THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING THIRD COMPONENT INTERACTIONS WITH THE URANIUM-BISMUTH SYSTEM Richard Eo\,alzhiser A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in The University of Michigan 1960 October, 1960 IP-471

-V),, ^"VI

ACKNOWLEDGEMENTS The author wishes to express his sincere thanks to Professor David VO Ragone who ignited and nurtured the motivation necessary to initiate and complete this program. His technical guidance and encouragement have been most gratefully appreciated throughout the course of this investigation. The author is also deeply indebted to Professor E. Eo Hucke for serving as Chairman in the absence of Professor Ragone and for his liberal contribution of time and advice. Appreciation is also expressed to Professor Philip Elving, Professor Kenneth Gordon, Professor Joseph Martin and Professor Maurice Sinnott for their contributions. Sincere appreciation is also expressed to the analytical group for their efforts throughout this program and to Mro John Verhoeven whose skill as an experimentalist is unsurpassed~ The author wishes to express his thanks to Professor Donald L. Katz and the Department of Chemical and Metallurgical Engineering for the opportunity of serving as a member of the staff during his enrollment in Graduate School., Acknowledgement is also made of the financial aid afforded by the Atomic Energy Commission through their Contract No. AT(ll-1)543o

TABLE OF CONTENTS Page INTRODUCTION..0....... 0 0 0. 0............. 1 REVIEW OF THE LITERATURE...................................... 4 Studies of Uranium-Bismuth Systems ooo.....O..O...0. 4 Interaction Studies........on...........n.....o.... 17 EXPERIMENTAL PROGRAM......0....................oooooooooooooooo 35 Theoretical Considerations.,.....oo.0000....o o...... 35 Equipment..oOOOoOO..OO....OO..o....O.ooo..ooo.o. 41 Materialas..o.......oooo o.... ooooooooooo o ooo ooooo o 45 1o Crucibles.o...000000000...00.000000000.00000 45 20 Uranium oooo oooo oo o o oooooo ooo ooo oo oo 47 3. Bismuth.oo.oo.oo. ooooooooooooooooooooooooooo 47 40 Palladiumoo000000..............00.000.00000000 48 70 CopperOO.00.00o.oooooooooooooooooooooooooooo 48 Procedure u000..D0000..0000000000000.......0000 00000 48 l.o Solubil.ity Studieso........ oo.........QOOO.O 49 2O Decomposition Studies...........o.......... 51 APPENDICES ooo 0000oo000000000000000000000000000000000000000000 72 BIBLIOGRAPHY o0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 o 0 0 0 0 0 0 0 0 a 0 0 0 0 0 0 0 0 0 0 0 108

LIST OF TABLES Page I. Summary of Solubility Relationships 5 II. Summary of Solubility Effects Induced by Third Components 15 III. Summary of Proposed Interaction Parameters 34 -iv

LIST OF FIGURES Page 1. Summary of Solubility Data for Uranium in Bismuth 7 2. Valve Arrangement on Equipment 42 30 Filter Type Crucible 42 40 Furnace Assembly for Equilibrium Study 44 5o Solubility of Uranium in Bismuth 54 6. Effect of Copper on Uranium-Bismuth Liquidus at 6000C 55 7. Evaluation of Cu) a600C 56 8. Suggested Bismuth Corner of Bismuth-Copper-Uranium Ternary 57 at 6000C 9. Effect of Palladium on Uranium-Bismuth Liquidus at 6000C 10l Effect of Copper on UC2 Decomposition at 8000C 61 11. Effect of Palladium on UC Decomposition at 8000C 62 2 12. Evaluation of e (Pd) at 8000C 63 U~~~~~~:

LIST OF APPENDICES Page Appendix A - Data 72 Solubility of Uranium in Bismuth 72 Effect of Copper on Uranium-Bismuth Liquidus 74 Effect of Palladium on Uranium-Bismuth Liquidus 75 Effect of Sodium on Uranium-Bismuth Liquid-us 76 Effect of Various Third Components on Uranium-Bismuth Liquidus 77 Determination of Equilibrium Uranium Concentration for Carbide Decomposition in Bismuth at 800~C 78 Effect of Copper on Carbide Decomposition 80 Effect of Palladium on Carbide Decomposition 81 App'endix B - Calculations 82 Calculation of Activity Coefficient of Uranium in Bismuth 832 Partial Molar Enthalpy Change of Uranium for Dissolution of UBio 82 Determination of Uninvestigated Portion of Cu-U-Bi Liquidus 83 Appendix C - Phase Diagrams 85 Uranium=Bismuth Phase Diagram 85 Copper-Bismuth Phase Diagram 86 Cbpper-4Uranism.i-m Phase Diagram 87 Palladlium-Uranium Phase Diagram 88 Palladium-Bismuth Phase Diagram 89 -vi -

Page Uranium-Carbon Phase Diagram 90 LeadAUranium-Bismuth Phase Diagram 91 Tin-Uranium-Bismuth Phase Diagram 92 Appendsix D - Analytical Procedures 93 Spectrophotometric Determination of Uranium in Bismuth 93 Spectrophotometric Determination of Copper in. Bismuth 98 Spectrophotometric Determination of Palladium in Bismuth 102 -vii

NOMENCIATURE AG~ free energy of formation yi activity coefficient of component i Ni mole fraction of component i a. thermodynamic activity of component i chemical potential of component i R gas constant T temperature OK (j) ) Wagner's interaction parameter Qi) interaction parameter of Turkdogan (i) i a interaction parameter of Turkdogan i ratio of 7i in ternary containing j to Yi in binary (j) X i interaction parameter of Ohtani and Gokcen i interaction parameter of Ohtani and Gokcen Hi partial molar enthalpy, cal/gm. mole S partial molar entropy, cal/gm. mole~K K equilibrium constant E electronegativity Subscripts e - electron o standard state H+ hydrogen ion j component i component -viii

INTRODUCTION The present status of metallic solution technology does not permit one to predict with any assurance the behavior of a particular solute in a solvent containing other solute species. Exploratory studies have shown that in many instances significant effects are produced by extremely small concentrations of foreign elements. Consequently much of the available information on, binary systems can not be used with confidence when additional components are present. Unfortunately many practical metallurgical operations involve multicomponent systems. Iron-base systems afford an excellent example of one such family in which a number of alloying elements and/or impurities are present in small but significant quantities. Their presence produces interactions which are of considerable interest to the steel-making industry. The magnitude of this industry and the possible importance of the effects has stimulated invesiti.gators to explore in detail many iron alloy system. These studies have yielded much informatioxl in recent years, the significance of which will be discussed in a subsequent section of this reporto Although bismuth solutions of uranium are of less commercial importance at the present time, they are also subject to such effects. The proposed use of these solutions as nuclear fuels has created a need for better understarnding the nature of the possible interactions. Small quantities of magnesium and zirconium have proven advantageous in minimizing corrosion problems. Since the uranium concentration is extremely critical in this instance, it was necessary to investigate experimentally the effects of these two elements. -1

-2However, corrosion as well as the fission process will continuously introduce many foreign elements to the solution which, if permitted to accumulate, could change appreciably the behavior of the uranium solute. The need for this type of information supplied much of the motivation for the investigation. Further impetus was derived from the anticipation that systematic investigations supplementing th'i-ron-base alloy data might be more revealing from a mechanistic point of view. Although bismuth as a solid exhibits many non-metallic tendencies, a transition toward metallic properties is observed in:the melting process. Thus, work on liquid bismuth systems should contribute to the overall goal of better understanding and predicting behavior in the liquid metallic state. In this study attention was focused on the interaction of third components with a molten solution of uranium in bismuth. Two specific reactions were utilized to evaluate the'interaction parameters. The first involved the equilibria occurring along the liquidus at the bismuth-ri.ch end of the system. At temperatures above the melting point of bismuth the two phase equilibrium involves a liquid solution of uranium in bismuth and the compound UBi2. The second reaction consisted of the decomposition of UC2 in a bismuth-rich medium according to the reaction UC2 —U + 2C. This decomposition proceeds until the activity of uranium in solution reaches its equilibrium value. The first of these reactions determines the maximum concentration of uranium which can be maintained in solution. Any factor which tends to alter this value is of significance in criticality calculations for nuclear fuel purposes. Consequently investigations were conducted at the Brookhaven National Laboratories to determine the significance of small concentrations

-3of additives and impurities on the position of the liquidus. Their results (to be discussed later) indicated that certain elements produce pronounced effects on this equilibrium. However, the studies were not extensive enough to warrant generalizations regarding effects of the other solute elements. A literature search revealed that although much effort had been expended in determining phase diagrams and thermodynamic properties of binary systems, relatively little work has been performed on ternary and higher systems. Thurs, this program was initiated with the purpose of exploring the bismuth corner of several carefully selected ternaries to obtain additional knowledge of the interaction phenomena. The results are discussed in terms of structural considerations, electron effects and thermodynamic behavior for the different alloys. Comparisons are made with the conclusions and correlations developed by previous investigators.

REVIEW OF LITERATURE Studies of Uranium-Bismuth Systems The original study of the uranium-bismuth system was performed by Ahmann and Baldwin (1). Subsequent revisions by Teitel (50) and Ferro (20) produced the presently accepted phase diagram shown in Appendix C. Because of its importance as a liquid-metal, nuclear fuel the liqu.idus at the bismuth end of the diagram has been carefully studied in several additional investigations. The original studies by Bareis (4) covered the range from 2710 to 700~C and yielded the equation listed in Table I. His values were later confirmed by Teitells work which extended the upper temperature to 9000C. However, Teitel's results above 6000C were approximately 10% higher than the previous work of Greenwood (23). At lower temperatures Greenwood s results agreed well with those of Bareis. Cotterill and Axon (14) employed differential thermal analysis techniques to study the system from 0-35 atomic percent uranium. Their work also confirmed Bareis's earlier results obtained using filtration and settling methods. Later 0tudies by Schweitzer and Weeks (47) at Brookhaven disagreed with these results. Their data produced two linear segments when plotted as the natural logarithm of weight percentage vs 1/T. The temperature range from 3000C - 7250C was explored with the discontinuity in slope occurring at 480~C. The equations representing these two segments can be found in Table I. Their expressions yield values which fall below those observed by previous investigators for this temperature region. Filtration techniques utilizing VycorGi-taphitite "G" and molybdenum samplers were employed for their

-5TABLE. I Summary of'Solubility Relationships Predicted Values wt/o Investigator. Equation Range 400oC 6000C 800~c Bareis log W/6-U = 3.13 255~ 271~-700~C 0.251% 1.78% Greenwood log /oU = -.oo 244~ 5150-960~C -- 1.62% 5.37% TOK Schweitzer and log /= 2.585o22 00-48OC 0.182 -- -- Weeks Schweitzer and log w/o= 3.263744 4800-7250C -- l.52% 5.15% Weeks Armour 350~-600~C 0.140% 1.65% Barton 4000-800~C 0.165% 1.30%o 5o9% Author log W/oU = 3 -26O9 400~-8ooc.19o 1.55 5.9% 5U.272-TO- or0800.5

-6 studies. No differences wiere observed in the results for the three different types of samplers and crucibles. This factor' was also checked by other investigators, all of whom concurred with the Brookhaven findings. Recent work by P. J. Barton (5) produced values for the uranium solubility at 4o00, 6000~and 800~C. His values at lower temperatures approximate those of Schweitzer and Weeks reasonably well. His value at 8000C, of 5.9 weight percent, was below the values reported by Greenwood but in fair agreement with the work of Cotterill and Axon. An even more pronounced effect at 4700C was observed by investigators at Armour Research Foundation (3). Their solubility curve below 5600C was considerably lower than those obtained in other studies. The maximum deviation occurred at 470 Cwhere Armour reported a.-value of 2600 ppm compared to values of 4000 ppm and 4700 ppm predicted by Weeks and Bareis respectively. Sampling techniques similar to those used at Brookhaven were employed to obtain this data. A resistivity change which occurs when crystallization begins in. the melt was used to confirm the results. For a melt of known composition resistivity was determined as a function of temperature. A discontinuity in the slope of the curve was used to determine the liquidus temperature for a particular concentration. Bareis's data was corroborated by Teitel, Greenwood and Cotteri'll between 4000and 6000C,wherea-s the data of Barton, Schweitzer and Weeks and that from Armour deviated considerably to the lower side (see Figure 1). Results are known. to be sensitive to certain impurities which might account for the discrepancies. Analytical problems could also generate such deviations.

-7TEMPERATURE - OC 800 700 600 500 400 I0.0 9.0 8.0 7.0 6.0 5.0 l RPJ. BARTON 0 GREENWOOD 4.0 0 | BARCIS -— ARMOUR ------— SCHWEITZER a WEEKS \,3.0 I -~ —---- ___PRESENT WORK 2.0 = 1.0 z 0.9 w ) 0.8 w a. 0.7 z 0.6 w 0.5 0.4 0.3 0.2 0.1, 0 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 IOOO/T OK Figure 1. Summary of Solubility Data for Uranium in Bismuth.

-8The thermodynamics of the binary system, uranium-bismuth, has been the subject of several recent investigations. EganVs emf measurements (18) using a Us/KiC1-KCleut-5%U Cl13/UBiI cell estimated the activity coefficient of uranium in bismuth to be 10-5 at 5000C. His measurements also suggested that the system exhibits regular solution behavior in the liquid region. More recent investigations by Cosgarea (12) and Gross, Levi and Lewin (25) have revised Egan's initial estimate upward. Gross et al. used effusion measurements to determine- the bismuth vapor pressure over alloys of different compositions. They assumed bismuth obeyed Raoult's law in the liquid region and used the Gibbs-Duhem equation to calculate uranium activities. Their measurements were made at 7420C and yielded an activity of 2.7 x 10-4 for a saturated solution of uranium in bismuth. Using Greenwood s solubility data the activity coefficient at saturation would equal approximately 5. x 10-? Cosgarea also measured bismuth vapor pressures across the binary using optical absorption methods. He reported activities for uranium from 745~C to 8420C. At 7450C the activity of uranium in the saturated solution was 2.19 x 10-3. Assuming the solubility to be approximately 5 atomic percent, this value produces an activity coefficient of 4.38 x 10-2 which is an order of magnitude above the value reported by Gross, Levi and Lewin. His data also displayed a very pronounced temperature effect. The Henry's law parameter at 7415~CC was estimated to be about 10-3. This is an order of magnitude greater than Egan's corrected value of 1.6 x 10-4 at 7420C. A correction of Gross's et al. data using Cosgareals measurements.in the liquid region to establish the deviation from ideality yielded a value of 2.5 x 10-4 which agrees well with Egan's extrapolated value. These measurements all concur in the fact that the sytem exhibits strong negative deviations from ideality.

-9Gross, Levi and Lewin also reported free energies of formation for the compounds in the U-Bi systemso Cosgarea calculated partial and integral molar enthalpies and entropies for the various regions of the diagram. He also concludes from the non-linear behavior of the regular solution parameter that the system does not conform to regular solution behavior, even in the very dilute region~ Grieveson and Alcock (24) have determined the thermodynamics for the Au-U system by measuring the vapor pressure of gold. The Gibbs-Duhem equation was applied to obtain uranium activities for the system. This information permitted the calculation of the free energy of the formation of UC2 and UC from carbide decomposition studies conducted in gold. They reported the following values in cal/gm "mole: For UC AG0 (U as solid) = - 22,200 + 1.5 T A(U as liquid) = - 25,200 + 3.6 T For UC?2 as soliG = - 29,610 + 15 T (U as solid) AGO(U as liquid) = - 32,610 + 3.6 T These values compare favorably with the best values previously available in the literature. Bowman (7) measured by combustion'the heat of formation of UC to be -21,300 + 1000 cal/gm mol. An accurate value for the free energy of formation of UC2 is essential to the evaluation of myU to be made in this program.

