A^ *J 4 O 6. ^^.Q* 4 o O > :. '-^.0^ o, .^ Q V • ^^ ^* »*^ ■^.s^' lV>. /^♦'^^ %^ 4 o •->' V -.1/ •- C/ K_vV ''^.<, - ^% '' • <^" >* o. i^-v ^^ .».- '-^ V*- .^^^ %. .c°y>;^'> X-^;.%V ./.•;^:^^'^-o -^^' -- v* .. *u •" '$y ^ •• r ... ■■■■\-^ .. %-^-'/ \^-f^^-y "%--!^v*' v----^- .- *-.. Aluminum From Domestic Clay Via a Chloride Process The State-of-the-Art By A. Landsberg UNITED STATES DEPARTMENT OF THE INTERIOR Information Circular 8923 Aluminum From Domestic Clay Via a Chloride Process The State-of-the-Art By A. Landsberg UNITED STATES DEPARTMENT OF THE INTERIOR James G. Watt, Secretary BUREAU OF MINES Robert C. Norton, Director Mm MO. 8933 «!P m ^^^ This publication has been cataloged as follows: Landsberg, Arne Aluminum from domestic clay via a chloride process. (Information circular / United States Department of the Interior, Bu- reau of Mines ; 8923) Bibliography: p. 12-15. Supt. of Docs, no.: I 28.27:8923. 1. Aluminum— Metallurgy. 2. K^aolinite. 3. Chlorination. I. Title. II. Series: Information circular (United States. Bureau of Mines) ; 89 23. -fN^aS^W-^ {TN775] 622s [669'. 722] 82-600392 CONTENTS Page Abstract 1 Introduction 2 Background 2 Technical requirements 3 Aluminum chloride production 4 Chlorlnatlon of domestic clay 5 Purification of anhydrous aluminum chloride 10 Conclusions 12 References 12 ILLUSTRATIONS 1 . Processing routes to aluminum 5 2 . Generalized process flowsheet 5 3 . Pertinent vapor pressures 7 TABLES 1 . Pertinent compositions 3 2. Reactions pertinent to clay chlorlnatlon 6 3. Quantities of materials per pound of aluminum 6 4. Reactions pertinent to aluminum chloride purification 10 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT atm atmosphere kcal kilocalorie Btu British thermal unit lb pound ° C degree Celsius pet percent K Kelvin yr year ALUMINUM FROM DOMESTIC CLAY VIA A CHLORIDE PROCESS The State-of-the-Art By A. Landsberg ^ ABSTRACT Kaolinitic clays are potentially a vast domestic resource for alumi- num. Utilization of this resource could decrease or eliminate the near- ly complete dependence of the United States on foreign raw materials for this important metal. Furthermore, processing of clay to aluminum through anhydrous chloride metallurgy could reduce the high electrical energy requirements of the conventional Hall-Heroult aluminum reduction process. Several anhydrous chloride processes have been proposed; how- ever, unresolved technical problems have prevented their coimnercializa- tion. In particular, an acceptable chemical means has not been found to extract aluminum from clay as a highly pure anhydrous aluminum chloride. This Bureau of Mines report identifies and discusses the important chem- ical problems involved in achieving an acceptably rapid, self -heating, selective chlorination reaction and the subsequent separation of iron chloride byproduct from the anhydrous aluminum chloride. — — 'Chemical engineer, Albany Research Center, Bureau of Mines, Albany, OR. INTRODUCTION The worldwide use of aluminum exceeds that of all other metals except iron. Fortunately, aluminum is one of the most abundant elements in the earth's crust and bauxite, the principal ore of alumi- num, is plentiful and accessible to sur- face mining. However, the prime bauxite deposits are situated in tropical areas so that the largest producer and consumer of aluminum, the United States, must import over 90 pet of its aluminum raw material (24). 2 The present commercial production of aluminum is dependent upon two processes: the Bayer process for producing pure alu- mina from bauxite and the Hall-Heroult process for electrolytically reducing the alumina to aluminum metal. These two processes are very energy intensive, re- quiring over 244 million Btu to produce 1 ton of aluminum. In comparison, only 24 million Btu are needed to produce 1 ton of steel (_3 ) . Furthermore, over two-thirds of the energy used in the alu- minum industry is consumed as expensive electrical energy in the Hall-Heroult reduction process. Research and develop- ment efforts have increased the effi- ciency of the Hall-Heroult process since its discovery nearly 100 yr ago to the point that further major improvements are unlikely. Producing aluminum from domestic re- sources with less electrical energy has been the goal of several proposed alumi- num processes ( 15 , 37). While a few of these schemes have been developed through the pilot plant scale, technical and economic realities have precluded adop- tion of new aluminum technologies. Nevertheless, the search continues. The success of anhydrous chloride metallurgy in the commercial-scale production of the reactive metals titanium and zirco- nium has provided motivation to develop an analogous chloride metallurgy for aluminum. Recent operation of pilot- plant-scale aluminum chloride reduction cells has revived interest in the chlo- ride process for aluminum (42). However, as is often the case with new and promising technology, factual information has often been strongly ex- ceeded by publicity and speculation, fre- quently colored by a lack of technical understanding, and sometimes completely neglected in an overpowering urge to secure the benefits and rewards of a more efficient process. Therefore, it is timely that a review be made to de- scribe the chloride metallurgy of alumi- num, to evaluate recent developments, and to outline the problems that remain unresolved. BACKGROUND The technical literature abounds with suggested methods for making aluminum but the Hall-Heroult reduction process con- tinues to be the commercial method used and alumina to feed the reduction cells continues to be made by the Bayer process from bauxite (14) . There are good rea- sons for this. First, the raw material is found in