. . ;;‘; o X aw e AR R S OAK RIDGE NAT'ONAL LABORATORY operated by UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION r""- "\‘ /( Z ORNL- TM- & Ui ] CoPYNO. - /7 DATE — December 6, 1961 LABORATORY-SCALE DEMONSTRATION OF THE FUSED SALT VOLATILITY PROCESS G. |, Cathers, R. L. Jolley, and E. C. Moncrief ABSTRACT The feasibility of processing enriched irradiated zirconium-uranium alloy fuel by the fused salt-fluoride volatility procedure has been demonstrated in laboratory tests with fuel having a burnup of over 10%. Uranium recoveries were good and decontamination factors for radioactive fission products were 106 to 10”. The UF¢ product contained significant quantities of nonradicactive impurities, and additional work in this area is needed. For review by Nuclear Science and Engineering. NOTICE This document contains information of a preliminary nature and was prepared primarily for internal use at the Oak Ridge National Loboratory. It is subject to revision or correction and therefore does not represent a final report, The information is not to be abstracted, reprinted or otherwise given public dis- semination without the approval of the ORNL potent branch, Legal and infor- mation Control Department. INTRODUCTION The fused salt~fluoride volatility process for zirconium=-uranium reactor fuel consists of (1) hydrofluorination and dissolution of the fuel in molten salt, (2) fluori- nation to volatilize UFy from the melt, and (3) complete decontamination of the UF, in an absorption-desorption cycle (1,2). As a nonaqueous process it has not received the large-scale development effort that has been expended on aqueous processing methods. Its advantages include low waste volumes (<1 liter/kg Zr-U alloy), high decontamination from fission product activities, a greatly decreased criticality prob- lem with enriched fuel due to the absence of neutron moderators, and the form of the product, UFg, which eliminates some of the chemical conversion steps needed in the uranyl nitrate~uranium metal cycle. Some possible disadvantages are the corrosion at high temperatures in a fluoride system, the necessity of a gas-tight system, and the difficulty of manipulating molten salt. The tests, carried out in a hot cell, included study of the hydrofluorination and fluorination reactions, the behavior of various fission product activites, and the degree of uranium recovery and decontamination. The tests were conducted primarily in preparation for operation of the ORNL Volatility Pilot Plant, which has been adapted to process Zr-U reactor fuel (3). If this operation is successful, the pilot plant may be modified to test variations of the process with other types of fuel. PROCESS DESCRIPTION AND EXPERIMENTAL PROCEDURE The major process steps of fuel dissolution, UF volatilization, absorption, and desorption were used (see Fig. 1). In each of 12 tests, 650 g of 2- to 3-year- Refrigerant (-60°C) T + Of:E;as HF Trap } £ v 2 - 6 aste HF <5 3 NOF L. Oo D T = . Metal Fuel = cte. and He Ist Naf Bed He NaF-LiF Salt 100°C Adsorption ‘ 400°C Desorption HF IT F2 — LU Salt Transfer . P O 2 T U - o g§| .g °8 LT 3 = oste Soit Disposal Can -'—-_- -— __! T Hot Cell UNCLASSIFIED ORNL-LR-DWG. 49010 2nd NaF Bed 400°C UF6 Cold Trap Cold Laboratory Fig. 1 Schematic of laboratory process test equipment. decayed zirconium alloy fuel, with gross B and y activity levels of 1.8 x 107 and 1.0 x 107 cpm/mg U, respectively, was dissolved at 500-700°C by hydrofluorination in molten 57-43 mole % LiF-NaF with a liquidus temperature of 670°C. As hydro- fluorination and dissolution proceeded, the composition of the salt was changed as represented by the line shown in the phase equilibrium diagram (Fig. 2). Dissolu- tion was completed at a salt composition of 31-24-45 mole % LiF-NaF-ZrFy, i.e., close to a eutectic composition melting at 449°C. The resuiting final UF4 concen- tration in the melt was <1%. The system LiF-NaF-ZrF,4 is one of the few fluoride systems known in which liquidus temperatures are so low for large concentrations of ZrF4 (4). The composition of the initial dissolution salt was chosen so as to mini- mize the liquidus temperature encountered in the 0-20 mole % ZrF 4 region of the phase diagram. The dissolution product salt containing UF4 was fluorinated at 500°C with elemental fluorine, and the volatilized UF was absorbed on sodium fluoride at 100°C. The UF, vapor