-10Third element interactions with the bismuth-uranium system have been investigated reegntl9 by several groups. In 1953 Bryner (58) studied the effect of iron, chromium'and nickel on the uranium solubility at 4500C using a filtration technique. Neither iron nor chromium produced measurable changes in the binary liquidus at that temperature. Both of these elements are virtually insoluble in bismuth. However, nickel which readily dissolves in bismuth at 450~C effected drastic reductions in the solubility. The filtrate contained o01l0 uranium and 1.8% nickel. This corresponded to a 30 to 1 reduction in uranium solubility. U Ni2 and U Ni5 were both detected in the residue but UBi2 was apparently absent. NiBi3 was found in both the filtrate and residue after cooling. Weisman (58) studied the effect of nickel concentration on the observed solubility depression. His studies were conducted at 3500C and 5000C. At both temperatures the solubility was reduced by a factor of 10 for 1% additions of nickel. Marked effects were observed with additions of only 0.01% nickel in his studies. Weisman also studied the effect of magnesium on the liquidus composition at 5000C. He reported a slight depression for concentrations greater than 0.0Ol (0.087 atomic percent). In 1955 Weeks et a1.(58) studied the effect of magnesium as a function of temperature over the range 3000-6350C. For 1275 ppm magnesium a definite reduction was observed. The addition of 400 ppm zirconium to the above solution produced no further solubility changes. These tests contained approximately 10 ppm chromium and iron at its normal saturation. Schweitzer and Weeks (46) studied the Zr-U-Bi ternary liquidus in considerable detail at 3500, 3750, 400~ and 4250C. Their studies indicated a negligible decrease in uranium solubility at low zirconium concentrations

-11 over this temperature range. A very sharp reduction was shown to occur for zirconium~n uranium ratios above 0.125. However above this ratio the liquid was no longer in equilibrium with UBi2, but with a ternary field of varyl.n.g composition. Thus-the sharp reduction in solubility can not be interpreted in terms of the U + 2Bi,-UBi2 reaction. For zirconium concentrations below the 0.125 ratio the zirconium appears to have no effect on the preceding equilibrium. Although magnesium additions of 1% had not affected the Zr-Bi and U-Bi binary liquidusesinearlier studies, Schweitzer and Weeks reported. an increased solubility of both uranium and zirconium along th.e U-Zr-.Bi ternary liquidus for magnesium additions. The Armour investigators (3) also studied interaction effects of various solutes. Their results for zirconium confirmed the sharp decrease at certain zirconium concentrations. However, the liquidus curves at 3700C and 4000C which they determined differed considerably from the Brookhaven family of curves. Particular inconsistency was observed at low zirconium concentrIations. Additions of magnesium to the ternary melts producer an. increased uranium solubility. Magnesium effects on the uranium-bismuth binary were not studied. Sodium additions to a quaternary composed of bismuth-uranium-magnesilum-zirconium produced little effect on uranium solubility at lower temperatures. However, an increased solubility was observed at higher temperatures for 5000 ppm.sodium additions. Brookhaven investigators (59) studied the entire alkali metal group. Their results showed lithium and rubidium to have little or no effect on the binary uranium-bismuth liquidu.s. Sodium increased uranium solubilities up to 30%. The maximum effect was observed at 5000C foro.5% sodium~ Potassium (3000-6000 ppm) approximately doubled the uranium solubility over the temperature range investigated. 1.5% cesium increased the solubility about 30% at 450Qc.

Solubility determinat;ions were also performed on, a uranium-bismuth alloy containing 250 ppm zirconium, 350 ppm magnesiLum, and 120 ppm of mixed fission products (60 ppm neodymium, 15 ppm samarilum, 15 ppm stronti:lum, 10 ppm cesium and 8 ppm ruthenium). Values slightly below those obtained for the U-Bi + 250 ppm zirconium curve were observed. Of the elements present bothb magnesium and cesium had been shown to exert positive effects, if any. Hence, the solubility was fairly sens:itive to one or more of the remaining elements. Studies in our laboratories have shown cerium to exert a depressing effect. One might expect,therefore,that both neodymium and samarium would behave similarly. BartonVs interaction studies (5) included many of the transition elements which had not been previously investigated. His studies were conducted at 4500C, 6000C and 7500C for varying thilrd component coneentrations. Manganese was observed to increase the uranium solubility slightly. The effect appeared to be more pronounced at higher temperatures where:increases of approximately 25% were observed for 3.5%(atomic)manganese. At 4500C the effect appeared difficult to distinguish. The increase appeared to be directly proportional to the manganese concent:ration for the condition.s investigated. Rhodium was studied at 4000, 6000 and 800~C. At 6000~C two tests yielded rather inconclusive data. One determination for 0.5% (atomic) produced a 20% increase, whereas a second test containing0.45% (atomic) showed no change. These were the only two points reported at that temperature so it is difficult to evaluate the actual behavior. At 800~C a very definite suppression of about 20% seemed to occur for 1.7% and 2.0o (atomlc) rhod:ium,

but at 400~C 0.18% (atomic) rhodium increased the solubility by a factor of three. Cobalt from the same group produced 20% reductions in. solubility for both 6.7% and 20% (atomic) cobalt at 800~C and 10o reductions for 2o0% and 5.4I%(atomic) cobalt at 600oC. At 4009C 1% (atomic) produced no detectable change in the liquidus composition. Nickel was studied at 3500, 5000 and 600~C with a definite depression. occurring over this entire temperature range. At 6000~C 325 ppm nickel reduced the solubility almost 50%. These results agreed reasonably well with Weisman s earlier determinations. The copper group was examined by Barton at 4500, 6000 and 750~Co The data for copper additions produced considerable scatter at 800~C. However, at both 450~C and 6000C no significant effects were detected for copper concentrations up to 4.50 (atomic). Silver which lies immediately below copper in the periodic chart produced a slight increase in the uranirum concentration. The results at 7500C showed a 10o increase for 4o (atomi.c) silver. The effect at 6000~C seemed to confirm this tendency. Deviations at 4500C were difficult to detect on. the available plot. Gold, the third member of the copper group, displayed the most pronounced effect. A 10% increase was effected by 1.2% (atomicd) gold at 800~C. An increase o:f close to 50%o was produced by 1% (atomic) gold at 6000~C. Even.the lower-temperature deviations were distinguishable for 0.8% (atomic) gold. Teitel has studied the Sn-U-Bi and the PbJU-Bi *ternary diagrams (51). His work produced no evidence of a U-Pb-Bi intermetallic compound at 8000ol The phase diagram (Appendix C ) indicates a decreasing urani.um solubility

as the lead content increases with the liquid solut~ion remainil.ng in, eqtilAli'brium with UBi2. This behavior would necessitate a compensating increase in yU as NU decreases. Support for such an. increase was provided by earlier Brookhaven. carbide-formation studies (58)o'Their attempts to form uranium carbide on the walls of graphite cru~cibles from uranium-bismuth solutions were unsuccessful until uranium concent:rations reached 4.7% uranium.at 11000C. 120 hour runs at 10000C for 1% uranium solutions in bismuth were unsuccessful in producing a detectable carbide layer at the interface. Later carbide decomposition studies (54) performed by the author showed this concentration to be well below the necessary concentration at 10000C. However, lead-bismuth eutectic solutions containing 1% uranium formed a detectable uranium carbide layer after only 96 hours. A kinetic explanation for the results seemed unsatisfactory since conditions in the two runs were essentially the same. However, an increased yU resulting from the presence of lead could increase the aU for 1% to that required for the formation of the carbide. The 3500C -section of the Sn-U-Bi ternary (:Appendix C ) was also studied. Tin was shown to decrease the solubility of uranium and bismuth. The bismuth-rich, liquid solution was found to be in equilibrium with UBi2 as in the preceding diagram. Thus one would expect similar behavior in'this'systen as the 7U must increase to compensate for the decreasing uranium concentration. The decreasing solubility of uranium in this system was confirmed by Barton's work. The results of interaction phenomenea in bsmuth-uranium systems have been summarized in Table II. With the exception of Barton.'s recent work all

Table II SUMMARY OF SOLUBILITY EFFECTS INDUCED BY THIRD COMPONENPS Element Investigator Temperature Effect Magnitude* Na Brookhaven 5000C Increase Weak Author 600~C Increase Weak K Brookhaven 5000C Increase Strong Li Brookhaven 5000C None Rb Brookhaven 5000C None Cs Brookhaven 4500C Increase Moderate Mg Weisman 500~C Decrease Weak Weeks 300-6350~C None Author 600~C None Zr Schweit zer&Weeks 350-4.25C Decrease Strong Armour 370-420~C Decrease Strong Author 4006000~C Decrease Strong Ce Author 6000C De'crease Weak Fe & Cr Brookhaven 450~C None Ni Weisman 4500C Decrease Strong Ni Barton 350-600~C Decrease Strong Author 600~C Decrease Strong Mn Barton 450,600&750~C Increase Moderate Co Barton 400o,600&8000C Decrease Moderate Rh Barton 40oo600oo&9O0~C Decrease Moderate Cu Barton- 450,600&7500C None Author 600~C Decrease Moderate Ag Barton 450,9 600&750~C Increase'Moderate Au Barton 450 600&750~C Increase Mod-Strong Pb Teitel 8000C Decrease Author 6000C Decrease Hayes&Gordon 600~C Decrease Sn Teitel 3500~C Decrease Hayes&Gordon 3500~C Decrease Pd Author 600~C Decrease Moderate Slizght corresponds to a change of 5% or less for 1 atom percent Moderate corresponds to a change of 5-25% per 1 atom percent Strong corresponds to a change of 30% or more per 1 atom percent~

-6 of the previous interaction studies on bismuth-uranium systems were performed in conjunction with nuclear fuel development. The third component elements studied were determined primarily by the anticipated fuel composition, with little concern given to a systematic control of variables such as valence, size or electronegativity, which are known to have an effect on the extent of interaction. Only the study by Schweitzer and Weeks of Bi-U-,Zr inters actions possessed the depth required to evaluate the interaction process parameters. The Brookhaven investigation of the effect of alkaldipmetals furnished a semi-quantitative indication of the behavior wi.thin a group (or of the size effect), but the available information was insufficient to permit definite conclusions. Barton's work when completed should shed considerably more light on the nature of the interaction phenomena. His selection of the elements should permit an evaluation of thei relative importance of the parameters cil.ted above. His preliminary results seem to suggest a definite size effect in the copper and cobalt group. Likewise, his data permits evaluation of valence effects for the fourth period transition elements mapganese, cobalt, nickel and copper. Each element was studied at three temperatures thus permitting assessment of this effect. In most cases the difference in solubilities remained about constant at the two upper temperatures, 600_-8000c. At the lower temperatures, 3500 and 4OsC, the effects appearediLmuch less apparent. Solubility results can not be used to calculate activity coefficient changes without first establish ng the activity of uraniu.m in the solid phase. If a ternary field does not replace the UB~2, then the activity of uranium remains unchanged in the presence of the third component and the act;ivity

-_17coefficient change is proportl.onal to the reciprocal of the solubility change. However the formation of a ternary field will likely change the value of the uranium actlvity along the liquidus, and thus necessitate additional measurements. Interaction Studies Despite recent contributions from X-ray and neutron diffraction experiments, the structural aspects of the liquid state still remain uncertain. Recent work (19,21,36,42',55))has confirmed the existence of short range ordertand has yielded the coordination number of most low melting liquid metals. Liquid alloys have also been studied by several investigators and the results suggest that a type of micro-lnhomogenity may occur among clusters in certain liquid alloy systems. X-ray diffraction work by Danil.ov and Danilova (15) on lead-bismuth and tin-lead eutectics demonstrates a fore— hadowing of the solid eute-ct:ic at temperatures just above the liquidus. The radial distribution curves for these systems produce certain peak.-s which correspond to those obtained from theljpure tin and bismuth liquids. Examination of these phase diagrams shows that in each case the eutectic mixture is composed of a lead solution and essentially the pure constituent tin or bismuth. Thus, the formation. of tin and bismuth clusters Sn the respective systems at temperatures just above the liquidus seems quite probable. Hendus (28) in his studies of the gold-tin system observed a doubling of the main peak for dompositions near that; of the intermetallic compound. He concluded that clusters possessing a coordlnation similar to that exhibilted

by the compound were present along with clusters of gold and tin randomly arranged. However, the validity of this model has not been established. Gingrich and Henderson (22) have explored the sodium-potassium system for different compositions at 115C. A -systematic shift in breadth and main peak positions was observed for different compositions. Recent work in this field has been' voluminous and future'efforts seem certain to disclose many enlightening features pertaining to the coordination and structure in molten media:. Evidence suggests that in many respects the liquid state differs only slightly from that of the solid. Physical properties of true metals undergo only a slight change during the melting process and hence the bonding force cannot change radically. However at the melting point a very definite stability transition occurs which produces fluidity at the expense of long range order. Despite these changes, which are also characteristic of the gaseous state, proximity of-the atoms is maintained such that a definite density is assured. Numerous theories have been proposed to account for this behavior, but each possesses apparent shortcomings. Consequently, one can only conjecture as to the actual structure and type of forces persisting in liquid metallic media. Evidence to date Suggests a sea of electrons containing charged atoms migrating continuously from cluster to cluster. These clusters or cybotactic groups possess orderly structures very similar to the solid state. Densities of these groups are imagined as being equal to or greater than deasities in the solid state. However the free- volume between clusters

contributes to the overall reduction in density observed for mnot metals upon melting. Some investigators have postulated that these micro-groups possess five fold axes of symmetry. This would certainly explain the absence of long range order in the liquid structure. Solution of a metallic constituent appears to involve a transfer of electrons between solvent and solute. Certain elements with relatively free peripheral electrons (conduction electrons) contribute them to the sea, whereas others extract electrons to fill lower energy levels. Consequently these atoms become charged particles which may attract or repel one another as ordinary charged bodies. The electron sea maintains electroneutrality by adjusting to the fluctuating local charge. Hume-Rothery (30), Raynor (43), Jones (31) and others have demonstrated the significance of the electron/atom ratio for phase diagrams. Sieverts and Krumbhaar (4k8) have shown, that the solubility of hydrogen in copper is cic decreased considerably by tin and aluminum additions, but is increased by additions of nickel and platinum. Bever and Floe (6) have likewise confirmed the results for tin~ Himmler (29) observed a decrease of the hydrogen solublllty in solid copper for zinc additions and confirmed the increases for nickel and platnuMm,i cited above. These results have been interpreted by ascribing a chemical potential, A, to the free electrons. The condition of thermodynamic equilibrium may be stated by the following equation: 1/2 RH2~ -aC dE e (] )

Elements such as aluminum, tin and zinc pbssess electron/atom ratios greater than that of copper. Their dissolution in a copper solvent results in an increase in the free electron concentration. This increases and necessitates a corresponding decrease in AH+ at a given hydrogen activity. Both nickel and platinum have been shown by magnetic measurements to dissolve in copper as essentially neutral. atoms. This produces a decrease in the electron concentration which reduces e'. The equilibrium tequli.rez an increase in p+ which necessitates an increased solubility. Although. the systems considered in. the present study are far more complex than hydrogen and copper, the foregoing considerations provide a possible interpretation of the results. Hargreaves (27) has shown that additions of aluminum, which contribute three valence electrons per atom to a brass, produce a decrease in the zinc vapor pressure; whereas nickel with no effective valence electrons produces an increase. Wagner (57) has concluded from the foregoing studies that the activity of a solute metal, 2, will be increased by a third component, 3, if 2 and 3 change the electron atom ratio in the same direction. Conversely if the ~electron/atom ratios differ from that of the solvent in opposite directions, a decrease in the activity coefficient of 2 is to be expected. Wagner points out that the preceding considerations have neglected the interdependence of iH+ and [Le. Since the electron concentration is likely to influence [H+, this factor must be considered when utilizing this concept. Opie and Grant (40) have shown that the behavior of copper additions to an aluminum solution containing hydrogen contradicts this theory. Further study will likely explain the apparent discrepancies.

