carried out using a relatively high discard level of the order of atm for all species, and this served to establish an accurate estimate of the input distribution for a second run. The second calculation was then able to establish equilibrium levels of gaseous species to below atm. This was the fundamental feature that enabled a predominance area diagram to be used for a wide range of gas compositions. The method of checking the accuracy of calculation of these low component levels may be demonstrated by referring to Figure 4. The triple point for Fe-FeS-FeO was calculated by conventional means, using data from Kubaschewski et al. (1967). The unique values for sulfur and oxygen pressures a t equilibrium a t lOOOOK were 0.96 X 10-lo atm, respectively. The triple point could and 2.33 X be reproduced by the minimization program by equilibrating a closed system containing 3.6 mol of SO2, 6.4 mol of CO and excess iron. The equilibrium distribution was found to contain the solid components a t sulfur and oxygen and 2.4 X atm, respectively. pressures of 1.2 X This degree of precision was attained in other similar systems, confirming that the use of the low calculated partial pressures was justified. L i t e r a t u r e Cited Buntin, A,, et al., Tr. Tomsk. Gos. Univ.: Ser. Khim., 154, 52 (1962). Chao. T.. Ph.D. Thesis, Pennsylvania State University, 1966. Finke, C., de Marchi, V., Trans. Electrochem. Soc.,74, 469 (1938). Foley, E., Tittle, K., Proc. Aust. lnst. Min. Metall., 239, 59 (1971).
Glass, A.. Ross, 0.. Can. J. Chem., 50, 2537 (1972). HaUpin, W. E. (to Aluminum Co. of America), U S . Patent 3,755,099 (Aug 28, 1973). Ketteridge, i., Ph.D. Thesis, University of Adelaide, Adelaide, Australia, 1969. Khundkar, M.. Ahmad, N., J. lndlan. Chem. SOC.,18,109 (1955). Kubaschewski, O., et ai., “Metallurgical Thermochemistry,” Pergamon. London, 1967. McIntosh, P., Holliday, R. D. (to Conzinc Riotinto of Australla), Australian Patent 445,623 (Mar 4, 1874). Oliver, S.,et el., Chem. Eng., 80, 121 (1962). Perieres. R. (to Cie. Pechiney), French Patent 1,495,002 (Sept 15, 1967). Russell, A. S., et ai. (to Aluminum Co. of America), German Patent 2,244,041 (Mar 22, 1973). Semenenko, T., Naumova, T., Do&/.Akad. Nauk SSSR, 154, 169 (1964). Spitzvin, V., 2.Anorg. Allg. Chem., 180, 337 (1930). van Zeggeren, F., Storey, S., “The Computation of Chemical Equilibria,” Cambridge University Press, Cambridge, 1970. White, W., et al., J. Chem. Phys., 28, 751 (1958).
Receiued for reuiew December 2,1974 Accepted May 14,1975
Supplementary Material Available. Details of the mathematical procedure will appear following these pages in the microfilm edition of this volume of the journal. Photocopies of the supplementary material from this paper only or microfiche (105 X 148 mm, 24X reduction, negatives) containing all of the supplementary material for the papers in this issue may be obtained from the Business Office, Books and Journals Division, American Chemical Society, 1155 16th St., N.W., Washington, D.C. 20036. Remit check or money order for $4.00 for photocopy or $2.50 for microfiche, referring to code number PROC-75-442.
Experimental Evaluation of Routes for Purification of Bauxite by Gas-Solid Reactions Robin D. Holliday and Davld J. Mllne’’ University of Newcastle, New South Wales 2308, Australia
Conditions for preparation of AIC13 of low iron content were studied in a thermogravimetric apparatus designed so that sensitive components were shielded from the gaseous reagents. Reaction of natural bauxite pisolites of diameters between 1.5 mm and 12 mm with gaseous reagents based on S02-C0, HCI-CO, HCI, HPS-CI~, S2C12-CO, S02-C12 and CO-CI2 was analyzed. Predictions that Sop-CO mixtures of high sulfur potential would be particularly suitable for preparation of FeS were confirmed. Maximum sulfidization rates occurred at the SO21 CO ratio of 3565. Rates were proportional to (p(S2)2 p(COS).p(Sd p(COS)2). FeS could be rapidly removed to residual Fe levels of 0.1 to 0.3% by direct chlorination below 750OC. Direct chlorination of FepO3 using CO-CI2, HCI, or HCI-CO gave much less favorable results. The possibility of removing AIC13 while retaining FeS is unique to the H2S-C12 reagent system, but the reaction kinetics were unfavorable. The sulfidizationchlorination route appears to afford the best combination of characteristics sought in a beneficiation process suitable for integration with AIC13 production.