large, easily mined deposits that have been concentrated and upgraded over geo- logic time by the action of water. ^Underlined numbers in parentheses re- fer to items in the list of references at the end of the report. The resulting bauxite contains alumina relatively free of impurities. Second, the established, large-scale Bayer proc- ess uses common reagents to refine this bauxite to cell-grade alumina. And fi- nally, the purified alumina is reduced to aluminum metal in well-engineered Hall- Heroult cells. This makes up today's aluminum industry: a large-scale, highly efficient technology which has enjoyed nearly a century of development through huge investments of human effort and financial resources ( 13 ) . In order to be a viable competitor any new alumi- num technology must be economically superior and an almost instant tech- nical success. Nonetheless, with current TABLE 1, - Pertinent compositions, percentages Kaolinite (AI2O3 '28102 •2H2O) Typical clay (calcined) Jamaican Georgia kaolin Used for calculations bauxite dried' AloO:. 39.5 46.5 NAp NAp 14.0 40 -45 49 -52 .5- 3 1 - 2 NAp 45 51 2 2 NAp 49 "~^2^3 ••••••••••••• Si02 2 Fe90T.. 20 TiOo 2.5 Loss on ignition. . 26 NAp Not applicable, 'Reference 49. emphasis on using domestic raw materials and with the escalating cost of energy, an increasing interest has been shown in new aluminum producing technology. Among potential domestic resources of aluminum, high-alumina clays appear to be the most favorable ( 36 , 39 , 50). In par- ticular, the kaolinitic clay deposits in the southeastern United States are at- tractive because of their size, accessi- bility, and quality. The richest depos- its, found in Georgia, are estimated to have resources in excess of 3 billion tons accessible to mining and process- ing (50). Such clay can be considered of high quality because of its high alu- mina content and because of its lack of complexing impurities. Composition of a typical high-quality Georgia clay is shown in table 1. For processing purposes this clay can be considered as a mixture of silica, alumina, titania, and iron oxide chem- ically separable but not separable by mechanical means. The silica in the quantities present can be considered a waste product, the titania a valuable by- product; and the iron a nuisance by- product. It is the qualitative differ- ences between clay and bauxite that give the competitive edge to the latter. If political and strategic considera- tions dictate the use of clay as a domes- tic aluminum source, it is conceivable to extract alumina suitable for reduction in existing primary aluminum plants (38). However, if this new resource is exploit- ed it is also conceivable to extract the aluminum in a form suitable for reduction in a new, more energy efficient process. Recent patent disclosures and successful large-scale operation of an aluminum chloride reduction cell offer such a process (9^, 17 , 42 , 44-45) . There now exists a potential domestic resource and a developing, energy-efficient technology for producing aluminum; lacking is a proven technology for converting the kao- linitic clay into anhydrous aluminum chloride suitable for electrolytic reduc- tion to metal. TECHNICAL REQUIREMENTS I The nature of kaolinitic clay and the requirements for the aluminum chloride product should be examined before consid- ering the chlorination process. Kao- linite, a hydrated aluminum silicate, consists of alternate single layers of AI2O3 and Si02 held together by the water of hydration. On a microscopic level, kaolinitic clay is seen to be made up of stacks of hexagonal sheets much like pages of a book (22). The major physical impurity is separate grains of quartz sand. The iron and titanium impurities appear to be within the kaolin structure because they cannot be completely removed without chemical destruction of the par- ent kaolin. The free and structural water associ- ated with clay cannot be tolerated in a chlorinator. Under the reaction condi- tions water and chlorine as a chloride combine to give the corresponding oxide and hydrogen chloride, a waste product that must be scrubbed from the exiting gases. For every one part of water near- ly four parts of chlorine are thus wasted as hydrogen chloride. Therefore, raw clay must be dried of free moisture and calcined to remove the water of hydra- tion. A calcination temperature of 700° C is required ( 12 , 19). At this and higher temperatures, and depending upon time, atmosphere, and the presence of other substances, kaolinitic clay can be- come quite refractory to subsequent chlo- rination (25). Therefore, this step in clay preparation can be critical to the entire chlorination process and should be carefully considered. The calcination step can also be used to fix the form of clay feed. The plas- ticity of wet clay and its strength after the removal of hydration water allow it to be made into any form from fine pow- ders to pellets of any size and shape. Prior to calcination, catalysts and reductants may be added to assure inti- mate contact in the final feed. But as a good thing can be overdone so can the property of clay to make strong fired shapes become a disadvantage. If the final feed form is made too dense there is a reduction in accessible surface area where chemical reaction occurs and the rate of chlorination becomes exceedingly slow. In general, kaolinitic clay can be considered an ideal raw material with respect to preparation as a reactor feed, but the preparation itself must be care- fully controlled so that the desirable characteristics are not lost before the feed enters a reactor. It is important that the anhydrous alu- minum chloride product of chlorination and subsequent purification processes be very pure (41) . The aluminum chloride will be fed as a vapor or solid into the high-temperature, molten-salt electro- lytic cell that discharges only re- cyclable chlorine gas and product liquid aluminum metal. All impurities entering with the aluminum chloride must exit with these two or build up in the cell, caus- ing premature shutdown ( 17 ) . Some impu- rities, notably oxides, would react with cell components causing deterioration of carbon electrodes. Of course, small quantities of impuri- ties such as iron, silicon, sulfur, phos- phorus, and other metals can be removed in association with the product aluminum but these must be kept below rigid speci- fications (O. Likewise some gaseous impurities and reaction products such as carbon monoxide, carbon dioxide, and sul- fur chlorides , can be removed in the chlorine stream. In addition, a system to circulate the molten salt electrolyte and filter out impurities prior to intro- duction into the cell, much like the oil system in an automobile engine, has been described in the patent literature (17) . Any such system would, of course, require more expensive equipment, add to the com- plexity of operation, and increase costs. Preparation of feed for electrolytic re- duction must keep the impurities to with- in a few hundredths of a percent (41) while processing aluminum chloride at tons per hour. ALUMINUM CHLORIDE PRODUCTION Several schemes for processing aluminum mineral resources to aluminum metal are shown in figure 1. At the far left is the traditional Bayer-Hall-Heroult com- bination; the double line branching from this indicates the path taken by the recent development and operation of a pilot production facility (43) ; and the dashed lines represent the possibilities of interest here. The latter two show a combination of chlorination and purifica- tion operations. Going from right to left across the upper part of figure 1, the degree of purification prior to chlorination in- creases. At the far left the alumina intermediate is of sufficient purity so that the chloride produced from it may be used directly in reduction cells. The center path represents removal of nui- sance impurities, such as iron, prior to chlorination. Attempts to remove iron from bauxite and clay indicate that this Mineral resource A Chemical processing \ \ \ Purification and chlorination \ \ \ Chlorination \ \ \ \ \ Chlorination and purification / / \ / Aluminum chloride Molten chloride electrolysis Aluminum metal FIGURE 1. - Processing routes to aluminum. does not now seem possible short of total chemical modification of the mate- rial such as in the Bayer process (7^, 20 , 32-33). Therefore, the following discussion will consider the path on the far right representing chlorination of domestic kaolinitic clay followed by purification of the anhydrous aluminum chloride product as outlined in the gen- eralized flowsheet of figure 2. CHLORINATION OF DOMESTIC CLAY All of the metal oxides associated with domestic kaolinitic clay chlorinate in the presence of carbon as noted by the thermodynamic data in table 2 (28). (A negative value for the Gibbs function indicates that a reaction is possible at the temperature indicated.) Similar- ly, all of the chlorides will react with oxygen to give the oxides and free chlorine. Thus it would seem to be a simple matter to completely chlori- rinate the clay, separate the chlo- rides by fractional vaporization accord- ing to the vapor pressures shown in Energy- Energy- — Catalyst ■Reductont- Chlorine- Clay i Drying and preparation Calcination Chlorination Residue -Water ■Water Chloride vapors and gases Reagents and/or - energy Energy Exit gases ,1 Condensation T Scrubbing or recovery Purification T Reagents and/or energy _^To waste, use, or recycle AICI3 product FIGURE 2. - Generalized process flov/sheet. TABLE 2. - Reactions pertinent to clay chlorination Reaction Enthalpy , Gibbs ^^9 00 K> function. kcal ^^9 00 K» kcal ^56.4 NAp -21.5 -66.1 40.0 -61.4 -83.0 -70.7 -67.4 -73.9 -89.8 -154.8 -71.2 -115.4 -51.8 -64.7 -60.8 -54.9 -74.6 -47.2 59.5 .9 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Al203'2Si02'2H20 ^ [AI2O3 + 2Si02] + 2H2O. 2Si02] 2Si02] 2Si02] 2Si02] 2Si02] 3Cl2 + 3Cl2 + 3Cl2 + 3C0C1 2 7Cl2 + 1.5C -»■ 2AICI3 + 2Si02 + I.5CO2. 3C ^ 2AICI3 + 2Si02 + 3C0 3C0 ->- 2AICI3 + 2Si02 + 3CO2 -»■ 2AICI3 + 2Si02 + 3C02 3.5C ->► 2AICI3 + 2SiCl4 + 3.5C02. [AI2O3 [AI2O3 [AI2O3 [AI2O3 [AI2O3 ^ Fe203 + 3Cl2 + 1.5C ->- 2FeCl3 + 1.5C02 Ti02 + 2CI2 + C -»- TiCli^ + CO2 SiCl^ + O2 ->■ Si02 + 2CI2 2FeCl3 + 1.5 O2 ->■ Fe203 + 3Cl2 3SiCli, + 2AI2O3 ->■ 4AICI3 + 3Si02 Not applicable. ^At 298 K, for the water as vapor. NAp figure 3, reclaim the chlorine from the unwanted chlorides by reaction with oxy- gen, and electrowin aluminum chloride to the desired metal and recyclable chlo- rine (6^). But the realities of the phys- ical world are not so simple. These realities will be examined with reference to each operation, process, and output stream in the generalized flowsheet in figure 2 and the relative quantities of materials listed in table 3. Although considerable energy is re- quired to remove the water, both free and combined, which typically totals some 1.43 lb per pound of aluminum, clay prep- aration, drying, and calcination present no problems. As was stated earlier, care need only be taken to insure that the innate reactivity of the clay is not destroyed in these operations. A solid carbon reductant and/or suitable catalyst can be added prior to calcination. Carbon added at this point must be very finely divided and evenly distributed be- cause its reducing effect appears to oc- cur only at the immediate clay-carbon interface. The chlorination process itself is com- plex and several reactions occur simulta- neously with the chlorination of the alu- minum oxide portion of the clay. It is desirable to have this latter reaction proceed rapidly and completely, so that the reaction is as small as possible, and so that the chlorine is entirely consumed and need not be separated from the prod- uct gases and recycled. Although the negative value of the Gibbs function TABLE 3. - Quantities of materials per pound of aluminum, pounds Free water in wet clay (assumed, -14 pet) 0.76 Combined water 67 Oxygen associated with aluminum 89 Reductant for aluminum oxide: Carbon, C ^ CO 67 Carbon, C > CO2 33 Carbon monoxide, CO -> CO2 1.56 Chlorine in aluminum chloride 3.94 Aluminum chloride 4.94 ^Assuming total chlorination. ■^Assuming no chlorination. Silicon tetrachloride-'^ 6.05 Chlorine in silicon tetrachloride^. 