pressure over the UF 4-3NaF complex at this temperature is ~2 x 1073 mm, and essentially all the UF . is absorbed out of the Fo-UF 4 gas stream (5). Desorption consisted in heating the UF s~NaF complex bed from 100 to 400°C while passing F, through to a second NaF bed held at 400°C. The dis- sociation pressure of the UF ,-3NaF complex exceeds 760 mm af 400°C. The final UFé product was cold-trapped at -60°C, then hydrolyzed witha 1 M AI(NO3)3 solution for analysis. UNCLASSIFIED ORNL -LR-DWG 38115R ZrF, 912 TEMPERATURE IN °C COMPQSITION IN moile % = INDICATES SOLID SOLUTION P-537 E£-512., 7 NoF - 6 ZrFy - £-500, P-544 4, P40, /T ZNQF'ZFFq’ e o SNaF-2Z1F,— M 3NaF - Zrfy— #‘ ‘ £-747 “ \:‘ R '3 \625 8 - 9. %0 Co_ s, 603 Qo S 950 o NGF ! | N\ %% £-652 oas Fig. 2. LiF-NaF-ZrF4 phase diagram with process composition line. _g_ Equipment for the laboratory tests was installed in a hot cell equipped with Argonne Model 8 slave manipulators (see Fig. 1). It consisted of a dissolution reactor, fluorination vessel, NaF absorption beds, cold traps, and the necessary pneumatically operated valves for coupling the system together. The Hastelloy N dissolver was 18 in. deep and 3 in. i.d., with a 250-mil-thick wall, and had a loading chute. The L~nicke! fluorinator was also 18 in. deep, 3 in. i.d., with a 250-mil wall. Both vesseis were heated by a 5-in.-dia 12-in.-long tube furnace, supported vertically. The salt transfer lines of 3/8 in.-dia Inconel tubing (30 mils wall thickness) were heated auto-resistively with high-amperage current. The salt transfer line between the two salt reactors was also used as a common gas inlet line for the two vessels. In runs 1 through 7, U-tube nickel absorption reactors containing 200 g of NaF, 12-20 mesh, were used. Since less than 10 g of uranium was involved, smaller nickel vessels containing 50 g of NaF on a grade H sintered nickel filter were used in runs 8 through 12. In these last runs the UF6 was desorbed in a “"cold" laboratory, the absorption bed having a maximum activity of about 200 mr/hr at contact, with a large part of this being due to external surface contamination, The stainless steel cold traps for waste HF and product were cooled by trichlo- roethylene and dry ice. The waste HF was jetted into ice water, warmed to ambient temperature, sampled, and poured into a waste drain, The molten salt was sampled by a dip rod-frozen salt technique before being disposed of in heavy iron cans, sealed over with a high-melting wax while still warm, RESULTS Dissolution The total dissolution time varied from 16 to 62 hr, the HF efficiency from 19 to 48%, and the average dissolution rate from 0.17 to 0.64 mg em=2min=) (Table 1). These rates, although lower than in early laboratory work, are comparable to those obtained in engineering studies. The variables in the dissolution tests were flow rate, temperature, gas phase reaction, salt purification, zirconium hydriding, and HF con- centration. Dissolution was most rapid in run 12, in which the salt had received some prepurification, the zirconium was prehydrided, and a high HF flow rate was used. Entrained or volatilized material carried over in the HF stream was usually <0.1% (Table 11), and the maximum uranium loss, probably as UF 4, was 0.03%. Entrain- ment of Na, Cs, Sr, and rare earth fluorides was similar in magnitude to that of uranium. Volatilization combined with subsequent entrainment is indicated for fluorides of Zr, Nb, and Ru. Some of the variables present in the test dissolutions have not been evaluated adequately in cold laboratory work, but they were used in an effort to reduce the dissolution time. The probable effect of some of the variations are noted in Table ]II. The nature of the tests precluded, however, the drawing of firm conclusions about optimum conditions for dissolution. The effect of impurities in the salt on dissolution has not been definitely established although the piating-out of nickel on the zirconium Table I. Typical Dissolution Results Average HF Average HF HF Utilization Dissolution Dissolution Run Flow Rate, Concentration, Efficiency, Time, Rate, No. ml/min % % hr mg cm™2-min~! ] 820 100 20.1 62 0.17 3d 1130 100 30.3 34 0.30 10° 1200 70 27.3 29 0.35 12° 1400 100 47.7 16 0.64 an. , e Direct gas phase reaction occurred in first 6 hr. bSdI’r prehydrogenated; fuel hydrided to ZrHo 33" “Salt prehydrogenated; fuel hydrided to ZrH; . Table