+
A detailed thermodynamic analysis of routes for removal of iron from bauxite under anhydrous conditions defined a number of reactions that merited experimental study (Milne and Holliday, 1975). The experimental program aimed to establish which of the prospective routes could be realized under the least severe reaction conditions. An attractive process was defined as one which afforded the best Address correspondence to this author at Comalco Ltd., 95 Collins St., Melbourne, Australia.
+
combination of the desired characteristics: low temperatures to minimize corrosion, high reaction rates to minimize reactor size, together with low reactant costs and ease of recovering or recycling intermediates. The essential criterion was that iron removal be highly efficient, so in practice only processes found capable of producing iron levels below 0.5% were selected for detailed study. Further, the temperature of reaction must be kept below 75OOC to avoid serious impairment of reaction rates in the subsequent chlorination of yA1203. Id.Eng. Chem., Process Des. Dev., Vol. 14, No. 4. 1975
447
FWACTIONA;
CONVERSlOh. X
8 D 100
c ga
i
:I
9
7
1 6
TIME,
7
MINS
Figure 1. The sulfidization of bauxite pisolites. Relation between reaction time, t , and F(X) where F(X) = 0.45(1 - (1 - X ) l I 3 ) t 0.55 (1 - 3(1 - X ) 2 / 3t 2(1 - X)). Experimental conditions are given in Table V.
Table I. Analysis of Bauxite from Weipa, Australia ( w t %)
*bo, 58
7.5
Si02 5 .O
Ti02 2.5
L.O.I. 27
Experimental Section The progress of reactions was followed thermogravimetrically using a cantilever beam balance designed so that sensitive components were shielded from contact with corrosive gases. Strain gauges attached to the beam were monitored to provide a continuous record of weight changes. The reaction vessel and sample suspension system were the only components exposed to the reactant gases, and these were constructed from pure fused silica. A silica fiber basket was used to support spherical reactant particles so that access of reactant gas was unimpeded. The normal precautions were taken to ensure completeness of gas mixing and preheating, and to prevent gas-starvation effects. Materials. One specific type of bauxite was used in the main part of the present investigation. This was material from the Weipa deposit in Queensland, Australia, having a typical analysis shown in Table I. The material occurs predominantly as spherical particles (pisolites), which afford a natural range of particle diameters between 1.6 mm and 12 mm well suited to the experimental requirements. The principal aluminous component is the trihydrate, A1203-3H20, which on heating to 5OOOC and above dehydrates and develops a submicroscopic (below 100 A diameter) internal pore system with an extremely large specific surface area. The chemical reactivity is strongly dependent on the surface area developed, and unless otherwise stipulated, reactant samples were preheated to 75OOC to remove 448
Ind. Eng. Chem., Process Des. Dev., Vol. 14, No. 4, 1975
Figure 2. The chlorination of sulfided bauxite. Relation between reaction time, t , and ( 1 - ( 1 X)'l3). Experimental conditions are given in Table V.
