MOLTEN-SALT REACTOR CHEMISTRY W. R. GRIMES Oak Ridge National Labovatory, Oak Ridge, Tennessee 37830 Received August 4, 1969 Revised October 7, 1969 This document summarizes the lavge program of chewmical research and development which led to selection of fuel and coolant compositions for the Molten-Salt Reactor Experiment (MSRE) and for subsequent reactovs of this type. Chemical ‘behavior of the LiF-BeF,-ZvF,~-UF, fuel mixture and behavior of fission products duving powev op- evation of MSRE are presented. A discussion of the chewmical veactions which show promise for recovery of bred ***Pa and for vemowval of fission product poisons from a molten-salt breeder rveac- tor is included. INTRODUCTION A single-fluid molten-salt thermal breeder (MSBR) of the type described by Rosenthal et al.,’ Bettis and Robertson,” and Perry and Bauman® makes very stringent demands upon its fluid fuel.*”® This fuel must consist of elements having low capture cross sections for neutrons typical of the energy spectrum of the chosen design. The fuel must dissolve more than the critical concen- tration of fissionable material (**°U, *°°U, or 23%9py), and high concentrations of fertile material (?32Th) at temperatures safely below the tempera- ture at which the salt leaves the MSBR heat exchanger. The mixture must be thermally stable, and its vapor pressure needs to be low over an operating temperature range (1100 to 1400°F) sufficiently high to permit generation of high quality steam for power production. The fuel mixture must possess heat transfer and hydro- dynamic properties adequate for its service as a heat-exchange fluid. It must be relatively non- NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 1970 KEYWORDS: fused salt fuel, chemical reactions, reactors, beryllium fluorides, zirconium fluorides, lithium fluorides, vranium hexafluoride, repro- cessing, protactinium, separa- tion processes, breeding, fission products, MSRE aggressive toward some otherwise suitable ma- terial of construction and toward some suitable moderator material. The fuel must be stable toward reactor radiation, must be able to survive fission of the uranium (or other fissionable ma- terial) and must tolerate fission product accumu- lation without serious deterioration of its useful properties. We must also be assured of a gen- uinely low fuel-cycle cost; this presupposes a low-cost fuel associated with inexpensive turn- around of the unburned fissile material, and effective and economical schemes for recovery of bred fissile material and for removal of fission- product poisons from the fuel. A suitable secondary coolant must be provided to link the fuel circuit with the steam-generating equipment. The demands imposed upon this cool- ant fluid differ in obvious ways from those im- posed upon the fuel system. Radiation intensities will be markedly less in the coolant system, and the consequences of uranium fission will be ab- sent. The coolant salt must, however, be com- patible with metals of construction which will handle the fuel and the steam; it must not undergo violent reactions with fuel or steam should leaks develop in either circuit. The coolant should be inexpensive, possessed of good heat-transfer prop- erties, and it should melt at temperatures suit- able for steam cycle start-up. An ideal coolant would consist of compounds which would be easy to separate from the valuable fuel mixture should they mix as a consequence of a leak. This report presents, in brief, the basis for choice of fuel and coolant systems which seem optimum in light of these numerous—and to some extent conflicting—requirements. CHOICE OF FUEL COMPOSITION Compounds which are permissible major con- stituents of fuels for single-fluid thermal breeders 137 Grimes MSBR CHEMISTRY are those which can be prepared from beryllium, bismuth, boron-11, carbon, deuterium, fluorine, lithium-7, nitrogen-15, oxygen, and the fissionable and fertile materials. As minor constituents one might tolerate compounds containing the other elements in Table L Many chemical compounds can be prepared from the several ‘‘major constituents’’ listed above, Most of these, however, can be eliminated after elementary consideration of the fuel re- quirements.””® No hydrogen- (or deuterium-) bearing compounds possess overall properties that are practical in such melts. Carbon, nitro- gen, and oxygen form high melting binary com- pounds with the fissionable and fertile metals; these compounds are quite unsuitable as constit- uents of liquid systems. The oxygenated anions either lack the required thermal stability (i.e., NO;™ or NO; ) or fail as solvents for high concen- trations of thorium compounds (i.e., CO;=). It quickly develops, therefore, that fluorides are the only suitable salts indicated in this list of ele- ments. Fluoride ion is capable of appreciable neutron moderation, but this moderation is by itself in- sufficient for good neutron thermalization. An additional moderator is, accordingly, required. TABLE 1 Elements or Isotopes Which may be Tolerable in High-Temperature Reactor Fuels Absorption Cross Section Material (barns at 2200 m/sec) Nitrogen-15 0.000024 Oxygen 0.0002 Deuterium 0.00057 Carbon 0.0033 Fluorine 0.009 Beryllium 0.010 Bismuth 0.032 Lithium-7 0.033 Boron-11 0.05 Magnesium 0.063 Silicon 0.13 Lead 0.17 Zirconium 0.18 Phosphorus 0.21 Aluminum 0.23 Hydrogen 0.33 Calcium 0.43 Sulfur 0.49 Sodium 0.53 Chlorine-37 0.56 Tin 0.6 Cerium 0.7 Rubidium 0.7 138 NUCLEAR APPLICATIONS & TECHNOLOGY The only good moderator material truly compati- ble with molten-fluoride fuel mixtures is graph- ite,*”° Phase Behavior Among Fluorides Uranium tetrafluoride and uranium trifluoride are the only fluorides (or oxyfluorides) of uranium which appear useful as constituents of molten- fluoride fuels. Uranium tetrafluoride (UF,) is relatively stable, nonvolatile, and largely non- hygroscopic. It melts at 1035°C (1895°F), but this freezing point is markedly depressed by useful diluent fluorides. Uranium trifluoride dispropor- tionates at temperatures above ~ 1000°C by the reaction 4UFs;= 3UF, + U° . (1) It is unstable®’ at lower temperatures in most molten-fluoride solutions and is tolerable in reac- tor fuels only with a large excess of UF4 so that the activity of U° is so low as to avoid appreciable reaction with moderator graphite or container metal. | Thorium tetrafluoride (ThF4) is the only known fluoride of thorium. It melts at 1111°C (2032°F) but fortunately its freezing point is markedly de- pressed by fluoride diluents which are also useful with UF4. Consideration of nuclear properties alone leads one to prefer as diluents the fluorides of Be, Bi, 'Li, Mg, Pb, and Zr in that order. Equally simple consideration of the stability of these fluorides®® toward reduction by structural metals, however, eliminates the bismuth fluorides from considera- tion. This leaves BeF:and 'LiF as the preferred diluent fluorides. Phase behavior of systems based upon LiF and BeF:as the major constitu- ents, has, accordingly, been examined in detail.™ Fortunately for the molten fluoride reactor con- cept, the phase diagrams of LiF-BeF.-UF,; and LiF-BeF:-ThF4 are such as to make these ma- terials useful as fuels. The binary system LiF-BeF: has melting points below 500°C over the concentration range from 33 to 80 mole% BeF..'”'" The phase diagram, pre- sented in Fig. 1, is characterized by a single eutectic (52 mole% BeF:, melting at 360°C) be- tween BeF, and 2LiF-BeF:. The compound 2LiF-BeF, melts incongruently to LiF and liquid at 458°C. LiF.BeF: is formed by the reaction of solid BeF. and solid 2LiF-BeF: below 280°C. The phase behavior of the BeF:-UF.'”' and BeF,- ThF,'” systems are very similar. Both sys- tems show simple single eutectics containing very small concentrations of the heavy metal fluoride. ThF4 and UF4 are isostructural; they form a con- tinuous series of solid solutions with neither maximum nor minimum. VOL. 8 FEBRUARY 1970 Grimes MSBR CHEMISTRY 900 ’ 848 800 ~ 700 A\, LiF+ LIQUID o > 600 555 % \ N — E:IJ \ .__/ < s00 P w ~ 458 / . BeF, (HIGH QUARTZ TYPE) \ / +L1QUID 400 \\/ 360 LiF+Li2BeF4 Li,Bef,+BeF, (HIGH QUARTZ TYPE) 1 ' | | 300 < ! | 1 280 m . @ Li,BeF, » | LiBeF; + BeF, (HIGH QUARTZ TYPE) 5 + ® LiBeFy+ BeF, (LOW QUARTZ TYPE)\ =~ | 220 LiBeF. @ ; 1 I 3 -4 | | I N | 200 LiF 10 20 30 40 50 60 70 80 90 BeF, BeF, (mole %) Fig. 1. The system LiF-BelF,. The binary diagrams LiF-UF" and LiF-ThF," are generally similar and much more complex than the binary diagrams discussed immediately above. The LiF-UF, system shows three com- pounds (none are congruently melting) and a single eutectic, at 27 mole% UF4, melting at 490°C. The LiF-ThF4 system contains four binary compounds, one of which (3LiF-ThF4) melts congruently, with two eutectics at 570°C and 22 mole% ThF, and at 560°C and 29 mole% ThFy. The ternary system LiF-ThF4-UF,," shown in Fig. 2, shows no ternary compounds and a single eutectic freezing at 488°C with 1.5 mole% ThF, and 26.5 mole% UF,;. Most of the area on the dia- gram is occupied by the primary phase fields of the solid solutions UF4-ThF,, LiF-4UF4-LiF-4ThF,, and LiF-UF,-LiF-ThF,. Liquidus temperatures decrease generally to the LiF-UF4 edge of the diagram. The single-fluid molten-salt breeder fuel will need a concentration of ThFs much higher than that of UF4. Accordingly, the phase behavior of the fuel will be dictated by that of the LiF-BeF2- ThF, system. Figure 3 gives the ternary system LiF-BeF.-ThF,, this system shows a single ter- nary eutectic at 47 mole% LiF and 1.5 mole% ThF,, melting at 360°C.'