No 8963 
 

 
 
 
 ^.-'^'^ 
 
 <. 
 
 o V 
 
 »- o. 
 
 
 
 .♦^"V 
 
 v ^ 
 
 <• 
 
 - -^o. .o^,-^^.% .v^\^r.-.'^o. ./.^J.^%'^.. ..^^'A'i:. V .^^ •-*^'* -^ ^ 
 
 
 '^^ ••«<> 
 
 
 
 iP" .LV,L'* "^^ 
 
 
 
 
 'b *vT7V* A 
 
 C° ."l^^'. °o 
 
 ^'^* 
 
 .V 
 
 
 
 c'?^" »'<!!Siiai'. -^^ A^ /^^^A*^ >^ ,^'^" 
 
 *'oV'' 
 
 
 
 
 »v 
 
 '<J> * « o ' ^-ft 
 
 o» •■ , 
 
 0* .'VL'» ^ 
 
 
 
 
 <=^ aP 
 
 
 
 <• o 
 
 ^Jt^rfi oV'^^Mll?^'- 'J'^^'*- -^i^M^^^. ^^J^r^ oV^^SiA'- ^>!^<-^ 
 
 
 <^°^ 
 
 
 0^ oo-,.v -^* 
 
 '^o^ 
 
 "oV 
 
 0' . vv: 
 
 v^-^ 
 
 -^^ -^^ 
 
 
 /\v;z^^\ .^°/^i>- /^>^%\ .^°^•^•°- /^>^^\.^ -" 
 
 ^t^ 
 
 
 
 
 */ V'^\** '>v-^--*/ *^.'-^\** 'V'^-* 
 
 ^:o<i^ 
 
 
 
 
 
 <r^ '°.« 
 
 
 
 o^ *.Tvr» A 
 
 "^^-^ 
 
 A" .<■" . •* 
 
 
 -O^ r « ■ • -t O 
 
c"^ 
 
 "oV" 
 
 
 v.'i"^"' /i§ia\ ^^..♦^ yjmA\ "v.^* /^I^\ ^^..♦^ .^^(^tto ^^^..^^ 
 
 
 
 
 
 :. -^Ao^ o^, 
 
 
 <J> ' ' » o ' . 
 
 'U.Q^ 
 
 'bV 
 
 
 
 
 % -^^0^ f- 
 
 
 
 .4> . V ' . . "<P. 
 
 
 
 -- S . . . <A j^ -^ 
 
 
 
 <. '0-.7-- 0^^ % 
 
 
Bureau of Mines Information Circular/1984 
 
 Review of Anhydrous Zirconium-Hafnium 
 Separation Techniques 
 
 By Robert L. Skaggs, Daniel T. Rogers, and Don B. Hunter 
 
 2 UNITED STATES DEPARTMENT OF THE INTERIOR 
 
Information Circular 8963 
 
 Review of Anhydrous Zirconium-Hafnium 
 Separation Techniques 
 
 By Robert L. Skaggs, Daniel T. Rogers, and Don B. Hunter 
 
 UNITED STATES DEPARTMENT OF THE INTERIOR 
 William P. Clark, Secretary 
 
 BUREAU OF MINES 
 Robert C. Norton, Director 
 
1 
 
 Library of Congress Cataloging in Publication Data: 
 
 Skaggs, Robert L 
 
 Review of anhydrous zirconium-hafnium separation techniques. 
 
 (Information circular / United States Department of the Interior, Bu- 
 reau of Mines ; 8963) 
 
 Bibliography: p. 22-25. 
 
 Supt. of Docs, no.: I 28.27:8963. 
 
 1. Separation (Technology). 2. Zirconium tetrachloride. 3. Hafni- 
 um tetrachloride. I. Rogers, Daniel T. II, Hunter, Don B. III. United 
 States. Bureau of Mines. IV. Title. V. Series: Information circular 
 (United States. Bureau of Mines) ; 8963. 
 
 ■--Tf^^a5LjU4 [QD63.54] 622s [669'.735] 83-20940 
 
CONTENTS 
 
 Page 
 
 1 
 
 2 
 
 4 
 
 5 
 
 7 
 
 10 
 
 12 
 
 13 
 
 16 
 
 17 
 
 17 
 
 18 
 
 18 
 
 21 
 
 22 
 
 Abstract 
 
 Introduction 
 
 Separation techniques based on relative volatility 
 
 Thin-film sublimation 
 
 Extractive distillation from molten salts 
 
 High-pressure liquid-vapor distillation 
 
 Chemical methods of separating hafnium from zirconium 
 
 Preferential reduction of ZrCl^ 
 
 Fluoride redox exchange 
 
 Chloride-oxide exchange reaction 
 
 Preferential decomposition of alkali metal salts 
 
 Differential oxidation of chlorides 
 
 Ranking of processes 
 
 Conclusions 
 
 References 
 
 ILLUSTRATIONS 
 
 1. Diagram of fractional sublimation apparatus used by Goldberger and Gillot.. 6 
 
 2. Diagram of Spink extractive distillation process 8 
 
 3. Relationship between decomposition temperature of hexachlorozirconates and 
 
 metal ion radius for selected alkali and alkaline earth elements 9 
 
 4. Pechiney process for extractive distillation 10 
 
 5. Vapor pressure of ZrCl4 and HfCl4 11 
 
 6. Density of coexisting liquid and vapor phases of ZrCl4 and HfCl4 11 
 
 7. Apparatus used by Ishizuka to separate HfCl4 from ZrCl4 12 
 
 8. Pressure of the tetrachloride gas over HfCl4(s), ZrCl4(s), and lower chlo- 
 
 rides over Zr 15 
 
 9. Thermal stability of the lower chlorides of zirconium 15 
 
 TABLES 
 
 1. Typical analysis of crude ZrCl4 3 
 
 2. ASTM specification B349-73 for nuclear-grade zirconium sponge 4 
 
 3. Chemical requirements for reactor-grade hafnium metal 4 
 
 4. Physical constants of hafnium and zirconium tetrachlorides 4 
 
 5. Comparison of the thermal stability of the alkali chlorozirconate and chlo- 
 
 rohaf nate compounds 10 
 
 6. Operating conditions and results for a three-stage separation of HfCl4 
 
 from ZrCl4 by the method of Frampton and Feldman 16 
 
 7. Comparison of the distillation processes for separating H.fCl4 from ZrCl4... 19 
 
 8. Comparison of the chemical processes for separating HfCl4 and ZrCl4 20 
 
 t/F JAH 
 
 nwfi 
 

 UNIT OF MEASURE ABBREVIATIONS USED IN 
 
 THIS REPORT 
 
 
 
 A 
 
 angstrom 
 
 kj 
 
 kilo joule 
 
 atm 
 
 atmosphere 
 
 kJ/mol 
 
 kilo joule per gram mole 
 
 °C 
 
 degree Celsius 
 
 kPa 
 
 kllopascal 
 
 cm 
 
 centimeter 
 
 lb 
 
 pound 
 
 cm^/mol 
 
 cubic centimeter per 
 gram mole 
 
 Ib/h 
 
 pound per hour 
 
 
 
 m 
 
 meter 
 
 cm/h 
 
 centimeter per hour 
 
 
 
 
 
 mm 
 
 millimeter 
 
 ft 
 
 foot 
 
 
 
 
 
 ym 
 
 micrometer 
 
 g 
 
 gram 
 
 
 
 
 
 mm Hg 
 
 millimeter of mercury 
 
 g/cm^ 'h 
 
 gram per cubic cen- 
 
 
 
 
 timeter per hour 
 
 mln 
 
 minute 
 
 h 
 
 hour 
 
 mol pet 
 
 mole percent 
 
 in 
 
 Inch 
 
 pet 
 
 percent , usually weight 
 percent 
 
 K 
 
 Kelvin 
 
 
 
 
 
 ppm 
 
 part per million 
 
 kcal/mol 
 
 kllocalorle per gram 
 
 
 
 
 mole 
 
 psl 
 
 pound per square Inch 
 
 kg/cm^ 
 
 kilogram per square 
 
 
 
 
 centimeter 
 
 
 
REVIEW OF ANHYDROUS ZIRCONIUM-HAFNIUM SEPARATION TECHNIQUES 
 
 By Robert L, Skaggs, Daniel T. Rogers, and Don B. Hunter 
 
 ABSTRACT 
 
 Sixteen nonaqueous techniques conceived to replace the current aque- 
 ous scheme for separating hafnium and zirconium tetrachlorides were re- 
 viewed and evaluated by the Bureau of Mines. The methods are divided 
 into two classes: separation by fractional volatilization of the tet- 
 rachlorides, which takes advantage of the higher volatility of hafnium 
 tetrachloride; and separation by chemical techniques, based on differ- 
 ences in chemical behavior of the two tetrachlorides. 
 
 The criteria used to evaluate separation methods were temperature, 
 pressure, separation factor per equilibrium stage, complexity, compat- 
 ibility with existing technology, and potential for continuous oper- 
 ation. Three processes were selected as being most promising: 
 (1) high-pressure distillation, (2) extractive distillation from a 
 molten salt, and (3) preferential reduction of gaseous ZrCl4 . Any of 
 the proposed nonaqueous Hf-Zr separation schemes must be supplemented 
 with additional purification to remove trace impurities. 
 
 'Metallurgist, Boulder City Engineering Laboratory, Bureau of Mines, Boulder City, 
 NV (now with Engineering Department, University of Nevada, Las Vegas, NV) . 
 
 ^chemist, Boulder City Engineering Laboratory, Bureau of Mines, Boulder City, NV 
 (now with Albany Research Center, Bureau of Mines, Albany, OR) . 
 
 ■^Metallurgist, Boulder City Engineering Laboratory, Bureau of Mines, Boulder City, 
 NV (now with Albany Research Center, Bureau of Mines, Albany, OR). 
 
INTRODUCTION 
 
 This Bureau of Mines report focuses on 
 one aspect of the problem of producing 
 nuclear-grade zirconium, via zirconium 
 tetrachloride, in a nonaqueous system. 
 In 1981, reports of promising, nonaqueous 
 Hf-Zr separation work in France and Japan 
 led the Bureau to initiate a project to 
 reexamine the field. During 1982 a lit- 
 erature search covering 1955-82 was con- 
 ducted. The search included HfCl4-ZrCl4 
 separation techniques and related thermo- 
 dynamic data and phase equilibria. Six- 
 teen separation processes were critically 
 evaluated and rated by a semiquantitative 
 method based on cost parameters. On the 
 basis of this evaluation, three processes 
 were identified as promising candidates 
 for commercial-scale separation. The 
 three processes are — 
 
 1. High-pressure liquid-vapor 
 distillation. 
 
 2. Extractive distillation at 1 atm. 
 
 deposits, with Zr:Hf ratios averaging 
 about 50 to 1 (40) . The most common 
 sources of zirconium are zircon (Zr02 
 •Si02) and baddeleyite (Zr02). Zircon, 
 the more abundant of the commercially 
 important minerals, occurs domestically 
 in beach sands of Florida but is also im- 
 ported from Australia and South Africa 
 (40). 
 
 The cost of nuclear-grade zirconium in- 
 got, $11.60/lb in January 1983, is con- 
 sidered high for such a plentiful ele- 
 ment, but there are several reasons for 
 the high cost: 
 
 1. The zircon ore is very stable. Al- 
 though a number of techniques have been 
 proposed for unlocking the ore, only 
 three have achieved widespread use: 
 
 a. Conversion to ZrCN in an elec- 
 tric arc furnace followed by 
 chlorination. 
 
 3. Separation by the preferential re- 
 duction of ZrCl4(g) to ZrCl3(s). 
 
 The techniques reviewed were limited to 
 Hf-Zr separation. In no case was the 
 problem of the removal of trace elements 
 considered. The evaluation is incomplete 
 in that it does not address this impor- 
 tant question. Any nonaqueous Hf-Zr 
 separation scheme will entail the use of 
 additional purification to achieve trace 
 element removal. The most promising 
 technique for accomplishing this is 
 molten-salt scrubbing. 
 
 Zirconium, once considered a rare ele- 
 ment, is actually common in the earth's 
 crust. Current estimates indicate that 
 zirconium makes up 0.028 pet of the lith- 
 osphere and is about as plentiful as car- 
 bon and much more abundant than the less 
 expensive metals copper, nickel, lead, 
 and zinc (37) .^ Zirconium is always 
 accompanied by hafnium in its natural 
 
 ^Underlined numbers in parentheses re- 
 fer to items in the list of references at 
 the end of this report. 
 
 b. Direct carbochlorination in a 
 
 gas-solid reactor. 
 
 c. Caustic fusion. 
 
 In the process used by Teledyne Wah 
 Chang, OR, and by Western Zirconium, UT, 
 carbochlorination is employed twice: 
 first to unlock the ore, and a second 
 time to convert the oxide back to the 
 chloride form after aqueous solvent ex- 
 traction separation of hafnium. 
 
 2. The hafnium content of nuclear- 
 grade zirconium must be less than 100 
 ppm. Hafnium has no perceptible effect 
 on strength, ductility, or corrosion re- 
 sistance, but it negates the utility of 
 zirconium in a nuclear reactor. Nuclear- 
 grade zirconium must have a low neutron 
 absorption cross section (a^, = 0.18 barn 
 in zirconium) , but the presence of even 
 small amounts of hafnium (Oq = 105 barns) 
 greatly increases absorption of neutrons. 
 Separation of these two metals is costly 
 because their chemical properties are 
 nearly identical. 
 
3. The presence of other minor impuri- 
 ties not only affects the ductility and 
 corrosion resistence of zirconium, but 
 also increases neutron absorption. Fis- 
 sionable elements such as uranium and 
 thorium are particularly detrimental be- 
 cause of the degradation of mechanical 
 properties that can occur on exposure to 
 radiation. 
 
 Since mineral acids other than Hf will 
 not attack, zircon (6^, 60 ) , traditional 
 extraction processes rely on high- 
 temperature treatment in the presence of 
 carbon and chlorine. In these processes 
 zircon is reacted with carbon in an arc 
 furnace at 3,500° C to form zirconium 
 carbonitride, while silicon is driven off 
 as SiO gas. The carbide then reacts exo- 
 thermally with CI2 to form zirconium tet- 
 rachloride (ZrCl^), which is condensed in 
 a separate nickel-lined chamber. 
 
 The present commercial process (carbo- 
 chlorination) involves treatment of zir- 
 con sand-carbon mixtures with chlorine at 
 1,150° C in a f luidized-bed reactor (59, 
 61) . The silicon tetrachloride byproduct 
 is readily separated from the less vola- 
 tile ZrCl^ . Table 1 shows a typical 
 analysis of the crude ZrCl4 product. 
 
 TABLE 1. - Typical analysis of crude 
 ZrCl^ , ' percent 
 
 steel vessel at 
 frit: 
 
 600° to 650° C to form a 
 
 Element 
 
 ZrCl4 
 
 Zr 
 
 metal basis 
 
 Zr 
 
 38.3 
 .84 
 .09 
 .19 
 .03 
 .08 
 .01 
 .008 
 .03 
 .5 
 Balance 
 
 
 Balance 
 
 Hf 
 
 Al 
 
 2.2 
 .23 
 
 Fe 
 
 .5 
 
 P 
 
 .08 
 
 Si 
 
 Th 
 
 .2 
 .03 
 
 Ti 
 
 .02 
 
 U 
 
 Insoluble in H2O. 
 CI" 
 
 .08 
 NAp 
 
 NAp 
 
 
 NAp Not applicable. 
 
 ^Supplied by Teledyne Wah Chang, Al- 
 bany, OR, Dec. 1981. 
 
 In another commercial method used to 
 obtain aqueous process solutions, the 
 zircon ore is reacted with dry NaOH in a 
 
 Zr02-Si02 + 4 NaOH -»- Na2Zr03 
 
 + Na2Si03 + 2 H2O. (1) 
 
 The solid frit is washed with water to 
 remove soluble silicates. Nitric or sul- 
 furic acid is added to dissolve the zir- 
 conium values. Caustic fusion has been 
 used for unlocking the zircon ore by Co- 
 lumbia National in Pensacola, FL, in con- 
 duction with the tributyl phosphate(TBP)- 
 nitric acid separation process. TBP in 
 n-hexane is used as the organic solvent 
 for removing zirconium from the nitric 
 acid solution. 
 
 Another aqueous method used commercial- 
 ly for achieving Hf-Zr separation is the 
 methyl isobutyl ketone (MIBK)-thiocyanate 
 process. Thiocyanate solutions in MIBK 
 are used for removing hafnium from the 
 aqueous phase. The less commonly used 
 TBP-nitric acid process produces reactor- 
 grade hafnium oxide as a byproduct only 
 with great difficulty and is not being 
 used commercially at present. Hafnium 
 production is important because its high 
 neutron absorption cross section makes it 
 valuable as a control rod in nuclear re- 
 actors. Hence the MIBK-thiocyanate pro- 
 cess is preferred. 
 
 After removal of the hafnium, hydrous 
 zirconium oxide must be precipitated from 
 solution, washed with water to remove im- 
 purities, and calcined to the oxide, and 
 the oxide must be rechlorinated to the 
 tetrachloride. Thus, removal of hafnium 
 from zirconium by solvent extraction is a 
 costly, energy-inefficient process. 
 
 Reduction of the tetrachloride to pro- 
 duce zirconium (or hafnium) metal must 
 be carried out in the absence of air, be- 
 cause the absorption of oxygen or nitro- 
 gen renders these metals brittle. Before 
 the invention of the Kroll process, the 
 only practical way of preparing ductile 
 zirconium metal was to decompose Zrl^ on 
 a hot tungsten wire. This is a slow, ex- 
 pensive process. The Kroll process was 
 
developed in the mid-1940' s, and commer- 
 cial production of zirconium metal was 
 achieved in 1953. 
 
 The Kroll process is a batch technique 
 and consists of the following steps: 
 
 1. Sublimation of impure ZrCl4 in an 
 inert atmosphere such as argon, to leave 
 behind undesirable oxides and other non- 
 volatile impurities. 
 