-
water and to develop a correspondingly high surface area. The reactivity of the A1203 matrix in chlorination is seriously impaired by heating to above 75OOC. Results Reaction w i t h SO&O Followed by Chlorination. The possibility of using SO2-CO mixtures to convert FepO3 to FeS without reaction of A1203makes this system of particular importance, since FeS reacts readily a t low temperatures to form FeC13 which can thus be separated from the bauxite by volatilization. The AG-minimization technique affords sufficient insight into the process to permit experiments to be tailored so that reactant gas of the highest sulfur potential can be selected and the temperature chosen consistent with the production of the desired FeS rather than FeS2 or Fez03 (Milne and Holliday, 1975). On this basis, exploratory experiments were carried out using a 3:7 ratio of SO2/CO at 700°C, although this reactant composition is not stoichiometric with the expected reaction Fez03
+ 2502 + 7CO = 2FeS + 7 c o 2
Sulfidization was followed by chlorination in the absence of a reducing agent. The initial experiments showed that conversion rates were very favorable (Figures 1 and 2), and accordingly an extensive study of the system was undertaken. Rate-Controlling Factors-Sulfidization. The relation between the fractional conversion to sulfide, and time of reaction, t , was not simple. By computer-aided trial and error curve-fitting, it was established that the best fit of the data was provided by equations of the form
x,
r--
1 \
2-
1 D
‘C
fi1
91
Figure 4. Relationship between relative pseudo reaction rate constants, K/Kma and the function G ( p ) = ( p ( S z ) *f p(COS).p(Sn)+ p(COS)*) a t 74OOC.
I
L_
/
I
\ 20
40
60
I
80
i3
Figure 3. Change in measured reaction rate and calculated sulfidizing component concentration for SOdCO mixtures at 730°C.
Reaction rates are rehtive to the maximum rate shown in the upper curve, Le., K/K(max).
Kt
A(l
- (1- X)1’3)IB(l
- 3(1 - X ) 2 / 3+ 2(1 - X)) (i)
where K is a pseudo rate constant and A and B are numerical coefficients which depend on reaction conditions. This is the relation to be expected if diffusion of reactants through the product layer proceeds a t the same rate as the chemical reaction a t the surface of a spherical core of unreacted material (Lu, 1963; Seth and Ross, 1965). There are a number of reasons why an exact fit over the entire range of variables is not to be expected in the present system. The principal of these are that the equation was derived for a single component reactant gas reacting a t a sharply defined interface. In the present system, the reactant composition is complex, and the pore structure is such that diffusion in the Knudsen regime may occur throughout the particle so a relatively diffuse reaction front exists. Further, in the bauxite system, nonisothermal conditions may occur within particles because the sulfidization reactions are exothermic; additional complications arise because some variation in stoichiometry of the iron sulfide is t o be expected as the sulfur potential of the reactant gas is changed. Finally, some inhomogeneity of iron distribution is characteristic of bauxite, so the assumption of uniform solid reactant composition is inexact. In fact, a slight spread in the best values for the coefficients was encountered, and in four out of 28 determinations no satisfactory fit could be established. Despite these limitations, the correlation provided a useful basis for estimating rate constants to describe the effects of variation of reactant gas composition (Figure 3). Relation i was the only one of a very large number tested that provided an acceptable fit for the bulk of the data. In the SO&O system a t high temperature, a complex mixture of gases exists, and this must be taken into account in interpreting reaction rate measurements. The experimental evidence strongly indicated that the predicted equilibrium was in fact established in the reactor. Thus, on in-
troducing the premixed gases a temperature rise was detected which showed that exothermic reaction was occurring; further, sulfidization reaction rates did not depend upon gas flow rate, showing there was no variatiorr of reactant composition with residence time in the hot zone. In addition, sulfur deposition occurred in the gas off-take for compositions where high p(S2) was predicted. Most important, a simple explanation of the observed rates could be formulated on the basis of the calculated equilibrium. The dependence of the pseudo rate constant on gas composition is shown in Figures 3 and 4. In these plots, K is shown as a fraction of the value K(max) observed a t the SO&O ratio of 35:65. In Figure 4,these K values are superposed on a plot of the function G ( p ) = (p(S2).p(COS) + ~ ( S Z ) P~( C O S ) ~calculated ) from the equilibrium compositions of the reactant mixtures. G ( p ) also attains B maximum near the 35:65 composition and is plotted as a fraction of this maximum value. Evidently, M is proportional to Gb),so that differentiation of (i) leads to relation (ii) for the conversion rate
+
dX/dt = K”(p(S&p(COS) f P ( S ~f ) ~
P(COS)~)/F’(X) = K”G(p)/F’(X) (ii) where K”G(p) = F(X)/t, the pseudo rate constant. In the present system, a second-order dependence of the rate constant on the partial pressures of the major sulfidizing components is found. For a single-component reactant, K is directly proportional to reactant pressure (IA, 1963). Dilution of reactants by addition of hT2 apparently led to further complication of the gas diffusion mechanism, and no simple relation between the rate constant and p ( N 2 ) could be established. However, the salient observation was that reaction rates were reduced far less by dilution than by departing from the optimum stoichiometry (Table 11). Most significantly, the rate measurements confirm that maximum reaction rates are indeed established a t the point , to an optimum SO2/CO of maximum ~ ( S Z )corresponding ratio of 35:65. Under optimum conditions, sulfidization was essentially complete within I5 min a t 70OOC for particles up to %s-in. diameter. Rate Controlling Factors--Chlorination o f FeS. In the subsequent chlorination of the sulfide intermediate the conversion could be represented (Figure 2) by the relation K’t = (1 -- (I -. X)”?)
