ORNL/TM-11792 Chemical Technology Division REACTION OF URANIUM OXIDES WITH CHLORINE AND CARBON OR CARBON MONOXIDE TO PREPARE URANIUM CHLORIDES P. A. Haas D.D. lLee J. C. Mailen Date Published — November 1991 Prepared by the OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee 37831 managed by L MARTIN MARIETTA ENERGY SYSTEMS, INC. © 7~ for the U.S. DEPARTMENT OF ENERGY under contract DE-AC05-840R21400 i VA, SAE rear g &S5, Wy o= o ST TR e e et L -.v_.);':g y ]1} Y Bl BAS e e el L L “EE IR g B T iR g e g o T T B D i - T . i:tr"';;t:fi'{?;- L iryr e e DISCLAIMER This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their empioyees, make any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefuiness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. DISCLAIMER Portions of this document may be illegible in electronic image products. Images are produced from the best available original document. | 31 3.2 33 34 3.5 3.6 CONTENTS INTRODUCTION . ... ittt et ie i rseraanaenannenns 2. EXPERIMENTAL APPARATUS AND PROCEDURES ................. 3. RESULTS .. i i it ittt eatasansenaanaonns REACTIONS OFCHLORINE ...... .. i iiiieiieaaes UTILIZATION OFCHLORINE ...... . it EFFECTS OF URANIUM OXIDE PROPERTIES ................. REDUCTIONS BY CARBON ... ...ttt iiiieneenn REDUCTIONS BY CARBON MONOXIDE ...............oolss VOLATILIZATION AND CONDENSATION OF URANIUM CHLORIDES ... ittt iiieiiaentetnaneaanaaaeans 37 PRELIMINARY RESULTS WITH A LARGER REACTOR AND BOTTOM CONDENSER. . .+« e oo oo e e e e 4. CONCLUSIONS - oo oo 5. REFERENCES - ..o e APPENDIX - - o oo e e A, THERMOCHEMISTRY . .. eo e e e e e e e e B. INDIVIDUAL TEST RESULTS . .. uonttere e C. EQUIPMENT DETAILS AND CALIBRATIONS . ...cueueuvunnnn.. iil 10. 11. 12. 13. 14. TABLES Chlorination test parameters and conditions ........... ... .. .ot 11 Chlorination test results .. ... ii ittt eennennaereeenns 14 Changes in process conditions and results .......... ... ... oLt 18 Changes in apparatus and procedures . .........c.cciiiiiieiiiiiia 19 Chlorination test material balances: chlorine ............... ... ... ... 20 Chlorination test material balances: catbonandoxygen ................... 21 Chlorine losses versus Cl, feed rates using carbon black at ~730°C .......... 23 Properties of uranium oxide, UO,-C, and carbon feed solids ....... e 24 Gas flows and uranium volatilization .............. . it 29 Thermochemical data ..........ouiiiiiiiieiieiieonnnennaennenns 44 Heats of formation for U-O-Cl compounds at 298°K ..................... 45 Free energies of formation for U-O-Cl compounds at 900°K (627°C) ......... 45 Vapor pressure eqUations ... ......eeenioriactaatii e 47 MS-22 gas flow calculations from totalizer readings ....................... 53 FIGURES Conversion reactions for U-O-Clcompounds .......... .. ...t 4 Phase diagram for UCL,-UO, ...... . ittt 5 Chlorination reactor (4-cmID) ....... ... . .o 7 Photograph of reactor as removed after MS-8 ............. ... ..ol 8 Chlorination reactor (68-mm ID) with bottom condenser .................. 33 Vapor pressures of uranium chlorides .......... .. ... oo 48 CO, concentrations during MS-22 ........ ... .. it 52 vii ABSTRACT The preferred preparation concept of uranium metal for feed to an AVLIS uranium enrichment process requires preparation of uranium tetrachloride (UCl,) by reacting uranium oxides (UO,/UQ,) and chlorine (Cl,) in a molten chloride salt medium. UOQ, is a very stable metal oxide; thus, the chemical conversion requires both a chlorinating agent and a reducing agent that gives an oxide product which is much more stable than the corresponding chloride. Experimental studies in a quartz reactor of 4-cm ID have demonstrated the practicality of some chemical flow sheets. Experimentation has illustrated a sequence of results concerning the chemical flow sheets. Tests with a graphite block at 850°C demonstrated rapid reactions of Cl, and evolution of carbon dioxide (CO,) as a product. Use of carbon monoxide (CO) as the reducing agent also gave rapid reactions of Cl, and formation of CO, at lower temperatures, but the reduction reactions were slower than the chlorinations. Carbon powder in the molten salt melt gave higher rates of reduction and better steady state utilization of ClL. Addition of UO, feed while chlorination was in progress greatly improved the operation by avoiding the plugging effects from high UO, concentrations and the poor Cl, utilizations from low UO, concentrations. An UO, feed gave undesirable effects while a feed of UO,-C spheres was excellent. The UO,-C spheres also gave good rates of reaction as a fixed bed without any molten chloride salt. Results with a larger reactor and a bottom condenser for volatilized uranium show collection of condensed uranium chlorides as a loose powder and chlorine utilizations of 95-98% at high feed rates. 1. INTRODUCTION The feed for an Atomic Vapor Laser Isotope Separation (AVLIS) process for uranium will be uranium metal.! The principal production of uranium metal for nuclear fuel cycles has previously been by batch metallothermic reductions of uranium fluoride (UF,) using magnesium or calcium metal. If this batch metal production were used for a large AVLIS enrichment plant (=10* ton Ufyear), the costs of the hydrogen fluoride (HF) feed, the calcium or magnesium feed and the disposal of magnesium fluoride (MgF,) or calcium fluoride (CaF,) wastes would be major parts of the total uranium enrichment costs. Alternate processes for preparation of uranium metal from UCI, allow recycling of Cl, from electrolytic cells.? The products of uranium ore refineries are uranium oxides— most commonly UO;. The objective of this AVLIS development program was to determine practical process conditions for efficient production of UC], from uranium oxides. There is extensive background literature on the production of UCI, based on reaction of carbon tetrachloride with uranium oxides.> Such a process was used at Oak Ridge for producing calutron feed material. Despite this previous experience, a process based on the use of CCl, would not be desirable for the proposed continuous metallothermic reduction process in which chlorine values are recycled because of both the risks associated with CCl, and the fact that recycle of chlorine values would require process equipment for CCl, synthesis. Such difficulties could be avoided with a process involving direct reaction of chlorine, carbon and/or carbon monoxide, and uranium oxides. A search of the technical literature did not reveal any report of the preparation of pure UCl, from uranium oxides, Cl,, and C or CO. Canning demonstrated nearly complete utilization of Cl, for up to 90% chlorination of impure uranium oxides,* as a first step of an overall process for producing purified metal. He sparged Cl, through a graphite diffuser into molten KCI-NaCl to which he fed the impure oxides. After chlorination, the molten salt mixture was treated with magnesium metal for reducing uranium compounds and electro- refined to improve the metal purity. Lyon reported rapid reaction of uranium oxides with chlorine in molten NaCl-KCl at 850°C to produce UO,Cl,°> Gens studied the volatilization of uranium chlorides from nuclear fuels and also found the formation of some nonvolatile UO,CL,* Gibson reported complete chlorination of UQ, using a block of carbon and Cl, in KCl-NaCl at 860°C as a first step in a process for producing purified UO,.” These results, along with those using CCl, rather than Cl,, report chlorination of uranium oxides and removal of oxygen as CO or CO.,. The use of published thermochemical data for uranium oxide, oxychloride, and chloride compounds is a first step for identifying probable reactions for the desired chemical conversion. However, there are several reasons for uncertainties for such predictions. Uranium chemistry is complex with stable valence states of 3, 4, 5, and 6. Oxides, one or more oxychlorides, and chlorides have been identified at most of these valence states. The volatility of the chloride compounds increases with valence state while chemical stability decreases. All of the compounds have large negative heats and free energies of formation; hence, calculation of free energies of reaction generally involves the inherent uncertainties of small difference of large numbers. A detailed compilation of available thermochemical data for the various uranium oxide, oxychloride, and chloride compounds is presented in the Appendix. Evaluation of this body of data leads to the following general findings: ® At a given valence state, the oxychlorides are more stable than the chlorides. ® At a given valence state, the oxides are more stable than the chlorides. e All additions of chlorine to oxides or oxychlorides of lower valence [less than U(VI)] are favorable to give oxychlorides of higher valance. e The oxychlorides can be formed both by direct reaction of chlorine and by reaction of an oxide with a chloride. | e Production of UCI, requires a reducing agent whose oxide product is much more stable than the chloride; that is, it does not react with uranium chlorides. Carbon and carbon monoxide meet this criteria. Hydrogen does not and the oxide (water) reacts with uranium chlorides to produce HCL Figure 1 presents a summary of the expected thermochemistry of the uranium oxide, oxychloride, and chloride compounds at various uranium valence states. The phase relationships between tetravalent uranium oxide and chloride are shown in Fig. 2.2 The phase diagram shows that there are three stable compounds over the entire range of composition—UCl,, UOC],, and UO,. There is a eutectic reaction between UCI, and the intermediate compounds, UOCI,(UCl, + 50 mol % UO,). The melting point of pure UC], is 590°C. A minimum melting point of 545°C occurs at the eutectic composition of UCl, + 6.9 mol % UO,. A maximum solubility of about 13 mol % UO, in molten UC], is reported at 810°C. At temperatures from 810 to 855°C UCl, vapor is in equilibrium with solid UOCl,. UOCI, decomposes at 855°C. At higher temperatures, vapor and solid UO, are in equilibrium. The reasonably high solubility of UQ, in molten UC], over the temperature range of 545 to 810°C suggests a desirable precondition for achieving rapid rates of reaction between chlorine, carbon/carbon monoxide, and UG, dissolved in molten UCl,. The present investigations were carried out to characterize the production of UCI, by a direct carbochlorination of uranium oxides in a molten salt medium and to identify the optimum conditions for accomplishing the conversion in the most economic fashion. Variables that have been considered include the physical and chemical characteristics of the oxide feed material, the use of carbon versus carbon monoxide as a reductant, the physical characteristics of carbon reductants, the effects of feed rates, and the effect of temperature. ORNL DWG 891A-10 VALUES IN PARENTHESES: MINUS FREE ENERGY OF REACTION, 900 K, kJ/equiv MULTIPLIER OUTSIDE PARENTHESES: EQUIVALENTS/REACTION, URANIUM 0/U, ATOM/ATOM VALENCE 0 1 2 3 u(o) U U U +3/,Cl, +1/20,+1/2Cl, 3(233) 3(260) +1/202—C|2 +0 +8/0 ______> 6 u(in) UCI3\ 5(20) UO({ 4(233) 195(187) +'%4Ci, +C0O-CO, +4Cl, +C0O-CO, 95 —2(48) 120 -2(8) +1/202—C|2 +1/202-"C|2 ' _—.....’ + 15l + 140l +15cl +2/3C0-%/3C0, 14 100, 0. -\ %(52) \ +Y20,-Cly +120,-Cl> _—_> ___> u(v) UCls— a0 UOCls 5 U0 Cl +C0—CO; 2(71) +1/4C[z 1, (100) +1/2C| +C0-CO, +1/2C|2 + V50, —-3/4Cl ' o -2(11) 20 1,[(U0, ) ,Cl4] ——%63%—2»’/3[0308] +"4Cly +'/60 4 '/,(110) —%4(8) U(Vl) +1/202—C12 +V202—C|2 +1/202—C|2 —_— —_— —_— Fig. 1. Conversion reactions for U-O-Cl compounds. | Wt % 99 97 95 90 85 . TPl Py 1 ol P | 900 ) Vopor + U0, — 800 810° _| ©00 — \ - S00H - | | 400H — | 300 +_ UC|4 ss + UOCIZ — I | 2001 = 100 S| - = ol 1 o104 ovor o1 ot | 19 i S UCIa 2 6 o 8 O U0, — Mol % Fig. 2. Phase diagram for UCl,-UO,.? 