I, Entrainment in Waste HF of Dissolution Step Amount, % of total in feed Run Gross Gross No. U Zr Na B Yy Ruy Zry Nby Csy Srp TREB 1 0.007 - - 0.02 0.3 0,005 0.02 - 0.1 - - 3 0,007 - - 002 0.2 0.04 0.1 7 0.07r - - 10 0.01 0.3 0.04 0.03 0.2 0.3 0.04 0.9 0.05 0.02 0.004 12 0.007 0.1 0.05 0.08 0.2 2.0 0.4 1.2 0.04 0.02 0.006 ~-10-~ Tabie I Variation and Effect on Dissolution Rate Remarks Increasing HF flow rate increases rate logarithmically Increasing temperature increases rate Nonsubmergence of fuel in salt promotes direct HF-alloy fuel reaction HF dilution with Hp decreases rate Prehydriding possibly increases rate Salt purification by H22reducfion of Fe2t, Ni2t and Cr+ possibly removes inhibition of reaction Probably due to more chance of direct gas-solid reaction Generally true, but effect does not involve much over a factor of 2 in available temperature range Kinetic studies have shown that initial rates of 100% HF with unconditioned massive metal at 700°C results in run-away reaction Ho arises from reaction with metal or metal hydride, or is added as diluent Indefinite at low H/Zr ratio, since much surface cracking and swelling generally occur only above a H/Zr ratio of 1.5/1 Effect not proved although plating-out of Ni, Fe, and other elements on Zr surfaces has been noted =11- metal surface has been observed, and this presumably hinders dissolution of the zir- conium. Other elements such as iron, molybdenum, tin, and chromium are also elec- tronegative relative to zirconium and would presumably act similarly. All of these elements are either fission products, arise from corrosion of the Inor-8 (71% Ni, 16% Mo, 5%Fe, and 7% Cr), are present in the fuel alloy, or are impurities in the initial salt. Two reactions involved in the nickel cycle are 2 Nify + Zr——> ZrF4+2Nij, AF°® = -130 kcal Ni + 2HF ——> Hy + NiFg, 4F°® = +6 keal Reduction of NiF5 to nickel metal at the zirconium fuel surface probably proceeds mainly by the first reaction although the second reaction in reversal leads also to reduction if the Hp/HF mole ratio is sufficiently high. At the end of zirconium fuel dissolution the production of hydrogen becomes small and the second reaction leads to total redissolution of the amorphous nickel metal formed in the earlier stage. Uranium Recovery Greater than 99% UF 4 volatilization was obtained only in the last four tests because of inexperience with the equipment and the use of low fluorine flow rates (Table 1V). Volatilization was repeatedly 99.8% or more in earlier work with simu- lated process tests. The NaF absorption-desorption cycle also operated effectively, resulting in little uranium loss and duplicating the behavior observed in earlier laboratory and pilot plant work. The total uranium retention on NaF beds wos less than 0.1%, and -12- Table V. Typical Uranium Volatilization in Fluorination Step Fluorination Fo U Run Temp, Time, Flow Rate, Volatilized,® No. °C hr ml/min % of total inventory 8 500 4 300 97.9 9 520 4 300 99.4 10 500 3 340 99.2 11 520 3 340 99.9 12 500 3 430 99.8 a . . .ys . . Based on analysis of fluorinated salt. Initial salt contained ~0.5% uranium. - 13- this probably represents an upper limit with reuse of the beds and the tendency for refluorination of any retained residue. PRODUCT DECONTAMINATION Overall Overall gross B and y decontamination factors ranged from 106 to 107 (Table V). In all but a few cases the amount of activity in the product UF, was less than 10- fold the "natural™ activity of U-235 (Table V1). In run 11 the ruthenium y activity was high because an accidental pressure buildup and release entrained dust from the NaF into the UFg cold trap. In runs 1 and 6 the product UF 4 was trapped on NaF instead of in a cold trap, with the result that subsequent hydrolysis prior to analysis gave an excessively dilute solution and a high and uncertain background correction led to more apparent activity than in later runs. Runs 2 through 5 are not included since they were not complete flowsheet tests. In runs 7 through 12 the services of a special low-activity-level analytical laboratory were used, which, in conjunction with an improved product hydrolysis method, gave o more accurate picture of the activity in the product. The principal chemical impurities in the UF, product were molybdenum, techne- tium, neptunium, and chromium, in order of decreasing concentration (Table VII). Molybdenum, the