2? >> .V 5x ■- vO V "oV' •^o^ ^^o^ . ^oV •» o ^"V %.^^*/ \*^'^\^^' "o^^^-/ \/^^\/ '^"^ «^^ "^. - '»bv* .•» O , .«- i^o' ^,^ ♦;'%•' ^^'«" •"-• »v .«-"^ • 'J'. -** oil"* ^x. ,vV ,-. V. ••• A^ IC 8952 Bureau of Mines Information Circular/1983 Rates of Chlorination of Aluminous Resources By N. A. Gokcen UNITED STATES DEPARTMENT OF THE INTERIOR Jnformation Circular 8952 Rates of Chlorination of Aluminous Resources By N. A. Gokcen UNITED STATES DEPARTMENT OF THE INTERIOR James G. Watt, Secretary BUREAU OF MINES Robert C. Horton, Director Library of Congress Cataloging in Publication Data: Gokcen, N. A Rates of chlorination of aluminous resources. (Information circular / Bureau of Mines ; 8952) Bibliography: p. 14. Supt. of Docs, no.: I 28.27:8952. 1. Aluminum oxide. 2. Chlorination. 3. Aluminum chloride. I. Title. II. Series: Information circular (United States. Bureau of Mines) ; 895 2. TN295.U4 [QD181.A4] 622s [546'. 6732] 83-600192 CONTENTS Page Abstract 1 Introduction 1 Raw materials 2 Calcination 2 Chlorlnatlon 3 Experimental procedure and results 4 Pure AI2O3 5 Bauxite 7 Clays 8 Effect of added NaCl 9 Effect of SICI4 10 Purification of AICI3 12 Conclusions 12 References 14 ILLUSTRATIONS 1. Residual water and porosity as specific surface area of A1(0H)3 after cal- cination at various temperatures 2 2. Chlorlnatlon of AI2O3 In a thermogravlmetrlc balance at various temperatures 5 3 . Chlorlnatlon of Y-AI2O3 In CO/CI2 = 1 7 4. Chlorlnatlon of bauxite and AI2O3 with CO + CI2 + Ar 8 5. Chlorlnatlon of clay with CO + CI2 + Ar and COCI2 + Ar 9 6. Typical elemental changes during chlorlnatlon of clays 9 7. Chlorlnatlon of NaCl-clay = 1/10 In CO + CI2 + Ar 10 8. Effect of NaCl addition on chlorlnatlon of clays at 650° C In CO + CI2 + Ar mixtures 11 TABLE 1. Compositions of typical bauxite and clays 2 RATES OF CHLORINATION OF ALUMINOUS RESOURCES By N, A, Gokcen ^ ABSTRACT This Bureau of Mines report reviews and summarizes recent studies of the rates of chlorination of aluminous resources with CO and CI 2 mixtures with and without COCI2. No reaction mechanism could be obtained from the existing results; however, diffusional barriers in the gas and solid layers probably control the chlorination rate in thermogravimetric experiments. Fluidized beds of particles smaller than 0.1 mm appeared to show very little solid layer diffusional barrier. For optimum chlorination the particle size should probably be less than 8 mm, the calcination temperature approximately 700° C, and the chlorination temperature from 650° to 750° C. Relatively rapid chlorination with COCI2 in themnogravimetric experiments was attributed partly to the simultaneous supply of reductant and chlorinator by one gas to the sample reaction site. Comparable results for chlorination of fluidized beds with COCI2 are not available. Equimolar mixtures of CO and CI 2 produced the optimum rate of reaction. Addition to the calcine of 10 to 20 wt-pct NaCl acceler- ated the rate of chlorination, and addition of SiCl^ to the gas mixture decreased the rate of chlorination of Si02 drastically, but at the expense of chlorination of sig- nificant fractions of AI2O,. Further research in various areas is suggested. INTRODUCTION Chlorination of domestic nonbauxitic resources is a thermodynamically possible process for the production of aluminum chloride (AICI3). However, thermodynamic data on gaseous metal chlorides are not sufficient to evaluate the purification of chlori- nation products. Further, reaction rate considerations are necessary to evaluate the feasibility of this process for producing AICI3 of purity suitable for electrowinning aluminum by direct electrolysis in a single-compartment cell (2, 5, 16, 18). 2 Since de Beauchamp's extensive review (2) on AICI3 preparation was published in 1969, sev- eral interesting rate studies have been made. This report reviews and summarizes recent investigations on rates of chlorination of AI2O3 and clays for the production of pure anhydrous AICI3.3 This critical review and summary is part of the Bureau of Mines effort to advance mineral technology and energy economy. ^Research supervisor, Albany Research Center, Bureau of Mines, Albany, OR. ^Underlined numbers in parentheses refer to items in the list of references at the end of this report. ■^Various patents in this field are not discussed to avoid possible controversies. RAW MATERIALS The raw materials used in the past, and discussed in the present report, are pure aluminum hydroxide [A1(0H)3], bauxite, and clays. Pure A1(0H)3 is obtained from pure aqueous solutions containing chemically pure aluminum salts and therefore con- tains virtually no impurities. The remaining materials contain various amounts of impurities, such as compounds of Fe, Si, and Ti, which are chlorinated simultaneously with the compounds of Al, Typical analyses of some aluminous resources are shown in table 1. TABLE 1. - Compositions of typical bauxite and clays on dry