2. EXPERIMENTAL APPARATUS AND PROCEDURES The experimental system was assembled from small flowmeters, 0.25-in. OD metal tubing and fittings, and quartz or borosilicate glass components fabricated by the Oak Ridge National Laboratory (ORNL) glass shop. For hot chlorine gas, the materials of construction were borosilicate glass for up to 450°C and quartz for the chlorination reactor up to 900°C. For chlorine gas at room temperature, Monel, Teflon, and Viton rubber were also used with some stainless steel for short time periods. Apparatus after the Cu-foil trap for Cl, was mainly stainless steel and plastic. A diagram of the apparatus as used for Tests 5 to 18 is shown in Fig. 3 The initial arrangements as used for Tests 1 to 4 were described with the preliminary results.> The quartz reactor was 4-cm 1D, 69-cm long, with a closed bottom and a 65/40 ball-joint socket at the top end (Fig. 4). The mating half of the ball joint had the fittings for all connections, including a gas sparger and the gas outlet. The initial charge was loaded into a quartz crucible of 3.2-cm ID, which was lowered on to a small pad of quartz wool on the reactor bottom. The initial reactor was fabricated with a quartz jacket for cooling from 38 to 53 cm above the bottom end. A simple borosilicate glass sleeve with "O" rings was used for air cooling after the initial unit. The gas sparger and the thermowell or UO, feed line were installed through the reactor cap using slip-fittings ("O" rings or Viton rubber sleeves) to allow length adjustments. The apparatus and procedures allowed three different material balances for each test. The weight balances were probably the most accurate but gave only the overall run material balances. The weight measurements were: 1. The weight loss by a CuO reactor indicated the oxygen used to oxidize CO to CO.,. 2. The weight gain in a final Ascarite absorber showed the CO, that was produced from CO. 3. The weight gain for the first Ascarite absorber showed the CO, in the gas leaving the chlorination reactor. 4. ‘The weight gain by the Cu-foil trap showed the Cl, in the gas leaving the chlorination reactor. 5. The weight gain by all reactor components equals the Cl, reacted plus the solids feed minus the C or O as CO, or CO, shown by a, b, and c. (Note that these quantities must allow for whether the oxidation reaction is C to CO,, C to CO, or CO to CO,.) ‘(d] wo-§) 1030831 UONRULIONY) "¢ ‘T Y 1SNVHX3 QOOH 90 L —v086 Y3ILINONYNA dILIAN 183t L ] 1w _ ¥ ) o 3 2 v | N 1] : n 4 - [3ovnanid] 704 N VUI ’ . i - o] SHILIAMOTS SSYAN 94 0oL 14 SYALINVLIOY 9¥ Ol 1Y OMa INYO xr e—— o N 02 '8-SIN 191JE PIACWIAL SE 1070831 JO ydesSoroyq p Sy The second material balances were from gas flow rates times concentrations. The concentrations and flows were: NN R WD 100% N, for diluent gas feed, 100% N, for wet-test meter flow out, 100% Cl, for Cl, feed, 100% CO for CO feed, in-line CO, measurement for gas out of the Cu-foil trap for Cl,, in-line CO measurement for gas out of the first Ascarite absorber, and the N, content of the gas out of the first Ascarite is 1 minus CO. The third, and generally least accurate material balances, are from differences in flow rates. These differences are: el reactor out minus Cu-foil trap out is the Cl, that leaves the chlorination reactor. Cu-foil trap out minus the first Ascarite out shows the CO, absorbed. the first Ascarite out minus the wet-test meter shows the CO oxidized to CO, and absorbed on the final Ascarite. the Cl, feed, plus the N, feed, plus the CO feed, plus CO, or CO from C in the charge, minus the reactor out indicates the Cl, reacted with the charge. The results found in the literature provide little information for selection between the chlorination process alternates. The reducing feed when using Cl, might be: P massive carbon or graphite blocks as reported, carbon particles or powder, carbon particles or powder mixed with uranium oxide powder and compacted, or CO gas. Any of the final three alternatives appear better than the massive blocks with respect to the reactivity, cost, and ease of replacement when consumed. Further, the reaction medium might be predominantly: 1. inert, low melting chloride salts, 2. uranium chlorides, 3. a fixed bed of uranium oxide-carbon granules or chunks, or 4. a fluidized bed of uranium oxide and carbon solids. The phase diagram (Fig. 2) for UO,-UCI, shows formation of UOC], with all liquid for up to 7 mol % UQ, at the melting point of UCl,. This solubility of UO, in molten UdCl, appears to be a desirable condition for easy and rapid chlorination and reduction reactions. 10 The chlorination condition of greatest interest for the AVLIS feed process was to react UQ, and Cl, in a molten sait medium. Chlorination studies were planned for these conditions, and the selection of other conditions proceeded as follows: 1. The first tests were with a block of carbon for diffusers as favorably reported by Canning* and Gibson.’ 2. After good reaction of Cl, and formation of CO, were demonstrated with the carbon diffusers, CO was tested as a more practical reducing agent. 3. After results with CO showed that the reduction reactions were much slower than the chlorination reactions, carbon powder was used to determine the effects of this reducing agent as compared to CO. 4. Since high UQO, concentrations resulted in sparger plugging problems and low UQO, concentrations gave poor utilizations of Cl,, the experimental apparatus was modified to allow UO, feed while chlorination was in progress. 5. Tests were made with ball-milled UQ, powder and with UQO,-carbon black spheres for comparison with the UQO, spheres and the petroleum coke first tested. 6. Comparison tests were made with UQO, spheres as feed and with a fixed-bed of UO,- carbon black spheres without any molten salt. 7. The experimental apparatus was then modified to use a larger reactor with a bottom condenser. The first four experimental tests were previously reported.’ After eighteen tests (including the initial four) in a small apparatus, the equipment was modified to provide a larger reactor and a more useful condensation arrangement for product vapors. 3. RESULTS Eighteen experimental tests were made in a quartz reactor of 4-cm ID and were directed toward the molten salt chlorination of uranium oxide. Three tests were not completed as planned because of failure of reactor components. One test was to check the procedures and material balances without chlorination. The remaining fourteen gave useful chlorination results. The test conditions are summarized as a tabulation (Table 1). 1 0 2o 0 8L0 8t°0 8L'0 £8°0 wo 9¢°0 o v, 06°1.2-60 s 0 670 0 Lo 80 8L°0 80 ¥00 $T0 O olL 06-5¢-60 \4) 0] s810 0 0L0 LLO 8L0 Tl 000 620 o 089 06-L0-60 tl 089 0 Lo 0 +9°0 LLO 8L'0 0Tl 00 y1°0 ) L9 06-¥0-60 7l 069 0| 9500 0 A 0 0 L90 0 70 o SL9 06-TC-80 Dl 0 0 0 0 0 0 L90 0 Yo o $t9 06°12-80 il 0 0 0 0 ¢ 0 L90 0 yeo JUON $t9 06-0T-80 Vil 089 0 0 0 Al 0 0 LSO 200 0 ) 0z D 8LL 16-¥1-S0 8 0 0 0 990 £0 £0 'l 0£0 8l D 8LL 16°10°50 7z o| sol 0 L0°E £0 £0 862 020 ¥l D vLL 16-¥2-90 9 szo | 90 0 L6’ £0 £0 002 $1°0 71~ 2 LLL 16-€0-$0 T LLL | 0 8¢ €0 £0 002 sI'0 960 D 8EL 16-92-£0 e cLlom 0 Lt 120 120 e $1'0 gl e 1£L 16-91-10 €2 SIL seo | 1L0 0 $T 120 120 Ly $T'0 Lt 9 ¥69 16-€0-10 w 669 0{ 90 0 I't 0 0 (A 50'0 £1 O LL9 06-81-21 12 o wo 0 99°0 0 0 Sy $I°0 'l O 569 06-90-C1 0z 0 0 0 0 0 0 0 0 0 QUON $L9 06-82-11 6l YASNIANOD WOLLOE “HOLDVHY MHDUVT OL AHONVHD WALSAS HOLOVAY 0 0 0 0L'0 0 0 vl PE'0 0 2D 00L 06-90-11 81 o| 1o 0 SE0 801°0 SZI'0 60 700 860 S 089 06-0¢-01 L1 $L9 €1 | 1€0 0 8L'0 801°0 STI'0 0 +0°0 ST0 D soL 06-91-01 91 1Y 5| on ) 10 0 e | GDIN D ‘on DN ade .) sep 'ON Supapay amnjeradwy 199, 1591 SW (low) ameradway 1e poag (jow) 231eyd jemu 0RUIN (ponuguoa) SUONIPUOS pue s1ojowered Js9) uoneUuloD I JqEL 13 The progress of the chlorinations was indicated by the measurement of gas flow rates and concentrations during the tests. The overall or average results are shown by the material balances from reactor and trap weights after cool down to room temperature. Some results for the useful chlorination tests are tabulated (Table 2). Additional details by test numbers are in the Appendix. Each individual result can be interpreted in a number of ways since the overall process reactions result from a number of multiple step reactions. Conclusions for important process parameters are discussed as separate sections. In general, a specific conclusion cannot be proven by one test result. Instead, the test results (Table 2 and the Appendix), along with thermochemical data, must be considered as a whole to justify the conclusions. Rapid reaction rates and removal of oxygen as CO or CO, were demonstrated in the first four experiments as previously reported.” The results of these initial tests are included in the tabulations, but the details in the initial report® are not repeated. The test conditions (Table 1), and the brief tabulation of results (Table 2) do not clearly show how the process conditions, apparatus, and procedures were changed between experiments. Some of the most important changes in process conditions and the results are listed in Table 3. Important changes in apparatus and procedures, and the corresponding results are in Table 4. The material balances tabulated for chlorine (Table 5) and oxygen and carbon (Table 6) are averaged values with more weight to the more accurate mass measurements. 3.1 REACTIONS OF CHLORINE All of the chlorination studies were done with uranium oxides and the reducing agent in the reactor at temperature when the chlorine flow was started. All tests show that the initial reactions of chlorine go very well without any significant concentration of chlorine in the exit gases. This was true when the initial charge contained melts of UCl,, MgCl,-NaCl or MgCl,- LiCl, UCl,-MgCl,-NaCl], or a fixed bed of UO,-C particles. The reaction rates for chlorine can be much higher than the rates of reduction indicated by CO or CO, flows. This result, thermochemical data, and some melt analyses indicate that the chlorine reacts by "oxidizing" U(IV) to higher valances; that is, the chlorine adds to the U(IV) compound to give U(vl) or perhaps U(V) compounds. Reaction of Cl, with UO, or oxychlorides is more favorable than reaction with UCl, to give UCL or UCl,. After U(IV) oxides or oxychlorides are depleted in concentration, UC, is reacted to give UCl; or UClI; that are much more volatile. 14 Table 2. Chlorination test results Overall/average Cl, reacted MS Furnace run result® (%) UO, feed test temp. Cl,feed O,in UO, atom Cl No. (°0) reacted evolved mol UQ, Principal conclusions and results 1 850 ~50 ND? ~3.0 UO,+Cl, + graphite - CO,+UCl,+UO,ClL. 2 700 ~50 ND small Most of Cl, reacted with Monel sparge tube. 3 650 ND ND small UQ;+CO - UO,+CO,. Severe plugging of 790 ND ND ND sparger (perhaps UOCI,). Nearly all of uranium volatilized out of crucible. 4 625 42 40 2.8 Consistent material balances, gas flow rates and gas concentrations. Condensed solids were predominantly UCl, while charge was half UCl, and half oxychlorides. 5 650 ~ - — Quartz reactor broken. 6 650 0 — 0 Operator qualification completed. 7 655 38 <10 23 UO,+CL+CO - UOLCL+CO with little removal of O as CO, Apparently UCl, is necessary for reduction by CO. g 645 30 55 28 The results of MS-4 were duplicated with good - 680 material balances. The rate of CO, evoluation showed little variation with temperature, Cl, rate, and sparger immersion. Complete plug by condensed solids. 9A 645 40 70 2.5 The C gave CO, without any detectable CO. 9B 645 0 0 Condensed solids plug of same appearance as MS-8. Leaks at the in-reactor sparger coupling resulted in low CL, utilizations. 10 645 36 >100 42 Condensed solids plug of same appearance as 680MS-8 and MS-9. Total material balalnce for MS-9 and MS-10 indicates complete conversion of UO, to UCI,. 11A 635 No condensed UCI, without gas sparge. 