 2. Reduction of the ZrCl4 vapor with 
 excess magnesium metal at 825° to 875° C 
 to produce zirconium sponge. 
 
 3. Vacuum treatment at 
 to remove Mg and MgCl2 
 
 920° to 960° C 
 This step is 
 carried out with the container inverted 
 to allow partial drainage of liquid Mg 
 and MgCl2 from the solid 
 
 sponge, 
 wall. 
 
 which clings to the 
 
 zirconium 
 container 
 
 for structural components within such 
 reactors. 
 
 TABLE 2. - ASTM specification B349-73 for 
 nuclear-grade zirconium sponge (2) 
 
 Element 
 
 Max 
 
 cone, 
 
 ppm 
 
 Al 75 
 
 B 0.5 
 
 C 50 
 
 Cd 0.5 
 
 CI 1,300 
 
 Co 20 
 
 Cr 200 
 
 Cu 30 
 
 Fe 1,500 
 
 Hf 100 
 
 Element 
 
 Max 
 
 cone, 
 
 ppm 
 
 Mn 50 
 
 Mo 50 
 
 N 50 
 
 Ni 70 
 
 1,400 
 
 Si 120 
 
 Ti 50 
 
 W 50 
 
 U (total). 3.5 
 
 TABLE 3. - Chemical requirements for 
 reaction-grade hafnium metal (25) 
 
 Zirconium sponge produced in this way 
 meets impurity limits of ASTM specifica- 
 tion B 349-73 (2^), shown in table 2. 
 Standards for hafnium are not as strict, 
 although 2 to 5 pet Zr content is the ac- 
 cepted upper limit for hafnium metal. 
 The chemical requirements for reactor- 
 grade hafnium metal are given by Goodwin 
 (25) and are listed in table 3. 
 
 After the zirconium sponge is compacted 
 and melted into ingots by the consum- 
 able electrode method, it is fabricated 
 and used as cladding for the fissonable 
 uranium in nuclear reactors (39). Zirco- 
 nium and zirconium alloys are also used 
 
 Element 
 
 ppm^ 
 
 Element 
 
 ppm^ 
 
 Al 
 
 50 
 
 <5 
 
 50 
 
 <1 
 
 <10 
 
 <10 
 
 <50 
 
 100 
 
 2<30 
 
 <10 
 
 <10 
 
 Mo 
 
 <10 
 
 B 
 
 N 
 
 20 
 
 c 
 
 <10 
 
 Cd 
 
 
 
 500 
 
 Co 
 
 Pb 
 
 <10 
 
 Cr 
 
 Si 
 
 <10 
 
 Cu 
 
 Fe 
 
 Sn 
 
 Ti 
 
 <10 
 <10 
 
 H 
 
 W 
 
 50 
 
 Mg 
 
 Mn 
 
 Zr pet. . 
 
 2.25 
 
 'Except zirconium. 
 ^Vacuum melting would result in 5 to 10 
 ppm H. 
 
 SEPARATION TECHNIQUES BASED ON RELATIVE VOLATILITY 
 
 of their 
 
 The close similarity between haf- 
 nium and zirconium is reflected in the 
 
 physical constants 
 shown in table 4. 
 
 chlorides , 
 
 TABLE 4. - Physical constants of hafnium and zirconium tetrachlorides 
 
 Properties 
 
 HfCl4 
 
 ZrCl4 
 
 Properties 
 
 HfCl4 
 
 ZrCl4 
 
 Sublimation temp'..°C.. 
 Triple point: 
 
 Temperature °C. . 
 
 Pressure kPa. . 
 
 317 
 
 432.0 
 4,501.3 
 
 331 
 
 437.0 
 2,235.92 
 
 Critical point: 
 
 Temperature °C. . 
 
 Pressure kPa. . 
 
 Volume cm-' /mol . . 
 
 449.2 
 
 5,776.12 
 
 303.6 
 
 505.0 
 
 5,766.31 
 
 319.3 
 
 'Lange (37). 
 Source: Denisova (14) for all except sublimation temperature. 
 
Three separation concepts iJ^S^, 2^, 29- 
 32 , 58-59) based on the relative volatil- 
 ity of HfCl4 and ZrCl4 have been studied: 
 
 1. Thin-film sublimation at atmospher- 
 ic pressure. 
 
 2. Molten-salt distillation at 1 atm 
 (101.3 kPa). 
 
 3. High-pressure fractional distilla- 
 tion at 40 to 60 atm (4,050 to 6,080 
 kPa). 
 
 THIN-FILM SUBLIMATION 
 
 HfCl4 and ZrCl4 have a relative vola- 
 tility PHfCu/PzrCU Of 1.9 at 250° C, 
 enabling separation at 1 atm. However, 
 sublimation columns do not operate as 
 efficiently as countercurrent fractiona- 
 tion columns, because the solids adhere 
 to the surfaces and thus do not flow 
 countercurrent to the gas. To overcome 
 this problem, mechanical transfer of sol- 
 ids down the column must be accomplished. 
 
 The vapor pressures of solid ZrCl4 and 
 of HfCl4 may be represented by equations 
 2 and 3 ( JJ- ) : 
 
 log Pzrci4 = 11.4632 - ^^^ 
 (below 426° C) , 
 
 where 
 
 a = 
 
 (2) 
 
 log PHfci4 = 11.6726 - ^-^ (3) 
 
 (below 412° C), 
 
 where P is in mm Hg and T is in kelvins. 
 
 As the relative volatility of these two 
 tetrachlorides is nearly constant over 
 the temperature range 150° to 350° C, the 
 Fenske -Underwood equation (51) can be 
 used to estimate the number of theoreti- 
 cal plates needed to achieve the desired 
 separation, under a condition of total 
 reflux : 
 
 n = 
 
 Xi/Xo = 
 
 1 . (Xi/X2)p 
 
 (4) 
 
 relative volatility (ap- 
 proximately 2 in the 
 range of interest, 150° 
 to 350° C), 
 
 number of theoretical 
 plates required, 
 
 ratio of mole fraction of 
 more volatile component 
 
 (1) to mole fraction of 
 less volatile component 
 
 (2) on a lower plate o 
 and an upper plate p, 
 respectively. 
 
 This approach predicts approximately 30 
 theoretical plates for the separation. 
 
 The concept was explored in 1949 by 
 Plucknett, Hansen, and Duke (54) , who at- 
 tempted to separate zirconium and hafnium 
 tetrachlorides in a 10-plate column with 
 mechanical transfer of the solid material 
 down the column. Practical separation 
 was not accomplished because only the 
 surface of the solids came into equilib- 
 rium with the traveling gases. 
 
 The opposite situation, stationary sol- 
 ids and an inert gas as a carrier for the 
 volatilized tetrachlorides, was investi- 
 gated by Jacque and IXimez (32) in 1967. 
 Temperature of the heated column was 300° 
 to 400° C at the bottom, decreasing to 
 150° to 230° C at the top; column height 
 was 2 m, and column ID was 16 mm. The 
 authors stated that ZrCl4 with an HfCl4 
 content of 250 to 1,000 ppm could be pro- 
 duced by repeating the batch process sev- 
 eral times. The authors postulated a 
 process capable of producing 0.3 kg of 
 "dehafniated" ZrCl4 in a 4-h cycle. A 
 bundle of parallel columns, 8 m long and 
 10 mm in ID, with a total cross section 
 of approximately 80 cm^ , was required. 
 
 A process was devised by Goldberger and 
 Gillot (22^, 24 ) to overcome the disadvan- 
 tages of the two solid-gas fractiona- 
 tion concepts described above. A column 
 4.5 cm in ID by 1 m long, maintained at 
 
1 
 
 280° C, was filled with inert glass 
 spheres of 0.42- to 0.59-iDm diam, which 
 moved down the column by gravity at a 
 controlled rate of 50 cm/h. A drawing of 
 the apparatus is shown in figure 1 . The 
 crude ZrCl4 was vaporized in a chamber at 
 the bottom of the column and carried up- 
 ward by a stream of nitrogen gas. The 
 HfCl4-rich vapor was condensed at the top 
 of the column. Fresh inert material was 
 fed into the top of the column as re- 
 placement for that withdrawn at the bot- 
 tom; the purified ZrCl4 was vaporized 
 from the glass spheres and condensed in a 
 separate chamber. The patent by Goldber- 
 ger and Gillot also covered separation by 
 means of a temperature gradient without 
 the nitrogen gas stream. 
 
 In this patent (24) , it is stated that 
 a mixture of 2.64 g ZrCl4 and 2.51 g 
 HfCl4 (59.3 mol pet ZrCl4), after frac- 
 tionation for 3 h in the column de- 
 scribed, had the following compositions: 
 
 The repeated sublimation and condensa- 
 tion of the tetrachlorides concentrates 
 the more volatile HfCl4 in the upper part 
 of the column. Goldberger and Gillot 
 found a height equivalent to a theoreti- 
 cal plate of 22 cm and postulated that 
 the column length would have to be 4.2 m 
 to achieve the desired purification of 
 the ZrCl4 , but no practical work was done 
 to confirm this. The process has the 
 following advantages and disadvantages: 
 
 1. Process operates at atmospheric 
 pressure; the need for high-pressure pro- 
 cess components such as valves and seals 
 is eliminated. 
 
 2. Temperatures do not exceed 350° C. 
 Thus , material requirements are less 
 stringent. 
 
 3. Process is simple in operation and 
 could be operated continuously if 
 required. 
 
 Height of sampling point Composition of 
 
 above bottom of heating 
 
 sample, mol 
 
 jacket, cm 
 12 
 
 ZrCl4 
 68.7 + 0.8 
 
 32 
 
 57.3 + 3.0 
 
 52 
 
 38.5 ± 4.0 
 
 Condensed 
 oroduci. 
 
 Condensed 
 ^ product. 
 ZrCI,-rich 
 
 Panicle conveye 
 
 FIGURE !• - Diagram of fractional sublimation ap- 
 paratus used by Goldberger and Gillot (24). 
 
 4. No process reagents are used, and 
 no byproduct streams are created. 
 
 5. The main disadvantage is tempera- 
 ture control, particularly if larger 
 diameter columns are used, because of the 
 difficulty in achieving uniform radial 
 temperature. 
 
 6. Secondary operations are required 
 for stripping the inert solids exiting 
 the bottom of the column, so that carrier 
 material can be recycled. 
 
 7. The diffusional nature of the pro- 
 cess requires that the uniform film 
 thickness be less than 1 ym in order to 
 achieve equilibrium between the solid 
 and vapor within a realistic time peri- 
 od. This implies a very inefficient uti- 
 lization of column volume. The mass 
 throughput per equilibrium stage for the 
 work described was approximately 0.004 
 g/cm^*h, compared to a corresponding val- 
 ue of 5,800 times greater for the liquid- 
 vapor separation of Ishizuka (29) . 
 
 8. The need to protect the hydroscopic 
 solid from contact with the atmosphere 
 
complicates the problem of column 
 operation. 
 
 EXTRACTIVE DISTILLATION 
 FROM MOLTEN SALTS 
 
 Researchers (15-16 , 24 , 43 ) have 
 pointed out the technical difficulties 
 associated with the high-pressure liquid- 
 vapor distillation of ZrCl4-HfCl4 mix- 
 tures. The operating pressures needed to 
 achieve liquid-vapor distillation are in 
 the range of 40 to 60 atm and require the 
 use of special high-pressure components. 
 A serious drawback associated with high 
 operating pressure is the difficulty in 
 achieving continuous operation that in- 
 volves feeding a solid into the system. 
 Only a narrow operational temperature 
 range exists between the triple point 
 (437° C) and the critical point (505° C) 
 of ZrCl4. This condition places a strin- 
 gent temperature control requirement on 
 the separation system. 
 
 A number of attempts have been made to 
 achieve distillation using solutions of 
 (Zr,Hf)Cl4 in molten salts. The use of 
 molten salts decreases the activity of 
 the tetrachlorides and permits separation 
 at atmospheric pressure and at tempera- 
 tures below 400° C. 
 
 In 1976, Spink (_58) obtained a patent 
 covering the distillation of a feed mix- 
 ture of 63 mol pet tetrachlorides, 29 mol 
 pet KCl, and 8 mol pet NaCl. This is the 
 ternary eutectic composition with an in- 
 variant freezing temperature of 218° C. 
 As shown in figure 2, the crude feed eu- 
 tectic solution is fed into a distilla- 
 tion column operating between 325° C at 
 the top and 400° C at the bottom. Fifty 
 theoretical plates are specified in or- 
 der to obtain the desired end products, 
 nuclear-grade ZrCl4 and commercially pure 
 HfCl4. This number of plates corresponds 
 to a relative volatility of 1.7, which is 
 assumed constant over the entire range 
 of operating temperature. Reflux of the 
 HfCl4-rich overhead vapor stream and re- 
 cycle of the stripped ZrCl4-rich salt 
 bottoms are included in the scheme. This 
 necessitates the handling of hot molten- 
 salt streams ; the feed and reflux flows 
 
 can be transferred by gravity, but the 
 recycle stream must be pumped. Pumping 
 of molten salts in the range of 300° to 
 400° C requires special consideration. 
 Alternatively, the molten salts could be 
 transferred using inert gas pressure. A 
 variation that avoids molten salt is to 
 add the solvent salt at the top of the 
 column as a granular solid. To eliminate 
 the need to crush the solidified salt, 
 Spink has experimented with "prilling" or 
 dropwise solidification. 
 
 Removal of the product tetrachloride 
 from the overhead and bottom streams is a 
 major problem in molten-salt distilla- 
 tion. Spink and Jonasson (59) concluded 
 that a combination of high temperature 
 and vacuum was necessary to completely 
 remove ZrCl4 from the bottoms salt be- 
 cause of the high stabilities of K2ZrClg 
 and Na2ZrCl6. 
 
 Three possible approaches for removal 
 of the ZrCl4 from the bottom stream 
 will be considered in the following 
 paragraphs: 
 
 1. Molten-salt electrolysis . — Electro- 
 winning of zirconium from the ZrCl4-rich 
 bottoms stream is an attractive approach 
 in that the solvent salt mixture would 
 be returned to the mixing tank with its 
 original composition. Spink and Vijayan 
 (60) , however, have demonstrated that the 
 tetrachloride-rich solution is unsuitable 
 for electrowinning because of the high 
 vapor pressure exerted. For this reason, 
 only dilute ZrCl4 solutions are used in 
 electrowinning. Martinez and Couch (41) 
 were able to produce ductile zirconium 
 crystals from NaCl melts, but the level 
 of zirconium was only 2 pet. It is ap- 
 parent that the bottom stream of the 
 distillation column cannot be treated 
 directly on a commercial scale by molten- 
 salt electrolysis. Dilution is a possi- 
 bility, but an extremely large electro- 
 lytic recovery vessel would be needed. 
 
 2. Partial thermal stripping with re- 
 cycle . — Spink and Jonasson (59) have sug- 
 gested that the addition of MgCl2 to the 
 solution salt might increase the activity 
 of ZrCl4 according to the reaction 
 
K2ZrCl6 + MgCl2 ->■ K2MgCl4 
 + ZrCl4 ( g ) , 
 
 (5) 
 
 as reported by Tverskov and Morozov ( 63 ) . 
 Examination of the KCl-NaCl-MgCl2 ternary 
 system (32) shows three double salts: 
 KCl'MgCl2~mp - 490° C) , NaCl«MgCl2 (dec 
 ^ 465° C) and 2NaCl-MgCl2 (dec ^ 485° C) 
 (36) (mp = melting point; dec = decompo- 
 sition point.) A ternary eutectic point 
 occurs at 385° C and approximately 46 mol 
 pet MgCl2, 21 mol pet KCl, and 33 mol pet 
 NaCl. ZrCl4 and MgCl2 form a simple eu- 
 tectic system with the invariant point 
 at 426° C and 94 mol pet ZrCl4 . Thus, 
 the addition of MgCl2 to the metal chlo- 
 ride solvent not only increases the 
 activity of ZrCl4 by incorporating KCl 
 
 and NaCl into double salts , but also low- 
 ers the melting point of the 
 depleted return solvent salt. 
 
 ZrCl4- 
 
 Dutrizac and Flengas (15-16, 18 ) made a 
 systematic study of the stabilities of 
 double salts of zirconium and hafnium 
 with alkali and alkaline-earth chlorides 
 and found a relationship between decompo- 
 sition temperature and the radius of the 
 metal ion. Decomposition temperature was 
 defined as the temperature at which the 
 vapor pressure of the ZrCl4 over the dou- 
 ble salt was equal to 1 atm. Figure 3 
 shows a plot of decomposition temperature 
 versus metal ion radius. The smaller 
 metal ions are more able to pull a cova- 
 lently bonded Cl~ ion from the octahedral 
 ZrClg^- ion. The addition of Li'^, Mg2+, 
 
 HfCl4-ricti gas 
 
 Crude (Zr, Hf]CI 
 
 330°- 
 340°C 
 
 Feed tank 
 
 1 
 
 Feed stream 
 
 Overhead vapor 
 
 32 5°C 
 
 Makeup'salt 
 
 Retlux melt tank 
 
 Retlux line 
 
 Fractionation column- 
 50 theoretical plates 
 up to 90 actual plates 
 
 400°C 
 
 J 
 
 ZrCl4 vapor, 
 50 ppm HfCU 
 
 ^ 
 
 Bottoms, ZrCl4-rich salt 
 
 4 2 0^ 4 5 °C 
 
 Vaporizer 
 
 Recycle line 
 
 I Purge salt 
 FIGURE 2. - Diagram of Spink extractive distillation process (58). 
 

 l,OUU 
 
 
 1 
 
 1 
 
 \ . 
 
 ►Cs+ 
 
 ^ 
 
 
 
 
 
 Rb+ X 
 
 
 < 
 
 a. 
 
 lU 
 
 a. 
 5 
 
 1.100 
 
 — 
 
 
 Na+ . 
 