>'" The system is complicated to some extent by the fact that the compound NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 1970 3LiF'ThF, can incorporate Be”’?' ions in both interstitial and substitutional sites to form solid solutions whose compositional extremes are rep- resented by the shaded triangular region near that compound. Liquidus temperatures < 550°C (1022°F) are available at ThF, concentrations as high as 22 mole%. The maximum ThF4 concen- tration available at liquidus temperatures of 500°C (932°F) is seen to be just above 14 mole%. Inspec- tion of the diagram reveals that a considerable range of compositions with > 10 mole% ThF,4 will be completely molten at or below 500°C. As expected from the general similarity of ThF, and UFs—and especially from the substitu- tional behavior shown by the LiF-UF4-ThF4 sys- tem (Fig. 2)—substitution of a small quantity of UF,s for ThF. scarcely changes the phase be- havior. Accordingly, and to a very good approxi- mation, Fig. 3 represents the behavior of the LiF-BeF,-ThF4-UF4 system over concentration regions such that the mole fraction of ThF, is much greater than that of UF4. Oxide Fluoride Equilibria Phase behavior of the pure fluoride system LiF-BeF,-ThF;-UF,, as indicated above, is such that adequate fuel mixtures seem assured. The 139 Grimes MSBR CHEMISTRY ThF4 114 TEMPERATURE IN °C COMPOSITION IN mole % LiF- 4ThF, LiF- 2ThF4 P 8I7 AN LiF-ThE A P 762 P 609 P 597 £ 568 3Lil—'-ThF4 (b) & > © D E 565 {\ 5 0 LiF 848 P 610 4LiF-UF;” P500" £ 490 (. Fig. 2. behavior of systems such as this, however, is markedly affected by appreciable cbncentrations of oxide ion. When a melt containing only LiF, BeF., and UF4 is treated with a reactive oxide (such as H, 0O) precipitation of transparent ruby crystals of UO2, 00 results.”” If the melt contains, in addition, an appreciable concentration of ZrF, the situation is markedly altered. ZrO; is less soluble than is UO: in such melts, and the monoclinic ZrO: (the form stable below ~ 1125°C) includes very little UO: in solid solution. Thus, inadvertent oxide contamination of a LiF-BeF;-ZrFs-UF, melt yields monoclinic ZrO. containing 250 ppm of UO:." Precipitation of cubic UO. (containing a small concentration of ZrO.) begins only after precipitation of ZrO, had dropped the ZrF, con- centration to near that of the UF,. Slow precipitation of UO. followed by a sudden entrance of this material into the reactor core could result in undesired increased in reactivity. This possibility was assumed to represent a dan- ger to the Molten-Salt Reactor Experiment. Ac- cordingly, the MSRE fuel was chosen to contain 5 mole% of ZrF, to eliminate such a possibility. 140 (d)\@ = o, \% Q S Q = % o & 6:5\ £ 500 b 5\ 52 W\ 3 \ \ \\ £488 1 \l ¥ F NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 PRIMARY-PHASE AREAS a) UF,~ThE, (ss) b) LiF-4UF,~LiF-4ThF, (ss) ) LiF-2Th(U)F, (ss) ) TLiF-BUF,~7LiF-6ThF, (ss) ) 3LiF-Th(U)F, (ss) ( ( (c (d (e (f) LiF © o \V4 PT75 Uy LiF-4UF4 1035 iF-UF, The system LiF-ThF,-UF, When a mixture of LiF and BeF, containing ThF, and UF4 is treated with a reactive oxide a homogeneous cubic phase is produced; this phase is a solid solution of UO; and ThQ. which is very rich in UO.." Careful studies have shown that the reaction ThO:(ss) + Uy = UOa(ss) + Thip (2) [where the subscripts (ss) and (f) indicate the solid-phase solid solution and the molten-fluoride solution, respectively| approaches equilibrium with reasonable speed. Values for the equilibrium quotient @ for this reaction 4 + NUO 2(ss_). NTh (f) (3) + NTRO o(ss)’ N4U(f) Q increase with UO; concentration of the oxide phase and decrease markedly with temperature. Since values of @ for mixtures similar to those chosen as fuel compositions are typically 300 to 1000, it is clear that oxide contamination of such salts will selectively precipitate the uranium. FEBRUARY 1970 LiF-ThFa 3LiF-ThF4 SS Grimes MSBR CHEMISTRY The, 4114 TEMPERATURE IN °C COMPOSITION IN mole % BeF. 848 2LiF- BeFZ/scBo?zxso 400! 400 450 500 548 P 458 £ 360 Fig. 3. The system LiF-BeF;-ThFs. It is likely, though not certain, that addition of some ZrF, would afford protection of the sort ob- tained with the MSRE fuel. Such addition is un- desirable, however, since the presence of ZrF, would certainly complicate the separation pro- cesses described later in this paper and else- where in this series. The successful operation of the MSRE over a three year period (discussed later) lends con- fidence that oxide contamination of the fuel system can be kept to adequately low levels. This confi- dence, when added to the prospect that the breeder fuel will be reprocessed (and its oxide level re- duced) at regular intervals, suggests very strongly that successful operation can be achieved without added ‘‘oxide protection.’’ Tolerance levels for oxide concentration in LiF-BeF.-UF4 and LiF-BeF:-ZrF,-UF, systems have been studied in detail and are relatively well understood.’®**~® Analogous values for the LiF- BeF,-ThF4-UF, system are still largely lacking. It is known, however, that processing of these quaternary melts with anhydrous HF and H: serves to remove oxide to a level below that required for precipitation of the solid solutions. There seems little doubt, therefore, that initial processing of NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 the type used for MSRE fuels can be successfully applied on a large scale to the LiF-BeFz:-ThF4- UF4 system. MSRE and MSBR Fuel Compositions The fuel chosen for operation of MSRE (with a 23517_238y jsotopic mixture containing 33% of the fissionable isotope) was a mixture of 'LiF, BeFz, ZrFs, and UF, consisting of 65, 29.1, 5, and 0.9 mole%, respectively. The uranium concentration was fixed at ~1% so that there was less possibility of fissile U precipitating (~ 0.3 mole% **°U was necessary to achieve criticality and to provide a small excess of fissionable material for power operation of the machine). The ZrF4 was added, as indicated above, to preclude possible inad- vertent precipitation of UOz. Beryllium fluoride is an extremely viscous material; its viscosity is markedly lowered by addition of LiF. The ratio of LiF to BeF: in the MSRE fuel was chosen to opti- mize the conflicting demands for low viscosity and a low liquidus temperature for the molten fuel. The single-fluid breeder requires a high con- centration of ThF4; concentrations near 12 mole% seem to be reasonable for good reactor FEBRUARY 1970 141 Grimes MSBR CHEMISTRY performance. Criticality estimates suggest that such a fuel could be made critical in a practicable reactor with somewhat < 0.3 mole% 2**UF,. The ratio of "LiF to BeF, should, to decrease vis- cosity, be kept at a value as high as is practicable. If the liquidus temperature is to be kept at or below 500°C (932°F) for a melt with 12 mole% of ThF4, the beryllium concentration limits range from 16 to 25 mole%. The most likely choice for the MSBR fuel—and the present design composi- tion—is, accordingly, "LiF-BeF,-ThF.-UF, at 71.7-16-12-0.3 mole%, respectively. Choice of Coolant The secondary coolant is required to remove heat from the fuel in the primary heat exchanger and to transport this heat to the power generating system. In the MSBR the coolant must transport heat to supercritical steam at minimum tempera- tures only modestly above 7T00°F. In the MSRE the heat was rejected to an air cooled radiator at markedly higher salt temperatures. The coolant mixture chosen for the MSRE and shown to be satisfactory in that application is BeF: with 66 mole% of "LiF. Use of this mixture would pose some difficulties in design of equip- ment for the MSBR since its liquidus temperature is 851°F; moreover, it is an expensive material. The eutectic mixture of LiF with BeF, (48 mole% LiF) melts at near 700°F (see Fig. 1) but it 1s relatively viscous and is expensive, especially if "LiF is used. The alkali metals, excellent coolants with real promise in other systems, are undesirable here since they react vigorously with both fuel and steam. Less noble metal coolants such as Pb° or Bi’ undergo no violent reactions, but they are not compatible with Hastelloy-N, the Ni-based alloy used in MSRE and intended for use in MSBR'’s. Several binary chloride systems are known to have eutectics melting below (in some cases much below) 700°F.” These binary systems do not, however, appear especially attractive since they contain high concentrations of chlorides (TICI, ZnClz, BiCls, CdClz, or SnClz), which are easily reduced and, accordingly, corrosive; or chlorides (AlCls, ZrCl, HICls, or BeCly) which are very volatile. The only low-melting binary systems of stable, non-volatile chlorides are those containing LiCl; LiCl-CsCl (330°C at 45 mole% CsCl), LiCl- KC1 (355°C at 42 mole% KCl), LiCl-RbCl (312°C at 45 mole% RbCl). Such systems would be relatively expensive if made from "LiCl, and they could lead to serious contamination of the fuel if normal LiCl were used. Very few fluorides or mixtures of fluorides are known to melt at temperatures below 370°C. Stan- 142 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 nous fluoride (SnF:) melts at 212°C. This ma- terial is probably not stable during long term service in Hastelloy-N; moreover, its phase dia- grams with stable fluorides (such as NaF or LiF) show high melting points at relatively low alkali fluoride concentrations. Coolant compositions whichwill meet the low liquidus temperature spec- ification may be chosen from the NaF-BeF, or NaF-LiF-BeF, system. These materials are al- most certainly compatible with Hastelloy-N, and they possess adequate specific heats and low vapor pressures (discussed in the next section). They (especially those including LiF) are moder- ately expensive, and their viscosities at low temperature are certainly higher than desirable. It is possible that substitution of ZrF,; or even AlF; for some of the BeF. would provide liquids of lower viscosity at no real expense in liquidus temperature. It now appears that the best choice for the MSBR secondary coolant is the eutectic mixture of sodium fluoride and sodium fluoroborate. The binary system NaF-NaBF, has been described as showing a eutectic at 60 mole% NaBF, melting at 304°C (580°F).