15 Table 2. Chlorination test results (continued) Overall/average Cl, reacted MS Furnace ___run result® (%) UO, feed test temp. Cl,feed O,in UO, atom Cl No. (°CO) reacted evolved mol UO, Principal conclusions and results 11B 635 Low UC], volatility in agreement with UCI, vapor pressure without Cl,. 11¢ 635 40 >80 ~5.0 The feed of uranium oxide spheres into the UClL,-C 690 with Cl, gives an immediate evolution of CO, that peaks in a few minutes. 12 635 65 115 4.6 The rates of CO, evolution and the utilization of 680 Cl, increased as the oxygen inventory was increased by successive additions of uranium oxides. A 50% increase in Cl, rate gave a 20% increase in CO,. A 45°C higher temperature gave 60 to 80% increases in CO,. 13 680 60 100 4.0 At 680°C, some point utilizations of Cl, were 100% and average values for 30 to 60 min were 80%. 14 710 92 102 4.0 Excellent Cl, utilization and conversion without excessive volatility of uranium. 15 745 63 106 3.9 Same type of plug as observed for 100% UCI, in MS-8, -9, -10, and -11. 16 705 92 115 42 Flow rates showed =99% utilization of Cl, until end 675 of test when oxides were depleted. At 705°C, about 10% of the oxygen evolution was as CO. This decreased to a low value at 675°C. Carbon black is more reactive than coke. 17 680 65 145 3.8 UO, feed gives more complex reactions than UO,. Uranium is more volatile with UO, feed. Cl, utilization is lower. 18 700 88 120 3.8 UO,-carbon black gel spheres react well without molten salt. Typical dark plug of condensed solids stopped run while CO, rate was still high. Reactor System Changed To Larger Reactor; Bottom Condenser 19 675 Mechanical and thermal performance of new components are good. 16 Table 2. Chlorination test results (continued) Overall/average Cl, reacted MS Furnace ___run result® (%) UO, feed test temp. Cl,feed O,in UO, atom Ci No. (°C) reacted evolved mol UQO, Principal conclusions and results 20 695 &5 117 4.9 About 0.17 mol condensed U collected as powder. Deposits on quartz wall above furnace and inside insulation below furnace. 21 677 94 107 5.0 0.77 mol condensed U collected as powder. 699 Condensed U on vessel walls was about 0.59 mol or ~40% of volatilized U. Good operation except for two feed-line (UO,) plugs. 22 694 82 120 4.4 Cl, utilization near 100% for 1 h. When Cl, 715 utilization decreased, addition of 2 g C-black (48 g C-core in charge) doubled CO, evolution. N, purge eliminated UQO, feed plugs. Different UO, feeds showed same reaction rate. 23 731 72 75 34 Quartz crucible cracked — probably during heatup. A quartz baffle above crucible prevented UCI, deposits in top of reactor. Other results compromised by cracked crucible. 24 138 95 130 43 Reduction rate equaled chlorination rate. 777 CO, rate = 0.5 Cl, rate = UO, feed rate; 141 g in product jar. Rates of reaction determined by Cl, feed rate without effects from temperature or the amounts of UO, or C inventory. 25 777 92 140 4.5 The MS-24 results confirmed with doubled UO, (0.60 mol/h) and Cl, (1.25 mol/h) feed rates. Condenser cross section plugged by dense solids. 26 774 92 120 4.4 The MS-24 and MS-25 results confirmed at higher UO, and Cl, feed rates; 329 g in product jar or 67 mol % of UO, feed. 27 718 95 160 4.2 Material specimens exposed for 6 h. Cl, utilization and percentage collection of product solids about same at the low rate and five times higher rate. 28 718 93 120 4.3 The MS-24 to MS-26 results confirmed at flow rates up to 2.8 mol/h Cl, feed. Continuous feed for UO, demonstrated, but plugs in feed line were troublesome. Reactor cross section plugged above condenser. 17 Table 2. Chlorination test results (continued) Overall/average Cl, reacted MS Furnace __ run result® (%) UO, feed test temp. Cl,feed O,in UO, atom Cl No. (°C) reacted evolved mol UO, Principal conclusions and results 29 778 98 145 4.2 Material specimens exposed for 6 h with CO in the sparger gas. Cl, utilization and product collection results of MS-27 were confirmed. Thermowell in melt shows salt temperatures of about 730°C. 30 725 97 125 4.4 Good Cl, utilization at lower temperatures, salt temperatures of about 650°C. High N, flow to sparger gave high U vaporations at lower melt temperatures; 73% of condensed U in product jar. 31 690 97 110 3.4 Short period of Cl, flow did not complete conversation of UO,. Cl, utilization good at melt temperature of 645°C. Dip samplers demonstrated removal of samples of melt. ®ND - not determined (measured). "AnieIoA n 19ySy pue uoneznn %D Jamo] Juipnput $190]J0 PRIUBMUR [RISAS ‘UONIBal XIidiwoo IO 109132 1231¢] s1 awn dwes 1e uonppe [DeN-YIDFW Ing (Amejoa (1 Taso] Ajqeqoid) YD Jo suoneziin 191 Yony "Do01L 10 $89 1e sgnyd 9jqonis ou ‘) Jo ANIRIOA JoMO] YININ 18 -aInssaxd odea ‘1DN UM JUaRITe Ul ‘IS ST [] JO UOHBZNEIOA ‘11 JO uonez|nejoa ON ‘uaf8£xo Jo _m.onE areidwod 20w ‘suoneznn 4y (10ysiy) Jenag ‘suonezZInn 4D pood nq ‘0D 10 FQD AGrINIAP ON "sonel onfon udy e 18eds Jo Suiddnyd v1asag ‘uonnjoas Q) Jo selel Ieysiy yony ‘sanesadwa) Jamof 18 10D Jo uonnjoaq Auo L1 [A! €T TT L1oel Aluo g11 Auo V11 6 Aluo ¢ Auo T oh oD t ‘on Jo 20eid ut “ON JO pIdd voneiado Suunp 0N Jo paag "o J0J WANIp B S8 213093 [DeN-IDIN (%D oN) #81eds *N yim *'1on a81eds sed noyim DN 02 Jo 20eid ur sajonted 9302 WN|ONdJ "o noynm Jaw [DeN-08W 10 DITIDSN (1DEN-IDIT-4D3 I ou) [es apuIofyd Ajuo ay se PN Jopunks anydead pauuy Jo aoerd ui JueIPNPal 0D 23ueyo woly nsay a8ueyo i 3893 18T a8uey) §}jnsoJ pue sUONIpuod ssaoold ul saguey) "¢ S[qE], 19 Table 4. Changes in apparatus and procedures Results from change First test Change with change Cu foil trap for Cl, in place of 3 thiosulfate scrubbers. In-line measurement of CO, and 3 CO concentrations. Monel sparger for Cl, feed. 2 only Open end to perforated sparger; 4 perforated sparger to open end. 10 Shop calibrations of flow meters. 9 Long crucible with top extending 9 above insulation. Long quartz sparger without coupling 10 in reactor. Feed of UQ, particles during 11C chlorination. Larger reactor and bottom condenser. 19 Good measurement of amount of Cl, trapped; better operation of Ascarite traps for CO,. Real-time indications of the rates of reaction. Cl, reacted with Monel, then bypassed without reaction. Not imtl))ortant to plugging; a 45° cut open end is best. Much better accuracy for Cl,, CO, CO rates from flow differences; better material balances from flow rates and concen- trations, However, calibrations change with exposure to Cl,. Allows easy removal of condensed solids or return to charge by melting. Eliminated leaks and coupling failures as cause of low Cl, utilizations. Much better operation with respect to Cl, utilization, sparger plugging, and reduced volatility of U. 60% of condensed U as powder on the bottom. 20 "UOTIZNUIOUO0D SN MO sed soredipul D4, "SO1B1 MO[J Bl $IOUDIDJJIP SIIBINPUI (Qx, *SIUILIINSEINU JYdl1oMm SOIBIIPUI IM, L1 £L L9 w 89T 87 €7 L0T 91 L0 SL'O €T T 98 8 8 6£T 897 Wt 1€ 86’1 LEO vr'0 w vz 98 6 89'C 897 0€T 8€0 L1'0 12 6 80°¢ 787 LT 18 98 990 89°0 $S'0 £1°0 600 0z gl 8 L8 98 oL0 090 SL'0 £5°0 90 L0'0 960°0 81 9] 0s 89 69 9¢'0 9¢'0 SE0 810 V20 810 1o LY 1T ¥6 6 6 9L'0 8L0 080 €L°0 vL'0 8¥0°0 7900 91 7T ) £9 86 wo LLO 080 €50 0S°0 020 0£'0 Sl A/ $6 16 16 1L°0 8L0 Lo vL'O 90 00 790°0 ¥1 02 £9 SS 9¢ 1.0 1L0 690 S0 8€°0 LT0 10 €1 ¥ €L £9 9 ¥9'0 L90 £9°0 6¥0 or'o 810 £2°0 Al LT oy Sy o¢ £€°0 0£0 8¢0 10 L1'0 810 120 o1l 97 ve 9¢ 8¢ LYo o SY'0 S10 91°0 620 620 V0!I 01 6¢ 8¢ vt £°0 9’0 or'o 600 s1°0 v1°0 $T0 6 A 61 0g €y 901 6L'0 98°0 0z 92'0 650 09°0 8 [~ pI 8¢ 8t 950 050 950 L0'0 170 €70 SE'0 L 160 poInsEdW JON | SE 137 $€0 | pomseawoN | 1€0 S60°0 10 $T0 070 v D: pa3) pue ad M a4 oM (10wy/j0w) dd m M den oy | DO podg ad WM 101083y | 10eoy | denn) den n) ‘ON panedl *oNfo (%) wonezpnn ' (tow) ¥1D 110, (1ow) 10npord D o QULIO[YO :SPOUE[RQ [ELIDJBW 1$9) UONRULIO[YD °C d[qe], 21 "§91R1 MO[J Ul SIOUIISJIP soredipul (d, “SIUOWIINSEIW 1YTIom SIIBDIPUT 1M “UONBIIUIOUOD SOW mO[) sed $a1ed1put D, 8v0 SL 180 L60 $60 0 SO0 <00 200 0 €C §S°0 0c1 171 STl 60’1 0 0 <00 0 0 [é poInsesul 10 001 €01 88°0 £0'1 0 00 10N 100> 0 1Z wo ril LT0 yZ0 ¥eo0 0 0 0 100> 0 474 LSO 0ct sy S0 8¢0 S0'0 600 900 L00 0 .wfi 1 ori1 LT0 12°0 L0 00> 0 0 100> 0 Ll sS0 0zl 8t’0 Iv°0 144" 90°0 00 00 €00 0 91 €50 o1l 070 €20 820 0 0 0 100 0 ST LvO 001 LT0 £€0 ye0 painseail JON 0 0 0 0 14! €S0 001 910 S1°0 170 100> 0 0 0 0 tl L0 ort 970 ¥20 1€°0 palnseaul JON 0 0 100 0 Cl1 80 00Z~ 90°0 <10 ¢10 0 0 0 0 0 11 6’0 00C~ 110 1o 10 0 0 0 0 0 Vo1 90 oL LOO 600 010 palnseaul 10N 0 0 0 0 6 80 ss a0 €20 070 a0 S0 LE0 LE0 6v°0 8 1'0> o1~ 0~ 100 00 950 L0 ¥9°0 L90 680 L 80 )4 ¥80°0 LLOO 980°0 poInseawt JON veo o 120 £C0 v (jowy/[our) no JIM IM M poedI 49 uodkxo *ory | G WOl | D WoId | °sqe woid -] WwoIg S wold | 'sqewoly | Ond wolj (jour) ‘ON Ano Q) JO 120194 D 4q 1S9, (four) 10 QD (four) 1030831 JO MO 0D P39} 0D uoFAxo pue UoqIed :$OUR[Eq [RIIDIEW )53 UOHEUMIOYD ‘9 dJqBL 22 than UCl,. This operating condition results in large losses of Cl, to the off-gas. Either the completeness of reaction of Cl, is limited by a less favorable equilibrium or the UCls or UCl decomposes as the vapors cool to give condensed UCI, and Cl,. Operation with UO; as the feed instead of UO, also gave higher volatilities of uranium and larger losses of Cl,. This result also indicates that U(IV) oxides or oxychlorides are very helpful to high utilizations of chlorine. 3.2 UTILIZATION OF CHLORINE High utilization of chlorine required a good inventory of U(IV) oxides and oxychlorides in the reactor charge. The initially charged UO, supplied this inventory so that chlorine utilizations were always high when the chlorine feed was started. Continued high utilizations with complete chlorination required the reduction of U(VI) back to U(IV) at rates that prevented depletion of U(IV). Continued high utilizations were demonstrated using carbon as the reductant and reactor temperatures of =2670°C. The minimum conditions to assure high chlorine utilizations for steady-state operation were not clearly determined. The following conditions gave poor chlorine utilizations of <70% (i.e., more than 30% of the feed chlorine was trapped from the exit gas (see Table 5). 1. All tests with CO as the reductant. (The best combination of higher temperatures and an optimum continuous feed of UO, might give higher chlorine utilization using CO.) 2. All tests with reactor temperatures near 630°C. 3. All operation with low concentrations of oxygen in the reactor charge. The limit on oxygen concentration depends on the Cl, feed rate and the concentration and reactivity of the carbon. Melts as low as 2 mol % oxides—98 mol % chloride can give good chiorine utilizations, but there is not enough information to determine limits. Because of this effect, operation at conditions intended to complete the conversion of UO, to UC], gave higher chlorine losses to the off-gas. When chlorine utilizations were good, the Cl, removed in the Cu trap was only a small percentage of the total gas flow. The calibrations of the two flow meters changed continually, and calculations of flow differences using separate calibrations of the two meters did not give useful measurements of the small Cl, losses. Finally, the whole run Cl, losses measured by the Cu trap weight gain were used to calculate meter factors that match the whole run flow differences to the Cu trap result. These meter factors gave much more consistent results for the short period Cl, losses. Some results from the large reactor tests using carbon black (all melt depths of 7 to 10 cm) show Cl, losses of 2 to 4% for a range of Cl, feed rates (Table 7). 