Phase Equilibrium between Ni S Melt and CaO Al 2 O 3 Based Slag in CO CO 2 SO 2 Gas Mixtures at 1773 K

Transcription

1 Materials Transactions, Vol. 43, No. 11 () pp. 873 to 879 c The Japan Institute of Metals Phase Equilibrium between Ni S Melt and CaO Al O 3 Based Slag in CO CO SO Gas Mixtures at 1773 K Hector M. Henao, Mitsuhisa Hino and Kimio Itagaki Institute of Multidisciplinary Research for Advanced Materials, Tohoku University. Sendai , Japan To provide thermodynamic data for converting the nickel matte to liquid nickel, an experimental study was conducted in the phase equilibrium between the Ni S alloy and the CaO Al O 3 based slag melted in a MgO or Al O 3 crucible at 1773 K under controlled P SO of 1.1 kpa and P O in a range between and 1.6 Pa by using CO CO SO gas mixtures. The contents of nickel in these slags at P O of 1.6 Pa, above which solid NiO is expected to precipitate in equilibrium state, were found to be 5.6 and 7. mass% for the slags melted in the MgO and Al O 3 crucibles, respectively. The contents of sulfur and oxygen in the nickel melt at P O of 1.6 Pa were and.4 mass%, respectively. The CaO Al O 3 based slag with the mass% ratio (CaO/Al O 3 ) of about 1 melted in the MgO crucible was homogeneous in the investigated P O range. While, that with the ratio of about.5 melted in the Al O 3 crucible formed two phases composed of a liquid and an Al O 3 NiO solid solution whose nickel content was increased with increasing P O. The dissolutions of nickel and sulfur in the slag melted in the MgO crucible were discussed on the basis of distribution ratio and sulfide capacity, respectively. (Received August 8, ; Accepted September 5, ) Keywords: nickel smelting, nickel refining, phase equilibrium, distribution ratio, activity coefficient, CaO Al O 3 based slag, nickel matte, nickel converting, magnesia crucible, alumina crucible 1. Introduction The pyrometallurgical process of nickel sulfide ores consists of two stages of smelting and converting. The concentrates are oxidized in the smelting stage to produce the Ni Cu Fe S matte, which is further converted to the Ni Cu S matte. The converting matte is separated into the sulfides of Ni 3 S and Cu S by the controlled slow cooling method. Mostly, the Ni 3 S is converted to pure nickel or a nickel compound by hydrometallurgical processes. The alternative pyrometallurgical route to convert Ni 3 S to liquid nickel was employed in a process developed by INCO, 1) in which oxygen was injected in the top blown rotary converter (TBRC), or a top submerged lancing process developed by Ausmelt. ) To discuss the pyrometallurgical route for the conversion of Ni 3 S to liquid nickel from the standpoint of thermodynamics, the equilibrium studies for the Ni S melt and the slag or the flux are of practical importance. Hence, following the experimental studies 3 5) for the Ni Fe S or Ni Cu Fe S matte/slag equilibration, the phase equilibrium between the Ni S melt/iron silicate or calcium ferrite based slag has been investigated in the previous study. 6) It is considered that a study for an iron-free slag or flux will be more practical because the iron content in the nickel sulfide produced by the controlled slow cooling method is considered small at less than 1 mass%. A CaO Al O 3 based system may be available as the iron-free slag or flux, which has been used for refining steel 7) as well as for deoxidation and denitrogenization of liquid nickel. 8) Nevertheless, no data have been hitherto reported on the equilibrium between the CaO Al O 3 based slag and the Ni S melt. Hence, the equilibrium between both phases in a MgO or Al O 3 crucible was investigated at 1773 K in the present study. The experiments were made under the partial pressure Doctor Student, Tohoku University. Corresponding author: henaocol@iamp.tohoku.ac.jp of SO controlled at 1.1 kpa, and the partial pressures of O and S were also controlled in the ranges Pa and kpa, respectively, by using CO CO SO gas mixtures.. Experimental The experimental set-up and procedure have been presented in the previous paper. 