. - ) ORNL-TM-4056 THE EQUILIBRIUM OF DILUTE UF3 SOLUTIONS CONTAINED IN GRAPHITE L. M. Toth L. O. Gilpatrick MASTER INSTINTER 27 7003 oincivnaeny 1 UNLITER Ce '3‘ 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 usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. ORNL-TM- 4056 Contract No. W-7405-eng-26 REACTOR CHEMISTRY DIVISION THE EQUILIBRIUM OF DILUTE UF3 SOLUTIONS CONTAINED IN GRAPHITE L. M. Toth and L. 0. Gilpatrick 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, DECEMBER 1972 OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37830 operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION DISTRERUTIGN OF THIS DOCUMENT IS URLIMITED U THE EQUILIBRIUM OF DILUTE UF3 SOLUTIONS CONTAINED IN GRAPHITE L. M. Toth and L. 0. Gilpatrick ABSTRACT The equilibrium of dilute UF3-UF, molten fluoride solutions in contact with graphite and UC2: 3UF, + UC, < 4UF, + 2C has been studied as a function of temperature (370-700°C), melt com— position and_atmospheric contamination. Equilibrium quotients, Q = (UF3)4/(UF4)3 for the reaction were determined by measuring the UF4 and UF, concentrations spectrophotometrically. The solvents used were primarily LiF-BeF2 mixtures. Results from this solvent system were related to the reactor solvents LiF-BeF9~-ZrF,(65.6-29.4-5 mole %) and LiF-BeFp-ThF,;(72-16-12 mole %). It has been found that the equilibrium quotient is very sensitive to both temperature and sol- vent changes increasing as either the temperature increases or the alkali-metal fluoride content of the solvent decreases. INTRODUCTION The relative stability of dilute UF3-UF, molten fluoride solutions contained in graphite is of practical importance to Molten Salt Breeder Reactors, MSBR, in which these solutions are used as nuclear fuels. Be- cause the reactors contain a large amount of graphite in the core serving as a neutron moderator, reaction of UF4 with graphite: * 4UF3(d) + 2C 1'3UF4(d) + U02 (1-1) to form UF4 and uranium dicarbide has long been recognized1 as a major factor limiting the amount of UF3 which can be maintained in solution. Although typical fuel mixtures consist essentially of 1 mole ¥ U235F4 or less in a solution of LiF and BeFZ, the ease of UF4 reduction to UF3 by the chromium in the metal containment vessel** o - + -— ZUFA(d) Cr™ & 2UF3(d) + Cer(d) (1-2) * The subscript "d" indicates that the component is in solution. **Hastelloy N, a nickel-based alloy containing Cr, Fe and Mo has been used? to fabricate the metal containment vessel for the Molten Salt Reactor Experiment, MSRE. necessitates the consideration of UF; chemistry as well. The effect of the corrosion reaction of Eq. 1-2 is to leach chromium from the structural metal and cause it to appear in solution as CrFj. In order to minimize the corrosion, the equilibrium of Eq. 1-2 is shifted to the left by reducing a small percentage (approximately 17%) of the UF, to UF3 through the addition of beryllium metal:3 2UF + Be® T 2UF + BeF (1-3) 4(d) 3(d) 2 Although a small amount of UF3 is beneficial in reversing the corrosion mechanism, it produces complications due to possible reaction via Eq. 1-1 and the resulting undesirable formation of an insoluble uranium carbide. Reference to "UFg3 stability" in this paper will therefore mean specifically the equilibrium concentration of UFj3 relative to UF, as determined by Eq. 1-1. This equilibrium has never been experimentally measured despite the fact that it is the major factor in determining UF3 stability for molten salt reactor systems. Although they used indirect means, Long and Blankens‘nip4 are the only investigators who have attempted to measure the equilibrium. Since their work is the basis on which all previous estimates of UFy stability have been made, it will be reviewed in detail, with the equilibrium expressions in fractional coefficients as used by the authors. They studied the reduction of UF, (both pure solid phase UF, and in molten fluoride solution) with hydrogen: 1 T - - SH, + UF, [ UF, + HF (1-4) and determined the equilibrium quotients for the above reduction: X Y R 3 Fyp R UF, Q=g 77 =K (1-5) Xop, pl/2 YuF 4 H2 3 by measuring HF and H, ratios evolved from a reaction vessel containing UF, and UF,. From the solid-phase UF4 reduction they obtained equilibrium 4 3 constants, KR, for the reduction. These, combined with the equilibrium R X . . . quotient, Q , for the dilute solutions and the activity coefficient for UF3, YUF33 obtained from solubility data, enabled them to calculate the activity coefficient for UF,, YUF,» in the molten fluoride solution. By combining the free energy expression for Eq. 1-4 with one for the decompo- sition of UF3 into UF& and UO: —>§_ 1.0 - UF3(d) 7 QUFA(d) +Z{I (1-6) they obtained an expression for the equilibrium quotient, QD, of Eq. 1-6, in terms of the equilibrium quotient for Eq. 1-4, QR, and the activity co- efficients of UF, and HF. (c.f. p 18, Ref. 4, part II). Using free ener- gies of formation for UCy and UC from Rand and Kubachewski,5 which were acceptable at the time, they estimated uranium activities in the carbides and concluded that solutions in which up to 607 of the initial 1 mole % of UF, is converted to UFq5 are expected to be stable in the presence of graph- ite. In addition they concluded that temperature and solvent changes should have little effect on the equilibrium mechanism of Eq. 1-1 since they found no significant effect from them on the H2 reduction mechanism of Eq. 1-4. In view of the significance of Eq. 1-1 to Molten Salt Reactor Technol- ogy, a closer examination of it is clearly warranted. The development of spectrophotometric techniques for the study of molten fluorides and the realization of solvent effects on molten fluoride chemistry, have given impetus to the study. We have aiready identified6 UC,, as the stable uran- ium carbide phase in equilibrium with UF3-UF, solutions in graphite. The object of this report is to describe the effects of temperature, solvent, and atmospheric contamination on the equilibrium. Both the forward and the back reaction of Eq. 1-1 in the reference solvent system LiF-BeF, have been followed. The data in the reference solvent system have been related to practical reactor solvents such as the Molten Salt Reactor Experiment, MSRE, solvent, LiF-BeF,-ZrF, (65.6-29.4~5 mole %) and the proposed Molten Salt Breeder