23 Table 7. Chlorine losses versus Cl, feed rates using carbon black at ~730°C. Test Cl, feed Cl, losses period (cm®/min) (%) 27 43 2.6 24A 230 1.8 24B 230 4.3 25A 460 2.9 25B 460 33 26A 635 2.2 26B 770 4.3 28A 622 3.3 28B 838 4.0 28C 842 34 28D 1032 4.5 28E 1026 6.8 The higher losses near the end of test MS-28 may have resulted from an unintended depletion of the UQ, inventory. After the UO, feed was ended, the appearance of high Cl, losses and the sharp decrease in CO, concentration indicated only 2 to 3 min (12 to 18 g of UO,) before the inventory was grossly depleted. | These results show little variation in chlorine losses for a wide range of UO, and Cl, feed rates. Using carbon black at 730°C allowed high rates of reaction, but the CI, losses remained =>2% at the most favorable conditions. 3.3 EFFECTS OF URANIUM OXIDE PROPERTIES The feeds to an AVLIS plant will be uranium ore concentrates from refineries. Since a representative sample or a specification of typical properties was not available, the chlorination tests were made with several uranium oxides available at ORNL. Some measurements for these materials are in Table 8. Results of chlorination tests indicated that the chlorination reactions are simpler and operation is better if the feed is predominantly UO, instead of higher oxides. The U(IV) is favorable to formation of UCI, while U(VI) allows excessive vaporizations of uranium as UCl. 24 Table 8. Properties of uranium oxide, UO,-C, and carbon feed solids Surface Bulk Particle area density size Nominal composition Description (m%g) (g/em?) (um) UQ, Ball-milled powder 0.102 4.90 20—150° UoQ, UOQ, gel spheres in H, to 740°C 7.37 1.3 300500 UO,-C UO,-C gel sphere in Ar to 740°C 30.52 0.80 400—800 C/U = 4.2 atom/atom C Petroleum coke 1.34 1.15 50—400° C C-black for UO,-C spheres 96.0° 0.2° 0.03" C Compacted C-black granules c 0.68 100—-1000° *Some finer particles. *Manufacturer’s data. “Probably ~ 100 m%g and 0.03 xm true basic particle size. This effect has been reported for chlorinations using CCl,, COCl, and other chlorination agents. The additions of uranium oxide feeds generally gave bursts of gas pressure that were largest for UQ,, smaller for U;O4, and smallest for feed that was nearly UO,. Most of these chlorination tests have been with UO, feed even though the usual uranium ore concentrates are UO,. Test MS-17 was made using a feed of UO, spheres to the UCl,-carbon black charge remaining after test MS-16. The test was carried out to observe the UO;-melt reactions and the chlorination reactions separately by adding the UO; without Cl, feed, and then starting the Cl, after the UO;-melt reactions approached completion. The simplest and best result would be the reaction of UO, and carbon black to give UO,, and then the same chlorination behavior after Cl, flow was started as for UO, feed. However, the reactions observed for UO, are much more complicated and less desirable than the simple reactions just mentioned. It appears the UQ; also reacts with UC], to form oxychlorides and that the U(VI) feed yields more UCI, than U(IV) feed. It is not certain exactly which reactions occur, but the observed results include the following: 25 1. The amount of CO, evolved before the start of Cl, flow was <15% of the total oxygen added as UO; (or about 40% of the oxygen for reduction to UQO,). 2. The amount of CO, evolved after the start of Cl, feed was much more than that for UO, and confirms that more than 85% of the oxygen introduced as UO; remains in the melt without reacting with the carbon until Cl, is fed. 3. The chlorine utilizations were only in the 50 to 70% range, while.chlorine utilizations for the same conditions with UO, would have been =90%. 4. A plug of condensed uranium chlorides plugged the crucible cross section at the end of the test. This amount of condensed uranium chlorides was not expected based on results with UO, feed at similar conditions. The chlorination reactions did not show any dependence on the physical properties of the uranium oxide feeds. The reactions of the uranium oxides with UCIl, were rapid so that the availability of oxychlorides for oxidation by Cl, and reduction by C or CO did not vary with the physical properties of the uranium oxide feed. During test MS-22, batches of the ball- milled UO, of low-surface area and the uranium oxide gel spheres of high-surface area were alternated as feeds without any detectable differences in results. For either feed, the effects of the addition peaked as quickly as the feed addition was completed; both uranium oxides were available for reaction without any significant delay. The UCl, reacts with UO, to give UOCI,, while higher oxides probably give mixtures of UOCI, and UO,Cl,. The carbon and oxygen balances (Table 6) and the CO, collected per mol of Cl, reacted show consistent effects. The early tests with CO were commonly ended by reactor plugs before oxygen removal was complete. The extra oxygen in U;0, or UO; as compared to UQO, was noticeable. Exposures to air during shutdowns resulted in extra oxygen and loss of chlorine from the reaction of UCl, with water vapor. The material balances as compared to the amounts of UO, feed commonly showed CO,/UO, larger than 1 mol/mol, and Cl,/UO, larger than 2 mol/mol. The excesses over the stoichiometric amounts would be expected from reactions of UCl, with water vapor during handling or from water, hydrocarbon, or oxygen impurities in the feeds. Any water in the feeds during chlorination would give both CO, and HCI that would collect on the first Ascarite trap, in addition to the CO, from chlorination of UQ,. The effects of other impurities in the uranium oxide or the carbon have not been studied. Logically, some impurities would accumulate in the chlorination reactor melt. Other impurities would form volatile chlorides and be transferred with UCI, to the uranium metal product or transferred to the waste purge streams. 26 3.4 REDUCTIONS BY CARBON Use of carbon as the reducing agent has important advantages over the use of CO. An excess of carbon remains in the melt ready for later use while an excess of CO separates and is lost to the exit flow of gases. The excess CO results in a toxic and flammable waste gas. ~The CO can react with excess Cl, to give COCl, (phosgene), which is more toxic than Cl, alone. Carbon feeds are easily stored and are commercially available in several forms while a large feed of CO would require special preparation or storage facilities. The chlorination behavior using C and CO appears to agree with two thermodynamic concepts. An excess of solid carbon has an activity of one for chemical equilibriums. The CO is always diluted by other gases and has activities of less than one for atmospheric pressure. The free energies of formation for CO, and two molecules of CO are equal at 700°C. Because of the activities mentioned above and kinetic effects, some CO is possible from use of carbon at temperatures below 700°C. The chlorination reactions in this study have shown better utilizations of chlorine and higher rates of reaction with carbon as compared to CO. The best steady-state utilizations of Cl, were 30 to 40% with CO feed, but Cl, utilizations were 290% for complete tests using carbon with some steady-state periods near 100%. Temperatures of =>670°C were necessary for high Cl, utilizations using petroleum coke. The higher reactivity of carbon black was most clearly demonstrated during test MS-22. When the reactions appeared to be near steady state with 48 g of petroleum coke in the melt, an additional 2 g of carbon black was added. The rate of CO, formation doubled and the utilization of Cl, returned to nearly 100%. The peak in CO, evolution agreed with the amount of carbon black. A later addition of 2 g of petroleum coke did not give any noticeable change in CO, rate. Another indication of the more rapid reaction of carbon black as compared to the petroleum coke is the more rapid decrease in CO, evolution after the Cl, feed is stopped. With melt temperatures over 700°C and carbon black, the CO, rate decreases almost immediately and rapidly. With petroleum coke, the CO, evolution may taper off over a 20 or 30 min period. This difference indicates that the petroleum coke results in a much higher inventory of U(VI) [or U(V)] that reacts with the coke over the 20 to 30 min period. Rapid formations of CO, during chlorination were demonstrated with three types of carbon. A petroleum coke was used in the form of free-flowing, granular particles that were easy and clean to handle and feed. The frozen charge after cool down showed a floating bed 27 of salt-wetted coke particles. Carbon black dispersed in UO, gel spheres provided fine carbon particles of high surface area and very intimate mixing with the UO,. The frozen change after cool down did not show any separation of carbon in the melt or as dust above the melt (the UQ, dissolved in UCl,). Test MS-1 was made at 850°C with a finned graphite cylinder as the source of carbon. The formation of CO, with little CO in spite of the 850°C temperature indicates a limited availability of carbon for reaction. Graphite shapes are not a practical carbon source for a production process. Either the petroleum coke or the carbon black mixed with UQ, appear more practical. The carbon black gives high reaction rates at lower temperatures, otherwise, the choice probably depends on convenience of use, significance of impurities, and costs rather than the suitability of the compound as a reactant. A fine, dusty carbon black powder by itself might cause feeding problems, but carbon black is available as compacted granules. 3.5 REDUCTIONS BY CARBON MONOXIDE The use of CO has the advantage of providing a highly pure reactant that is easily metered and fed. Excess reactant is easily separated from the reactor charge or from condensed products. However, these advantages are much less important than the high rates of reduction and the other advantages of carbon. With reduction by CO, utilization of chlorine was high for a short period only until U(IV) oxychlorides were mostly oxidized to U(VI). This condition gave much higher volatilities of uranium and poorer utilizations of Cl,. It cannot be determined whether the chlorine escaped from the melt as Cl, gas or as UCl,, The result was mostly condensed UCl, and losses of Cl, to the trap, but the UCl; might decompose to UCl, and Cl, in the condenser. This type of mechanism is discussed more completely in Sect. 3.6. From 75 to 90% of the CO feed left in the exit gases without reacting (Table 6). The limitation of CO can be summarized as follows: Using CO, the rate of reduction is lower, and it is difficult to avoid a charge that is highly oxidized by the Cl,. This scenario is unfavorable with respect to chlorine utilization and the high volatility of uranium as UCl,. With C as the reductant and good (high) concentrations of oxychlorides, the reduction reactions are much faster than reduction by CO so that the charge has higher concentrations of U(IV). This situation is favorable to good utilization of Cl,. The lower U(VI) concentrations using C allow higher temperatures without excessive uranium volatility. The higher temperatures tend to give higher rates for both the chlorination and reduction reactions. 28 One test to expose materials of construction (MS-29) used both carbon black in the melt and CO with the Cl, gas feed. There was no detectable reaction of CO; all the reduction was apparently accomplished by the carbon black. The exit flow of CO did not vary for three Cl, feed rates (including no Cl, feed). 