(iii)
which applies when the rate is determined by the rate of chemical reaction a t the surface of a shrinking core of unreacted solid (Wen, 1968). The model was confirmed by microscopic examination of partially reacted particles and was found to apply up to 90% removal of iron in the range 550-800° C. Ind. Eng. Chem., PfQC0.W Des. Dev., VOl. 14, No. 4, 1975
449
Table 111. Effect of Particle Size a t Constant Temperature on Reaction Times for Sulfidization and Chlorination
Table 11. Effect of Diluent Gas on Reaction Rates for Sulfidization of Fez03 in Bauxite
Ratio CO/SO,
55:43 55:43 63:36 63:36 63:36
Pseudo N, rate Particle partial conTemp, diam., pressure, stant, "C mm atm K min-' 780 780 720 720 720
3.2 3.2 4.7 4.7 4.7
0.4 0.03 0.65 0.42 0.03
0.101 0.125 0.028 0.091 0.111
Ratio K/K (max) 0.81 1.0 0.25 0.82 1.0
Particle diam., mm
Time to 8 0 % reaction, min
Sulfidiz ation: 720°C. S0,'CO = 7:3
8.00 4.70 1.60 0.10
5.5 3.5 2.5 1.5
Chlorination: 720"C, p(C1,) = 0.97atm
6.35 4.76 3.17 0.10
3.36 2.58
Stage
2 .o 1.9
At very low reaction temperatures, the chlorination is better described by the relation
K"'t = -log (1- X )
t iv)
which is characteristic of a reaction taking place homogeneously throughout a porous solid. For both the sulfidization and the chlorination steps, the reaction models apply only up to about 90% conversion. Beyond this, inhomogeneity of the iron distribution exerts a marked effect, so that small pockets of high iron content remain after the passage of the main reaction front. Thus to secure complete conversion, the reaction time is significantly longer than predicted by models which assume uniform concentration of reactant. Particle Size. Most of the sulfidizing experiments were conducted on raw pisolites of 3.1 t o 8 mm diameter, and it was established that the bulk of the iron could be reacted without recourse to size reduction. For particles below 1.0 mm diameter, dependence of reaction rate on particle size is much reduced (Table 111).It is likely that diffuseness of the reaction front extends over a significant part of the particle volume, making the sharp interface model inexact. Rates for the chlorination step also showed a marked dependence on particle size for the larger diameters (Table 111). Below 2 mm diameter, the effect is much less, and again the interference is that diffuseness of the reaction front causes departures from the idealized model. Effectiveness of I r o n Removal. Under optimum conditions, pisolites containing 7.596 Fez03 could be beneficiated to final iron levels of 0.1 to 0.396, rapidly and at relatively low temperatures. Reaction times of the order of 10-20 min a t 750°C were required, using the optimum SO2/CO ratio of 35:65; similar reaction times were required for the chlorination a t 75OoC, and under these conditions A1203losses were approximately 2%. The loss of A1203 is attributable to the reaction 2A1203 + %S2C12 + 'hCl2 = 4AlC13 + 3so2 in which the S2C12 originates from chlorination of the FeS 2FeS
+ 4c12 = 2FeC13 + SzC12
Higher levels of iron could be treated by carrying out the chlorination in two stages. Thus, for a material containing 14% Fe203, the A1203 loss in a one-stage chlorination at 7OOOC was an unacceptably high 9-11%. However, on first removing 7040% of the FeS at 430°C and then continuing with a second chlorination at 68OoC, the iron could be removed with only 2% loss of A1203. A degree of variability of final iron level is to be expected because of the nonuniformity of the microstructure of the pisolites. Electron microprobe analysis indicated that local 450
Ind. Eng. Chem., Process Des. Dev., Vol. 14, No. 4, 1975
Figure 5. Thermogram for the reaction of Fen03 powder with HCl. Heating rate, 5'C/min.