basis, weight-percent Clayl (6) Bauxite (13) Bauxite (6) Kaolinic clay' (7) AI2O3, Fe203, SiOo.. TiO. 285 5.9 6.5 1.7 64.6 8.7 22.0 2.7 44.1 .6 53.6 1.7 40 -45 .5- 3 49 -53 1 - 2 'McNamee No. 1 from Bath, NC; composition of another commercial kaolinic clay from Georgia was within the range of composition in this column (6-7). ^Approximate; by difference. CALCINATION The calcination of A1(0H)3 and naturally occurring aluminous re- sources is necessary to remove all the water and to alter the clay structure to make it more reactive. The reasons for removing water are that (1) CI2 and H2O react to form HCl, which is not as effective in chlorination as CI2, and (2) H2O and HCl form hydrated AICI3 with n moles of H2O as AlCl3*nH20, which decomposes in the electrolytic cell to form the undesirable products HCl and AI2O3. The amount of re- tained water in the hydroxide, bauxite, or clay decreases with in- creasing calcination temperature, as shown in figure 1. The por- osity, as expressed in square meters per gram of calcine, de- creases with increasing temperature (fig. 1). Therefore, the rate of chlorination at a given temperature decreases with increasing calcina- tion temperature, partly because of the accompanying decrease in por- osity. The optimum calcination 0.6 o E O X "o E on LlJ I- < < 9 in a: 290 o> - 270 cm: -250 - 230 -210 - 190 < LlI < LiJ O £ q: z> CO o o 1x1 a. en 350 450 550 650 750 TEMPERATURE, °C 850 70 FIGURE 1, - Residual water and porosity as specific surface area of AI(0H)3 after calcination at various temperatures. Sol- id line is for residual water; broken I ine is for porosity, in square meters per gram, (From Alder (_1_),) temperature is probably in the neighborhood of 700° C (1, 12), where the volume of retained water is slightly greater than that at higher temperatures. Calcination of pure AI2O3 at temperatures above 1,050° C converts this oxide into a-Al203, which is known to be very slow to chlorinate. Below 1,050° C, pure AI2O3 occurs in several different crystalline forms, including y~A1203 » all of which chlorinate considerably faster than a-Al203. In this report, the AI2O3 formed below 1,050° C is referred to as Y~Al2^3 > ^^ accord with various investigators. CHLORINATION The chlorination process with gaseous reactants occurs according to the follow- ing reactions: Al203(s) + 3C0(g) + 3Cl2(g) = Al2Cl5(g) + 3C02(g), (1) Fe203(s) + 3C0(g) + 3Cl2(g) = Fe2Cl5(g) + 3C02(g), (2) M02(s) + 2C0(g) + 2Cl2(g) = MCl4(g) + 2C02(g). (3) In the last reaction, M represents Si or Ti. Instead of gaseous mixtures of CO and CI2, gaseous phosgene (COCI2) may be used in these reactions; namely. Al203(s) + 3C0Cl2(g) = Al2Cl5(g) + 3C02(g) (4) Chlorides of Al and Fe occur as monomeric and dimeric gaseous species, as discussed below. It is essential that the gas mixture contain a reduetant, such as CO, 4 and a chlorinator, such as CI2 ; in the case of COCI2, these are contained in the same mole- cule. Thermodynamic calculations based on recent compilations (3, 10) show that at equilibrium all the reactants in reactions 1 to 4 should be consumed at the usual chlorination temperatures below 1,000° C. It is useful to discuss the properties of the remaining compounds in these reactions, as well as the compounds NaCl, KCl, and NaAlCl^ , which are discussed in conjunction with the catalysts for chlorination in the section "Effect of Added NaCl." AICI3 (s,£) : The vapor pressure of solid AlCl3(s) is 1 atm at 169.7° C, and the during chlorination at considerably higher gas phase is nearly all Al2Clg(g); but temperatures, the gas phase consists of point of AlCl3(s) 4.57 atm. mixtures of AICI3 and Al2Cl( is 192.55 C, and at this temperature, the vapor The melting pressure is AlCl3(g); Al2Clg(g): These gaseous chlorides coexist in equilibrium so that their volume percentages at various temperatures and at 1 atm are as follows: Temperature. . . .°C. . 326.85 526.85 726.85 926.85 AICI3 pet.. Al2Clg pet. . 2.1 97.9 35.5 64.5 88.4 11.6 98.7 1.3 FeCl3(s,j;,) : The melting point of FeCl3(s) is 304° C; the boiling point, at which it virtually all becomes the dimeric chloride, Fe2Clg(g), is 332° C. FeCl3(g); Fe2Clg(g): These chlorides coexist in equilibrium so that their vol- ume percentages at 1 atm are as follows: Temperature °C. . 526.9 726.9 FeCl3(g) pet.. Fe2Cl^(g) pet.. 9.3 90.7 53.1 46.9 ^For brevity, gaseous compounds in reactions 1 to 4 hereafter will not always be denoted by (g) . SiCl^(jl,g): Tetrachlorosilane boils at 57.0° C. T±Cll^is,Siyg): Titanium tetrachloride melts at -24.1° C and boils at 136.9° C. COCloCg^: The reaction of CO with CI2 over activated charcoal at temperatures below 500 C generates this compound; thus CO + CI2 = COCI2. (5) At a total pressure of 1 atm, the following gases in volume percent are in equilibrium: Temperature °C. . CO pet.. CI2 pet. . COCI2 pet . . 