ORNL-TM-4047 _ BBOENVED BY 1ic | & MOLTEN SALTS AS BLANKET FLUIDS IN CONTROLLED FUSION REACTORS W. R. Grimes Stanley Cantor OPERATED BY UNION CARBIDE CORPORATION FOR THE U.S. ATOMIC ENERGY COMMISSION This report was prepared as an account of work sponsored by the United States Government. Neither the United States nor the United States Atomic Energy Commission, nor any of their employees, nor any of their contractors, subcontractors, or their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness or usefuiness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Contract No. W-7405-eng-26 REACTOR CHEMISTRY DIVISION MOLTEN SALTS AS BLANKET FLUIDS IN CONTRCLLED FUSION REACTORS W. R. Grimes and Stanley Cantor DECEMBER 1972 OAK RIDGE NATIONAL LABORATORY Osk Ridge, Tennessee 37830 operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION NOTICE This report was prepared as an account of work sponsored by the United States Government. Neither the United States nor the United States Atomic Energy Commission, nor any of their employees, nor any of their contractors, subcontractors, or their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, com- pleteness or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights, ORNL-TM- 4047 1ii CONTENTS Abstract . . . . i v e v e e e e e Introduction « « « « + ¢« + + o« o« . Behavior of LigBeF, in a Hypothetical CIR Effects of Strong Magnetic Fields Effects on Chemical Stability Effects on Fluid Dynamics Production of Tritium . e e . Recovery of Tritium . . . . . . Chemical Transmutations . . . . Compatibility of Li,Bel, with CTR Metals and Moderators Compatibility with Steam, Air, and Liquid Metals Choice of Most Promising Salts . . . . - Molten Salts in Laser-Induced Fusion Reactors. Summary: General Comparison of Molten Salts with Lithium in Fusion Reactors . . . . . . . . . Acknowledgments . . . . . . . . . . . References . - o A1 .13 .16 .17 .18 22 .24 025 .26 MOLTEN SALTS AS BLANKET FLUIDS IN CONTROLLED FUSION REACTORS W. R. Grimes and Stanley Cantor ABSTRACT The blanket of a fusion reactor serves to absorb and trans- fer the energy of the fusion reaction products, and to produce the tritium necessary to refuel the reactor. This report out- lines how these two functions are performed by lithium-bearing molten salts. The strong magnetic fields may have a considerable effect on the chemical stability and a less significant effect on the fluid dynamics of a flowing salt. A salt melt flowing across a strong magnetic field induces an electric field, which in turn produces an emf between the walls of the conduit and the adjacent salt. The emf can be lessened to minor proportions by careful design--causing the salt to flow parallel to the mag- netic field wherever possible and using a system of small bore tubes where the flow must cross the magnetic field. Although flow in the magnetic field parallel to the lines of force suppresses turbulence (necessary in a salt for adequate heat transfer), this effect on molten salts will be negligible owing to their low electrical conductivities. Breeding of tritium in a molten-salt blanket is at best marginal when the lithium is in its natural isotopic abundance; however, the tritium-breeding ratio can be improved by including blanket regions of lithium or of beryllium, or by enriching the salt in lithium-6. LiF and its mixtures with Bel,; are the best molten-salt coolants in which to breed tritium in quanti- ties adequate for fueling a reactor. Molten LiF-BeF, is advan- tageous in recovering tritium since, in contact with metallic Ni, Mo, or W, virtually all of tritium is present as TF. While fluorides are adequate heat-transfer agents and possess good radiation stability, neutronic transmutation of Be and F in the salt ca le&d 10 corrosion unless a redox buffer (analogous to U3 /U4 in a fission reactor) is included in the melt. In a blanket which has two coolants, one being metallic lithium, salts other than LiF-BeF, could be considered. For example, LiCl-KCl, melting at 354°C, may be adequate as vacuum wall coolant and as the fluid to transport heat to the steam system of the reactor. INTRODUCTION A controlled thermonuclear reactor (CTR) that fuses deuterons with tritons yields 17.6 Mev per fusion event mostly in the form of very energetic neutrons. ©Such a device requires a blanket system, a more or less complex composite of several materials, capable of performing at least two functions. These functions are (a) absorption of the energy carried by the energetic fusion reaction products and transfer of heat generated in the blanket to the power-producing portion of the reactor, and (b) generation of tritium, in a manner such as to enable its ready recovery, to replace that consumed in the fusion reaction. The first of these functions certainly requires a suitable heat transfer fluid. The second requires that the blanket contain a sufficiency of lithium, from which tritium can be effectively bred. Though these two functions are separable in princip]_e-,1 there is probably a considerable advantage, other than elegance, if the coolant fluid can be a liquid with suffi- cient lithium to sustain the required tritium production. The coolant fluid must meet several criteria. These are generally similar to the requirements imposed on molten salts as fuels for fission reactor fuels (1). It must not adversely interact with neutrons necessary for breeding of tritium. It must be a good heat transfer fluid and its heat transfer and hydrodynamic behavior must (for most applications) be adequate in the presence of large magnetic fields. It must be non- corrosive toward metals of construction in the blanket region, the pumps, and the power generation equipment. It must be suitably stable to the intense radiation fields within the blanket and it must not react violently if, upon failure of the heat transfer equipment, 1t is inadvertently mixed with the power-generating fluid (steam, potassium, etec.). It should, in addition, possess a relatively low vapor pressure 1o minimize stresses on the blanket structure, and it should be compatible with auxiliary blanket subsystems such as neutron moderators (graphites) or neutron multipliers. Finally, this fluid should also be capable of efficient 1Tt is possible, at present, to visualize a blanket in which tritium is bred in (essentially) stationary bodies of lithium metal, lithium alloys, or lithium salts, with the necessary cooling accomplished by another fluid--molten sodium, pressurized helium, gas, or molten salt. generation of tritium and it should be of a nature such as to permit easy recovery of the tritium for return to the fusion cycle,. This paper attempts to assess the potential of molten salts as blanket fluids in controlled thermonuclear reactors. As matters stand at present, several entirely different types of CTRs are potentially feasible. This fact coupled with the possibility (even the likelihood) that the coolant and the breeding functions are separable makes 1t impossible to examine in a single document all the credible combinations. We have chosen instead, in what we hope will prove to be a continuing examination of the overall problem, to do the following: 1. To assess the problems for the most difficult case, a single molten salt coolant and breeder blanket in a closed magnetic field device. We will use molten Li,BeF,, probably the most studied and best understood of the possible candidates, for this assessment in a hypothetical Tokamak; g Stellerator would be equivalent in nearly every regard. 