areas of high iron concentration, 50-100 p in diameter, were retained in beneficiated pisolites. No discrete particles of iron compounds exceeding 1 p diameter were discernible. I t is probable that the local high iron levels originate from goethite and hematite particles of similar size that were found sparsely distributed in the original pisolites. Such particles would develop a markedly less porous structure than the A1203 matrix during calcination and would be relatively unreactive. The lowest iron levels detected by microprobe analysis of individual beneficiated pisolites were of the order of 0.0596, and this possibly constitutes the ultimate level of beneficiation for this type of bauxite. Reaction w i t h HCl. Removal of iron with HC1 or HC1 with a reducing agent has the particular advantage that HC1 is probably easier to regenerate from iron chloride than is Cl2. Previous investigators (Spitzvin, 1931; Buntin, 1962; Khundkar, 1955) have reported some success with HCI. In the present investigation HC1 has not been studied in as much det,ail as the SOz-CO system, but results are presented here to show that while some beneficiation is attainable, the overall reaction conditions and yields compare unfavorably with the S O y C O system. The complexity of the system is reflected in the thermogram, Figure 5. At low temperatures, iron oxide reacts rapidly according to Fez03
+ 6HC1*
2FeC13
+ 3Hz0
However, complete conversion does not occur and above 130°C the weight decrease is due to reversal of the reaction
rather than progressive volatilization of t h e water and FeC13. Above 400°C mixed di- and trichlorides are formed, the dichloride vaporizing above 700°C. Large excess amounts of HC1 gas are required to attain good iron conversion. H2/HCl mixtures result in production of the dichloride only. T o attain complete conversion the HC1 must again be supplied in greater than stoichiometric quantities. The requirement for recycling large volumes of gas may lead to difficulties in large-scale application. T o attain low levels and satisfactory rates of removal of iron as dichloride a temperature of 800°C is required. The kinetics of removal are not as favorable compared to the sulfidization route. A best result of about 0.7% residual iron in 120 min was attained. A feature of this route was the tendency to produce a proportion of the iron as lower oxides with H2/HC1 and CO/HC1 rather than all as the dichloride. Thus HC1 offers some prospects as a beneficiating agent under certain conditions and is receiving further study. However this reagent appears to have some significant disadvantages. Sulfidization with HzS-Clz. This system affords the possibility for selectively chlorinating A1203while retaining iron as nonvolatile FeS. In practice this gas system proved very difficult to handle, because the predicted highly exothermic reaction to produce S2Cl2 took place in the mixing chamber. In a limited series of experiments, the best results were obtained with a 1:1 mixture a t 700°C. Under these conditions, complete chlorination of the A1203 content with retention of 20% of the iron was achieved in 110 min. With 60% HzS, chlorination of A1203 proceeded slowly, although formation of FeS was rapid. Rapid formation of AlC13 occurred in 60% Cl2, but with concurrent volatilization of FeC13. Thus it appears that the desired separation, unique to the H2S-Cl2 system, is possible in principle, but reaction conditions would lead to impractically low conversion rates. Reaction with S2ClrCO. Pure Fez03 reacted very rapidly with S2ClpCO mixtures below 330°C to form FeC12, and the product could be rapidly converted to FeC13 by pure Clz. Below 330°C reaction of 7-A1203with S2ClrCO was very slow, but above 360°C the rate increased significantly. Thus an attractive process for iron removal below 330OC seemed possible. In fact, the reaction of S2C12-CO with bauxite or with mechanical mixtures of Fez03 with 7-A1203was much more complex. The reactant produced only a weight gain stoichiometric with Fez03
