The Effects of Iron, Silicon, Nickel and Copper on the Reaction of Aluminum Subchloride Process*

Transcription

1 The Effects of Iron, Silicon, Nickel and Copper on the Reaction of Aluminum Subchloride Process* By Takeaki Kikuchi** The effects of iron, silicon, nickel and copper on the reaction of aluminum subchloride process were investigated by the flow method of aluminum trichloride at reduced pressure and in the temperature range of In all cases, the extraction rate of aluminum decreased with increasing contents of alloying elements, and increased as the temperature and supplying rate of aluminum trichloride were raised. The extraction rate of aluminum could be expressed by the following equation including the activity of aluminum and the partial pressure of aluminum trichloride. The contents of silicon and iron in the deposited aluminum were about 0.2% and 0.05%, respectively, and a minute amounts of copper and nickel transfered to the products. In addition, the experimental results were discussed on the basis of thermodynamic calculation. (Received January ) I. Introduction Concerning the subehloride process of aluminum, a large number of reports have been published on equilibrium. However, there are a few reports on other important problems such as the effects of impurity metals involved in a crude alloy one the extraction rate of aluminum or transfer of the impurities into the refined aluminum. Tanabe et al.(1) and Mitani et al.(2) studied the above problems with an artificial alloy, and Yerkes et al.(3) examined the behavior of iron and silicon using the alloy produced by direct electric smelting of clay. In previous papers the present author previously reported the equilibrium(4) and reaction rate(5) of the subchloride reaction. This paper describes the effects of iron, silicon, nickel and copper on the reaction rate of aluminum. Generally, iron and silicon are the main elements related to this process, because these elements are derived from the aluminous raw materials in the course of the reduction. Furthermore, the effects of copper and nickel were also tested in the present work, because the addition of the materials containing these elements were efficient for carbothermic reduction (6) (7). Therefore, investigations were made on the effects of the above-mentioned elements, temperature and flow rate of aluminum trichloride on the extraction rate, and also on the purity of the deposited aluminum. II. Experimental Apparatus and Procedures 1. Experimental apparatus The experimental apparatus is essentially the same as * This paper was published originally in Japanese in J. Japan Inst. Metals, 35 (1971), 527. ** National Research Institute for Metals, Meguro-ku, Tokyo. Present address: Niihama Technical College, Niihama, Japan. (1) I. Tanabe, H. Konno, Y. Sawada and T. Takahashi: Kagaku, 32 (1964), 285. Denki used in the measurement of the reaction rate between pure aluminum and aluminum trichloride(5), and it consists of a trichloride evaporator, a reaction tube, a trichloride condenser and a vacuum pump. The reaction tube was a silica pipe (40mm in inner diameter, 800mm in length), and a high-alumina pipe (35mm in inner diameter, 600mm in length) was inserted in the reaction tube as a place for aluminum deposition. A Siliconit furnace with an opening and shuting device was used, and the temperature was maintained within The temperature was measured by a calibrated Pt- Pt.Rh thermocouple located on the outside of the reaction tube just above the position of the alloy, but no temperature difference was found between the inside and the outside of the tube. Aluminum trichloride of chemical grade was used after distillation refining. Aluminum trichloride unreacted and that decomposed due to the subchloride reaction were condensed by liquid nitrogen. The aluminum alloys with a desired composition were prepared in vacuum by means of the high-frequency furnace. The compositions of the alloys were 9.98 and 5.01% Si in the Al-Si system, 10.0 and 2.97% Fe in the Al-Fe, system and 9.97% Ni in the Al-Ni system, and and 9.91% Cu in the Al-Cu system. The aluminum used was 99.99% in purity and the silicon was a very pure material prepared by the hydrogen reduction of trichlorosilane. The iron was a electrolytic grade (99.9%), the nickel was carbonyl nickel (99.9%) and the copper was a electrolytic grade. (2) H. Mitani and H. Nagai: J. Japan Inst. Metals, 34 (1970), 752. (3) L. A. Yerkes and O. C. Fursman: U. S. Bureau of Mines, Rep. Inv., 5773 (1961) (4) T. Kikuchi, T. Kurosawa and T. Yagihashi: J. Japan Inst. Metals, 28 (1964), 9. (5) T. Kikuchi, T. Kurosawa and T. Yagihashi: J. Japan Inst. Metals, 33 (1969), 305. (6) A. Schneider and O. Hihner: Metall, 3 (1960), 186. (7) T. Kikuchi, T. Kurosawa and T. Yagihashi: J. Japan Inst. Metals, 34 (1970), 643. Trans. JIM 1972 Vol.13