*“?? Studies here have shown that this publication is seriously in error. Boric oxide substantially lowers the freezing point of NaF- NaBF; mixtures; the original authors may well have used quite impure materials. Use of pure NaF and NaBF, leads to the conclusion that the system shows a single eutectic at 8 mole% NaF and a melting point of 385°C (725°F),2® as shown in Fig. 4. At elevated temperatures the fluoroborates show an appreciable equilibrium pressure of gas- eous BFs;. The equilibrium pressure®® above a 1000 900 3 fl Q o TEMPERATURE (°C) | (o)) O o o O 3 300 200 NaoF 20 40 60 80 NaBF, (mole % ) NoBF4 Fig. 4. The system NaF-NaBF,. FEBRUARY 1970 melt maintained at the eutectic composition (8 mole% NaF, 92 mole% NaBF4) is given by 5,920 U @) log PTorr = 9.024 - PHYSICAL PROPERTIES OF FUELS AND COOLANTS Tables II andIII list some of the pertinent phys- ical properties® for MSRE and MSBR fuels and secondary coolants. Many of the properties shown are estimates rather than measured values. These estimates have been carefully prepared from the best avail- able measurements on several salt mixtures of similar composition. The values given are un- likely to be in error sufficient to remove the fluid from consideration. It is clear, however, from the fact that estimates rather than measured values are shown that an experimental program must be devoted to firming up the physical properties of these materials. The densities were calculated from the molar volumes of the pure components by assuming the volumes to be additive. The heat capacities were estimated by assuming that each gram atom in the mixture contributes 8 calories per degree centi- grade. The value of 8 is the approximate average from a set of similar fluoride melts. The vis- cosity of the MSBR fuel and the BeF:-based cool- ants were estimated from other measured LiF-BeF: and NaF-BeF: mixtures. The vapor pressure of the fuels and the BeF'» based coolants are considered negligible; extrapo- lation of measurements from similar mixtures TABLE II Composition and Properties of MSRE and MSBR Fuels MSRE MSBR Fuel Fuel Composition LiF 65 LiF T1.7 (mole%) BeF, 29.1 | BeF: 16 ZrFs 5 ThF, 12 UF, 0.9 | UF,4 0.3 Liquidus °C 434 500 °F 813 932 Properties at 600°C (1112°F) Density, g/cm’ 2.27 3.35 Heat capacity, cal/(g °C) or Btu/(lb °F) 0.47 0.33 Viscosity, cP 9 12 Vapor pressure, Torr Negligible | Negligible Thermal conductivity, W/(°C cm) 0.014 0.011 Grimes MSBR CHEMISTRY TABLE III Composition and Properties of Possible MSBR Secondary Coolants C C. Cs Composition (mole%) NaF 8 | LiF 23 NaBFs 92 | NaF 41 NaF 57 BeF, 36 BeF: 43 Liquidus Temperature: °C 385 328 340 °F 725 622 644 Physical Properties al 850°F (454°C)* Density, 1b/ft® 121 136 139 Heat capacity, Btu/(1b °F) 0.36 0.47 0.44 Viscosity, cP 2.5 40 65 Vapor pressure at 1125°F (607°C),> mm 200°¢ Negligible | Negligible Thermal conductivity, W/(°C cm) 0.005 0.01 0.01 aMean temperature of coolant going to the primary heat exchanger. b.'I-Iighest normal operating temperature of coolant. cRepresents pressure of BF; in equilibrium with this melt compo- sition. yielded pressures < 0.1 mm. The partial pressure of BFs above the fluoroborate coolant mixture was calculated from measurements on very similar mixtures. CHEMICAL COMPATIBILITY OF MSBR MATERIALS Details and specific findings of the large pro- gram of corrosion testing are presented by McCoy et a1? as a separate article in this issue. In brief, compatibility of the MSBR materials is assured by choosing as melt constituents only fluorides that, insofar as is possible, are thermodynamically stable toward the moderator graphite and toward the structural metal, Hastelloy-N,® a nickel alloy containing ~ 12% Mo, 7% Cr, 4% Fe, 1% Ti, and small amounts of other elements. The major fuel components (LiF, BeF2, UF4, and ThF4) are much more stable than the structural metal fluorides (NiF2z, FeF3z, and CrF2); accordingly, the fuel and blanket have a minimal tendency to corrode the metal. Such selection, combined with proper pur- ification procedures, provides liquids whose cor- rosivity 'is well within tolerable limits. The chemical properties of the materials and the nature of their several interactions are described briefly in the following. NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 aHastelloy-N used in MSRE was Ni with 17% Mo, 7% Cr, 59 Fe. Probable composition of modified Hastelloy-N for MSBE is that shown above. FEBRUARY 1970 143 Grimes MSBR CHEMISTRY Thermodynamic Data for Molten Fluorides A continuing program of experimentation over many years has been devoted to definition of the thermodynamic properties of many species in molten LiF-BeF: solutions. Many techniques have been used in this study. Many of the data have been obtained by direct measurement of equilib- rium pressures for reactions such as Hzg) + FeFyq)= Fe?c) + 2HF () (5) and 2HF () + BeO(c)= BeF:(1) + H20(y) (6) [where (g), (c), and (d) represent gas , crystalline solid, and solute, respectively] using the molten fluoride as reaction medium. Baes has reviewed all these studies and by combining the data with the work of others has tabulated thermodynamic data for many species in molten 2LiF.BeF,.?® TableIV below records pertinent data for the major com- ponents of MSRE and MSBR fuels and for cor- rosion products in molten 2LiF-BeF,. From these data one can assess the extent to which UFs-bearing melt will disproportionate ac- cording to the reaction 4UFs(q) = 3UF4q) + U . | (7) TABLE IV Standard Free Energies of Formation for Species in Molten 2LiF - BeFo (773 to 1000°K) -AG/ -AG1000°k) Material? (kcal/mole) (kcal/mole) LiF (1) 141.8-16.6 x 10~3T°K 125.2 BeFz (1) 243.9 - 30.0 x 10”3 7°K 106.9 UFs3 (q) 338.0-40.3 x 10 T°K 99.3 UF4(q) 445.9 - 57.9 x 102 T°K 97.0 ThFs (g 491.2-62.4 x 10 T°K 107.2 ZrFa(q) 453.0 - 65.1 x 103 7°K 97.0 NiFz (g 146.9 - 36.3 x 10> T°K 55.3 FeFa (g 154.7-21.8 x 1072 T°K 66.5 CrFz(q) 171.8-21.4 x 1073 T°K 75.2 MoFs (g) 370.9-69.6 x 10> T°K 50.2 *The standard state for LiF and BeF2 is the molten 2LiF - BeF2 liquid. That for MoFe (g) is the gas at one atmosphere. That for all species with subscript (d) is that hypothetical solution with the solute at unit mole fraction and with the activity coefficient it would have at infinite dilution. 144 NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 For the case where the total uranium in the mol- ten solution of 0.9 mole% (as in the MSRE) the activity of metallic uranium is near 10~ '° with 1% of the UF, converted to UFs and is near 2 X 107" with 20% of the UF4 so converted.?” Op- eration of the reactor with a small fraction (usually <2%) of the uranium present as UF;s is advantageous insofar as corrosion and the conse- quences of fission are concerned (see subsequent sections of this report). Such operation with some UFs; present should result in formation of an extremely dilute (and experimentally undetectable) alloy of uranium with the surface of the container metal. Operation with > 50% of the uranium as UFs would lead to much more concentrated (and highly deleterious) alloying and to formation of uranium carbides. Fortunately, all evidence to date demonstrates that operation with relatively little UFs is completely satisfactory. Oxidation (Corrosion) of Metals The data of Table IV reveal clearly that in reactions with structural metals (M) 2UF4 ) + M()= 2UF3(4) + MF2(q) (8) Cr is much more readily attacked than is iron, nickel, or molybdenum.?” Few thermodynamic data exist for the fluorides of titanium in the pure state and none for dilute solutions in molten fluoride solvents. Estimates®® of free energies of formation suggest that Ti is somewhat more reac- tive than Cr. Titanium on the surface layers of the metal should, therefore, be expected to react with the available oxidants, such as UF,. Such oxidation and selective attack follow from reac- tions such as the following: Impurities in the melt Cr + NiF, - CrF, + Ni (9) or Cr + 2HF — CrF, + H, (10) Oxide films on the metal NiO + BeFz2 — NiF: + BeO , (11) followed by reaction of NiF, with Cr. Reduction of UF4 to UF; Cr + 2UF4 = 2UF; + CrF» (12) Reactions (9), (10), and (11) above will proceed essentially to completion at all temperatures within the MSBR circuit. Accordingly, such reac- tions can lead (if the system is poorly cleaned) to a noticeable, rapid initial corrosion rate. How- ever, these reactions do not give a sustained corrosive attack. FEBRUARY 1970 The reaction of UF4 with Cr, on the other hand, has an equilibrium constant with a small tempera- ture dependence; hence, when the salt is forced to circulate through a temperature gradient, a pos- sible mechanism exists for mass transfer and continued attack. If nickel, iron, and molybdenum are assumed to be completely inert diluents for chromium (as is approximately true), and if the circulation rate in the MSBR is very rapid, the corrosion process can be simply described. At high flow rates, uniform concentrations of UFs; and CrF. are maintained throughout the fluid circuit; these concentrations satisfy (at some intermediate temperature) the equilibrium constant for the reaction. Under these steady-state conditions, there exists some tem- perature (intermediate between the maximum and minimum temperatures of the circuit) at which the initial surface composition of the structural metal is at equilibrium with the fused salt. Since the equilibrium constant for