5.05 Silicon dioxide residue^ 2.14 Iron Iron trichloride ^ Chlorine in iron trichloride Titanium Titanium tetrachloride^ Chlorine in titanium tetrachloride^ .059 .17 .11 .050 .20 .15 Q r— 71 1 — I I I I FeaCI Feci. Circles indica melting points ]te^ 50 100 TEMPERATURE,^ C FIGURE 3. - Pertinent vapor pressures. 500 1,000 predicts that a reaction is favorable, it does not indicate the rate of the reac- tion. In order to obtain a sufficiently rapid reaction, a temperature of between 550" and 700° C is necessary (26). (Table 2 is based on 900 K or 627° C.) In this temperature range, several catalysts have been reported to enhance the chlorination rate of the aluminum portion of clay. These include sodium chloride (26) , boron trichloride ( 51 ) , and sulfur compounds ( 52 ) . Each of these increases the reaction rate of alumina but, with the exception of higher concen- trations of boron trichloride, they do not appear to change the rate of silica chlorination. The net effect is there- fore an apparent selective chlorination of alumina. The mechanism by which these catalysts work continues to be investigated. It is known that sodium chloride and perhaps boron chloride combine with some of the aluminum chloride product to form a liq- uid phase. These liquids are capable of dissolving small amounts of the aluminum oxide portion of the clay, and it is sus- pected that this dissolved aluminum oxide chlorinates easily and leaves as the volatile chloride. The fate of the catalyst as well as its effectiveness is of importance, A sodium chloride catalyst could remain in large part with the unreacted residue. It would be necessary to remove this soluble salt before disposing of the solid resi- due. The sulfur and boron most likely would leave the reactor as obnoxious gaseous chlorides that require very low condensation temperatures (refer to fig- ure 3) or efficient chemical scrubbing for their removal. If scrubbing is nec- essary the catalyst is not likely to be available for reuse (_5) . It is also pos- sible that any of these catalysts would combine with or be adsorbed on the condensed aluminum chloride product and require special removal procedures. Finally, the corrosive nature of the catalyst products needs to be considered with respect to equipment construction. Liquid phase sodium aluminum chloride could attack refractories in the re- actor and the combination of sulfur and chlorine is known to be quite corrosive (34). The carbon reductant used in the chlo- rination reaction must be chosen to ful- fill several requirements. First, it must be reactive, i.e., it must sustain the desired alumina chlorination reaction with respect to rate and extent, and it should be somewhat selective towards the chlorination of alumina over silica. Carbon monoxide and phosgene best fill these requirements (i.»^e reactions 4 and 5, table 2; some silica does chlorinate but this is not indicated in the equation) . Solid carbons (reactions 2, 3, and 6) promote rapid reactions only if they are finely divided or have a high sur- face area characteristic of expensive activated charcoal (26) . According to published data, solid carbon is less selective in promoting chlorination of the alumina over the silica portion of clay (26). However, elemental carbon has a higher capacity for reaction with ox- ides if the final product is carbon di- oxide and therefore less carbon is needed (compare reactions 2, 3, and 4). When handling and storage are consid- ered, solid carbon would be the most desirable reductant. Extremely toxic carbon monoxide gas would have to be stored under pressure, requiring special precautions. Phosgene too is toxic but can be detected by its odor and immediate physiological effects. It has the advan- tage of being able to be stored as a liquid under reasonable pressure and in- corporates both the reducing agent and chlorine in one compound. The enthalpy of reaction is of con- siderable importance when selecting the reductant. A chlorinator of the size to be used in the aluminum industry must be internally heated. External heating would require that the outer containment vessel be the hottest part of the reactor and any chlorinating agent reaching it would react violently with the metal at the temperatures required. With internal heating the outer vessel can be kept as cool as desired. Heat can be generated within the re- actor by feeding hot reactants , burning some of the carbon reductant with added oxygen, and by the chlorination reaction itself. Burning with added oxygen not only requires oxygen and more carbon but it also dilutes and adds to the volume of the product gases. The clay itself can be fed hot directly from the calcination step. As indicated by the values of enthalpy in table 2, only two of the important reactions occurring in the chlorinator, reactions 3 and 11, are endo thermic and should be avoided. The carbon monoxide generated in reaction 3 would most likely be reacted