2. To compare several molten salts for this and other less demanding possible CTR uses, 3. To consider, briefly, application of molten salts for a laser- powered fusion device whose problems are quite different, and finally 4, To compare and contrast these molten salts with molten lithium as the blanket fluid. BEHAVIOR OF Li,BeF, IN A HYPOTHETICAL CTR The most difficult problems which must be surmounted by a molten salt in a CTR blanket system are, almost certainly, those imposed if the salt must serve as the single-fluid blanket-coolant for a closed magnetic field (and essentially steady state) CTR. The precise magnitude of these problems will depend upon the detailed design of the device--1ts power level, size, temperature and temperature differential, and coolant passage patterns within the blanket region. To illustrate the nature and general magnitude of the problems we have chosen a toroidal CTR, blanketed and cooled with molten Li,Be¥,, of the size and characteristics shown in Table 1.° The blanket dimensions are essentially those proposed by 2This design is a hybrid from sources indicated in the text. It is pre- sented only for illustration of several problems to be described. Fraas (2) for a lithium-metal cooled device; the magnetic filelds are those indicated in an Osk Ridge National Leboratory study (3); flow rate and temperature differential are scaled from a proposed 1000 MW (e) molten salt fission reactor (4). It is clear from Table 1, that the efficiency of the device (ca 45%) presupposes temperatures above 7OOOC for the coolant outlet and that mean residence time of the fluid within the active blanket region is (since only a small fraction of the inventory is in the pumps, piping, and heat exchangers) approximately five minutes. It is assumed for this particular device that the Li,BeF, is pumped through the blanket region (and through the enormous magnetic field) at 120 cubic feet per second and directly to equipment for generation of high quality steam. These assumptions, as to pumping rates, AT, and steam generation compatibility, are not unlike those proposed for molten salt breeder (fission) reactors (MSBRs) (4). Table 1. Characteristics of a Hypothetical Controlled Thermonuclear Reactor Power 2250 Mw{(th) [1000 MwW(e)] Ma jor Diameter 21 meters Minor Diameters: First Wall 7 meters Blanket Region 9 meters Shield Region 11 meters Magnetic Fields: Maximum Toroidal Field at the Coils 80 kgauss Toroidal Field at the Center of the Plasma 37 kgauss Poloidal Field (pulsed) in the Plasma 7 kgauss Blanket Characteristics: Salt (LiyBeF,) 60% Graphite (or Be) 40% Inventory of LipBeF, 2 x 10° kg Flow Rate 4 % 10° kg/min (120 ft?/sec) AT 167°C (300°F) It is worthy of note that the mean radistion load on the Li;BeF, (ca 1.1 watts/gram) is more than 10-fold below that proposed for MSBRs. This comparison is, however, deceptively favorable toward MSBRs in that - these reactors can assure moderately uniform radiation levels within their fuels while CTRs cannot do so for their blankets. It is likely that radiation densities in CTR blanket near the plasma-confinement (first) wall will be larger than those proposed for the MSBR. Thig CTIR radiation density is not, however, likely to approach the maximum radia- tion density at which molten salts have been tested (5). Such a device as that described above will, on the other hand, pose several problems for the Li,BeF,. Some of these problems are similar and some are quite different from those posed by use as fuel solvents in fission reactors. These problems are defined and discussed in the foliowing sections. Effects of Strong Magnetic Fields Pumping a conducting fluid into (across the magnetic field lines of) a closed magnetic field device poses a problem. For a liquid metal this problem manifests itself as a large pumping power loss due to magneti- cally induced turbulence; for molten salts the effect appears as chemical destabilization. After the fluid is within the magnetic field one can, in principle (and perhaps, with considerable difficulty, in practice) make the flow channels conform closely to the magnetic lines of force. In that case the magnetic field may exert a pronounced effect upon the fluid dynamics of the flowing strean. Effects on Chemical Stability From electromagnetic theory it is known that the electric field induced in a conducting fluid crossing a magnetic field is given by the cross-product of fluid velocity and magnetic field: e R =VxB Molten LisBeF, (or any conducting fluid) flowing at 10 meters/ second in a pipe of 5 cm diameter with its axis aligned perpendicular to the lines of force of an 80 kgauss (8 volt-sec/meterz) magnetic field will have induced, at right angles to both the magnetic field and the flow direction, a potential difference of 4 volts between the salt and the pipe wall. Potential differences of such magnitudes are clearly intolerable; though LiF and BeF, are both very stable compounds (6) an induced voltage such as this (equivalent to destabilization by 92 kcal/ mole) would make these compounds quite corrosive to the metallic tube walls. Homeyer (7), who seems to have been the first to consider such electrolytic corrosion in a CTR blanket system, noted that such corro- sion should be alleviated by (a) reducing fluid velocities perpendicular to the magnetic field, and (b) using a series of parallel pipes to reduce the pipe dimension where flow across the magnetic field lines are necessary. If, for example, the 120 ft?/sec flow of Li,BeF, required of our hypothetical 1000 MW(e) CTR were supplied at a flow rate of 4 meters/sec through pipes of 4 cm diameter perpendicular to the 25 kgauss field?® the emf induced in each pipe would be 0.4 volts; some 675 pipes would be required. If these pipes penetrated the field at 300 to the field lines this emf would be reduced to 0.2 volts. These conditions would seem to be tolerable. Alternatively, by supplying external cooling to each pipe, as it crosses the magnetic field lines, so as to form a poorly conducting layer of frozen salt on the inner pipe wall, it may be possible to reduce the number of pipes and to somewhat increase their size. Periodic replacement of corroded pipe sections might also be considered, since they are located at the periphery of the torus.* Plasma stability in a reactor such as this requires a pulsed poloidal magnetic field transverse to the main field and, accordingly, perpendicu- lar to the fluid flow parallel to the fuel lines of the main (toroidal) field. Chemical effects of this poloidal field (which may reach 7 kgauss) may not be trivial, but they would seem, in general, to be tolerable. 3This is roughly the maximum field between the coils (3) in a pipe entering the outer edge of the torus. “Other alternatives exist. It should be possible to penetrate the field by shafts of mechanical pumps to permit use of LiyBelF, to transfer heat to (for example) a boiling potassium cycle within the magnetic field. Such alternatives are beyond the scope of this document. It is clear that the problems outlined above deserve experimental study especlally in the area of kinetics of de- and re-stabilization of real fluids upon passage through intense magnetic fields. Effects on Fluid Dynamics To avoid induced electric fields as discussed above, flow of the blanket fluid within the torus will be aligned with the magnetic lines of force. However, in that case the magnetic field will exert a force opposed to eddies within the fluid and will tend to damp turbulent flow. Heat loadings in the blanket structure of a CTR, and especially at the first (plasma-confining) wall, will be very large, and molten LisBely, must develop turbulent flow if it is to ccol this wall effectively. It is important, therefore, to assess this damping effect of the field on turbulent flow in the salt. No experimental