Reactor, MSBR, solvent, LiF-BeF;-ThF, (72-16-12 mole %Z)}. Our find- ings are compared with earlier observations which have not been reviewed before. EXPERIMENTAL Equilibrium quotients for the back-reaction* (Eq. 3-2) were determined by measuring UF3 and UF, concentrations spectrophotometrically with a Cary Model 14-H recording spectrometer. The sample system consisted of a control- led temperature, inert atmosphere furnace shown in Fig. 2-1 which held a diamond-windowed graphite spectrophotometric cell.7 Molten fluoride salt solutions and reagent uranium carbides were contained in this cell. Ab- sorption spectra of the molten salt solution were measured against an air reference. Net spectra due to UF, and UF4 were determined by subtracting independently determined solvent slank spectra using standard digital com- puter techniques. Spectra were measured in the near infra-red and visible regions from 4000 to 33000 eml. The absorption spectra of UF5 and UF, served as the primary means of monitoring these components in solution as a function of temperature, time, and solvent composition. Materials - Molten salt solvent compositions were prepared by mixing calculated amounts of the pure component fluoride salts. Optical quality crystal fragments from the Harshaw Chemical Co. was the source of LiF. Beryllium fluoride was prepared by vacuum distillation8 from a large special purchase supplied by the Brush Beryllium Co. The water-—clear, glass-like product contained no spectrographically detectable cation im- purities, but was exceedingly hydroscopic and had to be stored under very anhydrous conditions. Uranium tetrafluoride was taken from a laboratory purified spectroscopic standard which contained less than 10 ppm of total cation impurities. Thorium tetrafluoride was part of a special purchase from the National Lead Co. which contained no greater than 100 ppm in any cation impurities. Storms and coworkers of the Las Alamos Scientific Laboratory supplied each of the uranium carbides used in this study and supplied the following analysis: Uranium dicarbide - UC, wt % C = 8.83 or 75.74 mole % C 02 = 200 ppm *See Results and Discussion Section for an explanation of why the back- reaction was measured. **Although uranium dicarbide is a substoichiometric com.pound,13 it will be identified as UCy in this paper. SAMPLE BEAM s REFERENCE BEAM s QO~NOOC WM . 2-1 High Temperature Furnace Molten Fluorides. ORNL-DWG 72-10653 NEUTRAL SCREEN LAVITE SUPPORT _ MINERAL WOOL - QUARTZ - VACUUM WINDOW O-RING SEAL BRASS SHELL TRANSFER CHAMBER (REMOVABLE) _ VACUUM SHUTTER { O-RING SEAL ] . TEFLON SLIDING SEAL CELL SUPPORT TUBE . STAINLESS STEEL STiRRER RGO . VACUUM OR HELIUM TEFLON GLAND a-in. TUBE FITTING .. NEOPRENE O-RING WATLOW HEATERS (NOT SHOWN) - COPPER BLOCK .. INCONEL. FURNACE . DIAMOND-WINDOWED GRAPHITE CELL _ Pt-10% Rh STIRRER TITANIUM SPONGE GETTER - CELL SUPPORT System for Absorption Spectra of 3.5251 + 0.00054 5.9962 + 0.00083 Uranium sesquicarbide - U,Cy wt “ZC=6.99 or 59.83 mole % C 0, = 50 ppm crystal lattice by X-ray a = 8.0889 1_0.0009R crystal lattice by X-ray a c I These high purity carbides were received as lusterous grayublack granules which ranged in size from 1/2 to 1 mm3. They were shipped sealed in glass ampules and stored in a helium filled dry box. Exposure to even the dry box conditions was kept to the absolute minimum needed for weighings and cell loadings. Procedure - Even though the reagent salts were quite free of cation impurities, they were not free of oxides and H90 to the degree needed. All compositions were therefore treated while molten at 600°C for oxide removal by sparging for several hours with reagent HF gas or HF-Hy gas mixtures.9 Residual HF was then stripped from the melt with He prior to cooling. Clean portions of the recovered salt "button" were then crushed and used to charge the spectrophotometer cell, by weighing out the fluorides in a hel- ium drybox which was maintained at a water vapor content < 0.1 ppm and at an 07 content < 2 ppm. Between 0.5 and 0.6 gm of salt solvent made a con- venient cell loading to which was added from 5 to as much as 100 mg of the uranium carbide under study. Poco AXF-5Ql grade graphitelo spectrophoto- metric cells were used which were purified after fabrication by heating in an Hyp gas stream to 1000°C and then flushed free of Hy, with He. Subsequent dry box handling and loading techniques have been discussed earlier.1l A "dash pot" stirrer made from platinum-10% rhodium (see Fig. 2-1) was used to hasten the attainment of equilibrium which is otherwise dependent large- ly on diffusion. It proved to be a great aid in shortening the time needed to reach equilibrium. We observed a small but temporary loss of transmig- sion directly after stirring in some cases which was equal to 0.15 absorb- ance units at 4000 cm™l. We have assumed this recoverable loss to be caused by the temporary suspension of fine particles which later settle. Whole grains of the carbide were used after early attempts to increase the surface area by crushing caused the carbide to collect at the window and interfere with the optical transparency of the cell. A large excess of the golid carbide phase was always maintained in the cell. On some occasions, the experimental sequence was interrupted and additional uranium carbide was added to demonstrate that an excess was indeed present. No change was observed in the concentrations of UFg or UF, in the homogeneous solutions as a result of these additions. Spectral Measurements - Molar concentrations of dissolved UF3 and UF, were determined simultaneously in solution at a series of temperatures a- bove the melting point by measuring optical densities at 9174 and 11360 cm-l. These wave numbers represent the maximum absorbance values for dissolved UF,4 and UFq4 respectively in the near infra-red region as shown in Figs. 2-2 and 2-3. The strong UF3 absorption in the visible region from 16000 to 33000 cm™ ! was in general too intense to be useful since the solutions studied had initial UF,; molarities in the range of 0.04 to 0.10. Figures 2-2 and 2-3 show that for spectra of pure UF, and UF4 there is a contribu- tion from each at the most sensitive absorbance region of the other member. Stated differently, the absorbance at 9174 em™1 in a mixed solution is primarily due to UF4, but not entirely so. This condition is solved unique- ly for the contriburion from each species by the solution of a set of simul- taneous linear equations equal to number to the number of components in the system which contribute to the net spectra, in our case 2. Using Beer's law -log I/I = A, = (e,) (DL (2-1) where I = measured optical intensity of the sample IO = measured optical intensity of the reference solvent A = total absorbance at a given frequency, V. (8\))T = molar absorption coefficient at v and temperature T (M), = molarity of component in solution 2 = cell path length = 0.635 cm The following set of equations are sufficient to determine the separate molar concentrations in a mixed solution at a particular temperature. _ 3 4 _ b 3 Ajp T AL T A T g WM e (M) R (2-2) 3 b 4 3 sz - Av2 + sz - E:\)2(1\{4)1‘£ + E\)2(M3)T2 (2-3) ABSORPTION COEFFICIENT (liters mole ™! cm™!) 