further ac- cording to reaction 4. Reaction 11 will be discussed later. Finally, the purity of the reductant need be considered. Hydrogen, sulfur, and alkaline metals and their oxides associated with some carbon materials would be detrimental. These would be chlorinated in the reactor leaving a sticky residue of the nonvolatile chlo- rides or add undesirable vapors to the exiting gas, and all would consume valu- able chlorine. In this latter case small amounts of hydrogen are particularly wasteful because only one part of hydro- gen combines with 35 parts of chlorine by weight to give hydrogen chloride. Silica, being the predominant species in clay, requires special consideration with respect to conservation of chlorine. Even with the best selectivity reported for the preferential chlorination of the alumina fraction of clay, some silicon tetrachloride also is produced (26) . With no market for the large amount that would be a byproduct of clay chlorina- tion, disposal of silicon tetrachloride would be necessary. This would not be easy because it readily hydrolyzes to silica and hydrogen chloride, a waste of valuable chlorine. Three solutions to this problem have been proposed. The first and least acceptable is to scrub the exit waste gases containing the silicon tetrachloride vapor with a caus- tic solution, losing the chlorine value as salt. Another solution is to react the silicon tetrachloride with oxygen to give disposable silica and recyclable chlorine (27). An efficient, proven technology for this option is not now available. The third proposal is to re- cycle silicon tetrachloride through the chlorinator to take advantage of reac- tion 11 listed in table 2 (^i, 41). In the temperature range considered for a clay chlorinator, this reaction may go in either direction. At higher tempera- tures and higher concentrations of sili- con tetrachloride, the reaction favors the formation of aluminum chloride. At slightly lower temperatures and higher ■concentrations of aluminum chloride, the aluminum chloride will react with silica to form silicon tetrachloride and alu- mina. Therefore, by proper selection of temperature and the amount of silicon tetrachloride fed with the incoming chlo- rine, it is conceivable to prevent further chlorination of silica in the chlorinator. Both of the latter methods require the recovery of silicon tetra- chloride from the gaseous products. Condensation to accomplish this separa- tion requires that the entire gaseous product stream be cooled to below 0° C and probably as low as -40° C where the vapor pressure for silicon tetrachloride is about one-hundredth of an atmosphere. Such a recovery system would require sizable equipment and considerable energy ( 53 ) . Even if it were economical to recycle silicon tetrachloride, circumstances have been found which could prevent its feasi- bility. It has been reported that clay exposed to silicon tetrachloride vapor at high temperatures becomes nonre- active with respect to chlorination (26). Catalysts such as sodium chloride may re- verse this condition ( 31 ) . The most attractive solution to the silicon tetrachloride problem is to limit its formation to a level where recovery of the chlorine value associated with this byproduct is not necessary. Limit- ing silica chlorination can be quite com- plex (30). In addition to the reductant and catalysts used, the chlorination re- actor configuration can influence silicon tetrachloride formation ( 29 ) . The opera- tion of an ideal fluidized bed chlo- rinator illustrates this point. The solid feed would consist of clay, a solid reductant, and catalyst. Chlorine would be the fluidizing gas. Outlet gas- es ideally would be only carbon dioxide and aluminum chloride vapor. Conse- quently, the solids overflow would con- sist of excess reductant, catalyst, and silica residue, with very little alumina. This is also the composition of the bed. Even though the alumina may be much more reactive than the silica, in this arrangement there is a much greater op- portunity for the silica to react. In contrast to the fluidized bed con- cept, a countercurrent reactor in which the gas stream flows counter to the sol- ids could enhance selective chlorination of the alumina. Even though incoming chlorine would encounter reacted clay consisting primarily of silica, any sili- con tetrachloride formed would have an opportunity to react with alumina in fresh clay as the gas stream left the re- actor (reaction 11). On the other hand, the countercurrent reactor allows any silicon tetrachloride formed to pass over yet unreacted clay with the possibility of inactivation as mentioned above. Clay chlorination in a liquid phase re- actor with molten salts as the liquid and catalyst has been suggested (18). In this case, residual powdery silica from the unreacted clay would be difficult to separate and remove from such a reactor. Whatever type of reactor is considered, a solid residue will be formed. Disposal 10 of this residue must be considered. Ideally the residue would consist of in- nocuous silica, but in practice there would also be some chlorinated products. These, especially any toxic chlorinated carbon compounds , would have to be de- stroyed or disposed of safely. The amounts and nature of such compounds de- pend upon the minor constituents of the clay, reductant, and catalyst and have not as yet been reported. PURIFICATION OF ANHYDROUS ALUMINUM CHLORIDE As it emerges from the chlorinator, product aluminum chloride is but a part of a mixture of gases and vapors from which it must be separated and purified. Under the most ideal chlorinator opera- tion, represented by reaction 2 in ta- ble 2, aluminum chloride vapor would con- stitute only slightly more than half