726.85 48.16 48.16 3.68 In the usual temperature range of chlorination, an equimolar mixture of CO and CI2 may contain small amounts of COCI2. However, when pure COCI2 is used, the decomposi- tion into an equilibrium gas mixture might be inhibited in the absence of catalysts. NaCl(s,£) ; Sodium chloride melts at 800.7° C, and its vapor pressure is en- tirely negligible at chlorination temperatures since its boiling point is above 1,500° C. KCl ( s , £ ) ; Potassium chloride behaves like NaCl; namely, it melts at 770.9° C and exhibits negligible vapor pressure at chlorination temperatures. NaAlClt^(£ ,g) : This complex chloride is a part of the binary phase diagram NaCl- AlCl3^ The total vapor pressure over the liquid is 0.129 atm at 650° C and 1 atm at 749.5° C, according to Linga (9^). KAlCli^ (jl ,g) : The vapor pressure of this liquid is roughly one-fourth the vapor pressure of NaAlCl^(jl), on the basis of limited data (9). It should be noted that any equilibrium data at two temperatures in the fore- going results may be extrapolated or interpolated by first writing an equilibrium constant Kp , where the subscript p indicates that Kp is a function of pressure, then obtaining an equation linear in 1/T, where T is the temperature in kelvins, so that Jin Kp = (A/T) + B, where A and B are constants related to the standard enthalpy and entropy changes. EXPERIMENTAL PROCEDURE AND RESULTS (6) The experimental procedures for accomplishing gas-solid contact and removing the gaseous products containing the chlorides of aluminum and impurity elements play very important roles in the rate of chlorination. It is evident that in a perfect gas- solid contact, nearly achieved in a vertical reaction chamber in which solid par- ticles fall against rapidly rising gases, the rate of reaction is predominantly con- trolled by the diffusion of reaction products from the pores to the gas stream. When the sample is in a crucible inside a vertical tubular furnace, the gas-solid contact is not intimate, and the rate of reaction is controlled significantly by the gas-phase diffusion barrier. It must therefore be emphasized strongly that the rate of chlorination obtained in the laboratory must be applied with great caution to a pilot-plant or industrial-plant operation. Only certain results, based on compara- tive experiments, can lead to certain conclusions with confidence (see Conclusions). The results of investigations on different resources are sufficiently different to justify the discussion in three sections; namely, under pure AI2O3, bauxite, and clays. Pure AI2O3 280 o — - -^(~; Numbers indicate °C Pure AI2O3 used for this type of investigation was obtained by calcining chem- ically pure A1(0H)3 to form AI2O3. Alder (_1_)5 shows (fig. 1) the residual water in 150 mg of initial charge of A1(0H)3 maintained at each temperature for 30 min to obtain AI2O3 , The average particle size of the resulting calcine was 0.088 mm. The specific internal surface area is also shown in figure 1. In Alder's experiments (l) , approximately 150 mg of AI2O3 was placed in a crucible resting on an arm of a thermogravimetric balance. Approximately 2 i/hv of an equimolar mixture of CO and CI2 were passed downward over the crucible containing AI2O3. Each sample, dehydrated in nitrogen at a certain temperature, was subsequently chlorinated at the s£une tem- perature. The progress of chlorination was recorded by weight loss because the prod- ucts of chlorination were gaseous compounds. The results for their particu- lar sample geometry and the corres- ponding gas-solid contact indicated that the chlorination was 52 pet complete at 400° C and 87 pet com- plete at 600° C after 40 min (fig. 2). The results were nearly iden- tical at 600°, 700°, and 800° C. The percentage of conversion (pet conv) appears to follow (pet conv) = (constant x /time, which suggests qualitatively that the combined solid- and gas-phase dif- fusion had probably controlled the rate of reaction in Alder and Mul- ler's experiments (J^, 14-15) , though they presented no conclusive explanations. The percentage of conversion for pure y-Al'^O^ — dehy- drated at 1,000° C, weighing ap- proximately 0.2 g, and chlorinated at 500° C under 0.4 atm CO, 0.4 atm < LU < LU < cc 3 (/) o o UJ CO CI 2» and 0.2 atm Ar — followed a nearly identical pattern in Lands- berg's experiments (6), as dis- cussed in the section on "Bauxite." Additional experiments with various molar ratios of CO to Clo •-*Data from this investigation were used in two similar reports ( 1 4-1 5 ) , together with various com- ments and interpretations. 