study of magnetic damping in molten salt has been reported, but experiments with liquid mercury have been performed (8,9). These experiments with mercury appear to provide a means of estimating the Reynolds number at which the transition from laminar to turbulent flow occurs. A dimensionless quantity which characterizes the magnetic forces that affect flow is called the Hartmann number (M), and is defined: M = Bfl(c/n)% where B is the applied field, £ 1s a characteristic length, usually the half-width of the flow channel, o is the fluid's electrical conductivity, and n is the viscosity In the absence of a magnetic field the laminar-turbulent transition occurs at Reynolds number (RO) of about 2200; in the presence of the field this transition occurs at a higher Reynolds number (Rt)' This increase (more commonly the ratio (Rt/Ro) is a function of the Hartman number (M). Three such correlation functions have been published (8,9,10). For Li,BeF, at 600°C, the electrical conductivity (11) is 220 (ohm-meter) "1, the viscosity is & x 10~° kg/(sec—meter). Accordingly, for this material in a 80 kgauss (8 webers/meterz) field and assuming a 6 cm (3 cm half-width) channel the value of the Hartman number is 40. If we apply each of the three correlations (8,9,10) developed from experi - ments with mercury to a fluid of M = 40 we find the transition Reynolds number should rise from RO = 2200 to RJG = 2720, 3740, and 2400, respectively. The effect of magnetic field on this fluid is, therefore, predicted to be relatively small. These three wvalues for Rt are very much less than values estimated for flows in the first wall region. As an approximation let us assume that to cool the vacuum wall we require 40 ft3 per second of salt (one third the quantity required to cool the entire blanket) and that this salt flows through a 6 cm wide annulus around the 7 meter diameter first wall. The linear velocity of salt is 0.85 meters/sec, the density (11) (at 600°C) of Ii,BeF,; is 1990 kg/meter? and the viscosity is 8 centipoise. The Reynolds number under these conditions is about 13,000. These data strongly suggest that molten Li;BeF, can be made to flow turbulently within the blanket system, but direct experiments with molten salts would certainly appear desirable.’ Any transverse magnetic field will also act to suppress turbulence. Hoffman and Carlson (10) propose the formula, R, = 500M for calculating the transition Reynolds number of Mercury flowizg transverse to the magnetic field. The same formula applied to Li,BeF, at 6OOOC flowing in a 6-cm thick channel transverse to 8 kgauss (the poloidal field strength) yields Rt = 500M = 2000. This result suggests that an 8 kgauss poloidal field in a toroidal reactor may not affect turbulence in Li,BeF,; at most, the suppression of turbulence by this field will be comparable to the effects of the toroidal field. “Metallic lithium is expected to be quite different. The Hartman number for Li is 12,000 for the condition where that of LiyBeF, is 40; the corresponding R+ for lithium is, accordingly, above 700,000. It would appear that Li will be constrained to laminar flow; it seems likely, however, that (because of its good thermal conductivity) it can adequately cool the first wall. Production of Tritium It is obviously necessary to use the neutrons produced in fusion to breed the tritium required to fuel a D-T reactor. If the only sources of tritium were today's transmutation facilities then the fuel cost alone would be about 1.2 cents per kilowatt-hour.® The only practical neutron reactions which will yield tritium sufficient for the needs of a D-T fusion reactor are ®Li(n,Q)T and 7Li(n,on')T. The latter is a high-energy reaction with a threshold at 2.5 MeV. Neutron-capture cross sections of 6Li become significant only at energies of 0.5 MeV and less. The ’Li reaction is particularly favorable since the product neutron (n') can react with °Li to yield a second tritium atom, but because the blanket contains elements which scatter and absorb high-energy neutrons, the production of tritium from 7Ii is not very efficient. When the lithium in the blanket is in natural isotopic abundance (92.58% 7"Li, 7.42% ®Li), the greater fraction of tritium is produced from °Ii. Several studies have been reported concerning the breeding of tritium in fusion reactor blankets (12-16). In cases where the lithium is in natural isotopic abundance, tritium-breeding ratios clearly greater than unity are calculated for lithium metal blankets; the breeding ratios for lithium salts are distinctly less favorable. Table 2 presents some very recent calculations for four salts and for lithium metal all in the same blanket configuration. ©Some of the neutron cross sections, partic- ularly those for the n,Q, and n,n'Y reactions of fluorine (17), are uncertain. The values in Table 2 are, therefore, more useful for com- 7 yncertain- parison than for accurate prediction of tritium production; ties of perhaps lO% in the calculated breeding ratios are possible. Three options, either alone or in combination, can be considered for upgrading tritium production when the breeding ratio is marginal as appears to be the case for L12B6F4.8 These are: ®Based on a tritium cost of 10 cents/curie and 22 MeV/fusion utilized in a plant operating at 40% thermal efficiency. 7It should be noted that Blow, et al. (14) calculate a tritium breeding ratio of 1.027 for Lip,BeF, in a blanket assembly similar to that of Table 2. 8possible use of LiF, LiCl, and Li,CC3 is discussed briefly in a subse- quent section of this report. 10 Table 2. Tritium Breeding Calculations®’® Breeding Ratio® from Total 3 Breeding Coolant oLi "Li Ratio LiF (850°0) ¢ 0.80, 0.24, 1.05 LiyBeF, (850°C) 0.785 0.14, 0.93 Li,C00; (900°K) 0.64, 0.16+ 0.81 LiCL (900°K) 0.61, 0.13, 0.75 L1 (850°¢C 0.98, 0.45, 1,44 &p. Steiner, Osk Ridge National Laboratory, personal communication, March 1972. bBlanket Configuration: (1) first wall - 0.5 ecm Nb, (2) 94% coolant, 6% Nb - 3 cm, (3) second wall - 0.5 cm Nb, (4) 94% coolant, 6% Nb - 60 cm, (5) graphite - 30 em, (6) 94% coolant, 6% Nb - 6 cm. CDefined as tritium atom produced per fusion neutron incident on the first wall. dLithium in natural isotopic abundance. eTem.perature at which atom densities were calculated. (2) design of blanket to include a region of metallic lithium, (b) increasing Be content of the blanket by adding a region of Be or Be,;C to increase neutron multiplication and to provide more 61,1 ———94§—§§9——> SLi, and ZBe(n,a) SHe (¢) modest enrichment of the blanket material in °Li. The second option was briefly treated by Bell (16) who showed that if a blanket region (40 cm thick adjoining a 1 em first wall of molybdenum) were changed from Li,BeF, to an equal thickness of Be and LipBeF,, the tritium-breeding ratio would increase from 0.95 to 1.50. The third option has also received some attention. Impink (13) reported that small increases in 61.i enrichment of the Li,BeF, blanket led to modest gains in breeding ratio. For example, increasing the 811 isotopic fraction to 0.2 in a 6.25-cm thick coolant region next to the first wall improves the total breeding ratio by about 3%. Although 11 enrichment costs are high, these costs would be partly offset by improved shielding of the magnhet coils and by reduced radiation damage to the first wall through reduced resonance capture (13). In light of present knowledge of the pertinent cross sections, it appears that the breeding capability of Li,BeF, is marginal in devices such as our hypothetical torus. This material would, therefore, probably need to be augmented by one of the methods indicated above, or by other means. It is clear that better