20 18 12 10 ORNL-DWG 72-9634 l I ! l | | I I l i 10 " 12 13 14 16 18 20 22 24 (x 103) WAVENUMBER {cm™!) 2-2 UF, Spectrum (approximately 1 mole %) in LiF-BeF;-ThFy (72-16-16 mole %) at 575°C. ORNL-DWG 72-9635 19 O D n 1 - n N W L &) O U o g O ABSORPTION COEFFICIENT ( liters mole™! cm S I | | ! | | ;’ 1 1 —~ 900 — 800 700 K 600 - - 500 - 400 300 : / 200 — l ' i | | | i ! i I | l | 6 7 8 9 10 4 12 13 14 18 22 26 30 34 (x103) WAVENUMBER (cm™) UF3 Spectrum (approximately 0.3 mole %) in LiF-BeFp (66-34 mole %) 10 where: 1 it 9174 cm 11360 cm~ 1 " 1 2 3 = UF5 component 4 fl UF, component Solving Eqs. (2-2) and (2-3) simultaneously for (M4)T and (M3)T gives the desired molarities since AVl and sz are measured and the € values are known from previous calibrations. 4 3 oLy = (Evz Ayl T Eu1 sz) k2-4) 4°T 2(83 84 _ €3 €4 ) v2 vl vl v2 4 4 0Ly = (€51 A2 ~ €1 A (2-5) 3T ., 3 4 3 4 (€52%01 = Ev1fu2) Because spectra were recorded versus a neutral screen in the reference beam (see Fig. 2-1) it was always necessary to subtract a solvent spectrum, or blank, which was independently determined for each experimental spectrum, to get the net absorbance due to species in solution. Analyzing composite spectra required making calibrations for € with solvent melts containing a known concentration of pure UF, and UF5. Values of € are reduced with increasing temperature because of two effects: the change of molarity caused by thermal expansion and temperature effects on the absorption spectra themselves. Changes in molarity due to temperature changes were adjusted by using S. Cantor's data12 for the molal volume of various fused fluoride salts and assuming that the molal volumes are additive to within + 3% according to the following general relations: NT[xl(\)l)T + xz(vz)T —————— ] = 1000 ml (2-6) (Ml)T = NT x; (2-7) where NT = moles per liter of solution at temperature T x, = mol fraction of component #1 (\)l)T = molar vol of 1 at temperature T in cc/mole M 1)T = molarity of 1 in moles/l at temperature T 11 Molar absorptions were first measured for pure UF, solutions at various temperatures using a known concentration and at molarities adjusted ) and (832) are recorded in Table for expansion. Measured values of (E4 T v1°T 2-1. A corresponding calibration was performed for pure UF3 under identical conditions. This was most easily achieved by adding an excess of a reduc- ing agent. Both zirconium and uranium metal were used for this purpose, they react as shown in Eqs. (2-8) and (2-9). Zr + 4UF, T 4UF. + ZrF (2-8) 4 3 4 > U + 3UF4 < 4UF3 (2-9) The effect on the properties of the solutions caused by the production of ZrFA in Eq. 2-8 was very small and hence neglected for these dilute solu- tions. When uranium was used concentrations had to be increased by 1/3 over those calculated for UF4 in the initial solutions as shown in Eq. (2-9). Pure UF45 solutions in contact with graphite result in the loss of uranium from solution by the formation of UC2 as shown in Eq. (1-1). Fortunately this reaction is rather slow under the conditions that we have studied, and it was possible to correct for this loss by measuring absorbances as a function of time to determine the rate of loss (dAv/dT), and correcting for the loss by extrapolating back to zero time. Reducing UF4 with uranium does not result in a loss of UF3 from solution. The addition of Eq. (1-1) and (2-9) results in the cyclic conversion of U and C to UC2 with no net change of UF, concentration in solution as shown in Eq. (2-10). 3 4UF3 + 2C ~> 3UF, + UC, (1-1) 3UF4 + U~ 4UF3 (2-9) U+ 2C -+ UC (2-10) 2 An alternate approach to determining the molar absorption coefficients (€v) for UF3 in solution has also been used. Since UF4 solutions are more stable than UF3 solutions under our experimental conditions, the calibra- tion results for UF4 are more reliable and associated with less error than are those for UF3. Using this fact the uncertainty associated with the UF3 calibration can be reduced by measuring the absorption spectrum of a mixture Table 2-1 Molar Absorption Coefficients for Molten Fluoride Solutions of UF4 and UF3 Solution in L2B Solvent:LiF°BeF2 LB SolventILiF'Ber MSBR So:’LventzLiF'BeFZ'ThF4 Mole 4 (66.7-33.9) (48-52) (72-16-12) Spectra UF3 UF4 UF3 UF4 UF3 UF4 Mol ti Cge?;ii?zgip ton €11360 €9170 €11360 €9170 €11360 €9170 £11360 E9170 €11360 E9170 €11360 E€9170 Temperature °C 370 44,2 7.6 3.1 18.4 400 43.2 7.8 3.1 17.8 450 46,0 10.0 3.90 18.7 41.7 8.1 3.1 16.9 500 44,2 10.0 3.85 17.9 40.1 8.4 3.1 16.0 58.5 14.5 2.80 19.2 550 41.7 10.0 3.80 17.1 38.5 8.6 3.1 15.1 56.8 14.5 2.70 18.2 600 39.0 10.0 3.75 16.3 36.9 8.9 3.1 14.2 55.0 14.5 2.65 17.2 650 36.2 10.0 3.70 15.4 35.3 9.2 3.1 13.3 53.3 14.5 2.55 16.2 700 33.5 10.0 3.65 14.8 33.7 9.4 3.1 12.4 51.6 14,5 2.50 15.2 750 30.7 10.0 3.60 14.1 49,9 14.5 2.40 14.5 800 48.0 14.5 2.35 13.8 ¢t 13 of UF5j and UF, where the UFj is generated by partially reducing a dilute UF, solution of known concentration. (The reductant chosen for partial reduction was UCp.) The spectrum is then converted to digital form along with a UF4 reference spectrum. Using iterative computer techniques, varying amounts of the UF4 spectrum (i.e., k x (UF4 spectrum) where k is the coef- ficient which is varied in the iteration process) are subtracted until the resulting spectrum visually matches that of previously measured (uncalibra- ted) UF3 spectra. When a match is found for a particular value of k, the concentration of