500 :o. — 400 _ s ^--500 ^400 100 C3 < H- 80 Z Q. to OJ 60 < Q UJ 1- < 40? on _l X 20 30 60 REACTION TIME, min FIGURE 2. - Chlorination of AI2O3 in a thermogravimetric bal- ance at various temperatures. Solid lines represent percent of chlorination of initial AI2O3; broken lines represent porosity as specific surface area, in square meters per gram. (From Alder (1).) showed that the rate of chlorination increased rapidly with increasing values of CO/ CI2 up to CO/CI2 = 1 but decreased as the CO/CI2 ratio increased further. This is understandable from reactions 1 to 3, wherein equimolar amounts of CO and CI2 are required. Alder and Muller (1, 14) also caried out experiments with a f luidized-bed reac- tor (75 mm ID) with 250 g of 0.06-mm AI2O3 particles and approximately 65 mmol/min of an equimolar CO/CI2 mixture. The bed height was 66 mm. The extent of chlorina- tion was measured by the outlet gas composition since the concentrations of CO2 or CI 2 in the outlet gas for a constant gas inlet flow rate can determine the extent of reaction. The results showed that (1) dehydration at 600° C and the subsequent chlorination at 600° C followed a linear behavior when the weight-percent of AI2O3 chlorinated was plotted versus the time, unlike the experiments in a thermogravimet- ric balance, (2) the results at 700° C were nearly identical to those at 600° C, (3) utilization of CO was nearly constant and roughly equal to 83 pet for chlorination at 600° C, and (4) the corresponding results for a sample dehydrated at 600° C and chlorinated at 400° C showed a somewhat lower rate of chlorination, with CO utiliza- tion decreasing from 80 pet at 20 min to 60 pet at 180 min. The outlet gas con- tained 9 mol pet COCI2 at 400°, but only 1.4 mol pet at 600° C. Doubling the AI2O3 height from 66 to 132 mm in the fluidized bed increased the CO utilization from 83 to 87 pet at 600° C. Varying the bed height from 33 to 132 mm and the gas flow rate by a factor of four did not indicate reaction breakthrough at 600° C for full utilization of the reaetants. When the gas flow rate was increased from 50 to 270 mmol/min, the CO utilization decreased from 83 to 60 pet for 66 mm of bed height at 600° C. Similar behavior was observed for the bed heights of 33 and 132 mm, also at 600° C. At the maximumm rates of gas flow in these experiments, the carryout velocities of AI2O3 particles were attained. Another similar investigation was carried out by Milne (12) , who used pure AI2O3 for chlorination. Samples calcined at 750° C and weighing 1 to 10 g were each sus- pended from a thermogravimetric balance. Samples smaller in particle size than 2.5 mm were placed in a crucible (its dimensions were not given) , but those with particle size greater than 2.5 mm were placed in a silica (Si02) fiber basket. The extent of chlorination with equimolar mixtures of CO + CI2 is shown for various particle sizes in figure 3. At 700° C, essentially complete chlorination required 50 min for 0.125- mm particles (fig. 3A) and 60 min for 7.9-mm particles (fig. 3B^) . Overall results showed similar time requirements for particles 7.9 mm or smaller at all chlorination temperatures, but for particles larger than 7.9 mm, the time required for completion of chlorination differed significantly at higher temperatures. These differences show that for a given set of experiments, the initial bauxite particle size should probably be less than 8 mm. Further, figure 3A shows that the chlorination at 420° C is considerably faster with CO + CI2 + COCI2 than with CO + CI2 alone. It is interesting to note that the silica containers used by Milne (12) , as well as those used by Landsberg (6-7), do not chlorinate, as indicated by blank runs, and yet the Si02 in clays does chlorinate at a significant rate. The types of curves obtained from tests in crucibles have very little practi- cal or theoretical significance because of the complexities involving poros- ity, diffusivity, adsorption, and desorption. Adsorption and desorption probably play lesser roles above 500° C. Milne ( 12 ) shows that, at 360° to 475° C, CI2 is adsorbed first, followed by CO. (See the initial dip below zero in the curve for 420° C in figure 3A and the curve for 430° C in figure 3B.) The desorption .00 1 \ ^ __l£20° C 700°, 800° 4^' , 900° C^ ^-^■^'^600° C 2 'y^\^0° C O .75 ' f/ / y''oAO° C < ^Z 01 o / ^^1,000° c _J f ^^ X 1 / X y'^^ o .50 — J / / ^/^ ^^ — -J < I / / y ^ 420° C ^^^ C0Cl2^^^^20° C z X ^,y^ o \ / / /^ y / ^/^ 1- \ 1 / / / ^>^ o < .25 - \l / / / y^ - q: I//X IL. I A - 1 .00 1 1 _,^=r>700° C- ^^y^"^^ ^l^^-'-^^y 575° c^s^;^:^^^^^ \ yy y^:^'^ ^820° C^ >^ x^:^;^x^«^545°C .75 f /y/^ ^y^ }920° C / /V^ y^ '^ / // y//^ ^--920° C .50 //^ yy'^^^^^}430° c - / ^/ /'x / .