cross section data are needed so that this point can be decided. Recovery of Tritium Approximately 270 grams of tritium are consumed per day by fusion in a2 2250 MW(t) D-T reactor. Slightly more than this, or approximately 300 grams per day, must be produced and recovered; this corresponds to some 50 moles of T, or to 100 moles of TF per day. 1In our hypothetical CTR the mean residence time of the fluid in the blanket is 5 minutes per cycle. Some 0.174 moles of Ts, or 0.348 moles TF is, accordingly, produced in the Li,BeF, in this interval. If the fluid entering the blanket confiained no tritium species the fluid emerging from the blanket will contain about 1.74 x 1077 moles T, (or alternatively about 3.48 x 1077 moles TF) per liter if complete homogeneity is assumed. The problems in recovery and management of the tritium depends significantly on whether the material exists as T, or as TF. These two situations, and the extent to which the mode of tritium behavior can be controlled, are briefly described in the following. Sclubility of H, in molten LisBel, has been shown to increase linearly with pressure of Hy; at lOOOOK, the solubility should be near 7 x 107° moles H, per liter of salt per atmosphere of Ho (18). No studies of tritium solubility have been reported. If the bred tritium occurs as Tp, and if the solubility behavior of T, and H; are similar, the emerging blanket fluid carries T, (generated during its pass through the blanket) equivalent to a saturating pressure of about 2.5 x 1073 atmospheres. Equilibration of the emerging salt with a relatively small volume of inert gas will result in stripping of a very large fraction of this dissolved T, from the salt. It is clear, however, that this process 12 (with 120 ft3 of salt and, for example, 1 ft° of He) will be difficult to engineer, and, moreover, that diffusivity of T, at such effective partial pressures through hot metal surfaces will pose problems. The solubility of HF in molten Li,BeF, also depends linearly on pressure of the solute gas, and its Henry's law constant is 1072 moles HF per liter of salt per atmosphere HF at 1000°K (19). The TF produced during each cycle of coolant through the blanket region will correspond to about 3.5 x 1077 moles TF per liter of Li,BeF,; this is equivalent to a saturation pressure of about 3.5 x 1077 atmospheres of TF. The TF will be more difficult to strip from the salt than will T,, but TF will not diffuse through the metal walls. If its reaction with the metal walls can be sufficiently minimized, the TF concentration can be allowed to increase and the rate of processing the blanket fluid can be corres- pondingly reduced. If, for example, the TF can be allowed to concentrate until its pressure is 10~> atmospheres, sparging of perhaps 5 ft2/sec of the fluid with helium should suffice for effective recovery of the bred tritium. It is, accordingly, worthwhile to examine whether the bred tritium can reasonably be maintained as TF. Tritium produced, for example, from SLiF +n — %He + T + F~ is, in principle, born as an oxidized species. The tolerable concentra- tion of TF, or of any other oxidized species, will, of course, be limited by the extent to which corrosive reaction with the CTR metal can occur. If the containment metal is sufficiently inert, useful concentrations of TF can be maintained without appreciable reaction. By way of i1llustration, let us examine the reaction of HF with nickel, HF (g) + iNi(c) = 1H,(g) + $NiF,(d) where (g), (c), and (d) indicate, respectively, that the species is gas- eous, crystalline solid, or dissolved in molten LisBeF,. From the data of Table 3 AG° = 10.9 keal for this reaction at 1000°K. The equilibrium constant is given by . . X ='§§£Ea—ifi3 = 4 x 1072 o Fur 13 where N is the mole fraction of dissolved NiF,, a is the activity of nickel (unity in this case), and P is the partial pressure of the desig- nated gaseous species. If we set NNiF2 = 3.2 x 107 (equivalent to 6 parts per million of Ni2+, a value that seems likely to be tolerable) we calculate pressures of H, of 6 x 10-° atmospheres and 5 x 10-% atmos- | pheres, respectively, in equilibrium with HF pressures of 3.5 x 1072 and 10=3 atmospheres. These results suggest that if the CIR metal were Ni a very large fraction of the tritium could be maintained as TF and stripped as such. Examination of Table 3 suggests that the situation may be even more favorable for molybdenum and, perhaps, for tungsten as the containment metal. However, if the containment metal were iron, chromium, niobium, tantalum, titanium, or, probably, vanadium the tritium must, of necessity, be stripped and handled as T,. If one of these more reactive metals proves necessary as the CTR material, some way of preventing corrosion due to xIF + M - MFX + %T2 must be provided. This would seem to be possible by incorporation of a redox buffer (described in more detail in the subsequent section) in the molten Li,BeF,. At this stage in the technology of fusion reactors one should probably not dismiss the possibility of using stainless steel or a chromium-containing nickel-based alloy at temperatures at or below 1000°K. Such materials can, perhaps, be coated with molybdenum, tungsten, or nickel, by electrodeposition (22) or by plasma spraying (23). Chemical Transmutations Several types of chemical transmutations will occur in molten Li,BeF, in its service as the blanket fluid in a fusion reactor. The most important of these, and means for maintenance of the blanket to minimize or avoid their deleterious effects, are the following: Transmutation of lithium is, of course, essential to production of tritium. The overall reactions can be represented as TiF +n —» %He + TF + n', and 6LiF + n — %He + TF. 14 Table 3. TFree Energies of Formation of Fluorides s 259 5009k (kcal/g-atom of fluorine) Reference MoFg (g) ~50.2 (20) W (2) ~56.8 (&) NiF, () ~55,32 (20) VFs(g) _58° (21) VF, (cr) 5602 (21) HF () -66.2 (6) FeF, (@)% -66.5% (20) NbF5(g) -72.5, (20) CrF, () -75.2 (20) TaFs (g) -82.2 (20) TiF, (g) -85.4 (6) LiF (¢) -125.2: (20) BeF, (£) -106.9 (20) ®Standard free energy of formation in molten LiyBeF,. bEstlmated from the relatlon, Aglooo = AHggg - 1000 (¢ 8298) taking AHggg and most 889g from NBS Technical Notes 270-3,4,5 (21). Other S%9¢g estimated from analogous compounds. These reactions are not inherently oxidizing or reducing, though, as described in the previous section, the generated TF can oxidize reactive structural metals to form metal fluorides which will dissolve in the melt. Transmutation of beryllium (as BeFz in Li,BeF;) leads to corrosion of any system metal since disappearance of Be2+ is equivalent to release of fluorine. The two reactions may be represented as: BeF, + n — 2n + 23He + 2F (or Fp), and BeFy, + n — 2He + 8He + 2F, followed by S e 0.8 sec half life > §Ti, and 81i + F —» °OLiF. In our 2250 MW reactor, these reactions yield, respectively, the equivalent of 500 g and 70 g of fluorine per day. This problem is generally similar to that encountered in fission of uranium (as UF,) in the MSRE ) ; 15 it is clearly necessary 10 provide a redox buffer in the molten salt (the UF3-UF, couple does this in fission reactors), capable of oxidizing FO to F~. It is also necessary, if Ni, Mo, or W constitutes the container system, that this redox buffer be consistent with maintenance of the tritium as TF. The couple Ce3+ = CetT may possibly serve this function. If, for example, the concentration of cerium in the melt is set at 10~% mole fraction the blanket will con- tain 6 x 10% mole of Ce3% + Ce*?, and the Ce?