 >&/ 
 
 
 - 
 
 900 
 
 
 
 
 A 
 
 > 
 
 
 z 
 C 
 
 M 
 O 
 
 a. 
 2 
 O 
 
 o 
 
 UJ 
 
 O 
 
 700 
 500 
 
 — 
 
 -V 
 
 
 r. 2 + 
 
 ^ Ba 
 1 
 
 
 0.5 
 
 1.0 
 
 1.5 
 
 2.0 
 
 METAL ION RADIUS, A 
 
 FIGURE 3, • Relationship between decomposition 
 temperature of he xachloro zircon at es and metal ion radi- 
 us for selected alkali and aJkalineearth elements. (Af- 
 ter Morozov and Sun (45)) 
 
 or Ca^"^ to the melt breaks up the ZrCl5 2- 
 ion and increases the activity of ZrCl4. 
 Such a solvent melt, high in LiCl and 
 MgCl2, should be easily stripped at mod- 
 erate temperature. 
 
 3. Reduction in the salt solution with 
 electrolysis of the product salt . — The 
 entire bottoms stream may be fed directly 
 into a vessel, where zirconium is recov- 
 ered from the molten salt by reducing the 
 ZrCl4 to the metal, which precipitates. 
 The zirconium-depleted salt is drained 
 from the reduction vessel. During the 
 reduction step, the original eutectic 
 salt composition is altered due to the 
 generation of either MgCl2 or NaCl, de- 
 pending on the reducing metal agent used. 
 Electrolysis of the resulting NaCl-KCl- 
 MgCl2 salt solution could be used to re- 
 store the composition and at the same 
 time generate reductant metal. Based on 
 the relative oxidation potentials, the 
 resulting metal would be predominately 
 magnesium. 
 
 The above practice would provide a 
 completely closed system. The distil- 
 lation column becomes an integral com- 
 ponent in a closed system consisting 
 of extractive distillation column, 
 
 reduction reactor, and a molten-salt 
 electrolysis cell for the recovery of 
 the high-magnesium alloy. Opportunities 
 exist for energy conservation by the 
 use of hot metal and hot salt transfer 
 from stage to stage. However, it is not 
 consistent with existing practice, which 
 involves reduction of the pure tetra- 
 chloride with magnesium. New techniques 
 and equipment would be required if re- 
 duction in molten-salt solution were 
 adopted. 
 
 The process developed by Besson (_5) for 
 Pechiney Ugine Kuhlmann, whose subsidiary 
 Cezus is reported by Brun (^) to be using 
 it commercially, is based on mixing ZrCl4 
 and HfCl4 with molten aluminum chloride 
 and potassium chloride and distilling the 
 mixture at atmospheric pressure. It is 
 referred to as the Cezus-Pechiney pro- 
 cess. The resulting ZrCl4 contains less 
 than 50 ppm HfCl4. This process has been 
 termed "extractive distillation" by its 
 developers and is similar to the process 
 investigated by Spink. Two critical dif- 
 ferences are evident: 
 
 1. The ratio of AICI3 and FeCl3 to KCl 
 in the melt must be maintained above 
 0.94:1, and preferably between l.OA to 
 1.10:1, by periodic addition of AICI3 or 
 FeCl,. 
 
 2. After the HfCl4 has been removed 
 4 , the latter is stripped 
 
 from the ZrCl 
 out of the solvent salt by a nitrogen 
 stream and condensed. The molten solvent 
 salts are recirculated to the top of the 
 column so that the operation is continu- 
 ous. Figure 4 shows a schematic repre- 
 sentation of the process. 
 
 Although the patent claim of Besson (_5) 
 cites use of either FeCl3 or AICI3 in the 
 solvent salt, in the process described by 
 Brun (8^) only AICI3 is used. 
 
 Based on the correlation of Dutrizac 
 and Flengas (15), FeClj or AICI3 should 
 be quite effective in unlocking the 
 ZrClg2- con^lex. They have shown that 
 the stability of the metal chlorozircon- 
 ate will vary as (r^ + rci)2/q^, where r^, 
 is the ionic radius of the metal, Xq\ is 
 
10 
 
 Recycle salt 
 
 Condenser 
 
 Crude feed 
 
 Reboiler 
 
 Reservoir 
 
 Salt pump 
 
 FIGURE 4. - Pechiney process for extractive 
 distillation (5). 
 
 the covalent radius of chlorine in the 
 ZrCl5 2~ complex, and q^ is the charge on 
 the metal ion. Table 5 shows the decom- 
 position temperatures of several alkali 
 chlorozirconate and chlorohafnate com- 
 pounds. Because of the relatively small 
 
 TABLE 5. - Comparison of the thermal 
 stability of the alkali chlorozir- 
 conate and chlorohafnate compounds 
 
 Compound 
 
 Dec 
 
 omposition 
 
 Reference 
 
 
 temp 
 
 eraturej °C 
 
 
 Li2ZrCl6.. .. 
 
 
 501 
 
 3 
 
 Li2HfCl6 
 
 
 513 
 
 3 
 
 Na2ZrCl6 
 
 
 634 
 
 3 
 
 Na2HfCl6 
 
 
 -648 
 
 35 
 
 K2ZrCl6 
 
 
 831 
 
 3 
 
 K2HfCl6 
 
 
 863 
 
 34 
 
 Rb2ZrCl6.... 
 
 
 904 
 
 3 
 
 Cs2ZrCl6. ... 
 
 
 1,040 
 
 3 
 
 Cs2HfCl5 
 
 
 953 
 
 3 
 
 P(Zr,Hf)CI = 1 atm. 
 
 size and trivalency of Fe^"*" or Al^"*", the 
 removal of ZrCl4 should be accomplished 
 readily. Besson (5) reports that at 
 500° C and 13 mm Hg (1.7 kPa) , the resid- 
 ual ZrCl4 was reduced to 0.6 g/100 g 
 KAICI4. No mention was made of FeCl3 or 
 AICI3 contamination of the product tetra- 
 chloride. This should occur because of 
 the volatilities of these two substances, 
 but the product reported by Brun (8) 
 yields a nuclear-grade sponge. 
 
 This extractive distillation process, 
 termed by its inventors the "S" process, 
 is being used in a pilot plant, replacing 
 the MIBK-thiocyanate process. However, 
 nothing in either the patent claim by 
 Besson or the article by Brun indicates 
 how contamination of the purified ZrCl^ 
 by AICI3 is prevented. There must be a 
 practical reason for using AICI3 instead 
 of FeCl3, because ASTM specifications 
 permit 1,500 ppm Fe but only 75 ppm AICI3 
 in the zirconium metal. 
 
 The role of trace impurity chlorides 
 must be considered in any molten-salt ex- 
 traction process. The main impurities 
 are the chlorides of Fe, Al, Si, P, and 
 Ti (table 1); in the Cezus process (7^), 
 these are removed by a preliminary subli- 
 mation of the crude ZrCl4-HfCl4. An al- 
 ternative method is molten-salt scrubbing 
 of the ZrCl4 , described by Spink (53) . 
 Greenberg (26) and Frey (20) describe 
 other patented methods for selective im- 
 purity removal, Greenberg claims that 
 aluminum halides can be removed by dis- 
 tilling the ZrCl4 through CaCl2. Frey 
 states that the use of highly viscous oil 
 that carbonizes below the sublimation 
 points of ZrCl4 and HfCl4 will remove 
 FeCl3, A patent was issued to Ross (55) 
 for removal of CO, COCI2, and CI 2 from 
 crude ZrCl4, The impure ZrCl4 was dis- 
 solved in a KCl-NaCl bath partitioned 
 into chambers, and the purified tetra- 
 chloride was removed as a vapor. 
 
 HIGH-PRESSURE LIQUID-VAPOR DISTILLATION 
 
 Distillation techniques require heavy- 
 duty components necessary to withstand 
 pressures of 587,6 to 881.4 psi (4,050 to 
 
11 
 
 6,080 kPa) and temperatures up to 505° C. 
 Materials of construction must be resist- 
 ant to ZrCl4 , Hf CI4 , and impurity chlo- 
 rides. Despite these requirements, sev- 
 eral processes have been devised for the 
 high-pressure liquid-vapor separation of 
 ZrCl4 and Hf CI4 . 
 
 In 1958, Bromberg (7^) patented a method 
 for purification of ZrCl4 by fractional 
 distillation. The patent claims that the 
 temperature should be between 455° and 
 520** C at the bottom of the column and a 
 minimum of 440° C at the top. The criti- 
 cal temperature for ZrCl4 was probably 
 not known in 1958 when the Bromberg pa- 
 tent was written because 520° C is above 
 the critical temperature for ZrCl4 (505° 
 C). A line leading from the top of the 
 column enables the more volatile HfCl4 to 
 be condensed in a receiving vessel. The 
 purified condensed ZrCl4 is collected in 
 a receiving vessel at the bottom. The 
 impure ZrCl4 (1.6 pet HfCl4) is vaporized 
 from a boiler at the side of the frac- 
 tionation column. Valves on all three 
 storage vessels enable the HfCl4-rich 
 distillate (92 pet HfCl4) and the puri- 
 fied ZrCl4 (60 ppm HfCl4) to be withdrawn 
 at intervals, and fresh, impure ZrCl4 to 
 be added periodically. Preferred operat- 
 ing temperatures were 495° C at the bot- 
 tom and 460° C at the top. The column 
 was constructed of type 316 stainless 
 steel, and was 26 ft high by 3 in. in ID. 
 Either a plate or packed-column design 
 is claimed to be effective. Bromberg 
 claimed that a 36-ft column, operating at 
 a reflux ratio of 100:1, would produce 
 nuclear-grade ZrCl4 in the bottom receiv- 
 er, while purified HfCl4 would be taken 
 off from the top. 
 
 The narrow operating range of this pro- 
 cess was made clear in 1967 when Deni- 
 sova, Safronov, Pustil'nik, and Bystrova 
 published their study of liquid-vapor 
 phase equilibria of ZrCl4 and HfCl4 (14). 
 In figure 5, solid-vapor and liquid-vapor 
 behavior are shown. The authors' exten- 
 sion of the log P versus lO^/x plot into 
 the supercritical region is unexplained. 
 Figure 6 shows the densities of coex- 
 isting liquid and vapor phases at con- 
 stant temperature. The temperature at 
 
 TEMPERATURE. C 
 496 468 441 
 
 1.20 125 130 135 140 
 
 10''/T. deg'V 
 
 145 1,50 
 
 FIGURE 5. - Vapor pressure of ZrCl^ and HfCI^. 
 (After Denisova (14)) 
 
 510 
 
 500 — 
 
 490 — 
 
 480 
 
 UJ 
 
 ^ 470 
 
 h- 
 < 
 
 ifJ 460 
 
 Q. 
 
 2 
 
 450 — 
 
 440 — 
 
 430 
 
 420 
 
 KEY — I 
 
 C Critical point 
 
 5 10 15 2.0 
 
 DENSITY, g /cm ^ 
 
 FIGURE 6. - Density of coexisting liquid and vapor 
 phases of ZrCI^ and HfCI^. (After Denisova (14)) 
 
12 
 
 the bottom of the column is limited 
 by the critical conditions for ZrCl4 
 (5,766.3 kPa and 505.0° C) . The lower 
 temperature limit of operation is the 
 triple point for Hf CI4 , 432° C. 
 
 The necessity for a narrow range of 
 operating conditions was also reported by 
 Ishizuka (29). The 1974 patent appli- 
 cation stated that the temperature of 
 the boiler was 469° C, bottom rectifier 
 466° C, top rectifier 461° C, and con- 
 denser 454° C, at an operating pressure 
 of 40 kg/cm2 (3,900 kPa) . Column height 
 and ID were 700 mm and 20 mm, respective- 
 ly. Figure 7 shows the type of apparatus 
 used by Ishizuka. The crude feed to this 
 batch process contained 2 pet HfCl4 and 
 produced, after 24 h of operation, a con- 
 denser product with 32 pet HfCl4 and a 
 ZrCl4 boiler product with 50 ppm HfCl4. 
 Before the column was frozen to col- 
 lect product fractions, the boiler prod- 
 uct contained only 8 ppm Hf CI4 . Although 
 specification HfCl4 (<5 pet ZrCl4) was 
 not obtained, extrapolation to the 26-ft 
 column used by Bromberg (7) indicated 
 that the Ishizuka column was more 
 efficient. 
 
 Operation of a unit for processing 5 
 tons of crude ZrCl4 is reported in a Eu- 
 ropean patent application by Ishizuka 
 (30) . A mild steel column was good for 
 20 to 50 runs before it needed substan- 
 tial repairs. A second distillation was 
 necessary to convert hafnium-rich over- 
 head chloride to nuclear-grade HfCl4. 
 Removal of impurity chloride was achieved 
 by adding of small amounts of NaCl or KCl 
 to form nonvolatile complexes with AICI3 
 and FeClj. 
 
 fpij Pressure gages (B) 
 
 r 
 
 4XH=^ 
 
 Sampling valve 
 
 . Reflux condenser 
 
 — yU — . — Reflux reservoir 
 
 Overhead product 
 
 KEY 
 
 Heated portion 
 
 — Insulated portion 
 q Ttiermocouple 
 
 Sampling valve 
 
 @- 
 
 Packed column 
 
 Reboiler 
 
 Sampling valve 
 
 I I I 
 
 FIGURE 7, - Apparatusused by Ishizuka to separate 
 HfCI^ from ZrCI^ (29). 
 
 The narrow operating range for frac- 
 tional distillation shown in these pa- 
 tents would require close temperature 
 control for successful separation of zir- 
 conium and hafnium tetrachlorides. 
 
 CHEMICAL METHODS OF SEPARATING HAFNIUM FROM ZIRCONIUM 
 
 Because of the difficulty of separat- 
 ing hafnium from zirconium by sublimation 
 or fractional distillation, chemical 
 methods have been investigated. Unlike 
 the methods based on relative volatility, 
 chemical separation techniques cannot be 
 easily categorized. As a broad general- 
 ity, hafnium compounds are slightly more 
 
 stable than the corresponding zirconium 
 analogs . This is probably because the 
 Hf-X bond is stronger and displays a more 
 covalent character than the Zr-X bond. 
 These small differences have been ex- 
 ploited in the separation of hafnium from 
 zirconium by chemical methods . 
 
13 
 
 A number of schemes have been proposed. 
 Each is unique, so the chemical separa- 
 tion methods must be treated individual- 
 ly. Several of the most promising candi- 
 date processes follow: 
 
 Preferential reduction of ZrCl4 (Newn- 
 ham - 1957) (46). 
 
 Fluoride-redox equilibrium (Megy - 
 1979) (44). 
 
 Chloride-oxide exchange (Chandler - 
 1966) (9^). 
 
 Preferential decomposition of salts 
 (Flengas and Dutrizac - 1977) (18). 
 
 The reduction reaction may be repre- 
 sented by 
 
 Differential oxidation of 
 (Berl - 1961) (4). 
 
 chlorides 
 
 Each process has inherent advantages 
 and disadvantages. The high separation 
 factor and closed cycle nature of the 
 Nevmham process are offset by the neces- 
 sity of handling pyrophoric solids. The 
 Megy process has the highest separation 
 factor, but the use of fluoride and the 
 lack, of compatibility with existing Kroll 
 or electrolytic technology are serious 
 disadvantages . 
 
 PREFERENTIAL REDUCTION OF ZrCl4 
 
 In 1957, Newnham (4_6) obtained a patent 
 for the separation of HfCl4 from ZrCl4 
 based on the observation that ZrCl4 is 
 more easily reduced to the lower chlo- 
 ride form than is Hf CI4 . For example, at 
 427° C (700 K) the Gibbs energy change 
 for the reaction (23) 
 
 HfCl3(s, + ZrCl4(g) > HfCl4(g) 
 
 + ZrCl3(s) (6) 
 
 is -22 kcal/mol (-92 kj/mol). The lower 
 chloride of zirconium remains in the con- 
 densed form, while HfCl4 and unreacted 
 ZrCl4 may be sublimed. The separation is 
 much more effective than one based on the 
 relative volatilities of HfCl4 and ZrCl4. 
 
 Zr(s) + 3 ZrCl4(g) ^ 4 ZrCljcs) 
 
 (7) 
 
 A number of reducing agents may be used, 
 but zirconium metal is preferred because 
 no impurities are introduced into the 
 system. The more volatile HfCl4 and the 
 unreacted ZrCl4 remain in the gaseous 
 form. 
 
 ZrCl3 is subsequently heated to 420° to 
 460° C, where it disproportionates: 
 
 2 ZrCl3(s) > ZrCl2(s) + ZrCl4(g). (8) 
 
 The low-hafnium ZrCl4 product is recov- 
 ered, and the resulting ZrCl2 solids are 
 recycled as a reducing agent in subse- 
 quent stages: 
 
 ZrCl2(s) + ZrCl4(g) -^ 2 ZrCl3(3), 
 
 340° to 420° C. (9) 
 
 The patent proposes a process that is 
 closed and cyclic. 
 
 In 1959, Newnham obtained a second pa- 
 tent (47) that extended the original con- 
 cept to carry out the reduction in a 
 molten-salt medium, such as AlCl3-NaCl, 
 LiCl-KCl, or other mixtures containing at 
 least one alkali chloride salt. The 
 molten-salt medium keeps the temperature 
 close to the optimum required for selec- 
 tive reduction. In addition to separa- 
 tion of ZrCl3 and HfCl4 ^y volatility, 
 the patent claims that this separation 
 can be carried out through decantation or 
 filtration, as ZrCl3 is a solid in a liq- 
 uid medium. 
 