the chemical reaction in- creases with increasing temperature, the chromi- um concentration in the alloy surface tends to decrease at temperatures higher than T and tends to increase at temperatures lower than 7. [In some melts (NaF-LiF-KF-UF4, for example) AG for the mass transfer reaction is quite large, and the equilibrium constant changes sufficiently as a function of temperature to cause formation of dendritic chromium crystals in the cold zone.] For MSBR fuel and other LiF-BeF2-UF4 mixtures, the temperature dependence of the mass-transfer reaction is small, and the equilibrium is satisfied at reactor temperature conditions without the for- mation of crystalline chromium. Thus, in the MSBR, the rate of chromium removal from the salt stream by deposition at cold-fluid regions is controlled by the slow rate at which chromium diffuses into the cold-fluid wall; the chromium concentration gradient tends to be small, and the resulting corrosion is well within tolerable limits. Titanium, which must be presumed to undergo similar reactions, diffuses less readily than does Cr in Hastelloy-N. It seems most unlikely that the 1% of Ti in the alloy will prove detrimental. The results of numerous long-term tests have shown that Hastelloy-N has excellent corrosion resistance to molten-fluoride mixtures at temper- atures well above those anticipated in MSBR.” The attack from mixtures similar to the MSBR fuel at hot-zone temperatures as high as 1300°F is barely observable in tests of as long as 12 000 h. A fig- ure of <0.5 mil/year is expected.’® In MSRE, attack was ~ 0.1 mil/year. Thus, corrosion of the container metal by the reactor fuel does not seem to be an important problem in the MSBR.™ It is likely that the NaF-NaBF4 coolant mixture will also prove compatible with Hastelloy-N, but NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 1970 Grimes MSBR CHEMISTRY experimental study of this is much less extensive. The free energy change for the chemical reaction BFs(p) + 3/2Cr (s— 3/2CrFzq) + B(s) (13) is ~+30 kcal at 800°K.” The reaction is, there- fore, quite unlikely to occur, and similar reactions with Fe, Mo, and Ni are much less so. In addition, the above reaction becomes even less likely (per- haps by 5 kcal or so) when one considers the energetics of formation of the compound NaBF, and dilution of the NaBF4 by NaF. However, re- actions which produce metal fluorides and metal borides are those which must be anticipated. Ti- tanium, a minor constituent of Hastelloy-N, has a quite stable boride.’® The reaction 2BFs(p) + 2Ti(o= TiB2(c)+ 2TiFsa (14) probably shows a small negative free energy change and must be expected to proceed. The very small diffusion rate of Ti in Hastelloy-N would be expected to markedly limit the reaction. Thermo- chemical data for the borides of Fe, Cr, Ni, and Mo do not seem to have been established. Unless the borides of these materials are very stable (AGf values more negative than -25 kcal B atom) the Hastelloy-N should prove resistant to pure NaF-NaBF4 coolant. Compatibility of Graphite with Fluorides Graphite does not react with molten fluoride mixtures of the type to be used in the MSBR. Available thermodynamic data®® suggest that the most likely reaction 4UF4 + C = CF4 + 4UF; (15) should come to equilibrium at CF4 pressures £ 10°% atm. CF4 concentrations over graphite-salt systems maintained for long periods at elevated temperatures have been shown to be below the limit of detection (< 1 ppm) of this compound by mass spectrometry. Moreover, graphite has been used as a container material for many NaF-ZrF,- UF., LiF-BeF:-UF4, and other salt mixtures with no evidence of chemical instability. CHEMICAL BEHAVIOR IN MSRE The Molten-Salt Reactor Experiment was, as detailed in an article by Haubenreich and Engel™ operated during much of 1966, 1967, and early 1968 with the original fuel charge in which the uranium was 33% 2°°U with the balance consisting primarily of >**U. The reactor accumulated nearly 70 000 MWh of operation during this interval with the major fraction of this accumulated at the max- imum power of ~ 8 MW(th). 145 Grimes MSBR CHEMISTRY Behavior of Fuel Components Samples of the molten salts were removed routinely from the fuel and coolant circuits and were analyzed for uranium, major fuel constitu- ents, possible corrosion products, and (less fre- quently) for oxide ion contamination. Early in the power runs, standard samples of fuel were drawn three times per week; this schedule was markedly diminished as confidence in the system grew. At late stages in the power runs, analyses for these items was generally done on a one-per-week basis. The LiF-BeF: coolant system was sampled every two weeks initially and less frequently dur- ing 1967 and 1968. Chemical determinations of uranium content of the fuel salt were run by coulometric titrations® of dilute aqueous solutions. Such analyses showed good reproducibility and high precision (+0.5%). (On-site reactivity balance calculations proved to be about ten-fold more sensitive than this in es- tablishing changes in uranium concentrations with- in the fuel circuit.) Agreement of the chemically determined uranium concentrations (in a statis- tical sense) with the reactor inventory values diminished by the uranium burnup during opera- tion was excellent. The inventory value at end of operation on °*°U showed the uranium concentra- tion to be 4.532% by weight.*®* The difference (200 g of uranium out of 220 kg) is <0.1%. These data strongly indicate that uranium losses (as to the purge gas stream) have been extremely small and that, as anticipated, the fuel salt is chemically stable during long periods of reactor operation. A further check on these numbers will be possible when the **°U-?°*U mixture removed by fluori- nation from the MSRE fuel is recovered and assayed. While analyses for UFs and for ZrF, agree quite well with the inventory data, the values for "LiF and BeF: have never done so; analyses for LiF have shown higher and those for BeF, have shown lower values than the inventory since start-up. Table V shows a comparison of typical analysis with the original inventory: value. This TABLE V Typical and Original Composition of MSRE Fuel Mixture Original Value Constituent (mole%) Typical Analysis "LiF 63.40 + 0.49 64.35 BeFs 30.63 + 0.55 29.83 ZrF4 5.14 £ 0.12 5.02 UF4 0.821 + 0.008 0.803 146 NUCLEAR APPLICATIONS & TECHNOLOGY discrepancy in LiF and BeF: concentration re- mains a puzzle, but there was nothing in the an- alyses (or in the behavior of the reactor) to suggest that any changes occurred. Oxide concentration in the radioactive fuel salt were performed by careful evaluation of H: O pro- duced upon treatment of the salt samples with anhydrous gaseous HF. All samples examined showed less than 100 ppm of O%; no perceptible increase in the values with time or with opera- tional details was apparent.’* These facts are quite reassuring insofar as maintaining oxide contamination at very low levels in future reactor systems is concerned. MSRE maintenance operations have necessi- tated flushing the interior of the drained reactor circuit on numerous occasions. The salt used for this operation consisted originally of a "LiF-BeF, (66.0 to 34.0 mole%) mixture. Analysis of this salt before and after each use shows that ~ 215 ppm of uranium is added to the flush salt in each flushing operation; this corresponds to removal of 22.7 kg of fuel-salt residue (~0.5% of the charge) from the reactor circuit. Behavior of Corrosion Products Analysis of many samples at regular intervals during reactor operation showed relatively high values for iron (120 ppm) and nickel (50 ppm) in the salt.”””™° These values scattered significantly but showed no perceptible trends. They do not seem to represent dissolved Fe®* or Ni**. Molyb- denum concentrations were shown, in the few attempts made, to be below the detectible limit (~ 25 ppm) for chemical analysis. The chromium concentration, as determined by chemical analysis, rose from an initial value near 40 ppm to a final value after 70 000 MWh of ~85 ppm. A considerable scatter (+ 15 ppm) in the numbers was apparent but the increase with time and reactor operation is clearly real. All evidence suggests that the analytically determined chromium was largely, if not entirely, present as dissolved Cr**. This observed increase in chro- mium concentration corresponds to removal of <250 g of this element from the MSRE circuit. If this were removed uniformly it would deplete the chromium in the ailoy to a depth of < 0.2 mil. Such an estimate of corrosion seems consistent with observations (reported elsewhere in this series) showing very slight attack by the fuel during MSRE operation. This virtual absence of corrosion—though in general accord with results from a variety of en- gineering corrosion tests—is mildly surprising since the MSRE was operated for appreciable periods with the fuel salt more oxidizing than VOL. 8 FEBRUARY 1970 necessary. The analysis for ratio of U*t to U has proved especially difficult in the radioactive salts, but it seems certain that the MSRE fuel has, at times, contained < 0.2% of the total uranium as U The fission process (see following section of this report) appears to be mildly oxidizing toward dissolved U°t in the fuel. Accordingly, and al- though the corrosion product analysis did not seem to require it, a convenient means was de- veloped for adding U°Y to the reactor circuit as desired. Beryllium metal (as 3-in. rods of 3-in. diameter encased in a perforated capsule of nickel) was introduced through the sampling sys- tem and suspended in the flowing salt stream in the MSRE pump bowl. This active metal reacts with UF4 by Be(()c) + 2UF4q)— BeFza)+ 2UF3(q) (16) at a rate such that some 600 g of UFs is produced during an 8-h treatment. The salt flowing past the Be appears to be locally over-reduced so that at least 6% of the UF4 becomes UFjs; crystalline Cr was observed upon the nickel capsule from one such reduction attempt. The over-reduced salt mixture clearly reacted and achieved equilibrium with the large excess of unreduced salt in the pump bowl and reactor circuit. A molten-salt reactor which included a repro- cessing circuit could use that external circuit to maintain the UF,/UFs; ratio at the desired level. Use of techniques such as the Be addition for such a reactor would seem to be quite feasible but unnecessary. Behavior of Fission Products When fission of UF4 occurs in a well-mixed high-temperature molten-fluoride, four fluoride jons associated with the U** are released and the two fission fragments must come quickly to a steady state as common chemical entities. This steady state is made very complicated by the radioactive decay of many species. The valence states that the fission product assume in the mol- ten system are, presumably, defined by the re- quirements that cation-anion equivalence be maintained in the liquid and redox equilibria be established both among the components of the melt and the surface layers of the container metal.