UF, in solution and thus the absorption coefficient can be 3 calculated knowing the total amount of UF4 before reduction. Comparison of € values by this method with the total reduction method showed agreement within a 5% uncertainty. In Table 2-1, absorption coefficients are listed for the various solu- tions and temperature ranges that have been studied. Values were taken from smoothed functions which within the limits of our precision are a linear function of temperature. RESULTS AND DISCUSSION An equilibrium expression such as the one written in Eq. 1-1 implies that certain criteria are valid: (1) The equilibrium expression should in- clude all reactants and products which are involved in the reaction and these components should combine in the stoichiometry indicated by the ex- pression. (2) The entire process must be reversible. Before quantitative data for the equilibrium in Eq. 1-1 were measured, the above criteria were examined in the following manner: The equation represents a heterogeneous equilibrium between a molten-fluoride solution of UF3 and UF4 and two solid phases, U02 and graphite. The identification of the UF3 and UF4 was made by the characteristic absorption spectrum of each component in the near-infrared and visible regions (4000-33000 cm_l). The identification of these two solute components is well established be- cause their absorption spectra have been thoroughly documented.11 In view of the extensive spectroscopic work which has preceded, there is no spec- tral evidence for any cations in the solution other than U+3, U+4. 14 The solid phase components, UC, and graphite, exhibit no measurable 2 solubility in molten fluorides. These phases were identified by their respective X-ray diffraction patterns. A serious anomally arises as a result of the U02 phase identification since its formation is contrary to the established phase diagram13 for the U-C system which shows UCZ to be metastable with respect to U203 and graphite at temperatures less than 1500°C. On the basis of the uranium-carbon phase diagram and the accepted free energies of formation for the uranium carbides at temperatures less than 1000°K, U,C, should be the carbide phase which was identified. Never- 273 theless, UC, has been repeatedly shown to form at these temperatures and 2 has been established as the stable carbide phase in the equilibrium of Egq. 1-1. The reader who is interested in the details of this identification i . 6 : is referred to an earlier paper. In the present paper we have included a series of equilibration experiments where excess U2C3 was used to reduce UF4 solutions via the back reaction of Eq. 1~1. Results are compared with similar experiments where U02 was used as a reductant. One of the simplest and yet most convincing observations to offer for the equilibrium is that the stoichiometry of the soluble uranium fluoride species follows the four-to~three relationship of Eq. 1-1. When a solution of approximately 0.1 mole % UF3 in LiF-—BeF2 is allowed to react with graph- ite it is observed that 4 moles of UF3 form 3 moles of UF4. For example, when a 0.068 molar solution of UF, was allowed to react via Eq. 1-1 to form 3 UFA’ it was observed that under conditions where reaction was more than 99% complete, a 0.049 molar UF4 solution resulted. If the process were merely one of UF3 oxidation, then 4UF3 should form 4UF4. For example: 4UF3(d) + 2MF < 4UF 2(d) * M (3-1) 4y T where M is a metal such as Ni. Finally the reversibility of the reaction was demonstrated by the re- versible temperature dependence of the equilibrium. From a particular tem— perature at which the system had attained equilibrium and concentrations of UF3 and UF4 measured, the temperature could be repeatedly raised or lowered causing the relative concentrations of UF, and UF4 to shift and attain equi- 3 librium concentrations at these new temperatures. When the system was 15 returned to the initial temperature, the original concentrations of UF3 and UF& were reproduced. Quantitative aspects of the temperature dependence will be given in the following sections. The criteria tests established the equilibrium process as written in Eq. 1-1. We found it more practical to measure the back reaction mechanism: 3UF, + UC, < 4UF, + 2C (3-2) since, by intentionally adding excess UC2, we could insure that the molten fluoride solution was always in contact with all the reactive solid phases. Furthermore, we could interrupt the equilibration and add fresh UC2 to demonstrate that the original carbide had not been consumed or altered dur- ing the course of the reaction. This procedure also insured that more ac-— tive reducing agents, including other uranium carbides, were not present. An equilibrium quotient for Eq. 3-2 can be written: 4 (UF 5) Q = (3-3) (UF,)> where UF3 and UF4 are expressed in mole fractions of the solution. Q is simply the reciprocal of the equilibrium quotient, Q', for the forward reaction of Eq. 1-1. The data in the following paragraphs will be presen-— ted as Q values in terms of the back reaction and should not be confused with forward action. The effect of variables such as temperature, melt composition, carbide composition and atmospheric contamination on the equilibrium of Eq. 3-2 in the solvent system LiFwBer have been measured and are treated separately in the following sections. Since the equilibrium of Eq. 3-2 (also Eq.1-1) is the central theme of this paper, it will often be cited as simply “the equilibrium." Effect of Temperature on the Equilibrium Previous results4 from the hydrogen reduction of UF4 in molten fluor- ide solutions indicated that the temperature effect on the equilibrium of Eq. 1-1 should be small. However, when we measured the equilibrium by either the forward or the back reaction, we found it to be very sensitive to temperature. This can be seen qualitatively by examining the molten 16 fluoride absorption spectra of Fig. 3-1 for equilibrium mixtures of dilute UF3 and UF4 in LiF-BeF2 (66~34 mole %), in LZB’ over excess UC2 at various temperatures. The spectra are due only to the UF3 and UF4 components of the solution. Therefore, by comparing these spectra with the spectra of pure UF4 and pure UF3 (Figs. 2-2 and 2-3 respectively), it can be seen that at 500°C, most of the uranium in solution is present as UF4 whereas at 700°C, enough UF, of Fig. 2-3. The composite spectrum at 600°C resembles neither of the two is present to make the composite spectrum resemble that pure component spectra but instead an intermediate mixture of the two. The quantitative aspects of these spectra were calculated by the pro- cedure described in the experimental section. From absorption spectra such as those in Fig. 3-1 concentrations of UF,_, and UF4 were determined in mole fractions and used in Eq. 3-3 to calculatg equilibrium quotients, Q, at various temperatures. The data are presented in Table 3-1 along with Q values which are then presented in Fig. 3-2 as 10g10Q versus l/TK (where TK is the Kelvin temperature). At the top of the figure is shown the centi- grade scale and at the right side of the figure, the equilibrium ratio, [UF31 R=_—__..