 A related patent was issued in 1973 to 
 Larsen and Gil-Arnao (38) . In this case 
 the crude ZrCl4 is reacted with a metal- 
 lic reducing agent (Al or Zr) in a pure 
 molten AICI3 medium. Whereas Newnham in- 
 sisted on the presence of an alkali chlo- 
 ride salt to maintain atmospheric pres- 
 sure, Larsen implies that the resultant 
 
14 
 
 advantage of a faster reaction at lower 
 ten^erature (260° C) is more important 
 than avoiding higher pressure, which 
 could be as high as 8 atm (810 kPa) for 
 pure AICI3 at 260° C (12). Recycling of 
 ZrCl2 recovered is the same as in the 
 Newnham method. Separation factors dem- 
 onstrated by Larsen vary from 5.7 to 
 19.8, while Newnham (48) demonstrated 
 values up to 200. Frampton and Feldman 
 (19) report separation factors from 6.6 
 up to 22 for the technique. Separation 
 factor (SF) is defined by 
 
 Hf in feed, pet 
 Hf in product, pet 
 
 (10) 
 
 Larsen points out that the liquid-phase 
 reaction mechanism overcomes the disad- 
 vantages of the solid-gas reaction of 
 Newnham, where the ZrCl3 coats the sur- 
 face of the zirconium metal and impedes 
 the 
 ZrCl4 
 phase 
 
 reaction. With molten AICI3, the 
 
 first forms a blue intermediate 
 that is soluble in the melt, so 
 
 that the melt turns blue. Brown ZrCl 
 
 3» 
 
 which is insoluble in molten aluminum 
 chloride, forms later, and the molten 
 bath becomes colorless, which indicates 
 the end of the reaction. Related patents 
 are claimed by Newnham (48-49) . 
 
 The disproportionation of ZrCl3 to re- 
 generate ZrCl2 is complicated by a series 
 of reactions to form nonstoichiometric 
 compounds. Shelton and others (10, 54 ) 
 summarize the reactions and their temper- 
 atures of occurrence, as follows: 
 
 12 ZrCl3(s) > 10 ZrCl2.8(s) 
 + 2 ZrCl4(g), 115° to 300° C. (11) 
 
 10 ZrCl2.8(s) ^ 5 ZrCli.6(s) 
 + 5 ZrCl4(g), 310° to 450° C. (12) 
 
 5 ZrCli.6(3) > 4 ZrCl(s) 
 + ZrCl4(g), 500° to 600° C. (13) 
 
 4 ZrCl(s) -> 3 Zr(s) + ZrCl4(g), 
 570° to 700° C. (14) 
 
 The equilibrium ZrCl4 pressures for the 
 first two of these reactions are given by 
 Copley and Shelton (10) : 
 
 log P = -6,138/T + 13.288, (15) 
 
 log P = -9,870/T + 15.555 (16) 
 
 respectively , where P is given in mm Hg 
 and T in kelvins. The latter two reac- 
 tions are not important in the currently 
 conceived reduction process cycle. 
 
 Shown in figure 8 is a plot of log P 
 versus 10^/T for the sublimation of solid 
 HfCl4 and ZrCl4 (56). On the same plot 
 is shown the decomposition pressure of 
 
 ZrCl 
 
 4(g) 
 
 over lower chlorides of zirconi- 
 
 um. The stability of the lower chlorides 
 of zirconium as a function of temperature 
 and pressure is shown in figure 9 (56) . 
 
 In 1968, Mauser (42) studied the selec- 
 tive reduction reaction occurring in a 
 rotating stainless steel reactor filled 
 with stainless steel balls. The rolling 
 balls crush the particles and break up 
 the ZrCl 3 coating that forms on the re- 
 ductant and quenches the reaction. From 
 this gas-solid reaction study, the fol- 
 lowing conclusions were drawn: 
 
 1. The reactor grinding balls were ef- 
 fective in eliminating sintering (agglom- 
 eration) of the reacting particles. 
 
 2. The dichloride (ZrCl 2) was not an 
 effective reductant for ZrCl4. Zirconium 
 in the form of sponge, minus 325-mesh 
 fines, or machine turnings had to be 
 used. Regeneration was carried out at 
 900° C in order to drive the dispropor- 
 tionation reactions to completion and 
 yield finely divided metallic zirconium. 
 Improved yields and separations were re- 
 ported with recycled zirconium. This was 
 attributed to the increased surface re- 
 sulting from repeated reduction and dis- 
 proportionation. The zirconium became 
 increasingly pyrophoric with each cycle. 
 
 3. Zirconium sponge and fines were 
 equally effective reductants, but the 
 turnings were less effective. 
 
15 
 
 KEY 
 
 8 ZrCl4(35^z=z=± ZrCl^^g) 
 
 ^ 2^30C'90(s}^ Z^29Cl86(s] + ^rCUcg) 
 
 e 2ZrCl2 8 -. ZrClig^g^ + Z^CI^^g) 
 
 3 -1 
 
 10 /T. deg K 
 
 FIGURE 8. - Pressure of the tetrachloride gas over 
 HfCL, ,,ZrCi., ,, and lower chlorides over Zr, (After 
 
 4( s)' 4( s )' 
 
 Gjpley and Shelton (10)) 
 
 National Distillers and Chemical Corp. 
 obtained the rights to the Newnham pa- 
 tents and devoted considerable effort to 
 bring the dry process to commercial prac- 
 tice. Frampton and Feldman (19) have de- 
 scribed this work. Although ZrCl2 is re- 
 ported to be a satisfactory reducing 
 agent, the temperature at which the pre- 
 ferential reduction is carried out is 
 critical. In the temperature range 330° 
 to 370° C, a nonselective lower chloride 
 complex (Zr3Cl8 'Hf CI4 ) is formed and de- 
 creases the separation factor. Above 
 
 400 600 
 
 TEMPERATURE. ° C 
 
 1000 
 
 FIGURE 9." Thermal stability of the lower chlorides 
 of zirconium, (After Shelton (56)) 
 
 420° C, the disproportionation of ZrCl3 
 occurs at an appreciable rate. The ZrCl^ 
 formed mixes with Hf CI4 , and the separa- 
 tion factor is decreased. These two con- 
 ditions restrict the temperature of the 
 reduction operation to 400°±20 ° C. The 
 close temperature control was obtained by 
 immersing the apparatus in a bath of mol- 
 ten tin; the authors suggested using 
 molten sodium-potassium alloy (NaK) for 
 large-scale operations. The equipment 
 was constructed of type 316 or 347 stain- 
 less steel. To insure thorough mixing of 
 the reactants, an anchor-type stirrer 
 that scraped the bottom and sides of the 
 container and prevented any buildup of 
 solids was used. 
 
 Disproportionation of ZrCl3 was car- 
 ried out in the range 420° to 460° C and 
 yielded product ZrCl4 and regenerated 
 ZrCl2. The initial ZrCl2 bed was pre- 
 pared by reacting finely divided zirco- 
 nium sponge with ZrCl4 vapor at 430° C 
 for an extended time and subsequently in- 
 creasing the temperature to 460° C to 
 cause disproportionation. 
 
 The authors proposed a pilot plant in 
 which the reactions would be carried out 
 in horizontal tube, screw-fed reactors 
 that would produce 25 Ib/h of hafnium- 
 free ZrCl^ . The separation factor used 
 in the hafnium concentration stage is a 
 very conservative 1.6; three to four 
 
16 
 
 stages of separation were required to 
 produce ZrCl4 containing < 100 ppm Hf/(Hf 
 + Zr). 
 
 Frampton and Feldman report a three- 
 stage separation using solid feed. Oper- 
 ating conditions and results are given in 
 table 6. The solid-gas process ( 19 , 46 ) 
 studied by Frampton and Feldman has the 
 following advantages: 
 
 1. The process is closed and cyclic 
 and does not require reagents. The only 
 raw materials are crude ZrCl4 and makeup 
 zirconium sponge. 
 
 2. The National Distillers work has 
 already provided a process scheme with 
 material balances and a tentative cost 
 estimate. 
 
 3. The process lends itself to contin- 
 uous countercurrent operation. 
 
 Disadvantages are — 
 
 1. Extremely close temperature control 
 is required. 
 
 2. Reducible impurities, such as 
 
 FeCl 
 
 3» 
 
 (ZrCli.e) 
 life. 
 
 will collect in the ZrCl2 
 bed and shorten its useful 
 
 3. Although the process is potentially 
 continuous, initial designs will probably 
 
 TABLE 6. - Operating conditions and re- 
 sults for a three-stage separation of 
 HfCl4 from ZrCl4 by the method of 
 Frampton and Feldman (19) 
 
 Results 
 
 Time h. . 
 
 Hf content, pet: 
 
 Feed 
 
 Product 
 
 Yield, pet: 
 
 Per stage 
 
 Net 
 
 Separation factor 
 per stage 
 
 Temperature range, C 
 
 301-328 
 
 1.1 
 
 2.4 
 0.31 
 
 65 
 65 
 
 7.7 
 
 302-319 
 
 0.9 
 
 0.29 
 0.05 
 
 54 
 35 
 
 5.5 
 
 318-338 
 
 3.2 
 
 0.05 
 0.01 
 
 77 
 27 
 
 4.8 
 
 be batch with considerable manual 
 dling of equipment and materials. 
 
 han- 
 
 4, Mechanical agitation is necessary 
 to expose fresh zirconium surface to va- 
 por. Several techniques are available to 
 accomplish this: 
 
 a. Stirred reactor. 
 
 b. Rotating ball mill reactor. 
 
 c. Fluidized-bed reactor. 
 
 5. The effectiveness of ZrCl2 as a re- 
 ducing agent is questionable. If Mau- 
 ser's observations are correct, a consid- 
 erably higher regeneration temperature 
 (900° C) will be required in order to 
 produce metallic zirconium. 
 
 FLUORIDE REDOX EXCHANGE 
 
 In 1978, Megy (43) improved yields on 
 the exchange reaction 
 
 ZrFg^- + Hf J HfF6 2- + Zr 
 
 (17) 
 
 by the addition of molten zinc to dis- 
 solve the zirconium metal produced. The 
 zinc shifted the reaction to the right 
 because zirconium is preferentially dis- 
 solved in molten zinc and also increased 
 the reaction rate by improving transport 
 in the molten zinc so that conversion was 
 essentially complete in 5 min. 
 
 The equilibrium constant (Kq) for the 
 
 reaction as a function 
 in kelvins) is 
 
 of temperature (T 
 
 log Kq = -1.565 + 4,320/T 
 
 (18) 
 
 for systems using Na2ZrF6 , plus NaCl and 
 KCl to lower salt phase melting tempera- 
 tures (700° to 900° C). 
 
 A reductant, preferably aluminum metal, 
 must be used to convert hafnium and zir- 
 conium salts to the metal so that the ex- 
 change reaction can occur. The presence 
 of aluminum salts does not interfere with 
 the separation. 
 
17 
 
 A similar reaction using chlorides 
 rather than fluorides has the disadvan- 
 tage of producing intermediate oxida- 
 tion states (2+, 3+) for hafnium and 
 zirconium. 
 
 Operating in the temperature range with 
 molten zinc and fluoride salts poses for- 
 midable containment problems. Megy and 
 Freund (44) found that during screening 
 tests employing temperatures of 800° C 
 for 1 h, vitreous quartz, boron nitride, 
 alumina, and glassy carbon were attacked. 
 Even tungsten and graphite were only mar- 
 ginally adequate. Graphite containers 
 contributed 100 ppm C to the metal phase 
 in a 1-h test (44). For this reason the 
 Megy process is of limited interest. 
 
 CHLORIDE-OXIDE EXCHANGE REACTION 
 
 In 1966 Chandler (9) patented a method 
 of separating HfCl^ from ZrCl4 by prefer- 
 ential conversion of HfCl^ to Hf O2 : 
 
 Zr02(s) + HfCl4(g) > Hf02(s) 
 
 + ZrCl4(g). (19) 
 
 This is achieved by passing the mixture 
 of tetrachloride gases over a bed of Zr02 
 and Hf02, where the hafnium preferential- 
 ly enters the solid phase. 
 
 Hafnium is removed from the gas phase 
 because Hf02 is more stable than Zr02 
 relative to the respective chlorides. 
 This is a thermodynamic rather than a 
 kinetic effect (4^, 2J^) . Equilibrium con- 
 stant calculations show little change (Kg 
 = 2.3 to 2.8) between 25° and 950° C. Kg 
 is determined from 
 
 Ke = 
 
 ^ [Hf02] [ZrCl4] 
 [Zr02] lHfCl4] 
 
 (20) 
 
 The constant agrees approximately with 
 that estimated from Chandler's experi- 
 ments (Ke = 5) . 
 
 In the limited work that Chandler per- 
 formed, crude ZrCl4 freshly prepared by 
 carbochlorination was passed through 
 a 15-in bed of crude Zr02 and removed 
 
 two-thirds of the hafnium from the tetra- 
 chloride vapor stream (at 950° C over a 
 2-h period) . No impurity removal was 
 reported. 
 
 PREFERENTIAL DECOMPOSITION 
 OF ALKALI METAL SALTS 
 
 Most physical methods for separating 
 anhydrous hafnium and zirconium tetra- 
 chlorides make use of the higher volatil- 
 ity of HfCl4. 
 
 Flengas and Dutrizac (J^, 18 ) have dis- 
 covered a separation method in which 
 ZrCl4 is the more volatile species. The 
 chlorides are converted to alkali metal 
 salts, M2ZrCl6 and M2HfCl6. The salts 
 are heated, and preferential decomposi- 
 tion of the less stable zirconium salt 
 occurs at >450° C: 
 
 M2ZrCl6(s) -^ 2MCl(s) + ZrCl4(g). (21) 
 
 Table 5 shows the decomposition temper- 
 atures of the double alkali metal chlo- 
 rides that were compiled by Flengas and 
 and others. Recent studies show that 
 potassium is the preferred alkali metal 
 cation for the reaction because all oth- 
 ers show lower separation factors (34- 
 35 ) . The method used (16 , 18 ) involves 
 equilibration of 1 mol of HfCl4-ZrCl4 at 
 330° C with slightly more than 2 mol of 
 KCl held at 450° C. Equilibration takes 
 up to 3 days before a separation factor 
 of 1.6 is achieved. If the ZrCl4:KCl 
 ratio is increased, the separation fac- 
 tor decreases. If the reaction time is 
 decreased, the tendency of the zirco- 
 nium salt to form more quickly than the 
 hafnium salt (4^) greatly decreases 
 efficiency. 
 
 The disadvantages are — 
 
 1. The 3-day reaction time required 
 for static equilibration greatly reduces 
 production rates. 
 
 2. Small separation factors (1.6 to 
 1.9) for KCl systems require a large num- 
 ber of separation stages. 
 
18 
 
 3. Continuous processes in a packed- 
 salt column, where ZrCl4 reacts quickly 
 and decomposes quickly at low separation 
 efficiency, are plagued by the problem of 
 salt swelling; that is, volume change as- 
 sociated with the cyclic formation and 
 decomposition of the double salt. The 
 swelling causes plugging of the column 
 (18). 
 
 Although both the reaction kinetics and 
 the material throughput might be improved 
 by development work, this process is not 
 particularly promising. 
 
 DIFFERENTIAL OXIDATION OF CHLORIDES 
 
 Berl (4) has suggested a novel method 
 for separating zirconium and hafnium com- 
 pounds in a fluidized-bed reactor at tem- 
 peratures above 600° C. The basis for 
 separation is that the following exother- 
 mic reaction for ZrCl4: 
 
 ZrCl4(g) + 02(g) ^ Zr02(s) 
 
 + Cl2(g) (22) 
 
 proceeds more rapidly than the corre- 
 sponding reaction for Hf CI4 . Zr02 Pi^e- 
 ferentially builds up in the solid phase, 
 while enriching the hafnium content in 
 the gas phase. HfCl4 is easily separated 
 from product CI 2 by selective condensa- 
 tion at 0° to 300° C. 
 
 Funaki and Uchimura (21) confirmed 
 the selective reactivity and measured 
 the rate constants (k^ , mm Hg/h) as a 
 function of absolute temperature. For 
 zirconium, 
 
 log kp = 4.25 -5,300/T, 
 
 (23) 
 
 and for hafnium, 
 
 log kr- = 2.9 -4,100/T. (24) 
 
 Calculations show that the rates are 
 equal at 615° C. Berl (4_) ran tests at 
 temperatures where rate differential 
 is small (620° to 800° C) and obtained 
 a maximum separation factor of 2.5 at 
 620° C. 
 
 Berl (4) used hafnium-free zirconium or 
 zirconium oxides as catalysts (seed crys- 
 tals). In consideration of Funaki's (21) 
 work, the catalyst must be essential to 
 the separation at 620° C, where ZrCl4 and 
 HfCl4 react at equal rates. 
 
 The method offers no advantages over 
 other methods that require numerous 
 stages to produce reactor-grade zirconi- 
 um. As in other methods, prepurif ication 
 to remove AICI3, FeCl3, etc., is neces- 
 sary. The most serious problem is the 
 requirement that the zirconium catalyst 
 be hafnium-free, which makes the approach 
 self-defeating. If a countercurrent pur- 
 ification process is set up, with ZrCl4 
 containing 2 pet HfCl4 fed into one side 
 and O2 + Zr02 (2 pct Hf02) into the other 
 side, the steps using high-hafnium Zr02 
 will be highly inefficient, especially at 
 620° C, where rates are nearly equal. 
 This result contradicts the results of 
 Chandler (9^) , where preferentially oxi- 
 dized Hf02 concentrated in the solid 
 phase. The separation method is not 
 worthy of further study. 
 
 RANKING OF PROCESSES 
 
 All process options must be judged on 
 the basis of relative costs. To prepare 
 realistic production cost estimates on 
 which to base process selection is not 
 now possible. It is possible to identify 
 critical process characteristics and to 
 relate them to costs in a qualitative 
 manner. Six characteristics were rated 
 for each of the processes studied as 
 follows: (--) very unfavorable, (-) un- 
 favorable, (0) neutral, (+) favorable, 
 (++) very favorable. The results are 
 
 summarized in tables 7 and 8. Caution 
 should be exercised in the use of the 
 tables because the parameters are not 
 equally important. The assignment of 
 relative weights would imply an unjusti- 
 fied precision for the method. It is not 
 intended that the processes be compared 
 quantitatively on the basis of point to- 
 tals. The tables should be viewed as a 
 systematic attempt to consider identifi- 
 able factors that contribute to process- 
 ing cost. 
 
19 
 
 o 
 
 I 
 
 o 
 
 *> 
 
 a. 
 O 
 i/i 
 
 u 
 O 
 
 « 
 
 I. 
 a 
 
 o 
 o 
 
 
 
 
 
 
 
 
 
 , 
 
 
 , 
 
 
 1 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ^ 
 
 • 
 
 c 
 
 E 
 
 L. 
 