3.6 VOLATILIZATION AND CONDENSATION OF URANIUM CHLORIDES During chlorination tests, major fractions and sometimes essentially all of the uranium is vaporized from the crucible and condensed outside the furnace. Examination of the condensed solids by X-ray diffraction shows the lines of UCl, only without any known lines of other uranium compounds. The published data indicate that the volatilities of uranium oxides and oxychlorides are negligible at the chlorination temperatures. However, the observed amounts of condensed solids are too large to agree with the vapor pressure of UC], (Appendix). The vaporization behavior of UC, in the reactor without chlorine gas is consistent with the UCI, vapor pressure. There was no detectable condensed uranium from molten UCI, without a sparge gas (Test MS-11A). This fact indicates that a thermal convection cell is not a controlling means of vapor transport. With a nitrogen sparge, the amount of condensed uranium is small in approximate agreement with the vapor pressure of UCl, at the test temperature (Test MS-11B). With Cl, feed, the amount of condensed uranium is about five to ten times that calculated from the temperature and saturation of the noncondensable gases with UC], vapor (test 11C and others) (Table 9). The uranium condenses outside the reactor furnace and gives deposits of two distinctly different visual appearances. Both deposits show X-ray diffraction lines for UC], only. The quartz reactor walls inside the furnace remain transparent and free of deposits throughout the chlorination tests. After a sparge flow is started, a haze of fine powder deposits on the quartz walls above the insulation. This deposit becomes opaque immediately above the insulating blocks in a short time and decreases in thickness to a partly transparent film in the cap. This type of deposit remains loose and powdery and can be discharged from a tilted reactor by moderate tapping or jolting. The color of the solids is the greenish-black of UCI, with some yellow-green tints for the deposits farthest from the furnace. The other type of condensed solid is distinctly different in properties and appearance. This type of deposit forms the first 2 in. outside the reactor heated zone. The color is dark 29 N o3ind ourp paoj ton Juipnpug, -soanjesndwo) sovUIng oY) URY) JOMO] D,0Z INOQR 210M $2INJEIDUWD) 291eyn D1LINJ0E 1SOW PIIDPISUCD soouE(eq [eLIEWw 1YTom iim Po1aojos sMol YD pie 40D ‘QD tuonnjoad Q) 10 pud [nun podj i) Jo 1els Wol) smo)j Sen, 0 19°0 ST S0 SLO 00°1 1€L X4 89°0 {(58°0) (SS°P) 00 ov'o S1'Y SIL 890 yeo 08’1 YA oro STl v69 (44 t0'1 669 o1~ 2o 0s'l oga L10 t0'1 LL9 1 870~ ¥80°0 6v°0 910 600 70 <69 0z S0 €10 - LTl vLO 800 S0 00L 81 SI'0 LO0 . ¥6'0 960 . o _ 9’0 089 Ll o000~ 600 o'l 50 200 N4 SL9 900~ 600 o'l 50 900 SY'0 SOL 91 <10 17°0 p0'1 90 0£'0 870 SrL S1 SO0~ 890°0 60’1 69°0 900 ye0 01L vl 00~ 200 w0l 50 ot 0z’o 089 tl 100~ 1000~ 60’1 9¢°0 €70 0£0 089 100~ 1000~ 60’1 960 70 0t’0 L9 4! tl'0 00 90'1 190 €0 10 069 £10 2200 901 19°0 £e'0 (441 $t9 ort €10~ 0r0'0 £6'0 0S0 670 ¥1'0 089 £1'o~ L70°0 €60 050 LYAY Y10 Sv9 Vo1 €10~ wo 9’0 1o $T0 oro Y9 6 124V €10 tL'l Lo 980 090 089 Pyo !0 L'l Lo 980 090 S¥9 8 9000 S000 Lyl 120 950 oLo $S9 L 00 0o 190 1'0 0zo 1£°0 $79 y "puG-2UOU (jouw) (jow) Je10L IN D 0D+ 0D (.) ‘ON 159, PIsUOpUC) Jodea "N almesadwoy wniuer) poenoled (jow) 219101 JO N0 SMOY] ooeuIn,g UOHEZI[IE[OA WNIUEIN PUE SMO[J SeD) "6 9[qeL 30 black-purple with a mirror surface against the quartz wall and coarse crystals on the inside of the reactor. These solids can completely plug the cross section to completely block gas flow. The deposits are dense and strong and fix the gas sparge tube in place so it cannot be withdrawn. The plugs can be broken into chunks by rodding and removed without breaking the quartz reactor. Removal of chunky deposits leaves a clean quartz wall while the powdery deposits of the other type do not. Both types of deposits are found for all chlorination tests, but their rates of accumulation differ. A photograph for MS-8 shows both types of deposits and an almost complete vaporization of uranium compounds out of the crucible and heated zone (Fig. 4). The green- black powder deposits accumulate at moderate rates that increase with temperature in reasonable agreement with the vapor pressure of UCl,. The dark, dense solids have highly variable rates of accumulation. The high rates occur at conditions that give higher losses of chlorine to the Cu trap in the off-gas cascade. The tests with CO as the reductant were most commonly ended by plugs of dark, dense deposits that restricted the gas flow before the intended run plan was completed. Some tests with carbon as the reductant were ended with only thin layers of the dark deposits on the quartz walls. The thin-layer result occurred only when the Cl, flow was stopped as soon as the off-gas flow rates indicated increased Cl, removal in the Cu trap. The vaporization and condensation behavior is consistent with the following explanation. The vaporization of the uranium occurs as both UC, and UCl,. With an excess of carbon and oxychlorides in the charge, the steady-state concentration of U(VI) is low and the amount of UC] vaporized is small. The chlorination gives an "oxidation" of U(IV) to U(VI). If the conditions for the reduction reaction are less favorable, the steady-state concentration of U(VI) and the amount of UCI, vapor increase. Conditions that lower the rate of reduction and thus increase the vaporization of UCl, are: 1. CO as the reductant (CO is less effective than carbon); 2. low concentrations of oxychlorides (inadequate UQ, feed) as C or CO cannot reduce UCl, to U(1V); and 3. lower temperatures as the rates of reduction are more temperature dependent than the rates of chlorination. 31 The vapor pressures of both UCl, and UCl, increase with temperature. Nevertheless, the dark plugs of condensed solids were formed at some tests below 650°C while other tests at higher temperatures gave much less dark solids. The experimental data do not explain how the UC], vapor results in the deposit of dense UC], solids. The thermochemical data indicate that a large partial pressure of Cl, is required to keep UC], stable. Therefore, the UCl; might decompose to UCl, and Cl, as a surface catalyzed reaction. As an alternate explanation, the dark deposits may result from condensation at higher temperatures and higher UCI, concentrations while the powdery solids form at lower temperatures and lower UCl, concentrations. The weight material balances for the large (68-mm-ID) reactor tests have given ten good measurements of uranium chloride vaporization for a range of conditions. The observed amounts of condensed solids have been compared with the calculated rates of UCI, vaporization. These calculations assume a UCI, vapor pressure given by an ideal melt of nonvolatile carbon solids and molten MgCl, and NaCl with the remainder of the charge as UCl,. The N, sparge to the crucible and the CO, product are saturated with UCl, vapor at the melt temperature and composition. The results show agreement with the calculated results for the effects of temperature (i.e., UCl, vapor pressure), the effects of MgCl,—NaCl as nonvolatile diluents, and the effects of N, sparge gas flow rates. However, the relative oxidation-reduction of uranium salts in the melt is also very important. The amounts of uranium vaporized compared to the calculated amounts of UCl, vapor without consideration of this factor have been as follows: a. For no Cl, flows or for very small Cl, flows with good charges of carbon black, the experimental amounts have been less (80 to 100%) than the calculated amounts. b. For the high Cl, and UO, feed rates with good charges of carbon black, the experimental amounts have been 100 to 150% of the caluclated amounts. ¢. For useful Cl, flow rates and petroleum coke in the melt, the experimental amounts were about 300% of the calculated amounts. d. Some tests in the small reactor system with CO as the reducing agent showed 500 to 1000% of the calculated amounts. Chemical analyses of the deposited solids give little additional information. The UC], is so reactive with water vapor (to form HCI) that sampling and analyses without oxygen contamination is extremely difficult. Reaction with O, to release Cl, is also possible, but the 32 rates are low below 100°C. As previously mentioned, x-ray diffraction shows lines for UCl, only without any lines for other uranium compounds. The chemical analyses for chlorine, total uranium, and U(IV) (Appendix B) showed small amounts of U(VI) and more chlorine than needed for UC],, but not enough to give all UCl, and UCI,. It is believed that the difference is oxygen from reaction with water vapor during removal and sampling. In summary, the chemical analyses can be explained by deposits that are mostly UCl, with a small amount of UCI and a small amount of reaction with water vapor to form oxychlorides before analysis. Several of the plugs of dark, dense solids were melted and drained to the charge in order to allow continued chlorination. This operation was only possible when a tall crucible was used with the top extending above the insulation so the dark solid formed inside the crucible. The melting was done by either lowering the reactor into the furnace or by increasing the furnace temperature to over 800°C (this high temperature was necessary to give >600°C in the insulation above the furnace). A supplementary heater was procured for occasional use in melting plugs, but its use was not tested. 3.7 PRELIMINARY RESULTS WITH A LARGER REACTOR AND BOTTOM CONDENSER After Test MS-18, the chlorination reactor was replaced with a reactor of larger (68-mm) diam designed to give a downflow of vapor and an additional flow of nitrogen diluent gas to a condenser at the bottom of the unit. The salt charge was in a crucible of 5.6 cm ID. The schematic flow sheet (Fig. 5) shows only small changes from that for the 4-cm diam reactor, but some flowmeters and reactors in the exit gas train are also larger. This larger system was operated to provide information on the condensation behavior of the uranium chlorides. The test conditions and results are included in the tabulations for the small reactor (Tables 1 through 6). Tests MS-20 and MS-21 were with a charge of UCl, and C (petroleum coke particles) and UO, feed. Test MS-20 with 0.21 mol of UO, and moderate Cl, rates showed a steady accumulation of very dark gray-green solids falling to the bottom of the condenser chamber. Removing the bottom cap dropped 58 g of solids into a plastic sack (0.15 mol UC], and an estimated 0.02 mol remaining in the condenser). Test MS-21 was made with 80 g/h of UQ, 33 "JOSUSPUOD WONOq Yum ((] ww-g9) 103108l uoieuloy) s S -t CNOAn l Y [ 3OVNENS 1SNVHX3 dooH L0¥1—-¥06 OMO INHO > —13Mm d313N 1S3l 00 _m— vall oA —+—1w 4ILINONVYIN %09 104 D m van [3ovnEnd] 1 o SYILINMOTd SSYAN 94 OL 4 SHILIAVIOY 94 O1 | 431714 o\ A201 334 34 feed and gave 236 g of solids in a collection bottle below the condenser (0.62 mol UCl,). The test went well with the exception of two plugs of feed lines for the UO, feed. The average Cl, utilization was 95% even though the chlorine feed was twice continued until the CO, showed a significant drop off from depletion of oxide from the charge. After Test MS-21, the crucible was removed and the apparatus cleaned to obtain weight material balances for MS-20 and MS-21 together. The measured MS-20 and MS-21 results arc: 1.47 mol UCI, in the initial charge; 4.5 mol C (petroleum coke) in the charge; 1.19 mol UO, fed to crucible; 1.27 mol CO, on traps; 2.8 (weight) or 3.3 (feed flowmeter) mol total Cl,; 0.26 mol Cl, on Cu-foil trap; >90% Cl, utilization; 0.77 mol UC], into product receivers; 0.38 mol UCI, released from the reactor walls by a wiper blade treatment; 0.09 mol UCl, remained on reactor walls; 0.12 mol UCI, deposits on cap, sparger tube, and feed tubes; and ~ 1.3 mol UCl, in the crucible after MS-21. This discharge of UCI, powder as noncaking, flourlike particles shows a partial demonstration of the pilot plant product collection concept. Over 60% of the vaporized uranium discharged in this way, and 40% deposited on the apparatus walls. Part of the deposits were hard, dark, and difficult to loosen from the quartz wall. The 1.27 mol CO, compared to 1.19 mol UO, shows the effect of oxide impurities, water, or incomplete reduction of uranium oxides to UQ,. Test MS-22 was like test MS-21 with the following changes: additional charge to give about 10 cm depth of melt in the crucible; addition of MgCl, and NaCl to give a charge composition (before reaction of UCI, with UO,) of 1.7 mol UCl,, 0.21 mol MgCl,, and 0.21 mol NaCi; a large purge flow of nitrogen to the UO, feed line and a reduced nitrogen purge flow into the reactor top; and several different UO, feeds to observe the differences in exit gas rates and composition vs the feed characteristics. 