__—____. {UF3]+[UF4] (3-4) where [UFB] and [UF4] are the concentrations in solution. (Note that the denominator of Eq. 3-4 represents the total uranium fluoride in solution.) These R values have been the customary manner in which UF -UF4 concentra- 3 tions are expressed within the MSRE program. The two lines drawn through the data points represent the experimental uncertainty of the data which arises mainly from the baseline error in the absorption spectra. Equilib- ria at various temperatures were approached from both the high (open circles) and low (closed circles) temperature direction. The system was initially held at ca. 50°C above the temperature desired until the UF, con- 3 centration had ceased to grow (UF4 reacting with UC2 via Eq. 3-2). Then the temperature was dropped 50° and the UF3 concentration was allowed to fall by reaction of UF3 with graphite until no further change could be detected. The equilibrium could be shifted repeatedly in this manner by varying the temperature of the system. The train of points at any given 17 ORNL-DWG 72-9633 1 l [ I | I ABSORBANCE 5000 1O,OOQ1 15,000 WAVENUMBER (cm ) 3-1 Spectra of Dilute UF3-UF, Mixtures in LiF-BeFp (66-34 mole %) Showing Temperature Effect on the Equilibrium: 4UF3 + 2C Z 3UF, + UCj. 18 Table 3-1 Typical Equilibrium data used in Figures 3-2 to 3-5 where Q and R are defined by Eqs. 3-3 and 3-4. Run Solvent Carbide Temp Measured Absorbance Mole Fraction Q R Phase (°C) 11360 9170 UF.(10%)UF, (10%) (x108) cm™ em 1 3 4 1 LZB uc, 500 0.130 0.515 0.264 7.25 0.128 } 0.035 2 L,B uc, 550 0.217 0.505 0. 804 7.22 11,12 10.10 3 L,B uc, 600 0.485 0.517 2,76 6.68 1951.0f 0.29 4 L,B uc, 650 0.692 0.327 4,57 6.14 19080.0{ 0. 43 5 LZB uc, 700 1.305 0,780 9.83 7.60 212300,0f 0.56 6 L,B U,Cq 500 0.254 0.538 0.99 7.17 26.0{0.12 7 ! " 600 0.409 0.416 2.36 5.29 2070.0; 0,31 8 " " 700 1.083 0.567 8.29 4.74 443000.00.64 9 LB uc, 370 0.147 0.384 0.53 5.52 4.6]0.087 10 " " 400 0.215 0.330 1.04 4.67 115.0(0.18 11 " " 450 0.380 0.328 2,23 4.36 2950.010. 34 12 " " 500 0.482 0.274 3.09 3.21 27600,0{0.49 13 MSBR UC2 550 0.925 1.317 3.97 18.31 401.010.18 14 " " 600 1.257 1.165 6.13 15.16 4050,.0[0.29 15 " " 650 2,575 1.433 14.0 14.32 129000,0{0.49 19 ORNL-DWG 72-10719R TEMPERATURE (°C) 2500 9550 573 600 622 650 700 < I I I | /I / / 0.6 ® ® -3 / / / ® — 0.5 @ / !/ — 0.4 -4 > L/ o — 0.3 ¥ -5 _g If QS $ u? o 2 g s/ / 0.2 -6 ‘/ %5 I @ 7 :t/ — 0.4 _ / 8 i/ -9 — 0.03 1.30 1.25 1.20 145 .40 1.05 1.00 1000/7 (k) 3-2 Equilibrium quotients, Q = (UF3)4/(UF4)3, versus temperature for UCy + 3UF4(d) 2z 4UF3(d) + 2C in the solvent LiF-BeFy (66-34 mole 7). 20 temperature represents the approach to equilibrium with only the lowermost (for open circles) and uppermost (for closed circles) being the best measur- ed equilibrium value. The large temperature effect on the equilibrium is exemplified by Fig. 3.2 where the quotient, Q, shifts by 106 in going from 500 to 700°C. 1In practical terms, this means that the concentration of UF3 relative to the total uranium fluoride in solution is increased from ca. 5% at 500°C to ca. 60% at 700°C. The same large temperature effect on the equilibrium is found when U203 (in place of UC2) is equilibrated with UF4 solutions. These data are presented in Table 3-1 and the resulting Q values are plotted in Fig. 3-3. Here the Q values are greater at a given temperature than in Fig. 3-2 and therefore support the identification of UC2 as the stable carbide phase of Eq. 1-1. Furthermore the U,C, equilibration experiments demon- 273 strate that the UF, stability in dilute fluoride solutions as well as the 3 temperature effect on the equilibrium would not be far different from that presented in Fig. 3-2, even if the identification of UC2 as the stable carbide phase of Eq. 1-1 were not correct. The data of Fig. 3-2 can be used to calculate the change in enthalpy for the equilibrium. By defining the standard state of the solutes UF3 and UF, as one mole percent in L, B, their activity coefficients are unity and 4 2 then the equilibrium quotients become equilibrium constants, K. The change in enthalpy, AH, for the reaction in the temperature range of 500-650°C can be calculated from the slope of the line in Fig. 3-2 using the expression: d (1nkK) d(1/T) where R is the gas constant. The value obtained for AH of Eq. 3-2 is 99.3 AH = -R (3-5) Kcal/mole which is surprisingly large in view of the enthalpy change calcu- lated from enthalpies of formation for the pure, undiluted components at either 298° or 800°K. These values are given in Table 3-2 and yield AH° -10 Kcal/mole for the undissolved components of Eq. 3-2 at 298°K and ARC = ~12.90 Kcal/mole at 800°K. The process of solvation is not included in the calculation since no heats of solution for UF3 and UF4 are available. It should be noted that even the sign of the AH is different: We measure an endothermic process whereas a slightly exothermic process is expected. 