 
 
 • 
 
 
 
 
 
 
 
 
 c 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 T? 
 
 s 
 
 O 
 
 — 
 
 
 
 a. 
 
 
 — 
 
 
 i 
 
 
 
 
 O 
 
 
 
 
 
 
 
 
 C 
 
 
 
 ■«- 
 
 • 
 
 o 
 
 9 
 
 o 
 
 4- 
 
 (J 
 
 
 
 o 
 
 
 (0 
 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 
 
 ID 
 
 ■D 
 
 I. 
 
 E 
 
 c 
 
 4- 
 
 9 
 
 
 • 
 
 4- 
 
 in 
 
 •— 
 
 
 l_ 
 
 
 
 
 in 
 
 
 
 
 
 
 
 
 
 » 
 
 
 
 
 M 
 
 
 .^ 
 
 o 
 
 L. 
 
 
 — 
 
 
 9 
 
 o 
 
 
 l- 
 
 
 
 
 3 
 
 
 
 
 
 
 
 
 +- 
 
 ■♦- 
 
 
 ■o 
 
 — 
 
 
 ID 
 
 c 
 
 J3 
 
 
 
 ID 
 
 4- 
 
 4- 
 
 L. 
 
 
 o 
 
 
 
 
 •»- 
 
 
 C71 
 
 
 
 
 
 
 « 
 
 c 
 
 
 9 
 
 o 
 
 4- 
 
 — 
 
 3 
 
 
 4- 
 
 
 > 
 
 ID 
 
 (D 
 
 ffl 
 
 
 o 
 
 
 
 
 *4- 
 
 
 C 
 
 9 
 
 
 
 
 
 
 0) 
 
 
 O 
 
 Ul 
 
 1_ 
 
 O 
 
 
 E 
 
 — 
 
 
 o 
 
 
 4- 
 
 E 
 
 • 
 
 
 
 C 
 
 
 — 
 
 
 — 
 
 4- 
 
 
 
 
 
 ^ 
 
 ^ 
 
 
 D 
 
 
 ID 
 
 
 Ul 
 
 o 
 
 (D 
 
 • 
 
 
 •D 
 
 — 
 
 E 
 
 T3 
 
 • ^ 
 
 
 O 
 
 
 -o 
 
 • 
 
 4- 
 
 ID 
 
 
 
 
 
 — 
 
 
 -D 
 
 ID 
 
 O. 
 
 9 
 
 4- 
 
 t_ 
 
 CO 
 
 4- 
 
 9 
 
 — 
 
 — 
 
 <s 
 
 9 
 
 9 
 
 
 — 
 
 
 
 Ul 
 
 ID 
 
 i- 
 
 
 
 ••- 
 
 
 -t- 
 
 n 
 
 
 O 
 
 
 
 Li. 
 
 Ul 
 
 »*- 
 
 
 — 
 
 l_ 
 
 — 
 
 — 
 
 E 
 
 > 
 
 
 4- 
 
 
 >- 
 
 
 1_ 
 
 o 
 
 
 
 c 
 
 
 Ifl 
 
 o 
 
 
 1- 
 
 Ul 
 
 •— 
 
 
 o 
 
 
 
 3 
 
 
 o 
 
 U 
 
 
 
 ~ 
 
 
 ID 
 
 
 X) 
 
 9 
 
 XI 
 
 
 
 i 
 
 
 
 i_ 
 
 
 ■t- 
 
 ID 
 
 
 »*- 
 
 o 
 
 ^ 
 
 « 
 
 O 
 
 •«3- 
 
 Ul 
 
 ID 
 
 
 ID 
 
 Ul 
 
 
 I- 
 
 
 
 — 
 
 Q. 
 
 ID 
 
 
 
 
 T3 
 
 ••- 
 
 
 C 
 
 
 
 O 
 
 
 
 4- 
 
 — 
 
 
 
 **- 
 
 • 
 
 — 
 
 c 
 
 • 
 
 9 
 
 
 T3 
 
 J3 
 
 O 
 
 — 
 
 
 
 B 
 
 
 
 
 
 — 
 
 c 
 
 >> 
 
 
 « 
 
 o 
 
 — 
 
 ^4- 
 
 CJ 
 
 Ul 
 
 
 — 
 
 O 
 
 9 
 
 (/> 
 
 o. 
 
 
 9 
 
 O 
 
 
 9 
 
 
 
 8 
 
 
 ^ 
 
 ^ 
 
 
 
 E 
 
 CM 
 
 — 
 
 Ul 
 
 1- 
 
 3 
 
 »4- 
 
 1_ 
 
 ID 
 
 — 
 
 ID 
 
 
 o. 
 
 § 
 
 o 
 
 
 4- 
 
 i- 
 
 «+- 
 
 
 
 
 
 ■♦- 
 
 ID 
 
 
 4- 
 
 3 
 
 — 
 
 ID 
 
 9 
 
 M 
 
 o 
 
 — 
 
 M 
 
 
 
 > 
 
 c 
 
 X 
 
 
 
 — 
 
 o. 
 
 O 
 
 • ^ 
 
 
 
 
 
 o 
 
 O 
 
 
 — 
 
 
 o 
 
 > 
 
 Ul 
 
 
 — 
 
 "O 
 
 
 ■D 
 
 — 
 
 o 
 
 O 
 
 9 
 
 — 
 
 4- 
 
 
 E 
 
 
 
 Ul 
 
 • 
 
 
 
 
 ^ 
 
 
 
 « 
 
 O 
 
 c 
 
 o 
 
 Ul 
 
 »♦- 
 
 *4- 
 
 
 Ul 
 
 9 
 
 A 
 
 i 
 
 • — 
 
 
 XI 
 
 C 
 
 
 
 • * 
 
 9 
 
 c 
 
 4- 
 
 
 
 
 
 1 
 
 
 U) 
 
 O 
 
 to 
 
 E 
 
 O 
 
 O 
 
 »♦- 
 
 c 
 
 ■o 
 
 ■o 
 
 — 
 
 4- 
 
 4- 
 
 O 
 
 ID 
 
 • 
 
 — 
 
 T3 
 
 CT) 
 
 O 
 
 c 
 
 
 
 
 V) 
 
 
 
 
 
 9 
 
 
 
 — 
 
 O 
 
 
 ■o 
 
 o 
 
 i_ 
 
 ID 
 
 C 
 
 l_ 
 
 
 ■D 
 
 
 — 
 
 c 
 
 *— 
 
 9 
 
 
 
 
 o 
 
 £ 
 
 
 ■»- 
 
 *♦- 
 
 Ul 
 
 l- 
 
 
 — 
 
 T3 
 
 ~ 
 
 (D 
 
 ID 
 
 ^ 
 
 
 O 
 
 9 
 
 Ck. 
 
 Q. 
 
 9 
 
 Ul 
 
 — 
 
 ID 
 
 4- 
 
 K 
 
 
 
 
 c 
 
 o 
 
 
 c 
 
 o 
 
 4- 
 
 
 CSl 
 
 ID 
 
 
 4- 
 
 
 
 
 '* 
 
 — 
 
 §. 
 
 
 
 E 
 
 Ul 
 
 o 
 
 I. 
 
 • — 
 
 
 
 
 — 
 
 
 
 9 
 
 
 l_ 
 
 9 
 
 — 
 
 > 
 
 4- 
 
 (D 
 
 CN 
 
 4- 
 
 l«-l 
 
 ^ 
 
 — 
 
 c 
 
 4- 
 
 
 9 
 
 Ul 
 
 
 ■o 
 
 •— 
 
 
 
 
 -Q 
 
 •o 
 
 
 > 
 
 a. 
 
 ID 
 
 E 
 
 o 
 
 o 
 
 — 
 
 — 
 
 — 
 
 — 
 
 
 CJ 
 
 O. 
 
 — 
 
 O 
 
 O 
 
 ID 
 
 O 
 
 
 9 
 
 c 
 
 3 
 
 
 
 
 E 
 
 c 
 
 
 — 
 
 O 
 
 O. 
 
 O 
 
 c 
 
 E 
 
 ID 
 
 3 
 
 o 
 
 ID 
 
 o 
 
 1- 
 
 Q. 
 
 3 
 
 — 
 
 
 
 — 
 
 O 
 
 c 
 
 TJ 
 
 O 
 
 cr 
 
 
 
 
 cS 
 
 ID 
 
 
 O 
 
 4- 
 
 
 </5 
 
 m 
 
 9 
 
 Ul 
 
 U 
 
 Ol 
 
 to 
 
 
 M 
 
 (D 
 
 CT 
 
 Ul 
 
 — 
 
 U 
 
 1_ 
 
 • — 
 
 •^ 
 
 u 
 
 9 
 
 
 
 
 
 
 to 
 
 
 (M 
 
 
 
 tr 
 
 
 
 2 
 
 
 < 
 
 
 
 LU 
 
 
 Q_ 
 
 
 CL 
 
 
 S 
 
 
 
 
 1/1 
 
 o 
 
 2 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 3 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 C 
 
 ■t- 
 
 1 
 
 
 
 «Mh 
 
 
 ^m^ 
 
 
 
 + 
 
 
 
 + 
 
 
 + 
 
 
 
 1 
 
 
 1 
 
 
 ««*» 
 
 
 ^-^ 
 
 
 
 
 — 
 
 c 
 
 1 
 
 
 
 + 
 
 
 + 
 
 
 
 + 
 
 
 
 + 
 
 
 + 
 
 
 
 1 
 
 
 1 
 
 
 + 
 
 
 1 
 
 
 
 
 ■t- 
 
 (D 
 
 %„^ 
 
 
 
 %.^ 
 
 
 
 
 
 
 
 
 ^^ 
 
 
 •^^ 
 
 
 
 ^^ 
 
 
 %_^ 
 
 
 ^.^ 
 
 
 s.^ 
 
 
 
 
 C 
 
 +- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 cS 
 
 O 
 o. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1 
 
 >■ 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ■♦- 
 
 ^^ 
 
 
 
 + 
 
 
 + 
 
 
 
 + 
 
 
 
 + 
 
 
 + 
 
 
 
 + 
 
 
 + 
 
 
 + 
 
 
 + 
 
 
 
 o 
 
 ^ 
 
 — 
 
 o 
 
 
 
 + 
 
 
 + 
 
 
 
 + 
 
 
 
 + 
 
 
 + 
 
 
 
 + 
 
 
 + 
 
 
 + 
 
 
 + 
 
 
 
 t- 
 
 •^ 
 
 '<,.>> 
 
 
 
 <^^ 
 
 
 *m^ 
 
 
 
 <ta>r 
 
 
 
 ^.^ 
 
 
 <h^ 
 
 
 
 *m^ 
 
 
 *h^ 
 
 
 «MI' 
 
 
 «M« 
 
 
 
 ^ 
 
 o 
 o 
 
 J3 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1 
 
 X 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 a> 
 
 
 ,.~« 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ^^ 
 
 
 ^% 
 
 
 
 
 ^>^ 
 
 
 
 
 
 >- 
 
 1 
 
 
 
 ^■^ 
 
 
 ^mt, 
 
 
 
 ^-N 
 
 
 
 ^■^ 
 
 
 ^l«k 
 
 
 
 + 
 
 
 + 
 
 
 ^•^ 
 
 
 1 
 
 
 
 
 a. 
 
 ■f- 
 
 1 
 
 
 
 1 
 
 
 1 
 
 
 
 + 
 
 
 
 + 
 
 
 1 
 
 
 
 + 
 
 
 + 
 
 
 + 
 
 
 1 
 
 
 
 
 E 
 
 •— 
 
 ^.^ 
 
 
 
 ^^ 
 
 
 ^^ 
 
 
 
 «H^ 
 
 
 
 Sw^ 
 
 
 ^•^ 
 
 
 
 ^■^ 
 
 
 N_^ 
 
 
 ^■^ 
 
 
 ^^ 
 
 
 
 
 cS 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 c 
 
 i- 
 
 § 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 9 
 
 — 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ID 
 
 
 
 
 Q. 
 
 I. 0) 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 • 
 
 
 
 ••- 
 
 
 -o a; 
 
 •— 
 
 *-% 
 
 
 r^ 
 
 ^ 
 
 r- 
 
 --* 
 
 
 P- 
 
 ^ 
 
 
 r- 
 
 ^^ 
 
 r~- 
 
 ^^ 
 
 
 t- 
 
 ^ 
 
 r- 
 
 *^ 
 
 o 
 
 ^ 
 
 CM 
 
 ^^ 
 
 
 a 
 
 I. 
 
 — ID 
 
 • 
 
 1 
 
 
 • 
 
 1 
 
 • 
 
 1 
 
 
 • 
 
 1 
 
 
 • 
 
 1 
 
 • 
 
 1 
 
 
 • 
 
 1 
 
 • 
 
 1 
 
 • 
 
 1 
 
 1 
 
 1 
 
 
 i- 
 
 o 
 
 — -t- 
 
 — 
 
 1 
 
 
 — 
 
 ,.^ 
 
 »— 
 
 
 
 ■ 
 
 
 
 *— 
 
 
 ^- 
 
 
 
 •— 
 
 
 . — 
 
 %^ 
 
 CSl 
 
 >^ 
 
 
 1 
 
 
 o 
 
 •«- 
 
 — in 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 • 
 
 
 
 a. 
 
 u 
 
 3 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 »— 
 
 
 
 e 
 
 « 
 
 a- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 (/) 
 
 ■*" 
 
 ID 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ^ 
 
 
 
 
 
 
 
 
 
 
 
 
 
 • 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 irt 
 
 
 
 ^-^ 
 
 
 
 ^^ 
 
 
 .—^ 
 
 
 
 ^^ 
 
 
 o 
 
 
 
 ^^ 
 
 
 O 
 
 ^^ 
 
 
 ^^ 
 
 
 ,-^ 
 
 O 
 
 <-^ 
 
 
 
 in 
 
 ^ 
 
 
 + 
 
 
 
 4- 
 
 
 4- 
 
 
 
 4- 
 
 
 ID 
 
 ^^ 
 
 
 + 
 
 
 r- 
 
 1 
 
 o 
 
 1 
 
 
 4- 
 
 CM 
 
 1 
 
 
 
 e 
 
 ^- 
 
 + 
 
 
 «— 
 
 4- 
 
 "— 
 
 4- 
 
 
 •— 
 
 4- 
 
 
 > 
 
 4- 
 
 '— 
 
 + 
 
 
 1 
 
 1 
 
 U^ 
 
 1 
 
 •— 
 
 4- 
 
 1 
 
 1 
 
 
 
 t- 
 
 a 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 
 
 
 
 
 
 
 
 
 Q. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 •~ 
 
 
 
 
 
 rO 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 
 
 o 
 
 
 o 
 
 
 
 o 
 
 
 
 O 
 
 
 o 
 
 
 
 O 
 
 
 o 
 
 
 o 
 
 
 O 
 
 
 
 
 h 
 
 
 
 
 
 o 
 
 
 in 
 
 
 
 o 
 
 
 
 o 
 
 
 o 
 
 
 
 (N 
 
 
 o 
 
 
 ■^ 
 
 
 o 
 
 
 
 
 a. 
 
 
 ^ 
 
 ^-v 
 
 
 TT 
 
 *-v 
 
 ^ 
 
 ^-* 
 
 
 VO 
 
 ^^ 
 
 
 in 
 
 ^^ 
 
 u^ 
 
 ^•^ 
 
 
 U^ 
 
 -^ 
 
 in 
 
 *-s 
 
 fn 
 
 ^-v 
 
 r-- 
 
 ^^ 
 
 
 
 E 
 
 o 
 
 1 
 
 + 
 
 
 1 
 
 4- 
 
 1 
 
 4- 
 
 
 1 
 
 1 
 
 
 1 
 
 O 
 
 1 
 
 O 
 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 4- 
 
 1 
 
 1 
 
 
 
 « 
 
 e 
 
 ^ 
 
 N.^ 
 
 
 o 
 
 
 o 
 
 
 
 o 
 
 
 
 o 
 
 
 o 
 
 
 
 O 
 
 • 
 
 o 
 
 -^ 
 
 o 
 
 4- 
 
 o 
 
 1 
 
 
 
 1— 
 
 
 
 
 
 
 
 in 
 
 (SI 
 
 
 
 in 
 
 
 
 in 
 
 
 in 
 
 
 
 in 
 
 
 o 
 
 
 CO 
 CM 
 
 
 CM 
 
 ^ 
 
 
 
 
 
 
 
 
 9 
 
 
 9 
 
 
 
 1 
 
 
 
 1 
 
 
 -D 
 
 
 
 1. 
 
 
 
 
 • 
 
 
 
 
 
 
 
 
 
 
 
 T3 
 
 
 — 
 
 • 
 
 
 10 
 
 
 
 ID 
 
 
 9 
 
 
 
 o 
 
 
 
 
 c 
 
 
 1 
 
 • 
 
 
 
 
 
 
 
 
 — 
 
 
 •^ 
 
 (Nl 
 
 
 ^ 
 
 1 
 
 
 — 
 
 1 
 
 — 
 
 
 
 Q. 
 
 
 
 
 o 
 
 
 1- 
 
 c 
 
 
 
 
 
 
 >- 
 
 
 I. 
 
 1. 
 