6) The salient features are as follows. CO CO SO gas mixtures were used to control the partial pressures of SO,O and S. The flow rate was regulated at m 3 /s by using capillary flow meters. The partial pressure of SO was fixed at 1.1 kpa and those of O and S were varied by changing the CO/CO ratio in the gas mixtures. These were calculated by using a ChemSage Ver software (GTT-Technologies, Aachen, Germany). A furnace assembly consisted of a silicon carbide heating element and an alumina reaction tube. Temperature of a sample was controlled at 1773 K within K by a SCR controller with a Pt/Pt Rh thermocouple. About 6 g of pre-melted slag were equilibrated with about 6 g of the total amounts of Ni 3 S and Ni with a Ni/S molar ratio close to a predicted equilibrium composition to achieve the equilibration in a short time. In the previous paper, 6) it was clarified that the equilibrium in the alloy phase can be made in a restricted time of less than 1 ks by adjusting the sulfur content of starting alloy phase so that it may be a little smaller than the equilibrated one. In the present study, the holding time was set at 19.6 ks and the sulfur content of starting Ni S melt was adjusted to be about mass% smaller than the pre-estimated equilibrium value reported by Meyer et al. 9) The compositions of premelted starting CaO Al O 3 slag melted in the MgO crucible were 49 mass%cao and 51 mass%al O 3, while those of the CaO Al O 3 slag melted in the Al O 3 crucible were 35. mass%cao,

2 874 H. M. Henao, M. Hino and K. Itagaki 6.3 mass%al O 3 and 4.5 mass%mgo. The sample was put in the MgO or Al O 3 crucible of m i.d and 6 1 m height. After equilibration, the sample was taken out of the furnace and quenched in a jet stream of argon. The slag and alloy specimens were analyzed by the inductively coupled plasma spectrometry (ICP). For the chemical analysis of CaO and Al O 3, the slag was preliminarily fused with NaNO 3 to make it amenable to the acid dissolution. The total oxygen content in the alloy specimen was determined by inert gas fusion-infrared absorptiometry (EMGA- 65, HORIBA Co, Ltd., Kyoto, Japan), while the total sulfur content in the slag specimen by a combustion-infrared spectrometer (EMIA-58, HORIBA Co, Ltd., Kyoto, JAPAN). 3. Results Compositions of the slag and alloy phases after equilibration are listed in Tables 1 and for the CaO Al O 3 slag melted in the MgO crucible and the Al O 3 crucible, respectively. The summation of the analytical values for both slag and alloy are a little more or less than 1 mass% due to the error of the analysis. The partial pressures of oxygen and sulfur are expressed as dimensionless ones, defined in the cases of oxygen by p O = (P O )/(1135 Pa) and sulfur by p S = (P S )/(1135 Pa). 3.1 Alloy phase The relationship between the sulfur content in the alloy and log p S or log p O for both slag systems is shown in Fig. 1. The limits of p O and p S, between which the experiments were carried out, are indicated with the dash-dot lines. The right limit corresponds to the formation of Ni 3 S in the alloy phase, while the left limit to log p O of 4.8 where the precipitation of solid NiO is anticipated. Meyer et al. 9) determined p S in relation to the sulfur content in the Ni S melt by equilibrating with H H S gas mixtures at 1773 K, which is shown with a dashed line in the Fig. 1. It is found that the present results agree fairly well with the data by Meyer et al. 9) within scatter of the data. As shown in Fig., the oxygen content in the Ni S melt increases very gently with increasing log p O and it ranges between.1 and.4 mass% at log p O of 7.3 and 4.9, respectively. Within scattering of data, no obvious difference in the content of sulfur or oxygen in the alloy is observed between the slags melted in the MgO and Al O 3 crucibles. Table 1 Compositions of CaO Al O 3 based slag and nickel sulfur alloy melted in MgO crucible at 1773 K. Slag composition (mass%) Alloy composition (mass%) No log p S log p O Ni Al O 3 CaO MgO S γ NiO Ni S O Table Compositions of CaO Al O 3 based slag and nickel sulfur alloy melted in Al O 3 crucible at 1773 K. Slag composition (mass%) Alloy composition (mass%) No log p S log p O Ni Al O 3 CaO MgO Ni S O Composed of the liquid and solid phases