2 360 The Effects of Iron, Silicon, Nickel and Copper on the Reaction of Aluminum Subchloride Process 2. Experimental method Aluminum trichloride was charged quickly into the evaporator and weighed out. The evaporator was then connected with the reaction tube by a ball joint. The boat used for melting alloy was a high-alumina sectional view, its bottom was semicircular to fit the inserted pipe, and the other, about two-thirds of the depth was rectangular to maintain a constant surface area in spite of the decrease of aluminum due to the reaction. Prior to the reaction, the boat containing the aluminum alloy was placed in the uniform temperature zone. Then the evaporator and the reaction tube were connected with each other and then evacuated. When the reactor was heated to a desired temperature, a furnace was set in the evaporator to supply aluminum trichloride gas. The vapor pressure of trichloride was measured every five minutes by a silicone-oil manometer set between the evaporator and the reaction tube. The relation between the vapor pressure of aluminum trichloride and its flow rate has been reported (5). The refined aluminum produced by the subchloride reaction deposited many small globular particles in the hot side of the inserted tube, while it became dendritic crystals in the low temperature zone. Aluminum trichloride effusing from the reaction tube was trapped completely in the condenser. After the reaction the apparatus was cooled rapidly, and the reaction ratios of aluminum and aluminum trichloride were estimated in terms of the weight loss of the alloy and the evaporated amount of trichloride. The contents of iron, silicon, nickel and copper in the deposited aluminum were determined by chemical analysis. The experimental temperature was 900, 1000 and Moreover, the difference in the concentration of alloying metals could not be found between the surface and the inside of the remaining alloy by means of an electron probe micro-analyser. As shown in these figures, iron delayed the extraction rate remarkably in comparison with other metals at the same concentration. In addition, almost equal rates Fig. 1 Relation between flow rate of AlCl3 and reaction rate of Al in Al-Si alloys. Fig. 2 Relation between flow rate of AlO3 and reaction rate of Al in Al-Fe alloys. III. Experimental Results and Discussion Figures 1 to 4 indicate the reaction ratios of aluminum (g/cm2.min) in the Al-Si, Al-Fe, Al-Cu and Al- Ni alloys as functions of the flow rate and partial pressure of aluminum trichloride. As these figures show, the reaction ratios increase as the flow rate of trichloride and the reaction temperature are raised. Furthermore, the rates decrease with increasing concentrations of additional elements. Figures 5 to 7 show a comparison of the reaction rate under the conditions where trichloride flow rates were 0.2, 0.4 and 0.6g/min with respect to the molar concentrations of the alloying metals. In these figures, the extrapolated points to zero concentration in each system almost coincided with the value calculated from the rate equation for pure aluminum(5). Fig. 3 Relation between flow rate of AlCl3 and reaction rate of Al in Al-Cu alloys.

3 Takeaki Kikuchi 361 Fig. 4 Relation between flow rate of AlCl3 and reaction rate of Al in Al-Ni alloys. Fig. 6 Comparison of reaction rate in the Al-Si, Al-Fe, Fig. 5 Comparison of reaction rate in the Al-Si, Al-Fe, were found among silicon, copper and nickel, at 900 and rates were smaller than the values on the broken line which was the product of the reaction rate of pure aluminum by its molar concentration. As described previously(5), the rate of the subchloride reaction 2Al(l)+AlCl3(g)=3AlCl(g) in the pure aluminum system can be written as follows. where A is the frequency factor, m and n are the reac- or partial pressure of each substance. The above equation was obtained from the conditions involving a constant surface area of aluminum, a small pressure, i.e. a low concentration of aluminum trichloride, and surfacereaction control. Fig. 7 Comparison of reaction rate in the Al-Si, Al-Fe, In the alloyed system, the subchloride reaction 2Al (in liq. alloy)+alcl3(g)=3alcl(g) occurs at the interface between melted alloy and gas. Thus it may involve the following processes; (1) aluminum trichloride gas reaches the surface of melt; (2) chemical reaction ocuurs at the interface of melt and gas; (3) aluminum in the melt transfers to the interface; (4) AlCl gas removes from the interface. In the present work, as the reaction takes place under the conditions of low pressure and fast flow of reacting gas, the processes (1) and (4) may not be considered as a rate-determining step(5).