-21Wagner (57) with cons;:ideration to the chemical potential of the electrons, derived a quantitative relationship expressing the effect of one solute on the activity' coefficient of a second. The derivation neglected di rect interaction between ions of like charge and ascribed all deviations from ideality to electronic effects. Hence for his assumptions d(RT ln yi) = 0 if dJe = 0 The following expression relating behavior in the binary systems 1-2 and 1-3 to that in the ternary resulted: in 3 2 a in y a i 2 n 32 ~,,L ~ln(2) ar3 aN2 -= aN- 2 Qualitative considerations described previously must be utilized to evaluate the sign. Wagner proceeds to show that for a system i.n which ion in.teractio can be neglected, for any solute in a given solvent must have the same sign. Hence, the expression appearing to the 1/2 power should always be positive. This supposition is contradicted in mercury systems where bismuth, lead, tin and zi.nc are known to produce negative values while cadmium thallium, lithium, sodium and potassium yield positive expressions. This demonstrates the li.mitations o:f th.e derivat.ion and of the underlying assumption of negligible ionic interaction. Si.nce these interactions generally occur, this expression is probably valid only for systems where the binaries exhibit pronounced deviations from ideality. Wagner obtained experimental data for the effect of thallium on. the activity coefficient of potassium, sodium and lithium amalgams. His observed values agreed remarkably well with values predicted from his express ion.

-22The most extensive studies of interaction!phenomena have been performed on iron-base systems.-. The actual values obtained for interaction parameters are of secondarr interest for purposes of this study, but the models and correlations used in presenting and discussing the data are quite pertinent to the problem at hand. Wagner (56) using a Taylor series expansion for the excess partial molar free anerges (or the logarithm of the solute activity coefficients) developed the following expression: in y2(N2.,N3N4...) = ln 2 +N N 2 + NN + N4 (3) 3,9YQ+ Ngae+ N5 Y- + N)14 4 This expression neglects second and higher order derivatives which'eems reasonable for dilute solutions. He defined the interaction coefficients as (j) _ lnyi and evaluated them for the limiting condition of zero concentration for all solutes. Wagner has also shown a simple proof of the reciprocity of solute effects on one another such that e(2) - (2) This equation, or a simple modification permitting the direct use of weight percentage and common logarithm, has been used with fair success by many investigators to correlate their results. The elimination of second order terms implies that different solutes will have a negligible influence on the effect exerted by other solutes. If both solutes are in the dilute range, the probability of interaction should be small and the assumption valid. Chipman (11), Turkdogan (52), and Ohtani and Gokcen (39) have all published summary articles concerning interactions in iron-base systems.

Chipman lists values for the parameter, (X) for cases where "X" represents C, Al, Si, P, S, V, Cr, Mn, Ni, Cu and Mo, and "Y" signifies H, C, N, 0. Si and S. The' values for carbon and silicon substantiated Wagnervs reciprocity relationship. Chipman also discusses the parameter values in terms of interatomic attraction. He suggests that negative values can be explained using a simple model. He considers the Fe-Cr-O system as an example where (Cr) -8.8. He pictures chromium atoms replacing iron atoms as the nearest neighbors of oxygen, The greater bonding energy for chromium-oxygen than for iron-oxygen, results:in the oxygen atoms being held more firmly by the solution. This decreases its activity and hence its activity coefficient. Vanadium has a greater effect than chromium and -manganese a lesser effect. Aluminum yields a very large negative coefficient (Al)= _ 1340. Although this value is subject to considerable experimental uncertainty, it is definitely indicative of a large aluminum-oxygen bonding energy. Chipman has postulated that nickel which has little affinity for oxygen. atoms associates itself with iron atoms in the solution. This competition with oxygen to coordinate iron results in weaker oxygen bonding and an increased oxygen activity coefficient. Carbon, silicon, aluminum and phosphorous react similarly in iron-sulfur melts thus producing an increased activity coefficient for sulfur. Chipman evaluated Wagnerrs expression for interaction effects between. positive metal ions and electrons. In the Fe-C-Si system the values for (SC and c () were equal as predicted,but their magnitude was considerably

greater than that predicted by Equation 2. The Fe-O-S system yielded an exceptionally large value for c(S) of + 130 which Chipman attributed to a combination of ion interactions and electron competition. Both atoms would apparently diminish the electron chemical potential thereby raising the activity coefficients. Turkdogan (52),recognizing the limitations of very dilute solutions for Wagner's derivation, developed a correlation which he felt was applicable over greater concentration ranges. Wagner's expression (Equation 3) for the effect of different solutes on the activity coefficient of a particular solute can be written as: (2)7(3)7(4) (4) Y2 = 2 2 2 (2) where (2) represents activity coefficient of 2 at some concentration. (3) (2) 2' represents the effectiof 3 on y2 or 7(2) for N3 0, N4 = 0 _(3) (5 rn /) From Equation 3 and 4 it can be seen that in 2 3)= N - l7 N (3) (5) c3 was evaluated at zero concentration for all solutes. Consequently, in Y2(3) is seen to be a linear function of the atom fractions in this expression. Turkdogan's examination of data from the Fe-Si-C, Fe-Mn-C and Fe-Cr-0 systems indicates that the individual coefficients (3),(4) etc. may also vary with the concentration of component 2. The variation is generally not as great as with the third components themselves, particularly at low concentrations, but it does restrict the composition range over which the relationship is valid.

To overcome this difficulty Turkdogan has suggested plotting the data as ANy/N4 vs NX where ANXy = Ny-Y Ny represents the atom fraction of component Y in the Z-rich ternary, X-Y-Z, and NY the atom fraction of Y in the binary Y-Z, both evaluated at the same activity of solute Y. AN~ was known to be relatively insensitive to temperature for many alloys. His calculations demonstrated that the ratio, ANt: -T,was independent of the activity of Y in most systems for which data was N~Y~~~ ~X available. He defines a parameter(c' Qy,by the following procedure. Division X Y of the expression for ANy by Ny yields: Y Y Ny X + 1=- - for constant activity of Y. (5) X Y ~ = ~Y at a given temperature and activity of Y.' (6) T,ay A term PY is defined for a given temperature and fixed concentration of Y as follows X =7 P=Y Yt (7) X This term corresponds to the 7X used previously by Wagner in his expression 72 32 2.... As Ny approaches zero, Q and P become equal. A plot o ln y vs NX yields a curve which is independentf.'thercuncentration of Y.

-26Of the alloys for which data was available only the Fe-Mn-C system failed to yield values for the parameter which were independent of activity. Interactions in multicomponent systems were handled in a. similar!-manner to that proposed by Wagner. The procedure assumed the effects of various solutes on the activity coefficient of a particular solute were additive and that second order effects were negligible. Thus the following expression re sults n Q = n Q + in + In Q3 +..... (8) This equation reproduced the data extremely well in'solutions of O-Al-Si-P-M4-S in iron for values of log QX <0.5. However, recent evidence (53) has shown that the- individual effects of silicon and phosphorus on the solubility of graphite in iron are not additive. One would also expect that, although this relationship is valid at higher concentrations for ternaries, difficulty with second order effects might enter the picture when additional components are added at these higher concentration levels. Ohtani and Gokcen (39) have also derived a relationship which they claim to be valid for any concentration range. For a three component system they derive theYJfollowing expression which defines the parameter, 4ff) l M in Y2 N1 + N i = (3) (9) N1 N1,Where N1 stands for the atom fraction df component 1 etc. Hence (3) N1 + N3 (2) __ 2 = a:and for 1NF O (10) = N1 + N2 2 and N-O this relationship reduces to Wagner's derivation.

-2. An attempt was made to correlate interaftion parameters with atomic numbers. A plot of self-interaction parameter. vs atomic number suggested a linear relationship for'elements within the same period. Carbon, hydrogen and nitrogen as solutes in an iron matrix yielded virtually a straight line for such a plot. A different line segment was generated by altminum, silicon and sulfur from the second period. In both cases the value of the parameter decreased with increasing atomic number. Although the results appear promising, additional information will be required to establish a definite correlation between the atomic number and self-interaction parameter. An investigation of the effect of third elements on the carbon solubility in liquid iron produced rather suggestive results. The suggested by Turkdogan and discussed earlier in this review was shown to produce a straight line when plotted vs the atomic fraction of X. Ohtani and Gokcen have deyonstrated that the term when plotted v's the atomic number of X alTo yields a series of straight lines. Data for aluminum, silicon, phosphorus and sulfur all fell on one straight line whereas points for vanadium, chromium, iaSanga&sle iron, )',:r cobalt and nickel fell on a second straight line. The available information for copper produced a point which fell slightly below the latter curve. A definite periodicity is also displayed by two other parameters considered by these authorso. which they defined as follows, (a ln YC ln NC(11) XC \ 7aa 6 NX a NX~~~~~

and aft discussed earlier, both increase for increasing atomic number within a period. This corresponds to an increasing value of %Xfor decreasing carbon solubilities. Positive values are obtained for aluminum, silicon, phosphorus, sulfur, cobalt, nickel and copper and negative values for titanium, vanadium, chromium and manganese. The authors conclude that electron contributions of the solutes alone cannot account for this behavior. However they do attach significance to the fact that the negative values occur for elements with strong carbide-forming tendencies. They conclude that, if the X-C bond energy exceeds that for iron-carbon, the solubility will be increased and the activity coefficient decreased. If the ironcarbon bond energy is greater, then the reverse is expected to be true. Kitchener, et al (32) utilized a crude structural model to explain the phenomena displayed in iron-base systems. They consider the interaction process as a competition between solutes to coordinate adjacent'solvent atoms in-; a preferred orientation. Carbon and sulfur were used as an example of solutes which prefer to coordinate iron in different forms which display little compatability. Consequently the different clusters tend to isolate themselves with a resulting increase in the activity coefficient of each solute. Their structural interpretation of the process emphasizes the interaction of solutes with the solvent as well as with each other. Additional consideration is given -to size effects and the state of ionization in rationalizing the observations. In the solid state carbon- islknovn to fit with slight strain into the intersticcesJ of the iron lattice. Resonating covalent

bonds between iron and carbon atoms are postulated as contributling to the metallic-like structure, Transference measurements on. carbon in austenite have revealed that the bonding must be partially ionic and that the carbon atom possesses a positive charge of from 1-4. The electrons transferred are shared among coordinated iron atoms. Such polarization in metallic.bond"ing is quite common according to Kubaschewski and Reinartz (35) and reflects differences in the electronegativity between atoms. The effect is greatest between systems exhibiting intermetallic "electron" compounds such as Mg3Bi2. As additional carbon is added to the structure, the octohedral holes existing between iron atoms are gradually filled; and the carbon atoms begin to repel one another as suggested by their reluctance to occupy adjacent holes. This explanation proves consistent with the observations of the iron-carbon system. There is little reason to believe the behavior inth- liquid. adateiiotuld differ:ignificantly from that in the solid. The iron-sulfur system is assumed to possess a different type of structure. Sulfur is virtually insoluble in the iron crystal lattice, whereas iron-suLfiideis completely misible in molten iron and in iron-oxygen and FeO-SiO2 melts. The latter are certainly ionic so that iron-sulfur must possess a transitional character that will blend with either. Analogy with iron-sulfur suggests that in liquid iron sulfur ions would teand to coordinate six iron atoms. Increasing sulfur concentration enhances the ionic character of the melt at the expense of the metallic, as the average charge per iron atom is increased. This change is postulated as effecting the reduction in aS as the NS increases.

These two structural pictures provide a consistent picture when applied to the Fe-C-S ternary. Increasing sulfur tends to create an ionic type melt in contrast to the metallic environment preferred by carbon. Thus, carbon is no longer accomodated as 1ina purely metallic medium and an increasing activity coefficient is observedo Other more metallic solutes such as nickel, vanadium, manganese and chromium are discussed in less detail by the Kitchener, et,,al. Nickel, similar in size and electronic structure to iron is assumed to substitute for it readily with little effect on solutes. Manganese although similar in size evidently differs appreciably in electron affinity. Smoluchowski and Koehler (49) have shown, by comparing optical and excitation levels of atomic iron and manganese that the 3d manganese electrons occupy higher'energy levels than the 3dd iron ele-ctrans. In solution manganese is assumed to lose them to the lover energy levels available in iron andto.-acqulre a positive charge such tha't it- interacts stroingly with negatively charged sulfur. This accounts for the depression of yS by manganese. The authors admit to the speculative- nature of their interpretations in reaching the foregoing conclusions, but fee1 in the absence of a sound quantitative theory these considerations provide r satisfactory grounds for comparisono Alcock and Richardson (2) have discussed the electrons atom ratio theory proposed by Himmler and further developed by Wagner. They observed inconsistencies which substantiated Wagner's conclusion with respect to the importance of the chemical potential-of the ions. Their experiments showed that sulfur in a gold-copper solution possessed an activity coefficient 80 times greater than in puree copper. Since the electron: atom ratios are

approximately the same for copper and gold, the change must be attributable to the different chemical potential of the,..sulfur ion in the two systems. Furthermore, in accordance with Himmler s theory, the activity coefficient of hydrogen in copper is decreased by both platinum and nickel, while gold displays no effect. However, platinum and gold have been shown to increase the activity coefficient of sulfur in copper, whereas nickel produces a decrease. These findings demonstrate the inadequacies of the electron~ atom ratio theory for the copper-sulfur system and emphasize the importance of the ion potentials in. rationalizing behavior in certain alloy systems. Since these individual potentials can neither be measured nor calculated accurately, these authors suggest attacking the problem from the chemical or thermodynamic standpoint. Their considerations assign energy to the various types of bonds existing in the ternary solution. Assuming the atoms to be randomly distributed and to possess the-same coordination number as the solvent, they proceed to derive an expression for the partial heat of solution of component,P, in —the binary solvent, X-Y. If component Y is present only at high dilution so that Y-Y bonds have a low probability of existing, -lc-. the following expression can be derived. The terms AeS(X), MY(X) etc. represent the partial heats of solution of. P o:i>; and Y in X etc. Furthermore they assume for low concentrations of Y that Sp( )Sp<(x) so that: -a Nj*(o Q aN A,

-32For the following assumptions: 1) 1Sp(y) = 6Sp(x). at low concentrations Ef P.; 2) Y7(y) and FP(X) are taken relative to the same standard state; and 3) 2SY(X) is Raoultian, (nY) (Y) 7(X) Y(X) =E n y - in - in y (12) The preceding expressions permit an evaluation of solute effects if a knowledge of binary behavior is available. For systems in which P represents the metalloids, an appreciable difference between AHIp(y) and EII(X) is not likely to be reversed by LHy(X). since this is generally small for true metal-metal mixtures. In other instances where the difference is small the Y-X interaction has been shown to be of significance. For purely metallic systems one might expect the AHy(x) contribution to have the same:significance as AHIp(). Neverthelessgin view of the assumptions necessary the values of e generated by these equations cannot be expected to be of great accuracy. The authors give detailed consideration to the validity as well as to the magnitude of the error introduced by these simplifying assumptions. For cases where the coordination number of 1P'differs from that of the solvent X-Y it has been shown that the term HY(x) must be corrected by the ratio C/C where CP represents the coordination of, and, C that iOf ithe solven.t Certainly the assumption of complete randomness has been shown to be invalid

in many systems possessing high interaction -energies. Preferential clustering of Y relative to X about -: P- atoms would produce X-Y bonds in excess of those predicted by the equation. This necessitates multiplying. by ___ actual to correct for this phenomenon. The authors in subsequent yrandom work-have given greater consideration to this correction. Despite these considerations this theory can only yield qualitative results. It needs to be subjected to further tests in which the three terms on the right are known more- accurately.