+ 2SzC12 + 7CO = 2FeC12 + 3 c o 2 + 4COS
when pure Fez03 was used. In the presence of 7-A1203, an initial weight gain was followed by a weight loss always equivalent to removal of about 70% of the iron. It seems likely that the marked divergence in behavior of mechanical mixtures and naturai bauxite compared to pure Fez03 is caused by a radical change in the equilibrium under closed-box conditions. With restricted access of gas reactants, the presence of excess Fez03 will cause deposition of liquid sulfur, which can then produce pore-blockage and virtual cessation of reaction of Fe2O3. Analysis of the reacted bauxite confirmed the presence of elemental sulfur. Removal of iron probably occurs because reaction to FeC13 is favored under pore conditions in the presence of A1203. Removal of elemental sulfur mandates the use of temperatures above 444"C, in the region where A1203losses become excessive. The attractiveness of the route is thus negated, and detailed investigation was not warranted.
Table IV. Low-Temperature Preferential Chlorination of Fez03 in Bauxite Using 1:l CO/C12 ChloriDehydration nation temp, "C temp, "C
Time, min
Residual % Fe
60 48 36 36 35 35
5 .O 0.97 0.80 2 .o 1.o 0.70
Al,O, loss, '% ~~
430 456 480 4 80 430 1100
300 320 340 345 345 3 50
0 .o 1.o
5.3 10 .o
14 .O 4 .O
Reaction with SOrCIz. In equilibrium with this system, FeCl2 may be produced under conditions where A1203 is unreactive. Exploratory measurements showed that rates of formation and removal of FeClz were relatively low, even at 880°C. By injection of small amounts of 0 2 into the reaction mixture, it was possible to enter the zone of stability of FeS04, in accordance with the predictions of the AG-minimization calculations. FeS04 formed rapidly at temperatures as low as 52OoC, but did not respond to chlorination. The alternate, removal by aqueous leaching, was outside the scope of the present investigation. Reaction with CO-CIz. The thermogravimetric study confirmed that excessive losses of A1203 were encountered under conditions where iron removal is efficient. Rates of iron removal were highest and A1203 losses were lowest after precalcination to 1100°C (Table IV). However, such treatment greatly impairs the reactivity of A1203 in subsequent conversion to AlC13. Conclusions Sulfidization of Fez03 to FeS followed by chlorination to FeC13 permits beneficiation of bauxite to final iron levels of 0.1 to 0.3%. High sulfidization rates are attained a t temperatures below 750°C with inexpensive S 0 + 2 0 reactants, and rapid removal of iron as FeC13 is attained by treatment with pure chlorine a t the same temperature. This route approaches the ideal combination of characteristics sought in a dry process for integration with AlC13 production more closely than any available hitherto. At the present stage of development, the iron levels are too high to permit use of the AlC13 as reduction cell feed without a further purification. Iron a t the residual levels can probably be eliminated economically by reduction with molten A1 or HZgas. The maximum reaction rates are attained with 35:65 ratio SO2/CO mixtures. Sulfur potentials a t this composition are predicted to be a t a maximum over a wide range of compositions by AG -minimization calculations for the complex gas equilibrium. Rate constants for the sulfidization