3 Phase Equilibrium between Ni S Melt and CaO Al O 3 Based Slag in CO CO SO Gas Mixtures at 1773 K 875 MgO crucible Al O 3 crucible 4 spinel liquid periclase liquid CaO Al O 3 liquid liquid lime liquid mass (MgO NiO) Al O 3 mass Al O 3 CaO Fig. 3 Compositions of slags equilibrated with Ni S alloy melted in MgO or Al O 3 crucible at 1773 K. Fig. 1 Relationship between sulfur content in Ni S melt and log p S or log p O in equilibrium with CaO Al O 3 based slags melted in MgO or Al O 3 crucible at 1773 K. mass O in Alloy.6.4. MgO crucible Al O crucible 3 log p O Fig. Relationship between dissolution of oxygen in alloy and log p O for CaO Al O 3 based slags melted in MgO or Al O 3 crucible at 1773 K. 3. Slag phase 3..1 Compositions The compositions of slags melted in the MgO and Al O 3 crucibles are reproduced in Fig. 3 on the (MgO + NiO) Al O 3 CaO ternary diagram. The phase boundaries in the MgO Al O 3 CaO system without NiO at 1773 K are shown in Fig. 3 with the dashed lines, which are reported in a reference. 1) The compositions of starting slags are shown with filled circle ( ) and filled box ( ) for the experiments with the MgO and Al O 3 crucibles, respectively. According to Fig. 3, the slag melted in the MgO crucible would be equilibrated not with solid MgO but with a (MgO, Al O 3, CaO) solid solution of periclase if NiO were not involved in the slag system. While, the slag melted in the Al O 3 crucible would be equilibrated with solid MgAl O 4 spinel and CaAl O 4. A microphotographic investigation for the specimens with the MgO crucible clarified that no distinct layer corresponding to solid periclase was formed at the interface between the slag and the MgO crucible. Instead, the EPMA analysis showed that Ni diffused in the MgO crucible with the concentration gradient, as typically shown in Fig. 4(A). No diffusion in the MgO crucible was observed for Ca or Al. As for the slag, precipitation of solid particles was not found in the solidified phase. This result shows that the CaO Al O 3 based slag melted in the MgO crucible was homogeneous at 1773 K. No distinct layer corresponding to the spinel or CaAl O 4 was found at an interface between the slag and the Al O 3 crucible. Ca and Ni were found to diffuse into the Al O 3 crucible with the concentration gradients, as typically shown for Ca and Ni in Figs. 4(B) and (C), respectively. It is noted that, as shown in Fig. 5, precipitation of solid particles was found in the slag phase equilibrated with the Ni S melt under the entire range of p O investigated in the present study. This result shows that the CaO Al O 3 based slag melted in the Al O 3 crucible was inhomogeneous, composing of liquid and solid phases at 1773 K. The EPMA analysis in the precipitated particles clarified that it is a solid solution composing mainly of Al O 3 and NiO and that the content of NiO increases with increasing p O, as shown in Fig. 6. Once it was recognized that the Al O 3 CaO based slag melted in the Al O 3 crucible was inhomogeneous, special care has to be paid so that the sample for a chemical analysis might be taken from the whole part of the solidified slag and thoroughly mixed after grinding the sample. 3.. Nickel content The nickel contents in the CaO Al O 3 based slag equilibrated with the Ni S alloy in the MgO crucible or Al O 3 crucible are plotted in Fig. 7, in relation to log p O. It is shown that the nickel content in the CaO Al O 3 based slag melted in the Al O 3 crucible increases with increasing p O from.3 mass% at log p O of 7.3 to 5.7 mass% at log p O of 5.. It is noted that the nickel content in the CaO Al O 3 based slag equilibrated in the Al O 3 crucible is a little larger than that in the MgO crucible. This difference can be ascribed to the higher content of nickel dissolved in the solid solution reported in the CaO Al O 3 based slag melted in the Al O 3 crucible. It is also observed in Fig. 7 that the content of nickel for the CaO Al O 3 based slag, at a given p O, is smaller than that for the FeO X CaO or FeO X SiO based slag. 6) 3..3 Sulfur content The sulfur contents in the CaO Al O 3 based slags are shown in Fig. 8, in relation to log p S. It is noted that the sulfur content in the CaO Al O 3 based slag melted in the Al O 3 crucible increases abruptly in the range of higher log p S and