4 362 The Effects of Iron, Silicon, Nickel and Copper on the Reaction of Aluminum Subehloride Process Activation energy of the reaction rate is occationally useful for discussing the rate-determining step. Activation energies calculated from the Arrhenius equation were 12.9, 12.2, 12.1 and 11.9kcal/mol in the Al-Si almost coincided with the value of 11.8kcal/mol obtained in the pure aluminum system, and it is large in comparison with the value of several kcal/mol which is usually found in the diffusion control. As the chemical reaction on the melt surface can be considered as a rate-determining step and the concentration of the alloying metals is relatively small, the rate equation can be written by using the activation energy of the pure aluminum system. Therefore, the transfer of aluminum of the melted alloy to the interface between melt and gas, process (3), is not a rate-determining step because of the mutual diffusion of aluminum and alloying metals as well as convection of the melt. This agrees with the fact that the difference in concentration could not be found in the remaining alloy by means of a micro-analyser. considered to be equal in all systems because of the same mechanism of the subchloride reaction. It was difficult to get a proper explanation from Figs. 5 to 7 representing the concentration of alloy in mol fraction. Hence, the activity of aluminum was introduced instead of the molar concentration in the same way as in a kinetical study of desulphurization by Peters et al.(8), in which the rate is dexpressed by the activity of sulphur in the Fe-S melt. Figures 8 and 9 show the reaction rate of aluminum as a function of the activity of aluminum under the conditions of 0.2 and 0.4g AlCl3/min. In these figures, the dotted lines represented by m=2 are the rates given Fig. 8 Relation between reaction velocity of Al in the alloys and aluminum activity. (Flow rate of AlCl3, 0.2g/min) (8) R. J. W. Peters, C. R. Masson and S. G. Whiteway: Trans. Faraday Soc., 61 (1965), (9) Z. Kozuka, S. Yasugi, W. Nakajima and J. Moriyama: Denki Kagaku (in Japanese), 34 (1966), 35. (10) S. Oyama, T. Matsushima and K. Ono: Bulletin of the Research Institute of Mineral Dressing and Metallurgy (Senkoseiren Kenkyujo Hokoku, in Japanese), 21 (1965), 23. Fig. 9 Relation between reaction velocity of Al in the alloys and aluminum activity. (Flow rate of AlCl3, 0.4g/min) by the product of the rate in the pure aluminum system by the square of aluminum activity aal. The activities of aluminum were quoted from the values by Kozuka and Moriyama(9) in the Al-Si system, by Matsushima and Ono(10) in the Al-Cu system, by Coskin and Elliott(11) and Mitani and Nagai(12) in the Al-Fe system, and by Jeannerod(13) in the Al-Ni system. When no measured data are available within the experimental temperature and concentration ranges investigated, the extrapolated values were used. The the Al-Ni system. As shown in Figs. 8 and 9, a better result was obtained as expressed in terms of the aluminum activity than the case shown by the molar concentration, and the order m=2 was more suitable than m=1 or 3. Therefore, the rate equation can be expressed as follows: Figure 10 shows the reaction ratio of aluminum trichloride in the Al-Ni system. This figure indicates that the fractional reaction increases at a higher temperature and a lower concentration of nickel, and it decreases as the flow rate is raised. Furthermore, a similar tendency was also found in the other systems. Figure 11 indicates the purity of the deposited aluminum obtained from the Al-Fe and Al-Si systems. The contents of iron and silicon were mostly below 0.1 and 0.2%, respectively. The silicon content was usually higher than the iron content, and showed a tendency to increase with temperature. Furthermore, the copper content was below 0.001% and nickel was scarcely detected in the deposited aluminum. The mechanism of contamination due to impurities is not so clear, but it can be discussed on the assumption that it is caused only by the reaction without (11) A. Coskin and J. F. Elliott: Trans. Met. Soc. RIME, 242 (1968), 253. (12) H. Mitani and H. Nagai: J. Japan Inst. Metals, 32 (1968), 752. (13) M. Jeannerod: The papers for the degree of master of science at MIT, (1966).