®”® Fluorine and uranium species higher than 4 are, in MSRE and MSBR fuels, unstable toward reduc- tion by the active constituents of Hastelloy-N; the fission product cations must satisfy the four fluoride ions plus the fission product anions. Should they prove inadequate, or if they become adequate only by assuming oxidation states incom- patible with Hastelloy-N, then oxidation of UFs NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 Grimes MSBR CHEMISTRY must occur or the container must supply the de- ficiency. Operation of MSRE has indicated that the presence of UF; in relatively low concentrations (see preceding sections) suffices to avoid oxida- tion or corrosion from this source. Behavior of the fission products in MSRE was studied®®*™*° during the entire period of operation with 23°U. Samples of MSRE fuel, drawn from the sampling station in the MSRE pump bowl, have been routinely analyzed for numerous fission prod- uct species. Since samples drawn in open metal cups were shown (for reasons described below) to give erroneously high values for gas-borne spe- cies, evacuated bulbs of Ni sealed with a fusible plug of salt were used to obtain many of the data presented here. Samples of the helium gas purge within the MSRE were obtained by use of identical evacuated sample bulbs in which the fusible plug was per- mitted to melt with the bulb exposed to the gas phase.*"»%® In addition to these salt and gas samples, as- semblies of surveillance specimens were exposed as the central stringer within the MSRE core. These assemblies included specimens of various graphites and many specimens of Hastelloy-N and some other metals. Their removal at reactor shutdowns permitted study of fission product de- positions after 8000, 20 000, and 70 000 MWh of operation. Samples of salt removed from the pump bowl were dissolved in analytical chemistry hot cells (after careful leaching of activities from the out- side of capsules where necessary); the fission products were identified and their quantity de- termined radiochemically. Analysis of the gas samples was accomplished similarly after careful leaching of the activities from within the capsules. Deposition of fission products upon the metallic surveillance specimens was determined in the same way after repeated leachings of the metal surface. Deposition on the graphite specimens was es- tablished as a function of depth of penetration. This was accomplished by careful milling of suc- cessive thin layers from each surface of the rec- tangular specimens. The removed layers were separately recovered and analyzed for several fission product isotopes.’®™*° The fission gases Kr and Xe form no com- pounds under conditions existing in a molten-salt reactor,*’ and these elements are only very spar- ingly soluble in the molten fluoride.*” ** The helium purge gas introduced into the MSRE pump bowl serves to strip these gases from the in- coming salt to charcoal filled traps downstream in the exit gas system. This stripping ensures that the fission product daughters of these gases will FEBRUARY 1970 147 Grimes MSBR CHEMISTRY appear within the reactor system at lower than the theoretical concentration. The MSRE graphite is, of course, permeable to Kr and Xe; radioactive daughters of these gases are, accordingly, ex- pected in the graphite specimens. The fission products Rb, Cs, Sr, Ba, the lan- thanides and yttrium, and Zr all form quite stable and well-known fluorides which are soluble in MSRE fuel.*>*® These fission products were ex- pected (except to the extent that volatile precursor Kr and Xe had escaped) to be found almost com- pletely in the molten fuel. These expectations have been confirmed. Isotopes such as °°Zr, °'Sr, and "*Cewhich have no precursors of consequence are typically found in the salt at 90%% of the cal- culated quantity. Isotopes such as ®°Sr and “°Ba whose Kr and Xe precursors have appreciable half-lives are found in less than the calculated quantity; these isotopes are, as expected, found in the graphite specimens and the gas samples as well, The elements (molybdenum, niobium, ruthen- ium, and tellurium) whose higher valence fluorides are volatile and relatively unstable toward re- duction by UF;%'%2%%%%7 are virtually absent from the salt. Typical analyses of samples (which have been carefully protected from the pump bowl gas by use of the evacuated bulb samples) show < 2% of the calculated concentration of these iso- topes. Deposition of fission products on the ~ 7- to 10-mil outer layers of graphite specimens ex- posed for 8000 MWh in MSRE is shown in Table VL It is clear that, with the assumption of uniform deposition in or on the moderator graphite, ap- preciable fractions of Mo, Te, and Ru and a larger fraction of the Nb are associated with the graph- ite. Subsequent examinations of graphite after 22 000 and 70 000 MWh have shown somewhat smaller fractions deposited for all these isotopes. The data of Table VII described below show values more typical of the longer exposures.*° A careful determination of uranium in or on the graphite specimens led to values < 1 pug/cm?.% This quantity of uranium—equivalent to a few grams in the moderator stack—seems completely negligible. Figure 5 shows the change in concentration of the fission product isotopes with depth in the graphite. Those isotopes (such as ™°Ba) which penetrated the graphite as noble gases show straight lines on the logarithmic plot; they seem to have remained at the point where the noble gas decayed As expected, the gradient for “°Ba with 16- sec **Xe precursor is much steeper than that for **Sr, which has a 3.2-min **Kr precursor. All the others shown show a much steeper concentra- tion dependence. Generally the concentration 148 NUCLEAR APPLICATIONS & TECHNOLOGY TABLE VI Fission Product Deposition on Surface* of MSRE Graphite Graphite Location Top Middle Bottom Percent of Percent of Percent of Isotope Total? Total? Total? PMo 13.4 17.2 11.5 1%27e 13.8 13.6 12.0 193pu 11.4 10.3 6.3 *Nb 12 59.2 62.4 131y 0.16 0.33 0.25 Bz 0.33 0.27 0.15 %oe 0.052 0.27 0.14 ®sr 3.24 3.30 2.74 140 1.38 1.85 1.14 Hloe 0.19 0.63 0.36 137cs 0.07 0.25 0.212 *Average of values of 7- to 10-mil cuts from each of three exposed graphite faces. aPercent of total 1n reactor deposited on graphite if each cm® of the 2 x 10° cm? of moderator had the same con- centration as the specimen. drops a factor of 100 from the top 6 to 10 mils to the second layer. More recent examinations which used cuts as thin as 1 mil show the materials to be concentrated in an extremely thin layer. Typ- ically 95% of the metallic species are within 3 mils of the surface. TABLE VI Fission Product Distribution for MSRE at End of **®*U Operation Quantity Found (% of Calculated Inventory) On In On Hastelloy- | Purge Nuclide In Fuel Graphite? N2 Gas? | Total Mo 0.17 9.0 19 50 78 132e 0.47 5.1 9.5 74 89 129ne 0.40 5.6 11.5 31 48 193 pu 0.033 3.5 2.5 49 55 1%%pu 0.10 4.3 3.2 130 138 **Nb 0.001 to 2.2 41 12 11 64 Bzr 94 0.14 0.06 0.43 95 8%sr 83 8.5 0.08 17 109 1op, 96 1.9 0.05 0.48 98 Hloe 0.33 0.03 0.88 e 0.92 0.03 2.7 131y 60 0.11 0.3 19 80 2Calculated on assumption that average values for surveillance specimens are representative of all graphite and metal surface in MSRE. PThese values are the percentage of production lost assuming mean values found in gas samples apply to the 4-liter/min purge. VOL. 8 FEBRUARY 1970 DISINTEGRATIONS PER MINUTE PER GRAM OF GRAPHITE 0 10 20 30 40 50 DISTANCE FROM SURFACE OF GRAPHITE (mils) Fig. 5. Concentration profile of fission products in MSRE core graphite after 8000 MWh, Examination of the Hastelloy-N surveillance specimens exposed in the MSRE core reveals very low concentrations of the stable fluoride-forming fission products such as *8r, ®*Zr, and *'Ce. The more noble metals such as Mo, Nb, and Ru are deposited, along with Te, in appreciable though not in large quantities. If the surveillance specimens are truly representative of all the metal surface in MSRE then (see Table VII) only a relatively small fraction of these metals are so deposited. However, flow conditions in the MSRE heat ex- changer (where most of the metal surface exists) is turbulent while laminar flow occurs over the core specimens. It seems very likely that the actual fraction deposited on MSRE metal is higher than that shown here. Analysis of samples of gas from MSRE yielded some real surprises. Isotopes such as 8gr and 0By are found in amounts consistent with con- sideration of the half-lives of their Kr and Xe NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 Grimes MSBR CHEMISTRY precursors. A small quantity of *’Zr appears in the gas phase; this isotope, along with several others with stable soluble fluorides, almost cer- tainly appears in the gas as fine particles of salt mist.