of the total volume of the product stream. Furthermore, at the temperatures neces- sary for condensation, aluminum chloride forms dimers and complex molecules as indicated in reactions 12 and 14 (35) of table 4. As such it would represent only 40 pet of the product volume (and pressure) . Therefore, according to the vapor pres- sures shown in figure 3, to recover by condensation 99 pet of the product at a total pressure of 1 atm, the temperature would have to be lowered to 114° C. This is about 75° C below the melting point of aluminum chloride. The above exercise in stoichiometry brings to light the two factors that complicate any process for purifying aluminum chloride: the tend- ency of aluminum chloride to form complex molecules with itself and other chlorides and its high vapor pressure at ing point (16). its melt- It would not be practical to recover aluminum chloride as a liquid from chlo- rinator gases; the total pressure re- quired is too high. Instead it would be condensed as a fluffy powder unless special conditions such as fluidized bed condensers were employed. The more dense, granular, more easily handled product of the fluidized bed condenser is preferred ( 23 , 46 ) . Because aluminum chloride is hygroscopic, it must not come into contact with atmospheric moisture and, therefore, any transport or storage between the chlorination and purification processes requires specially designed equipment. A free-flowing material such as the product of the fluidized bed would be advantageous. As a first consideration, controlled condensation appears to be the solution to obtaining a pure aluminum chloride from the chlorination product stream (2). According to the simplistic presentation of vapor pressures in figure 3, the vapor pressures of aluminum chloride and ferric chloride differ by a factor of about 2,000, a ratio near the acceptable limit for iron in the aluminum chloride prod- uct. Two condensers in series would ideally deliver this product. The first condenser, operating at about 190° C, would remove ferric chloride to an acceptably low level without condens- ing any aluminum chloride. The second condenser, held at 100° C, would remove essentially all of the aluminum chloride. Remaining in the gas exiting the second condenser along with the carbon dioxide and carbon monoxide, would be titanium TABLE 4. - Reactions pertinent to aluminum chloride purification Reaction Enthalpy , ^^5 00 K' kcal Gibbs function, kcal (12) 2AICI3 ->► AI2CI6 (13) 2FeCl3 ->■ FeaCle (14) AI2CI6 + FeaCle ^ (AlFeCl6)n (15) FeCla -y FeCl2 + 0.5 CI2 -30.0 -34.2 ±1 26.9 -12.8 -17.2 ±1 19.1 11 tetrachloride and any silicon tetrachlo- ride, catalyst products, and unreacted chlorine. These chlorides could be con- densed at a much lower temperature, sub- sequently separated from one another by fraction distillation, and used. If not recovered they would have to be scrubbed and disposed of as waste. There is no mention in the technical literature of difficulties in separating these lower boiling compounds from aluminum chloride; however, the separation of chlorides with low vapor pressures from aluminum chlo- ride is reportedly not as simple as it might seem. Separation of aluminum and ferric chlo- rides is complicated by the formation of the mixed dimer AlFeCls according to re- action 14 of table 4 (47-48). This molecule and multiples of it have been reported from mass spectrometer observa- tions of the vapor over mixtures of the two simple chlorides (11, 21 , 35 , 47 ) . Measured properties of the mixed iron- aluminum chloride indicate that its vapor pressure and the equilibria for reaction 14 are such that the iron is more inti- mately associated with aluminum chloride than would be expected from simple vapor pressure data. There is a tendency for the mixed chloride to form and its vapor pressure appears to be closer to that of the aluminum chloride dimer than is the vapor pressure of ferric chloride. Considerable effort has been put into developing a scheme to separate iron and aluminum chlorides but, notably, success of suitable technology for the solution to this problem has not been reported in the literature. Although fractional dis- tillation, fractional sublimation, and reduction of ferric chloride to less vol- atile ferrous chloride have been sug- gested, the operation of a practical scheme to separate iron chloride from aluminum chloride has not been reported. Reducing the iron chloride to the less volatile ferrous chloride or iron metal followed by separation by volatilization of the aluminum chloride may be possible (4^) if the appropriate reducing agent can be found (8^) . Aluminum metal is a prime candidate for the reductant since it would not add another material to the system, only more aluminum chloride. It is, however, an expensive reagent and there are technical problems. If solid aluminiom is used (i.e., if the reduction is carried out below 660° C) large sur- face areas of aluminum are needed to con- tact the liquid or vapor aluminum chloride-iron chloride mixture. This surface soon becomes covered with reac- tion products and loses its reactivity. If the impure aluminum chloride vapor is passed through molten aluminum in order to reduce and remove the iron impurity, high melting iron-aluminum intermetallic compounds form. These solids in the molten aluminum bath quickly hinder the operation. Fractional sublimation on a large scale has never been a success owing to the difficulties inherent in the mechanics of moving large quantities of condensing solids. In the case of aluminum chloride the strict avoidance of contact with the atmosphere would only add to these difficulties. Although other schemes have been pro- posed (40) , fractional distillation is the most attractive method for separating ferric chloride from aluminum chloride. The feasibility of this operation depends upon the nature of the vapor-liquid equi- libria for the binary system, for which there are yet no data in the literature. Finally, there are minute quantities of other elements such as calcium, magne- sium, and phosphorus that have not been considered. These may present problems as impurities which associate with the aluminum chloride ( 10 ) . Only extended pilot scale operation of an aluminum chloride-f rom-clay process could identify the importance of these potential impurities. 