21 ORNL-DWG 72-10717 TEMPERATURE (°C) 1 500 550 573 600 622 650 700 | | l | ] l -2 / * 06 /s . / — 0.5 s 0 :v . — 0.4 -4 o ‘f O O © 8 5 L o 2 g g o —0.3 -5 s o 5 | i x W/ o / 1.30 1.25 1.20 1.45 110 1.05 1.00 1000 /7 (o) 3-3 Equilibrium quotients, Q = (UF3)4/(UF4)3, versus temperature for 1/2UC3 + 3UF4(q) < 4UF3(gq) + 3/2C in the solvent LiF-BeF; (66-34 mole %). -8 22 Table 3-2 Enthalpy Data in (Kcal/mole) Sources of the Data are from Tabulations referenced as Superscripts UC2 UF4 UF3 C MY 2013 ~450(5)° ~345(10)° 0 o .o 13 14 15 16 B Eo0g 8.79 14.99 11.8 1.83 Table 3-3 Equilibrium Quotients, Q, and Ratios, R, for UF3, UF4 Solutions in Atmospheric Contaminated System (Taken from Ref. 24) Solution 675°C 575°C (Mole %) LiF-BeF, R Q R Q 6634 025 2.7x107%0 o004 1.8x10713 7 10 48-52 .13 3x10° .03 6.2x10 23 Heats of solution could plausibly account for a large amount of the dis- crepancy since it is observed that the heat of solution for CeF3 in LZB (600-800°C) is 17 Kcal/mole whereas only 10 to 12 Kcal/mole is predicted.17 From the enthalpy data, without the heats of solution, we can only conclude that the thermodynamic data is not adequate to predict the change in en- thalpy for the reaction. Effect of Solvent on the Equilibrium In the same way that temperature shifted the equilibrium, changes in the solvent composition did also. The original purpose of this research was to demonstrate that changes in the fluoride ion concentration (which have already been shown to affect the c60rdination behavior of dilute UF4 solutionslS) might be related to shifts in redox equilibria as well. The effect of changing the solvent composition on the equilibrium is exempli- fied by comparing the equilibrium quotients, Q, for LiF—BeF2 (48-52 mole %), LB, in Fig. 3-4 with those previously shown in Fig. 3-2 for the LZB compo- sition. At any given temperature (e.g., 500°C), Q is considerably larger in the LB composition than in the corresponding LZB melt. Therefore the equilibrium of Eq. 3-2 is shifted to the right by increasing the concentra- tion of BeF2 in the solvent, i.e., by making the solvent more F deficient through the addition of a component which coordinates strongly with fluo- ride ions. The ratio, R, of Eq. 3-4 at 500°C has been increased from ca. 0.05 for the L 3-4 at 500°C). 2B solvent to ca. 0.55 for the LB solvent (c.f. Figs. 3-2 and The magnitude of this change can be compared with that which is pre- 3+ C ;o bt . 1 dicted from Baes' activity coefficients for M~ and M cations. ? Real- izing that the only difference in equilibrium quotients between the two melts is the ratio of the respective activity coefficients, v, for UF3 and UF, raised to the appropriate powers: 4 3-4 370 TEMPERATURE (°C) 400 450 ORNL-DWG 72-40718 500 I T / v 1.\l\\_ ;/ | / / / ] v/ % / / 1.55 1.50 1.45 1.40 1.35 1.30 1000/7 (ok) 0.6 0.5 0.1 Equilibrium quotients, Q = (UF3)4/(UF4)3, versus temperature for UCy + 3UF4(d) bl 4UF3(d) 4+ 2C in the solvent LiF-BeFj (48-52 mole 7%). 25 &) @) QLB K=4Q L,B L3 o3 (&UFQ;) (zY ;) where L2B and LB denote the solvent systems. Since Baes defines all activ- 3-6 ity coefficients as unity in the reference composition, LZB’ then: &) o5 \'UF%, QL2B <:?3§;:f 3 19 The right-hand term can be estimated from Baes' data at 600°C™" where YUF3::NbeF3 4::YThF = 10 by extrapolating to LiF—ger (48-52 mole %). By this procedure g/ g is estimated to be 4 x 107, 2 = 0,7 and YUF From our data, at 600°C can be determined by extrapolating QLB/QLQB the double lines to 600°C and comparing this value, Q read from Fig. 3~2. We find Q LB® with the value, QLzB’ LB/QL2B = 5 x 10" agrees reasonably with the estimate from Baes' data. Even better agreement could be obtained if a non-linear extrapolation (which is suggested by the trend in the data of Fig. 3-4) is made. Furthermore, it should be noted that Baes' activity coefficients are only approximate for UF, since they are actually based on 3 data for CeF,. The comparison serves to show that the magnitude of the solvent effegt is in reasonable agreement with previous data and consequen- tly must be considered when estimating UF3 stabilities in other molten fluoride solvent systems. This leads then to the practical question, '"What UF3 stability is ex- pected in the MSRE and the MSBR solvents?". In these ternary systems the relative measure of F concentration is more difficult to determine than in the binary system LiF-BeF, since there are two "acidic™* cations in each competing for fluoride ions. It is currently regarded that the MSRE sol- vent is more F deficient than LyB and results of a previous electrochemi- cal study of UFj3 stability by Manning20 support this contention. Little *By "acidic" we mean, in the Lewis acid concept, the tendency to coordinate with F . 26 attention has been given to the UF3 stability in MSBR solvents. We have first attempted to predict it and finally we have measured it directly. Realizing the solvent effects on UF, stability arise from changes in 3 the available free F , an attempt was made to estimate the F concentration . . 18 in MSRE and MSBR solvents based on the earlier observation™ that the co- ordination equilibrium of U4+ ions in LiF-BeF. solvents depended upon F 2 according to: 4"' 3 3- - UF8 Z UF7 + F 3-8 We have previously suggested21 that the F~ could be measured by determining the concentration of UF8 " and UF73— and then estimating the F concentra- tion by Eq. 3-8. This method was found to work for LiF—BeF2 solutions with Ber concentrations of up to 52 mole 7 and for the MSRE solvent which is essentially a LiF—BeF2 solvent. The method then was used to estimate the F concentration in the MSBR solvent. Because the spectrum of UF4 in this solvent was largely UF8 y & F concentration greater than that in LZB was suggested from Eq. 3-8. We concluded that the stability of UF, would be very much less than in L,B, 3 ] , 21 and in fact, some earlier UF, stability measurements” tended to support this conclusion. ’ In contrast to this viewpoint, were activity coefficient data by C. F. Baes19 and BF3 solubility data by S. Cantor22 which suggested that the UF3 should be slightly more stable in the MSBR solvent than in LzB. We examined this discrepancy by experimentally measuring the stability of UF3 in the MSBR solution over excess UC2 in the graphite spectrophoto- metric cell. The results are shown as logloQ Vs l/TK in Fig. 3-5 in the same form as that used for previous figures. These data show that UF3 is more stable in MSBR than in LZB' From the standpoint of reactor operatioms, concentration ratios, R, of UF3 (c.f. Eq. 3-4) of up to 0.03 can be main- tained safely down to the ca. 500°C freezing point of the solution. The discrepancy in our earlier21 predictions can only be rationalized by allowing a more complex coordination mechanism for the MSBR solvent than + is described in Eq. 3-8. This probably involves U4 which are fluoride bridged to neighboring Th4+ or Be2+ so that, through bridging, the log,g @ 3-5 for UCy + 3UF, (72-16-12 mole % 27 ORNL-DWG 72-12321 TEMPERATURE (°C) 500 550 600 650 700 4 l T l / -2 7 - 06 ® 4 05 -3 / 33' /s @ / - 04 < _q / 5 o/ 103 o s ~oL : 3 / > O " -5 L 8 Jo2 3 x S/ - 01 7 7 / -8 130 125 420 1145 110 105 100 ‘1000/T (°K) Equilibrium quotients, Q = (UF3)4/(UF4)3 versus temperature (d 3 < 4UF3(q) + 2C in the solvent LiF-BeFp-ThFy 28 coordination number (and accordingly by Eq. 3-8, the F concentration) ap- pears much larger. There is some evidence for this in LiF—BeF2 solvents o 2 where the Ber concentration is greater than 52 mole Z%. 