*/Ce*™ ratio would require chemical adjustment on a cycle time of many days. If, on the other hand, the container metal is Nb (or some other metal which will reduce TF in dilution solution) the redox couple must be chosen so as to be consider- ably more reducing. It must deal with the F, generated by transmutation of beryllium but it must also reduce the 100 moles per day of TF produced by transmutation in the LiF. Such a buffer system would require adjust- ment on g cyecle of a few days. In addition, transmutation of fluorine occurs upon capture of neutrons of energy above about 3 MeV. This reaction may be represented by 18F" +n - 18N” + %He. This nitrogen isotope decays, with a 7.3 sec. half-life, to an oxygen isotope 16y~ o 160 4 g7, and the result is probably, although the mechanism may be complex, grow- in of 0°”., The asbsolute quantity of 1°N formed by this reaction is relatively uncertain; it is estimated to be, within a factor of three, 120 grams/day. The very short half-life of this isotope guarantees that all the °N decays within the CTR blanket. The concentration of 16N, in whatever chemical form, within the Li;BeF, cannot exceed 1.1 parts in 1011, However, some fraction of this material will react with the CTR containment metal; decay of this isotope will lead to formation of metal oxide in the CIR metal. This may, especially if it concentrates within the grain boundaries, prove troublesome. If all the 1oy~ decayed within the blanket salt, the oxide concentration of our hypothetical CIR would increase about 60 parts per billion per day. Since 10 to 50 16 parts per million of oxide is almost certainly tolerable, a process for removal of oxide on a cycle time of several months to several years should suffice. Finally, it should be noted that the transmutation reactions shown all generate He. For the hypothetical CTR the daily production of helium is gbout 125 gram atoms or nearly 100 standard cubic feet. Helium is relatively insoluble in molten Li.BeF, (24); the solubility at 1000°%K is 1.7 x 10™% moles He per liter salt per atmosphere. Helium produced per pass of blanket salt corresponds to a saturation pressure of 2.6 x 1072 atmospheres. If no sparging were attempted the helium pressure would reach 1 atmosphere in about 30 hours. Compatibility of LipBeF, with CTR Metals and Moderators As indicated in Table 3 above, LiF and Bel; in molten LipBel, are very stable materigls. Both are much more stable than the structural metal fluorides; consequently, corrosion due to chemical reactions with these major blanket constituents should prove minimal. Indeed, experi- ence with the Molten Salt Reactor Experiment (25) has shown negligible corrosion by this fluid on a nickel-base alloy (Hastelloy N). However, such salts are excellent fluxes for metallic oxides and halides, and films of such substances afford no protection against oxidizing agents carried by such melts; accordingly, as described above, HF (or TF) may react with the containment metal, and impurity ions such as Ni?+ will react with metallic iron or chromium in the container metal (1). Melts such as Li,BeF, are chemically inert toward, and do not wet, graphite (1). However, the possibility that such salts will transfer graphite and carburize metals such as Mo or Nb cannot be discounted. It is not likely that a system built of Mo, Nb, or V can use molten Li,BeF,; and unclad graphite without adverse interactions. Similarly, metallic Be cannot react appreciably with LisBeF, (but the Be could certainly react with TF or with the CeB+/Ce4+ couple proposed as & redox buffer in the system). Any real use of metallic Be as a neutron multi- plier in the blanket system, therefore, presupposes that the Be is clad with an inert metal. 17 Other processes which could conceivably give rise to corrosion can be dismissed as highly improbable. Direct dissolution of structural metals in Li,BeF, has never been observed. ©Salt decomposition caused by the slowing down of energetic particles should not lead to corrosion provided that the salt is kept at elevated temperatures. Experience gained in the Molten Salt Reactor Experiment (26) and in an extensive in-pile radiation testing program showed that as long as the temperature was greater than 15000 (27), radiolytic decomposition was of no impor- tance to corrosion of structural metals or graphite. Compatibility with Steam, Air, and Liquid Metals In any system of heat-exchangers and hot flowing liquids, there exists a real and finite probability that leaks will occur. In this section we examine the consequences of leaks and intermixing of other fluids and LipBeF,. The reaction of steam with Li,BeF, yields HF and BeO HoO(g) + BeFp (1) = BeO(e) + 2HF(g) though the reaction is not particularly exothermic. Both H,0 and HF are likely to corrode the metal in contact with the salt; corrosion-product fluorides will dissolve or be otherwise carried by the salt. Since BeO is only very slightly soluble (125 ppm at 500°¢) in Li,BeF, (28), a large in-leakage of steam would soon lead to the precipitation of BeO in the salt circuit. Leakage of air into LiF or Li,Be¥, will have trouble- some, but not hazardous, consequences. Dry air will not react directly with either salt; however, air oxidation of surfaces in contact with the salt will result in dissolution by the salt and, if continued, in ultimate precipitation of BeO. Moisture in the air will also react, as does steam, with LisBeF,. The molten LisBeF, can, if necessary, be freed of oxide by treatment at elevated temperatures with anhydrous HF (29). In some CTR designs suggested in g subsequent section of this paper, LiyBeF, (or other salt) could inadvertently be mixed with liquid alkali metals. From LipBeF,, metallic Li, Na, or K react to precipitate Be metal, but the reaction is not highly exothermic. 18 In general, although inadvertent mixing of Li,BeF, (or most other molten salts) with other CTR fluids would prove troublesome, such mixing would not lead to violent or explosive reactions. CHOICE OF MOST PROMISING SALTS In this section we attempt to answer two questions. These are: (8) if tritium must be bred in the blanket-coolant which lithium-bearing salt is best, and (b) if the coolant and breeding function of the blanket can be separated which are the most promising molten salt coolants? In answer to the first question, it must be conceded that obtaining breeding ratios greater than unity with molten salts alone may pose a real diffieulty. Table 2 above suggests that LiCl and Li;CO3 show, in reasonable (though not optimized) blanket configurations, breeding ratios that are unsatisfactory. Breeding ratios have also been calculated for LiNO, (13) and LiNO; (15); the results tend to be gquite unfavorable. Impink (13), for example, obtained the value 0.82 for LiNO,.? No calcu- lations appear to have been made for Li,S0O,, but the high cross sections for S(n,x) and S(n,p) reactions almost certainly will reduce the breed- ing ratio below that for T.i,CO03. Moreover, LiNO, and LiNO3; lack the thermal stability required of truly high temperature coolants, and Li,CO04 and LiSO, will oxidize many CIR structural materials. Lithium hydroxide seems to be eliminated, even if (as is unlikely) its properties are otherwise satisfactory, because its hydrogen would excessively dilute the bred tritium. Lithium oxide (1Li,0) has a lithium density nearly 50% above that of metallic lithium, but its melting point of nearly 1470°C (6) eliminates it as a major constituent of a blanket fluid. Lithium chloride melts at 610°C (6) and should be reascnably compatible with CTR metals, but its breeding ratio (see Table 2) appears inferior. | The salt with the most favorable breeding ratio is LiF. This salt is inert toward graphite and to metals under consideration for the blanket structure. The major drawback of LiF is its melting point of 848°c. °Tn the same configuration he calculated 1.15 for Li,BeF,. 