 4- 
 
 —~ 
 
 
 o 
 
 1 
 
 
 o 
 
 1 
 
 4- 
 
 
 
 ID 
 
 
 
 
 
 
 o 
 
 o 
 
 
 
 
 
 
 I. 
 
 
 o 
 
 o 
 
 a 
 
 o 
 
 
 > 
 
 4- 
 
 
 > 
 
 4- 
 
 — 
 
 
 
 > 
 
 
 
 
 •1- 
 
 
 
 — 
 
 
 
 
 
 X 
 
 « 
 
 
 
 
 — 
 
 C 
 
 
 c 
 
 C 
 
 
 c 
 
 C 
 
 x> 
 
 
 
 
 
 
 
 10 
 
 
 
 4- 
 
 
 
 
 
 • 
 
 > 
 
 
 x: 
 
 « 
 
 o 
 
 CO 
 
 
 o 
 
 9 
 
 
 o 
 
 9 
 
 o 
 
 
 
 ■o 
 
 
 
 
 c 
 
 
 4- 
 
 ID 
 
 
 
 
 
 
 o 
 
 
 u 
 
 _ 
 
 > 
 
 1 
 
 
 c 
 
 > 
 
 • 
 
 c 
 
 > 
 
 E 
 
 
 
 •^ 
 
 
 JC 
 
 
 o 
 
 
 — 
 
 — 
 
 
 
 c 
 
 
 i 
 
 o 
 
 
 
 o 
 
 
 1 
 
 
 
 ^ 
 
 
 
 
 — 
 
 
 • 
 
 
 3 
 
 
 o 
 
 
 
 
 ID 
 
 — 
 
 
 
 O 
 
 
 9 
 
 
 a 
 
 ^ 
 
 n 
 
 4- 
 
 
 ID 
 
 O 
 
 o 
 
 ID 
 
 O 
 
 ID 
 
 4- 
 
 
 CT 
 
 
 +■ 
 
 
 +■ 
 
 
 in 
 
 •^ 
 
 
 
 -" 
 
 
 o 
 
 L. 
 
 
 
 
 
 C 
 
 
 
 Ul 
 
 ^ 
 
 
 U) 
 
 
 C 
 
 
 *^ 
 
 
 ID 
 
 
 u 
 
 
 
 4- 
 
 
 
 ■♦- 
 
 
 u 
 
 1 
 
 
 c 
 
 « 
 
 c 
 
 9 
 
 
 C 
 
 
 1 
 
 C 
 
 
 c 
 
 9 
 
 
 _ 
 
 
 n 
 
 
 10 
 
 
 9 
 
 Ul 
 
 
 
 o. 
 
 
 
 1 
 
 
 *— 
 
 ^ 
 
 — 
 
 > 
 
 
 — 
 
 9 
 
 
 — 
 
 9 
 
 » 
 
 > 
 
 
 
 • 
 
 
 • 
 
 i_ 
 
 
 ^ 
 
 >— 
 
 
 
 — 
 
 
 o 
 
 C 
 
 
 
 o 
 
 
 — 
 
 
 
 X) 
 
 o 
 
 
 •o 
 
 
 — 
 
 
 9 
 
 c 
 
 9 
 
 c 
 
 «4- 
 
 
 JH 
 
 T3 
 
 
 
 I. 
 
 
 +- 
 
 O 
 
 
 c 
 
 « 
 
 c 
 
 O 
 
 
 c 
 
 
 ID 
 
 c 
 
 — • 
 
 c 
 
 O 
 
 
 l_ 
 
 o 
 
 I- 
 
 O 
 
 
 
 3 
 
 
 
 
 o 
 
 
 
 '— 
 
 
 o 
 
 z 
 
 o 
 
 Ul 
 
 
 O 
 
 l_ 
 
 z 
 
 O 
 
 l_ — 
 
 O 
 
 Ul 
 
 
 3 
 
 •— 
 
 3 
 
 — 
 
 c 
 
 
 o 
 
 TJ 
 
 
 
 tfl 
 
 
 c 
 
 +- 
 
 
 
 1 • 
 
 
 
 
 
 O 
 
 
 
 O O 
 
 
 
 
 Ul 
 
 4- 
 
 Ul 
 
 4- 
 
 o 
 
 
 "O 
 
 C 
 
 
 
 9 
 
 
 o 
 
 c 
 
 • 
 
 +- 
 
 1 — 
 
 4- 
 
 9 
 
 
 4- 
 
 
 u 
 
 4- 
 
 — ID 
 
 4- 
 
 9 
 
 
 Ul 
 
 ID 
 
 Ul 
 
 ID 
 
 — 
 
 
 
 ID 
 
 
 
 Q 
 
 
 — 
 
 ^ 
 
 m 
 
 c 
 
 •t- o 
 
 « 
 
 ■D 
 
 
 ID 
 
 ^ 
 
 — 
 
 ID 
 
 ^ z 
 
 ID 
 
 ■o 
 
 
 9 
 
 — 
 
 9 
 
 — 
 
 4- 
 
 
 ■o 
 
 
 
 
 
 
 in 
 
 — 
 
 — 
 
 — 
 
 c b<: 
 
 — 
 
 — 
 
 
 — 
 
 u 
 
 4- 
 
 — 
 
 O 1 
 
 — 
 
 — 
 
 
 1- 
 
 — 
 
 L. 
 
 — 
 
 ID 
 
 
 9 
 
 C 
 
 
 
 
 
 I. 
 
 •^ 
 
 CJ 
 
 — 
 
 9 1 
 
 — 
 
 I- 
 
 
 — 
 
 
 y 
 
 — 
 
 (Nl 
 
 — 
 
 L 
 
 
 Q. 
 
 — 
 
 CL 
 
 — 
 
 E 
 
 
 c 
 
 o 
 
 
 
 
 
 « 
 
 ■*- 
 
 qL 
 
 — 
 
 > — 
 
 — 
 
 O 
 
 
 •— 
 
 9 
 
 9 
 
 — 
 
 9 — 
 
 — 
 
 o 
 
 
 1 
 
 4- 
 
 i 
 
 4- 
 
 
 
 •^ 
 
 •^ 
 
 
 
 
 
 > 
 
 in 
 
 
 4- 
 
 — CJ 
 
 4- 
 
 
 
 4- 
 
 
 4- 
 
 4- 
 
 — O 
 
 4- 
 
 
 
 ^ 
 
 Ul 
 
 Ul 
 
 — 
 
 
 a 
 
 4- 
 
 
 
 
 
 c 
 
 — 
 
 •♦- 
 
 in 
 
 o a 
 
 Ul 
 
 ^ 
 
 
 U) 
 
 — 
 
 3 
 
 Ul 
 
 — C71 
 
 Ul 
 
 -C 
 
 
 C71 
 
 •~- 
 
 CD 
 
 — 
 
 ^ 
 
 
 E 
 
 ID 
 
 
 
 
 
 (S 
 
 ■o 
 
 o 
 
 — 
 
 U) z 
 
 — 
 
 u 
 
 
 — 
 
 4- 
 
 9 
 
 — 
 
 ■t- Z 
 
 — 
 
 u 
 
 
 
 T3 
 
 
 ■o 
 
 3 
 
 
 cS 
 
 E 
 
 
 
 
 
 
 
 o 
 
 
 Q 
 
 
 
 Q 
 
 
 
 o 
 
 
 o 
 
 
 
 X 
 
 
 X 
 
 
 to 
 
 
 
 
 
 
 
 .. 
 
 
 
 
 
 • 
 
 
 
 
 
 
 
 
 • 
 
 
 
 • 
 
 
 • 
 
 
 ^^ 
 
 
 • 
 
 
 
 
 
 ,-^ 
 
 ^~^ 
 
 
 
 
 
 ^-. 
 
 
 
 
 
 
 ^-^ 
 
 
 • 
 
 
 
 • 
 
 
 • 
 
 
 CJ> 
 
 
 • 
 
 
 
 
 
 •t- 
 
 ^ 
 
 
 
 
 
 r- 
 
 
 
 
 
 
 ■o 
 
 
 • 
 
 
 
 ^^^ 
 
 
 ,,1^ 
 
 
 <o 
 
 
 ^m^ 
 
 
 
 
 
 C 
 
 ir\ 
 
 
 » 
 
 
 
 m 
 
 
 
 
 
 
 9 
 
 
 • 
 
 
 
 CO 
 
 
 VO 
 
 
 ON 
 
 
 <N 
 
 
 
 
 
 e 
 
 c^ 
 
 X 
 
 c 
 
 
 
 0» 
 
 
 
 
 
 
 
 
 ^^ 
 
 
 
 m 
 
 
 r» 
 
 
 
 
 f^ 
 
 
 
 
 m 
 
 4- 
 
 
 9 
 
 o 
 
 ^^ 
 
 
 
 
 
 ^^ 
 
 
 
 >*- 
 
 
 r» 
 
 
 
 o« 
 
 
 o\ 
 
 
 ^^ 
 
 
 o\ 
 
 
 
 
 i/t 
 
 a 
 
 ^^ 
 
 _ 
 
 — 
 
 o 
 
 
 ^ 
 
 
 
 lO 
 
 
 
 tmm 
 
 
 r- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 e 
 
 o. 
 
 
 ^ 
 
 ■♦- 
 
 in 
 
 
 
 
 
 r~ 
 
 
 
 ■o 
 
 
 a> 
 
 
 
 ^^ 
 
 
 ^^ 
 
 
 L. 
 
 
 ^ 
 
 
 
 
 u 
 
 
 m 
 
 c 
 
 o 
 
 
 4- 
 
 
 
 <y> 
 
 
 
 o 
 
 
 
 
 
 
 
 
 
 9 
 
 
 
 
 
 
 o 
 
 •*- 
 
 3 
 
 o 
 
 ^ 
 
 
 
 4- 
 
 
 
 
 
 
 E 
 
 • 
 
 ^^ 
 
 
 
 CT) 
 
 
 n 
 
 
 CT) 
 
 
 s 
 
 
 
 
 I. 
 
 o 
 
 — 
 
 <J 
 
 — 
 
 *^ 
 
 
 9 
 
 
 
 s^ 
 
 
 
 
 ^^ 
 
 
 
 
 1. 
 
 
 j£ 
 
 
 I. 
 
 
 
 
 
 0. 
 
 
 — 
 
 
 •^ 
 
 
 
 c 
 
 
 
 
 
 
 
 •— 
 
 c 
 
 
 
 9 
 
 
 3 
 
 
 9 
 
 
 3 
 
 
 
 
 
 L. 
 
 e 
 
 in 
 
 ■♦- 
 
 c 
 
 
 J« 
 
 
 
 J£ 
 
 
 
 j: 
 
 QO 
 
 O 
 
 
 
 ^ 
 
 
 N 
 
 
 X> 
 
 
 N 
 
 
 
 
 
 « 
 
 c 
 
 — 
 
 m 
 
 O 
 
 
 U 
 
 
 
 c 
 
 
 
 c 
 
 O 
 
 U) 
 
 
 
 1 
 
 
 — 
 
 
 ■o 
 
 
 — 
 
 
 
 
 
 • 
 
 t. 
 
 O 
 
 — 
 
 ♦- 
 
 
 3 
 
 
 
 •^ 
 
 
 
 — 
 
 
 U) 
 
 
 
 
 £ 
 
 
 — 
 
 
 jC 
 
 
 
 
 
 >> 
 
 e 
 
 Q. 
 
 TJ 
 
 « 
 
 
 — 
 
 
 
 Q. 
 
 
 
 o. 
 
 N^ 
 
 9 
 
 
 
 1. 
 
 
 Ul 
 
 
 O 
 
 
 Ul 
 
 
 
 
 
 ^^ 
 
 u. 
 
 
 
 Ul 
 
 
 a. 
 
 
 
 «/) 
 
 
 
 to 
 
 
 GO 
 
 
 
 CD 
 
 
 
 
 CD 
 
 
 — 
 
 
 
 s.^ -- ^ O 
 
 <2 
 
20 
 
 o 
 
 M 
 
 •o 
 
 c: 
 
 10 
 
 o 
 
 4- 
 
 X 
 
 O. 
 (D 
 10 
 
 I- 
 
 o 
 
 <D 
 (/) 
 I/) 
 
 (J 
 O 
 t- 
 Cl 
 
 O 
 ui 
 
 I. 
 
 ID 
 
 a. 
 
 E 
 O 
 
 o 
 
 CD 
 < 
 
 
 
 
 
 
 <D 
 
 
 
 
 
 
 
 
 (D 
 
 
 
 
 
 
 • 
 c 
 
 
 
 
 
 
 ^ 
 
 
 
 
 
 J- 
 
 
 
 — 
 
 
 
 
 
 
 O 
 
 
 
 
 
 
 x> 
 
 
 
 
 
 o 
 
 1 
 
 
 XI 
 
 • 
 
 1 
 
 
 • 
 
 1 
 
 
 
 
 
 • 
 
 
 •— 
 
 1 
 
 
 c 
 
 
 
 •— 
 
 
 — 
 
 C 
 
 1- 
 
 
 +■ 
 
 — 
 
 ■H 
 
 
 
 
 l_ 
 
 
 in 
 
 o 
 
 
 O 
 
 
 •D 
 
 l_ 
 
 
 in 
 
 O 
 
 o 
 
 
 o 
 
 in 
 
 10 
 
 
 
 I- 
 
 o 
 
 
 in 
 
 3 
 
 
 
 
 — 
 
 O 
 
 
 in 
 
 
 o 
 
 
 
 
 in 
 
 c 
 
 
 
 O 
 
 ■(- 
 
 
 O 
 
 — 
 
 
 ■t- 
 
 
 — 
 
 ^ 
 
 
 O 
 
 ■1- 
 
 
 **. 
 
 Q. 
 
 
 
 •— 
 
 
 
 >i- 
 
 o 
 
 
 Q. 
 
 ^- 
 
 • 
 
 10 
 
 •o 
 
 o 
 
 
 
 CL 
 
 10 
 
 
 
 in 
 
 in 
 
 o 
 
 I- 
 
 -H 
 
 
 
 ID 
 
 
 
 
 I. 
 
 -t- 
 
 
 
 in 
 
 ■(- 
 
 
 
 c 
 
 L. 
 
 
 O 
 
 
 
 O 
 
 C 
 
 
 0) 
 
 (D 
 
 
 • •* 
 
 l_ 
 
 O 
 
 . — 
 
 (- 
 
 O 
 
 c 
 
 
 • ^ 
 
 ■— 
 
 3 
 
 
 
 i_ 
 
 c 
 
 
 0} 
 
 
 +- 
 
 l- 
 
 
 in 
 
 O 
 
 •H 
 
 O) 
 
 L. 
 
 3 
 
 
 
 • 
 
 in 
 
 £ 
 
 +- 
 
 — 
 
 Q. 
 
 
 j: 
 
 E 
 
 
 ID 
 
 
 
 u 
 
 ^4- 
 
 O 
 
 ID 
 
 — 
 
 
 l_ 
 
 c 
 
 O 
 
 ID 
 
 10 
 
 J3 
 
 
 +- 
 
 o 
 
 E 
 
 
 T3 
 
 Ifl 
 
 
 •— 
 
 
 10 
 
 
 ■(- 
 
 >*- 
 
 t. 
 
 — 
 
 •— 
 
 -t- 
 
 1- 
 
 o 
 
 
 
 o 
 
 
 
 cS 
 
 
 *— 
 
 •o 
 
 
 4- 
 
 
 
 
 
 — 
 
 in 
 
 
 3 
 
 — 
 
 -1- 
 
 c 
 
 
 
 1- 
 
 cn 
 
 3 
 
 1- 
 
 
 ■o 
 
 — 
 
 
 
 
 -H 
 
 1_ 
 
 ID 
 
 
 «^ 
 
 o 
 
 J^ 
 
 
 
 
 
 Q. 
 
 Q. 
 
 c 
 
 ■o 
 
 
 
 
 c 
 
 — 
 
 
 c 
 
 ID 
 
 
 O 
 
 • *^ 
 
 L. 
 
 L. 
 
 
 c 
 
 o 
 
 E 
 
 
 ID 
 
 o 
 
 ID 
 
 
 
 ID 
 
 O 
 
 
 •— 
 
 ■o 
 
 in 
 
 •— 
 
 -o 
 
 o 
 
 
 
 — 
 
 . — 
 
 
 
 
 C 
 
 l_ 
 
 1_ 
 
 
 
 
 O 
 
 in 
 
 
 ^ 
 
 — - 
 
 ■o 
 
 c 
 
 
 
 +■ 
 
 -H 
 
 ID 
 
 ^ 
 
 to 
 
 ■(- 
 
 o 
 
 1 
 
 Q. 
 
 10 
 
 
 
 
 O 
 
 
 
 ■o 
 
 — 
 
 10 
 
 ■o 
 
 o 
 
 C 
 
 •t- 
 
 
 
 
 
 U) 
 
 
 
 
 
 
 ■o 
 
 3 
 
 
 S 
 
 c 
 
 — 
 
 ^ 
 
 
 
 ID 
 
 3 
 
 c 
 
 I. 
 
 o 
 
 ^ 
 
 in 
 
 c 
 
 CM 
 
 -t- 
 
 
 
 O 
 
 
 
 O 
 
 ID 
 
 o 
 
 o 
 
 
 
 
 
 O 
 
 O 
 
 O 
 
 
 O) 
 
 o 
 
 O 
 
 O 
 
 10 
 
 
 
 o 
 
 M- 
 
 
 
 (J 
 
 in 
 
 
 
 c 
 
 i- 
 
 o 
 
 N 
 
 O 
 
 < 
 
 
 L. 
 
 
 L. 
 
 ■1- 
 
 
 
 o 
 
 
 
 1/1 
 
 
 
 s 
 
 
 
 
 
 Q- 
 
 
 X 
 
 
 
 M 
 
 
 U) 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 3 
 
 — 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 ID 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 3 
 
 • — 
 
 
 
 
 
 
 
 ^-^ 
 
 
 
 
 
 
 
 
 
 
 ^^ 
 
 
 C 
 
 +- 
 
 *-^ 
 
 
 
 *-N 
 
 
 
 + 
 
 
 
 
 
 ^% 
 
 
 ^-^ 
 
 
 
 + 
 
 
 •— 
 
 c 
 
 + 
 
 
 
 1 
 
 
 
 + 
 
 
 
 
 
 1 
 
 
 1 
 
 
 
 + 
 
 
 ■(- 
 
 (D 
 
 %_f 
 
 
 
 >.^ 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 c 
 
 ■t- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 (S 
 
 O 
 o. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 .i 
 
 >- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 — -H 
 
 ■(- 
 
 
 
 
 
 
 
 ^-k 
 
 
 
 
 
 
 
 ^•^ 
 
 
 
 ^-^ 
 
 
 — ID 
 
 •— 
 
 -— N 
 
 
 
 ^^ 
 
 
 
 + 
 
 
 
 
 
 <*-^ 
 
 
 1 
 
 
 
 1 
 
 
 P 9- 
 
 — 
 
 + 
 
 
 
 + 
 
 
 
 + 
 
 
 
 
 
 + 
 
 
 1 
 
 
 
 1 
 
 
 1- E 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 i^ O 
 
 ^ 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 O 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 i 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ® 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 «^^ 
 
 
 
 
 
 _ 
 
 >- 
 
 <— s 
 
 
 
 ^^ 
 
 
 
 «>-H 
 
 
 
 
 
 ^^ 
 
 
 1 
 
 
 
 ^•^ 
 
 
 Q. 
 