>® However, high concentrations of “°Mo, 1327¢ ®Nb, ©Ru, and Ru are found in the gas phase. Indeed, as Table VII indicates, the gas samples represent the largest fractions observed for each of these isotopes except *°Nb. In addition, appreciable fractions of '°Ag and of palladium isotopes have been observed in the purge gas samples. An appreciable quantity of '°'I appears in the gas phase; it seems very likely that its appearance there is due to ‘‘volatilization’’ of its tellurium precursor. The mechanisms by which the Mo, Nb, Te, and Ru isotopes appear in the gas phase are still not fully understood. NbFs, MoFs, RuFs, and TeFq are all volatile (although Ag and Pd have no volatile fluorides); these volatile fluorides, however, are far too unstable, with the possible exception of NbFs, to exist at the redox potentials in MSRE. Stability of lower fluorides of these elements is poorly known; it is possible that some of these may contribute to the ‘‘volatility.”’ It seems much more likely, however, that all these species exist in the reduced MSRE fuel in elemental form. They originate as (or very rap- idly become) individual metal atoms. They ag- gregate in the fuel at some slow but finite rate, and become insoluble as very minute colloidal particles which then grow at a slower rate. These colloidal particles are not wetted by the fuel, they tend to collect as gas-liquid interfaces, and they can readily be swept into the gas stream of the helium purge of the pump bowl. They tend to plate upon the metal surfaces of the system, and (as extraordinarily fine ‘‘smoke’’) to penetrate the outer layers of the moderator. While there are difficulties with this interpretation it seems more plausible than others suggested to date. Deposition of fission products on or in the graphite is, of course, undesirable since there they serve most efficiently as nuclear poisons. Of the species studied only Nb (whose carbide is . stable at MSRE temperatures) seems to prefer the graphite. It seems likely that in a MSBR the molten fuel will contain virtually all of the zirconium, the lanthanides and yttrium produced, and a large fraction of the iodine, rubidium, cesium, stron- tium, and barium. Thess species would, therefore, be available for removal in the processing plant. The salt will contain very little Mo, Nb, Te, Ru, and (presumably) technetium. These will report to the MSBR metal, graphite, and purge gas sys- tems in fractions which are probably not very different from those indicated for MSRE. FEBRUARY 1970 149 Grimes MSBR CHEMISTRY SEPARATIONS CHEMISTRY Molten-salt breeder reactors will require ef- fective schemes for recovery of the fissionable material which is produced and for removal of fission product poisons from the fuel. MSRE pro- duced no significant quantity of fissionable isotope and made no provision for on-stream removal of fission products other than the noble gases Kr and Xe. Recovery of uranium from molten fluorides by volatilization as UFs and the subsequent decon- tamination of this UFe by sorption-desorption on beds of NaF is well demonstrated. Recovery of uranium from MSRE or MSBR fuel by this scheme is clearly feasible. However, performance of a single-fluid MSBR is markedly improved if the bred protactinium is removed from the circulating fuel and permitted to decay to 2**U outside the reactor core. Fluorination is not effective in re- moval of protactinium from molten fluoride mix- tures. Development of other processes for Pa recovery has, accordingly, been necessary. In molten-salt reactors, from which Kr and Xe are effectively removed, the most important fis- sion product poisons are among the lanthanides. These fission products, as indicated above, form stable fluorides which are soluble in the fuel and will report almost quantitatively with the fuel to the separations plant. Their removal especially from fuels with high ThFs concentrations is diffi- cult. Rare earths have, accordingly, received more attention than other fission products. Processes for recovery of Pa and the lan- thanide fission products from MSBR fuels are described in some detail by Whatley et al.*® The chemical basis for such processes is described very briefly below. Separation of Protactinium Protactinium is produced in the MSBR fuel by reaction of a neutron with **?*Th to yield 2**Th which transmutes to ***Pa; the ***Pa decays with a 27.4-day half-life to °**U. Clearly, a separation of **Pa from the fuel needs, if it is to be effec- tive, to be accomplished on a cycle time short compared with this half-life. It is desirable to process the fuel on a cycle of three to five days. The process needs, therefore, to be relatively simple. Removal of Pa from LiF-BeF;-ThF,; mixtures by addition of BeO, ThO., or even UO: has been demonstrated in laboratory-scale experiments?®® Addition of these oxides to MSBR fuel composition would, as indicated in a previous section of this report, result in a uranium-oxide-rich solid solu- tion of UOz-ThO2. Whether the protactinium oxide 150 NUCLEAR APPLICATIONS & TECHNOLOGY was adsorbed on the added oxide or participated in the solid solution formation is not known. The process was shown to be reversible; treatment of the oxide-fluoride system with anhydrous HF dis- solves the oxide and returns the Pa to the fluoride melt from which it can readily be reprecipitated. It seems likely that Pa might be removed from the MSBR fuel by effective oontact of a fuel mix- ture side stream with UO»-ThO, solid solution. Such a possibility—which is still under considera- tion in laboratory-scale equipment—would add appreciably to the **U inventory through holdup of UO: in the solid solution. It would also probably require that the oxide ion concentration in the fuel after contact be diminished (as by treatment with HF) before the melt could be returned to the reactor. Separations based upon reduction of the Pa (present in the fuel as PaF,) to metal and extrac- tion into liquid metals appears at present to be more attractive.*® Such separations have the fol- lowing basis. When thorium metal, dissolved as a dilute alloy in bismuth (Thg;), is equilibrated with a molten mixture of LiF, BeF:, and ThF, several equilibria must be satisfied simultaneously. These include LiF (4) + 1 Th(gj) = i ThFq4(q) + Li(g;) (17) BeF2(d4)+ 3Th(pi) =3ThFs + Be(gj (18) 1 ThF4() + i Th(gy = ThFs g , (19) where (d) and (Bi) indicate that the species are in fluoride solution and alloyed with the Bi, respec- tively. Reaction (19) seems to be of little im- portance since only a small quantity (if any) ThFs; forms. Reaction (18) does not occur perceptibly; Be is “‘insoluble’’ in Biand forms no intermetallic compounds with that liquid metal. Reaction (17) occurs to an appreciable extent even though LiF (see Table IV) is much more stable than ThF,.'° Uranium forms stable dilute alloys in Bi, and its fluorides are less stable than ThF4.' Accord- ingly, the reactions UF4(q)+ sTh(pi) = 1 ThF4(q) + UFs(q) (20) and UFs(q) + 1 Thgi)= U(gi) + 3 ThF4g) (21) proceed very far to the right with low concentra- tions of Th (and Li) in the Biat equilibrium. Uran- ium 1is, therefore, very readily extracted from MSBR fuel melts. *'%2 It the LiF-BeF:-ThF, melt contains PaF,, the reaction PaF4(d) + Th(Bi) - ThF4(d) + Pa(Bi) (22) VOL. 8 FEBRUARY 1970 proceeds to quite a convenient extent®® "*°as shown in Fig. 6 although protactinium is reduced less easily than is uranium. Accordingly, if a melt containing both UFs and PaF,4 is contacted with bismuth and thorium metal added in small incre- ments the uranium and the protactinium are ex- tracted. Preliminary experiments® disclosed that Pa could indeed be removed from the salts by reduc- tion with Th but material balance data were very poor and little Pa appeared in the Bi. More recent experiments have used equipment of molybdenum, which seems essentially inert to the salt and the alloy, and very careful purification of the ma- terials and the cover gas.*” > By these techniques two sets of investigators®’™ have shown values in excess of 1500 for the separation factor = —_— (23) Th Npar, NTh a where N indicates mole fraction of the indicated species either in metal or in LiF-BeF2:-ThF4 so- lution. The separation factor for U from Pa has been shown to be above 25. These separation fac- tors are, of course, quite adequate for process design, as discussed in detail by Whatley et al.*® 100§ VN | 90 80 ¢4 a o o PROTACTINIUM IN BISMUTH o PROTACTINIUM IN SALT s THORIUM METAL IN BISMUTH 70 0 | FLOWING ARGON STATIC ARGON 50 “ FLOWING - . FLOWING ARGON A HELIUM ] o | | 40 30 20 Z 4t o > I THORIUM METAL ADDED OR TOTAL PROTACTINIUM (percent) L l | I I | I I | I | I | | | I A I | | I. I | | I I l I | | A A | a A A | | | I I CONTACT TIME (h) Fig. 6. Distribution of protactinium and material bal- ance for equilibrium of LiF-BeF,-ThF, (72-16- 12 mole%) with Bi-Th alloy (0.93 wt% Th). NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 b L 0 S 25 45 65 85 105 125 145 165 Grimes MSBR CHEMISTRY Separation of Rare Earths There is no real doubt that the rare-earth elements, which form fluorides that are more stable than ThF.' will be present in MSBR fuel sent to the processing plant. The limited solubility of these trifluorides (though sufficient to prevent their precipitation under normal MSBR conditions) has suggested a possible recovery scheme. When a LiF-BeF:- ThF,-UF. melt (in the MSBR concentration range) that is saturated with a single rare-earth fluoride (LaFs, for example) is cooled slowly, the precipi- tate is the pure simple trifluoride. When the melt contains more than one rare-earth fluoride the precipitate is a (nearly ideal) solid solution of the trifluorides.