35 Test MS-22 gave results that are favorable with respect to steady-state operation. The chlorine feed was maintained for 4 h at 0.01-mol Cl,/min and an overall chlorine utilization of 82%. Flow differences indicate an initial chlorine utilization of near 100% for 1 h. After the flow differences showed a decrease in chlorine utilization, a feed addition containing 2 g of carbon black gave a short period of doubled CO, evolution and a return to near 100% chlorine utilization. These increases indicate that the conversion was limited by the reaction of the petroleum coke and that the addition of 2 g of carbon black to the charge containing 48 g of petroleum coke more than doubled the rate of the reduction reaction. After the chlorine utilization had again dropped, an additional 2 g of petroleum coke did not yield a significant change (<10%) in the rate of CO, evaluation. The nitrogen purge to the UO, feed line eliminated the plugs or deposits that occurred in Test MS-21 without a purge. There was no observable difference in reaction results for dense UO, of low surface area (ball milled), as compared to porous UQO, of high-surface area (gel process spheres). The condensed uranium was again divided about 60% as discharged powder and 40% as deposits on the reactor walls in agreement with tests MS-20 and MS-21 results. The moles of condensed uranium was 65% of the feed UO, as compared with about 115% for Tests MS-21 and MS-22. The lower amount of uranium vaporized agrees with previous results showing less volatility of uranium with MgCl, and NaCl in the charge. The deposit of dense dark crystals on the reactor wall at the low end of the furnace was much smaller for Test MS-22. Test MS-23 was intended to test a more steady-state operation using a combination of UO,-C spheres and UO, powder to give a feed UO,/C ratio of 1 mol/mol. Test MS-23 run schedule was completed as planned with the exception of a severely restricted gas flow inside the quartz reactor just after the chlorine feed was shut off. A quartz baffle with nitrogen flow to the reactor top completely eliminated the deposition of condensed uranium above the furnace. The total UO, added and reacted (0.15 mol prerun and 0.96 mol at temperature) was greater than any of the previous tests. The quartz crucible probably cracked during heatup and leaked part of the charge into the bottom region of the reactor. The frozen melt probably obstructed the vapor flow to the condenser and affected the other test results. Tests MS-24 through MS-30 were intended to provide information important to the design of pilot-plant chlorination equipment. A series of tests were completed to determine results for short-term steady-state operation with increasing rates of UO, and Cl, feed (see Table 7). The Cl, feed rates were up to 770 cm*/min (34 mmol/min) for test MS-26, and 36 1030 cm®min (46 mmol/min) for test MS-28. Test MS-28 demonstrated a continuous screw feeder for UO, powder, while all previous tests were run with periodic batch additions of UO,,. The primary objective of tests MS-27 and MS-29 was to expose three proposed materials of construction (carbon composites) to the chlorination reactor conditions. The MS-27 and MS- 29 results also give chlorine utilizations at low chlorine feed rates for comparison with the higher rates. The vaporization and condensation behavior of the uranium chlorides was observed for a wide range of Cl, feed rates and some variations in the gas and vapor flow configurations. The most favorable reaction conditions of carbon black in the melt and melt temperatures over 700°C were used, with the exception of a lower temperature for MS-30. Many of the MS-24 — MS-30 results are included where appropriate in Sects. 3.2 — 3.5. The tests of material specimens were part of a larger investigation and are reported elsewhere.’ A primary criteria for design and operation of the larger reactor system was to provide a separate condensation and measurement of the condensed uranium compounds. The simplest model for the vaporization behavior is that the gases leaving the melt are saturated with UCI, at the melt temperature. The results for MS-4 through MS-18 commonly show much larger amounts of volatilized uranium—often five to ten times the amounts calculated from the UC], vapor pressure. These results are discussed and an explanation suggested in Sect. 3.6. The results for MS-20, -21, and -22 with petroleum coke show condensed solids about three times the amounts calculated from the vapor pressure of UCl,. The utilizations of Cl, and the results from a small addition of carbon black in MS-22 indicate a need for better reductions of U(VI) [or U(V)] to U(IV). By using carbon black in melts at >700°C, the amounts of condensed uranium were much closer (typically 120%) to that calculated for the UCI, vapor pressure. The high chiorine utilizations (see Sect. 3.2) are consistent with this change in vaporization behavior. Discharge and collection of a UCl, powder are highly desirable to avoid the problems from deposition of UC], solids or the problems of collecting a UCl, liquid product. The proposed plant concept is to use a quench gas recycle to produce solids below the UCI, triple point of 15 mm Hg, 590°C. Our reactor system does not allow the large gas flows necessary for a valid test of this concept. The MS-20 to MS-30 results show 48 to 73% of the condensed uranium as powder into the product jar and 27 to 52% deposited on the system walls. The two most important effects on how the condensed solids collected appeared to be: *To be issued as an AVLIS program milestone report approximately September 1991. 37 1. Turbulence in the condenser results in deposit of condensed UCl, on the walls and reduces the fraction that discharges to the product jar. A quench gas inlet into the condenser gave poor results. 2. More dilution of the UCI, vapor increases the fraction of solids discharged, but the dilution must be done without turbulence in the condenser. The highest Cl, feed rates gave the highest UCl, vapor concentrations (230 mm of Hg calculated) and deposits that completely closed the condenser cross section. 4. CONCLUSIONS High rates of chemical reaction and good utilizations of Cl‘2 were demonstrated for reaction of UQ,, Cl,, and C to produce UCIl,. A molten salt medium containing UCI, allows dissolution of UQ, by the reaction UO, + UCl, - 2UOCI,, and this is favorable to high rates of reaction. Starting with UQ, dissolved in UCl,, the initial rates of reaction of chlorine can be very high with low concentrations of chlorine in the exit gas. The rates of reduction reactions to remove oxygen as CO, or CO control the overall rates of UCl, formation. With carbon black in a melt at about 730°C, the rates of CO, evolution matched the Cl, rates for Cl, feed rates up to 0.046 mol/min in a crucible of 5.6-cm ID. Less favorable conditions for reduction give lower utilizations of Cl, at steady state and greater volatilization of uranium—probably as UCl; or UCl;. Carbon black gave practical rates of reaction and good results for melt temperatures down to 645°C. Petroleum coke of low- surface area gave poorer results at 730°C than carbon black at 645°C. With CO as the reducing agent, the steady-state utilizations were <40% for Cl, and 10 to 25% for CO. The uranium oxide should be reduced to approximately UO, before use as chlorination feed. The results indicate that the chlorination reactions are simpler and operation is better for UQO, as compared to U0, or UO,;. A test with UO, feed gave lower chlorine utilizations and much more condensed uranium than for UO, feed at similar conditions. The physical properties of the UQO, feed do not appear to be important; UO, feeds of low- and high-surface areas did not show any detectable differences in results. The amounts of condensed uranium were compared with calculated rates of UCI, vaporization. The calculations assume that the N, sparge to the melt and the CO, product are saturated with UCl, vapor at the melt temperature and composition. The results showed agreement with the calculated results for the effects of temperature (i.e., UCl, vapor pressure), 38 the effects of MgCl,-NaCl as nonvolatile diluents, and the effect of the N, sparge gas flow rate. However, the relative oxidation-reduction of uranium salts in the melt is also very important. The experimental amounts were from 80 to 150% of the calculated amounts with good reducing conditions (carbon black or no Cl,), but were from 300 to 1000% of the calculated amounts for chlorinations with petroleum coke or CO reducing agents. The experimental system used here did not allow a valid duplication of a pilot-plant collection system for UCI, vapors. From 48 to 73% of the condensed solids were collected as a powder in a product jar below the condenser; the other 27 to 52% deposited on the apparatus walls. The deposits on the walls were partly powder and partly dark, dense crystalline solids that formed more rapidly when the Cl, feed rate exceeded the rates of reduction reactions. Examination of both types of solids by X-ray diffraction shows the lines of UCI, only without any lines of other uranium compounds. Chemical analyses for uranium and chlorine indicate mostly UCI, with small amounts of UCl,, and probably some oxychlorides from reaction with water vapor during sampling. Control of the UO, feed rate to the chlorination melt is very important to dependable, steady-state operation. A deficiency of UO, results in poor utilizations of Cl, and much higher (excessive) volatility of uranium. The UO, feed reacts rapidly with UCl, to give UOC,, which has a limited solubility in the molten chloride salts. Excessive UO, can solidify the charge so that the gas flow is obstructed or channels with poor contact. Carbon particles are wetted by and remain suspended in the melt; a wide range of carbon black inventories is acceptable. When the reduction conditions are favorable, all reaction rates are determined by the Cl, feed rate. Overall, good results were demonstrated for the chemical reactions to produce UCI, from UO,, Cl,, and carbon black. This result is much different than the results of an earlier study to prepare uranium metal by electrolysis of UQ, dissolved in fluoride salts. There, chemical reactions other than the desired ones always resulted in reductions of UF, in addition to UO, and in low current efficiencies. The engineering problems for UC], preparation (materials of construction, process control, UO,-C, and Cl, feed procedures, UCl, condensation and removal) can be addressed in pilot-plant systems with confidence in the chemical flow sheets for the process. 10. 11. 12. 13. 14. 39 5. REFERENCES Phillip G. Sewell and Norman Haberman, "AVLIS Program Powers Ahead in the United States,” Nucl. Eng. Int., October 1988. Lawrence Livermore National Laboratory and Martin Marietta Energy Systems, Inc., "Selection of an AVLIS Uranium Feed Process for Large-Scale Demonstration,” 1.