Table III Summary of Proposed Interaction Parameters 1) Wagner Ao Neglecting positive ion interaction's and considering only electron effects: a ln y2 a n a ln Z2 a In 1/2 - = + V a P aNTT7 Qualitative considerations determine sign. B. Taylor Series Expansion of solute activity coefficients: In y N 1:$3i Y2' in 2 / n / -t in Y2 N 2(W1)+ N ( 7) + N (aln 74) +. (3) _a in y2 3 T,N2 evaluated as N3 0 (3) (2) ~2 3 For expression: = 7(2 (s) 4) (2) n (5) Y2 2 22etclI' in Y2 = n Y2 in 2'2 + Y2 = where y(2 )represents activity coefficient of 2 in binary at N2 and (3) Y2 Y2 (n N (3) in2 = Ne2

-352) Turk(ogan 2Y2 T,a2 As N2 approaches zero, Q(3) p ( 3) (3) 2 2 = 2 3) Ohtani and Gokcen..(3) in Y2 (5)2 Va N ) 2 4) Alcock and Richardson (3) (a n n 72(3)- in 72(1)n Y3() where 1 represents solventnd 2 bnd l 37it0utes;hence. 2() represents activ.ity-.coeefftcient,of component 2 in binary system 2-3.

EXPERIMETAL PROGRAM Theoretical Considerations. The preceding theoretical considerations and correlations have been substantiated in part by experimental results in iron-base alloy Systems. One of the purposes of this Study was to'assess their applicability for'explaining interactions between metallic constituents in bismuth-rich systems. The present investigation was also intended to provide additional data on solute interactions with uranium in a bismuth solvent. Solubility effects were measured and used in conjunction with earlier data in an attempt to generalize on the behavior of the ternary liquidus for Bi-U-X systems. In addition carbide decomposition studies were undertaken to provide- a - more direct measure of uranium activity coefficient variations resulting from interactions with other solute species. Previous measurements of the solubility had produced numerousediscrepancies in the literature. Thereforelthe first phase of this study was concerned with the evaluation of the binary liquidus in the temperatureregion of interest (4oo00~C to 800oc). This curve was then used as the basis for measuring deviation's produced by the Ladded 7I solute elements. Although the curve is generally referred to as a solubility curve, the solid phase in equilibrium with the solution is UBi2 and not metallic uranium. A plot of the lnt vs 1/T (It = mole fraction of uranium at liquidus) yields the partial molal enthalpy change of uranium in passing from solution to the compoundg UBi2.

A number of factors were considered in selecting the third components to be studied. An observed periodicity of behavior for elements in bismuth aided in the selection. Solubility studies (59) in bismuth had shown that groups of elements tended to behave similarLy.o This was further confirmed by examination of binary alloy phase diagrams. Consequently, certain group's were omitted from consideration on the basis of the extremely low solubility demonstrated by certain members. Other groups presented extremely difficult analytical problems and thus were excluded from this study. These two considerations eliminated all but the titanium, cobalt, nickel and copper groups among the transition metals. Brookhaven studies had included zirconium from the titanium group and nickel. The latter had produced a tremendous depression of the solubility for relatively small additions, thus focusing attention on this region of the periodic chart. On the basis of the foregoing considerations,copper and palladium were studied, and the results compared with earlier data for neighboring elements. The effects of other elements including sodium, zirconium, qagnesium, cerium, rhodium, lead and zinc were examined semi-quantitatively. The bismuth-uranium system exhibits strong negative deviations from ideality as suggested by its compound-forming tendencie's (App:1' ). The bismuth-copper system (App.-O i) displays complete solubility in both the liquid and solid states,whereas the copper-uranium (A"pP."-C ) system demonstrates immiscible tendencies. The palladium-bismuth system (App:i:C ) appears to deviate negatively from ideal solution behavior. The uraniumpalladium phase diagram (;App.:~C') contains several high melting compounds suggesting negative departures from Raoult7-s law in this system also.

A temperature of 600ooC was selected for the study since it was near the range of practical interest for nuclear fuels and high enough to produce a reasonable: uranium concentration. Dilute concentrations of the perturbing element were used in an attempt to minimize second order effects arising from a'self -interaction of the thir:! Compopents Coefficients expressing the solubility change as a simple function of it hid componetc ronctrationL wre calculated. The equilibrium constant for the reaction U + 2Bi-~ UBi2 can. be.expressed as: a K; UBi2 a aTia If the addition of a third component does not alter the composition of the solid phase, then the value of the equilibrium constant and the aLBi should retain unchanged for small additions of a third component. If one can assume that bismuth, as the major component, obeys'iRqau-jt's law in this region, then aBi'should maintain a value close to unity. Negligible change in these three terms requires that the activity of uranium in solution also remain constant at the liquidus. The activity can be regarded as the product of the uranium mole fraction, NU, and its activity coefficient, 7U. Any change in the equilibrium value of NU must therefore be compensated for by a correspbnding change in yU. If the above assumptions are valia the measurement of solubility changes would permit an estimate of the effect of solute interactions on the thermodynamic properties of the components. It should be mentioned that coefficients expressing such effects might be expected to be functions of temperature anmduranium concentration as well as third component concentration.

A more drct y of evaluating interation effects on the activity coefficients of uranium in bismuth utilizes the second reaction presented above. The uranium-carbon phase diagram (.See figure 9) indicates a series of high melting uranium-carbon compounds formed sacroas the diagram. The UC2-C region is of particular interest in this investigation. In this two phase reginn the activity of both carboA and uranium must remain constant. The activity of uranium can be ecalculated by using the free energy of formation of'UC2. For the reaction: U + 204UC2 the equilibrium constant can be w-xpreWsed as follows: Ka = aCo'The phase diagram reports negligible Solubility of carbon in UC2 and hence both of the'W phases are present as pure solids J n'- thtir standard state and will have unit activity. Detenrminations of Ka from 4GO =-RTlnKa permits one to express aU = 1/Ka = constant for a given temperature. The equilibrium value for aUt is unaffected by the medium in which the' reaction occurs providing the' equilibrium is not disturbed. Consequently, the above re lationship remains valid for a decomposition occurring in bismuth sine the bimuth act's only as a solvent for the free uranium. Carbon and TC2: continue to iexist as pure, solid phases. The equilibration of molten bismuth and TC2 at 1ome teperature results in the decomposition of UC2 until the uranium con'entration achieveS its equilibrium value, Mea-surement of this concentration permits an evaluation of U since the activity had been previously evaluated from the UC2.free.energy. This calculation provides a check on other values of the activity coefficient reported in the literature.

For elements displaying weak carbide-forming tendencies this reaction can be used to evaluate'interaction effects on yU with aminimum of assumptions. The elements previously considered as third components are very weak carbide formers compared to uranium and should not interfere with the solid phases in the'equilibrium. The two solid phases remain at unit activity and the same equilibrium constant applies. The bismuth does not enter into the reaction at all; therefore, no assumption regarding its activity is necessary. Changes in the equilibrium uranium Oon'eentration can now be directly related to changes in 7U. The third component concentration can be varied continuously in the dilute region for any given temperature. The experimental portion of the investigation necessitated equilibrating solid and liquid phases. The "'solubility" portion of the study involved the reaction of metallic uranium with molten bismuth to form a bismuth solution of uranium and the compound UBi2. Interaction effects were studied by simply adding the desired amounts of copper and palladium to the charge. A filtration operation was used to separate phases following equilibration. The filtrate was analyzed for uranium and the third component using spectrophotometric techniques (See Appendix D). These results yielded the liquidus composition from which the'solubility parameters were determined. Whereas the above reaction required a complete transformation of the solid phase from metallic uranium to UBi2 and uranium'ihsolution, the carbide decomposition reaction maintained the'same'solid phase throughout. At equilibrium the uranium in the solution should have -an activity equal to that of uranium in the UC2-C region. Exce'ss carbide was added to insure an adequate supply of uranium and to decrease the time required to reach equilibrium. Although earlier

studies (13) had'suggested that decomposition would not occur, previous work (54) in our laboratory had shown a reasonable rate for this reaction. As in the solubility studies,third elements were introduced along with bismuth and UC2 in the charging operation. In this phase of the work none of the third components were pre-alloyed with the bismuth. A filtration operation again separated solid and liquid phases prior to chemical analysis. Equipment The reactions described above had to be performed in an inert atmosphere to minimize the oxidation of uranium. At the same time one had to suppress the volatilization of bismuth from the melt at reaction temperatures. Likewise, precise temperature control was important since temperature dif:t ferences would also produce changes in solution composition which might conceal interaction effects. To accomplish the above conditions, equipment designed for this'study which was somewhat different than the apparatus utilized by other investigators for solubility determinations~ A horizontal reaction chamber was employed which permitted eleven separate equilibrations. Each reaction was conducted in a graphite filter crucible (See Figure 3') from which a single filtrate sample could be obtained. The apparatus permitted varying the third element concentration at a fixed temperature for each run. Previous apparatus described in the literature allowed little composition control, but could be sampled repeatedly at different temperatures. Since composition, rather than temperature variations was of primary interest, this design seemed more advantageous.

THERMOCOUPLE GAGE COLD TRAP TO FURNACE A VACUUM PUMP HELIUM SUPPLY AVLVES Figure 2. Valve Arrangement on Equipment. FLUX MELT l -; S~ WCARBIDE 7" POROUS GRAPHITE FILTER DlSK FILTRATE Figure 3. Filter Type Crucible.

The reaction chamber consisted of a 2f-1/2" ID Mullite tube inserted in a split-wound resistance furnace (See Figure 42). The furnace was two feet long and contained compensating windings at each end to minimize endeffects. The furnace power supply was varied through a 0-220 volt variac. A Foxboro on-off temperature controller activated by a chromel-alumel thermocouple was used to maintain the reaction temperature. Thie control thermocouple was placed between the mullite tube -and::-the:furnace,-w.indixgs; to. iamjpr6ve 8sn'SSiviri:y Td!f'm'% ~ Xeatu're'AfluctUati&6ns w ithin- the: furnaae...The r'oi;:i cf the s tem was:SuffSiient t o maiatain-6tie - r4 a(t:lXn' tmtrurn e-a.aewic~.%in,'10:. i:,thei, ~temperature range of interest. Phe temperature profile was checked during the Early stages of theprogram by means of five chromel-alumel thermocouples spaced along the reaction zone. The couples were enclosed in stainless steel hypodermic tubing to eliminate corrosion from bismuth vapors. They were introduced to the furnace through conex glands mounted on the face of the brass end plate. The ends of the couples were then inserted in holes in the walls of every other crucible. At 6000C the reaction zone was practically isothermal. A maximum variation of 3_SC; -wi, zQ:- observed at 8o00.C, This differenc'e was ittr.ibutablte 1.rgely to the fact that, iesplite radiation'shields used at the open end of the tube, the'end crucible'"saw" temperature's somewhat below- the rest of the furnace. However,it acted a-s a radiation'.shield for the remaining crucibles so that the other ten were virtually isothermal. Most runs were made without utilizing the;.eleventh crucible. No problem was encquntered at the closed end of the tube since it remained three inches inside the heated zone. An insulating plug was

FURNACE ASSEMBLY FOR EQUILIBRIUM STUDY GRAPHITE CRUCIBLES (SEE FIG.3) ~~~.~ r i.~ -- n c31 3r J~dP u I j E C ) INSULATING. BRASS END PLATE 0 V MULLI OTENTOMETE C> 4 J.P tP -~\ -I; 3 C~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~HED TO VACUUM3Oj,j t' INSU LAT I ONTEPERTUE CNT L e ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~O TEMCUL rW C'fC FUNC WINDINGS Figure 1 P- A 9, e RTE UBB,1. -r re?lcl7 v 08r~~c cr:r ~~~~ ~Tg ~~- ~~cj ) d~ INSULATING R PLUG H~0e c I L 5' J-~ ~ ~ ~~~~~~~~~~~~~F INSULATING BRASS END PLATE~~~~~~~~~~~ADITI PLUG r\~~'~~~~,,~'~' —-------— ~~'~'~;~~:~ & ~rTHERCs. ELDS TO \\\\ ii~-Jo, TEMP c~~~~~~~~~~~~~~~~~~O HR Figure 4.~

inserted to minimize radiation and convection losses at that end. The latter stages of the investigation were conducted using only two thermocouples in the reaction zone. They rested beside the central crucibles in the charge. The equipment (See] Figure:2) permitted complete environmental control from pressures in the miicron range up to one atmosphere. A vacuum pump connected to the system through -&a cold trap was used to evacuate the system. The combination was capable of producing vacuums of about one micron as measured by a thermocouple gage. A mercury manometer indicated pressures up to atmospheric pressure. Commercial helium passed through a charcoal trap at liquid, nitrogen temperatures was used for environmental control and for providing the pressure differential necessary to induce ffiltration. An O-ring seal between the ground end of the mullite tube and a brass end plate provided access to the reaction zone. The sample tray was attached to the end plate. by stainless steel connecting rods. Thi-s connection facilitated charging and removing the crucibles and assured proper positioning for each run. Materials 1) Crucibles The graphite crucibles (Figure 31) were fabricated in our shops from Graph-i-tit.e "G", an impervious graphite obtained from Graphite: Specialties Corporation. Filters were machined from blocks of porous graphite stock obtained from the National Carbon Company. Both grades 50 and, 60 were used withoutdetectabledifferences in the results. Both have effective porosities

of 48a6,but the former retains particles 0.00079" compared to 0.00047" for the latter. The disks were 1/16" 1/8" thick and had a diameter slightly greater than the ID of t;he crua4ble Jutt above the shoulder. The filters Vwe force-fitted into position thus reducing the probability of leakage around the disk. The finished crucible was then outgas'sed at least 2000C above the run temperature to eliminate volatile constituents present in the impregnating medium. Crucibles used in the decomposition study at 8000C were outgassed initially at 14000C and then again at 10000C just prior to use. This treatment was adopted when outgassing studies indicated that significant desorption was not detected below 10000C. Wetting of the crucible was desired in these'studies as it further assured a carbon activity of unity, In all these outgassing treatments helium was introduced and reabsorbed prior to discharge. These crucibles were then charged immediately and the run initiated. Any crucibles which were exposed to the atmosphere for greater than 12 hours were given a rapid out;gassing at 10000C before being used. Both residue and filtrate were removed readily from the crucible. This permitted its re-use in subsequent runs with t;he same'element. New filters were installed and the outgassing treatment repeated. Since the equilibrium uranium concentration along the l.iquldus in the uranium-bismuth system was above that required toi- form carbide in the presence of graphite the use of graphite crucibles was initially discouraged. However, the results of earlier investigations (58) coupled with our experience in trying to form the carbide indicated that the kinetics of such a reaction were not rapid

4~7 enough to interfere. Graphite, alumina;;>, and vycor samplers had all yielded comparable results in the Brookhaven investigation, (58). -Excess uranium was added to permit some reaction with the crucibles, as well as with any oxygen in the systemwithout disturbing the uranium,-bismuth equilibrium. 2) Uranium Mallinckrodt's reagent grade uranium of 99.97o purity was used for the solubility studies. The one quarter pound slugs provided by the manufacturer were turned on a lathe to produce fine spirals. These turnings were first dropped in trich.loroethylene to remove the oil and then cleaned electrolytically in a solution of chromic oxide and H2S04 "to remove the oxide layer. The above treatment assured a solid phase which reacted readily with the molten bismuth at reaction temperatureso 3) Bismuth The bismuth used throughout this investigation was obtained from Belmont Smelting and. Refining Company and was reported to be 99. 998. It was received in 1" lumps which were vacuum cast into (7/8" cylinders. The slug was then cut into pieces of the desired weight and was charged directly to the crucible. All bismuth. used in the solubility phase of the work was filtered through. pyrex frit in the casting operation. This practice was eliminated in the later stages of the study. A hydrogen refining operation originally used by: Brookhaven and Armour investigators was found unnecessary with, the bismuth presently attainable (3).