 -H 
 
 + 
 
 
 
 + 
 
 
 
 + 
 
 
 
 
 
 1 
 
 
 1 
 
 
 
 + 
 
 
 E 
 
 •— 
 
 N..^ 
 
 
 
 ^^ 
 
 
 
 >.<r 
 
 
 
 
 
 S^ 
 
 
 %_« 
 
 
 
 
 
 cS 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 C 1- 
 
 § 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 
 o 
 
 • 
 
 
 
 
 
 O <D 
 
 • _ 
 
 
 
 
 o 
 
 
 
 o 
 
 
 
 
 
 • 
 
 
 o 
 
 
 
 
 
 — Q. 
 
 1- (D 
 
 
 
 
 • 
 
 
 
 • 
 
 
 
 
 
 o 
 
 
 o 
 
 
 
 
 
 +- 
 
 -Q Ol 
 
 o 
 
 ^-* 
 
 
 CS 
 
 ,->. 
 
 
 ON 
 
 ^-v 
 
 
 
 
 CM 
 
 ^^ 
 
 tf\ 
 
 ^^ 
 
 
 in 
 
 ^-N 
 
 ID l_ 
 
 — (D 
 
 • 
 
 + 
 
 
 1 
 
 1 
 
 
 1 
 
 + 
 
 
 
 
 
 + 
 
 
 + 
 
 
 • 
 
 1 
 
 1- 
 
 — ■(- 
 
 ro 
 
 
 
 r~ 
 
 
 
 o 
 
 
 
 
 
 O 
 
 
 o 
 
 + 
 
 
 IN 
 
 S_^ 
 
 (D -K 
 
 — If) 
 
 
 
 
 • 
 
 
 
 • 
 
 
 
 
 
 • 
 
 
 • 
 
 
 
 
 
 Q. O 
 
 3 
 
 
 
 
 « — 
 
 
 
 CM 
 
 
 
 
 
 to 
 
 
 o 
 
 
 
 
 
 ® ID 
 
 CT 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 
 
 
 
 CO H- 
 
 <D 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1/) 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 l/> 
 
 $ 
 
 
 + 
 
 
 
 ^-* 
 
 
 
 + 
 
 
 
 
 
 --* 
 
 
 + 
 
 
 
 + 
 
 <D 
 
 •— 
 
 + 
 
 
 ^— 
 
 + 
 
 
 ^— 
 
 + 
 
 
 
 
 in 
 
 1 
 
 « — 
 
 + 
 
 
 •— 
 
 + 
 
 1_ 
 
 ID 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 CL 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 O 
 
 
 
 o 
 
 
 
 O 
 
 
 
 
 
 o 
 
 
 o 
 
 
 
 O 
 
 
 •« 
 
 
 in 
 
 
 
 in 
 
 
 
 in 
 
 
 
 
 
 in 
 
 
 o 
 
 
 
 O 
 
 
 Q. 
 
 
 a\ 
 
 ^-* 
 
 
 '»• 
 
 ^^ 
 
 
 •* 
 
 ^^ 
 
 
 
 
 f 
 
 ^-« 
 
 c^ 
 
 j~^ 
 
 
 00 
 
 ^-^ 
 
 E 
 
 O 
 
 
 1 
 
 
 
 + 
 
 
 1 
 
 + 
 
 
 
 
 
 + 
 
 
 1 
 
 
 
 1 
 
 a> 
 
 o 
 
 o 
 
 1 
 
 
 o 
 
 
 
 in 
 
 
 
 
 
 in 
 
 
 o 
 
 1 
 
 
 o 
 
 1 
 
 1— 
 
 
 o 
 
 
 
 
 
 
 p- 
 
 
 
 
 
 hO 
 
 
 o 
 p- 
 
 
 
 CM 
 
 
 
 
 i 
 
 
 
 l_ 
 
 
 
 
 
 • 
 
 
 
 
 
 • i» 
 
 +-' 
 
 
 
 
 
 
 1- 
 
 
 
 o 
 
 
 
 
 
 
 
 
 
 • 
 
 
 
 -^ 
 
 
 
 
 
 
 o. 
 
 C 
 
 O 
 
 
 >4- 
 
 
 
 
 
 o 
 
 
 
 
 p 
 
 ■a 
 
 ID 
 
 in 
 
 c 
 
 CM 
 O 
 
 c 
 
 
 
 • « 
 
 — - 
 
 
 <•« 
 
 • ■^ 
 
 
 • ^ 
 
 
 M 
 
 
 
 
 c 
 
 1- 
 
 
 O 
 
 
 ID 
 
 
 
 c 
 
 in 
 
 
 c 
 
 ID 
 
 
 c 
 
 
 
 
 
 c 
 
 ID 
 
 o 
 
 — 
 
 — 
 
 >- 
 
 ^ 
 
 c 
 
 
 
 
 1- 
 
 
 O 
 
 
 
 O 
 
 
 + 
 
 
 
 •— 
 
 ■t- 
 
 3 
 
 < 
 
 +- 
 
 J^ 
 
 ■1- • 
 
 O 
 
 
 — 
 
 ® 
 
 
 •— 
 
 O 
 
 
 — 
 
 
 
 
 
 
 o 
 
 
 
 ID 
 
 
 T3 
 
 — 
 
 
 -1- 
 
 > 
 
 
 -1- 
 
 1_ 
 
 « 
 
 +■ 
 
 
 CN 
 
 
 
 •o 
 
 3 
 
 *4- 
 
 ^ 
 
 1- • 
 
 T3 
 
 I. 
 
 ■1- 
 
 
 o 
 
 c 
 
 • 
 
 o 
 
 M 
 
 c 
 
 o 
 
 t'l 
 
 
 
 
 © 
 
 T3 
 
 
 ■(- 
 
 -CI >. 
 
 
 
 -H 
 
 Q. 
 
 
 ID 
 
 o 
 
 (Ni 
 
 ID 
 
 CSI 
 
 O 
 
 10 
 
 
 o 
 
 
 
 > 
 
 
 
 O 
 
 
 — 1_ 
 
 ■t- 
 
 +- I- 
 
 •— 
 
 
 (D 
 
 o 
 
 o 
 
 
 
 ^ 
 
 — 
 
 
 
 o 
 
 i_ 
 
 
 
 ^ 
 
 l_ 
 
 +- 
 
 s 
 
 — 
 
 l- 
 
 in 
 
 l_ 
 
 
 1- 
 
 
 -4- 
 
 1- 
 
 
 +- 
 
 1_ 
 
 1_ 
 
 M 
 
 
 
 o 
 
 
 
 
 — > 
 
 
 
 ID > 
 
 O 
 
 
 
 — 
 
 X 
 
 
 v4- 
 
 • — 
 
 
 M 
 
 
 
 
 in 
 
 .. 
 
 c 
 
 c 
 
 3 O 
 
 > 
 
 ■4- C 
 
 in 
 
 
 X) 
 
 10 
 
 
 ■o 
 
 o 
 
 in 
 
 ■o 
 
 
 + 
 
 
 
 in 
 
 < 
 
 O 
 
 O 
 
 o- o 
 
 C 
 
 O 
 
 0) 
 
 
 • — 
 
 — 
 
 + 
 
 •— 
 
 
 O 
 
 
 ■\- 
 
 
 
 
 •— 
 
 
 
 
 
 
 O 
 
 CM U 
 
 Q 
 
 
 — 
 
 +- 
 
 
 — 
 
 c 
 
 a. 
 
 — 
 
 
 M^ 
 
 
 
 •o 
 
 
 in 
 
 -H 
 
 1_ 
 
 o 
 
 o 
 
 
 
 O 
 
 c 
 
 "* 
 
 O 
 
 o 
 
 
 o 
 
 ■* 
 
 
 
 
 
 1^ 
 
 L. 
 
 O 
 
 
 
 1- «* 
 
 
 
 in 
 
 (D 
 
 
 in 
 
 — 
 
 o 
 
 in 
 
 
 o 
 
 
 
 >* 
 
 
 
 
 3 
 
 ID (J 
 
 ^ 
 
 M — 
 
 
 
 t 
 
 1- 
 
 o 
 
 1 
 
 +- 
 
 o 
 
 I 
 
 O 
 
 i- 
 
 
 
 
 o 
 
 > 
 
 X) 
 
 ■1- C 
 
 
 „ O 
 
 
 
 in 
 
 ID 
 
 ^- 
 
 in 
 
 ID 
 
 
 
 in 
 
 1- 
 
 M 
 
 
 
 o 
 
 
 C 
 
 
 
 (B — 
 
 o 
 
 O t- 
 
 
 
 ID 
 
 M- 
 
 X 
 
 ID 
 
 E 
 
 ■o 
 
 ID 
 
 M 
 
 CS) 
 
 
 
 I. 
 
 < 
 
 O 
 
 1_ 
 
 E N 
 
 1_ 
 
 +■ X 
 
 
 
 CD 
 
 
 
 o 
 
 
 
 O 
 
 
 
 
 
 NJ 
 
 
 o 
 
 
 
 M 
 
 
 
 
 " 
 
 
 
 o 
 
 d) 
 
 
 
 
 
 
 
 
 • 
 
 
 
 
 
 
 
 P 
 
 
 
 
 
 ID 
 
 — 
 
 a 
 
 •^ 
 
 
 1 
 
 
 • 
 
 
 
 
 
 
 
 
 
 C 
 
 s 
 
 -o 
 
 
 N 
 
 J2 
 
 
 
 ^.^ 
 
 
 •+- 
 
 
 'S- 
 
 • > 
 
 o 
 
 
 
 
 
 
 
 (D 
 
 \o 
 
 — 
 
 
 ~ 
 
 3 
 
 O) 
 
 f^ 
 
 
 
 
 1 
 
 
 ^.^ 
 
 — l_ 
 
 
 • 
 
 
 
 
 {Tl 
 
 -1- 
 
 ON 
 
 X 
 
 
 1_ 
 
 o 
 
 C 
 
 in 
 
 -o 
 
 1- 
 
 
 
 o 
 
 ro 
 
 ID M 
 
 • • 
 
 X 
 
 
 
 
 lf> 
 
 10 
 
 
 o 
 
 
 ■1- 
 
 Q 
 
 ID 
 
 ON 
 
 C 
 
 CL 
 
 1_ 
 
 i_ 
 
 1^ 
 
 
 ,-^ 
 
 o 
 
 
 ^^ 
 
 
 ® 
 
 Q. 
 
 ^^ 
 
 <!) 
 
 • 
 
 3 
 
 
 ^ 
 
 
 ID 
 
 
 
 M 
 
 a\ 
 
 +- C 
 
 o\ 
 
 •o 
 
 
 — 
 
 
 O 
 
 
 
 
 
 Q 
 
 
 U 
 
 .s_^ 
 
 
 
 — 
 
 
 
 C O 
 
 r^ 
 
 
 
 
 lO 
 
 
 O 
 
 M- 
 
 i_ 
 
 13 
 
 O) 
 
 
 • • 
 
 X 
 
 
 1^ 
 
 • • 
 
 ID 
 
 c 
 
 ._^ 
 
 — 
 
 a\ 
 
 1- 
 
 
 cyi 
 
 
 L. 
 
 O 
 
 (D 
 
 — 
 
 c 
 
 in 
 
 ^^ 
 
 
 
 s 
 
 ^^ 
 
 ,-^ 
 
 •— 
 
 O 
 
 
 I- -H 
 
 
 
 
 
 
 a. 
 
 
 — 
 
 I- 
 
 ID 
 
 ID 
 
 P- 
 
 
 a\ 
 
 vD 
 
 -t- 
 
 — 
 
 c 
 
 o 
 
 ^^ 
 
 l_ 
 
 
 ^^ 
 
 
 
 I. 
 
 "O 
 
 o 
 
 ^ 
 
 O) 
 
 r— 
 
 ■t- 
 
 jz 
 
 in 
 
 vO 
 
 c 
 
 -1- 
 
 
 
 -^ 3 
 
 
 N 
 
 
 
 
 
 ID 
 
 c 
 
 — 
 
 o 
 
 c 
 
 CTv 
 
 
 c 
 
 o\ 
 
 ON 
 
 
 
 o 
 
 in 
 
 -o 
 
 >~ 
 
 ^ 
 
 
 ^ 
 
 
 
 
 
 ID 
 
 ^ 
 
 X 
 
 
 
 
 ID 
 
 » 
 
 
 
 I- 
 
 3 
 
 i_ 
 
 1_ 
 
 en 
 
 4- 
 
 
 1- 
 
 
 
 >- 
 
 ^ 
 
 o 
 
 
 
 — 
 
 ^.^ 
 
 in 
 
 
 
 «.^ 
 
 •^^ 
 
 
 
 ■o 
 
 ID 
 
 CL l_ 
 
 i 
 
 X 
 
 
 
 
 
 
 -' 
 
 O 
 
 
 
 Ll- 
 
 
 
 z 
 
 
 
 
 
 _1 
 
 
 
 
 CD 
 
 
 
 > 
 
 §. 
 
21 
 
 1. Temperature . — Elevated operating 
 temperatures involve increased energy 
 costs and the use of expensive materials 
 of construction. Increased maintenance 
 costs from corrosion and deterioration of 
 mechanical equipment also occur. 
 
 2. Pressure. — High pressure requires 
 the use of heavy-duty components and spe- 
 cial fabrication techniques. Added in- 
 spection is needed, and an element of 
 risk is added. Operation at pressure 
 less than 1 atm requires special vacuum 
 equipment and fabrication techniques. 
 
 3. Separation Factor . — Separation fac- 
 tor is important because it determines 
 the number of equilibrium stages required 
 to achieve separation. The number of 
 equilibrium stages needed reflects on the 
 amount of recycle, the number and size of 
 reaction vessels, material inventory, en- 
 ergy, and operating costs. The cost per 
 separation stage may not always be the 
 same . 
 
 4. Compatability with Kroll Technol- 
 ogy . — The inclusion of a separation pro- 
 cess that does not mesh well with the 
 Kroll process flow scheme would cause 
 the premature loss of usable facilities. 
 The design and construction costs for 
 
 replacement equipment are a deterrent for 
 such a choice. 
 
 5. Degree of Complexity . — Process com- 
 plexity is reflected in the number of 
 different steps required and in side 
 streams that must be treated. Process 
 complexity contributes to costs through 
 energy consumption, material inventory, 
 labor, and equipment. 
 
 6. Potential for Continuous Process- 
 ing. — Although a batch or semibatch pro- 
 cess is acceptable, a continuous process 
 is more desirable. Improved quality con- 
 trol, efficient use of energy, and lower 
 labor costs favor the continuous process. 
 In some cases the potential for continu- 
 ous operation is easy to assess. For 
 instance, the continuous operation of a 
 high-temperature, high-pressure distilla- 
 tion unit would be a difficult undertak- 
 ing. The continuous or even intermittent 
 introduction and removal of tetrachloride 
 from the high-pressure unit is a formida- 
 ble task and makes a batch operation a 
 more attractive alternative. On the oth- 
 er hand, continuous operation is already 
 claimed for the extractive distillation 
 separation process described by Besson 
 (5) and Brun (8). 
 
 CONCLUSIONS 
 
 At least one and possibly three non- 
 aqueous Hf-Zr separation processes show 
 promise for future commercial operation. 
 The economic impact of this development 
 on domestic zirconium producers who are 
 using the aqueous solvent extraction pro- 
 cess is unknown. 
 
 Only two separation processes are now 
 being studied. Of these, the extractive 
 distillation process of Cezus-Pechiney 
 has greater potential for commercial ap- 
 plication than the high-pressure dis- 
 tillation described by Ishizuka. This 
 judgment is based on considerations of 
 temperature, pressure, and potential for 
 continuous operation. Commercial-scale 
 production is already claimed for the 
 Cezus process (8). 
 
 Two promising separation techniques are 
 not being studied as far as can be deter- 
 mined. The Newnham process, based on the 
 selective reduction of ZrCl^, was studied 
 extensively during the 1960's by both the 
 Bureau of Mines and the National Distill- 
 ers Corp. The process is simple and has 
 high potential for continuous operation. 
 The extractive distillation process de- 
 scribed by Spink in Canada is similar to 
 that used by Cezus, with variations that 
 have promise for improved product purity 
 and more reliable operation. This work 
 was discontinued in 1981 because of lack 
 of funding. 
 
 The removal of minor impurities from 
 the products has not been solved. The 
 Brun description of the Cezus-Pechiney 
 
22 
 
 process claims a nuclear-grade ZrCl4 
 product but does not mention purification 
 steps for removal of iron, aluminum, and 
 other minor chloride impurities. The 
 
 authors believe that additional purifica- 
 tion is necessary regardless of the pri- 
 mary Hf-Zr separation process employed. 
 
 REFERENCES 
 
 1. American Chemical Society. Chemis- 
 try in the Economy. Washington, DC, 
 1973, pp. 61-62. 
 
 Application of Zirconium. Vancouver, 
 British Columbia, Canada, June 28-July 1, 
 1982, 11 pp. 
 
 2. American Society for Testing and 
 Materials. Standard Specification for 
 Zirconium Sponge and Other Forms of Vir- 
 gin Metal for Nuclear Application. B 
 349-73 in 1977 Annual Book of ASTM Stan- 
 dards. Part 8, Non-Ferrous Metals - 
 Nickel, Lead and Tin Alloys, Precious 
 Metals, Primary Metals, Reactive Metals. 
 Philadelphia, PA, 1977, pp. 260-262. 
 
 3. Asvestos, D. A., P. Pint, and S. N. 
 Flengas. Some Thermodynamic Properties 
 of the Solutions of ZrCl4 and HfCl4 in 
 CsCl Melts. Can. J. Chem. , v. 55, 1977, 
 pp. 1154-1166. 
 