-12038, September 1990. Joseph J. Katz and Eugene Rabinowitch, The Chemistry of Uranium, McGraw-Hill, 1951. R. G. Canning, "The Production of Uranium Metal Powders by Electrolysis in Molten Chlorides," Australian Atomic Energy Symposium - 1958, June 1958. W. L. Lyon and E. E. Voiland, The Preparation of Uranium Dioxide from a Molten Salt Solution of Uranyl Chloride, HW-62431, October 1959. T. A. Gens, Laboratory Development of Chloride Volatility Processes for the Recovery of Uranium Directly from Spent Rover Fuel or from its Combustion Ash, ORNL-3376, June 1963. A. R. Gibson et al., "Processes for the Production of Uranium Oxide,” U.S. Patent 3,117,836, January 14, 1964. Y. M. Sterlin and V. V. Artamonov, cited by E. M. Levin and H. F. McMurdie, Phase Diagrams for Ceramists: 1975 Supplement, The American Chemical Society, Columbus, OH, 1975, p. 396. J. C. Mailen et al., Laboratory Studies of the Production of Uranium Chlorides from Uranium Oxides, ORNL/TM-11609, Oak Ridge National Laboratory, October 1990. M. W. Chase et al., JANAF Thermochemical Tables, Third Edition, American Chemical Saciety, American Institute of Physics, and National Bureau of Standards, 1986. M. H. Rand and O. Kubaschewski, The Thermochemical Properties of Uranium Compounds, Interscience Publishers, 1963. E. H. P. Cordfunke et al., Thermochemical Data for Reactor Materials and Fission Products, Eur.-Contractno., ETSN-0005-NL, 1988. David Brown, "Compounds of Uranium with Chlorine, Bromine, Iodine," Gmelin Handbuch der Anorganischen Chemie, Vol. C9, 1979, Josef Krahe and Franz Miiller, "Zur Thermochemie der Stoffsysteme U, Th, Pa, C, O,, CL," Institute fiir Chemisehe Technologie - Jul 565 - CT, December 1968. 41 APPENDIX A. THERMOCHEMISTRY 43 A. THERMOCHEMISTRY The use of thermochemical data does not identify the probable reactions with any certainity or degree of confidence. There are several major causes of uncertainty. Uranium chemistry is complex with stable valences of 3, 4, 5, and 6, and with stable oxychlorides. UO,Cl, and UOCI, are well-known compounds and others are possible. All the possible products must be considered. The uranium chlorides are more volatile with increasing valence, but UCl; and UCI, are less stable with increasing temperature and decreasing Cl, partial pressure. All the uranium compounds have large heats of formation, and calculating the free energy of these reactions usually results in a small difference from two large numbers. Small percentage uncertainties for the large numbers give large uncertainties for the differences. Thermochemical data (Table 10) can be used to make calculations for chlorination reactor conditions.'®!? A temperature of 700°K (427°C) is about the lowest temperature of interest for practical rates of reaction, and the 1100°K (823°C) is above the boiling point of UCl; and is near the highest practical temperature. The relationships between the different U-O-Cl compounds can be illustrated by a matrix listing‘ (Tables 11 and 12). Data are available for heats of formation at 298°K for all the compounds (Table 11). The free energies of reaction of 900°K are more useful for the calculations (Table 12), but the data are much less complete. Table 12 shows estimated values for some of the compounds and one of several different published values for UO,Cl,. Tables 11 and 12 appear to show that the oxychlorides are more stable than either the chlorides or oxides. The free energies of formation show that either the U(VI) chloride or oxide are easily reduced to the corresponding U(IV) compounds, but the U(VI) oxychlorides are more difficult to reduce to U(IV) compounds. Since the oxychlorides are probable intermediates for conversion of uranium oxides to uranium chlorides, a conversion may appear favorable overall, but one of the steps involving an intermediate may be much less favorable. For example, consider the following: UO, + 2C + 2Cl, -» UC], + 2CO. At 900°K, AG = -246.5 kJ, but individual steps show: UO, + Cl, » UOClL, AG =-1742; UO,CL, + C » UOCL + CO, AG = 13.6; UOoCl, + Cl, - UOCI, AG = -119.8; and UOCl, + C - UC], + CO, AG =341. 44 Table 10. Thermochemical data Melting Boiling Free energy of formation, -AG® Ref. point point (kJ/mol) No. for Compound (°K) (°K) 700°K 900°K 1100°K data 172. 238 0 0 0 10 Cl, CCl, 250. 350 -1.7 -28.3 COodl, 169. 281 187.1 177.8 CO, 216. 195 (subl) 395.4 395.7 CcO 74. 81 173.5 1914 C, — >4000 0 0 Udl,, 1115 — 741.8 699.1 UuodCl - - ~T715. — UCl, 863. 1065 847.8, 794.5, UuoCl, a — 943 900 U0, 3150 - 963.5 930.8 UCl 600 800 845.3 ~780 UOC], - — - — UO,Cl — — — — 1/3[U304] d-1300 — 1038.0 998.1 UCl, 452 ~550 841.3 789.3 UQod, ~ — UO,Cl, >1050 1130 1105 UO, a — 1043.0 992.9 *Decomposes. 45 Table 11. Heats of formation for U-O-Cl compounds at 298°K - AHP° in kJ/mol U Reference No.: See Table 1 for Compounds Valance All Cl One O Two O >Two O 0 U U U U 0 0 0 0 3 Ucl, UOdl 893.32 947.3 4 UCL, UOC, Uo, 1051.4 1087.8 1083.7 5 UCl UOodC, UO,Cl 1094.1 1185.7 1188.3 5.33 (U0,),Cl, U,;04 or 5.5 1202.5 1191.2 6 UCl, UOC], Uo,Cl, UO, 1132.6 1238.5 12473 1225.9 *Values of 860 to 880 kJ/mol are also reported. Table 12. Free energies of formation for U-O-Cl compounds at 900°K (627°C) - AG in kJ/mol U Valance All Cl One O Two O >Two O 0 18] U U U 0 0 0 0 3 UCl, UoCl 699.1 ~ 780 est. 4 UCL, UOd, Uo, 794.5 900.2 930.8 5 UCl, UOdl, UO,Cl 780 ~ 1000 est. ~ 1000 est. 5.33 (U0,),Cl; U304 or 5.5 ~ 1050 est. 998.1 6 UCI UOCI, UO,Cl, U0, 789.3 ~ 1020 est. 1105 992.9 46 These numbers indicate that the reactions could stop at UO,Cl, or UOCI,. If using CO to give CO, as the product is considered, then the two reduction reactions change to: UOoCl, + CO - UOC], + CO, AG = 0.5 and Uuod, + CO - UCl, + CO,, AG =21.2. The uncertainties for A(G® values of UO,Cl, and UOC], are probably larger than either of the above AG values, so it is difficult to determine whether the reactions are thermodynamically favorable. The more favorable calculation for CO as compared to C may also be misleading. The C would be present as a solid with an activity of 1, while CO would be mixed with other gases and would have a lower activity for 1 atm total pressure. Many of the reactions to change between the U-O-Cl compounds are shown in Fig. 1. Large values for the negatives of the free energies of reaction (kJ/equiv at 900°K) show reactions that are thermodynamically favored. Negative values in Fig. 1 indicate that the reactions are not thermodynamically favorable. The data for the oxychlorides are uncertain and values ranging from -20 to +20 kJ do not justify predictions. The formation of UCls and UCl; appears to require excesses of Cl, and will not be complete. Otherwise, all additions of Cl, are favorable and oxychlorides should add chlorine to give UO,Cl, or UOCI,. The reduction of UO, or U,O; to UO, by CO are highly favorable. The reduction of UOCIL; by CO to give UCl; is very unfavorable. The other three reductions of oxychlorides by CO give free energies of reaction that are too near zero to justify predictions. A practical preparation of UCl, probably requires that the reductions of UO,Cl, and UOCI, are possible as they would otherwise accumulate as stable products. Some additional possible reactions not shown by the diagram include: udl, + U0, -» 2U0Cl, AG = -74.7kJ; UCLs + UO,Cl - 2UO0CL;, AG = ~-220 kJ; UCl + UO,(Cl, - 2U0Cl,, AG = ~-145 kI, ucl, + U0,Cl, -» 2U0CL,;, AG = -100 kJ; UCl, + UO, - UOCl, + UCl,, AG = -179 kJ; and UCk + UOCl, -» UOC], + UCl,, AG = -125KkIJ. The first five reactions indicate that the uranium chlorides will react to give oxychlorides. Krahe listed vapor pressure equations as shown in Table 13.* Calculated values from these equations are shown as Fig. 6 The decomposition of UCls (or UCl) into Cl, and UCl, must also be considered as the UCI or UCl, are only stable when excess Cl, is present. 47 Table 13. Vapor pressure equations® Temperature Compound A -B -C (°K) UcL (s) 19.224 15,760 3.02 298 - 1110 UCk (1) 24.044 14,340 5.03 1110 - 1950 uay, (s) 20.329 11,350 3.02 298 - 863 ucy, () 26.079 9,950 5.53 863 - 1062 UCk (s) 21.810 7,450 4.03 298 - 600 uc (1) 26.027 6,210 6.29 600 - 800 Uc (s) 22.317 4,765 5.03 298 - 453 ucy (1) 26.120 4,060 - 7.04 453 - 650 alcz»gp,m,,,=A+B/I‘+ClogT ORNL DWG 81A-~-11 1000 . | , | , , 800 600 400 200 | UClg 100 | UCls 80 60 | 40 20 A mp: 179°C A mp: 590°C VAPOR PRESSURE (mm Hg) A mp: 327°C UCls 4T >800°C | 2 i 1 ] i I 1 100 200 300 400 500 600 700 800 TEMPERATURE ( °C) Fig. 6. Vapor pressures of uranium chlorides. 49 B. INDIVIDUAL TEST RESULTS Some detailed results are given here as examples of the experimental studies. Some tests are not mentioned to minimize duplication or for the following reasons: ® MS-1, MS-2, MS-3, MS-4—details are in a preliminary report.? ® MS-5, MS-9B—mechanical failures; therefore, no useful chlorination results. ® MS-6, MS-19—training and calibration tests without any uranium or salt charge. e MS-11A, MS-11B—volatility tests without chlorine feed. B.1 MS-8 Test MS-8 was made with a charge of 103.5 g UCl, and 50.0 g of UQO, and other conditions similar to MS-4. The Cl, flow was started with an estimated sparger immersion of half of the melt depth as greater immersions resulted in zero gas flow. A good flow of CO, was indicated in 10 min in agreement with the system holdup times. This CO, evolution peaked in 30 min and tapered off to about half the peak value by 60 min. The CO, evolution rate continued for 3 h with only small effects from increased immersions of the sparger, a 30°C temperature increase, and changes in Cl, feed rate. After 4 h of Cl, feed, the chlorination reactor plugged and would only pass about 20% of the intended gas flow even though the feed gas pressure was increased to 6 psig. Disassembly of the cold system showed a complete closure of the reactor cross-section by a dark crystalline dome of solids above the furnace, but below the cooling jacket (Fig. 4). The sparge tube and the thermowell were frozen in position by the solids. The upper half of the solid deposits were a loose powder instead of dense crystals. The crucible and reactor bottom only contained a few crumbs of solids. A simple plug flow of gases that are saturated with UCl, cannot account for the transfer of uranium solids. The total amount of gas feed was about 1.6 mol to give a calculated UCI, concentration of 20 to 25 mol % in the vapor. The UCI, vapor pressures are reported to be 60 and 110 mm Hg at the two run temperatures of 650 and 680°C. The rate of CO, evolution showed little variation with run variables (temperature, sparger immersion, Cl, flow rate, crucible charge depletion). The CO rate was approximately constant throughout the test. 50 The results of the MS-8 material balances are: Weight measurements show: Feed UCl, 103.5 g, 0.272 mol U, 0.544 mol Cl.. Feed UO, 50.0 g, 0.185 mol U, 0.370 g-atom O,. Cu Trap 42.2 g Cl, or 0.592 mol Cl,. 1st Ascarite 8.7 g CO, or 0.198 mol CO,. CuO -5.9g O, or 0.369 g-atom O,. 2nd Ascarite 16.2 g CO, or 0.368 mol CO, or Co. Chlorination 15.3 g weight gain or 18.7 g Cl, (allowing for O, loss) Reactor to give 0.264 mol Cl, reacted. Gas flow rates times concentrations show: About 0.45 mol CO feed (vs 0.57 mol CO, on Ascarite traps), About 0.9 mol C}, feed (vs 0.86 volume from weights), 0.23 mol CO, (vs 0.20 on trap), and 0.45 mol CO (vs 0.37 on trap). Differences in flow rates show: 0.42 mol CO, a good check; 0.22 mol CO,, a good check; 0.68 mol CO + CO,, a reasonable check; and From 0 to 0.6 mol Cl, to CU trap, depending on which flow measures are used. Chemical analyses of product samples show: 97.2% of feed uranium, 110% of calculated Cl,, and 19% of calculated O,. Calculated compositions of: Dark crystalline Powdery solids, mol % solids, mol % udi, 93.8 81.4 Uo,Cl, 0.9 10.8 UClg 5.3 7.8 Examination of the two solids by X-ray diffraction show only UCl, with no indication of other uranium compounds. 