*®*® Accordingly, addition of an ex- cess of CeFs; or LaFs to the melt followed by heating to effect dissolution of the added tri- fluoride and cooling to effect crystallization ef- fectively removes the fission product rare earths from solution. It is likely that effective removal of the rare earths and yttrium (along with UFs and PuF;) could be obtained by passage of the fuel through a heated bed of solid CeFs or LaFs. How- ever, the price is almost certainly too high; the resulting fuel solution is saturated with the sca- venger fluoride (LaFs or CeFs, whose cross sec- tion is far from negligible) at the temperature of contact. Similar solid solutions are formed by the rare- earth trifluorides with UFs. It might, accordingly, be possible to send a side-stream (with the **Pa and 2**U removed by methods described above) through a bed of UF; to remove these fission product poisons; 2**UF; would presumably be used for economic reasons. The resulting LiF-BeF:- ThF, solution would be saturated with UFs after its passage through the bed. This ***U would have to be removed (for example, by electrolytic re- duction®® into molten Bi or Pb) before the salt could be returned to the cycle. The UF; bed could be recovered by fluorination to separate the ura- nium and rare earths. While this process probably deserves further study, the instability of UF; in melts with high UFs/UF, ratios and the ease with which uranium alloys with most structural metals would tend to make application of such a process unattractive. Removal of rare-earth ions (and other ionic fission-product species) by use of cation ex- changers has always seemed an appealing possi- bility. The ion exchanger would, of course, need (a) to be quite insoluble, (b) to be extremely un- reactive (in a gross sense) with the melt, and (c) to take up rare-earth cations in exchange for ions of low-neutron cross section. The bed of CeF; described above functions as an ion ex- FEBRUARY 1970 151 Grimes MSBR CHEMISTRY changer; it fails to be truly beneficial because it is too soluble in the melt. Unfortunately, not many materials are truly stable to the LiF-BeF:-ThF4.-UF, fuel mixture. Zirconium oxide is stable (in its low temperature form) to melts whose Zr*'/U*" ratio is in excess of ~3, and UO:-ThO: solid solutions are stable at equilibrium U/Th ratios. It is conceivable that sufficiently dilute solid solutions of Ce, O; in these oxides would be stable and would exchange Ce’" for other rare earth species. Intermetallic compounds of rare earths with moderately noble metals (or rare earths in very dilute alloys with such metals) seem unlikely to be of use because they are unlikely to be stable toward oxidation by UF4. Compounds with oxygenated anions (such as silicates and molybdates) are decomposed by the fluoride melt; they precipitate UO., from the fuel mixture. It is possible that refractory compounds (such as carbides or nitrides) of the rare earths, either alone or in solid dilute solution with an- alogous uranium compounds, may prove useful. By Reduction The rare-earth fluorides are very stable toward reduction to the metal. For example, at 1000°K the reaction 2/3LaFs () + Be()— 2/3La(s+ BeFzqy , (24) where (c) and (1) indicate crystalline solid and liq- uid, respectively, shows ~+30 kcal for the free en- ergy of reaction.'® With the LaF; in dilute solution and BeF; in concentrated solution in LiF-BeF, mix- ture the situation is, of course, even more un- favorable. However, the rare-earth metals form extremely stable solutions®™ in molten metals such as bismuth; the activity coefficient for La at low concentrations in Bi is near 10™%, Therefore, the reaction ZLaF; @) +Be) = %Lapj + BeFz2) , (25) where (d) indicates that the species is dissolved in 2LiF*BeF2, and (c) indicates crystalline solid can be made to proceed essentially to completion. Accordingly, LaF; can be reduced and extracted into molten Bi from LiF-BeF. mixtures. Figure 7 shows the behavior of several rare-earth ele- ments™ upon extraction into bismuth using Li° (which is more convenient to use) as the reductant. Such a process would seem to be useful, However, when the melt is complicated, as is necessary for the single-fluid breeder, by the addition of large quantities of ThF, the situation becomes considerably less favorable. Use of Li° concentrations as high as those shown in Fig. 7 is prohibited since the reaction 152 NUCLEAR APPLICATIONS & TECHNOLOGY o o NEODYMIUM o N LANTHANU o o o O SAMARIUM DISTRIBUTION OF RARE EARTH (mole fraction in metal / mole fraction in salt) 1072 1072 107! 10° To} LITHIUM FOUND IN METAL PHASE (at.%) Fig. 7. Effect of lithium concentration in metal phase on the distribution of rare earths between LiF- BeF, (66-34 mole%) and bismuth at 600°C, 2Bi(j) + $ThF4)+ Li(pi) = LiF(q)+ $ThBiz(c) (26) proceeds at lower concentrations than these to yield solid ThBi:. Accordingly, the most reducing metallic phase which is tolerable is that repre- sented by the (barely) saturated solutlon of ThBi, in Bi. When such solutions are used the rare- earth fluoride is much less completely extracted and relatively small separation factors from tho- . rium are obtained.®™:*® Figure 8 shows data obtained® in extraction of Ce at 600°C from sev- eral LiF-BeF:-ThF, mixtures (the points are labeled to indicate mole percentages of LiF, BeF;, and ThFy4, respectively) in this way. Separation factors appreciably above unity are obtained even in the worst case, and compositions containing 12 mole% ThFs with < 20 mole% BeFs; appear somewhat more attractive. Unfortunately, Ce’+ is one of the easiest of the rare earths to reduce. VOL. 8 FEBRUARY 1970 Others in the series show even smaller separation factors. Another paper in this series presents in- formation showing how processes with separation factors of this magnitude could be operated.*® There is little doubt, however, that improved (larger) separation factors would be desirable; there is reason to believe that improvement can be obtained by modification of the alloy phase. 6 T 74-20-6 5 . ® _ T z 9 / ® 0 | | Na . [ }] & : 8 i « Y O N o G3l——ti—FR 3 RS S o < © // g » ° / NOTE: aQ / VALUES CALCULATED n . AT THORIUM SATURATION 3 IN BISMUTH 1 M { o 0 w0 O -20 -10 0 10 20 30 EXCESS FREE FLUORIDE = mole % LiF —2(mole % Ber) -3(mole % ThFy,) Fig. 8. Effect of salt composition on the separation of cerium from thorium during reductive extrac- tion of cerium from LiF-BeF,-ThF, mixtures into bismuth at 600°C. SUMMARY Phase behavior of LiF-BeF:2:-ThF4-UFs mix- tures appears suitable to permit use of a high concentration of ThFs in melts whose freezing point will prove acceptable for single-fluid molten-salt breeder reactors. Oxide tolerances of such mixtures are not entirely defined, but laboratory experience and the MSRE operating experience combine to suggest that no oxide scavenger (such as ZrF,;) need be added. The design basis fuel, accordingly, consists of 71.7 mole% "LiF, 16 mole% BeF:, 12 mole% ThF,4, and 0.3 mole% 2**UF.. The secondary coolant is to be 8 mole% NaF with 92 mole% NaBF4. Physical NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 FEBRUARY 1970 Grimes MSBR CHEMISTRY properties of these fluids, estimated in most cases from data on similar materials, seem adequate for their use in MSBR. Compatibility of the LiF-BeF:-ThFs-UF4 melt with Hastelloy-N and with moderator graphite seems assured, and operation of LiF-BeF2-ZrF,- UF, melts in MSRE indicate very strongly that the compatibility will not be adversely affected by the consequences of radiation and fission of the ura- nium, Compatibility of the secondary coolant, while less satisfactory, appears to be adequate with Hastelloy-N if salt purity is maintained. Fission product Kr and Xe are virtually insolu- ble in the fuel and can be removed, if the moder- ator graphite is sufficiently impermeable, Dby simple equilibration with an inert gas (helium). Fission products with stable fluorides (Rb, Cs, Sr, Ba, the lanthanides, and Zr) appear almost entirely in the fuel as fluorides except as they are lost through volatilization of precursors. More noble fission products (Nb, Mo, Ru, and Te) are virtually absent from the fuel, but plate (in ap- preciable amounts) on the graphite and the Hastel- loy-N, and appear to a surprisingly large extent in the helium purge gas. Chemical separations, of which reductive ex- traction appears most attractive, for removing uranium and protactinium from the fuel salt and from each other have been demonstrated in small scale laboratory equipment. Separations of lan- thanides from the fuel are markedly more diffi- cult, but reductive extraction of these elements into molten bismuth appears possible. While much research and development remains to be accomplished there seems to be no funda- mental chemical difficulty with design and opera- tion of a single-fluid molten-salt breeder system. ACKNOWLEDGMENTS This research was sponsored by the U.S, Atomic Energy Commission under contract with the Union Carbide Corporation, REFERENCES 1. M. W. ROSENTHAL, P. R. KASTEN, and R. B. BRIGGS, ‘‘Molten Salt Reactors—History, Status, and Potential,’”’ Nucl. Appl. Tech., 8, 107 (1970), 2. E. S. BETTIS and R. C. ROBERTSON, ‘“MSBR De- sign and Performance Features,”’ Nucl. Appl. Tech., 8, 190 (1970). 3. A. M. PERRY and H. F. BAUMAN, ‘“MSBR Reactor Physics,”’ Nucl. Appl. Tech., 8, 208 (1970). 