19 Because of this high melting temperature, LiF cannot be used to transfer heat to the steam system of the reactor. If the blanket region were oper- ated at very high temperatures (>9OOOC), then LiF could be used in conjunc- tion with an intermediate heat-exchange medium--liquid Na, a lower melting salt, or perhaps a boiling alkali metal system. The melting point of LiF can be substantially lowered by many solutes; the ideal solute should lower the melting point below 374OC10 without affecting either the breeding gain or the generally favorable heat-removal and chemical properties of LiF. We know of no such solute. Dissolved Lio0O should increase the breeding ratio slightly, but considering the probable limited solubility of LisQ, the melting temperature (more accurately, liquidus temperature) of LiF will not drop below 800°C. Using AlF; and/or another slkali fluoride to lower the melting temperature to ~700°C should not have dire chemicsal conseguences, but the breeding ratio will almost certainly suffer. The nearest approxi- mation to an ideal solute in LiF is probably BeF,. The phase diagram for LiF-BeF, (30), presented in Fig. 1, shows that a melting temperature as low as 363°C is available in this system. Unfortunately, the viscosity of the melt increases with BeF, concentration, and mixtures with >40 mole % BeF; have viscosities greater than 50 centipoise at 450°C (31). The optimum salt mixture of low melting temperature and acceptable viscosity, and with a reasonably good tritium breeding ratic is at ~33 mole % BeFp, correspond- ing to the LiyBeF, used for illustrative purpcses in earlier sections of this report. Decreasing the BeF,; concentration below 33 mole % may have a modest beneficial effect upon breeding ratio; this increase might, possibly, offset disadvantages posed by the increased liquidus temperature and likely changes in chemical behavior. In summary, the answer to the first question posed above--the best blanket ccolant salt in which to breed tritium is LiF, but its melting point of 848°C limits its usefulness only for cooling a blanket that operates above this temperature. If the blanket coolant must also trans- fer heat to the steam system, Li,BeF,, or some modest variant of this composition, appears tc be the best choice. A partial separation of breeding and cooling functions, posed in the second question above, has been approached by Steiner (12). He calculated 10The critical temperature of H,0. TEMPERATURE (°C) 20 ORNL-DWG 71-5270R2 l I | [ [ I I | 900 — 500 ] T T T — /848 X (EUTECTIC) = 0,3280 0.0004 ,%Tmu, =459,1+0.2 800 . 450 LIQUID ] — X (EUTECTIC = 0.531 £0.002 700 400 |- Li,BeF, + — LiF + LIQUID LIQuID 600 350 ] 1 ! | 585 - 030 0.35 0.40 0.45 0.50 0.55 so0 — R —— 458.9+0.2°C [ BeF, (B-QUARTZ TYPE) | + LIQUID | 400 | — I 363.5 +0.5°C LiF + e — . : LiBeF3 + Befp LizBeFy LipBefy + BeF, (8-QUARTZ TYPE) 3 3| LigBeFa + u”| LiBeF, + Bef, (B-QUARTZ TYPE)} / LiBeFs @ 227°C 200 | l | I 2 | i 1 e / 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 XBon {mole fraction) Fig. 1. Phase Diagram of the System LiF-BelF,. 21 the tritium breeding ratio for a blanket design in which LiyBeF, cooled the vacuum wall and lithium metal assumed the rest of the heat-transfer function, and showed that the breeding ratio (1.22) was 8.3% less than that (1.33) for the same blanket with lithium as the sole coolant. This small loss in breeding ratio suggests that other salts, especially those melting below 37400, might also serve as vacuum-wall coolants. Moreover, the heat carried by the lithium might then be transferred ocutside of the blanket to a molten salt of the same composition as the vacuum-wall coolant. The salt streams from the vacuum wall and the lithium heat-exchanger would then be combined and pumped to the steam-raising system. The eutectic mixture of TiCl and KCl, which melts at 354°C (32), might be suitable for such service. The authors are presently assessing the implications of this "double-coolant"” concept. In principle, a complete separation of breeding and cooling functions might be embodied in a blanket design in which a molten salt cools the vacuun wall, a moderately thick region of quiescent lithium metal, and a 11 The coolant salt for this application graphite moderator-reflector. must have great chemical stability (to avoid deleterious destabilization in the magnetic field and to avoid corrosion of the CTR metal), sound heat transfer properties, and, preferably, a low freezing point. These specifications narrow the choice of major component to fluorides, chlorides, and oxides, of lighter alkali and alkaline earth metals which melt below 1200°C. The list seems to contain Li¥, LiC1, NaF; NaCl, Na,0, KF, KC1, BeF,, MgCl,, and CaCl, with, perhaps, other compounds of these families as possible minor constituents of mixtures. When one adds the further requirement that the coolant not lower the breeding ratio below unity, the list of useful major components becomes smaller. The high inelastic scattering cross sections for Mg, for example, probably eliminates any substantial concentration of MgCl, from the coclant. From consideration of this list of materials it would appear that LiF is best except for its high melting point, that Li,BeF, may well be the best overall choice, and 117 possible rationale for such a design is that lithium, though providing a comfortable breeding ratio, cannot be made to flow turbulently within the Wlanket and may require excessive power in being pumped through the major magnetic field. 22 that, if subsequent calculations show that the breeding ratio does not suffer unduly, the ternary eutectic LiF-NaF-KF (melting point 454OC) and the binary eutectic LiCl-KCl may be suitable coolants. Practical problems with these materials will differ in detail from those described earliier for Li,BeF,. Superficial examination of these problems reveal none that seem insuperable, but much experimental study would be necessary before use of these materials could be assured. MOLTEN SALTS IN LASER-INDUCED FUSION REACTORS Two major uncertainties in use of LisBeF, in CTRs such as our hypo- thetical device stem from (a) potentailly troublesome interacticns with the large magnetic field and (b) the fact that the first (plasma-contain- ing) wall and blanket structure degrade the neutron spectrum so that the r’\’Li(n,0:'-1’1')'.[' reaction is reduced and tritium breeding becomes marginal. It is, accordingly, a matter of some interest to examine briefly the potential of molten salts in a CTR device which possesses neither a magnetic field nor a first wall. Lubin and Fraas (33) have described a device in which pellets of deuterium and tritium produce a plasma upon ignition by an energy pulse from a suitable laser. The blanket-coolant fluid (Lubin and Fraas pro- posed metallic lithium) is pumped through the reaction vessel and through external power generation equipment. The liquid is pumped into the (essentially spherical) reaction vessel tangentially to provide a very rapid swirl; ignition of the pellet occurs in the vortex so formed on the vertical center-line of the vessel. The plasma generated in this pulsed device requires no magnetic containment, and the blanket coolant liquid is exposed directly to radiation from the plasma. In addition to its coolant and breeding functions such a liquid must also provide attenuation of the severe shock waves sufficient to assure feasibility of the reactor vessel. Lubin and Fraas proposed 10 assist this function by introduction of (compressible) gas bubbles into the swirling liquid. These shock