^^^ // Ay y^ // // ^ y^ .25 III // /y W/ U B - 1 1 25 50 TIME, min 75 25 50 TIME, min 75 FIGURE 3, = Chiorination of y-Al203 in C0/Ci2 = It A, Solid lines are for 9,5'=mm particles; broken lines are for 0,125=mm particles; one curve for COCI2 is also indicated, B, Solid lines are for 7.9=mm particles; broken lines are for 3.2=mm particles. Samples were calcined at 750° C, (From Milne (12),) of AlCl3is probably slower and conceivably requires dimerization of AICI3 on the sur- face and then vaporization as gaseous Al2Clg ( 12 ) . Alumina transforms into the alpha form upon heating above 1,050° C. This yields a denser, less porous, and less reactive material. Hille and Durrwachter (4_) ob- tained only half as much AICI3 from a-Al203 as they did from Y-AI2O3 under the same experimental conditions, the same temperature range (700°-900° C), and the same time period (8 hr) for chiorination of 200 g of charge in a vertical furnace. Bauxite A pisolitic bauxite having the approximate composition shown in table 1 was used by Milne (12) . The reaction rates were slightly slower than those for pure AI2O3 be- cause of the smaller surface area per gram for bauxite. The chiorination rates of pure AI2O3 and bauxite were nearly identical in his experiments. For AI2O3 and baux- ite, previously dehydrated at 700° C, nearly complete chiorination with CO + CI2 was possible in about 1 hr at 600° to 800° C. A bauxite of composition shown in table 1 was used by Landsberg {b) for chior- ination with 0.4 atm CO, 0.4 CI2 , and 0.2 atm Ar. A 0.2-g sample of previously cal- cined bauxite was placed in a fused silica boat, 20 mm diam by 10 mm high, suspended from a recording balance with a fused silica fiber (8). From Landsberg' s results, shown in figure 4, it is evident that the amount of chlorinated sample, as shown by samples A to D, increased with increasing temperature up to 600° C. Blank runs with- out calcined bauxite showed that the fused silica components of the sample support assembly did not chlorinate. The amount of chiorination was not greatly different . 60- < CO lOOr^c — ^i 1 1 1 i 1 for the experiments at 500° , 600° , and 800° C as shown by C, D, and E. Sample F, calcined at 1,200° to form a-Al203, chlorinated much more slowly at 800° C for up to 150 min, as can be seen by comparing it with sample E, which was calcined at 1,000° C and chlorinated also at 800° C. The chlorination could have been faster had the bauxite been calcined at lower temperatures, as can be seen by comparing figure 3 with figure 4. However, part of the difference might be due to the lower Si02 contents of the bauxite used by Milne (12). Clays The rates of chlorination of bauxite and clays differ because of the differences in concentrations of impurities. The composition of the McNamee No. 1 clay used by Lands berg (6) is shown in table 1. The aluminum silicate, also used by Landsberg, was reagent-grade ^28120^ containing 0.3 pet Fe as the only significant impurity. The clay and the silicate samples were calcined at 1,000° C to remove all the volatile constituents. Preliminary experi- ments showed that both materials chlorinated similarly in mixtures of CO and CI 2. Figure 5 shows weight loss versus time for McNamee No. 1 clay. The extent of chlori- nation was considerably different between 600° and 750° C but not between 750° and 900° C. Chlorination with COCI2 was markedly faster at 700° C. It would be inter- esting to observe in future experiments if this is also true for fluidized beds. Typical changes in concentrations of elements during chlorination of clays are shown in figure 6. Experiments with various CO/CI2 ratios showed that, again, CO/CI2 = 1 accomplished the fastest chlorination rate for clays. Both figures 5 and 6 show that the chlorination of clays requires very long time periods. 100 150 TIME, min 250 FIGURE 4, - Chlorination of bauxite and AI2O3 with C0 + Ci2 + Ar, (From Landsberg (6)t) In another series of experiments, Landsberg (6) investigated the effect of cal- cination temperature on the extent of chlorination. For this purpose he calcined the kaolinic clays listed in table 1 at 750°, 1,000°, and 1,200° C, and the resulting calcines were each chlorinated at 750° C in 0.4 atm CO, 0.4 atm CI2, and 0.2 atm Ar. The extent of chlorination averaged 22 pet higher for samples calcined at 750° C than for those calcined at 1,200° C. This result is in qualitative agreement with those for pure AI2O3 and for bauxite, because the porosity decreases with increasing cal- cination temperature (1, 6). The rates of chlorination with carbon are not discussed here because the process is slow unless O2 or CO2 is admitted to convert a part of the carbon into CO for ac- celerating contact between the reductant and the calcine (7). For a possible indus- trial process, similar to the iron blast furnace practice, it might be feasible to use coke as both reductant and fuel in the charge, and feed O2 and CI 2 as reacting gas mixtures. 100 o Q. o UJ cr >- < o 4 6 8 TIME, h FIGURE 5, - Chlorination of clay with CO + CI 2 + Ar and COCI 2 + Ar. (From Landsberg (6).) 