51 B.2 MS-22 Material was added to the crucible removed after MS-21 (541.7 g of charge, 7-cm depth) as follows: 20.0 g MgCl,, 12.0 g NaCl, 12.0 g C (petroleum coke), 60.0 g UO, (ball-milled powder), and 42.5 g UCl, (the condensed MS-21 wall deposits). The reactor furnace was controlling at 685°C at 10:30 a.m., but the sparger appeared to show solids and signs of plugging when lowered to less than 2 cm from the crucible bottom. The Cl, flow was started at 10:55 and 20 g batches of UQO, feed added at 11:35 and at 15-min intervals thereafter with some omissions. The sparger was lowered to the crucible bottom at 11:18 a.m. with no signs of solids or plugging. The evolution of CO, is shown by the recorder chart for the CO, concentration '(Fig. 7) and were confirmed by post-run gas flow calculations (Table 14). After 1 h with little or no flow differences for flows in and out of the Cu trap for Cl,, these flows appeared to show chlorine losses. At 13:10, 10 g UO, — 2 g carbon black was added. This resulted in a very high rate of gas evolution that overpressurized the gas train and showed about triple the previous CO, rate. This high rate decreased back to the base rate in about 20 min. The remaining UO, feed additions and an addition of 2 g of petroleum coke were completed without any noticeable effects on the CO, concentrations or rates. About 45 min after the last UO, feed, the CO, rate decreased sharply and the Cl, losses increased, indicating a depletion of oxide from the charge. The Cl, feed was stopped and the CO, rate decreased to a low value in about 30 min. After cool down, the weights showed: +31.2 g or 0.44 mol Cl, on the copper trap; +47.9 g or 1.089 mol CO, on the 1st Ascarite; ~0 g or ~0 mol CO oxidized by the CuO; +0.8 g or 0.018 mol CO, from CO on the second Ascarite trap; 134.1 g or 0.353 mol UCI, powder in the product jar; 66.1 g or 0.174 mol UCI], collected by scraping the reactor walls; 0.29 g or <0.001 mol UCl, on the filter; 838.4 g of charge in the crucible as about 10.5 cm depth of melt; and 37 g or 0.097 mol of UCI, collected by washing the reactor, the cap, sparger, and feed line. CLOCK TIME, h AND min 52 ORNL DWG 91-405 N,FEED 270 ¢cm3¥min THROUGHOUT STARTED 10 g UO,-2 g CARBON 30 40 50 60 CO, CONCENTRATION, % «— SPARGER DOWN 3 ¢m (TO BOTTOM) F1515 «— Cl, FEED OFF p —1.0 g U0 ,-0.2 g CARBON BLACK - 1445 = 2 b a5 <«— 10th UD, FEED (20 g) - «— 2 g PETROLEUM COKE <—9'" YO, FEED (20 g} L 1345 «—8'" UO, FEED (20 g) . «— 7" YO, FEED (20 g) 2 1515 <«—— COMPLETED ! | | | 0 10 20 - 5'" U0, FEED (20 g) - 1245 N «— 4'" U0, FEED {20 g) | | | 1| 1 _ 0 30 68 118% 180 270 405 CQ, cm®/min 1215 <— 3" U0, FEED (20 g) L1145 "« 2% UO, FEED (20 g) <—— 1"'U0, FEED (20 g) - 1115 Cl, ON AT 1055 218 cm>/min /MS-ZZ 60 g UO, 50 g C{COKE) IN CHARGE Fig. 7. CO, concentrations during MS-22. 53 (uww/ o) (unu £$7) 211 79¢ - €LE (unw £T) L1T 0LT $9¢T S9C moy Ay sed JnJj 08L'LT - 0I¥'E - $09'tS 000901 | 089€0T | 089°€0l 0} PaNAII0D 950°8E 9EL'TPT v10'y 0sL'svl $90't9 000901 | 089601 | 089°€0l 16€ unJ 20UM 1£€1 015'T 06¥'11 09§ 050'C1 006'1 01Z‘8 0666 086'8 €0S1 330 D 9¢ Syl 0cT'E 0L9'6 008 0Ly 01 £8L'S 0L0'9 01€9 0SH9 €2 eyl 091'y 0Z0'€1 T8¢ 0LSEl 0r1's 02’6 0£8°8 0988 $p33J ‘0N OML 7€ 00v1 oTL'y 062C1 ) C8T 2l ovs'L 001'8 0128 oL1'8 $P33J ‘0N oML Ot otel 0sS'1 0SLE 0 00L°€ 0L0'C 000~ 0EY'T 00Z'T 6 12€1 0L0'Y 015’6 ove 0SL'6 orE'S 00S'S~ 008'S ovv's $S0] sed “0-“0N 1T ootl ort'y oLYCl OLL €l or6'L 001°8 08C'8 0£0'8 $pa3J ‘0N oML 0t 0gCl 020°S 0L9°ZI 09 0El'El 00S'L 099'L 065 L 089'L P33 ‘0N MO 67 1021 0897 056'C1 0sy 00v'€1 020'8 081°'8 0£0'8 0.7'8 $pa3J ‘0N oML 1€ 0El1 0€L'1 0£S'S 0 ory's 0L8'E orl'y 0l6€ 008’ <1 SITL 0001 00L'S 0 or9's 018°¢ 089'y 066'€ 00L'Y <1 0011 0§ OvE ¥ 0 oved 0LT'l 00S'¥ 010'% 06C ¥ §SO1L o D S SpOT 0921 0.T'8 0 000'8 0 026'L 056'L 010°L D oN 0 101 0£¢C 0£0°0T 008 0£8°0C 0 000°t! 0SE8I 00861 dmesf] St 0060 - o'l - L1’ 121 0z’ 10°1 00’1 00 + 00 nd uo ) 159} 1om ‘0D + 00 +°N S-Wd 90 + %00 + 00 +°N ‘0 02 +°N ‘N ‘N snurw - S-WA snui WA WA 1-Wd 9- WA TIWd | 1591 M S)UAWWOD fifi_v salun 301D $10)0B] J3)dUl pue ‘sed ‘SI9PW AINb3 N JO (WD “SmOf sed paiegnofe)d s3urpeal 19z1je)0) wolj suone[noes moy sed ZZ-SW v1 Sl9EL 54 The flowmeter totalizer readings and the average of pre- and post-run calibration factors were used to calculate flows for fourteen intervals (Table 14). There were some losses of gases to allow breakup of a cake on the top of the 1st Ascarite absorber at 12:10 p.m., and from the overpressurization at 1:10 p.m. B.3 MS-28 Test MS-28 was made using a continuous feeder for UO, powder and Cl, feed rates of 620, 840, and 1030 cm®/min. This feeder is a glass and Teflon unit with a screw of Teflon flights inside a glass tube and a closed, gas-purged hopper. The screw uses an O-ring seal and gives controlled and reproducible powder feed rates using a variable-speed drive motor. The powder bridged in the hopper above the screw several times, but the bridges were casily seen through the glass and were broken by moderate taps on the hopper wall. Feed was interrupted twice by plugs at the end of the feed tube in the chlorination reactor. These plugs were hard, tight solid cylinders in the last 0.5 cm of the feed tube and appeared to result from reaction of condensed UCl, with UO,. After the second plug, the feed tube end position was changed to about 1 cm into the crucible instead of 4 cm; the test was then completed without further feed difficulty. The tube end was partly restricted by solids when removed after cooldown. Test MS-28 was made in three parts as a result of the two replacements of plugged feed lines. Nitrogen purge gas was lost during these replacements and through a leak early during the test, but all other material balances were excellent. The weight balance showed 0.1% loss, and the weights were confirmed by the flow rates times concentrations and by the flow differences. The overall results included: U0, feed 521 g (1.93 mol) Product jar 390 g (1.03 mol) Reactor wall deposits 432 g (1.13 mol) Cu trap 22.0 g (0.310 mol Cl,) Other traps 103 g (2.34 mol CO,) Total Cl, feed 316 g (4.46 mol Cl,) Average Cl, utilization 93.1% The reactor wall deposits included a nearly complete closure of the reactor cross-section by deposits between 2 and 5 cm below the bottom end of the furnace. This is inside the end insulating brick and just above the cooling jacket. The deposits were about 50% removed as powder by moderate scraping, 35% hard, dense chunks removed with difficulty, and 15% as thinner films removed by water washes. The product jar collection was 53% of the UO, feed or 48% of the condensed solids. 55 C. EQUIPMENT DETAILS AND CALIBRATIONS This equipment used quartz as the material of construction for hot chlorine gas. The components in contact with chlorine gas at room temperature were also selected to avoid excessive corrosion. Otherwise, the system was a series of small, special chemical reactors and commonplace types of laboratory flowmeters, temperature measurements, and control instruments. C.1 Flow Control and Measurement Feed flows of gases were set manually using needle valves and variable area flowmeters with ball floats. The primary flow measurements for material balances were the totalizer readings from Hastings mass flowmeters. One-time readings of indicators were not accurate for material balances because of delayed effects of both melt and gas inventory changes, a resonance-type oscillation of exit gas flows, and drifts of feed flows between flow adjustments. The flowmeters were calibrated with N, having one primary calibration at the ORNL instrument shop, and secondary calibrations against the wet-test meter just before or after experimental tests. The Hastings mass flowmeters exposed to Cl, (F1, F4, F5) showed continuing changes in calibration. The long-term change was to require an increasing multiplier to convert the indication to a true flow, but there were also some short-term decreases in multipliers. The calibrations remained linear; the multipliers were time dependent, but not flow dependent. The manufacturer’s factors were used to convert the indicated nitrogen flows to other gases. These were the following (air = 1.00) for the mass flowmeters. Gas Conversion Factor N, 1.02 CO 1.00 CO, 0.73 Cl, 0.85 The material balances were based on the N, equivalents being additive. For example, a combined flow of one liter each of N,, CO,, and Cl, would show as 1(1.02/1.02) + 1(1.02/0.73 + 1(1.02/0.85) = 3.60 L on the totalizer. The variable area flowmeters were convenient for short-term adjustments of feed gas rates. Their point indications had the same limitations as the mass flowmeter indications, and were of limited usefulness in two other ways. The ball flows gave very non-linear flows and did not give simple factors to convert for different gas compositions. The meters were frequently fouled or "sticky" from chlorine corrosion products or solids entrained in the exit gases. C.2 Product Gas Reactors and Traps Reaction and removal of individual components of the product gases by fixed beds of solids was the basis for all three types of material balances. The weight changes of the fixed beds showed the amounts of material reacted. The removal of individual components resulted in flow differences that were a measure of the amount and also allowed use of CO, and CO concentration measurements. The reactors were designed to minimize the total weights in order to give weight material balances to 0.1- or 0.01-g precision using laboratory balances. Attempts to use liquid scrubbers for the initial tests® were troublesome and unsatisfactory. The Cl, leaving the quartz reactor was trapped by copper foil in a Pyrex tube inside a furnace at 375 to 435°C. The reaction was visible as a sharp interface with plug flow, but the products at higher Cl, rates collected as solidified melt on the bottom wall of the tube. The CO, in the gas leaving the copper foil trap was collected by Ascarite II (8 to 20 mesh) in standard laboratory Drierite reactor units at ambient temperature. The reaction was observed to show a sharp interface by appearance and by heat generation. At high CO, rates, the Ascarite II became hot at the reaction interface and tended to soften and cake with high resistance to gas flow. This plugging was only a minor problem for the 4-cm chlorination reactor but resulted in serious flow interruptions for the 68-mm reactor. A 9.5-cm ID Ascarite absorber was then used without further difficulty. The CO was oxidized to CO, using a packed bed of CuO wires in a Pyrex tube in a furnace at 425 to 480°C. This reactor gave a sharp reaction interface by appearance and no problems. The CO, from oxidation of CO was absorbed with a similar reactor and results as for the CO, leaving the copper foil trap for Cl,. 57 ORNL/TM-11792 Dist. Categ?g UC-501 hemistry) INTERNAL DISTRIBUTION 1. D. H. Andrews 27. G. W. Parker 2. J.T. Bell 28. J. A. Pashley 3. W. Fulkerson 29. A. S. Quist 4. R. K Genung 30. M. H. Randolph 5-9. P. A Haas 31. J. E. Vasgaard 10. H. W. Hayden 32. R. L. White 11. W. H. Hermes 33. K-25 Records Dept. 12. J. R. Hightower 34. Enrichment Technol. 13. K. H. King-Jones 35. Cen. Research Library 14. F. E. Kosinski - 36. ORNL Y-12 Tech. Library 15-19. D.D. Lee 37-38. Document Ref. Section 20. L. A. Lundberg 39. Lab Records, ORNL RC 21-25. J. C. Mailen 40. ORNL Patent Section 26. J. R. Merriman EXTERNAL DISTRIBUTION 41. Office of Assistant Manager, Energy Research and Development, Oak Ridge Operations, P. O. Box 2001, Oak Ridge, TN 37831 42-53. Office of Scientific and Information Center, Department of Energy, P. O. Box 2001, Oak Ridge, TN 37831 54. 1. C.Hall, U.S. Department of Energy, P. O. Box 2001, FEDERAL, Oak Ridge, TN 37831 55. 1. D. Hughlett, U.S. Department of Energy, P. O. Box 2001, DNTNCON, Oak Ridge, TN 37831 56-60. N. Habermann, U.S. Department of Energy, Washington, DC 20585 61. J.T. Early, Lawrence Livermore National Laboratory, P. O. Box 808, Livermore, CA 94550 62. R. W. Feinberg, Lawrence Livermore National Laboratory, P. O. Box 808, Livermore, CA 94550 63. J. A. Horton, Lawrence Livermore National Laboratory, P. O. Box 808, Livermore, CA 94550 64-160. Given distribution as UC-501, Nuclear Energy (Chemistry).