have been found to be proportional to the empirical p(S2).p(COS)) of partial function (p(S2)z p(C0S)Z pressures of sulfidizing components. The sulfidization is best described by a rate equation developed for a mechanism where diffusion of reactants through the complex pore system of low-temperature dehydrated bauxite proceeds a t the same rate as chemical reaction a t the surface of a shrinking core of unreacted oxide. Marked size-dependence of the rate constants is found, down to 1 mm diameters. Below this size, penetration of reactant is such that reaction is proceeding throughout a significant fraction of the volume. The chlorination reaction proceeds in accordance with a chemically controlled, shrinking-core mechanism. The particle size-dependence decreases below 2 mm diameters,
+
+
Ind. Eng. Chem., Process Des. Dev.. Vol. 14. No. 4, 1975
451
Table V. Data for Experiments on Beneficiation of Bauxite Summarked in Figures I and 2 Sulf idization
Chlorination .-
Figure Particle reference diam, number mm 33 38 42 45 89 90 92 94 96 98 99 100 119 135 137 a
4.7 4.7 6.4 4.7 4.7 4.7 4.7 3.2 3.2 3.2 3.2 3.2 9.6 3.2 3.2
Temp, "C 720 720 580 720 740 740 780 780 690 696 783 778 607 700 700
Partial pressure, atm
co
coeff rate eq
0.49 0.20 0.34 0.34 0.22 0.13 0.34 0.25 0.13 0.40 0.26 0.26 0 -29 0.34 0.34
0.49 0.35 0.64 0.64 0.76 0.84 0.64 0.72 0.85 0.56 0.46 0.33 0.69 0.64 0.64
0.995 0.997 N.D. (' N.D. 0.999 0.997 0.990 0.985 0.999 0,920 0.998 0.986 N.D. x.1). N. n
0.98 0.98 0.98 0 "98 0.98 0.98 0.98 0.98 0.98 0.77 0.60 0.40 0.98 0.98 038
N.D. N.D. N.D. Y.D. 0.990 0.999 0.999 0,998 0,999 0.990 0.999 0.999 0.999 N.D. N.D
Resid. Fe, o/c
_--
0.50 0.30 0.10 0.10 0.20 0.40 0.50 0.29 0.40 0.70 0.47 0.70 0.68 0.17 0.28
N.D. = not determined.
again pointing to the existence of a relatively diffuse reaction zone. Chlorination of Fez03 using CO-Cl2, HC1, or HCI-CO gave much less favorable results than attained by reaction of FeS with Cl2. Under conditions where FeClz was produced, it was not as easy to remove from bauxite as experiments on pure Fez03 suggested. This was apparently due to blockage of the pore system by solid and liquid FeC12. The possibility of removing AIC13 with retention of FeS is unique to the H2S-Clz reactant system, but a clean separation with favorable kinetics was not attained. Predictions of conditions for formation of FeS04 using SOz-Cl2 with small additions of 0 2 were verified, bvt re moval of FeS04 could not be effected. However, under cvnditions where FeC12 is formed by reaction with S0.~-C12, reasonable iron elimination was achieved by direct volatilization a t the expense of using temperatures near 850OC.
Nomenclature K , K', K"' = pseudo rate constants, min-l
452
c or 1'e 1
SO2
.___I_.
_ I _
_ I
~
Ind. Eng. Cham., Process Des. Dev., Vol. 14, No. 4, 1975
~ ( S Y )etc , = partial pressures of sulfid.izing components, atrn t = reaction time, min X = fractional conversion of solid reactant F(X) = function of X in aulfidization rate equation F'(X) = derivative of F(X) with respect to X
Literatwe Cited Buntin, A., et ai., ti.. Tomsk. Goo. Univ. Ser. Khim., 191,52 (19823. Khundkar, M., Ahmad, N.. J. lndian Ch6m. Soc., 18, 109 (1955). LM,W., Trans. Met. Sac. 41ME,227,203 (1963). Milne, D. J., Holiday, R. D.,hd. Eng Chom, Process Des. Dei1....64, 442 (1975). Reeve, S.,J. iron SteelInsf. London, ?el, 26 (1955). Seth. 8..ROSS, ti., rrzns. et. SOC. AIME, 233. 180 (1965). Spitavin, K . , Z. .Anorg. A/@. Chem 189, 337 ('1930). Wen, C., Ind. Fng. Chsm , 60 (9). 34 (lW38).
Hecaiued ,for reuiew December 2, 1974 Accepted May 14,1975
authors wish to thank Comalco Ltd., 95 Collins St., Melhoume. for financial assistance toward this work. The