100 Chlorinated at 700° C 0.4 atm CO and 0.4 otm CIg Effect of Added NaCl Sodium chloride reacts with AI2O3 to form NaAlCl^, which has a catalytic effect in chlorinating aluminum-bearing calcines , as ob- served by Seferovich (17). Hille and Durrwachter (4^) added large amounts of NaCl to AI2O3 and chlor- inated their charge with CO + CI 2 + COCI 2. They were successful in condensing NfiAlCl^ and refluxing it into their calcine charge. In their experiments, with 120 g of NaAlCl4 and 200 g of AI2O3 for 8 hr, the ultimate yield of AICI3 in- creased for each run from 400° to 530° C, but remained constant from 530° to 900° C. The yield was roughly 60 pet chlorination of AI2O3 when the initial NaAlCl^/ AI2O3 ratio was 120/200. However, the yield was increased to 90 pet when NaAlCl4/Al203 = 220/200 at the start of a run. Landsberg ij) used various amounts of NaCl with calcined ka- olinic clays, listed in table 1, to 2 observe its catalytic effect. All samples were calcined at 700° to 750° C and ground to approximately 0.5 mm in particle size. Approxi- mately 0.2 g of calcine was placed in a VycorS bucket, 12 mm diam by 8 mm deep, that was suspended from an automatic recording balance in a _n vertical Vycor tube, of 25-mm diam. Approximately 500 ml/min of an equimolar mixture of CO and CI 2 containing 20 pet Ar was directed toward the open end of the bucket for chlorination. Figure 7 shows that, for NaCl/calcine = 1/10, the extent of chlorination was slow at 500° C but relatively rapid at 600° and 650° C, and considerably slower at 700° C than at 600° C. The ini- tial increase in weight during the first 30 min at 600° and 650° C was Ti 12 3 4 TIME, h FIGURE 6. = Typical elemental changes during chlorination of clays, (From Landsberg (6),) "Reference to specific manu- facturers/ brands of equipment, or trade names is made for iden- tification only and does not imply endorsement by the Bureau of Mines. 10 100 u Q. CD LJ cn iii _i Q_ < C/) 40 TIME, min FIGURE 7. - Chlorination of NaCI/clay= 1/10 in CO + CI 2 + Ar, The initial increase in weight for the samples chlo- rinated at 600°and 650°C was due to the formation of NaAICl4, (From Londsberg (7_)t) due to the formation of NaAlCl^, which has a much lower vapor pres- sure than condensed AICI3, Figure 8 shows that at 650° C, when the NaCl/calcine ratio was increased from 1/20 to 1/5, the weight loss was significantly greater. At NaCl/calcine = 1/2, the formation of NaAlCl^ increased significantly so that there was a much smaller weight loss after 40 min; after 80 min, though, the weight loss at NaCl/calcine = 1/2 far exceeded that at other NaCl/ calcine ratios (fig. 8). The amount of chlorination with NaCl was considerably greater than with- out NaCl (_7). For example, the chlorination at 600° C and after 1 hr was approximately 50 pet for 1/ 10 NaCl/calcine mixture, but only 5 pet without NaCl. Further, with 10 pet added NaCl, as much as 90 pet of AI2O3 can be chlorinated simul- taneously with only 5 pet of Si02. Chlorides such as LiCl, KCl, and MgCl2 enhance chlorination to roughly the same extent, and fluorides such as NaF, MgF2, and CaF2 to a lesser extent (4^, 7), Lands berg (7) placed a calcined clay in liquid NaAlCl^ at various temperatures, and then cooled, crushed, washed, and recalcined the product at 700° C. Best results during the subsequent chlorination tests were accomplished at 625° C after treatment in liquid NaAlCl^ at 550° C, even when no NaCl was added prior to chlorination. The complex halides such as gaseous NaAlCl^ and KAICI4 can be sepa- rated from AICI3 by fractional condensation or evaporation, but the process might be energy intensive. Effect of SiCl4 When SiCl^ is added in CO + CI 2 in sufficient amounts, it may react to form Si02 on the calcine. It is possible to compute the equilibrium concentrations of gaseous SiCl^ from the following reaction and its equilibrium constant K : 2Al203(s) + asici.Cg) = 4Aici,(g) + asiOoCs), k = p4 AlCl. P^ SiCl, (7) The values of Therefore, for P/\|ci Kp at 726.85° and 926.85° C are 16.6 and 2,138, respectively. . + Psjci =1 3tm, 0.74 mol of AICI3 and 0.26 mol of SiCl4 are 11 40 TIME, min FIGURE 8, - Effect of NaCI addition on chlorination of clays at 650° C in CO + CI2 + Ar mixtures. The initial increase in weight for some of the samples was due to the formation of NaAlCl4. (From Landsberg (7).) in equilibritun with the solid ox- ides at 726.85° C, whereas at 926.85° C, the corresponding quan- tities are 0.93 mol of AICI3 and 0.07 mol of SiCl4. Consider, for example, an inlet gas containing 6 pet SiCl., 47 pet CO, and 47 pet CI 2, and assume that 80 pet of the CO and CI 2 are used to chlorinate only AI2O3 in a calcine containing only Si02 and AI2O3 at 726.85° C. The outlet gas at 1 atm and 726.85° C would contain very closely a sufficient amount of SiCl^ to prevent the chlorination of Si02, if equilibrium in reaction 7 prevailed. If SiCl4 in the inlet gas were 4 pet, with 48 pet CO and 48 pet CI 2, a small amount of Si02 would be chlorinated under the same conditions as required by equi- librium in reaction 7. The effect of SiCl^ on chlori- nation of bauxite was recently investigated by Milne (11). He chlorinated bauxite, previously calcined at 750° C, with an equi- molar mixture of CO and CI 2, with and without SiCl^, Approximately 60 pet of the AI2O3 and 10 pet of the Si02 were chlorinated in 90 min at 720° C using a mixture contain- ing 0.2 mol SiCl4, 0.4 mol CO, and 0.4 mol CI 2; whereas without SiCl4, both AI2O3 and Si02 were chlori- nated 