 4. Berl, L. (assigned to National Dis- 
 tillers and Chemical Corp., New York, 
 NY). Process for the Separation of Zir- 
 conium and Hafnium Values. U.S. Pat. 
 3,012,850, Dec. 12, 1961. 
 
 9. Chandler, H. W. (assigned to Iso- 
 met Corp., Palisades Park, NJ). Method 
 of Separating Hafnium From Zirconium. 
 U.S. Pat. 3,276,862, Oct. 4, 1966. 
 
 10. Copley, D. B. , and R. A. J. Shel- 
 ton. The Disproportionation and Non- 
 Stoichiometry of Zirconium Trichloride. 
 J. Less-Common Metals, v. 20, 1970, pp. 
 359-366. 
 
 11. Dean, J. A. (ed.). Lange's Hand- 
 book of Chemistry. McGraw-Hill Book Co., 
 New York, 11th ed. , 1973, various paging. 
 
 12. Denisova, N. D. , and E. K. Safro- 
 nov. (Behavior of Aluminum Chloride in 
 the ZrCl4-AlCl3 System.) Dok. Akad. 
 Nauk. SSR, v. 183, No. 3, 1968, pp. 648- 
 651; abs. in Chem. Abstr. , v. 70, 1969, 
 No. 41392t. 
 
 5. Besson, P., J. Guerin, P. Brun, and 
 M. Bakes (assigned to Ugine Aciers , Par- 
 is, France). Process for the Separation 
 of Zirconium and Hafnium Tetrachlo- 
 ride From Mixtures Thereof. U.S. Pat. 
 4,021,531, May 3, 1977. 
 
 13. Denisova, N. D. , E. K. Safronov, 
 and 0. N. Bystrova. (Vapor Pressure and 
 Heat of Sublimation of Zirconium and Haf- 
 nium Tetrachlorides.) Russian J. Inorg. 
 Chem., V. 11, No. 10, 1966, pp. 1171- 
 1173. 
 
 6. Blumenthal, W. B. The Chemical Be- 
 havior of Zirconium. D. Van Nostrand 
 Company, Ltd., New York, 1958, p. 227. 
 
 7. Bromberg, M. L. (assigned to E. I. 
 duPont de Nemours and Co., Wilmington, 
 DE) . Purification of Zirconium Tetra- 
 chloride by Fractional Distillation. 
 U.S. Pat. 2,852,446, Sept. 16, 1958. 
 
 8. Brun, P. , P. Thouvenin, and L. Mou- 
 lin. New Process for Zirconium and Haf- 
 nium Separation. Preprint for Trans. 
 ASTM 6th Internat. Conf. on Nuclear 
 
 14. Denisova, N. D. , E. K. Safronov, 
 A. I. Pustil'nik, and 0. N. Bystrova. 
 Pogranichnaya Krivaya Zhidkost '-Pari Dav- 
 lenie Nasyshchennogo (Boundary Liquid- 
 Vapor Curve and Saturated Vapor Pressure 
 of Zirconium and Hafnium Tetrachlorides). 
 Russian J. Phys. Chem., v. 41, No. 1, 
 1967, pp. 30-33. 
 
 15. Dutrizac, J. E. , and S. N. Flen- 
 gas. Pressure Temperature Measurements 
 on Alkali and Alkaline Earth Complex 
 Chlorides of Zirconium and Hafnium. 
 Proc. Advan. Extr. Met. Symp. , London, 
 
23 
 
 1967. London, Institution of Mining and 
 Metallurgy, 1968, pp. 572-599. 
 
 16. Dutrizac, J. E., and S. N. Flen- 
 gas. (assigned to the Minister of Na- 
 tional Defense, Ottawa, Canada). Separa- 
 tion of Hafnium From Zirconium. Canadian 
 Pat. 863,258, Feb. 9, 1971. 
 
 17. Eaton, R. B. (assigned to E. I. 
 duPont de Nemours and Co. , Wilmington, 
 DE) . Process for Separating Hafnium Tet- 
 rachloride From Zirconium Tetrachloride. 
 U.S. Pat. 2,744,060, May 1, 1956. 
 
 18. Flengas, S. N., and J. E. Dutri- 
 zac. A New Process for the Separation of 
 Hafnium From Zirconium. Metall. Trans., 
 V. 8B, 1977, pp. 377-385. 
 
 19. Frampton, O.D. , and J. Feldman. 
 Purification of Zirconium by a Dry Pro- 
 cess. Chemical Separation of Hafnium 
 from Zirconium. Ch. in Progress in Sepa- 
 ration and Purification, ed. by E. S. 
 Perry, Interscience, New York, v. 1, 
 
 1968, pp. 247-295. 
 
 20. Frey, W. (assigned to Saurefabrik 
 Schweizerhall, Schweizerhalle, Switzer- 
 land). Process for the Production of 
 Purified Anhydrous Zirconium Tetrahalide. 
 U.S. Pat. 2,682,445, June 29, 1954. 
 
 21. Funaki , K. , and K. Uchimura. 
 (Dechlorination of Zirconium and Hafnium 
 Tetrachlorides With Oxygen and Its Appli- 
 cation to the Separation of the Two Ele- 
 ments.) Denki Kagaku, v. 33, No. 2, 
 1965, pp. 171-175; abs. in Chem. Abstr. , 
 V. 63, 1965, No. 7890g. 
 
 22. Gillot, J., and W. M. Goldberger. 
 Separation by Thin Film Fractional Subli- 
 mation. Chem. Eng. Progr. Symp. Ser. , v. 
 65, No. 91, 1969, pp. 36-42. 
 
 23. Glassner, A. The Thermo chemical 
 Properties of the Oxides, Fluorides and 
 Chlorides to 2500° C. U.S. Atomic Energy 
 Commission ANL-5750, 1957, 70 pp. 
 
 24. Goldberger, W. M. , and J. Gillot 
 (assigned to National Distillers and 
 
 Chemical Corp., Columbus, OH). Separa- 
 tion Methods for Volatile Solids. U.S. 
 Pat. 3,457,049, July 22, 1969. 
 
 25. Goodwin, J. G. Melting of Hafni- 
 um. Ch. in the Metallurgy of Hafnium, 
 ed. by D. E. Thomas and E. T. Hayes. 
 U.S. Atomic Energy Commission, 1960, 
 p. 139. 
 
 26. Greenberg, H. , and H. R. Lubowitz 
 (assigned to National Distillers and 
 Chemical Corp., New York, NY). Purifica- 
 tion Method for Metal Halides. U.S. Pat. 
 3,053,620, Sept. 11, 1962. 
 
 27. Haendler, H. M. , S. F. Bartram, 
 R. S. Becker, W. J. Bernard, and S. W. 
 Bukata. The Reaction of Fluorine With 
 Titanium, Zirconium, and Oxides of Tita- 
 nium (IV), Zirconium (IV), and Vanadium 
 (V). J. Am. Chem. Soc. , v. 76, 1954, pp. 
 2177-2178. 
 
 28. Hudswell, F., and J. M. Hutcheon. 
 The Manufacture of Hafnium-Free Zirconi- 
 um. Ch. in Extraction and . Refining of 
 the Rarer Metals (Symposium on Extraction 
 Metallurgy of Some of the Less Common 
 Metals, March 22-23, 1956). Institution 
 of Mining and Metallurgy, 1957, pp. 402- 
 419. 
 
 29. Ishizuka, H. (Separation of Zir- 
 conium and Hafnium Chlorides) . Japan 
 Kokai 76, 65,093, June 5, 1976. 
 
 30. 
 
 Separation of Zirconium 
 
 and Hafnium Tetrachlorides From a Mixture 
 Comprising Such Chlorides and Apparatus 
 for the Process. Eur. Pat. Appl. EP 
 45,270, Feb. 3, 1982. 
 
 31. 
 
 Method for Preparing Zir- 
 
 conium Tetrachloride and Hafnium Tetra- 
 chloride. U.S. Pat. 3,671,186, June 20, 
 1972. 
 
 32. Jacque, L. , and P. Dumez. La Sub- 
 limation fractionee continue sans reflux 
 et son application a la separation du 
 zirconium et de I'hafnium (Continuous 
 Fractional Sublimation Without Reflux 
 and Its Application to the Separation of 
 
24 
 
 Zirconium and Hafnium) . Chimie Industrie 
 Genie Chemique, v. 97, No. 11, 1967, pp. 
 1677-1684. 
 
 33. Jane eke, E. The System KCl-NaCl- 
 MgCl2. In Phase Diagrams for Ceramists, 
 1969 Supplement. American Ceramic Soci- 
 ety, Inc., Columbus, OH, 1969, p. 334 
 (fig. 3262). 
 
 34. Kipouros, G. J., and S. N. Flen- 
 gas. Equilibrium Decomposition Pressures 
 of the Compounds K2ZrCl5 and K2Hf Clg . 
 Can. J. Chem. , v. 56, 1978, pp. 1549- 
 1554. 
 
 35. . Equilibrium Decomposition 
 
 Pressures of the Compounds Na2ZrCl5 and 
 Na2HfCl6. Can. J. Chem., v. 59, 1981, 
 pp. 990-995. 
 
 36. Klemm, W. , and P. Weiss. NaCl- 
 MgCl2 System. In Phase Diagrams for Ce- 
 ramists. American Ceramic Society, Inc. , 
 Columbus, OH, 1956, p. 219 (fig. 639). 
 
 37. Lange, N. A. (ed). Lange's Hand- 
 book of Chemistry. Handbook Publishers, 
 Inc. Sandusky, OH, 9th ed, 1956, p. 93. 
 
 38. Larsen, E. M. , and F. C. Gil-Arnao 
 (assigned to Wisconsin Alumni Research 
 Foundation, Madison, WI). Process for 
 Separation of Hafnium From Zirconium. 
 U.S. Pat. 3,734,689, May 22, 1973. 
 
 39. Lustman, B, , and F. Kerze, Jr. 
 The Metallurgy of Zirconium. McGraw-Hill 
 Book Co., Inc., New York, 1955, 776 pp. 
 
 40. Lynd, L. E. Zirconium and Hafni- 
 um. Ch. in Mineral Facts and Problems. 
 BuMines B 671, 1980, pp. 1045-1060. 
 
 41. Martinez, G. M. , and D. E. Couch. 
 Electrowinning of Zirconium From Zirconi- 
 um Tetrachloride. Metallurgical Trans. , 
 V. 3, No. 2, 1972, pp. 571-574. 
 
 42. Mauser, J. E. Hafnium-Zirconium 
 Separation by Selective Reduction. Bu- 
 Mines OFR 11-68, August 1968, 23 pp.; 
 NTIS PB 180 357. 
 
 43. Megy, J. A. (assigned to Teledyne 
 Industries, Inc., Los Angeles, CA) . 
 Method of Separating Hafnium From Zirco- 
 nium. U.S. Pat. 4,072,506, Feb. 7, 1978. 
 
 44. Megy, J. A., and J. Freund. The 
 Separation of Zirconium and Hafnium in a 
 Molten Salt-Molten Zinc System. Metall. 
 Trans., v. lOB, 1979, pp. 413-421. 
 
 45. Morozov, I. S. , and I. Sun. Ther- 
 mal Stability of the Compounds of Zirco- 
 nium and Hafnium Chlorides With the Al- 
 kali Metal Chlorides. Russ. J. Inorg. 
 Chem., V. 4, Nov. 11, 1959, pp. 1176- 
 1179. 
 
 46. Newnham, I.E. (assigned to Common- 
 wealth Scientific and Industrial Research 
 Organization, East Melbourne, Australia). 
 Process for Removal of Hafnium From 
 Zirconium-Containing Material. U.S. Pat. 
 2,791,485, May 7, 1957. 
 
 47. 
 
 (assigned to Mallory-Sharon 
 
 Metals Corp., a Corporation of Delaware). 
 Method of Purifying Zirconium Tetraha- 
 lide. U.S. Pat. 2,916,350, Dec. 8 1959. 
 
 48. 
 
 (assigned to Mallory-Sharon 
 
 Metals Corp., a Corporation of Delaware). 
 Purification of Zirconium Tetrahalides. 
 U.S. Pat. 2,953,433, Sept. 20, 1960. 
 
 49. 
 
 (assigned to Mallory-Sharon 
 
 Metals Corp., a Corporation of Delaware). 
 Purification of Hafnium Tetrahalide. 
 U.S. Pat. 2,961,293, Nov. 22, 1960. 
 
 50. Pankratz , L. , Private communica- 
 tion, 1982; available for consultation at 
 BuMines, Albany, OR. 
 
 51. Perry, R. H. Chemical Engineers 
 Handbook. McGraw-Hill Book Co., Inc., 
 New York, 4th ed. , 1963, p. 13-28. 
 
 52. Pint, P., and S. N. Flengas. 
 Pressure Effect on the Kinetics of Reac- 
 tion Between Solid NaCl and ZrCl4 or 
 HfCl4 Vapors. Can. J. Chem., v. 49, 
 1971, pp. 2885-2890. 
 
25 
 
 53. Plucknett, W. K. Separation of 
 Zirconium and Hafnium Halides. U.S. Pat. 
 2,816,814, Dec. 17, 1957. 
 
 54. Plucknett, W. K. , R. S. Hansen, 
 and F. R. Duke. Progress Report in Phys- 
 ical and Inorganic Chemistry. Groups II, 
 III, and IV. Iowa State College ISC-51, 
 Apr. 30, 1949, 14 pp. 
 
 55. Ross, D. K. Apparatus for Refin- 
 ing Metal Chlorides in Molten Salts. 
 U.S. Pat. 3,114,611, Dec. 17, 1963. 
 
 56. Shelton, R. A. J. The Lower Chlo- 
 rides of Zirconium. J. Less-Common Met- 
 als, V. 61, 1978, pp. 51-62. 
 
 57. Spink, D. R. Fused Salt Scrubbing 
 of Zirconium Tetrachloride. Tran. AIME, 
 V. 224, 1962, pp. 965-970. 
 
 58. (assigned to Amax Specialty 
 
 Metal Corp., Greenwich, CT). Separation 
 of Zirconium and Hafnium. U.S. Pat. 
 3,966,458, June 29, 1976. 
 
 59. Spink, D. R. , and K. A. Jonasson. 
 Separation of HfCl4 and ZrCl4 By Frac- 
 tional Distillation. Paper in Extractive 
 Metallurgy of Refractory Metals, (ed. by 
 U. Y. Sohn, 0. N. Carlson, and J. T. 
 Smith Proc. Symp. sponsored by the TMS- 
 AIME Refractory Metals Committee and 
 the Physical Chemistry of Extractive 
 
 Metallurgy Committee at the 110th AIME 
 Annual Meeting, Chicago, IL, Feb. 22-26, 
 1981). American Institute of Mining, 
 Metallurgical, and Petroleum Engineers, 
 New York, 1981, pp. 297-314. 
 
 60. Spink, D. R. , and C. P. Vijayan. 
 Hafnium Electrowinning Studies. J. Elec- 
 trochem. Soc. , v. 121, No. 7, 1974, pp. 
 879-881. 
 
 61. Stephens, W. W. Extractive Metal- 
 lurgy of Zirconium - 1945 to the Present. 
 Ch. in Zirconium in the Nuclear Industry: 
 Sixth International Symposium (ASTM, Van- 
 couver, British Columbia, June 28-July 1, 
 1982). ASTM STP 824, ed. by D. G. Frank- 
 lin, R. B. Adamson, and B. Cox. (Ameri- 
 can Society for Testing and Materials) 
 Philadelphia, PA, 1984. 
 
 62. Tsirel'nikov, V. I., G. K. Seli- 
 vanov, and A. A. Malt'sev. (Thermodynam- 
 ic Functions of Titanium, Zirconium, and 
 Hafnium Tetrahalides in a Gaseous State.) 
 Zh. Fiz. Khim. , v. 47, No. 4, 1973, p. 
 1046; abs. in Chem. Abstr. , v. 79, 1973, 
 No. 35749J. 
 
 63. Tverskov, V. A., and I. S. Mor- 
 ozov. The System KCl-MgCl2-ZrCl4 . In 
 Phase Diagrams for Ceramists, 1969 Sup- 
 plement. American Ceramic Society, Inc., 
 Columbus, OH, 1969, p. 340. 
 
 INT. -BU. O F MINES, PGH., PA. 27 27 5 
 
4 
 
 1 
 
o • . 
 
 <l • . 
 
 
 
 
 
 
 
 • <£5Atv^'- o 
 
 
 ^^0^ 
 
 
 
 ^""^ 
 
 ' ^ 
 
 ^oV* 
 
 ^^d« 
 
 
 
 
 
 
 
 4 o 
 
 V**^'^*'^^'' ~V'^^'^'"'\^'^^^ "V^'^**-^'' 
 
 
 
 ^.^<b^ 
 
 
 
 
 
 
 5°^ 
 
 
 
 

 
 ,-lc) 
 
 ^0 v-^ 
 
 
 
 
 
 • «.^ 
 
 
 ^^•b^ 
 
 
 ^ <.^' "^^ -.1 
 
 :. -t 
 
 
 
 V^' 
 
 
 ^_- •'TVT* v*^ 
 
 .* 
 
 
 
 cT ^'l 
 
 ♦.^777» ^ 
 
 
 • ^<{-' 
 
 ^- ^^ 
 
 
 'V 
 
 .*^\ 
 
 ;.\ 
 
 
 
 -^^0^ 
 
 <<^ 
 
 ^* 
 
 ' v^ 
 
 *** 
 
 .«_/»^3, 
 
 >* »•••• **« 
 
 V 
 
 
 •*^ 
 
 
 
 
 .0' 
 
 
 
 '^^<.^ 
 
 ^ . -.v^^/ .<> 
 
 4 o 
 
 0^ c»/^«.,'^o. ^-^^ .."•, -^f. 
 
 
 
 '.. .^* .-y^^:, v.,^° --^v. ■^,_.*^* .-irm 
 
 :S 
 
 " • 
 
 =^1 
 
 ■V -'ifrt 
 
LIBRARY OF CONGRESS 
 
 002 955 811 5 
 
 '*■