TM-1997, Oak Ridge Na- tional Laboratory (November 1967), 4. H. E. McCOY, ‘““An Evaluation of the Molten Salt Reactor Experiment Hastelloy N Surveillance Speci- mens—Second Group,”” ORNL-TM-2359, Oak Ridge Na- tional Laboratory, in press. 5. D. R. HARRIES, J. Brit. Nucl. Energy, 5, 74 (1966). 168 NUCLEAR APPLICATIONS & TECHNOLOGY 6. W. R. MARTIN and J. R. WEIR, in Flow and Frac- ture of Metals and Alloys in Nuclear Envivonments Spec. Tech. Publ., 380, pp. 251-267, American Society for Testing and Materials, Philadelphia (1965). 7. J. T. VENARD and J. R. WEIR, in Flow and Frac- ture of Metals and Alloys in Nuclear Envivonments Spec. Tech. Publ., 380, p. 269, American Society for Testing and Materials, Philadelphia (1 965). 8. W. R. MARTIN and J. R. WEIR, Nucl. Appl., 1, 160 (1965). 9. W. R. MARTIN and J. R. WEIR, ‘‘Postirradiation Creep and Stress Rupture of Hastelloy N,’’ Nucl. Appl., 3, 167 (1967). 10. H. E. McCOY and J. R. WEIR, ‘“‘Stress-Rupture Properties of Irradiated and Unirradiated Hastelloy N Tubes,”’ Nucl. Appl., 4, 96 (1968). 11. P. C. L. PFEIL and D. R, HARRIES, in Flow and Fracture of Metals and Alloys in Nuclear Envivonments Spec. Tech. Publ., 380, p. 202, American Society for Testing and Materials, Philadelphia (1965). 12. P.C. L. PFEIL, P. J. BARTON, and D. R. ARKELL, ‘““Effects of Irradiation on the Elevated-Temperature Mechanical Properties of Austenitic Steels,”’ Trans. Am. Nucl. Soc., 8, 120 (1965). 13. P. R. B. HIGGINS and A. C. ROBERTS, Nature, 206, 1249 (1965). 14. H. E. McCOY, Jr. and J. R, WEIR, Jr., Materials Development for Molten-Salt Breeder Reactors, ORNL- TM-1854, Oak Ridge National Laboratory (June 1967), 15. H. E. McCOY, Jr. and J. R. WEIR, Jr., ““Develop- ment of a Titanium-Modified Hastelloy with Improved Resistance to Radiation Damage,”’ Proc. Symp. on the Effects of Radiation on Structuval Metals, San Francisco, Calif., June 23-28, 1968, to be published. 16. R. E. NIGHTINGALE, Nuclear Graphite, Academic Press, New York (1962). 17. J. H. W. SIMMONS, Radiation Damage in Graphite, Pergamon Press, New York (1965). 18. R. W. HENSON, A. J, PERKS, and J. H. W. SIM- MONS, ‘‘Lattice Parameter and Dimensional Changes in Graphite Irradiated Between 300 and 1350°C,”’ AERE-R 5489, p. 33, Atomic Energy Research Establishment (June 1967), 19. C. R. KENNEDY, ‘“Gas Cooled Reactor Program Semiannual Progress Report, March 31, 1964, ORNL- 3619, pp. 151-154, Oak Ridge National Laboratory, 20. W. C. RILEY, High-Temperature Matevrials and Technology, p. 188, I. E. CAMPBELL and E. M. SHER- WOOD, Eds., Wiley, New York (1967). 21. W. H. COOK, ‘‘Molten-Salt Reactor Program Semi- annual Progress Report, July 31, 1964, ORNL-3708, p. 377, Oak Ridge National Laboratory. VOL. 8 FEBRUARY 1970 22. L. S. RICHARDSON, D. C. VREELAND, and W. D. MANLY, ‘“Corrosion by Molten Fluorides,”” ORNL- 1491, Oak Ridge National Laboratory (March 17, 1953). 23. G. M. ADAMSON, R. S. CROUSE, and W. D. MAN- LY, ‘“Interim Report on Corrosion by Alkali-Metal Fluorides: Work to May 1, 1953,”> ORNL-2337, Oak Ridge National Laboratory. 24, G. M. ADAMSON, R. S. CROUSE, and W. D. MAN- LY, ‘“‘Interim Report on Corrosion by Zirconium-Base Fluorides,’”’ ORNL-2338, Oak Ridge National Laboratory (January 3, 1961). 25. W. B. COTTRELL, T. E. CRABTREE, A. L.. DAVIS, and W. G. PIPER, ‘‘Disassembly and Postoperative Examination of the Aircraft Reactor Experiment,”’ ORNL-1868, Oak Ridge National Laboratory (April 2, 1958). 26. W. D. MANLY, G. M. ADAMSON, Jr., J. H. COOBS, J. H. DeVAN, D. A. DOUGLAS, E. E. HOFFMAN, and P. PATRIARCA, ‘““Aircraft Reactor Experiment—Metal- lurgical Aspects,”” ORNL-2349, Oak Ridge National Lab- oratory, pp. 2-24, (December 20, 1957). 27. W. D. MANLY, J. H. COOBS, J. H. DeVAN, D. A, DOUGLAS, H. INOUYE, P. PATRIARCA, T. K. ROCHE, and J. L. SCOTT, Progr. Nucl. Energy Ser. 1V, 2, 164 (1960). 28. W. D. MANLY, J. W. ALLEN, W. H. COOK, J. H, DeVAN, D. A. DOUGLAS, H. INOUYE, D. H. JANSEN, P. PATRIARCA, T. K. ROCHE, G. M. SLAUGHTER, A. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 McCoy et al. MATERIALS TABOADA, and G. M. TOLSON, Fluid Fuel Reaclovs, pp. 595-604, JAMES A. LANE, H. G. MacPHERSON, and FRANK MASLAN, Eds., Addison Wesley, Reading, Pa. (1958). 29, ‘“Molten-Salt Reactor Program Status Report,’’ pp. 112-113, ORNL-CF-58-5-3, Oak Ridge National Labora- tory (May 1, 1958). 30. J. H. DeVAN and R. B. EVANS, III, in Conference on Corrosion of Reactor Matevials, June 4-8, 1962, Pro- ceedings Vol. II, pp. 557-579, International Atomic Energy Agency, Vienna (1962), 31. J. W. KOGER and A. P. LITMAN, ‘‘Compatibility of Hastelloy N and Croloy 9M with NaBFs-NaF-KBF4 (90-4-6 mole%) Fluoroborate Salt,”” ORNL-TM-2490, Oak Ridge National Laboratory, in preparation, 32. J. W. KOGER and A. P. LITMAN, ‘‘Molten-Salt Reactor Program Semiannual Progress Report, Febru- ary 29, 1968, pp. 221-225, ORNL-4254, Oak Ridge National Laboratory, 33. ‘‘Molten-Salt Reactor Program Semiannual Pro- gress Report, August 31, 1968, ORNL-4344, Oak Ridge National Laboratory, in press. 34, W. R. GRIMES, G. M. WATSON, J. H. DeVAN, and R. B. EVANS, in Conference on the Use of Radioiso- topes in the Physical Sciences and Industry, September 6-17, 1960, Proceedings Vol. III, p. 559, International Atomic Energy Agency, Vienna (1962). FEBRUARY 1970 169