12 CONCLUSIONS There is a strategic incentive in the United States to use domestic clay as a resource for aluminum; and there is an economic incentive to develop a chloride metallurgy to produce aluminum from the clay. Together these make the chloride route from clay to aluminum attractive and worthy of serious consideration. However, there are several basic problems to be investigated and much development work to be done before it can be deter- mined if such a tc»chnology can in fact be made practical and economical on a large industrial scale. Only a sustained, well coordinated research and development effort can provide the data needed for technical and economic evaulations of any proposed chloride route for aluminum from clay. At the present time two areas need to be addressed: the clay chlorination reaction and the purification of anhy- drous aluminum chloride. In particular, the chlorination reaction must be under- stood so that it can be assessed with respect to rapid and complete use of chlorine; selectivity of alumina reactiv- ity over silica; use of a convenient, inexpensive reductant; and heat genera- tion to sustain the reaction. In addition, the identity and nature of product impurities and their interactions with anhydrous aluminum chloride must be determined. With such information a chlorination process and purification scheme can be proposed and tested. Only then will the technical operation prob- lems surface; and only if these can be solved with due regard to economic con- siderations will the chloride route from clay to aluminum become industrially possible. REFERENCES 1. American Society for Metals (Met- als Park, OH), Properties and Selections of Metals, Metals Handbook, 9th ed, , V. 2, 1979, 65 pp. 2. Arnold, P, M, (assigned to Phil- lips Petroleum Co., DE). Method of Puri- fying Anhydrous Aluminum Chloride. U.S. Pat, 2,387,228, Oct. 23, 1945. 3. Battelle Columbus Laboratories. Energy Use Patterns in Metallurgical and Nonmetallic Mineral Processing' (Phase 9 — ^Areas Where Alternative Tech- nologies Should be Developed To Lower Energy Use in Production of High-Priori ty Commodities) (Contract SO144093). Bu- Mines OFR 117(4)-76, August 1976, 68 pp.; NTIS, PB 261 153/AS. 5, Carpenter, L. G, , D. L. Kinosz, and V. Cedro, III (assigned to Aluminum Company of America, Pittsburgh, PA). Treatment of Off gas From Aluminum Chlo- ride Production. U.S. Pat. 4,209,496, June 24, 1980. 6. Chemical & Engineering News, Chlorination Process Upgrades Low-Grade Ores. Nov. 29, 1971, pp. 31-32. 7. Market PP Magnetic Breakthrough, 21-22. Separations Near Jan. 28, 1974, 8. Crouch, D. A., Jr. Purification of Anhydrous Aluminum Chloride In Situ in a Salt Melt. U.S. Pat. 3,832,452, Aug. 27, 1974. 4, Binnewies, V, M, Massenspektrom- etrische Untersuchungen der Systeme FeCl2/AlCl3 und FeCl2/FeCl3 (Mass Spec- trometer Investigation of the Systems FeCl2/AlCl3 and FeCl2/FeCl3) , Z, Anorg, und Allg. Chem. , v, 437, 1977, pp. 19-24. 9, Dell, M. B., W. E. Haupin, and A, S, Russell (assigned to Aluminum Com- pany of America, Pittsburgh, PA), Elec- trolytic Cell for Metal Production, U,S. Pat, 3,893,899, July 8, 1975, 13 10. Dewing, E, W. Gaseous Complexes Between Bichlorides and Trichlorides of Aluminum and Iron. Nature, v. 214, 1967, p. 483. 11. Fowler, R. M. , and S. S. Melford. FeAlCls a Volatile Molecule Formed by the Reaction of Aluminum Chloride and Ferric Chloride. Inorg. Chem. , v. 15, No. 2, 1976, pp. 473-474. 12. Freund, F. Beitrage zum kera- mischen Brennprozess. Teil II: Die Wasserabspaltung aus dem Gitter der Tonminerale (Ceramic Calcining Process. Part II: Water Splitting From the Lat- tice of Clay Minerals). Keramische Zeitschrift, v. 19, No. 7, 1967, pp. 433-436. 13. Grjotheim, K. , C. Krohn, M. Mali- novsky, K. Matiasovsky, and J. Thonstad. Aluminum Electrolysis. William Clowes and Sons, Ltd., London, 1977, 350 pp. 14. Grjotheim, K. , and J. B. See. The Hall-Heroult Process and Alternative Processes for the Manufacture of Metallic Aluminum. Miner. Sci. and Eng. , v. 11, No. 2, April 1979, pp. 80-98. 15. Grjotheim, K. , and B. Welch. Im- pact of Alternative Processes for Alumi- num Production on Energy Requirements. J. Met., V. 33, No. 9, September 1981, pp. 26-32. 16. Guttinger, E. Uber die Reindar- stellung von Aluminiumchlorid aus den Chlorierungsprodukte des Bauxites (On the Purification of Aluminum Chloride From the Chlorination Products of Bauxite) . Ph. D. Thesis, Dis. No. 4236, Zurich Tech. Univ., Zurich, Switzerland, 1969, 141 pp. 17. Haupin, W. E., and B. J. Racunas (assigned to Aluminum Company of America, Pittsburgh, PA). Preparation of Alkali Metal Chloride Melt for Use in Electro- lysis of Aluminum Chloride. U.S. Pat. 3,761,365, Sept. 25, 1973. 18. Haupin, W. E., J. A. Remper, and M. B. Dell (assigned to Aluminum Company of America, Pittsburgh, PA), Production of Aluminum Chloride. U.S. Pat. 4,039,648, Aug. 2, 1977. 19. Hindar, J., J. L. Holm, J. Linde- mann, and K. Lonvik. Investigation of Some Kaolins by Simultaneous DTA/TGA and Thermosonimetry. J. Therm. Anal., v. 2, 1980, pp. 313-318. 