3 Effect of Atmospheric Contamination on the Equilibrium In earlier attempts to measure the equilibrium quotients for Eq. 1-1, it was apparent that the equilibrium concentration of UF3 in LiF-BeF2 solvents was unusually low24 compared with the present results. These results are presented in Table 3-3 and were measured by following the re- action of UF3 in graphite with no uranium carbides added directly to the system. These reactions were always accompanied by the formation of UO2 and other unidentified solid phases. However after various improvements were made which eliminated obvious signs of atmospheric contamination, such as U0, formation, the stability of UF 2 3 sequently concluded that these earlier measurements involved equilibria of was greatly enhanced. We have sub- UF3 and UF4 solutions in graphite and an oxy-carbide phase (as opposed to a pure carbide phase). It was never possible to identify the oxy-carbide phase by X-ray analysis despite the fact that the equilibria were very easy to reproduce from Eq. 3-2. The effect of atmospheric contamination is clear -- it greatly reduces the stability of UF, and is therefore a major factor which cannot be ignor- 3 ed when considering UF3 stability in molten fluoride solutions. Effect of Temperature, Solvent and Contamination Compared All of the effects of the variables have been collected to compare their relative importance and are shown in Fig. 3-6 as loglOQ Vs T%l in the same fashion as the previous figures but with a substantial reduction in scale. The effect of increasing temperature is similar in all cases, causing an increase in the stability of UF There is no reason for the lines to be parallel to each other becauseBthey differ principally (except for the case of the atmospheric contamination) in the activity coefficients for UF3 and UF4 in the different solutions and these need not change pro- portionately for all solutions. Neither should it be necessary that the data be represented by straight lines, implying that AH for the reaction is constant. They are used here only because the data are insufficient to 29 ORNL~DWG 72-12322 TEMPERATURE (°C) o 370 400 450 500 550 600 650 700 I N I I | I | I | _1 . i i _2 | _ 0.6 -3 | / 105 ~ LL¢ o + » B 04 *, 2 ~ 0.3 5l Ty i 02 & ) o g © 0.1 ..8 L — _9 . —10 —— (A) LiF-BeF, (48-52 mole %) T (B) LiF-BeFp -Zr F4 (65.6 -29.4-5.0 mole %) _yy L (C) LiF-BeFy-ThF, (72-16-12 mole %) (D) LiF-BeF, (66- 34 mole %) (E) LiF-BeF, (48 -52 mole %) o OXIDE CONTAMINATED (F) LiF-BeF, {66 -34 mole %) OXIDE CONTAMINATED i3 | | I | 1.6 1.5 1.4 1.3 1.2 1.1 1.0 1000/7 (o) 3-6 A comparison of equilibrium quotients versus temperature for UCy + 3UF4(q) z 4UF3(d) + 2C in various solvent systems. 30 justify greater detail. Decreasing the F concentration by the addition of BeF2 is very bene- ficial in increasing UF3 stability whereas atmospheric contamination causes the opposite and most disasterous effects on UF, stability. 3 Since the MSRE results do not come from our work, the MSRE line is broken. The stability of UF, in the MSBR solvent is between that of the MSRE solvent and L2B. It iSBObViOUS that the region of greatest UF3 sta- bility is that of high temperature and low F concentration. We therefore suggest that little UF3 could be maintained in F rich solvents such as LiF-NaF-KF (46,.5-11.5-42.0 mole %) even if the reported K" reduction by UF (25) 3 were not to occur. Conversely, the greater stability of pure UF§4) (i.e., not dissolved in a molten fluoride solvent) is explained by the absence of solvating F . Other Considerations If the thermodynamic data are sufficiently accurate then it should be possible to calculate the free energy change for Eq. 3-2 in the solvent LiF-BeF, (66-34 mole %) and then the equilibrium constant by: AG = -RT 1n K (3-9) The expressions for the free energy of formation are given in Table 3-4 for UCZ’ UF4 and UF3 (where the latter two are for the standard state of Table 3-4 Free Energies of Formation for Pure UC, and the Solutes 2 3 and UF, in LiF-BeF, (66-34 mole %) with standard deviations, 0, in Kcal/mole for relationship: AG = A + B (TK/lDDO) UF Component A B g Source uc, ~15.82 8.2 - Storms™> UF, ~338.04 40.26 2 Baes?? 26 UF4 -445,92 57.85 2 Baes 31 one mole percent each in LZB)‘ These free energy functions, when combined in the proper stoichiometric proportions yield a change of free energy for Eq. 3-2 of AG = 1.42-4.31 (TK/lOOO) with a significantly large combined standard deviation of + 16 Kcal/mole. At 500°C, AG is -1.91 Kcal/mole and the equilibrium constant from Eq. 3-9 is 3.5. Since these are standard states for the UF3 and UF4 solutes, then Q is also equal to 3.5 and the ratio, R, of Eq. 3-4 is 0.89. (c.f. with Q = 1.5 x 10—9 and R = (.03 in Fig. 3-2). Therefore, from the existing free energies of formation for the components of the reaction, practically all of a dilute UF3 solution in contact with graphite should be stable. However, neither our results, nor those from any other investigators support this high a stability of UF3. A word of caution should be given at this point. It may seem obvious to demonstrate UF3 stability via Eq. 1-1 by holding UF3 solutions in graph- ite and allowing the UF3 to react with graphite. Furthermore, it may be most convenient to generate a UF3 solution by reducing a dilute UF4 solution with a strong reductant such as Be, Zr, or U metal within the same graphite vessel that will be used for the stability measurement. We have observed that this results in the formation of mixtures of UC, UZCB and UC2 phases accompanied by the consumption of more reducing metal than is expected for the complete UF4 reduction. The apparent anomally is caused by the revers- ibility of Eq. 1-1 since as soon as UF3 is formed in excess of