4) Palladium The palladium was obtained in powdered form from Chemical Commerce Company an.d was reported as 99.98% pure. During the early stages of the work palladium was added in the form of a master alloy. The alloy containing 5-6o palladium was filtered and cast under vacuum prior to its use. However, subsequent experimenlts demonstrated that it could be added quantitatively directly to the individual charges. This procedure was employed exclusively in the decomposition work. 5) Copper Fine copper wire, 99.99.99 pure, was used for copper additions. As in the case of palladium, copper additions were originally made through a master alloy. This procedure was abandoned in the decompoeition.studies as copper readily dissolved in' the bismuth solvent. 6) Uranium Dicarbide Three different types of uranium carbide were procured for this study A granular mixture of the mono and dicarbides supplied'by NUMEC was used in most te'sts. The carbon analysis of 7.95%C would indicate 71%UC2 and 29%oUC. X-ray diffraction studi'es off the'mixture confirmed quanitatively the above percentages. Negligible U02 was present and no free uranium was detected. Other impuritiees were maintained at a level. consistent with reactor grade quality. This carbide appeared more metallic than. either the UC2 or the UC, both of which were provided by Davison a's a partially sintered powder. In * Nuclear and Mat;e+rials Equipment Corporation.

4i9the absence of carbon the UC-UC2 mixture would generate a higher uranium activit;y consJistbnt with that region of the UC-UC2 phase diagraf~ BHowever, if cal bon is involved in the equilibrium the UC portion of the mixture would be converted to either UC2 or would decompose to uranium in solution and carbon. However, a uranium concentration in the melt above that required for the UC2 equilibrium should result in a reaction with graphite producing the UC2 and reducing uranium to its equilibrium value. Fluxing aided in assuring a more intimate contact between the graphite and the solution. The UC2 was used in several runs to confirm the results obtained with the more readily handled NUMEC product. The fines in the UC2 created numerous difficulties in both the handling and equilibrating operations. The larger particles were used but they did not afford the surface area that the granular carbide supplied. 7) Flux The flux used for promoting wetting in the decomposition studi.es was composed of 56%KCl and 44%NaCl by weight. This composition corresponds to the eutectic composition which melts at 6110C. Reagent grade chlorides were used for the mixture. Procedu:re 1) Solubility Studies After the pretreatmentis just described, the reactants were charged to the crucibles. Uranium, at least l4o in excess of that required to

saturate the solution, was charged first. The bismuth slugs were then inserted and ten crucibles placed in the furnace, The pressure'was reduced to'the micron regio:n before heating was begun. Runs were made with the vacuum system in operation. throughout the run and with its removal. The furnace proved to be'ufficiently vacuum-tight to permit the latter. The charge was equilibrated at temperature for periods ranging from 2 to 43 hours with very little difference observed in the results. Cosgarea s observations (12) confirmed this very rapid approach to equilibrium for this particular, reaction. Most runs were permitted to remain at temperatures for 12-24 hours before filtering. Filltration was effected by suddenly introducing helium to the system. An increase in pressure of several hundred millimeters was sufficient to push. the solution through. the filter disk very rapidly. The possibility of seepage aroi:nd or through the fi.lter was checked repeatedly and was never observed to occur. Even pressures of 50 mm of mercury have failed to filter some solutions. The filtrate and residue were removed from the crucible by inverting and tapping. The entire filtra'te slug was sent for analysis because of th-e segregation r:roduced on cooling. The residue was analy2zed in several instances in. an attempt to establicsh a secon.d poin:t on -the tiLe linet, Extrapolat;i.on of these li:nes permitted anr estimate of the composition of the solid phase in -the equilibrium. The binary liqulidus was studied from 400~C to 800~C with repeated runs. Each run, contained approximately 10 independent determinationr T ernary

L51btudies were confined to runs at 600~Co. Some were heated initially to 6500 or 700 C and then reduced to 6000C for 24 hours in an attempt to approach the equilibrium from both directions. Concurring results were obtained for both procedures. 2) Decomposition Studies After the pretreatments previously described -the reactants were charged to the furnace. The optimum charge size appeared to consist of 3 gm of the carbiEde, 20 gm of bismuth and the de'sired amount of third component. The uranium content of 3 gm of carbide was far in excess of that required to saturate the solution. However, the increased contact area was essential in producing reasonable equilibration times. Charges of up to 8 gm of carbide yielded results in 50 and 75 hour ru.ns similar to those for 3 gm charges. Bismuth charges of less than 20 g-mdid not distribute consistently over the en~ti;re filter and. thus failed to yield'samplesO. Larger charges required more decomposition. and. hence longer periods to reach equilibrium. The carbide was always charged first. If third components were involved'they were added along witb.the carbide. The bismuth slugs vere placed above the carbide and finally the flux (wh.en -used) was added. Graphite caps with a 1/87' hole were added to reduce bismuth. vaporization.~ The ini-tial pos-ition of the carbide was important as the density differences between. it and the melt were not significant en.ough to produ.ce a change in its position~ Consequently carlbide placed above the bismuth floated and only a smal-l percentage of the phase was contacted. Significantly lower concentrations were

observed in these'samples when compared to those with. the- carbide' beneath the: melt Assuring i.ntimate contact between carbide and solution afforded the major expe:rimental difficulty. Many of the data inconsiystencies apparently resulted f~rom poor wetting of the solid. The.equipment was not designed to permit agitation of any kind and thus it was necessary to rely on the bismuth wetting'and flowing axround the carbide~ Much time was devoted to the development of te'chniques to improve the contact problemo Thee carbide phase was cath;odically etched prior t~o charg8ing i:n, an at;tempt to clean'the surface~ Consistent improvement was not observed in, these charges. Size reduction was attempted using a mortar and pestle, but the pyrophoric nature of the carbide resulted in almost immediate oxidation of the newly created surfaces. A pre-wetting treatment accomplished by heating carbide and bismuth to above 800C in an aevacuated, agitated, vycor tube succeeded in producing wetting of the ITu. carbide, but not the Davison UC2 Howpver, the result failed to:improve the consistency of the data.

RESULTS 1) The solubility determinations over the temperature range 4000 to 8000C yielded the following expression for the liquidus composition' loglo0 wt.%U - 3.272 - 269 TOK No suggestion of a break ln. the curve was detected between 400O and 5000C as was reported by Schweitzer and Weekssand the Armour Research Foundation. Determinations at eight different temperatures yielded average values for solubility which deviated only slightly from the above relationship (See Figure 5). 2) The partial molar enthalpy change for uranium in the reaction: UBi.-U + 2Bi, equalled + 12,30gm mole This value was obtained from the slope of the solubility curve using the following relationship (See Appendix B for derivation)lnNi 6MU soln Hcpd Qa (T gR R 3) Copper additions to the uranium-bismuth system depressed liquidus uranium concentrations at 6000C,as is shown in Figure 6. Figure 7 shows that the reducti.,on when plotted as the n.atural logarithm of atomic percent uaranium vs atomic percent copper was linear for copper concentrations less than 3 atomic percent. A plot of the liquidus at 600~C (See Figure 8 ) suggests that the solid phase in equil.brium with the -established portion ~53

-54TEMPERATURE — ~C 800 700 600 500 400 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0: I.0O Z 0.9 a. 0.8 - 0.7 o 0,6 w 0.5 0.4 0.3 0.2 0.1 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 I000/T OK Figure 5. Solubility of Uranium in Bismuth.

WEIGHT PERCENTAGE URANIUM 0 0 - -7- -- - 0 0 CD c-No -. ~i ni C) 0~~ N~~~~~~~~~~~~ CD~dP1 0m ~C) o-; O ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ O~ m 0 -u z~~~~~~~~~~ cD~~~~~~~~~~~~~~~~~~~~~~~~ cD II, td G) 0 0~~~ I+ OD 0'o I'r 0 0 C) 0 o & o~ —, O~~~~~~~~~~~~~i iaiii

4.0 3.0 0 Cu -3__ o0 A Cu-$t 2.0 0 <3 2.0 -__ z 0 w IIzJo 0 0.9 D 0.8 z <3: 0.7 0.6 0.5..........0 2 4 6 8 10 12 COPPER CONCENTRATION-ATOMIC % FigurQ. 7. Evaluation of (cu) at 6oo0C.

-5720 18 16 0/10 ATOMIC U RANIUM Bi 0 DATA POINTS 2 CALCULATED 0 2 4 6 8 I 0 12 14 16 is 20 ATOMIC % URANIUM Figure 8. Suggested Bismuth Corner of Bi-Cu-U Ternary at 600~C.

-58of the liquidus is very likely UBi2. Consequently, the data permitted the evaluation of WagneP's interaction parameter, ECu), from Figure 7o (Cu) ln7rT alnNU EU = N - Cu =11.0 for constant uranium activity over that portion of the liquidus. Since ECu) = (U), it was possible to estimate the other extreme of Cu the liquidus where NU -e 0. This portion has been included in Figure 8o 4) Palladium additions to uranium-bismuth melts at 6000C effected a sharp reduction in the uranium liquidus concentration for palladium concentrations greater than 0.6 atomic percent (see Figure 9). 2.0.9 atomic percent palladium produced a 43% decrease in the uranium concentration. At palladium concentrations below 0.6 atomic percent there was no evidence of any change in uranium concentrationo 5) Qualitative studies showed that sodium increased the uranium concentration at 6000C. Rhodium, cerium, lead and nickel all produced decreaseso 6) The reaction, UC2 -U + 2C, produced an equilibrium uranium concentration of 1.86 weight percent in a bismuth medium. This concentration, when used in conjunction with Grieveson and Alcock's LG? for UC2, yielded an activity coefficient of 1,22 x 10-4 at 800~C (see Appendix B-1).

-591.6 1.5' 1.3 1.2 1.1 1.0 0.9 0.8 0.7 0 I.O 20 ATOMIC PERCENTAGE PALLADIUM Figure 9. Effect of Palladium on U-Bi Liquidus at 600c.

-60 7) Carbide decomposition studies in bismuth-copper solvents concurred. with the findings in the liquidus study. Although the data exhibited. considerably more scatter than the liquidus work, the interaction (Cu) parameter, CU, was also found to equal approximately +11.0 Figure 10 contain8 data from two different runs which produced slopes similar to that in Figure 6 for the liquidus at 600~Co. 8) Palladium produced an increase in the equilibrium uranium concentration for th.e carbide decomposition at 800~C. The effect appeared to vary linearly with palladium concentration below 5 atomic percent palladium (See Figure ll). A plot of the natural logarithm of atomic percent (Pd) uranium vs atomic percent palladium yielded a value for cU of -408 (Figure 12)o

WEIGHT PERCENTAGE URANIUM 0 0.. - 0 (D 0 r b p Oi (D 1-b, ~r:' o-bk o0 0 ('-) I —' (1 Fl~~~~~~~~~~~~~~, rvl CDCM "0 G) 0 om o o ~d 0 0 r 0+ c-t co 0 o o cO

-622.6 0T-24 2.2 0 z A_ 0 2.0 1.4.O I 2 3 4 5 6 ATOMIC PERCENTAGE PALLADIUM Figure 11. Effect of Palladium on UC2 Decomposition at 800c. Figure 1. Effect of Palladium on UC 2 Decomposition at 8000C.

5.0 z w Ld 0 W 4.0 1.,~ 3.0 z 0 z 0 z 0 0:::t~~~~~~~~)~~(P) Figure 12. Evaluation of at 8000c.'o m.o z.o 3.0 4.0 5.0 6o PALLADIUM CONCENTRATION - ATOMIC PERCENT Figure 12. Evaluation of C W at 800~C.

DISCUSSION OF RESULTS The data obtained in this study for the uranium-bismuth liquidus determination proved to be except.onal.ly consistent and Reproducible~ Individual samples varied le-ss than 35 percent from the average value used to establish, the curve in Figure 5o This variation is no greater tlhan the uncertainty introduced by the chemical analysis~ As Figure 5 demonstrates, the average values for the solub.lity at t;he eight temperatures; studied fallon a straightk line when plotted as in wt % vs 1/T. This dataf as Figure 1 indicates, also confirms the data of Bareis and Greenwood at temperatures below 6500C. Greenwood's data above 6500C deviates to the high side, but Barton's more recent determination at 8000C confirms precisely the values obtained in this study Although Schweitzer and Weeks reported values 15%1 below those found by the author -above 4800C, they did concur on the slope of *the culrve~ However, their data suggested. a break in the curve at 480~C which resulted in a second linear segment at lower temperatures. Armour investigators found a more exaggerated break a:round this temperature, but their data deviated so substant.ially from that of all previous investigators that the significance of the break appears questionable. No satisfactory expla:nation for such behavior has been offered by either o:f these groups. Two logical explanations for the different results reported by various investigators involve the analytical procedures and'the purity of the reactants. The former could very well account for the vertical displacement of the curve of Schweitzer and'Weeks from that prod.u.ced in this study. An error in the sstandard curve or difficulty with the stability -64

of the standard solution could produce the type of discrepancy observed. The extreme sensitivity of this curve to certain inmpurity atoms could likewise produce such deviations. Most elements likely to be present as impurity atoms i:n either bismuth or uranium have been shown to have depressing effects on the binary liquidus. This factor might account for the unusually low values reported by the Armour groupo An analysis of the thermodynamics of the liquidus permitted an evaluation of the change in the partial molar enthalpy of uranium for the reaction~ UBi2_ — U + 2B1.. (See Appendix B-2). The value for AU of +12,300 cal/gmmole was obtained by assuming the activity of uranium remained constant along the liquidus. If the activity is assumed to be a function of uranium concentration and temperature, the expression for the total differernt:ilof activity with respect to these two independent variables yields the relationship, - /T - R It was necessary to assume that ( ZU IT= 0. This condition is certainly satisfied in the Henry s law region, but is not necessarily valid at liquidus concentrations Although Egan has suggested the system conforms to regular solution theory, this fact was disputed by Cosgarea, and hence was not used to estimate ([j in the derivation. Regular solution. theory would produce a 3% decrease in H. Both copper and palladium were found to depress liquidus uranium concentrations,'but in slightly different manners. Copper, as shown in. Figures 6 and 8, effects a smooth decrease originating at zero copper concentration. An attempt was made to estimate the composition of the solid phase by analyzing the residue in, each crucible, and thus establish a second

-66m point on the equilibrium tie line. In all cases the copper concentration, in. the residue slightly exceeded that in the filtrateo This would suggest that the ternary liqu:idus is in equilibrium with a te:rncary solid field as *the tie lines are directed away from the'UBi2 omposition. It should be pointed out that the total uranium concentrati:ons in. these charges was approximately 3o. Since the filtrate concentration is between 1% and 1o5% uranium as compared. to 36% for IBI2, the two points on the tie line are fairly close together and could n.ot be extrapolated with a great deal of accuracy. Incomplete solutionr of the copper could also account for -the -hlgheTr copper analyses in the residues, A m-etallographie examination of the residue would. havepermitteda more positive identification of the solid phase. The presence of a ternary solid phase woulld li:kelyproduc- a definite change in the uranium activity from that in UBi2. Consequently, at the point along the llquldus where the transition occurs an abrupt change in, slope shou.ld occur. Nicke, immediately adjacent to copper in the periodic chart, exhibits such behavior a's does zirconiumo The liquidus established in this study for the Cu-U-Bi ternary did not possess such a break. The reluctance of copper to form compound.s with either u:ranium or bismuth (See phase diagr-ams Appendix C) diminished the likeli'hood of a ternary compound in this system. These considerations led to the construction of the bismuth corner of the ternary shown in Figure 80 For the equiblibrium as postulated above, it wasj possible to evaluate the interactlon parameters for copper and uranium solutes in, bismuth. Wagner Is parameter, iCa), as evaluated from Figure 7 and yielded a value of +11.0. Using his reciprocity relationship ~icj)= (U) it was possible to determine the liquidus for low uraniurm concentrations. Two 3poin.ts were

calculated~ NU = 0~0025 for NCUl = Q124 and NU 0.006 for NC = 0119o These are shown in Figure 8. The parameter proposed by Ohtan:i and Gokcen, Cu), is likewise equal to +11t0. It is defined similarly to Wagner's eU but i.t has the restriction of constant activity, as opposed to constan:t concentration for 4(C u) The former was true for'these studies, but the difference between the two restrictlons vanishes as NU approaches zero. Similarly for a, given copper concentration, the parameters QCu) and PU (cu) C.,U (Cu can.'be evaluated from the relationship ln PU NcuC = in Q(I Alcock and Richardson's analysis requi:res some speculation as to the thermodynamics of'the respective binary systems. The uraniumhbismuth system has been shown -to deviate negatively from Raoult s law. The copper-bismuth system has been. shown to be a slight positive d.eviator'o The uranium-copper system exhibits some immiscibility which suggests a positive deviationo The expression proposed by Alcock and Richardson (See Table III) for E produces the following results. Te:rms 1 and 2 are positive and. 3 is negative,o Term 2, -ln /U(Bi)r has been shown to equal approximately +10. Terms 1 and 3 are probably small in magnitude compared to ln 7U(Bi) and might'be expected to cancel each. other as a first approximation. Thus'this approach sllggests a value of approximately 10 which agrees remarkably well with, t:h.e measured value of 11.0. Mechanistically one can interpret'the results in terms of copper, aoms, which, have little affinity for uraniumcompeting with the uranium atoms'to coordinate'bismuth. The effective number of bismuth atoms or clusters