100 pet in 60 min. At 750° C, and with 0.10 mol SiCl4, 0.45 mol CO, and 0.45 mol CI 2» 80 SiOo were chlorinated in 125 min. pet of the AI2O3 and 15 pet of the The chlorination process virtually stopped after 90 min with 0.2 mol SiCl4 and after 120 min with 0.10 mol SiCl4. Therefore, SiCl4(g) in CO + CI2 mixtures decreased chlorination of Si02 significantly, but at the expense of chlorination of AI2O3, according to Milne. The chlorination time was considerably longer when SiCl4 was present. Landsberg (_7) found that calcined clay, previously exposed to SiCl4(g) at 800° C, did not chlorinate measurably at 625° C in 0.4 atm CO, 0.4 atm CI 2, and 0.2 atm Ar even when 10 pet NaCl was added to the calcine. Exposure to SiCl4 at tempera- tures below 800° C decreased the degree of inhibition for chlorination. Further experiments are needed to clarify the effect of SiCl4, particularly in f luidized-bed reactors. The chlorination of Si02 in clays consumes considerably large amounts of CI2. reuse. and the resulting SiCl4 requires special treatment for recovery of CI 2 for 12 PURIFICATION OF AICI3 The chlorination of aluminous resources containing significant amounts of Fe, Si, and Ti does not yield an AlCl-j condensate sufficiently pure for electrolysis to aluminum. The boiling point of SiCl^ is 57° C; therefore, separation of SiCl^ from Al2Clg should not present any difficulty, since the sublimation point of AICI3 is 169.7° C. Likewise, the boiling point of TiCl^ is 136.9° C, so it should be possible to distill out TiCl^. However, the boiling point of FeCl3(£) is 332° C, and this compound usually appears as an impurity in the AICI3 condensate. It is not certain whether Fe2Clg(g) and Al2Clg(g) form a complex gaseous species such as AlFeClg, which could contaminate AlCljCs). Fractional distillation of AICI3 appears to be attractive, but further research must be carried out in this area to ascertain the problems involved in formation of complex halides. CONCLUSIONS The rate studies using a thermogravimetric balance, with samples contained in crucibles, provide only qualitative answers to certain questions because the gas- condensed phase (CP) contact is not as intimate as in the fluidized beds. Even with a silica basket, the gas flow rate is not sufficiently large to provide turbulent contact between the chlorinating gas and CP. Therefore, no kinetic mechanism can be derived from thermogravimetric data. However, qualitatively, parabolic behavior of the fraction of chlorinated calcine versus time might indicate that the diffusion in the gas layer and in the CP layer (including the pores in the CP) probably controls the rate of chlorination. This conclusion is in agreement with Muller (15). In a fluidized bed, a constant rate of chlorination above 500° C indicates that the gas phase diffusion barrier is probably small, and the receding CP poses a small barrier to reaction in comparison with the thermogravimetric-type chlorination. For optimum chlorination rates, the particle size of the calcine should prefer- ably be less than 8 mm, and the calcination temperature should be 600° to 800° C, the optimum being roughly 700° C. Calcination should be at approximately the same tem- perature as chlorination if the latter is to be carried out at temperatures higher than 700° C, but both processes should be at temperatures sufficiently below 1,050° C, at which a-Al203 is formed. Chlorination with COCI2 is considerably faster than with equimolar CO + CI 2, particularly below 700° C, because both reduction and chlorination agents are sup- plied simultaneously to the CP by one gaseous compound in thermogravimetric experi- ments. Above 700° C, this may or may not be the case in a f luidized-bed reactor; clearly, additional experiments are necessary to resolve this point. A substantial degree of decomposition of COCI2 occurs above 600° C; therefore, it is doubtful that COCI2 could provide sufficient advantage over CO + CI2 in a fluidized bed, where more than 80 pet utilization of CO and CI2 is possible in bauxite chlorination. The rate of chlorination of all types of calcined aluminous resources with CO + CI 2 is at an optimum for equimolar gas mixtures. Since there are no difficulties in preparing equimolar mixtures, and because the overall chlorination reaction re- quires equimolar amounts of CO and CI2, no further research appears to be necessary to investigate this point. The optimum temperature for chlorination of aluminous resources with CO + CI 2 is in the range of 600° to 900° C, most likely within 650° to 750° C. Limited data by 13 Alder O) indicate that for a fluidized bed of AI2O3, 600° C might be quite satisfac- tory. Erosion and chlorination of refractories in industrial-scale chlorination chambers are considerably lower at lower chlorination temperatures; hence, chlorina- tion at as low a temperature as possible appears to be preferable. Sodium chloride is very effective in increasing the rate of