4 876 H. M. Henao, M. Hino and K. Itagaki Fig. 5 Microphotography showing NiO Al O 3 solid solution particles precipitated in CaO Al O 3 based slag melted in Al O 3 crucible at 1773 K. mass Ni in solid solution 3 1 log p O Fig. 4 Microphotography showing diffusion of Ni and Ca in MgO or Al O 3 crucible at 1773 K. (A) Ni intensity spectra in the MgO crucible, (B) Ca intensity spectra in the Al O 3 crucible, (C) Ni intensity spectra in the Al O 3 crucible. becomes 1.4 mass% at log p S of Discussion 4.1 Dissolution of nickel in slag The dissolution of nickel and the activity of nickel oxide in the CaO Al O 3 slag melted in the MgO crucible will be discussed on the basis of the distribution ratio of nickel between Fig. 6 Relationship between content of NiO and log p O for NiO Al O 3 solid solution particles precipitated in CaO Al O 3 based slag melted in Al O 3 crucible at 1773 K. the slag and alloy phases, which is defined by eq. (1). L s/ni Ni = (mass%ni in slag)/{mass%ni in alloy} (1) When nickel is dissolved in the slag as an oxide, the distribution ratio can be analyzed thermodynamically on the basis of the following reaction to form a mono-metallic oxide. Ni(l) + v/4o (g) = NiO v/ (s) ()

5 Phase Equilibrium between Ni S Melt and CaO Al O 3 Based Slag in CO CO SO Gas Mixtures at 1773 K 877 mass Ni in Slag CaO-Al O (MgO crucible) 3 CaO-Al O (Al O crucible) 3 3 FeO X-SiO -MgO FeO X-CaO-MgO log p O s/ni L Ni.1.1 NiO.1 log p O 1 Fig. 7 Relationship between content of nickel and log p O for CaO Al O 3 slags melted in MgO or Al O 3 crucible at 1773 K. Fig. 9 Relationship between distribution ratio of nickel and log p O for CaO Al O 3 based slag melted in MgO crucible at 1773 K. mass S in Slag log p S ity obtained in the present study, is plotted in Fig. 9 in relation to log p O. In the lower pressure range, a nearly linear relationship is observed between log L s/ni Ni and log p O with a gradient of about 1/. This suggests that nickel dissolves in this slag as NiO, according to eq. () with v =. The linear relationship observed in Fig. 9 also suggests that the sum of activity coefficient ratio and n T ratio in eq. (4) does not change appreciably in the range of log p O < 5.5. However, in a range of higher log p O above 5.5, log L s/ni Ni log p O plots make a line with a gradient considerably smaller than the expected value of 1/. This is considered to be ascribed to the change in the summation of both the ratios in eq. (4) against p O, mainly due to the large content of NiO in the slag, that may also result in the change of γ NiO. The Raoultian activity coefficient of NiO can be calculated on the basis of eq. (5), by using the eq. (3). Ni(l) + 1/O (g) = NiO(s) (5) Fig. 8 Relationship between content of sulfur and log p S for CaO Al O 3 based slag melted in MgO crucible at 1773 K. Where, the solid NiO v/ was taken as a reference of the NiO v/ activity for the lack in the thermodynamic data of liquid NiO v/. The equilibrium constant of eq. () is given by K = a NiOv/ /(a Ni p v/4 O ) (3) where a Ni and a NiOv/ are activities of Ni and NiO v/, respectively. By combining eqs. (1) and (3) and converting the mole fraction, N, in the activity relationship of a = Nγ with the Raoultian activity coefficient of γ into mass%, the following equation is obtained. log L s/ni Ni = v/4 log p O + log[{γ Ni }/(γ NiOv/ )] (4) + log[(n T )/{n T }] + log K where, ( ) and { }denote the slag and alloy phases, respectively. n T is the total number of moles in 1 g of each phase, which is calculated on the mono-cation base. The L s/ni Ni for the CaO Al O 3 slag melted in the Al O 3 crucible, which was obtained from the data on the nickel solubil- G /J mol 1 = T/K The standard Gibbs energy change and equilibrium constant of eq. (5) are given in a reference. 1) The Raoultian activity coefficient of nickel in the binary Ni S alloy, {γ Ni }, is obtained from the data reported by Larrain. 11) The calculated γ NiO is listed in Table 1. γ NiO is found to range between 5 and 1. These values are larger than those for the FeO X CaO based slag melted in the MgO crucible, which range between 3 and 5. 6) This trend may be ascribed mainly to a chemical property of NiO. NiO and Al O 3 make a compound of NiO Al O 3 with the Gibbs energy change of reaction of 6.9 kj/mol of cation at 1773 K, 1) that is less negative than 9.1 kj/mol of cation for NiO Fe O 3. 13) On the other hand, the chemical affinity between CaO and Al O 3 is considered to be very large with a Gibbs energy change of reaction of 16.8 kj/mol of cation at 1773 K. 14) Takeda and Yazawa 15) suggested that a solution tends to repel the component which has a low affinity with the main components when these have a considerably high affinity to form the stable solution. The observed trend for γ NiO is in concordance with this suggestion.