-68available for the uranium is thus decreased, and hence, an increased uranium activity is achieved for a given uranium concentration. The carbide decomposition studies in the binary -issmuth-copper solvent seemed to confirm these results. Although the data appears to exhibit considerable scatter, a critical evaluation of the individual samples provides some justification for the curves in Figure 10. Run T-23 yielded two particularly good samples in which the wetting and filtering both appeared satisfactory. The curve has been drawn to pass through these two points. Run T-18 appears not to have reached equilibrium for lower copper concentrations. The behavior of palladium provides a contrast to that of copper. The liquidus uranium concentration undergoes a sudden depression at 6000C when the palladium concentration reaches 0.6 atomic percento The sharp reduction suggests the presence of a new solid phase in which the uranium probably assumes a new activity. Hence the solubility studies, while defining the liquidus, were not useable in evaluating the interaction parameters. The absence of any change in uranium concentration for palladium concentrations up to o.6 atomic perent, the range in which the solution is likely to be in equilibrium with UBi2, suggests that the activity coefficient may be un.affected by palladium in dilut'e solutions. However later results from the carbide decompoosition studies ind-icated that palladium exerted a depressing effect on (Pd) the uranium activity coefficient~ The interaction parameters(P and. (Pd) were found to equal -4~8. Although the data exhibited some scatter within runs and appreciable vertical displacement between runs, most runs concurred in the'sign of the effect. The data from run T-24 plotted in Figure 11 appeared to be the most reliable. It converged at 1.89 percent

-69uranium for no palladium whith agreed with the earlier values for the decomposition in bismuth. A similar slope was observed for T -22 which is also plotted. Contact problems between the carbide and the solution provided difficulty in these studies. Poor wetting resulted in negligible decomposition and, hence, low results in some samples. It is possible that preheating at 1000 C caused the decomposition to proceed beyond that required for equilibrium at 800~C. This condition would necessitate the formation of UC2 from the excess uranium in solution and the graphite crucible. Consequently, poor wetting or insufficient time could also explain high results under these circumstances. Failure of the filter disk to function propertl*y could also contribute to the scatter. An analysis of these results using the expression of Alcock and Richardson is not particularly revealing. Both the palladium-bismuth and palladium-uranium systems appear to be negative deviators. The palladiumuranium interdttion appears exceptionally strong based on the high melting compound in the binary system. To confirm the value predicted in this study the term In yu(pd) would have to overcome the large positive contributions of the other two terms. Sufficient thermodynamic data is not available to postulate on the likelihood of such a confirmation. The process appears to involve a competition between uranium and bismuth for palladium. Apparently the strong interaction between uranium and palladium results in the uranium being held more strongly in solution. Thi.s'effect is apparently not overcome by the palladium:-bismuth interaction which would tend to reduce the bismuth available for combination with uraniuWm"

-701 Electron effects are difficult to evaluate for these systems. Mott and Jones (38) report bismuth to have 0.04 free electrons per atom. No value for uranium has been found in the literature. Both of these elements have complex electron structuresxuEhthat their behavior at elevated temperatures and in alloy systems is difficult to predict. Electronegativities provide some measure of the ability of elements to attract electrons. In this connection it might be noted that; positive electronegativity differences (Eso ) are displayed by the'solvent solute copper group elements whereas the nickel group shows negative values. This seems to indicate that bismuth has a greater affinity for electrons than copperbut less of an affinity than nickel or palladium. The electronegativity of lead which has been shown to increase the activity coefficient of uranium in bismuth differs only slightly from that of bismuth. Uranium exhibits a negative value of -.30. On the basis of this limited data it seems that solutes which possess electronegativities greater than bismuth and greater than uranium tend to depress the activity coefficient. Palladium with a difference of-0.49 (EBi- Epd) and nickel with a difference of -0.65 (EBi - ENi) both seem to conform to this postulate. Copper with an electronegativity difference of + 0.61 (EBi - ECu) produces an increased activity coefficient. Further study will undoubtedly reveal the significance of this term in predicting interaction effects. Certainly no definite conclusions can be reached with this limited data. Solubility studies to date seem to conform to no general pattern. Behavior in the copper group'suggests that as the size of the solute increase's

-71its effect becomes less depressing. Whereas copper was shown to decrease the solubility, both sllver and gold increased it, with the latter having the more pronounced effect. Although no consistent behavior is observed from group to group, definite regularity seems to occur within the cobalt, nickel, copper and tin groups. Other groups were not studied in sufficient detail to evaluate their consistency. The value obtained for the uranium activity coefficient in this investigation, 1.22 x 10-4 confirms reasonably well khe values obtained in earlier investigations. Egan's reported value of 10-5 at 5000C increases to 1.6 x 10-4 when corrected to 8000C using the regular'solution theory. Similarly the value reported by Gross, Levi and Lewin becomes 2.25 x 10-4 when their assumption of ideality in the liquid region is replaced with Cosgareas data over that portion of the diagram. However, Cosgareavs reported value still remains an order of magnitude higher. This discrepancy is presently under investigation and will likely be explained in the future.

APPENDIX -A - DATA Data: Solubility of Uranium in Bismuth Summary: Run Temperature C Length of Average Uranium Run-hrs. Concentration-wt o A-19 400 24 0.185 A-21 613 14.5 1.70 A-22 500 21 0.57 A-23 600 22 1.54 A-24 470 14 0 o434 A-25 450 12 0.350 W-19 800 25 5.8 tData: Run Uranium Concentrations-~ by weight A-19 0.182 0.188 0.184 0o 292 0.188 0.188 0.184 A-21 1. 63 1e5f 1.82 1.73 1.75 1.62 1.74 1o75 1.70 1.71 1.70 A-22 0.54 0.57 0.58 0.57 0.56 0.54 0,58 0.56 0,59 A-23 l,' 5 1Q50 1.57 1.57 1.55 1.58 1.53 1.56 1.53 -72

-73Run Uranium Concentrations-o by weight A-24 o.432 o. 31 0.447 0.440 0.434 o.424 0.437 0.422 A-25 0.354 0.307 0.343 0.353 0.356 0.343 W-10 5.9 5.9 5.6

-74 - Data: Effect of Copper on Uranium-Bismuth Liquidus Run Temperature0C Time-hr. Copper Conc. Uranium Conc. )Uranium AtoAtomic % Atomic. Weight, Cu-1 6000 28 None 1.34 1.53 None 1.37 1.56 0.05 1.35 1.54 0.174 1.34 1.53 0.262 1.30 1.48 0.323 1.30 1.48 0.525 1.28 1.46 0.602 1.25 1.42 1.05 1.23 1.40 1.19 1.20 1.37 1.60 1.16 1.32 Cu-3 6000 30 None 1.34 1.53 0.272 1.32 1.51 0.550 1.24 1.41 0.790 1.26 1.44 0.868 1.21 1038 1.25 1.18 1.34 2.20 1.03 1.18 3.11 0.956 1.09 6.88 0.776 0.885 Cu-4 6000 27 None 1.29 1.47 None 1.32 1.50 0.279 1.23 1.40 0.521 1.23 1,40 o.845 1.23 1.40 1.10 1015 1,31 1.36 1.13 1.29 2.07 1.01 1.15 3.03 0.949 1.08 7.01 0.761 0.867 9.64 0.722 0.823

-75Data: Effect of Palladium on Uranium-Bismuth Liquidus Run Temperature C Time-hr. Atomi.c % Atomic % Weight % Palladium Uranium Uranium Pd-l 6o00 24 None 1.34 1.53 None 1.37 1056 0.0622 1 26 1.44 0.0951 1.33 1.52 0.119 1.41 1o61 0o.168 1.35 1.54 0.206 1.37 1.56 0.252 1.31 1.49 O. 348 1.29 1.47 0.411 1.32 1.50 Pd-2 6000 20 None 1.25 1.43 None 1.o34 1.53 0.0o680 1.28 1.46 0.099 1.33 1.52 0.131 1.31 1050 0.180 1.29 1.4r 0.204 1.31 1.50 0.242 1.34 1.53 0.334 1.26 1. 44 0.423 1.33 1.52 0.709 1.22 1.39 Pd-4 6000 23 None 1.32 41 51 None 1o35 1.54 0o163 1.36 1l55 0.425 1.29 1.47 0.553 1.36 1o55 0.623 1. 30 1. o48 0.746 1,29.o 47 Pd-6 6000 27 None 1.32 1o50 None 1. 32 150 0.0557 1.26 1.44 0.1057 1031 1.49 0.219 1.32 1.50 o.306 1,32 1.51 0.528 1.31 1.49 0.753 1,26 1.44,.01 1.15 1.31 1 o196 1.03 1.18 2.09 0.736 0.84 * In end position during run

-76Data: Effect of Sodium on Uranium-Bismuth Liquidus Run Temperature~C Time-hrs Atomic % Atomic % Weight % Sodium Uranium Uranium Na-1 6oo 15 0.123 1.23 1i40 o.183 1o31 1.49 00.256 1.31 1.49 0.404 1.29 1.47 0.569 1.28 1.46 o.731 1.32 1.51 0.926 1.25 1.43 1.24 1.32 1.51 2.54 1.42 1.62 * 3.24 1.4o 1.60 * End Position

-77Data: Effect of Various Third Components on Uranium-Bismuth Liquidus Run Temperature ~C Time-hro Element Weight % Uranium S-1 600~ 27 None 1.44 Rhodium 1.21 Rhodium 1.16 Rhodium 1o 38 Nickel 1. 02 Nickel 1.45 Tin 1)44 Tin 1.44 Ti:l.n 1 43 S-2 6000 15 None 1. 39 None 1.41 None 1i442 Lead 1 39 Lead 1.533 Lead 1 113 Zinc 1.43 Zinc 1.41 Zinc 1.45 Nickel 00800 Nickel 0.208 Ce-1 6000 68 None 1 o47 None 1o51 Cerium 1o 39 Increaeing 1.40 o38 1o38 1.40 1.34 1 ~-3

-78Data: Determination of Equilibrium Uranium Concentration for Carbide Decomposition in. Bismuth at 8000~C Run Time-hr. Flux Preheat Carhide Uran;ium Concentrati on. Time -10000C wei.gh.t.o T-8 75 Yes None NUMEC o 103 75 Yes None NUMEC 3.50 75 Yes None NUMEC 1 29 75 No None NUMEC 1C14. 75 No None NUMEC 2 10 75 No None NUMEC:1.23 75 No None NUMEC 1 14 T-15 100 No None NUMEC 1 30 100 No None NUMEC 1 28 100 No None NIMEC 1i 46 100 No None NtEC 1. 31 100 No None NUMEC 1.19 100 No None NUMEC 1o 38 100 No None NUMEC 1. 20 T-16 70 No None NUMEC 1.862 70 No None NUMEC 1 813 70 No None NUMEC 1.899 70 No None NUMEC 1.857 70 No None NUMEC 1. 909 70 No None NUMEC 1, 844 70 No None NUMEC 1.862 70 No None NUMEC 1 878 70 No None NUMEC 1 o648 T-18 87 Yes None NUtMEC 1.37 87 Yes None RNMEC 1.- 538 87 Yes None NUMEC 1. 42 87 Yes None NUMEC 1.23 T-20 96 Yes 1 NUMEC 1..87 96 Yes 1 NUMEC 1053 T-22 90 Yes 5 NEC a1 42 90 Yes 5 NUMEC 1.53 90 Yes 5 NUMEC 1o26

-79Run Time-hr. Flux Preheat Carbide Uran.'ium Concentration. Time wei ght T-23 64 Yes 9 NUMEC 2.06 64 Yes 9 NUMEC 1. 40 T-24 50 Yes 7 NUMEC 1.89 50 Yes 7 NUMEC 1e 4, 50 Yes 7 Davison UC2 1.95 T-26 95 Yes 10 NUMEC 2.12 T-27 95+57 Yes 11 NUMEC 1o38 T-28 53 No 2 NMEC 1 o 86 53 No 2 NUMEC 1.89 T-29 120 No 2 NUMEC 0, 846 120 No 2 NUMEC 1. 21 120 No 2 NUMEC 5'7 T-30 100 Yes 12(11000~C) NUMEC 1.95 100 Yes 12 NUMEC 2.10 100 Yes 12 UMEC 2.1.7 T-31 50? 18 Davison UC 2 0.550 50 18 Davison UTC2 1.90 50? 18 Davison UC2 1.59 50?18 NUMEC 1.17 50? 18 NUMEC 2.68 T-32 105 Yes 2 NUEC 0 95'7 105 Yes 2 NUMEC o0877 NUM EC - A IUC-TC- mixture provided by Nuclear Mate:rials and Equil.pment Corporagion.