waves are caused by (a) partial conversion of neutronic energy into mechanical energy within the liquid, and (b) deposition of x-ray energy in the liquid at the vortex surface. The former is by far 23 the dominant perturbing forece (34). According to an analysis by Dresner (34), the impulse to the vessel wall due to the neutronically induced shock is proportional to OuCp™t Ryt where O is the volume expansivity, p is the sonie velocity, Cp is the specific heat, and Ry is & neutron attenuation distance which depends upon neutron scattering and absorption reactions. The values of these four quantities for Li,BeF, and for Li are: Li,BeF, (600°C) Li (600°C) a (°c1) 2.4 x 107% (11) 2.1 x 1074 (35) L (cm sec™l) 3.0 x 10°* 4.3 x 10° (36) Cp (erg g~* °¢~1) 2.4 x 107 (1) 4.2 x 107 (35) R, (cm) 23 (34) 33 (34) SEetimated. This analysis suggests that the wallsg of a vessel filled with molten LigBeF, will suffer an impulse almost twice that of an identical vessel filled with lithium metal. This conclusion, coupled with the fact that enhanced centrifugal forces (Li,BeF, is nearly four times as dense as Li) will make suspension of gas bubbles more difficult in the salt, would seem to place LisBeF, at some disadvantage.l? It seems apparent, however, that design and construction of this vessel to withstand repeated shocks over a long life will pose formidable problems. Metallic lithium at temperatures of 500°C and above is likely to prove compatible with relatively few (and generally expensive and exotic) materials. It is possible that Li,BeF, (or other salts), which are compatible with a much wider spectrum of metals may have real advantages in easing this difficult design problem. Problems with chemical transmutations and with recovery and manage- ment of tritium seem generally similar to, and should be handled by methods like, those described above for the hypothetical toroidal device. It 12pifferences in liquid properties will probably be relatively unimportant in attenuating the shock waves. The entrained bubbles almost certainly will be the principal shock-absorbers. 24 seems likely that the ease of tritium recovery may give the molten salts an additional advantage. Finally, it should be noted that the absence of the first wall leads to decidedly improved breeding. Metallic lithium will still prove to possess the highest ratios, but it seems certain that LipsBeF, will have values markedly above unity. Indeed, it is likely that several lithium- bearing salt compositions would be possible breeders in this sort of laser-powered CTR. SUMMARY: GENERAL COMPARISON OF MOLTEN SALTS WITH LITHIUM IN FUSION REACTORS In summary, and to supplement the several preceding discussions, we briefly compare liquid lithium with salts (especially Li,BeF,;) in several regards. Lithium metal is clearly superior to any molten salt in breeding of tritium; this seems certainly true in any CTR embodiment. Certain pro- posed designs, of which the laser-powered devices are the best examples, can certainly breed sufficient tritium using molten salts alone. However, the indication that tritium breeding is marginal for the salts in some (if not most) designs represents the worst drawback to their use. Fluo- ride salts are glso inferior to lithium in that, primarily because of fluorine's relatively high cross-section for ineleastic neutron scatter- ing, the salts are more intense gamma sources and cause increased gamma heating of the vacuum wall (12). Tritium recovery should prove considerably simpler if molten salts are used; this is particularly true if the tritium can be maintained as TF to minimize diffusion through metallic walls. Several of the physical properties of lithium (thermal conductivity, specific heat, viscosity and melting point, for example) are superior to those of the molten salt. However, since the magnetic field will prevent turbulent flow in the blanket for lithium, but not for the salt, it may be that the molten salt is a better heat transfer medium for such a CIR. Lithium is compatible with relatively few structural metals, niobium alloyed with 1% zirconium appears to be a suitable container (35). More- over, lithium reacts with graphite, and this material must certainly be 25 clad if it is to serve in a lithium-cooled blanket assembly. Molten LisBeF, is compatible with graphite and with a wide variety of structural metals. Corrosion by the salt is possible, through interaction with strong magnetic fields, but such corrosion can apparently be avoided by careful design. Transmutations within the salt, which provide potential for corrosive reactions, can be accommodated by relatively simple means. Reactions of salts with steam or with air produce accelerated corrosion but, unlike similar reactions of lithium, lead toc no inherently hazardous conditions. It is clearly not possible at this stage of the technology to predict with confidence how the problems inherent in use of molten salts (or of lithium) will be solved. It is entirely possible that both lithium and molten salts will be useful. At any event, and regardless of the ultimate choice, it is clear that many fascinating chemical research and develop- ment ventures lie ahead. ACKNOWLEDGMENTS The authors are indebted~to Don Steiner for his calculation of tritium-breeding ratios, to Lawrence Dresner for his analysis of shock attenuation, and to R. A. Strehlow and W. K. Sartory for much helpful discussion. 10. 11. 12, 13. 14, 15. 16. 17. 18. 26 REFERENCES W. R. Grimes, "Molten Salt Reactor Chemistry,” Nucl. Appl. Tech. &, 137 (1970). A, P. Fraas, Conceptual Design of the Blanket and Shield Region of a Full Scale Toroidal Fusion Reactor, ORNL-TM-3096 (1972). M. S. Lubell et al., "Engineering Design Studies on the Superconducting Magnet System of a Tokamak Fusion Reactor,"” Proceedings of the 4th Conference on Plasma Physics and Controlled Nuclear Fusion Research, Vol. 3, p. 433 (1971), IAEA Publication CN-28/K-10. E. S. Bettis and Roy C. Robertson, "The Design and Performance Features of a Single-Fluid Molten Salt Breeder Reactor," Nucl. Appl. Tech. &, 190 (1970). | W. R. Grimes, "Materials Problems in Molten Salt Reactors,” in Materials and Fuels for High Temperature Nuclear Energy Applications, ed. by M. T. Simmad and L. R. Zumwalt, the M,I.T. Press, Mass. (1969). JANAF Thermochemical Tables, 2nd ed., National Stand. Ref. Data Ser. NSRDS-NBS 37, U.S. National Bureau of Standards, June 1971. W. G. Homeyer, Thermal and Chemical Aspects of the Thermonuclear Blanket Problem, Tech. Report 435, M.I.T. Research Laboratory of Electronics (1965). F. W. Fraim and W. H. Heiser, "The Effect of a Strong Longitudinal Magnetic Field on the Flow of Mercury in a Circular Tube," J. Fluid Mech. 33(2), 397 (1968). S. Globe, "The Effect of a Longitudinal Magnetic Field on Pipe Flow of Mercury," J. Heat Transfer, Trans. ASME, 445 (1961). M. A. Hoffman and G. A. Carlson, Calculation Techniques for Estimating the Pressure Losses for Conducting Fluid Flows in Magnetic Fields, UCRL-51010, Lawrence Livermore Radiation Laboratory, (1971), pp. l-4. S. Cantor et al., Physical Properties of Molten-Salt Reactor Fuel, Coolant, and Flush Salts, ORNL-TM-2316 (Aug. 1968). D. Steiner, "The Nuclear Performance of Fusion Reactor Blankets," Nucl. Appl. Tech. 9, 83 (1970). A. J. Impink, Jr., Neutron Economy in Fusion Reactor Blanket Assemblies, Tech. Report 434, M.I.T. Research Laboratory of Electronics (1965) . S. Blow, V. 8. Crocker, and B. O. Wade, "Neutronics Calculations for Blanket Assemblies of a Fusion Reactor,"” Nuclear Fusion Reactors (Proc. Brit. Nuclear Energy Soc. Conf. on Fusion Reactors, Sept. 1969) UKAFA Culham Laboratory (1970), p. 492. W. B. Myers, M. W. Wells, and E. H. Canfield, Tritium Regeneration in a D-T Thermonuclear Reactor Blanket, UCID-4480, Lawrence Livermore Radiation Laboratory (1962). G. I. Bell, Neutron Blanket Calculations_ for Thermonuclear Reactors, LA-3385-MS, Los Alamos Scientific Laboratory (1965). J. R. Stehn et al., Neutron Cross Sections Vol. 1, Z =1 to 20, BNL.