its equilib- rium concentration within the graphite vessel, it reacts with graphite form- ing uranium carbide phases and UF4 in solution. The UF4 ig, in turn, reduced again by the excess reductant, forming more UF3. An example of the process using Zr metal is: 4UF4 + Zr T 4UF3 + 27:1?4 (3-10) 4UF3 + xC 2 3U:a'4 + UC, (3-11) so that the net reaction is: UF4 + Zr + xC 2 ZrF4 + UCX (3-12) This is one of the major reasons why we found it more practical to study the equilibrium by the back-reaction mechanism of Eq. 3-2. Although the UC and U203 phases do finally react leaving ultimately UCZ’ we found that 32 even for our small reaction system of less than 0.5 cc, it took an im- practical length of time. Larger systems with smaller surface-to-volume ratios would take even longer. The question of reaction times brings up the final point to be mention- ed, that is, the kinetics involved in achieving the equilibrium of Eq. 1-1. Since UF3 is reacting with graphite to form uranium carbides, the mechanism is obviously heterogeneous. It is consldered by these authors far too dif- ficult a mechanism to attempt to clearly describe; but 1f reaction rates are sought, the initial measurements should demonstrate that the mechanism is heterogeneous by varying the surface-to-volume ratios of the reacting system. We predict that the outcome of such a measurement will substan- tiate the heterogeneous mechanism. Another point of caution should be made. Since larger surface-to-volume ratios mean slower reaction rates, apparent high stabilities of UF., may appear whereas they actually involve metastable 3 states of the equilibrium mechanism which include uranium carbide phases other than UCZ' These other carbides will ultimately be converted to UC2 by the mechanism of Eq. 1-1; but, until the conversion is completed, the UF3 ratio, R, will remain fixed at a high value. The ultimate aim of the UF3 stability study has been to describe con- ditions under which certain UF3 ratios can be maintained in graphite. To demonstrate the validity of our measurements we mixed dilute UF3 and UF4 in the LB solvent so that the resulting solution had a UF3 ratio, R = 0.17. The solution was maintained for a period of a week in the graphite spectro- photometric cell at 475°C with no loss of UF. or UF4 from solution. (c.f. 3 Fig. 3-4 which shows the maximum R at 475°C to be 0.40-0.45). 33 References 1) W. R. Grimes, '"Chemical Research and Development for Molten Salt Breeder Reactors', ORNL-TM-1853, 1967. 2) M. W. Rosenthal, P. N. Haubenreich, H. E. McCoy, L. E. McNeese, At. Energy Rev., 9 [3], 601 (1971). 3) W. R. Grimes, Nuc. Application and Technology, 8, 137 (1970). 4) G. Long and F. F. Blankenship, "The Stability of Uranium Trifluoride", Part I and II, ORNL-TM-2065, 1969. 5) M. H. Rand and 0. Kubaschewski, The Thermochemical Properties of Uranium Compounds, Interscience Publishers, New York, 1963. 6) L. M. Toth and L. 0. Gilpatrick, "Equilibria of Uranium Carbides in Molten Fluoride Solutions of UF3 and UF4 J. of Inorg. & Nuclear Chem., In press. Contained in Graphite at 850°K", 7) L. M. Toth, J. P. Young and G. P. Smith, Anal. Chem. 41, 463 (1969). 8) Preparation of BeF; performed by B. F. Hitch of ORNL. 9) James H. Shaffer, "Preparation and Handling of Salt Mixtures for the Molten Salt Reactor Experiment', ORNL-4616 (Jan. 1971). 10) Poco Graphite Inc. a subsidiary of Union 0il Co. of California, Decatur,Texas 76234, 11) J. P. Young, Inorg. Chem. 6, 1486 (1967). 12) S. Cantor, Reactor Chem. Div. Annu. Progr. Rept. for Period Ending December 31, 1965, ORNL-3913, p. 27. 13) E. K. Storms, Refractory Materials, Vol. 2, "The Refractory Carbides", Academic Press, New York, 1967. 14) A. S. Dworkin, J. Inorg. Nucl. Chem. 34, 135 (1972). 34 15) C. E. Wicks and F. E. Block,*Thermodynamic Properties of 65 Elements, Their Oxides, Halides, Carbides, and Nitrides% Bureau of Mines Bulletin 605 (1963). 16) JANAF (Joint Army-Navy-Air Force) Interim Thermochemical Tables, Thermal Research Laboratory, Dow Chemical Co., Midland, Mich. 17) C. J. Barton, M. A. Bredig, L. O. Gilpatrick, J. A. Fredricksen, Inorg. Chem. 9, 307 (1970). 18) L. M. Toth, J. Phys. Chem., 75, 631 (1971). 19) C. F. Baes, Jr., MSR Semiannu. Progr. Rept. for Period Ending Feb. 28, 1970, ORNL-4548, p. 153. | 20) D. L. Manning, private communication, 1970. 21) L. M. Toth, L. O. Gilpatrick, MSR Semiannu. Progr. Rept. for Period Ending Aug. 31, 1971, ORNL-4728, p 77. 22) S. Cantor, MSR Semiannu. Progr. Rept. for Period Ending 23) 1. M. Toth, unpublished results. 24) L. M., Toth, L. O. Gilpatrick, MSR Semiannual Progr. Rept. for Perjiod Ending Aug. 31, 1971, ORNL-4728, p. 77. 23) F. F. Blankenship, B. H. Clampitt, W. R. Grimes, S. Langer, "The Stability of UF, in Molten Fluorides," unpublished data. 3 26) C. F. Baes, Jr., "The Chemistry and Thermodynamics of Molten Salt Reactor Fuels,' presented at AIME Nuclear Fuel Reprocessing Symposium at Ames Laboratory, Ames, Iowa, August 25, 1969, Published 1969, Nuclear Metallurgy Symp., Vol. 15 by the USAEC Div. of Tech. Information Extension. . L. Anderson F. Baes Bamberger Barton Bettis Blanco Blankenship . Bohlmann . Boyd Braunstein A. Bredig B. Briggs R. Bronstein D. Brunton Cantor Compere Culler Dale DeVan Doss Dworkin Eatherly Engel Ferguson . Ferris Frye Gabbard Gallaher Gilpatrick Grimes Guymon Handley N. Haubenreich R. Kasten HO=HEwm iy oo o= ? O OO ;:>!fi LM unmoOoO@Dn"mEanoEH"REO00Y z:;:nflo F Mo 2mYd"mwmes> o= e Technical 35 ORNL-TM-4056 INTERNAL DISTRIBUTION (99 copies) S. S. Kirslis J. W. Koger A. I. Krakoviak R. B. Lindauer M. I. Lundin H. G. MacPherson D. L. Manning H. E. McCoy L. E. McNeese A. S, Meyer J. P. Nichols E. L. Nicholson A. M. Perry G. D. Robbins M. W. Rosenthal (2) R. G. Ross C. E. Schilling Dunlap Scott J. H. Shaffer M. J. Skinner G. P. Smith R. A. Strehlow E. H. Taylor R. E. Thoma L. M. Toth (20) D. B. Trauger C. F. Weaver A. M. Weinberg M. E. Whatley J. C. White J. P. Young Central Research Library ORNL - Y-12 Technical Library (3) Laboratory Records Department (10) Laboratory Records, ORNL RC ORNL Patent Office EXTERNAL DISTRIBUTION Information Center, AEC-0R (2) Research & Technical Support Division, Oak Ridge . B. Deering, AEC c D. F. Cope, AEC Shaw, AEC —H R . W. McIntosh, AEC Al R. DeGrazia, AEC David Elias, AEC J. E. Fox, AEC Norton Haberman, AEC (11 copies)