-80Data: Effect of Copper on Carbide Decomposition Run. Temp.~C Time-hr' Flux Length of Atomic Atomic Weight Preheat-hr % Cu L U i U T-18 800~ 80 Yes None None 1.20 1. 357 None 1.21 1.38 None 1.24 1.42 0o573 1.17 1.33 0.998 0.956 x109 2,,05 0.825 0.942 3.07 0.596 0.68o 4.00 0.857 0 978 5.63 0.843 0 962 T-23 8000 73 Yes 9 (1000 C) None 1.81 2.06* None' 1.23 1.40** 0.585 1.14 1.30o* 1.16 1.30 1o48 2.24 1.35 1.54 3.58 1.24 1 41* 4.67 1.30 1,48 5.99 1.50 1.71** T-.29** 8000 120 No 2(10000~) None 1.57 None 1.21 None 0,846 0,612 l1.12 1.32 1 12 2.18 1.05 3.52 1.34 4.80 1.29 5.47 1.52 T-30** 8000 110 Yes 12(11000C None 2,17 None 2.10 None 1.95 1 1.0 1.82 3.88 2.12 6.21 2.28 7.02 2534 * Excellent wetting **:Poor wetting; Data not plotted in Figure 10 *** Run not at equilibrium at 800QC; Data not included in Figure

Data: Effect of Palladium on Carbide Decomposition Run Tempp,~C Time-hr Flux Length of Atomic % At % Weight Preheat Palladium U U T-20 8000 96 Yes 1(100lOOC)None 1.34 1.53 NQne 1.64 1.87 2-459 2.30 2.62 3.;66 2.28 2,60 T-22 800~ 93 Yes 5(1000~C)None 1.24 1.42 None 1.10 1.26 None 1.34 1,53. 0.543 0o. 561 0.64 1.045 1.47 1,68 2.48 1.32 1.51 2.61 1.68 1.92 3.65 2.27 2.59 5.30 1.79 2.04 T-24* 8000 57 Yes 7(1000C)None 1.71 1.95 None 1.66 1.89 None 1.26 1.44 0.725 1.61 1.84 1.15 1.79 2.04 2.08 1,73 1.97 3.21 1.96 2.24 3,86 2,08 2.37 4.94 2.16 2.46 T-28** 8000 55 No 2(10000C)None 1.63 1.86 None 1.66 1,89 0.459 1.54 1.76 1.55 1.62 1.85 2.61 1.61 1,84 4,30 1,03 1.18 T-352* 8000 107 Yes 2 (10000C)None 0.839 0,957 None 0,768 0,877 2,58 1.237 1.41 3.34 - 1.10 1,25 4.95 1.87 2,13 5.05 1.20 1.37 Run T-24 yielded nost consistent data; used to evaluate UPd) in Figure 12, ** Poor wetting

APPENDIX B -CALCUIATIONS 1) Calculation of Activity Coefficient of Urani.um in Bismuth. For reatction U + 2CPUC2 ~ = - RTlnK a K aUC2 a 2 aU aC Standard States: C - pure solid at 8000~C UC2 - pure solid at 8000 U - pure solid at 8000C 1 1 Ka = a NU-U for decomposition in bismuth ~-a - aU NU7U Using AO = - 29,610 + 1.5 T from Grieveson and Alcock log K -29,610 + 1.5 (1073) 5- 7 og Ka 1'.987)(1.073)(2'33 K = 501 x 105 U NU =.o0.0186 (20)= 0o.o0163 = 5.01 1 = 1.22 x 10YU (5.01 x 10_) 1.63 x 10o2 2) Partial Molar Enthalpy Change of Uranium for Dissolution of UBi2 Reaction: UBi2 — o U + 2Bi -82

Along liquidus the activity of uranium is assumed constant. Assume activity to be a function of T and Nu. The total differential for activity can be expressed as: da = (dT T dN+ (dT = 0 N)T dT)a dTN ~''\dT/ T + =0 a =N7 N T a+ N (d a ( N+ wdT) =0 Iy) = O (e.g. Henryts law region), then I ~d 1 dy Id lxlNy d ln y7 or I -7)ar 1n YU\ A Hu Since W7 R / 1n N HU(soln?1kH(UBi,) 3) Determination of Ulninvestigated Portion of C'u-U-B1 Liquidus (Cu) 8Jn7 a n 11 Cu Cu

-84bu ( Cu) (U) a in 7cu 1in Ncu ut C..TJ CU a in NC - il"O For NU = ~0025 in. Atomic % Cu = ln 1208 = 254 c =o0 A in A-tomnc % Cu = (11,,0(0O0025) -0O0275 in Atomic % Cu 2 =54 = -00275 in Atomic % Cu = 2.51 Atomic % Cu = 12,4 For Nu = oo6 I.n Atomic % Cu + 2054 O oC66; 2047 Atomic % Cu ii0.9

APPENDIX C -85

-86WEIGHT PER CENT URANIUM 10 20 30 40 50 60 70 80 90 1600 I!,48 TWO MEL 1400-14500 4 1400 T 1200 ___ u. 5!L}._ j(98)j / I I/ L~~~~~~~~~~~~~~~~~I~~~~~I 12001 1200'-'27 ( 30)/L — ".'1 ~~~~~~~~130 "27("30)~ 1130 ~ ~1133o 1100 --- (99 22(21)4. 10100 1000 / / /- _ 900 --- _ "' 800 0m._ ~~~~~~~~~7700 I,w 700 0. 2 _____ 6500 I — 600 ca-U 500 m in tZ 400 300 30.02 (0.02) 2700 273 200 0 10 20 30 40 50 60 70 80 90 100 Bi U ATOMIC PER CENT URANIUM Uranium-Bismuth Phase Diagram

WEIGHT PER CENT BISMUTH 192Q 310 49 505560 65 79 74767 0808284868 90 9 94 96 98 1100 82 88 9 HEYCOCK,NEVILLEREE I 1000 -A HIORNS, REE 2.x I I I Ix PORTEVIN,REE3 G JERIO MIN,REE 900 800 700 o~~~~~~~~~~~~~~~~~~~~~~~~~~~~O ljJ ~w x a 600 w a. w 500 400 300 2700 _ _95 (99.85 2710 200 0 10 20 30 40 50 60 70 80 90 100 Cu Bi ATOMIC PER CENT BISMUTH Copper-Bismuth Phase Diagram

WEIGHT PER CENT URANIUM 1020 30 40 50 60 70 80 90 I 1 1 300O T i _ _ I 1200 m =! TWO MELTS 0 1-22.5(52) 1/II(I I ~~~I110032 1083 _ 1080 1100 o _10520 9 _ _ (- 98.5) -~ -''C o~ 00 1 000 LJQI 9590,~L~ ~')' I I I 3. W Ix ~95 8900 e ~ ~ L.~so i 1760~ 700 0~~~~~~~~~~~~~~~~~~~ 650~ ~ o.o- 65 0~ ~~-a G e 600 0 10 20 30 40 50 60 70 80 90 100 Cu U ATOMIC PER CENT URANIUM Copper-Uranium Phase Diagram

-89WEIGHT PER CENT PALLADIUM 8_ 20 30 40 50 60 70 80 90 1800 1700 16400 1600 i! \I ~ 1552 II 1k''525 1500-w II I II i 1400 ~1 L II. II 0 1300 00 i 85(72 II 1305:5 Id ~~~~~~~~~~~~~~~~~~~~~~~II tLT~ ~ ~ ~ ~~~~~T 1200 <QI 78.0(61.4):11330 CL ~~~~~~~~~~~~~~1110~100 g I-.I III!Io:o I I I 100 w 1 1477 7 II I. II Pd ii r 5.0(2.3 ___ 37.4(21.1 90 T Ii9700' i II II 98~~00 78(6180+ II ____ 756~4_ 7 72p ~00 +1 6650 _____ I 6 6 I II 600 0 10 20 30 40 50 60 70 80 90 100 U Pd ATOMIC PER CENT PALLADIUM Uranium-Palladium Phase Diagram

WEIGHT PER CENT CARBON 2 3 4 5 6 7 8 910 121415 2900.... i L U N 2700 2500 23900 O /f: I, 2300,/ I 0 i \ I I 2100 / i -I 0~~~~~~~ O 1900 WEIGHT PER CENT BISMUTH W 1775 I 1700 J W: I l w M co co CL c 1500 I 0 sO 1300 - Nm 0 11330 __ __ 11 3.3 1100 (Pd) 900 0 I~~~~~~~~~~~~~~~~~~~ 770 __ _ 0 10 20 30 40 50 60 70 s U if 1 I I O0 ATOMIC PER CENT CARBON 0 20 40 60 8 0 PdBi ATOMIC PER CENT BISMUTH Uranium - Caroon Phase Diagram Palladium-Bismuth Phase Diagra 900 (P),~!1~~~~~~~~~~~~~~~~~a a- I I I I I I Ir I I I ~~~ ~~~~~~~~~~I.,,'~ 5~~~~~~70~. O eO 20 30 40 50 60 fO 80? 31~~~~~~~' "n \ 70 0 665niu - __po Ihs iga.aldu-"imt hs iga

U 90 10 80 20 70 / v v 30 8 ~ ~~~~~~ 90 60 50 2 SOL D PHASES o~~~~~~ED o/ r d50 5P 3-40 60 x~~K / \rU~Bi2 30, \ 70 UPb 3 b P\+ UBi + L 2 v soVv P UBi2+ IL 10. - U i+L 9c Pb Bi~~~~~~~~UBi+ 90 80 70 60 50 40 30 20 10 LEA —t~D, o/o Urqnium-loLasad-Bismut Phase- -:Diagrgm

-9290 10 80 20 o US +L 5 30 60 id \7J A i d 80 5090 70 50 6 Uranium-Tin-Bismuth Phase Diagram 30 70 S ISns + L io0/ v V V 1 -~5~ 80 Sn Bi 90 80 70 I. 60 50 40 30 20 10 TIN, a/o Uranium-Tin-Bismuth Phase Diagram

Appendix D A. SPECTROPHOTOMETRIC DETERMINATION OF URANIUM IN BISMUTH, I. INTRODUCTION Uranium is determined spectrophotometrically at 412 mj and 390 m; the dibenzoylmethane-uranium complex. The applicable range is 50 to 120 ig of uranium; up to 300 mg of bismuth can be tolerated. This method has been used for samples containing 99+% bismuth and 1000 ppm uranium with respect to bismuth. The bismuth is complexed with diaminocyclohexane tetraacetic acid (DCTA), and the uranium is complexed with dibenzoylmethane. The dibenzoylmethane-uranium complex is extracted into amyl acetate, and the absorbance of the resulting organic layer is measured at two wavelengths. II. APPARATUS Spectrophotometer, Beckman Model DU with tungsten lamp and bluesensitive cell; Absorption cells, 1 cm. path length, Corex or quartz (Pyrex cells also seem to be satisfactory). III. REAGENTS Nitric acid, concentrated, reagent grade Ammonium hydroxide, 5 M Hydrochloric acid, 3 M Saturated ammonium chloride solution Bromcresol green indicator solution, 0.04% -935

Dibenzoylmethane solution, 1% in ethanol (absolute or 95%). Amyl acetate, reagent grade. The isoamyl acetate was the only amyl acetate available in a reagent grade; consequently, it was used since the original reference did not specify which ester was used. Diaminocyclohexane tetraacetic acid (DCTA) solution: 100 g. is dissolved in 300 ml of 5 M NaOH; after dissolution, the pH is adjusted to 5.2 with HC1 and NaOH; finally the solution is diluted to 500 ml. IV. PREPARATION OF SAMPLES 1. Weigh the bismuth sample and dissolve in concentrated nitric acid. 2. Transfer to a volumetric flask and dilute to the mark with water. This dilution should be such that an aliquot no larger than 10 ml will be required for an analysis. V. PROCEDURE 1. Withdraw an aliquot (no more than 10 ml) containing 50 to 100 gigms. of uranium and no more than 300 mg of bismuth. Transfer to a 60-ml separatory funnel. (See Note 1) 2. Add water to bring volume up to 10 ml. 3. Add 5 ml of the DCTA solution. (See Note 2) 4. Add 3 drops of the bromcresol green indicator solution.. 5. Add dropwise ammonium hydroxide, 5 M, and/or hydrochloric acid, 5 M, until the solution just turns from blue to yellow. 6. Add 5 ml of saturated ammonium chloride solution. 7. Add water to make the volume about 20 ml. (See Note 3)

-958. Add 1.00 ml (pipet) of dibenzoylmethane solution while rotating the funnel. (See Note 4) 9. Immediately add from an eye dropper 12 drops of 5 M ammonium hydroxide. 10. Add 10.00 ml (pipet) of amyl acetate. 11. Shake the contents of the funnel thoroughly and allow the two layers to separate. (See Note 5) Centrifuge for 1 minute. 12. Pipet enough of the organic layer (top layer) into the absorption cell to fit it. 13. Measure the absorbance at 412 my and at 390 my against a blank which has been prepared in the same manner as the sample using water instead of uranium containing sample. (See Notes 6 and 7) VI. NOTES AND PRECAUTIONS 1. e A 10-ml aliquot is maximum because the final volume of the aqueous layer is 20 ml, and 10 ml of reagents must be added. 2. 5 ml of DCTA will keep in solution about 300 mg of bismuth. If a precipitate does occur, add more DCTA. 3. The diluted solution should be clear at this point; if not, probably insufficient DCTA was added. 4. After the additi.on of dibenzoylmethane, the solution will be turbid and a small amount of precipitate will be present. Keep the dibenzoylmethane in ethanol solution tightly stoppered to prevent volatilization.

5~ The two layers should be clear now with the aqueous layer colored blue and the organic layer yellow. The intensity of the yellow color will be determined by the amount of uranium present. 6. Unless one has tightly stoppered absorption cells, it will be necessary to work rapidly during the photometry because of the volatility of amyl acetate. 7. It has been found that good soapy water foll.owed by thorough. rinsing with water and a final rinse with acetone will clean the absorption cells very nicely. VII. CALCUIATIONS The concentration of uranium is read from a calibration curve of absorbance versus uranium concentration which is best prepared from a series of standard uranium solutions. These solutions should be treated in, the same manner as the sample and their absorbances read at the two wavelengths. = g of U read from calibration curve x dilution volume x 100 1000 x sample -wt;. in mg aliquot volume - g of U x 106 x dilution, volume ppm U =.. 1000 x sample wt. in. mg aliquot volume The concentration of the uranium as read from the 412 mCi curve should agree with that as read from the 390 mi curve. The two:readings taken together should show if any interference is present. The ratio of absorbances A(412 mp)o A(390 ma), is constant at 1.9 to 2.0 over the range studies, (50 to 120 ig U).

-97VIII. REFERENCES (1) Finston, H. Lo~ Brookhaven National Laboratory, private communicatibn. (2) Yoe, J. H.: Will III, F.; and Black, R. A.; Anal. Chem. 25,9 1200 (1953).

-98B. SPECTROPHOTOMETRIC DETERMINATION OF'COPPER IN BISMUTH ALLOYS I. INTRODUCTION The spectrophotometric method for the determination, of copper reported by G. Frederick Smith and W. Ho McCu:rdy, Jr. Anal. Chem., 24, (2) 371 (1952) and Y. Ysh.ihara and Y. Taguchi Japan Analyst, 6, (9), 588-89 (1957) has been successfully applied to the determination of copper in the presence of urani.um and large quantities of b:ismuth. Copper is extracted from an aqueous solution of pH 5-6 as a neocuproine complexl with isoamyl alcohol and measured at 454 mo. Berrrs law i.s obeyed over the rangeof 0.15 to 1.0O6 ppm copper in isoamyl alcohol. The -method is quite selective for cop-per and may be used for copper in the presence of many cations and anions. A minor modification of the original procedure was made in order to more rapidly determine the pH of the aqueous phase prior to extraction2. II. IN.ERFERENCES Smith et al, report that no other cati.on other than the cuprous:ion was found to form a colored. complex which was extractable under the conditions employed. The common anions chloride, sulfate, nitrate, perchlorate. tartrate, eitrate and acetate do not interfere. Many other anzions which either react wi.th hydroxylamine or give a yellow colored solution, may'be elliminated. employing suitable conditions. The following cati;.ons were found to produce no detectable interference~ Cation Weight Ratio Cati+on Cu+ 1 O2 +2 234 1

-99III. APPARATUS Beckman DU Quartz Spectrophotometer with Tungsten lamp. IV. REAGENTS (1) Concentrated nlitric acid. (2) p - Bromocresol GIreen., 0o04%, (W/v) aqueous sol.ution. (3) Ammonium Hydroxide - 5 M. (4) Hydrochloric acid.1 M. (5) Hydroxylamine hydrochloride, 30%, (W/V ) aqueous solution. (6) Sodium Citrate, 30% (W/V) aqueou.s solution. (7) Neo - cuproinre, 0.1% (W/V) ethano:l- water solution (l-o9 by volume). (8) Ysoamyl alcohol3 (Reagent Grade) V. PROCEDURE 1.o Dissol1ve te the bismuth alloy in concentrated ni.tric aci.d heat to expel n:itrogen oxides, cool and dilute to volume. 2. Transfer an al.iquot of sample containi.n.g 5 to 75 pg copper to a 60 ml separatory:funnel. 5. Add 10 ml hydroxylamine hydrochloride (30%o). 4. Add 2 ml of sodiu:m ctrate (30%)4 5. Add 2 drops of p-Bromocresol. Green5.m 6. Add ammonium hydroxide (5.M) and/o:r hydrochloric acid (1M) till the solution changes:from a blue to a green-yellow color. 7. Add 8 ml o:f sodium citrate. 8. Add 2 ml of neo-cuproine solution (0.1%) with mixing6.