20. Holliday, R. D. , and D. J. Milne. Experimental Evaluation of Routes for Purification of Bauxite by Gas-Solid Re- actions, I & EC Process Des. and De- velop., V. 14, No. 4, 1975, pp. 447-452. 21. Jorgensen, F. R. A., and F. J. Moyle. Solid- Vapor Equilibria in the FeCl3-AlCl3-Cl2 System. Proc. Australas. Inst. Min. Me tall. , No. 269, March 1979, pp. 29-35. 22. Kerr, P. F. Reference Clay Min- erals. American Petroleum Institute Re- search Project 49, Columbia University, New York, 1951, 699 pp. 23. King, L. K. , L. L. Knapp, R. C. Schoener, N. Kloap, B. M. Starner, and J. A. Remper (assigned to Aluminum Company of America, Pittsburgh, PA). Recovery of Solid Selectivity Constituted High Purity Aluminum Chloride From Hot Gaseous Effluent. U.S. Pat. 3,786,135, Jan. 15, 1974. 24. Kurtz, H. F., and L. H. Baumgard- ner. Aluminum. Ch. in Mineral Facts and Problems, 1980 Edition. BuMines B 671, 1981, pp. 9-34. 25. Landsberg, A. Chlorination Ki- netics of Aluminum Bearing Min- erals. Met. Trans. B, v. 6B, June 1975, pp. 207-214. 26. Some Factors Affecting the Chlorination of Kaolinic Clay. Met. Trans. B, v. 8B, September 1977, pp. 435-441. 14 27. Lippman, A,, and R, F, Sebenlk (assigned to Toth Aluminum Corp. , New Orleans, LA). Process for the Production of Aluminum Chloride and Re- lated Products. U.S. Pat. 4,083,923, Apr. 11, 1978. 28. Mah, A. D. Chemical Equilibria in Chlorination of Clay. BuMines RI 8696, 1982, 43 pp. 29. Martin, E, S, (assigned to Alumi- num Company of America, Pittsburgh, PA). Production of Aluminum Chloride From Clay Using Staged Reactors. U.S. Pat. 4,213,943, July 22, 1980. 30. Martin, E. S., and D, A. Wohleber (assigned to Aluminum Conq)any of America, Pittsburgh, PA), Production of Aluminum Chloride From Raw Materials Containing Aluminum Compounds and Silicon Compounds. U.S. Pat. 4,086,320, Apr. 25, 1978. 31. . Production of Anhydrous Aluminum Chloride From Clay Using Cata- lyst and Recycling of Silicon Chloride. U.S. Pat. 4,096,234, June 20, 1978. 32. Milne, D. J., and R. D. Holliday. Thermodynamics of Gas-Solid Reactions for Purification of Bauxite at Moderate Tem- peratures. I & EC Process Des. and De- velop. , V. 14, No. 4, 1975, pp. 442-447. 33. Milne, D. J., and L. J. Wibberley. The Removal of Iron From Bauxite Using Anhydrous Hydrogen Chloride. Light Metals. American Institute of Mining, Metallurgical, and Petroleum Engineers, New York, 1977, pp. 125-145. 34. Oden, L. L. , D. L. Paulson, and W. C. McBee. Industrial Service Fail- ures: Corrosion of Nickel in a Zirconia Chlorinator. Materials Protection and Performance, v. 10, No. 11, November 1971, pp. 22-24. 35. Patsos , H. S. A Mass Spectrom- etric Study of Gaseous Species Related to the Carbochlorination of Aluminum-Bearing Ores. M.S. Thesis, CO School of Mines, Golden, CO, 1982, 257 pp. 36. Patterson, S. H, Potential Alumi- num Resources of the World. U.S. Geol. Surv. Bull. 1228, 1967, pp. 129-135. 37. Peacey, J., and W. Davenport. Evaluation of Alternative Methods of Aluminum Production. J. Met., July 1974, pp. 24-28. 38. Peters, F. A., and P. W. Johnson, Revised and Updated Cost Estimates for Producing Aluminum From Domestic Raw Materials. BuMines IC 8648, 1974, 51 pp. 39. Pratt, W. P., and D. A. Brobst. Mineral Resources: Potentials and Prob- lems. U.S. Geol. Surv. Circ. 698, 1974, pp. 3-5. 40. Rado, T, A. (assigned to Kerr- McGee Corp., Oklahoma City, OK). Process for the Production of Aluminum Chlo- ride and Related Products. U.S. Pat. 4,283,371, Aug. 11, 1981. 41. Reynolds, J. E., and A. R. Wil- liams (assigned to Public Service Company of New Mexico, Albuquerque, NM), Process for Chlorinating Clays and Bauxite. U.S. Pat. 4,288,414, Sept. 8, 1981. 42. Russell, A. S. Pitfalls and Pleasures in New Aluminum Process De- velopment. Met. Trans. B, v. 12B, June 1981, pp. 203-215. 43. Russell, A. S., N. Jarrett, and P. T, Stoup (assigned to Aluminum Company of America, Pittsburgh, PA). Produc- tion of Aluminum Chloride. U.S. Pat. 3,842,163, Oct, 15, 1974. 44. Russell, A. S., and L. L. Knapp (assigned to Aluminum Company of Ameri- ca, Pittsburgh, PA), Furnace Structure. U.S. Pat. 3,773,643, Nov. 20, 1973. 45. Russell, A. S., L. L. Knapp, and W. E. Haupin (assigned to Aluminum Com- pany of America, Pittsburgh, PA). Production of Aluminum. U.S. Pat. 3,725,222, Apr. 3, 1973. 15 46, Schoener, R. C, L, K. King, L. L. Knapp, and N, Kloap (assigned to Aluminum Company of America, Pittsburgh, PA), Fluidized Bed Desubliming Apparatus for Recovery of Aluminum Chloride, U.S. Pat, 3,930,800, Jan, 6, 1976. 47, Semenko, K. N, , T, N, Navmova, L, N, Garokhov, G, A, Semenva, and A, V, Novoselova. Interaction of the Chlo- rides of Al and Fe, Dokl, Akad, Nauk SSSR, V. 154, No. 1, January 1964, pp. 169-170. 48, Shieh, C, and N, W, Gregory. Spectrophotometric Study of the Va- pors of Iron (III) Chloride and Mix- tures of Iron (III) Chloride and Aluminum (III) Chloride. Evidence for Forma- tion of Mixed Metal Dimer Molecules. J. Phys. Chem. , v, 79, No, 8, 1975, pp, 828-833, 49. Stamper, J, W, , and H, F, Kurtz, Aluminum. BuMines Mineral Commodity Pro- file, May 1978, 5 pp, 50. U,S, Bureau of Mines, Staff, Po- tential Sources of Aluminum. IC 8335, 1967, 148 pp. 51. Wyndham, R. (assigned to Toth Alu- minum Corp., New Orleans, LA). Con- trolled Carbo-Chlorination of Kaolinitic Ores. U.S. Pat. 4,083,927, Apr. 11, 1978. 52. Wyndham, R, , and J, C. Terry (as- signed to Toth Aluminum Corp,, New Or- leans, LA), Clay Halogenation Process, U,S, Pat, 4,082,833, Apr, 4, 1978, 53. . Preferential Chlorination of Alumina in Kaolinitic Ores. U.S. Pat. 4,139,602, Feb. 13, 1979. INT.-BU.OF MINES,PGH.,P A. 26701 ;V'.i'^>'',TC^ PD 18l * ♦ CiK^ • ^x- A\y * j{\ sa A. "^ ?i!'i4! ■.-iC-'Mi • I "I \ ■::: A;^:ii# "■ ■WMi ; *• ,.i-'Vi".'lffl LIBRARY OF CONGRESS 002 959 908 7