end product of several fission product decay chains and a hydro- fluorinator vessel corrosion product, varied in concentration from 1200 to 10,100 ppm. Volatile MoF 4 formed in the flucrination step complexes with NaF, similarly to the ~14- Table V. Overall Decontamination Factors in Complete Process Tests Run Decontamination Factors No. Gross B Grossy Ruy Zry Nby GCsy Sr B TRE B 8 2¢00% >1x10° 2108 8x105 4x107 3x10° 3x108 5x10° o 3x10° 2108 4x107 2x10° >2x108 1x107 33x107 >1x10'° 10 9x10® 5x107 2x106 3x10° 8x107 1x10° >1x10'0 37107 4 5 11 1x10° 210° 5x10% 1x10° 36x107 2x1010 >9¢10% 3x107 12 2%10° 6x107 ix107 ox10% 8x10® 2x10° »5x10° >1x10'! _]5.. Table VI. Ratio of Product Uranium Activity to Activity of Unirradiated U-235 Basis: 232 B cpm/mg U, 11 y cpm/mg U for 90% U-235 in equilibrium with daughters Run Activity Ratio, product/unjrradiated U-235 No. Ruy Zry Nb y Csy Sr B TRE P 1 <10 <10 <10 <10 1 <] 6 <10 <10 <10 <10 <1 <1 7 <] <1 <] <1 <1 <1 8 <<1 <1 <<] <<] <<1 <1 9 <1 <1 <<1 <<} <1 <<1 10 <10 <] << <] <] <<} 11 <100 <10 <<] <<] <] <1 12 <1 <10 <1 <<1 <] <<] -16- Table VII. Impurities in UF, Products Run Amount, ppm of U No. Mo Np Tc Crd 8 10,100 260 1,020 200 9 5,200 240 490 290 10 2,200 58 260 150 1 2,500 310 240 80 12 1,200 290 60 <100 oProbcbly from corrosion of the cold trap during the hydrolysis of the product to obtain a representative aqueous sample. -17- behavior of UF,. Investigation of the dissociation pressure of MoF,-NaF complex showed that it is approximately 1 atmosphere at 225°C, compared to 360°C for the UF 4-NaF complex (_(3) Some separation of MoF , and UF is thus achieved in the absorption step, dependent on the conditions of temperature, time, and gas flow. The low activity levels of the UF4 products in runs 8 through 12 and the fact that the usual individual fission product B activity contributors did not total up to the gross B activity indicated an unknown B contributor. This was found to be technetium by both chemical and radiochemical analyses. it was calculated that the feed contained >3900 ppm based on the uranium, assuming >10% burnup. The volatile compound TcFg is presumably formed in fluorination and possibly behaves in the same way as MoF ; and UF ¢ in absorption. Neptunium hexafluoride also appeared to follow UF¢ through the absorption step. Chromium may have been introduced by hydrolysis of the UF, product in a stainless steel vessel. Decontamination factors for the separate process steps are given for run 12 in Table VIII. Dissolution Step There was considerable but highly erratic disappearance of ruthenium and niobium activity during dissolution. The ruthenium y d.f.'s were 5 to 120, and the removal of Ru is believed due primarily to adsorption on the dissolver wall. Some ruthenium was volatilized with the excess HF. The niobium d.f.'s were 3 to 620. - 8- Table VIll. Step Decontamination Factors in Run 12 Decontamination Factors Step GrB Gry Ruy Zry Nby GCsy Sr P TREB Dissolution 2 ] 40 2 30 ] ] ] Fluorination 4x105 3x10° 4x10° 5x10% 4x10° 4x10° 1x107 2x107 3 Absorption- 2x103 2x10% 1x102 1 6 3x10° >5x102 >5¢10° Desorption 7 Overall 2107 6x107 1x107 9107 8x10° 2x10° >5x10° >1x10'] ~-19- The removal of niobium is believed primarily due to volatilization rather than ad- sorption on metal surfaces. Material balances for both ruthenium and niobium were low, possibly as the result of condensation of some NbFg (f.p. 225°C) in the top zone of the dissolver, and of adsorption of the ruthenium. The nondisappearance of cesium, strontium, and rare earth activities was expected since these elements form fluorides that are nonvolatile and difficult to reduce. Fluorination Step Decontamination from the most important volatile fission product activities, Ru y and Nb y (probably in the form of RuF 5 or RuF4 and NbFz) in the fluorination process was much higher at a fluorination temperature of ~500°C than in previous work at 600-650°C (1). In the first six tests conducted in the hot cell work, the absorption off-gas was trapped in caustic. From analyses of these solutions and of the first NaF beds, the gross B, gross y, Ru y, and Nb y activity decontamination factors were calculated for the fluorination step (Table IX). At 600-650°C the gross B and y d.f.'s were usually ~10° and the Ru y and Nb y d.f.'s ~5-10. Uranium is thus largely decontaminated from fission product activities in the fluorination