-325, Brookhaven National Laboratory, 2nd ed., Suppl. No. 2, (1964), pp. 9-0-6, -7, -1l. A. P. Malinauskas and D. M. Richardson, MSR Program Semiannu. Progr. Re& Feb. 28, 1972, ORNL“4782, b. 53. 19. 20. 21. 22. 23. 24 . 25. 26. 27. 28. 29. 30, 31. 32. 33. 34. 35. 36. 27 P. E. Field and J. H. Shaffer, "The solubilities of HF and DF in Molten Fluorides," J. Phys. Chem. 71, 3218 (1967). C. F. Baes, Jr., "The Chemistry and Thermodynamlcs of Molten Salt Reactor Fuels," Nuclear Metallurgy, Vol. 15 (ed., P. Chiotti), CONF-690801, U.S. Atomic Energy Commission (1969), p. 624. D. D. Wagman et al., "Selected Values of Chemical Thermodynamic Properties,"” Technical Notes 270-3, 4, 5, U.S. National Bureau of Standards, Washington, D. C., 1968, 1969, 1971. S. Senderoff and G. W. Mellors, '"Coherent Coating of Refractory Metals," Science 153, 1475 (1966). J. Wischhusen, Materials Engineering 74(1), 35 (1971). G. M. Watson, R. B. Evans III, W. R. Grimes, and N. V. Smith, "Solubility of Noble Gases in Molten Fluorides," J. Chem. Eng. Data 7, 285 (1962). R. E. Thoma, Chemical Aspects of MSRE Operations, ORNL-4658, (Dec. 1971) pp. 68-71, P. N. Haubenreich and J. R. Engel, "Experience with the Molten Salt Reactor Experiment,” Nucl. Appl. Tech. 8(2), 118 (1970). W. R. Grimes, Chemical Research and Development for Molten Salt Breeder Reactors, ORNL CF-66-7-41 (1966), pp. 45-55. B. F. Hitch and C. F. Baes, Jr., Reactor Chemistry Division Annual Progress Report Dec. 31, 1966, ORNL-4076, p. 20. J. H. Shaffer, Preparation and Handling of Salt Mixtures for the Molten Salt Reactor Experiment, ORNL-4616 (Jan. 1971). K. A. Romberger, J. Braunstein, and R. E. Thoma, "New Electrochemical Measurements of the Liquidus in the LiF-BeF, System. Congruency of Li,BeF,," J. Phys. Chem. 76, 1154 (1972). S. Cantor, W. T. Ward, and C. T. Moynihan, "Viscosity and Density in Molten BeF,;-LiF Solutions," J. Chem. Phys. 50, 2874 (1969). E. Aukrust, B. Bjdrge, H. Flood, T. F¢rland "Activities in Molten Salt Mixtures of Pota851um—L1th1um Halide Mixtures: A Preliminary Report," Annals New York Acad. Sci. 79, Art. 11, 830-837 (1960). M. J. Lubin and A. P. Fraas, "Fusion by Laser," Sci. Amer. 224, 21 (1971). L. Dresner, Osk Ridge National Laboratory, personal communication. J. 0. Cowles and A. D. Pasternak, Lithium Properties Related to Use as_a Nuclear Reactor Coolant, UCRL-50647, Lawrence Livermore Radiation Laboratory (Apr. 18, 1969). I. I. Novikov, Y. 5. Trelin, T. A. Tsyganova, Experlmental Data on the Speed of Sound in Lithium up to 1100 K " High Temperature (Trans- lated Russian Journal) 7(6), 1140 (1969). ’_ W~ udwmhnE 10. 11. 12. 13. 14, 15. 16. 17. 18. 19. 20. 21. 22. 23, 48. 49. 50. 51. 52. 53. 54 . 55. 56. 57. 58. 59. 60. 01, 29 INTERNAL DISTRIBUTION R. G. Alsmiller 24. L. E. McNeese E. G. Bohlmann 25. 0. B. Morgan G. E. Boyd 26. H, Postma R. B. Briggs 27. M. Roberts S. Cantor (10 copies) 28. M. W. Rosenthal J. F. Clarke 29. W. K., bBartory R. E. Calusing 30. F. J. Smith F. L. Culler 31l. A. H. Snell J. H. DeVan 32. Don Steiner L. Dresner 33. R. A. Strehlow D, E. Ferguson 34. M. L. Tobias L. M., Ferris 35. D, B. Trauger R. C. Forrester 36. J. S. Watson A. P, Fraas 37. C. F. Weaver John Googin (Y-12) 38. A. M. Weinberg W. R. Grimes 39. F. W. Wiffen P. N. Haubenreich 40, F., W. Young E. H. Kobisk 41-42. Central Research Library J. J. Keyes 43. Y-12 Document Reference Section H. M. Long 44-45. TLaboratory Records Department Martin Lubell 46, Laboratory Records, RC R. E. MacPherson 47. ORNL Patent Office A. P. Malinauskas EXTERNAL, DISTRIBUTION P. Bananos, Plasma Physice Laboratory, Princeton, N.J. 080540 Manson Benedict, MIT, Dept. of Nuclear Engineering, 77 Massachusetts Ave., Boston, Mass. 02139 Milton Blander, Argonne National Laboratory, Argonne, Ill. Steven Blow, A.E.R.E., Harwell, Didcot, Berks, England M. Bowman, Los Alamos Sc1ent1f1c Laboratory, P. 0. Box 1663, Los Alamos, New Mexico 87544 G. A. Carlson, Lawrence Livermore Laboratory, P. 0. Box 808, Livermore, California 94550 S. Dean, USAEC, Washington, D. C. 20545 E. E. Donaldson, Washington State University, Pullman, Wash., 99163 E, L. Draper, Jr., University of Texas, 200 W. 21st St,, Austin, Texas 78712 W. E. Drummond, University of Texas, Austin, Texas 78712 B. J. Eastlund, USAEC, Controlled Thermonuclear Resegrch Div., Washington, D. C. 20545 W. C. Gough, USAEC, Controlled Thermonuclear Research Div., Washington, D. C. 20545 R. W. Gould, USAEC, Controlled Thermonuclear Research Div., Washington, D. C. 20545 D. Henderson, Los Alamos Scilentific Laboratory, P. 0. Box 1663, Logs Alamos, New Mexico 87544 62. 63. 64 . €5. 66. 67 . 68, 69. 70. 71. 72. 73. T . 75, 76. 77 78. 79. 80. g1. 82. 83. 84. g5. ge. 87. 88. g9. 90. 30 R. G. Hickman, Lawrence Livermore Laboratory, University of California, P. 0. Box 808, Livermore, Calif. 94550 R. L. Hirsch, USAEC, Controlled Thermonuclear Research Div., Washington, D. C. 20545 M. A, Hoffman, Lawrence Livermore Laboratory, University of California, P. 0. Box 808, Livermore, Calif. 94550 G. Hopkins, Gulf Energy and Environmental Systems, P. O. Box 608, San Diego, Calif. 92112 R. A. Huse, Public Service Electric and Gas, 80 Park Place, Newark, New Jersey 07101 A. Impink, Carnegie-Mellon University, Frew Avenue and Margaret Morrison St., Pittsburgh, Pennsylvania 15213 E. Johnson, Princeton University, Dept. of Chemical Engineering, Princeton, N. J. 08540 J. D. Lee, Lawrence Livermore Laboratory, University of California, P. O. Box 808, Livermore, Calif. 94550 L. Lidsky, MIT, Dept. of Nuclear Engineering, 77 Massachusetts Ave,, Boston, Mass. 02139 Geoffrey Long, A.E.R.E., Harwell, Didcot, Berks, England R. E. McCarly, Ames Laboratory, lowa State University, Ames, Iowa 50010 Victor A. Maroni, Argonne National Laboratory, 9700 5. Cass Ave., Argonne, Ill. 60439 E. Mason, MIT, Dept. of Nuclear Engineering, 77 Mass. Ave., Boston, Mass. 02139 Bennett Miller, USAEC, Washington, D. C. 20545 R. G. Mills, Princeton Plasma Physics Laboratory, Princeton, N.dJ. 08540 J. Mitchell, UKAEA Culham Laboratory W. Blake Myers, Lawrence Livermore Laboratory, Div. of Mechanical Engineering, P. 0. Box 808, Livermore, Calif. 94550 R. S. Pease, UKAEA Culham Laboratory J. Phillips, Los Alamos Scientific Laboratory, P. O. Box 1663, Los Alamos, New Mexico 87544 R. F. Post, Lawrence Livermore Laboratory, P. 0. Box 808, Livermore, Calif. 94550 F. Ribe, Los Alamos Scientific Laboratory, P. O. Box 1663, Los Alsmos, New Mexico 87544 D. M. Richman, USAEC, Div. of Research, Washington, D. C. 20545 D. Rose, MIT, Nuclear Engineering Dept., 77 Massachusetts Ave., Boston, Mass. 02139 R. E. Stickney, MIT, 77 Massachusetts Ave., Boston, Mass. 02139 E. C. Tanner, Princeton Plasma Physics Laboratory, Princeton, N. J.. 08540 A. W. Travelpiece, Dept. of Physics and Astonomy, University of Maryland, College Park, Maryland 20472 A, R. Van Dyken, USAEC, Division of Physical Research, Washington, D. C. 20545 R. Vogel, Argonne National Laboratory, 9700 S. Cass Ave., Argonne, I11. 60439 R. W. Werner, Lawrence Livermore Laboratory, University of Californisa, P. 0. Box 808, Livermore, Calif. 94550 31 91. L. Wood, Lawrence Livermore Laboratory, University of California, P. 0. Box 808, Livermore, Calif. 94550 92. AEC Research and Technical Support Division, ORO 93-111'. Technical Information Center, P. O. Box 62, Oak Ridge, Tenn. 37830 (17 copies, ACRS)