-1009. Make up to a volume of 30 ml with distilled water. 10. Add 10 ml. of isoamyl alcohol and shake for 20 seconds. Allow phases to separate7. 11. Remove a portion. of the organic phase and measure the absorbancy at 454 m~u against a reagent, blank. VI. CALIBRATION CURVE A calibration curve was prepared:from a stock solution of copper (99.9%/) in lN n.itrlc acid. Amounts of bismuth up to 47,860 ig and uranium up to 400 4g were added to as small amount of copper as 1.71 I~g without any noticeable interference. VII. NOTES 1. The complex contains a molar rat:lio of copper to neo-cuproine of 1i2 with copper present as the cup:rous (Cu+l) ion. 2. p-Bromocresol gre.en indicator (pH 3-5) was introduced for determining the pH of the aqueous phase prior to extraction. The indicator exhibited an absorbancy of 0o010/1 drop at 454 mt, a sufficiently low value to permit its'use in the determination by.inclusion. in. the blank. 3. This solvent ex:hibits a nauseating odor and should be used in a hood. 4. Addition of sodium ciltrate prevents hydrolysis and prec:l.piLtati.on o:f heavy -metals during the subsequent neutralization operation. 5. Exactly 2 drops o:f indicator is used in both blank and sample ($See note 1) o

-1016. Complex formation, for all practical purposes, occurs instantaneous ly 7. Smith et al. performed two extractions, however, it was found in this laboratory that one 10 ml extraction was sufficient to remove the copper complex and proved through reproducib:ility of the reported molar extinction coefficient for the copper complex in isoamyl alcohol. The use of a single extraction significantly decreased the time required to perform the analysis.

-102C. SPECTROPHOTOMETRIC DETFTRINATION OF PAItADIUM IN BISMUTH ALLOYS I. INTRODLCTION The spectrophotometric method for the determination of Palladium reported by Oscar Menis and T. Co Rains, Anal. Chem., 27, (12), 1932 (1955) has been aplied to the analysis of bismuth alloys cont;aining Palladum concentrations as low as 20 ppm. The presence of Uranium and high concentrations of bismuth in the alloys necessitated a more rigid control of the acidity and time for color formation than is implied in the original reference. The absorbancy of the palladium-alpha-furildioxime complext is measured at 380 mi in chloroform, following extraction from a 13 N acid aqueous mixture. Time for color formation is exactly 5 minutes, if the time exceeds 30 minutes, bismuth causes interference An. acidity of 1.3 N in the aqueous phase prevents the precipitation of bismuth in. any of the alloys analyzed, including alloys containing the minimum amount of palladiume c ao 203 ppm and allows complete color development to occur in 5 minutes. IIo IoNTERFERENCES The following cations were reported to have no effect in quantities giveno

-103Elements Amt; present Mg Pd ( ) ir+4 0,06 59.0 Pt+4 0.7 59.0 Fe+g 2.0 59.0 Cd++ 2.1 59.0 Os+4 3~3 59.0 Cu+2 307 59'.0 Na+l 308 590 o U +6 8.42 590 Th+4 25,0 59.0 Rh+)3 0.8 59 o0 Zr+4 1o9 59.0 Pb+2 2.3 59.0 Mo+6 4O.0 59.0 Cr+3 4 3 59.0 Mn+2 lo0 59.0 Ni+2 1ol 59.0 Au+3 205 59.0 Co+2 2,52 59.0 Bi 321.0 57~5 Anions mg present Pd ( ) F 7.15 59~0 S04- 6,90 59.0 NO - 7.29 59.0 C1.4- 7.15 59.0 CN- 4.0o 0

Th~kunan It Q tz -ePatrophotlometer with tungsten lamp; absorption cells 1 cm..;'50 ml burette. IVt REAGENTS 1. Alpha - Furildioxime3 1% solution. Prepare by dissolving 1 gmn of the reagent in 30 ml of ethanol and diluting to 100 ml with watero 2o 00500 N Sodium Hydroxide, 3. Concentrated nitric acid. 4. Phenolphthalein Indicator. 50 Chloroform. Reagent Grade. V. PROCEDURE 1. Dissolve the sample in concentrated nitric acid and diluteg to volume. Final normality of the sample- solution should not exceed 3 N. 2. Remove an aliquot of sample solution and determine the free acid content using standard base and phenolphthalein indicator, e g. tdtal equivalents of base required minus the number of equivalents of bismuth pre s-ent4. 3. Transfer an aliquot of sample solution containing 5.75pg to 57,5 pg Pd to a 60 ml separatory funnel. 4. Add 2 N hydrochloric acid and water, sufficient to yield 35 ml total volume of a 1.3 N *solution The reagent blank should be

-105the same acidity as -the sample'solution and should contain. 321 mg blismuth40 5. Add 1 ml of - furildioxime (1% alcohol-water solution) while rotating the funnel. Allow the mixture to stand exactly 5 minutes. 6. Add 10 ml of chloroform, shake for 20 sec., allow the phases to sparate and transfer the!ower organic layer to a 25 ml volumetric flask. Time elapsed between the addition of alpha-furl.ldioxime and the final extraction should not exceed 10 minutes, 7. Dilute the organic extracts to 25 ml with. chloroform and mix thoroughly. 8. Add 0.3 gms. sodium sulfate, anhydrous, to chlodroform extracts to remove traces of water. 9. MeAsure the absorbancy of the sample at 380 mu against the reagent blank containing bismuth. VI. CALIA0ON C1RVE A stock solution of palladium was prepared by dissolving the pure metal (99.9,) in concentrated nitric acid contal.ning a small amount of hydrochloric acid, heating the mixture till the evolution of Cl2 ceased and diluting to volume. Aliquot portions of the palladium stock solution were then treated according to'the above procedure. Bismuth and uranium were added t-o the palladium aliquot's and produced no effect upon the observed absorbancy of the palladium alpha furlldioxime complex under the conditions of acld.i.ty and t+ime- foxr co-lor develropment cited in the procedure.

VIIo CALCULATIONS L. Normality of sample solution:. wtosample gins. aliquot sample ml N(as free acid) = (Volume Base ml)(N Base)..saple. alut s ml 69o6 Dilution ml (aliquot sample ml) 20 Volume of 2.00 N hydrochloric acid required to prepare 1.3 N sample solution for extraction:~ ml of 2.00 N HC1 required = (35 ml)(13 N) - (aliquot sample ml)(Ns) (2,00 N) NS = Normality of sample as determined from (1)o 3. Percent composit;ion,: % Pd = (.g Pd) (Dilution ml) (100) (ml aliquot) (gms. sample) (106) VIII. NOTES 1. The pall.adluma-alph.afurildioxime complex is stable in chloroform for at least 24 hours~ 2. Bism uth, cobalt and uranium were found to have no effect on the determination under the conditi:ons outlined in the above procedure,

30 The, eagent must be purifi-ed before use through recrystallization from water, If further purification is desired see, Reed, S. A., and Bank, C. U o, Proco Iowa Acado Scio, 55, 267 (1948). 4. 321 mg bismuth represent the largest amount of bismuth encountered with the alloys studied in the aliquot used for the analysis.

BIBLIOGRAPHY 1. Ahman.n, D H. and Baldrin, R. R. "The Uranium-Bismuth System", CT-2961, (November 12, 1945). 2. Alcock, C..B. and Richardson, F. D. "Dilute Solutions in Molten Metals and Alloys", Acta Metallurgica, 6, 385,(1958). 3. Armour Research Foundation? Solubility of Uranium in Liquid Bismuth. A Final Report to The Babcock a:d. Wilcox Co., foF CorhtractelRe'ebarch Chicago, (1959) 4. Bareis, Do W. "L'Lquid Reactor Fuels- Bismuth-Uranium.System," BNL-75, (September 15, 1950). 5. Barton, P. J. Private Communication. 6. Bever, M. B. and Floe, C. F. Trans. Am. Inst. Mining Met. En rs., 156, 149, (1944). 7. Bowman, M. G. Private Communication. 8. Brewer, Len. Private Communication. 9. Chipman, John. "Activities in Liquid Metallic Solutions'," Discussions Faraday Soc., No. 4, p. 3-49, (19948). 10. Chipman., John. "The Thermodynamics of Liq~u.id Metallic Solutions," A paper from Thermodynamics in. Physical Metallurgy, ASM, CIeveland0, Ohlo,(1950) 11. Chipman, John. "Atomic Interaction in Molten Alloy Steels," J-. Iron Steel Instltute, 179, 97, (1955). 12. Cosgarea, And:rew Jr. Some Thermodynam:ic Properties of Uranium-Bismuth Alloys. Ph.D. Thesi.s, University of Michigan, (1959). 135 Corrosion of Uo2C and UC2 by Bismuth," A Report:for Technological. Research Section II, C -423, (January 1943). 14. Cotterill, P. and Axon, H. J. "The Constlitution of Bismuth-Uranium Alloys in. the Compositi.on Range 0-35 Atomic Percent Uranium," J. Inst. of Metals, 87, 1.59-160, (1958). -- 15. Danil.ova, A. I. and Danilov, D. I, Problemy metallovendiya i fliziki metallov (Problems of Metallography and the Physics of Metals), part 2, 32, Metallurgizdat, (1951) 16. Darken, L. S. "Application of the Gibbs-Duherm Equation to Multicomponent Systems," Am. m Chem. Soc., 72, 2909, (1950)e -108

-10917. Darken, L. S. and Gurry, R. W. Physical Chemistry of Metals, McGraw-Hill. New York, (1953). 18. Egan, J. J. and Wiswall, R. H. Jr. "Applying Thermodynamilcs to LiquidMetal-Fuel Reactor Technology', Nucleonics, 15, 104, (1957)o 19. Es in., 0. A. and Geld, P. V. Physical Chemistry of Pyrometallurg:ical Processes, Pa20; IT. The Interacti.on of Ll.quids with Gases an.d Solid Phases, C-tr -3439- 9 1954. 20. Ferro, Ricardo, "Alloys of Uranium with. Bismuth II," Atti. accad. Nazl. Lincei. Rend.., Classe sci.. fiso mat. e nat., 14 89-94 (1953). 21. Frost; B. R. T.'"The Structure of Li.qu:d Met;als," Progress:i.n Metal Physics, Vol..., Inte:rscience Publ.ishers Inc. New Yo:rk ( 1953722. Gingrich, N. S. and Henderson, R. E. Journal of Chemi.cal Physics, 20, 117, (1952). 23. Greenwood, G. W. "The Solub:l.l.t:i.es of Uranium and Thorium in Li:qui.d Bismuth," A oE.R.E. M/R 2234. (June: 1957) 24. Grieveson, P. and Alcock, C. B. Private Communication. 25. Gross, P., Levi, D. L. and Lewin., R. H. "'Activities in Urani.um-Bismuth Alloys and the Free Energies' of Uranium-Bismuth Compounds, A paper from'The Physical Chemistry o:f Metallic Soluti.ons and. Inte:rmetall.l.c Compoun~ds, Lon.donro He-r Majest'Ss Statil.on.ery O:ffice, (1959). 26. Hansen, Max. Con.sti.tut;ion of Bi.nar:y Al1loys, McGraw Hill New York'(1958). 27. Hargreaves, BR, J. Inst.. Metals, 64, 115 (1939). 28. Hendus, H. "Atom Distri:.bution in. Molten Elements Pb,Tl,In,Au,Sn, Ga Bi, Ge, and in the Molt;en Alloys of -the Au-Sn System," Zei.tsch.tft fur, Naturforschung, Vol. 2A, 505,(1.947). 29. Himmler, W. Z. phbysik. Chem. 195, 244, 2539(1950). 30. Hume.-Rothery, W. The Structure of!Metals and Alloys, The Institute o:f Metal.s, Londonn(1c947l-) 31. Jones, H. Proc. Roy. Soco (London) A1k44k, 225,(1934). 32. Kitchener,J. A.o Bockris, J. 0o M. an.d Spratt, D. A. "Solutions in Li -quid Iron," Trans. Faraday Soc., 8 608, (1952). 33. Kleppa, 0. J. "Thermodynamics and Properties of Liquid. Solutions, " Liquid Metals and Solidification_, ASM, Cleveland, Ohio, (1958)

-110_ 340 KubaschlewlRkil 0O and Cotterall, J. A. Thermochemical Data of Alloys, Pergamon Pre'ss, London (1956). 35. Kubaschewski, 0. and Re-inartz. Z. Elektrochem, 52, 75,(1948) 36. Latin, A. "The Structure of Liquid Metal.sO A Review,' J. In.st Metals, 66., 177( 194o). 537. Mott, B. W. "Liquid Immiscibility in Metal Systems," Phil._ Ma 2, Ser. 8, 259,(1957). 38. Mott, N F. and Jones EH. The Theory of the Propertlies of Metals and Alloys, Dover, New Yo:rk7.1958) 39. Oh.tani, M. and Gokcen, N. H. "Thermodynamic Interaction Parameters of Elements.in Liquid Iron," Trans. AIME, 21.8, 533,(1960) 40. Ople, W. R. and Grant, N. J. Trans. Am. Inst. Mn.ing Met. Engrs 188, 1237,(1950) - 41. Pauling, Linus. The Nature of'the Chemical Bond. Cornell Univers:i'ty Press, Ithaca, New Yor:., (1948)- - 42. Radchenko, I. V. "The Structure of Liquid,Metals," Advances of the Physical Sciences, 61, 249,(1957). 45. Raynor, G. V. "Progress in -the Theory o:f Alloys," Progres's'in. Metal. Physics, Vol. 1., In.terscience Publishers Inco, New York,~9~-~7 44. Ric:hardson, F. D. "The Thermodynam:ics of Metallurgical Carbides anid Carbon in I:ron. J. Iron Steel. Inst., 175, 41.,(1953) ~ 450 Rough., F. A. and Bauer, A. A. Constitutional Diagrams o:f Uran.um and Thorium Alloys, Addison-Wesley,-eading, Mass., 195-.46. Schweitzer, D. Go and Weeks, J o R. "Li.quid Metal Fuel Constit'.t1.ion II. Liquidus Curves of the Uran.ium-Zirconum-:Bismuthb System," 1960.ASM Preprlnt No:196. 47. Schweitzer, D, G. and Weeks, J Re BL.:l.qud. Metal Fuel Const:itut ion I. T.he Solub ility of Uranium in Bismuth," 1960 ASM Preprint No. 195. 48. Sieverts, A. and Krumbh;aa, H. Ber. deuto. chemo Ges. 4 893n(1910). 49. Smoluchowski and Koehler. Ann. Rev. Physic. Chemo,2, 187,(1951). 50. Teitel, R. J. "Uranium Bismuth.System," Trans,. A..M.E., 1231360 (1957). ___

-11151 Teitel, R. Jo Work reported by Rough, F. A- and Bauer, A Ao Constitutional Diagrams of Uranium and Thorium Alloys, Addison Wes2ey, 90 and 91l(l958)T. 52~ Turkdogan, E. To "Interaction of Solutes in Liquid and Solid Solution. in Iron.," J. Iron Steel Inst. 183, 69,(1956). 53. Turkdogan, E.'T. and Leake, L. E. "Carbon Solubility in. Fe-P, Fe-Si, and Fe-Mn Melts," J. Iron Steel Inst., 179, 39,(1955). 540 University of'Michigan Research Institute. Studies on Uranium Carbide Formati.on. A finxal report submitted to $he Babcock -and Wilcox Co. Ann. Ar'bor ( 1959) o 55. Vineyard, G. H. "The Theory and Structure of Liquids," Liquid Metals and Solidification, American Society for Metals,(19-57)56. Wagner, Carl.. Thermodynamics of Alloys, Addison-Wesley Press, Cambr.dge Mass.,( (1952.952 57- Wagner, Carl. "Thermodynami.c Investigation on Ternary Amalgams, J. o;. Chem, Phy__es,19, 626.,(1951). 58. Weeks, J. R., Klamut, C. Jo., Silberberg, M., Moil.ler, H. E. and Gurinsky, D. H. Corrosion Problems with. Blismuth-Uranium Fuels. Geneva Conference Paper, "P/1i8,(-953' ) 59. Weeks, J. R. Private Communication.

UNIVERSITY OF MICHIGAN III 111 I 190 I 1 115 021 8998111111ilEI 1 3 9015 02493 8998