153 Grimes MSBR CHEMISTRY 4. W. R. GRIMES, F. F. BLANKENSHIP, G. W. KEIL- HOLTZ, H. F. POPPENDIEK, and M. T. ROBINSON, ““Chemical Aspects of Molten Fluoride Reactors,’”’ in Progress in Nuclear Enevgy, Series IV, Vol. 2, Per- gamon Press, London (1960), 5. W, R. GRIMES, D. R. CUNEO, F. F. BLANKENSHIP, G. W. KEILHOLTZ, H. F. POPPENDIEK, and M. T. ROBINSON, ‘“Chemical Aspects of Molten Fluoride Salt Reactor Fuels,”’ in Fluid Fuel Reactors, J. A. LANE, H. G. MacPHERSON, and FRANK MASLAN, Eds., Addi- son-Wesley, Reading, Mass. (1958). 6. W. R. GRIMES, ‘‘Materials Problems in Molten Salt Reactors,” in Materials and Fuels for High Tempera- ture Nuclear Energy Applications, M. T. SIMNAD and L. R. ZUMWALT, Eds., The M.I.T. Press, Mass. (1964), 7. W. R. GRIMES, ‘“MSR Program Semiannual Prog- ress Report, July 31, 1964, ORNL-3708, p. 214, Qak Ridge National Laboratory. 8. ALVIN GLASSNER, ‘““The Thermochemical Proper- ties of the Oxides, Fluorides, and Chlorides to 2500°K,”’ ANL-5750, Argonne National Laboratory, 9. L. BREWER, L. A. BROMLEY, P, W. GILLES, and N. L. LOFGREN, MDDC-1553 (1945) and L., BREWER in The Chemistry and Metallurgy of Miscellaneous Ma- terials; Therymodynamics, L. L., QUILL, Ed., McGraw- Hill, New York, pp. 76-192 (1950). 10. R. E. THOMA, Ed., ‘‘Phase Diagrams of Nuclear Reactor Materials,”” ORNL-2548, Oak Ridge National Laboratory (November 6, 1959). 11. L. V. JONES, D. E. ETTER, C. R. HUDGENS, A. A. HUFFMAN, T. B. RHINEHAMMER, N. E. ROGERS, P, A. TUCKER, and L. J. WITTENBERG, ‘‘Phase Equi- libria in the Ternary Fused-Salt System LiF-BeF - UF,’” J. Am. Ceram. Soc., 45, 79 (1962). 12, R, E. THOMA, H. INSLEY, H., A. FRIEDMAN, and C. F. WEAVER, ‘‘Phase Equilibria in the Systems BeFa- ThFy and LiF-BeF,-ThF,’’ J. Phys. Chem., 64, 865 (1960). 13. C. J. BARTON, H. A. FRIEDMAN, W. R, GRIMES, H. INSLEY, and R, E, THOMA, ‘‘Phase Equilibria in the Alkali Fluoride-Uranium Tetrafluoride Fused Salt Sys- tems: 1. The Systems LiF-UF, and NaF-UF,” J. Am. Cervam. Soc., 41, 63 (1958). 14. R. E. THOMA, H. INSLEY, B. S. LANDAU, H. A. FRIEDMAN, and W, R. GRIMES, ‘‘Phase Equilibria in the Fused Salt Systems LiF-ThF, and NaF-ThF,’’ J. Phys. Chem., 63, 1266 (1959). 15. C. F. WEAVER, R. E. THOMA, H. INSLEY, and H. A. FRIEDMAN, ‘Phase Equilibria in the Systems UF4ThF4 and LiF-UF4-ThF,,’’ J. Am. Cevam. Soc., 43, 213 (1960), 16. K. A. ROMBERGER, C. F. BAES, and H, H. STONE, ““Phase Equilibrium Studies in the Uranium IV ) Oxide- 154 NUCLEAR APPLICATIONS & TECHNOLOGY Zirconium Oxide System,’’ presented at 151st National Meeting of American Chemical Society, Pittsburgh, Pa., March 21-31, 1966, 17. C. F. BAES and C. E. BAMBERGER, ‘“MSR Pro- gram Semiannual Progress Report, August 31, 1968,” ORNL-4344, Oak Ridge National Laboratory. 18. A. L. MATHEWS and C. F. BAES, “Oxide Chemistry and Thermodynamics of Molten LiF-BeF, by Equilibri- um with Gaseous Water-Hydrogen Fluoride Mixtures,”’ ORNL-TM-1129, Oak Ridge National Laboratory (May 1965). » 19. C. F. BAES, ‘““The Chemistry and Thermodynamics of Molten Salt Reactor Fluoride Solutions,’”’ Proc. IAEA Symp. on Thermodynamics with Emphasis on Nuclear Materials and Atomic Transport in Solids, Vienna, Aus- tria (July 1965), 20. C. F. BAES and B. F. HITCH, ‘‘Reactor Chemistry Division Annual Progress Report, January 31, 1965,” ORNL-3789, pp. 61-5, Oak Ridge National Laboratory. 21. E. M. LEVIN, C. R. ROBBINS, and H. F. McMURDIE, Phase Diagvams for Cevamists, The American Ceram- ic Society, Inc., Ohio (1964). 22. V. G. SELIVANOV and V. V. STENDER, Zhur. Neorg. Khim., 3, 2, 448 (1958). 23. C. J. BARTON, L. O. GILPATRICK, H. INSLEY, and T. N. McVAY, ‘““MSR Program Semiannual Prog- ress Report, February 29, 1968, ORNL-4254, p. 166, Oak Ridge National Laboratory. 24. S. CANTOR, J. W. COOKE, A. S. DWORKIN, G. D. ROBBINS, R. E. THOMA, and G. M. WATSON, ‘Physical Properties of Molten Salt Reactor Fuel,”” ORNL-TM- 2316, Oak Ridge National Laboratory (August 1968), 25. H. E. McCOY, W. H. COOK, R. E. GEHLBACK, J. R. WEIR, C. R. KENNEDY, C. E. SESSIONS, R. L. BEATTY, A. P. LITMAN, and J. W. KOGER, ‘*New Developments in Materials for Molten-Salt Reactors,’’ Nucl. Appl. Tech., 8, 156 (1970). 26. C. F. BAES, Jr., “The Chemistry and Thermody- namics of Molten Salt Reactor Fuels,”” presented at AIME Nuclear Fuel Reprocessing Symposium at Ames Laboratory, Ames, Iowa, August 25, 1969. Published in the 1969 Nuclear Metallurgy Symp., Vol. 15 by the USAEC Division of Technical Information Extension. 27. G. LONG, ‘‘Reactor Chemistry Division Annual Progress Report, January 31, 1965, ORNL-3789, p. 65, Oak Ridge National Laboratory, 28. “MSR Program Semiannual Progress Report, Au- gust 31, 1961, ORNL-3215, p. 93, Oak Ridge National Laboratory. 29. C. E. WICKS and F. E. BLOCK, “Thermodynamic Properties of 65 Elements—Their Oxides, Halides, Car- bides, and Nitrides,’”” Bureau of Mines Bulletin 605 (1963). VOL. 8 FEBRUARY 1970 30. JANAF (Joint Army-Navy-Air Force) Interim Ther- mochemical Tables, Thermal Research Laboratory, Dow Chemical Co., Midland, Mich, 31. P. N. HAUBENREICH and J. R. ENGEL, “Operation of the Molten-Salt Reactor Experiment,”” Nucl. Appl. Tech., 8, 118 (1970). 32. R. F. APPLE, Method No. 9021206 (March16,1969), ““ORNL Master Analytical Manual,”” TID-7015 (Suppl. 8). 33. R. E. THOMA, ‘“MSR Program Semiannual Prog- ress Report, August 31, 1968,”> ORNL-4344, p. 110-112, Oak Ridge National Laboratory. 34, R, E. THOMA, “MSR Program Semiannual Prog- ress Report, February 29, 1968,”” ORNL-4354, pp. 88- 93, Oak Ridge National Laboratory. 35. R. E. THOMA, ‘“MSR Program Semiannual Prog- ress Report, August 31, 1967,° ORNL-4191, pp. 102- 108, Oak Ridge National Laboratory. 36. S. S. KIRSLIS and F. F. BLANKENSHIP, ‘‘Reactor Chemistry Division Annual Progress Report, December 31, 1966, ORNL-4076, pp. 45-54, Oak Ridge National Laboratory. 37. S. S. KIRSLIS, ‘““MSR Program Semiannual Prog- ress Report, August 31, 1966,”> ORNL-4037, pp. 165- 166, Oak Ridge National Laboratory. 38. S. S. KIRSLIS and F. F. BLANKENSHIP, ‘“‘MSR Program Semiannual Progress Report, August 31, 1967, ORNL-4191, pp. 116-120, Oak Ridge National Laboratory. 39. S. S. KIRSLIS and F. F. BLANKENSHIP, ‘“MSR Pro- gram Semiannual Progress Report, February 28, 1968,” ORNL-4254, pp. 94-114, Oak Ridge National Laboratory. 40, S. S. KIRSLIS and F, F. BLANKENSHIP, ‘“MSR Pro- gram Semiannual Progress Report, August 31, 1968, ORNL-4344, pp. 115-141, Oak Ridge National Laboratory. 41, F. F. BLANKENSHIP, S. S. KIRSLIS, J. E. SAVO- LAINEN, and W. R. GRIMES, ‘‘Reactor Chemistry Divi- sion Annual Progress Report, January 31,1963,”” ORNL- 3417, p. 17, Oak Ridge National Laboratory. 42, W. R. GRIMES, N. V. SMITH, and G. M. WATSON, “‘Solubility of Noble Gases in Molten Fluorides. I. In Mixtures of NaF-ZrF, (53-47 mole %) and NaF-Zrk4- UF, (50-46-4 mole %),”’ J. Phys. Chem., 62, 862 (1958). 43. M. BLANDER, W, R. GRIMES, N. V., SMITH, and G. M. WATSON, ‘‘Solubility of Noble Gases in Molten Fluorides. II. In the LiF-NaF-KF Eutectic Mixtures,’’ J. Phys. Chem., 63, 1164 (1959). 44, G. M. WATSON, W. R. GRIMES, and N, V., SMITH, ‘‘Solubility of Noble Gases in Molten Fluorides. In Lithium-Beryllium Fluoride,”’ J. Chem. Eng. Data, T, 285 (1962). 45. W. R. GRIMES, J. H. SHAFFER, N. V. SMITH, R. A. STREHLOW, W. T. WARD, and G. M. WATSON, ‘‘High- NUCLEAR APPLICATIONS & TECHNOLOGY VOL. 8 Grimes MSBR CHEMISTRY Temperature Processing of Molten Fluoride Nuclear Reactor Fuels,”’ Nucl. Eng.—Pavt VII, 55, 27, 65 (1959). 46. W. T. WARD, W. R. GRIMES, R. A. STREHLOW, and G. M. WATSON, ‘““Rare Earth and Yttrium Fluor- ides. Solubility Relations in Various Molten NaF-ZrFg4- UF,Solvents,”’” J. Chem. Eng. Dala, §, 137 (1960). 47, C. F. BAES, Jr., ‘“Reactor Chemistry Division Annual Progress Report, January 31, 1967, ORNL- 4076, pp. 49-50, Oak Ridge National Laboratory. 48. M. E. WHATLEY, L. E. McNEESE, W. L. CARTER, L. M. FERRIS, and E. L. NICHOLSON, ‘‘Engineering Development of MSBR Fuel Recycle,”” Nucl. Appl. Tech., 8, 170 (1970). 49, C. J. BARTON, ‘‘Reactor Chemistry Division An- nual Progress Report, December 31, 1965, ORNL- 3913, p. 45, Oak Ridge National Laboratory. 50. J. H. SHAFFER, W. R. GRIMES, G. M. WATSON, D. R. CUNEO, J. E. STRAIN, and M. J. KELLY, ‘‘The Recovery of Protactinium and Uranium from Molten Fluoride Systems by Precipitation as Oxides,’”’ Nucl. Sci. Eng., 18, 177 (1964). 51. J. H. SHAFFER, D. M. MOULTON, and W. R, GRIMES, ‘‘MSR Program Semiannual Progress Report, February 29, 1968, ORNL-4254, pp. 152-155, Oak Ridge National Laboratory. 52. L. M. FERRIS, J. F. LAND, J. J. LAWRENCE, and C. E. SCHILLING, ‘“MSR Program Semiannual Prog- ress Report, February 29, 1968,”> ORNL-4254, pp. 241- 247, Oak Ridge National Laboratory. 53. L. M. FERRIS, J. C. MAILEN, E. D. NOGUEIRA, D. E. SPANGLER, J. J. LAWRENCE, J. F. LAND, and C. E. SCHILLING, ‘“MSR Program Semiannual Progress Report, August 31, 1968, ORNL-4344, pp. 292-297, Oak Ridge National Laboratory. 54, C. J. BARTON and R. G. ROSS, ‘“MSR Program Semiannual Progress Report, August 31, 1968,”” ORNL- 4344, pp. 180-185, Oak Ridge National Laboratory. 55. J. H. SHAFFER, D. M. MOULTON, and W. R. GRIMES, ‘‘MSR Program Semiannual Progress Report, August 31, 1968, ORNL-4344, pp. 174-180, Oak Ridge National Laboratory. 56. J. H. SHAFFER, W. R. GRIMES, F. A. DOSS, W. K. R. FINNELL, and W, P, TEICHERT, ‘“Reactor Chem- istry Division Annual Progress Report, December 31, 1965, ORNL-3913, p. 42, Oak Ridge National Labora- tory. 57. J. J. EGAN and R. W. WISWALL, Nucleonics, 15, 104 (1967). 58, J. H. SHAFFER, W. P. TEICHERT, D. M. MOUL- TON, F. F. BLANKENSHIP, W. K. R. FINNELL, and W. R. GRIMES, ‘“‘Reactor Chemistry Division Annual Progress Report, December 31, 1966, ORNL-4076, p. 34-36, Oak Ridge National Laboratory. FEBRUARY 1970 155