5 Takeaki Kikuchi 363 Fig. 10 Relation between flow rate of AlCl3 and reaction ratio of AlCl3. Fig. 12 Calculated concentrations of elements in the deposited Al as a function of reaction temperature and total pressure. Fig. 11 Relation between reaction temperature and Si, Fe percentage in deposited Al. mechanical carryover. As described by Kitagawa(14) and Kozuka(9), iron, silicon, copper and nickel in the alloys may be assumed to react with aluminum trichloride as follows: Silicon tetrachloride may be formed, but it is probably very minute as discussed by Hirschwald(15). The result of calculation showed that the equilibrium partial pressure of AlCl in the reaction 2Al(l)+AlCl3(g)=3 AlCl(g) is very large in comparison with those of the other metal chlorides. Thus, it can be considered that the contamination is caused by the reverse process of the above reactions. (14) J. Kitagawa: Keikinzoku (in Japanese), 11 (1964), 59. (15) W. Hirschwald and O. Knake: Erzmetall, 3 (1958), 99. Figure 12 shows the calculated concentration of alloying metals in the deposited aluminum under total pressures of 0.1, 0.01 and atm. In this calculation, the thermodynamic data were quoted from the author's paper(4) and Kubaschewski's literature(16), and the activity of each metal was assumed to be unity. Figure 12 shows that nickel hardly transfer and copper, silicon, and iron follow in its order. Moreover, impurities transfer easily at higher temperature and larger pressure, and the temperature dependence is in the decreasing order of nickel, iron and silicon. Furthermore, the level of impurities is less than about 10-3%. A fairly good agreement was found between the experimental results and the tendency of this graph, for example, a small percentage of nickel in the deposit, ready contamination due to silicon and iron and larger temperature dependence of silicon. However, the contents of silicon and iron were large contrary to the estimated value of about 10-3%. This result is probably caused by the different reactivity of silicon and iron with trichloride(17), and a complex mechanism of contamination in the deposition part. Moreover, it may be caused by the marked subsidiary reaction between FeO, SiO2 and Cu2O in the reactor materials and AlCl gas as reported by Mitani and Nagai(2). The problem concerning the easy contamination due to silicon can be considered as follows. When the activities of iron and silicon are used, the equilibrium constants become (16) O. Kubaschewski, E. LL. Evans and C. B. Alcock: Metallurgical Thermochemistry, Pergamon Press, London, (1967). (17) M. F. Lee: J. Phys. Chem., 62 (1958), 877.

6 364 The Effects of Iron, Silicon, Nickel and Copper on the Reaction of Aluminum Subchloride Process Therefore, if the activity of an alloying metal is small, the partial pressure of the chloride becomes small and the contamination will be reduced. Referring the activity of iron to the value obtained by Pehlke(18), Mitani and Nagai(19) and the activity of silicon to the values by Kozuka, Moriyama(9), and Mitani and Nagai(19), in the concentration range of the present work, as, is more than ten times as large as are and PSiCl2 is expected to become larger than PFeCl2. Consequently, a larger percentage of silicon appeared to be introduced in the product. IV. Summary The effects of alloying elements, silicon, iron, copper and nickel on the rate of the aluminum subchloride (18) R. D. Pehlke: Trans. Met. Soc. AIME, 212 (1958), 486. (19) H. Mitani and H. Nagai: J. Japan Inst. Metals, 31 (1967), (1) The reaction rate of aluminum increased as the temperature and flow rate of aluminum trichloride were raised, while it decreased with increase in the concentration of alloying elements. (2) The reaction rate of aluminum in the alloy could be expressed by a combination of the activity of aluminum and the rate equation obtained in the pure aluminum system. (3) The silicon content in the deposited aluminum was mostly less than about 0.2% and the iron content was about 0.05%. In general, silicon was abundant in comparison with iron. Copper was below 0.001% and nickel was scarcely detected. (4) The problem concerning the impurites in the deposit was discussed on the basis of the thermodynamic calculation, and a fairly good agreement could be found. Acknowledgment This work was carried out during the period in which the author served at the National Research Institute for Metals. The author wishes to thank Dr. T. Yagihashi and Dr. T. Kurosawa for their kind discussion.