chlorination of aluminous resources when it constitutes roughly 10 to 20 pet of the calcine charge. Sodium chloride forms liquid NaAlCl4, which vaporizes during chlorination, but it can be condensed and refluxed into the calcine if the chlorination temperature is below the atmospheric boiling point of NaAlCl4 (750° C). The optimum temperature with re- fluxed NaAlCl4 and the mechanism of chlorination with NaAlCl4 require further ex- tensive investigations. When SiCl4 is added to CO + CI2, chlorination of Si02 in bauxite decreases mark- edly. However, the rate of chlorination of AI2O3 decreases simultaneously, and the retained AI2O3 in the calcine increases significantly. Further research is necessary with clay and in fluidized beds containing NaCl to clarify the effect of SiCl4(g). At lower temperatures, namely 600° to 625° C, with NaCl and possibly with as low as 4 pet SiCl4 in CO + CO2 , the chlorination of Si02 in aluminous resources of all types might be significantly reduced. The simultaneous effects of SiCl4 and NaCl must be investigated in detail with various clays, particularly in fluidized beds, to deter- mine possible advantages of SiCl4 in the chlorination gases. The physical state of Si02 is important in chlorination, and this aspect re- quires further extensive research. It is a remarkable fact that porous Si02 in cal- cined clays and Si02 in silicates do chlorinate at various rates; however, vitreous Si02 remains virtually unchlorinated in the range of 500° to 900° C over several hours. Construction materials for commercial-size chlorination chambers must be kept cool enough to avoid being chlorinated themselves. Therefore, externally heated chambers do not appear to be practical. An appropriate set of reactions must be selected so that the reactions could generate a sufficient amount of heat for main- taining a high internal reactor temperature with a temperature gradient sufficient to allow a relatively cool and virtually nonreacting wall. 14 REFERENCES 1. Alder, H,-P. , H.-P. Muller, and W, Richarz. Kinetic Study of the Alumina Chlorination. Light Metals, v. 1, 1977, pp. 219-232. 2. De Beauchamp , R. L. Preparation of Anhydrous Aluminum Chloride. BuMines IC 8412, 1969, 19 pp. 3. Dow Chemical Co. , Thermal Research Laboratory. JANAF Thermo chemical Ta- bles. NSRDS-NBS 37, 2d ed. , 1971, 1144 pp., plus supplements. 4. Hille, J., and W. Durrwachter. (Production of Anhydrous Aluminum Chloride from Y~alumina in a Fluidized Bed.) Angew. Chem. , v. 72, 1960, pp. 850-855. 5. Kirby, D. E. , E. L. Singleton, and T. A. Sullivan. Electrowinning Aluminum From Aluminum Chloride, Operation of a Single-Compartment Cell. BuMines RI 7353, 1970, 24 pp. 6. Landsberg, A. Chlorination Kinetics of Aluminum Bearing Minerals. Metall. Trans. B, v. 6B, 1975, pp. 207-214. 7. . Some Factors Affecting Chlorination of Kaolinic Clay, Metall. Trans. B, v. 8B, 1977, pp. 435-441. 8. Landsberg, A., C. L. Hoatson, and F. E. Block. The Chlorination Kinetics of Zirconium Dioxide in the Presence of Carbon and Carbon Monoxide Metall. Trans., V. 3, 1972, pp. 517-523. 9. Linga, H. , K. Motzfeldt, and H. A, Dye, Vapor Pressure of Aluminum Chlor- ide Containing Melts of Interest to Aluminum Electrolysis, TMS-AIME, Pap, Sel, , A 77-36, 1977, pp, 205-217, 10, Mah, A, D, Chemical Equilibria in Chlorination of Clay, BuMines RI 8696, 1982, 43 pp, 11, Milne, D. J. Chlorination of Bauxite in the Presence of Silicon Tetra- chloride. Metall, Trans, B, v. 6B, 1975, pp, 486-487, 12, , The Chlorination of Alumina and Bauxite With Chlorine and Carbon Monoxide, Proc, Australas, Inst, Min, and Metall,, No, 260, 1976, pp, 23-31, 13, Milne, D, J,, and L, J, Wibberley, The Removal of Iron From Bauxite Using Anhydrous Hydrogen Chloride, TMS-AIME, Pap, Sel,, A 77-68, 1977, pp, 125-145, 14, Muller, H,-P, , H,-P, Alder, A, Baiker, and W, Richarz, (Use of Flu- idized Bed Reactor for Chlorination of Alumina,) Chem, Ing,-Tech. , v, 51, 1979, pp, 124-127, 15, Muller, H,-P, , A, Baiker, and W, Richarz, (Thermogravimetric Investigation of Reduction and Chlorination of Alumina,) Helv, Chim, Acta, v, 62, 1979, pp. 76-85 (Engl, abstr.), 16, Russell, A, S., L, L, Knapp, and W. E. Haupin. Production of Aluminum, U,S, Pat, 3,725,222, Apr, 3, 1973, 17, Seferovich, Y. E, Preparation of Anhydrous Aluminum Chloride by Chlorina- tion of Kaolin in Presence of a Catalyst, Zh, Khim, Prom-sti, (J. Chem, Ind,), No. 10, 1934, pp, 62-64, 18, Singleton, E. L. , D. E. Kirby, and T. A. Sullivan, Electrowinning Aluminum From Aluminum Chloride, Operation of a Two-Compartment Cell. BuMines RI 7212, 1968, 15 pp, INT.-BU.OF MINES,PGH.,P A. 27135 s6 "^^^ 3.0 V\ V^-\** -v'^^%°' \-^^V V^ V \-^^*y 1 • -0,^ • .V^ "O •'VVT* ,0 5-^°'*> -.^Si^^/ J^^-n*.. '.''^^m: vi^-* •.cm^>c- .c .^ .,. 0' V-'*'^^^'... '^./-■^ i°-n^. V <► * TVi* ^6^ ^o. v^ ^ZZtL'* 'c> ^•" "W* .*»• \-/ :'^i'. \<^' .*». \/ : '*-..^* y% o V •^^0^ *' ^^^% USRARVOFCONGRES? lllJIIIlBiimmiiiimvv ° 002 959 878 ■^