6 878 H. M. Henao, M. Hino and K. Itagaki C S MgO crucible FeO X -SiO -MgO FeO -CaO-MgO X.1 log p O Fig. 1 Relationship between sulfide capacity and log p O for CaO Al O 3 based slag melted in MgO crucible at 1773 K. 4. Sulfide capacity When considering the dissolution of sulfur in the slag, two important chemical equilibria are represented by the following equations: 1/S (g) + (O ) = 1/O (g) + (S ) (6) 1/S (g) + 3/O (g) + (O ) = (SO 4 ) (7) and it is known that eq. (6) prevails against eq. (7) at lower oxygen pressures. On the basis of eq. (6), a sulfide capacity is defined by eq. (8). C S = (mass%s)(p O /p S ) 1/ = K 6 a O /γ S (8) Where K 6 is the equilibrium constant of eq. (6), and a O and γ S are activity of O and activity coefficient of S in the slag, respectively. C S for the present CaO Al O 3 based slag was calculated from the sulfur content and partial pressures of O and S determined in the present study and are shown in the Fig. 1, in relation to log p O. It is noted that C S increases with increasing p O. This suggests that (a O /γ S ) in eq. (8) is dependent on p O, provided that eq. (6) is applicable to the dissolution of sulfur in the present slag. It is shown in Fig. 7 that the Ni content in the slag remarkably varies with the oxygen pressure. This variation may be ascribed to the change in the (a O /γ S ) against p O. It is possible to make another interpretation for the change of C S against p O. It is reported by Sano 17) that C S for calcium ferrite and calcium-alumino-silicate melts and the dissolution of sulfur decrease with increasing p O and are minimum at the p O between log p O of about 4 and 5. With further increasing p O, the dissolution of sulfur as sulfate, as given by eq. (7), is prevailing. C S increase noticeably because the oxygen pressures in the present study are also considerably high at log p O between about 7 and 5. Hence, there might be a possibility that these oxygen pressures are to be in the transitional region between the sulfide and sulfate dissolution of sulfur in the slags. As shown in Fig. 1, at a given p O, C S for the CaO Al O 3 based slag and FeO X SiO is considerably lower than that for the FeO X CaO slag. This is considered to be ascribed to the more basic property of FeO X CaO based slag with larger a O. A direct comparison of the present experimental data with the previous investigations for the sulfide capacity of CaO Al O 3 based slags is difficult due to the different slag composition, temperatures and range of p O investigated. Nzotta 18) reported C S with the values ranging between and for the CaO Al O 3 MgO slag at (p S /p O ) 1/ of at 1773 K. These values are very close to the present results ranging between and at (p S /p O ) 1/ between 13 and 3 (log p O between 7. and 6.9). 4.3 Application to nickel smelting It was found in the present study that, at log p O of 5, that is just prior to precipitation of solid NiO, the compositions of alloys equilibrated with the CaO Al O 3 slags (fluxes) in the MgO and Al O 3 crucibles at 1773 K are Ni: 97.6, S:, and O:.4 mass% (the analytical results in Tables 1 and are normalized to 1 mass%). On the other hand, the contents of nickel in the slags are 5.6 and 7. mass% for the slag melted in the MgO and Al O 3 crucibles, respectively. Although the nickel contents in the alloys are relatively large at more than 98 mass%, considerable amounts of sulfur and oxygen are contained in the alloys as the impurities. Hence, when a process of smelting the Ni S alloy with the CaO Al O 3 based slags (fluxes) is considered, a refining process for the metal product will be indispensable. On the other hand, it is to be noted that the content of nickel in the slag