step if the temperature is kept as low as possible. Absorption-Desorption Decontamination factors in the absorption-desorption step were in the range 10-100 for the more volatile NbF5 and RuFg or RuF .. The amount of Ru y activity in the off-gas stream passing through the first NaF absorption bed was highly variable, -20- Table 1X. Measured Decontamination Factors Obtained in Fluorination I:\;J:. GrB Gry Ruy Zr y Nb y Cs B Sr B TRE B 1 3x10° 1x10° 4x102 7x10% 5108 2¢10° 4x10® 1x107 2 2:0° sxi0* 210 1x10° w0 - ; 6x10° 3 1x10° 3x10% 6x10% sx10* 1x10° - 8x10° - 4 1x10° 4x10% 1x10° 20% 2x10° - x10° - 5 8x10% 3x10* 2«10 3x10f 03 - - - 6 3x10% 2a0* 2107 1x0% 2103 4x10* 3x10° 5x10° -21- due principally to the small amount volatilized in the 500°C fluorination step: 82, 11, 8, 13, 11, and &% of the total volatilized Ru y activity in runs 1 through 6, re~ spectively. In previously reported work relatively more Ru y activity passed through the first NaF bed and resulted in much higher d.f.'s, since a larger amount of Ru y activity was volatilized in the fluorination step at 600-650°C. Actual absorption-desorption decontamination factors could be calculated only in runs 1 through é where the absorption off-gas activity was trapped and measured. The absorption-desorption d.f.'s presented in Table VIII for run 12 are minimum values since the absorption off-gas was not measured. Actually, inclusion of the off-gas activity in the calculation would affect only the Ru y and Nb y d.f.'s and those only slightly. The d.f.'s given for the separate process steps in run 12 (Table VIil) were calculated from prorated activities measured in the first absorp- tion bed since the same material was used in runs 8 through 12. CONCLUSIONS The fused salt-fluoride volatility process gave satisfactory decontamination of uranium from fission product activities at a burnup of >10%. The main impurities in the product UF4 were molybdenum, technetium, and neptunium. Some separation from molybdenum and technetium evidently occurs in fhhe absorption~desorption step, and further work to optimize this effect appears warranted. The results also indicate that use of a fluorination temperature of 500°C rather than 600-650°C minimizes to a large extent the carryover of Ru y and Nb y activities along with the UF¢ gas stream in the fused salt fluorination step. Evaluation of the absorption-desorption -22- step was only tentative due to the small carryover of all activities from the fluori- nation step. The overall results appear to confirm the chemical feasibility of the process with irradiated Zircaloy-2-U alloy fuel. Areas of chemical uncertainty exist, par- ticularly in the dissolution step, but these are apparently not serious. The dissolu- tion step involves gas-liquid-solid contact and is necessarily quite dependent on geometry. |t appears desirable, however, to study further the zirconium metal re- duction of NiF; and analogous impurities in the fused salt to determine how this affects the dissolution rate and the concurrent volatilization or deposition of rela- tively noble materials such as ruthenium and molybdenum. ACKNOWLEDGMENTS The assistance of T. E. Crabtree and C. J. Shipman in performing the labo- ratory work is gratefully acknowledged. The authors also express their appreciation for the work of personnel in the Analytical Chemistry Division of ORNL under the supervision of C. L. Burros, J. H. Cooper, W. R. Laing, C. E. Lamb, H. A. Parker, and G. R. Wilson. w REFERENCES . G. |. Cathers, Nucl. Sci. and Eng., 2, 768-777 (1957). . W. H. Carr, Jr., Chem. Eng. Symposium Series, 56, 57-61 (1960). R. P. Milford, S. Mann, J. B. Ruch, and W. H. Carr, Jr., Ind. and Eng. Chem., 53, 357-362 (1961). F. F. Blankenship et al.,, ORNL-2548, p. 60, Oak Ridge National Laboratory (1959). G. I. Cathers, M. R. Bennett, and R. L. Jolley, Ind. and Eng. Chem., 50, 1709-10 (1958). T G. |. Cathers, "Dissociation Pressure of MoF 4~NaF Complex and the Inter- action of Other Hexafluorides with NaF," paper presented at American Chemical Society Meeting, Sept. 3-8, 1961. 24~ DISTRIBUTION Nuclear Science and Engineering E. J. Murphy F. L. Culler D. E. Ferguson H. B. Graham F. R. Dowling, Wash. AEC E. L. Anderson, Jr., Wash. AEC . J. Vanderryn ORO AEC . G. |. Cathers . R. L. Jolley E. C. Moncrief Lab Records (RC) . Laboratory Records . Central Research Library Document Reference Section . DTIE