is not so high. This means that the loss of nickel in the slag can be kept down at a desired level if the amount of produced slag or used flux in the process is kept small. It is found in the present study that the solid solution was precipitated in the CaO Al O 3 based slag melted in the Al O 3 crucible at 1773 K. The use of this heterogeneous slag for the nickel smelting at 1773 K may cause a build-up problem with the accretion in the furnace or result in the enhancement of slag viscosity, both will make the smelting operation seriously difficult. Hence, the use of Al O 3 based refractory or the slag containing more than 6 mass%al O 3 has to be avoided. The present result suggests that the CaO Al O 3 based slag melted in the MgO crucible will be useful for converting the Ni S melt to liquid nickel. The distribution behavior of some minor elements between this slag and the Ni S melt will be reported in a separated paper. 5. Conclusion As a fundamental study of smelting the nickel sulfide to produce liquid nickel, experiments were made on the phase equilibrium between the Ni S melt and the CaO Al O 3 based slags in a MgO or Al O 3 crucible at 1773 K under controlled P SO at 1.1 kpa and P O in a range between and 1.6 Pa. It was found that the content of sulfur in the alloy decreased with increasing oxygen pressure (decreasing sulfur pressure) while the content of nickel increased. It was clarified that the content of oxygen in the alloy equilibrated with the CaO Al O 3 based slag changed very slightly with the

7 Phase Equilibrium between Ni S Melt and CaO Al O 3 Based Slag in CO CO SO Gas Mixtures at 1773 K 879 oxygen pressure. The content of nickel in the CaO Al O 3 slag was found to increase with increasing oxygen pressure and reach 6 mass% at log P O of 4.8 above which precipitation of solid NiO is anticipated. The content of sulfur in the CaO Al O 3 based slags decreased with increasing p O. The CaO Al O 3 melt in the MgO crucible was found to form a homogeneous liquid slag phase. On the contrary, the CaO Al O 3 melt in the Al O 3 crucible coexisted with a solid solution of Al O 3 NiO. REFERENCES 1) P. Queneau, C. E. O Neill, A. Illis and J. S. Warner: J. Met. July (1969) ) K. R. Robilliard, R. W. Matusewicz and B. R. Baldock: Proceeding of 5th CIM conference, (CIM, Toronto, Canada, 1994) paper No ) J. M. Font, M. Hino and K. Itagaki: Mater. Trans., JIM 39 (1998) ) J. M. Font, M. Hino and K. Itagaki: Mater. Trans., JIM 4 (1999) 6. 5) J. M. Font, M. Hino and K. Itagaki: Metall. Mater. Trans. 31B () ) H. M. Henao, M. Hino and K. Itagaki: Mater. Trans. 43 () ) S. Ban-Ya, M. Hino, A. Satoh and O. Terayama: Tetsu-to-Hagane 77 (1991) ) S. W. Cho and H. Suito: ISIJ Int (1994) ) G. A. Meyer, J. S. Warner, Y. K. Rao and H. H. Kellogg: Metall. Trans. 6B (1975) ) Slag Atlas: Verein Deutscher Eisenhüttenleute (VDEh), (1995) p ) J. M. Larrain: CALPHAD 3 (1979) ) O. Knacke, O. Kubaschewski and K. Hesselmann: Thermochemical Properties of Inorganic Substances, Second Edition, (Springer-Verlag, 1991) p ) O. Knacke, O. Kubaschewski and K. Hesselmann: Thermochemical Properties of Inorganic Substances, Second Edition, (Springer-Verlag, 1991) p ) O. Knacke, O. Kubaschewski and K. Hesselmann: Thermochemical Properties of Inorganic Substances, Second Edition, (Springer-Verlag, 1991) p ) Y. Takeda and A. Yazawa: Proceedings of Intern. Conf. on Productivity and Technology in the Metallurgical Industries, (TMS, Köln, 1989) pp ) J. Fincham and F. D. Richardson: Proc. Roy. Soc. A33 (1954) ) N. Sano, W. K. Lu, P. V. Riboud and M. Maeda: Advanced Physical Chemistry for Process Metallurgy. ed by N. Sano et al. (Academic Press, 1997) p ) M. M. Nzotta: Scan. J. Met. 6 (1997)